path: root/SOURCES/linux-surface.patch

From 36dd7450ff34e702af12d3a895810ee9e93abc63 Mon Sep 17 00:00:00 2001
From: Tsuchiya Yuto <kitakar@gmail.com>
Date: Sun, 18 Oct 2020 16:42:44 +0900
Subject: [PATCH] (surface3-oemb) add DMI matches for Surface 3 with broken DMI
 table

On some Surface 3, the DMI table gets corrupted for unknown reasons
and breaks existing DMI matching used for device-specific quirks.

This commit adds the (broken) DMI data into dmi_system_id tables used
for quirks so that each driver can enable quirks even on the affected
systems.

On affected systems, DMI data will look like this:
    $ grep . /sys/devices/virtual/dmi/id/{bios_vendor,board_name,board_vendor,\
    chassis_vendor,product_name,sys_vendor}
    /sys/devices/virtual/dmi/id/bios_vendor:American Megatrends Inc.
    /sys/devices/virtual/dmi/id/board_name:OEMB
    /sys/devices/virtual/dmi/id/board_vendor:OEMB
    /sys/devices/virtual/dmi/id/chassis_vendor:OEMB
    /sys/devices/virtual/dmi/id/product_name:OEMB
    /sys/devices/virtual/dmi/id/sys_vendor:OEMB

Expected:
    $ grep . /sys/devices/virtual/dmi/id/{bios_vendor,board_name,board_vendor,\
    chassis_vendor,product_name,sys_vendor}
    /sys/devices/virtual/dmi/id/bios_vendor:American Megatrends Inc.
    /sys/devices/virtual/dmi/id/board_name:Surface 3
    /sys/devices/virtual/dmi/id/board_vendor:Microsoft Corporation
    /sys/devices/virtual/dmi/id/chassis_vendor:Microsoft Corporation
    /sys/devices/virtual/dmi/id/product_name:Surface 3
    /sys/devices/virtual/dmi/id/sys_vendor:Microsoft Corporation

Signed-off-by: Tsuchiya Yuto <kitakar@gmail.com>
Patchset: surface3-oemb
---
 drivers/platform/surface/surface3-wmi.c           | 7 +++++++
 sound/soc/codecs/rt5645.c                         | 9 +++++++++
 sound/soc/intel/common/soc-acpi-intel-cht-match.c | 8 ++++++++
 3 files changed, 24 insertions(+)

diff --git a/drivers/platform/surface/surface3-wmi.c b/drivers/platform/surface/surface3-wmi.c
index ca4602bcc7dea..490b9731068ae 100644
--- a/drivers/platform/surface/surface3-wmi.c
+++ b/drivers/platform/surface/surface3-wmi.c
@@ -37,6 +37,13 @@ static const struct dmi_system_id surface3_dmi_table[] = {
 			DMI_MATCH(DMI_PRODUCT_NAME, "Surface 3"),
 		},
 	},
+	{
+		.matches = {
+			DMI_MATCH(DMI_BIOS_VENDOR, "American Megatrends Inc."),
+			DMI_MATCH(DMI_SYS_VENDOR, "OEMB"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OEMB"),
+		},
+	},
 #endif
 	{ }
 };
diff --git a/sound/soc/codecs/rt5645.c b/sound/soc/codecs/rt5645.c
index 620ecbfa4a7a8..b07d06d2971a8 100644
--- a/sound/soc/codecs/rt5645.c
+++ b/sound/soc/codecs/rt5645.c
@@ -3717,6 +3717,15 @@ static const struct dmi_system_id dmi_platform_data[] = {
 		},
 		.driver_data = (void *)&intel_braswell_platform_data,
 	},
+	{
+		.ident = "Microsoft Surface 3",
+		.matches = {
+			DMI_MATCH(DMI_BIOS_VENDOR, "American Megatrends Inc."),
+			DMI_MATCH(DMI_SYS_VENDOR, "OEMB"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OEMB"),
+		},
+		.driver_data = (void *)&intel_braswell_platform_data,
+	},
 	{
 		/*
 		 * Match for the GPDwin which unfortunately uses somewhat
diff --git a/sound/soc/intel/common/soc-acpi-intel-cht-match.c b/sound/soc/intel/common/soc-acpi-intel-cht-match.c
index 6beb00858c33f..d82d77387a0a6 100644
--- a/sound/soc/intel/common/soc-acpi-intel-cht-match.c
+++ b/sound/soc/intel/common/soc-acpi-intel-cht-match.c
@@ -27,6 +27,14 @@ static const struct dmi_system_id cht_table[] = {
 			DMI_MATCH(DMI_PRODUCT_NAME, "Surface 3"),
 		},
 	},
+	{
+		.callback = cht_surface_quirk_cb,
+		.matches = {
+			DMI_MATCH(DMI_BIOS_VENDOR, "American Megatrends Inc."),
+			DMI_MATCH(DMI_SYS_VENDOR, "OEMB"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OEMB"),
+		},
+	},
 	{ }
 };
 
-- 
2.39.2

From 3651879493891a6b14c68983d4730b6700acb4aa Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jonas=20Dre=C3=9Fler?= <verdre@v0yd.nl>
Date: Tue, 3 Nov 2020 13:28:04 +0100
Subject: [PATCH] mwifiex: Add quirk resetting the PCI bridge on MS Surface
 devices

The most recent firmware of the 88W8897 card reports a hardcoded LTR
value to the system during initialization, probably as an (unsuccessful)
attempt of the developers to fix firmware crashes. This LTR value
prevents most of the Microsoft Surface devices from entering deep
powersaving states (either platform C-State 10 or S0ix state), because
the exit latency of that state would be higher than what the card can
tolerate.

Turns out the card works just the same (including the firmware crashes)
no matter if that hardcoded LTR value is reported or not, so it's kind
of useless and only prevents us from saving power.

To get rid of those hardcoded LTR reports, it's possible to reset the
PCI bridge device after initializing the cards firmware. I'm not exactly
sure why that works, maybe the power management subsystem of the PCH
resets its stored LTR values when doing a function level reset of the
bridge device. Doing the reset once after starting the wifi firmware
works very well, probably because the firmware only reports that LTR
value a single time during firmware startup.

Patchset: mwifiex
---
 drivers/net/wireless/marvell/mwifiex/pcie.c   | 12 +++++++++
 .../wireless/marvell/mwifiex/pcie_quirks.c    | 26 +++++++++++++------
 .../wireless/marvell/mwifiex/pcie_quirks.h    |  1 +
 3 files changed, 31 insertions(+), 8 deletions(-)

diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
index 5dcf61761a165..84be9289caa63 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
@@ -1762,9 +1762,21 @@ mwifiex_pcie_send_boot_cmd(struct mwifiex_adapter *adapter, struct sk_buff *skb)
 static int mwifiex_pcie_init_fw_port(struct mwifiex_adapter *adapter)
 {
 	struct pcie_service_card *card = adapter->card;
+	struct pci_dev *pdev = card->dev;
+	struct pci_dev *parent_pdev = pci_upstream_bridge(pdev);
 	const struct mwifiex_pcie_card_reg *reg = card->pcie.reg;
 	int tx_wrap = card->txbd_wrptr & reg->tx_wrap_mask;
 
+	/* Trigger a function level reset of the PCI bridge device, this makes
+	 * the firmware of PCIe 88W8897 cards stop reporting a fixed LTR value
+	 * that prevents the system from entering package C10 and S0ix powersaving
+	 * states.
+	 * We need to do it here because it must happen after firmware
+	 * initialization and this function is called after that is done.
+	 */
+	if (card->quirks & QUIRK_DO_FLR_ON_BRIDGE)
+		pci_reset_function(parent_pdev);
+
 	/* Write the RX ring read pointer in to reg->rx_rdptr */
 	if (mwifiex_write_reg(adapter, reg->rx_rdptr, card->rxbd_rdptr |
 			      tx_wrap)) {
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
index dd6d21f1dbfd7..f46b06f8d6435 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
@@ -13,7 +13,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Pro 4"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Pro 5",
@@ -22,7 +23,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_SKU, "Surface_Pro_1796"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Pro 5 (LTE)",
@@ -31,7 +33,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_SKU, "Surface_Pro_1807"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Pro 6",
@@ -39,7 +42,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Pro 6"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Book 1",
@@ -47,7 +51,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Book"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Book 2",
@@ -55,7 +60,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Book 2"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Laptop 1",
@@ -63,7 +69,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Laptop"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{
 		.ident = "Surface Laptop 2",
@@ -71,7 +78,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Laptop 2"),
 		},
-		.driver_data = (void *)QUIRK_FW_RST_D3COLD,
+		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
+					QUIRK_DO_FLR_ON_BRIDGE),
 	},
 	{}
 };
@@ -89,6 +97,8 @@ void mwifiex_initialize_quirks(struct pcie_service_card *card)
 		dev_info(&pdev->dev, "no quirks enabled\n");
 	if (card->quirks & QUIRK_FW_RST_D3COLD)
 		dev_info(&pdev->dev, "quirk reset_d3cold enabled\n");
+	if (card->quirks & QUIRK_DO_FLR_ON_BRIDGE)
+		dev_info(&pdev->dev, "quirk do_flr_on_bridge enabled\n");
 }
 
 static void mwifiex_pcie_set_power_d3cold(struct pci_dev *pdev)
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
index d6ff964aec5bf..5d30ae39d65ec 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
+++ b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
@@ -4,6 +4,7 @@
 #include "pcie.h"
 
 #define QUIRK_FW_RST_D3COLD	BIT(0)
+#define QUIRK_DO_FLR_ON_BRIDGE	BIT(1)
 
 void mwifiex_initialize_quirks(struct pcie_service_card *card);
 int mwifiex_pcie_reset_d3cold_quirk(struct pci_dev *pdev);
-- 
2.39.2

From a239303211cb7eff1a6358c8514065a06032f31c Mon Sep 17 00:00:00 2001
From: Tsuchiya Yuto <kitakar@gmail.com>
Date: Sun, 4 Oct 2020 00:11:49 +0900
Subject: [PATCH] mwifiex: pcie: disable bridge_d3 for Surface gen4+

Currently, mwifiex fw will crash after suspend on recent kernel series.
On Windows, it seems that the root port of wifi will never enter D3 state
(stay on D0 state). And on Linux, disabling the D3 state for the
bridge fixes fw crashing after suspend.

This commit disables the D3 state of root port on driver initialization
and fixes fw crashing after suspend.

Signed-off-by: Tsuchiya Yuto <kitakar@gmail.com>
Patchset: mwifiex
---
 drivers/net/wireless/marvell/mwifiex/pcie.c   |  7 +++++
 .../wireless/marvell/mwifiex/pcie_quirks.c    | 27 +++++++++++++------
 .../wireless/marvell/mwifiex/pcie_quirks.h    |  1 +
 3 files changed, 27 insertions(+), 8 deletions(-)

diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
index 84be9289caa63..98be0d3cc41cf 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
@@ -368,6 +368,7 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
 					const struct pci_device_id *ent)
 {
 	struct pcie_service_card *card;
+	struct pci_dev *parent_pdev = pci_upstream_bridge(pdev);
 	int ret;
 
 	pr_debug("info: vendor=0x%4.04X device=0x%4.04X rev=%d\n",
@@ -409,6 +410,12 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
 		return -1;
 	}
 
+	/* disable bridge_d3 for Surface gen4+ devices to fix fw crashing
+	 * after suspend
+	 */
+	if (card->quirks & QUIRK_NO_BRIDGE_D3)
+		parent_pdev->bridge_d3 = false;
+
 	return 0;
 }
 
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
index f46b06f8d6435..99b024ecbadea 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.c
@@ -14,7 +14,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Pro 4"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Pro 5",
@@ -24,7 +25,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_SKU, "Surface_Pro_1796"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Pro 5 (LTE)",
@@ -34,7 +36,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_SKU, "Surface_Pro_1807"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Pro 6",
@@ -43,7 +46,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Pro 6"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Book 1",
@@ -52,7 +56,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Book"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Book 2",
@@ -61,7 +66,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Book 2"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Laptop 1",
@@ -70,7 +76,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Laptop"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{
 		.ident = "Surface Laptop 2",
@@ -79,7 +86,8 @@ static const struct dmi_system_id mwifiex_quirk_table[] = {
 			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Surface Laptop 2"),
 		},
 		.driver_data = (void *)(QUIRK_FW_RST_D3COLD |
-					QUIRK_DO_FLR_ON_BRIDGE),
+					QUIRK_DO_FLR_ON_BRIDGE |
+					QUIRK_NO_BRIDGE_D3),
 	},
 	{}
 };
@@ -99,6 +107,9 @@ void mwifiex_initialize_quirks(struct pcie_service_card *card)
 		dev_info(&pdev->dev, "quirk reset_d3cold enabled\n");
 	if (card->quirks & QUIRK_DO_FLR_ON_BRIDGE)
 		dev_info(&pdev->dev, "quirk do_flr_on_bridge enabled\n");
+	if (card->quirks & QUIRK_NO_BRIDGE_D3)
+		dev_info(&pdev->dev,
+			 "quirk no_brigde_d3 enabled\n");
 }
 
 static void mwifiex_pcie_set_power_d3cold(struct pci_dev *pdev)
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
index 5d30ae39d65ec..c14eb56eb9118 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
+++ b/drivers/net/wireless/marvell/mwifiex/pcie_quirks.h
@@ -5,6 +5,7 @@
 
 #define QUIRK_FW_RST_D3COLD	BIT(0)
 #define QUIRK_DO_FLR_ON_BRIDGE	BIT(1)
+#define QUIRK_NO_BRIDGE_D3	BIT(2)
 
 void mwifiex_initialize_quirks(struct pcie_service_card *card);
 int mwifiex_pcie_reset_d3cold_quirk(struct pci_dev *pdev);
-- 
2.39.2

From 7d5b4f5687544cadf760ebe14b28e31253c9e0e2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jonas=20Dre=C3=9Fler?= <verdre@v0yd.nl>
Date: Thu, 25 Mar 2021 11:33:02 +0100
Subject: [PATCH] Bluetooth: btusb: Lower passive lescan interval on Marvell
 88W8897

The Marvell 88W8897 combined wifi and bluetooth card (pcie+usb version)
is used in a lot of Microsoft Surface devices, and all those devices
suffer from very low 2.4GHz wifi connection speeds while bluetooth is
enabled. The reason for that is that the default passive scanning
interval for Bluetooth Low Energy devices is quite high in Linux
(interval of 60 msec and scan window of 30 msec, see hci_core.c), and
the Marvell chip is known for its bad bt+wifi coexisting performance.

So decrease that passive scan interval and make the scan window shorter
on this particular device to allow for spending more time transmitting
wifi signals: The new scan interval is 250 msec (0x190 * 0.625 msec) and
the new scan window is 6.25 msec (0xa * 0,625 msec).

This change has a very large impact on the 2.4GHz wifi speeds and gets
it up to performance comparable with the Windows driver, which seems to
apply a similar quirk.

The interval and window length were tested and found to work very well
with a lot of Bluetooth Low Energy devices, including the Surface Pen, a
Bluetooth Speaker and two modern Bluetooth headphones. All devices were
discovered immediately after turning them on. Even lower values were
also tested, but they introduced longer delays until devices get
discovered.

Patchset: mwifiex
---
 drivers/bluetooth/btusb.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index 952dc9d2404ed..b455885cdc0d8 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -64,6 +64,7 @@ static struct usb_driver btusb_driver;
 #define BTUSB_INTEL_BROKEN_INITIAL_NCMD BIT(25)
 #define BTUSB_INTEL_NO_WBS_SUPPORT	BIT(26)
 #define BTUSB_ACTIONS_SEMI		BIT(27)
+#define BTUSB_LOWER_LESCAN_INTERVAL	BIT(28)
 
 static const struct usb_device_id btusb_table[] = {
 	/* Generic Bluetooth USB device */
@@ -467,6 +468,7 @@ static const struct usb_device_id blacklist_table[] = {
 	{ USB_DEVICE(0x1286, 0x2044), .driver_info = BTUSB_MARVELL },
 	{ USB_DEVICE(0x1286, 0x2046), .driver_info = BTUSB_MARVELL },
 	{ USB_DEVICE(0x1286, 0x204e), .driver_info = BTUSB_MARVELL },
+	{ USB_DEVICE(0x1286, 0x204c), .driver_info = BTUSB_LOWER_LESCAN_INTERVAL },
 
 	/* Intel Bluetooth devices */
 	{ USB_DEVICE(0x8087, 0x0025), .driver_info = BTUSB_INTEL_COMBINED },
@@ -3957,6 +3959,19 @@ static int btusb_probe(struct usb_interface *intf,
 	if (id->driver_info & BTUSB_MARVELL)
 		hdev->set_bdaddr = btusb_set_bdaddr_marvell;
 
+	/* The Marvell 88W8897 combined wifi and bluetooth card is known for
+	 * very bad bt+wifi coexisting performance.
+	 *
+	 * Decrease the passive BT Low Energy scan interval a bit
+	 * (0x0190 * 0.625 msec = 250 msec) and make the scan window shorter
+	 * (0x000a * 0,625 msec = 6.25 msec). This allows for significantly
+	 * higher wifi throughput while passively scanning for BT LE devices.
+	 */
+	if (id->driver_info & BTUSB_LOWER_LESCAN_INTERVAL) {
+		hdev->le_scan_interval = 0x0190;
+		hdev->le_scan_window = 0x000a;
+	}
+
 	if (IS_ENABLED(CONFIG_BT_HCIBTUSB_MTK) &&
 	    (id->driver_info & BTUSB_MEDIATEK)) {
 		hdev->setup = btusb_mtk_setup;
-- 
2.39.2

From dbe93645ca7fe34649b6b9c964c18a1bc5e17cac Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sat, 27 Feb 2021 00:45:52 +0100
Subject: [PATCH] ath10k: Add module parameters to override board files

Some Surface devices, specifically the Surface Go and AMD version of the
Surface Laptop 3 (wich both come with QCA6174 WiFi chips), work better
with a different board file, as it seems that the firmeware included
upstream is buggy.

As it is generally not a good idea to randomly overwrite files, let
alone doing so via packages, we add module parameters to override those
file names in the driver. This allows us to package/deploy the override
via a modprobe.d config.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: ath10k
---
 drivers/net/wireless/ath/ath10k/core.c | 58 ++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)

diff --git a/drivers/net/wireless/ath/ath10k/core.c b/drivers/net/wireless/ath/ath10k/core.c
index 5eb131ab916fd..67f074a126d1f 100644
--- a/drivers/net/wireless/ath/ath10k/core.c
+++ b/drivers/net/wireless/ath/ath10k/core.c
@@ -38,6 +38,9 @@ static bool fw_diag_log;
 /* frame mode values are mapped as per enum ath10k_hw_txrx_mode */
 unsigned int ath10k_frame_mode = ATH10K_HW_TXRX_NATIVE_WIFI;
 
+static char *override_board = "";
+static char *override_board2 = "";
+
 unsigned long ath10k_coredump_mask = BIT(ATH10K_FW_CRASH_DUMP_REGISTERS) |
 				     BIT(ATH10K_FW_CRASH_DUMP_CE_DATA);
 
@@ -50,6 +53,9 @@ module_param(fw_diag_log, bool, 0644);
 module_param_named(frame_mode, ath10k_frame_mode, uint, 0644);
 module_param_named(coredump_mask, ath10k_coredump_mask, ulong, 0444);
 
+module_param(override_board, charp, 0644);
+module_param(override_board2, charp, 0644);
+
 MODULE_PARM_DESC(debug_mask, "Debugging mask");
 MODULE_PARM_DESC(uart_print, "Uart target debugging");
 MODULE_PARM_DESC(skip_otp, "Skip otp failure for calibration in testmode");
@@ -59,6 +65,9 @@ MODULE_PARM_DESC(frame_mode,
 MODULE_PARM_DESC(coredump_mask, "Bitfield of what to include in firmware crash file");
 MODULE_PARM_DESC(fw_diag_log, "Diag based fw log debugging");
 
+MODULE_PARM_DESC(override_board, "Override for board.bin file");
+MODULE_PARM_DESC(override_board2, "Override for board-2.bin file");
+
 static const struct ath10k_hw_params ath10k_hw_params_list[] = {
 	{
 		.id = QCA988X_HW_2_0_VERSION,
@@ -911,6 +920,42 @@ static int ath10k_init_configure_target(struct ath10k *ar)
 	return 0;
 }
 
+static const char *ath10k_override_board_fw_file(struct ath10k *ar,
+						 const char *file)
+{
+	if (strcmp(file, "board.bin") == 0) {
+		if (strcmp(override_board, "") == 0)
+			return file;
+
+		if (strcmp(override_board, "none") == 0) {
+			dev_info(ar->dev, "firmware override: pretending 'board.bin' does not exist\n");
+			return NULL;
+		}
+
+		dev_info(ar->dev, "firmware override: replacing 'board.bin' with '%s'\n",
+			 override_board);
+
+		return override_board;
+	}
+
+	if (strcmp(file, "board-2.bin") == 0) {
+		if (strcmp(override_board2, "") == 0)
+			return file;
+
+		if (strcmp(override_board2, "none") == 0) {
+			dev_info(ar->dev, "firmware override: pretending 'board-2.bin' does not exist\n");
+			return NULL;
+		}
+
+		dev_info(ar->dev, "firmware override: replacing 'board-2.bin' with '%s'\n",
+			 override_board2);
+
+		return override_board2;
+	}
+
+	return file;
+}
+
 static const struct firmware *ath10k_fetch_fw_file(struct ath10k *ar,
 						   const char *dir,
 						   const char *file)
@@ -925,6 +970,19 @@ static const struct firmware *ath10k_fetch_fw_file(struct ath10k *ar,
 	if (dir == NULL)
 		dir = ".";
 
+	/* HACK: Override board.bin and board-2.bin files if specified.
+	 *
+	 * Some Surface devices perform better with a different board
+	 * configuration. To this end, one would need to replace the board.bin
+	 * file with the modified config and remove the board-2.bin file.
+	 * Unfortunately, that's not a solution that we can easily package. So
+	 * we add module options to perform these overrides here.
+	 */
+
+	file = ath10k_override_board_fw_file(ar, file);
+	if (!file)
+		return ERR_PTR(-ENOENT);
+
 	snprintf(filename, sizeof(filename), "%s/%s", dir, file);
 	ret = firmware_request_nowarn(&fw, filename, ar->dev);
 	ath10k_dbg(ar, ATH10K_DBG_BOOT, "boot fw request '%s': %d\n",
-- 
2.39.2

From 0599e41661359b23763519f74cf74b87a7892df7 Mon Sep 17 00:00:00 2001
From: Dorian Stoll <dorian.stoll@tmsp.io>
Date: Thu, 30 Jul 2020 13:21:53 +0200
Subject: [PATCH] misc: mei: Add missing IPTS device IDs

Patchset: ipts
---
 drivers/misc/mei/hw-me-regs.h | 1 +
 drivers/misc/mei/pci-me.c     | 1 +
 2 files changed, 2 insertions(+)

diff --git a/drivers/misc/mei/hw-me-regs.h b/drivers/misc/mei/hw-me-regs.h
index bdc65d50b945f..08723c01d7275 100644
--- a/drivers/misc/mei/hw-me-regs.h
+++ b/drivers/misc/mei/hw-me-regs.h
@@ -92,6 +92,7 @@
 #define MEI_DEV_ID_CDF        0x18D3  /* Cedar Fork */
 
 #define MEI_DEV_ID_ICP_LP     0x34E0  /* Ice Lake Point LP */
+#define MEI_DEV_ID_ICP_LP_3   0x34E4  /* Ice Lake Point LP 3 (iTouch) */
 #define MEI_DEV_ID_ICP_N      0x38E0  /* Ice Lake Point N */
 
 #define MEI_DEV_ID_JSP_N      0x4DE0  /* Jasper Lake Point N */
diff --git a/drivers/misc/mei/pci-me.c b/drivers/misc/mei/pci-me.c
index 5bf0d50d55a00..c13864512229f 100644
--- a/drivers/misc/mei/pci-me.c
+++ b/drivers/misc/mei/pci-me.c
@@ -97,6 +97,7 @@ static const struct pci_device_id mei_me_pci_tbl[] = {
 	{MEI_PCI_DEVICE(MEI_DEV_ID_CMP_H_3, MEI_ME_PCH8_ITOUCH_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_ICP_LP, MEI_ME_PCH12_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_ICP_LP_3, MEI_ME_PCH12_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_ICP_N, MEI_ME_PCH12_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_TGP_LP, MEI_ME_PCH15_CFG)},
-- 
2.39.2

From ca537e0653e2843c08d900d4c3c6bb294b7e7693 Mon Sep 17 00:00:00 2001
From: Liban Hannan <liban.p@gmail.com>
Date: Tue, 12 Apr 2022 23:31:12 +0100
Subject: [PATCH] iommu: ipts: use IOMMU passthrough mode for IPTS

Adds a quirk so that IOMMU uses passthrough mode for the IPTS device.
Otherwise, when IOMMU is enabled, IPTS produces DMAR errors like:

DMAR: [DMA Read NO_PASID] Request device [00:16.4] fault addr
0x104ea3000 [fault reason 0x06] PTE Read access is not set

This is very similar to the bug described at:
https://bugs.launchpad.net/bugs/1958004

Fixed with the following patch which this patch basically copies:
https://launchpadlibrarian.net/586396847/43255ca.diff
Patchset: ipts
---
 drivers/iommu/intel/iommu.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index d4b5d20bd6dda..3965979c7bd41 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -37,6 +37,8 @@
 #define IS_GFX_DEVICE(pdev) ((pdev->class >> 16) == PCI_BASE_CLASS_DISPLAY)
 #define IS_USB_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_SERIAL_USB)
 #define IS_ISA_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA)
+#define IS_IPTS(pdev) ((pdev)->vendor == PCI_VENDOR_ID_INTEL &&	\
+			    ((pdev)->device == 0x9d3e))
 #define IS_AZALIA(pdev) ((pdev)->vendor == 0x8086 && (pdev)->device == 0x3a3e)
 
 #define IOAPIC_RANGE_START	(0xfee00000)
@@ -286,12 +288,14 @@ int intel_iommu_enabled = 0;
 EXPORT_SYMBOL_GPL(intel_iommu_enabled);
 
 static int dmar_map_gfx = 1;
+static int dmar_map_ipts = 1;
 static int intel_iommu_superpage = 1;
 static int iommu_identity_mapping;
 static int iommu_skip_te_disable;
 
 #define IDENTMAP_GFX		2
 #define IDENTMAP_AZALIA		4
+#define IDENTMAP_IPTS		16
 
 const struct iommu_ops intel_iommu_ops;
 
@@ -2630,6 +2634,9 @@ static int device_def_domain_type(struct device *dev)
 
 		if ((iommu_identity_mapping & IDENTMAP_GFX) && IS_GFX_DEVICE(pdev))
 			return IOMMU_DOMAIN_IDENTITY;
+
+		if ((iommu_identity_mapping & IDENTMAP_IPTS) && IS_IPTS(pdev))
+			return IOMMU_DOMAIN_IDENTITY;
 	}
 
 	return 0;
@@ -3019,6 +3026,9 @@ static int __init init_dmars(void)
 	if (!dmar_map_gfx)
 		iommu_identity_mapping |= IDENTMAP_GFX;
 
+	if (!dmar_map_ipts)
+		iommu_identity_mapping |= IDENTMAP_IPTS;
+
 	check_tylersburg_isoch();
 
 	ret = si_domain_init(hw_pass_through);
@@ -4788,6 +4798,17 @@ static void quirk_iommu_igfx(struct pci_dev *dev)
 	dmar_map_gfx = 0;
 }
 
+static void quirk_iommu_ipts(struct pci_dev *dev)
+{
+	if (!IS_IPTS(dev))
+		return;
+
+	if (risky_device(dev))
+		return;
+
+	pci_info(dev, "Passthrough IOMMU for IPTS\n");
+	dmar_map_ipts = 0;
+}
 /* G4x/GM45 integrated gfx dmar support is totally busted. */
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2a40, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2e00, quirk_iommu_igfx);
@@ -4823,6 +4844,9 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x1632, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x163A, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x163D, quirk_iommu_igfx);
 
+/* disable IPTS dmar support */
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x9D3E, quirk_iommu_ipts);
+
 static void quirk_iommu_rwbf(struct pci_dev *dev)
 {
 	if (risky_device(dev))
-- 
2.39.2

From 6fd49d6f1a0ead6f8e466fbac8f029b169e477f3 Mon Sep 17 00:00:00 2001
From: Dorian Stoll <dorian.stoll@tmsp.io>
Date: Sun, 11 Dec 2022 12:00:59 +0100
Subject: [PATCH] hid: Add support for Intel Precise Touch and Stylus

Based on linux-surface/intel-precise-touch@8abe268

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>
Patchset: ipts
---
 drivers/hid/Kconfig            |   2 +
 drivers/hid/Makefile           |   2 +
 drivers/hid/ipts/Kconfig       |  14 +
 drivers/hid/ipts/Makefile      |  14 +
 drivers/hid/ipts/cmd.c         |  62 +++++
 drivers/hid/ipts/cmd.h         |  61 ++++
 drivers/hid/ipts/context.h     |  51 ++++
 drivers/hid/ipts/control.c     | 495 +++++++++++++++++++++++++++++++++
 drivers/hid/ipts/control.h     | 127 +++++++++
 drivers/hid/ipts/desc.h        |  81 ++++++
 drivers/hid/ipts/hid.c         | 348 +++++++++++++++++++++++
 drivers/hid/ipts/hid.h         |  22 ++
 drivers/hid/ipts/main.c        | 127 +++++++++
 drivers/hid/ipts/mei.c         | 189 +++++++++++++
 drivers/hid/ipts/mei.h         |  67 +++++
 drivers/hid/ipts/receiver.c    | 249 +++++++++++++++++
 drivers/hid/ipts/receiver.h    |  17 ++
 drivers/hid/ipts/resources.c   | 108 +++++++
 drivers/hid/ipts/resources.h   |  39 +++
 drivers/hid/ipts/spec-data.h   | 100 +++++++
 drivers/hid/ipts/spec-device.h | 285 +++++++++++++++++++
 drivers/hid/ipts/spec-hid.h    |  35 +++
 drivers/hid/ipts/thread.c      |  85 ++++++
 drivers/hid/ipts/thread.h      |  60 ++++
 24 files changed, 2640 insertions(+)
 create mode 100644 drivers/hid/ipts/Kconfig
 create mode 100644 drivers/hid/ipts/Makefile
 create mode 100644 drivers/hid/ipts/cmd.c
 create mode 100644 drivers/hid/ipts/cmd.h
 create mode 100644 drivers/hid/ipts/context.h
 create mode 100644 drivers/hid/ipts/control.c
 create mode 100644 drivers/hid/ipts/control.h
 create mode 100644 drivers/hid/ipts/desc.h
 create mode 100644 drivers/hid/ipts/hid.c
 create mode 100644 drivers/hid/ipts/hid.h
 create mode 100644 drivers/hid/ipts/main.c
 create mode 100644 drivers/hid/ipts/mei.c
 create mode 100644 drivers/hid/ipts/mei.h
 create mode 100644 drivers/hid/ipts/receiver.c
 create mode 100644 drivers/hid/ipts/receiver.h
 create mode 100644 drivers/hid/ipts/resources.c
 create mode 100644 drivers/hid/ipts/resources.h
 create mode 100644 drivers/hid/ipts/spec-data.h
 create mode 100644 drivers/hid/ipts/spec-device.h
 create mode 100644 drivers/hid/ipts/spec-hid.h
 create mode 100644 drivers/hid/ipts/thread.c
 create mode 100644 drivers/hid/ipts/thread.h

diff --git a/drivers/hid/Kconfig b/drivers/hid/Kconfig
index 185a077d59cdd..1523ccdf73b51 100644
--- a/drivers/hid/Kconfig
+++ b/drivers/hid/Kconfig
@@ -1290,4 +1290,6 @@ source "drivers/hid/amd-sfh-hid/Kconfig"
 
 source "drivers/hid/surface-hid/Kconfig"
 
+source "drivers/hid/ipts/Kconfig"
+
 endmenu
diff --git a/drivers/hid/Makefile b/drivers/hid/Makefile
index e8014c1a2f8b6..e48300bcea9be 100644
--- a/drivers/hid/Makefile
+++ b/drivers/hid/Makefile
@@ -164,3 +164,5 @@ obj-$(INTEL_ISH_FIRMWARE_DOWNLOADER)	+= intel-ish-hid/
 obj-$(CONFIG_AMD_SFH_HID)       += amd-sfh-hid/
 
 obj-$(CONFIG_SURFACE_HID_CORE)  += surface-hid/
+
+obj-$(CONFIG_HID_IPTS)          += ipts/
diff --git a/drivers/hid/ipts/Kconfig b/drivers/hid/ipts/Kconfig
new file mode 100644
index 0000000000000..297401bd388dd
--- /dev/null
+++ b/drivers/hid/ipts/Kconfig
@@ -0,0 +1,14 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+
+config HID_IPTS
+	tristate "Intel Precise Touch & Stylus"
+	depends on INTEL_MEI
+	depends on HID
+	help
+	  Say Y here if your system has a touchscreen using Intels
+	  Precise Touch & Stylus (IPTS) technology.
+
+	  If unsure say N.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called ipts.
diff --git a/drivers/hid/ipts/Makefile b/drivers/hid/ipts/Makefile
new file mode 100644
index 0000000000000..0fe655bccdc0f
--- /dev/null
+++ b/drivers/hid/ipts/Makefile
@@ -0,0 +1,14 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+#
+# Makefile for the IPTS touchscreen driver
+#
+
+obj-$(CONFIG_HID_IPTS) += ipts.o
+ipts-objs := cmd.o
+ipts-objs += control.o
+ipts-objs += hid.o
+ipts-objs += main.o
+ipts-objs += mei.o
+ipts-objs += receiver.o
+ipts-objs += resources.o
+ipts-objs += thread.o
diff --git a/drivers/hid/ipts/cmd.c b/drivers/hid/ipts/cmd.c
new file mode 100644
index 0000000000000..7fd69271ccd52
--- /dev/null
+++ b/drivers/hid/ipts/cmd.c
@@ -0,0 +1,62 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/errno.h>
+#include <linux/types.h>
+
+#include "cmd.h"
+#include "context.h"
+#include "mei.h"
+#include "spec-device.h"
+
+int ipts_cmd_recv_timeout(struct ipts_context *ipts, enum ipts_command_code code,
+			  struct ipts_response *rsp, u64 timeout)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!rsp)
+		return -EFAULT;
+
+	/*
+	 * In a response, the command code will have the most significant bit flipped to 1.
+	 * If code is passed to ipts_mei_recv as is, no messages will be received.
+	 */
+	ret = ipts_mei_recv(&ipts->mei, code | IPTS_RSP_BIT, rsp, timeout);
+	if (ret < 0)
+		return ret;
+
+	dev_dbg(ipts->dev, "Received 0x%02X with status 0x%02X\n", code, rsp->status);
+
+	/*
+	 * Some devices will always return this error.
+	 * It is allowed to ignore it and to try continuing.
+	 */
+	if (rsp->status == IPTS_STATUS_COMPAT_CHECK_FAIL)
+		rsp->status = IPTS_STATUS_SUCCESS;
+
+	return 0;
+}
+
+int ipts_cmd_send(struct ipts_context *ipts, enum ipts_command_code code, void *data, size_t size)
+{
+	struct ipts_command cmd = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	cmd.cmd = code;
+
+	if (data && size > 0)
+		memcpy(cmd.payload, data, size);
+
+	dev_dbg(ipts->dev, "Sending 0x%02X with %ld bytes payload\n", code, size);
+	return ipts_mei_send(&ipts->mei, &cmd, sizeof(cmd.cmd) + size);
+}
diff --git a/drivers/hid/ipts/cmd.h b/drivers/hid/ipts/cmd.h
new file mode 100644
index 0000000000000..924758ffee671
--- /dev/null
+++ b/drivers/hid/ipts/cmd.h
@@ -0,0 +1,61 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_CMD_H
+#define IPTS_CMD_H
+
+#include <linux/types.h>
+
+#include "context.h"
+#include "spec-device.h"
+
+/*
+ * The default timeout for receiving responses
+ */
+#define IPTS_CMD_DEFAULT_TIMEOUT 1000
+
+/*
+ * ipts_cmd_recv_timeout() - Receives a response to a command.
+ * @ipts: The IPTS driver context.
+ * @code: The type of the command / response.
+ * @rsp: The address that the received response will be copied to.
+ * @timeout: How many milliseconds the function will wait at most.
+ *
+ * A negative timeout means to wait forever.
+ *
+ * Returns: 0 on success, <0 on error, -EAGAIN if no response has been received.
+ */
+int ipts_cmd_recv_timeout(struct ipts_context *ipts, enum ipts_command_code code,
+			  struct ipts_response *rsp, u64 timeout);
+
+/*
+ * ipts_cmd_recv() - Receives a response to a command.
+ * @ipts: The IPTS driver context.
+ * @code: The type of the command / response.
+ * @rsp: The address that the received response will be copied to.
+ *
+ * Returns: 0 on success, <0 on error, -EAGAIN if no response has been received.
+ */
+static inline int ipts_cmd_recv(struct ipts_context *ipts, enum ipts_command_code code,
+				struct ipts_response *rsp)
+{
+	return ipts_cmd_recv_timeout(ipts, code, rsp, IPTS_CMD_DEFAULT_TIMEOUT);
+}
+
+/*
+ * ipts_cmd_send() - Executes a command on the device.
+ * @ipts: The IPTS driver context.
+ * @code: The type of the command to execute.
+ * @data: The payload containing parameters for the command.
+ * @size: The size of the payload.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_cmd_send(struct ipts_context *ipts, enum ipts_command_code code, void *data, size_t size);
+
+#endif /* IPTS_CMD_H */
diff --git a/drivers/hid/ipts/context.h b/drivers/hid/ipts/context.h
new file mode 100644
index 0000000000000..3450a95e66ee8
--- /dev/null
+++ b/drivers/hid/ipts/context.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_CONTEXT_H
+#define IPTS_CONTEXT_H
+
+#include <linux/completion.h>
+#include <linux/device.h>
+#include <linux/hid.h>
+#include <linux/mei_cl_bus.h>
+#include <linux/mutex.h>
+#include <linux/sched.h>
+#include <linux/types.h>
+
+#include "mei.h"
+#include "resources.h"
+#include "spec-device.h"
+#include "thread.h"
+
+struct ipts_context {
+	struct device *dev;
+	struct ipts_mei mei;
+
+	enum ipts_mode mode;
+
+	/*
+	 * Prevents concurrent GET_FEATURE reports.
+	 */
+	struct mutex feature_lock;
+	struct completion feature_event;
+
+	/*
+	 * These are not inside of struct ipts_resources
+	 * because they don't own the memory they point to.
+	 */
+	struct ipts_buffer feature_report;
+	struct ipts_buffer descriptor;
+
+	struct hid_device *hid;
+	struct ipts_device_info info;
+	struct ipts_resources resources;
+
+	struct ipts_thread receiver_loop;
+};
+
+#endif /* IPTS_CONTEXT_H */
diff --git a/drivers/hid/ipts/control.c b/drivers/hid/ipts/control.c
new file mode 100644
index 0000000000000..2f61500b5119c
--- /dev/null
+++ b/drivers/hid/ipts/control.c
@@ -0,0 +1,495 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/delay.h>
+#include <linux/dev_printk.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/kthread.h>
+#include <linux/types.h>
+
+#include "cmd.h"
+#include "context.h"
+#include "control.h"
+#include "desc.h"
+#include "hid.h"
+#include "receiver.h"
+#include "resources.h"
+#include "spec-data.h"
+#include "spec-device.h"
+
+static int ipts_control_get_device_info(struct ipts_context *ipts, struct ipts_device_info *info)
+{
+	int ret = 0;
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!info)
+		return -EFAULT;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_GET_DEVICE_INFO, NULL, 0);
+	if (ret) {
+		dev_err(ipts->dev, "GET_DEVICE_INFO: send failed: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_GET_DEVICE_INFO, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "GET_DEVICE_INFO: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "GET_DEVICE_INFO: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	memcpy(info, rsp.payload, sizeof(*info));
+	return 0;
+}
+
+static int ipts_control_set_mode(struct ipts_context *ipts, enum ipts_mode mode)
+{
+	int ret = 0;
+	struct ipts_set_mode cmd = { 0 };
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	cmd.mode = mode;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_SET_MODE, &cmd, sizeof(cmd));
+	if (ret) {
+		dev_err(ipts->dev, "SET_MODE: send failed: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_SET_MODE, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "SET_MODE: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "SET_MODE: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	return 0;
+}
+
+static int ipts_control_set_mem_window(struct ipts_context *ipts, struct ipts_resources *res)
+{
+	int ret = 0;
+	struct ipts_mem_window cmd = { 0 };
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!res)
+		return -EFAULT;
+
+	for (int i = 0; i < IPTS_BUFFERS; i++) {
+		cmd.data_addr_lower[i] = lower_32_bits(res->data[i].dma_address);
+		cmd.data_addr_upper[i] = upper_32_bits(res->data[i].dma_address);
+		cmd.feedback_addr_lower[i] = lower_32_bits(res->feedback[i].dma_address);
+		cmd.feedback_addr_upper[i] = upper_32_bits(res->feedback[i].dma_address);
+	}
+
+	cmd.workqueue_addr_lower = lower_32_bits(res->workqueue.dma_address);
+	cmd.workqueue_addr_upper = upper_32_bits(res->workqueue.dma_address);
+
+	cmd.doorbell_addr_lower = lower_32_bits(res->doorbell.dma_address);
+	cmd.doorbell_addr_upper = upper_32_bits(res->doorbell.dma_address);
+
+	cmd.hid2me_addr_lower = lower_32_bits(res->hid2me.dma_address);
+	cmd.hid2me_addr_upper = upper_32_bits(res->hid2me.dma_address);
+
+	cmd.workqueue_size = IPTS_WORKQUEUE_SIZE;
+	cmd.workqueue_item_size = IPTS_WORKQUEUE_ITEM_SIZE;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_SET_MEM_WINDOW, &cmd, sizeof(cmd));
+	if (ret) {
+		dev_err(ipts->dev, "SET_MEM_WINDOW: send failed: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_SET_MEM_WINDOW, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "SET_MEM_WINDOW: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "SET_MEM_WINDOW: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	return 0;
+}
+
+static int ipts_control_get_descriptor(struct ipts_context *ipts)
+{
+	int ret = 0;
+	struct ipts_data_header *header = NULL;
+	struct ipts_get_descriptor cmd = { 0 };
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!ipts->resources.descriptor.address)
+		return -EFAULT;
+
+	memset(ipts->resources.descriptor.address, 0, ipts->resources.descriptor.size);
+
+	cmd.addr_lower = lower_32_bits(ipts->resources.descriptor.dma_address);
+	cmd.addr_upper = upper_32_bits(ipts->resources.descriptor.dma_address);
+	cmd.magic = 8;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_GET_DESCRIPTOR, &cmd, sizeof(cmd));
+	if (ret) {
+		dev_err(ipts->dev, "GET_DESCRIPTOR: send failed: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_GET_DESCRIPTOR, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "GET_DESCRIPTOR: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "GET_DESCRIPTOR: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	header = (struct ipts_data_header *)ipts->resources.descriptor.address;
+
+	if (header->type == IPTS_DATA_TYPE_DESCRIPTOR) {
+		ipts->descriptor.address = &header->data[8];
+		ipts->descriptor.size = header->size - 8;
+
+		return 0;
+	}
+
+	return -ENODATA;
+}
+
+int ipts_control_request_flush(struct ipts_context *ipts)
+{
+	int ret = 0;
+	struct ipts_quiesce_io cmd = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_QUIESCE_IO, &cmd, sizeof(cmd));
+	if (ret)
+		dev_err(ipts->dev, "QUIESCE_IO: send failed: %d\n", ret);
+
+	return ret;
+}
+
+int ipts_control_wait_flush(struct ipts_context *ipts)
+{
+	int ret = 0;
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_QUIESCE_IO, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "QUIESCE_IO: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	if (rsp.status == IPTS_STATUS_TIMEOUT)
+		return -EAGAIN;
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "QUIESCE_IO: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	return 0;
+}
+
+int ipts_control_request_data(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_READY_FOR_DATA, NULL, 0);
+	if (ret)
+		dev_err(ipts->dev, "READY_FOR_DATA: send failed: %d\n", ret);
+
+	return ret;
+}
+
+int ipts_control_wait_data(struct ipts_context *ipts, bool shutdown)
+{
+	int ret = 0;
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!shutdown)
+		ret = ipts_cmd_recv_timeout(ipts, IPTS_CMD_READY_FOR_DATA, &rsp, 0);
+	else
+		ret = ipts_cmd_recv(ipts, IPTS_CMD_READY_FOR_DATA, &rsp);
+
+	if (ret) {
+		if (ret != -EAGAIN)
+			dev_err(ipts->dev, "READY_FOR_DATA: recv failed: %d\n", ret);
+
+		return ret;
+	}
+
+	/*
+	 * During shutdown, it is possible that the sensor has already been disabled.
+	 */
+	if (rsp.status == IPTS_STATUS_SENSOR_DISABLED)
+		return 0;
+
+	if (rsp.status == IPTS_STATUS_TIMEOUT)
+		return -EAGAIN;
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "READY_FOR_DATA: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	return 0;
+}
+
+int ipts_control_send_feedback(struct ipts_context *ipts, u32 buffer)
+{
+	int ret = 0;
+	struct ipts_feedback cmd = { 0 };
+	struct ipts_response rsp = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	cmd.buffer = buffer;
+
+	ret = ipts_cmd_send(ipts, IPTS_CMD_FEEDBACK, &cmd, sizeof(cmd));
+	if (ret) {
+		dev_err(ipts->dev, "FEEDBACK: send failed: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_cmd_recv(ipts, IPTS_CMD_FEEDBACK, &rsp);
+	if (ret) {
+		dev_err(ipts->dev, "FEEDBACK: recv failed: %d\n", ret);
+		return ret;
+	}
+
+	/*
+	 * We don't know what feedback data looks like so we are sending zeros.
+	 * See also ipts_control_refill_buffer.
+	 */
+	if (rsp.status == IPTS_STATUS_INVALID_PARAMS)
+		return 0;
+
+	if (rsp.status != IPTS_STATUS_SUCCESS) {
+		dev_err(ipts->dev, "FEEDBACK: cmd failed: %d\n", rsp.status);
+		return -EBADR;
+	}
+
+	return 0;
+}
+
+int ipts_control_hid2me_feedback(struct ipts_context *ipts, enum ipts_feedback_cmd_type cmd,
+				 enum ipts_feedback_data_type type, void *data, size_t size)
+{
+	struct ipts_feedback_header *header = NULL;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!ipts->resources.hid2me.address)
+		return -EFAULT;
+
+	memset(ipts->resources.hid2me.address, 0, ipts->resources.hid2me.size);
+	header = (struct ipts_feedback_header *)ipts->resources.hid2me.address;
+
+	header->cmd_type = cmd;
+	header->data_type = type;
+	header->size = size;
+	header->buffer = IPTS_HID2ME_BUFFER;
+
+	if (size + sizeof(*header) > ipts->resources.hid2me.size)
+		return -EINVAL;
+
+	if (data && size > 0)
+		memcpy(header->payload, data, size);
+
+	return ipts_control_send_feedback(ipts, IPTS_HID2ME_BUFFER);
+}
+
+static inline int ipts_control_reset_sensor(struct ipts_context *ipts)
+{
+	return ipts_control_hid2me_feedback(ipts, IPTS_FEEDBACK_CMD_TYPE_SOFT_RESET,
+					    IPTS_FEEDBACK_DATA_TYPE_VENDOR, NULL, 0);
+}
+
+int ipts_control_start(struct ipts_context *ipts)
+{
+	int ret = 0;
+	struct ipts_device_info info = { 0 };
+
+	if (!ipts)
+		return -EFAULT;
+
+	dev_info(ipts->dev, "Starting IPTS\n");
+
+	ret = ipts_control_get_device_info(ipts, &info);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to get device info: %d\n", ret);
+		return ret;
+	}
+
+	ipts->info = info;
+
+	ret = ipts_resources_init(&ipts->resources, ipts->dev, info.data_size, info.feedback_size);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to allocate buffers: %d", ret);
+		return ret;
+	}
+
+	dev_info(ipts->dev, "IPTS EDS Version: %d\n", info.intf_eds);
+
+	/*
+	 * Handle newer devices
+	 */
+	if (info.intf_eds > 1) {
+		/*
+		 * Fetching the descriptor will only work on newer devices.
+		 * For older devices, a fallback descriptor will be used.
+		 */
+		ret = ipts_control_get_descriptor(ipts);
+		if (ret) {
+			dev_err(ipts->dev, "Failed to fetch HID descriptor: %d\n", ret);
+			return ret;
+		}
+
+		/*
+		 * Newer devices can be directly initialized in doorbell mode.
+		 */
+		ipts->mode = IPTS_MODE_DOORBELL;
+	}
+
+	ret = ipts_control_set_mode(ipts, ipts->mode);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to set mode: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_control_set_mem_window(ipts, &ipts->resources);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to set memory window: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_receiver_start(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to start receiver: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_control_request_data(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to request data: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_hid_init(ipts, info);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to initialize HID device: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int _ipts_control_stop(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	dev_info(ipts->dev, "Stopping IPTS\n");
+
+	ret = ipts_receiver_stop(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to stop receiver: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_control_reset_sensor(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to reset sensor: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_resources_free(&ipts->resources);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to free resources: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+int ipts_control_stop(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	ret = _ipts_control_stop(ipts);
+	if (ret)
+		return ret;
+
+	ret = ipts_hid_free(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to free HID device: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+int ipts_control_restart(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	ret = _ipts_control_stop(ipts);
+	if (ret)
+		return ret;
+
+	/*
+	 * Give the sensor some time to come back from resetting
+	 */
+	msleep(1000);
+
+	ret = ipts_control_start(ipts);
+	if (ret)
+		return ret;
+
+	return 0;
+}
diff --git a/drivers/hid/ipts/control.h b/drivers/hid/ipts/control.h
new file mode 100644
index 0000000000000..744bb92d682a0
--- /dev/null
+++ b/drivers/hid/ipts/control.h
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_CONTROL_H
+#define IPTS_CONTROL_H
+
+#include <linux/types.h>
+
+#include "context.h"
+#include "spec-data.h"
+#include "spec-device.h"
+
+/*
+ * ipts_control_request_flush() - Stop the data flow.
+ * @ipts: The IPTS driver context.
+ *
+ * Runs the command to stop the data flow on the device.
+ * All outstanding data needs to be acknowledged using feedback before the command will return.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_request_flush(struct ipts_context *ipts);
+
+/*
+ * ipts_control_wait_flush() - Wait until data flow has been stopped.
+ * @ipts: The IPTS driver context.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_wait_flush(struct ipts_context *ipts);
+
+/*
+ * ipts_control_wait_flush() - Notify the device that the driver can receive new data.
+ * @ipts: The IPTS driver context.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_request_data(struct ipts_context *ipts);
+
+/*
+ * ipts_control_wait_data() - Wait until new data is available.
+ * @ipts: The IPTS driver context.
+ * @block: Whether to block execution until data is available.
+ *
+ * In doorbell mode, this function will never return while the data flow is active. Instead,
+ * the doorbell will be incremented when new data is available.
+ *
+ * Returns: 0 on success, <0 on error, -EAGAIN if no data is available.
+ */
+int ipts_control_wait_data(struct ipts_context *ipts, bool block);
+
+/*
+ * ipts_control_send_feedback() - Submits a feedback buffer to the device.
+ * @ipts: The IPTS driver context.
+ * @buffer: The ID of the buffer containing feedback data.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_send_feedback(struct ipts_context *ipts, u32 buffer);
+
+/*
+ * ipts_control_hid2me_feedback() - Sends HID2ME feedback, a special type of feedback.
+ * @ipts: The IPTS driver context.
+ * @cmd: The command that will be run on the device.
+ * @type: The type of the payload that is sent to the device.
+ * @data: The payload of the feedback command.
+ * @size: The size of the payload.
+ *
+ * HID2ME feedback is a special type of feedback, because it allows interfacing with
+ * the HID API of the device at any moment, without requiring a buffer that has to
+ * be acknowledged.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_hid2me_feedback(struct ipts_context *ipts, enum ipts_feedback_cmd_type cmd,
+				 enum ipts_feedback_data_type type, void *data, size_t size);
+
+/*
+ * ipts_control_refill_buffer() - Acknowledges that data in a buffer has been processed.
+ * @ipts: The IPTS driver context.
+ * @buffer: The buffer that has been processed and can be refilled.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+static inline int ipts_control_refill_buffer(struct ipts_context *ipts, u32 buffer)
+{
+	/*
+	 * IPTS expects structured data in the feedback buffer matching the buffer that will be
+	 * refilled. We don't know what that data looks like, so we just keep the buffer empty.
+	 * This results in an INVALID_PARAMS error, but the buffer gets refilled without an issue.
+	 * Sending a minimal structure with the buffer ID fixes the error, but breaks refilling
+	 * the buffers on some devices.
+	 */
+
+	return ipts_control_send_feedback(ipts, buffer);
+}
+
+/*
+ * ipts_control_start() - Initialized the device and starts the data flow.
+ * @ipts: The IPTS driver context.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_start(struct ipts_context *ipts);
+
+/*
+ * ipts_control_stop() - Stops the data flow and resets the device.
+ * @ipts: The IPTS driver context.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_stop(struct ipts_context *ipts);
+
+/*
+ * ipts_control_restart() - Stops the device and starts it again.
+ * @ipts: The IPTS driver context.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_control_restart(struct ipts_context *ipts);
+
+#endif /* IPTS_CONTROL_H */
diff --git a/drivers/hid/ipts/desc.h b/drivers/hid/ipts/desc.h
new file mode 100644
index 0000000000000..c058974a03a1e
--- /dev/null
+++ b/drivers/hid/ipts/desc.h
@@ -0,0 +1,81 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2022-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_DESC_H
+#define IPTS_DESC_H
+
+#include <linux/types.h>
+
+#define IPTS_HID_REPORT_SINGLETOUCH 64
+#define IPTS_HID_REPORT_DATA	    65
+#define IPTS_HID_REPORT_SET_MODE    66
+
+#define IPTS_HID_REPORT_DATA_SIZE 7485
+
+/*
+ * HID descriptor for singletouch data.
+ * This descriptor should be present on all IPTS devices.
+ */
+static const u8 ipts_singletouch_descriptor[] = {
+	0x05, 0x0D,	  /*  Usage Page (Digitizer),            */
+	0x09, 0x04,	  /*  Usage (Touchscreen),               */
+	0xA1, 0x01,	  /*  Collection (Application),          */
+	0x85, 0x40,	  /*      Report ID (64),                */
+	0x09, 0x42,	  /*      Usage (Tip Switch),            */
+	0x15, 0x00,	  /*      Logical Minimum (0),           */
+	0x25, 0x01,	  /*      Logical Maximum (1),           */
+	0x75, 0x01,	  /*      Report Size (1),               */
+	0x95, 0x01,	  /*      Report Count (1),              */
+	0x81, 0x02,	  /*      Input (Variable),              */
+	0x95, 0x07,	  /*      Report Count (7),              */
+	0x81, 0x03,	  /*      Input (Constant, Variable),    */
+	0x05, 0x01,	  /*      Usage Page (Desktop),          */
+	0x09, 0x30,	  /*      Usage (X),                     */
+	0x75, 0x10,	  /*      Report Size (16),              */
+	0x95, 0x01,	  /*      Report Count (1),              */
+	0xA4,		  /*      Push,                          */
+	0x55, 0x0E,	  /*      Unit Exponent (14),            */
+	0x65, 0x11,	  /*      Unit (Centimeter),             */
+	0x46, 0x76, 0x0B, /*      Physical Maximum (2934),       */
+	0x26, 0xFF, 0x7F, /*      Logical Maximum (32767),       */
+	0x81, 0x02,	  /*      Input (Variable),              */
+	0x09, 0x31,	  /*      Usage (Y),                     */
+	0x46, 0x74, 0x06, /*      Physical Maximum (1652),       */
+	0x26, 0xFF, 0x7F, /*      Logical Maximum (32767),       */
+	0x81, 0x02,	  /*      Input (Variable),              */
+	0xB4,		  /*      Pop,                           */
+	0xC0,		  /*  End Collection                     */
+};
+
+/*
+ * Fallback HID descriptor for older devices that do not have
+ * the ability to query their HID descriptor.
+ */
+static const u8 ipts_fallback_descriptor[] = {
+	0x05, 0x0D,	  /*  Usage Page (Digitizer),            */
+	0x09, 0x0F,	  /*  Usage (Capacitive Hm Digitizer),   */
+	0xA1, 0x01,	  /*  Collection (Application),          */
+	0x85, 0x41,	  /*      Report ID (65),                */
+	0x09, 0x56,	  /*      Usage (Scan Time),             */
+	0x95, 0x01,	  /*      Report Count (1),              */
+	0x75, 0x10,	  /*      Report Size (16),              */
+	0x81, 0x02,	  /*      Input (Variable),              */
+	0x09, 0x61,	  /*      Usage (Gesture Char Quality),  */
+	0x75, 0x08,	  /*      Report Size (8),               */
+	0x96, 0x3D, 0x1D, /*      Report Count (7485),           */
+	0x81, 0x03,	  /*      Input (Constant, Variable),    */
+	0x85, 0x42,	  /*      Report ID (66),                */
+	0x06, 0x00, 0xFF, /*      Usage Page (FF00h),            */
+	0x09, 0xC8,	  /*      Usage (C8h),                   */
+	0x75, 0x08,	  /*      Report Size (8),               */
+	0x95, 0x01,	  /*      Report Count (1),              */
+	0xB1, 0x02,	  /*      Feature (Variable),            */
+	0xC0,		  /*  End Collection,                    */
+};
+
+#endif /* IPTS_DESC_H */
diff --git a/drivers/hid/ipts/hid.c b/drivers/hid/ipts/hid.c
new file mode 100644
index 0000000000000..6782394e8dde3
--- /dev/null
+++ b/drivers/hid/ipts/hid.c
@@ -0,0 +1,348 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2022-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/completion.h>
+#include <linux/gfp.h>
+#include <linux/hid.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+
+#include "context.h"
+#include "control.h"
+#include "desc.h"
+#include "hid.h"
+#include "spec-data.h"
+#include "spec-device.h"
+#include "spec-hid.h"
+
+static int ipts_hid_start(struct hid_device *hid)
+{
+	return 0;
+}
+
+static void ipts_hid_stop(struct hid_device *hid)
+{
+}
+
+static int ipts_hid_switch_mode(struct ipts_context *ipts, enum ipts_mode mode)
+{
+	if (!ipts)
+		return -EFAULT;
+
+	if (ipts->mode == mode)
+		return 0;
+
+	/*
+	 * This is only allowed on older devices.
+	 */
+	if (ipts->info.intf_eds > 1)
+		return 0;
+
+	ipts->mode = mode;
+	return ipts_control_restart(ipts);
+}
+
+static int ipts_hid_parse(struct hid_device *hid)
+{
+	int ret = 0;
+	struct ipts_context *ipts = NULL;
+
+	bool has_native_descriptor = false;
+
+	u8 *buffer = NULL;
+	size_t size = 0;
+
+	if (!hid)
+		return -ENODEV;
+
+	ipts = hid->driver_data;
+
+	if (!ipts)
+		return -EFAULT;
+
+	size = sizeof(ipts_singletouch_descriptor);
+	has_native_descriptor = ipts->descriptor.address && ipts->descriptor.size > 0;
+
+	if (has_native_descriptor)
+		size += ipts->descriptor.size;
+	else
+		size += sizeof(ipts_fallback_descriptor);
+
+	buffer = kzalloc(size, GFP_KERNEL);
+	if (!buffer)
+		return -ENOMEM;
+
+	memcpy(buffer, ipts_singletouch_descriptor, sizeof(ipts_singletouch_descriptor));
+
+	if (has_native_descriptor) {
+		memcpy(&buffer[sizeof(ipts_singletouch_descriptor)], ipts->descriptor.address,
+		       ipts->descriptor.size);
+	} else {
+		memcpy(&buffer[sizeof(ipts_singletouch_descriptor)], ipts_fallback_descriptor,
+		       sizeof(ipts_fallback_descriptor));
+	}
+
+	ret = hid_parse_report(hid, buffer, size);
+	kfree(buffer);
+
+	if (ret) {
+		dev_err(ipts->dev, "Failed to parse HID descriptor: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int ipts_hid_get_feature(struct ipts_context *ipts, unsigned char reportnum, __u8 *buf,
+				size_t size, enum ipts_feedback_data_type type)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!buf)
+		return -EFAULT;
+
+	mutex_lock(&ipts->feature_lock);
+
+	memset(buf, 0, size);
+	buf[0] = reportnum;
+
+	memset(&ipts->feature_report, 0, sizeof(ipts->feature_report));
+	reinit_completion(&ipts->feature_event);
+
+	ret = ipts_control_hid2me_feedback(ipts, IPTS_FEEDBACK_CMD_TYPE_NONE, type, buf, size);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to send hid2me feedback: %d\n", ret);
+		goto out;
+	}
+
+	ret = wait_for_completion_timeout(&ipts->feature_event, msecs_to_jiffies(5000));
+	if (ret == 0) {
+		dev_warn(ipts->dev, "GET_FEATURES timed out!\n");
+		ret = -EIO;
+		goto out;
+	}
+
+	if (!ipts->feature_report.address) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	if (ipts->feature_report.size > size) {
+		ret = -ETOOSMALL;
+		goto out;
+	}
+
+	ret = ipts->feature_report.size;
+	memcpy(buf, ipts->feature_report.address, ipts->feature_report.size);
+
+out:
+	mutex_unlock(&ipts->feature_lock);
+	return ret;
+}
+
+static int ipts_hid_set_feature(struct ipts_context *ipts, unsigned char reportnum, __u8 *buf,
+				size_t size, enum ipts_feedback_data_type type)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!buf)
+		return -EFAULT;
+
+	buf[0] = reportnum;
+
+	ret = ipts_control_hid2me_feedback(ipts, IPTS_FEEDBACK_CMD_TYPE_NONE, type, buf, size);
+	if (ret)
+		dev_err(ipts->dev, "Failed to send hid2me feedback: %d\n", ret);
+
+	return ret;
+}
+
+static int ipts_hid_raw_request(struct hid_device *hid, unsigned char reportnum, __u8 *buf,
+				size_t size, unsigned char rtype, int reqtype)
+{
+	int ret = 0;
+	struct ipts_context *ipts = NULL;
+
+	enum ipts_feedback_data_type type = IPTS_FEEDBACK_DATA_TYPE_VENDOR;
+
+	if (!hid)
+		return -ENODEV;
+
+	ipts = hid->driver_data;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!buf)
+		return -EFAULT;
+
+	if (rtype == HID_OUTPUT_REPORT && reqtype == HID_REQ_SET_REPORT)
+		type = IPTS_FEEDBACK_DATA_TYPE_OUTPUT_REPORT;
+	else if (rtype == HID_FEATURE_REPORT && reqtype == HID_REQ_GET_REPORT)
+		type = IPTS_FEEDBACK_DATA_TYPE_GET_FEATURES;
+	else if (rtype == HID_FEATURE_REPORT && reqtype == HID_REQ_SET_REPORT)
+		type = IPTS_FEEDBACK_DATA_TYPE_SET_FEATURES;
+	else
+		return -EIO;
+
+	// Implemente mode switching report for older devices without native HID support
+	if (type == IPTS_FEEDBACK_DATA_TYPE_SET_FEATURES && reportnum == IPTS_HID_REPORT_SET_MODE) {
+		ret = ipts_hid_switch_mode(ipts, buf[1]);
+		if (ret) {
+			dev_err(ipts->dev, "Failed to switch modes: %d\n", ret);
+			return ret;
+		}
+	}
+
+	if (reqtype == HID_REQ_GET_REPORT)
+		return ipts_hid_get_feature(ipts, reportnum, buf, size, type);
+	else
+		return ipts_hid_set_feature(ipts, reportnum, buf, size, type);
+}
+
+static int ipts_hid_output_report(struct hid_device *hid, __u8 *data, size_t size)
+{
+	struct ipts_context *ipts = NULL;
+
+	if (!hid)
+		return -ENODEV;
+
+	ipts = hid->driver_data;
+
+	return ipts_control_hid2me_feedback(ipts, IPTS_FEEDBACK_CMD_TYPE_NONE,
+					    IPTS_FEEDBACK_DATA_TYPE_OUTPUT_REPORT, data, size);
+}
+
+static struct hid_ll_driver ipts_hid_driver = {
+	.start = ipts_hid_start,
+	.stop = ipts_hid_stop,
+	.open = ipts_hid_start,
+	.close = ipts_hid_stop,
+	.parse = ipts_hid_parse,
+	.raw_request = ipts_hid_raw_request,
+	.output_report = ipts_hid_output_report,
+};
+
+int ipts_hid_input_data(struct ipts_context *ipts, u32 buffer)
+{
+	int ret = 0;
+	u8 *temp = NULL;
+	struct ipts_hid_header *frame = NULL;
+	struct ipts_data_header *header = NULL;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (!ipts->hid)
+		return -ENODEV;
+
+	header = (struct ipts_data_header *)ipts->resources.data[buffer].address;
+
+	if (!header)
+		return -EFAULT;
+
+	if (header->size == 0)
+		return 0;
+
+	if (header->type == IPTS_DATA_TYPE_HID)
+		return hid_input_report(ipts->hid, HID_INPUT_REPORT, header->data, header->size, 1);
+
+	if (header->type == IPTS_DATA_TYPE_GET_FEATURES) {
+		ipts->feature_report.address = header->data;
+		ipts->feature_report.size = header->size;
+
+		complete_all(&ipts->feature_event);
+		return 0;
+	}
+
+	if (header->type != IPTS_DATA_TYPE_FRAME)
+		return 0;
+
+	if (header->size + 3 + sizeof(struct ipts_hid_header) > IPTS_HID_REPORT_DATA_SIZE)
+		return -ERANGE;
+
+	temp = kzalloc(IPTS_HID_REPORT_DATA_SIZE, GFP_KERNEL);
+	if (!temp)
+		return -ENOMEM;
+
+	/*
+	 * Synthesize a HID report matching the devices that natively send HID reports
+	 */
+	temp[0] = IPTS_HID_REPORT_DATA;
+
+	frame = (struct ipts_hid_header *)&temp[3];
+	frame->type = IPTS_HID_FRAME_TYPE_RAW;
+	frame->size = header->size + sizeof(*frame);
+
+	memcpy(frame->data, header->data, header->size);
+
+	ret = hid_input_report(ipts->hid, HID_INPUT_REPORT, temp, IPTS_HID_REPORT_DATA_SIZE, 1);
+	kfree(temp);
+
+	return ret;
+}
+
+int ipts_hid_init(struct ipts_context *ipts, struct ipts_device_info info)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (ipts->hid)
+		return 0;
+
+	ipts->hid = hid_allocate_device();
+	if (IS_ERR(ipts->hid)) {
+		int err = PTR_ERR(ipts->hid);
+
+		dev_err(ipts->dev, "Failed to allocate HID device: %d\n", err);
+		return err;
+	}
+
+	ipts->hid->driver_data = ipts;
+	ipts->hid->dev.parent = ipts->dev;
+	ipts->hid->ll_driver = &ipts_hid_driver;
+
+	ipts->hid->vendor = info.vendor;
+	ipts->hid->product = info.product;
+	ipts->hid->group = HID_GROUP_MULTITOUCH;
+
+	snprintf(ipts->hid->name, sizeof(ipts->hid->name), "IPTS %04X:%04X", info.vendor,
+		 info.product);
+
+	ret = hid_add_device(ipts->hid);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to add HID device: %d\n", ret);
+		ipts_hid_free(ipts);
+		return ret;
+	}
+
+	return 0;
+}
+
+int ipts_hid_free(struct ipts_context *ipts)
+{
+	if (!ipts)
+		return -EFAULT;
+
+	if (!ipts->hid)
+		return 0;
+
+	hid_destroy_device(ipts->hid);
+	ipts->hid = NULL;
+
+	return 0;
+}
diff --git a/drivers/hid/ipts/hid.h b/drivers/hid/ipts/hid.h
new file mode 100644
index 0000000000000..62bf3cd486081
--- /dev/null
+++ b/drivers/hid/ipts/hid.h
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2022-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_HID_H
+#define IPTS_HID_H
+
+#include <linux/types.h>
+
+#include "context.h"
+#include "spec-device.h"
+
+int ipts_hid_input_data(struct ipts_context *ipts, u32 buffer);
+
+int ipts_hid_init(struct ipts_context *ipts, struct ipts_device_info info);
+int ipts_hid_free(struct ipts_context *ipts);
+
+#endif /* IPTS_HID_H */
diff --git a/drivers/hid/ipts/main.c b/drivers/hid/ipts/main.c
new file mode 100644
index 0000000000000..0f20c6c08c38c
--- /dev/null
+++ b/drivers/hid/ipts/main.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/completion.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/mei_cl_bus.h>
+#include <linux/mod_devicetable.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/stddef.h>
+#include <linux/types.h>
+
+#include "context.h"
+#include "control.h"
+#include "mei.h"
+#include "receiver.h"
+#include "spec-device.h"
+
+/*
+ * The MEI client ID for IPTS functionality.
+ */
+#define IPTS_ID UUID_LE(0x3e8d0870, 0x271a, 0x4208, 0x8e, 0xb5, 0x9a, 0xcb, 0x94, 0x02, 0xae, 0x04)
+
+static int ipts_set_dma_mask(struct mei_cl_device *cldev)
+{
+	if (!cldev)
+		return -EFAULT;
+
+	if (!dma_coerce_mask_and_coherent(&cldev->dev, DMA_BIT_MASK(64)))
+		return 0;
+
+	return dma_coerce_mask_and_coherent(&cldev->dev, DMA_BIT_MASK(32));
+}
+
+static int ipts_probe(struct mei_cl_device *cldev, const struct mei_cl_device_id *id)
+{
+	int ret = 0;
+	struct ipts_context *ipts = NULL;
+
+	if (!cldev)
+		return -EFAULT;
+
+	ret = ipts_set_dma_mask(cldev);
+	if (ret) {
+		dev_err(&cldev->dev, "Failed to set DMA mask for IPTS: %d\n", ret);
+		return ret;
+	}
+
+	ret = mei_cldev_enable(cldev);
+	if (ret) {
+		dev_err(&cldev->dev, "Failed to enable MEI device: %d\n", ret);
+		return ret;
+	}
+
+	ipts = devm_kzalloc(&cldev->dev, sizeof(*ipts), GFP_KERNEL);
+	if (!ipts) {
+		mei_cldev_disable(cldev);
+		return -ENOMEM;
+	}
+
+	ret = ipts_mei_init(&ipts->mei, cldev);
+	if (ret) {
+		dev_err(&cldev->dev, "Failed to init MEI bus logic: %d\n", ret);
+		return ret;
+	}
+
+	ipts->dev = &cldev->dev;
+	ipts->mode = IPTS_MODE_EVENT;
+
+	mutex_init(&ipts->feature_lock);
+	init_completion(&ipts->feature_event);
+
+	mei_cldev_set_drvdata(cldev, ipts);
+
+	ret = ipts_control_start(ipts);
+	if (ret) {
+		dev_err(&cldev->dev, "Failed to start IPTS: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static void ipts_remove(struct mei_cl_device *cldev)
+{
+	int ret = 0;
+	struct ipts_context *ipts = NULL;
+
+	if (!cldev) {
+		pr_err("MEI device is NULL!");
+		return;
+	}
+
+	ipts = mei_cldev_get_drvdata(cldev);
+
+	ret = ipts_control_stop(ipts);
+	if (ret)
+		dev_err(&cldev->dev, "Failed to stop IPTS: %d\n", ret);
+
+	mei_cldev_disable(cldev);
+}
+
+static struct mei_cl_device_id ipts_device_id_table[] = {
+	{ .uuid = IPTS_ID, .version = MEI_CL_VERSION_ANY },
+	{},
+};
+MODULE_DEVICE_TABLE(mei, ipts_device_id_table);
+
+static struct mei_cl_driver ipts_driver = {
+	.id_table = ipts_device_id_table,
+	.name = "ipts",
+	.probe = ipts_probe,
+	.remove = ipts_remove,
+};
+module_mei_cl_driver(ipts_driver);
+
+MODULE_DESCRIPTION("IPTS touchscreen driver");
+MODULE_AUTHOR("Dorian Stoll <dorian.stoll@tmsp.io>");
+MODULE_LICENSE("GPL");
diff --git a/drivers/hid/ipts/mei.c b/drivers/hid/ipts/mei.c
new file mode 100644
index 0000000000000..26666fd99b0c7
--- /dev/null
+++ b/drivers/hid/ipts/mei.c
@@ -0,0 +1,189 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/device.h>
+#include <linux/errno.h>
+#include <linux/jiffies.h>
+#include <linux/list.h>
+#include <linux/mei_cl_bus.h>
+#include <linux/printk.h>
+#include <linux/rwsem.h>
+#include <linux/types.h>
+#include <linux/wait.h>
+
+#include "context.h"
+#include "mei.h"
+
+static void locked_list_add(struct list_head *new, struct list_head *head,
+			    struct rw_semaphore *lock)
+{
+	down_write(lock);
+	list_add(new, head);
+	up_write(lock);
+}
+
+static void locked_list_del(struct list_head *entry, struct rw_semaphore *lock)
+{
+	down_write(lock);
+	list_del(entry);
+	up_write(lock);
+}
+
+static void ipts_mei_incoming(struct mei_cl_device *cldev)
+{
+	ssize_t ret = 0;
+	struct ipts_mei_message *entry = NULL;
+	struct ipts_context *ipts = NULL;
+
+	if (!cldev) {
+		pr_err("MEI device is NULL!");
+		return;
+	}
+
+	ipts = mei_cldev_get_drvdata(cldev);
+	if (!ipts) {
+		pr_err("IPTS driver context is NULL!");
+		return;
+	}
+
+	entry = devm_kzalloc(ipts->dev, sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return;
+
+	INIT_LIST_HEAD(&entry->list);
+
+	do {
+		ret = mei_cldev_recv(cldev, (u8 *)&entry->rsp, sizeof(entry->rsp));
+	} while (ret == -EINTR);
+
+	if (ret < 0) {
+		dev_err(ipts->dev, "Error while reading response: %ld\n", ret);
+		return;
+	}
+
+	if (ret == 0) {
+		dev_err(ipts->dev, "Received empty response\n");
+		return;
+	}
+
+	locked_list_add(&entry->list, &ipts->mei.messages, &ipts->mei.message_lock);
+	wake_up_all(&ipts->mei.message_queue);
+}
+
+static int ipts_mei_search(struct ipts_mei *mei, enum ipts_command_code code,
+			   struct ipts_response *rsp)
+{
+	struct ipts_mei_message *entry = NULL;
+
+	if (!mei)
+		return -EFAULT;
+
+	if (!rsp)
+		return -EFAULT;
+
+	down_read(&mei->message_lock);
+
+	/*
+	 * Iterate over the list of received messages, and check if there is one
+	 * matching the requested command code.
+	 */
+	list_for_each_entry(entry, &mei->messages, list) {
+		if (entry->rsp.cmd == code)
+			break;
+	}
+
+	up_read(&mei->message_lock);
+
+	/*
+	 * If entry is not the list head, this means that the loop above has been stopped early,
+	 * and that we found a matching element. We drop the message from the list and return it.
+	 */
+	if (!list_entry_is_head(entry, &mei->messages, list)) {
+		locked_list_del(&entry->list, &mei->message_lock);
+
+		*rsp = entry->rsp;
+		devm_kfree(&mei->cldev->dev, entry);
+
+		return 0;
+	}
+
+	return -EAGAIN;
+}
+
+int ipts_mei_recv(struct ipts_mei *mei, enum ipts_command_code code, struct ipts_response *rsp,
+		  u64 timeout)
+{
+	int ret = 0;
+
+	if (!mei)
+		return -EFAULT;
+
+	/*
+	 * A timeout of 0 means check and return immideately.
+	 */
+	if (timeout == 0)
+		return ipts_mei_search(mei, code, rsp);
+
+	/*
+	 * A timeout of less than 0 means to wait forever.
+	 */
+	if (timeout < 0) {
+		wait_event(mei->message_queue, ipts_mei_search(mei, code, rsp) == 0);
+		return 0;
+	}
+
+	ret = wait_event_timeout(mei->message_queue, ipts_mei_search(mei, code, rsp) == 0,
+				 msecs_to_jiffies(timeout));
+
+	if (ret > 0)
+		return 0;
+
+	return -EAGAIN;
+}
+
+int ipts_mei_send(struct ipts_mei *mei, void *data, size_t length)
+{
+	int ret = 0;
+
+	if (!mei)
+		return -EFAULT;
+
+	if (!mei->cldev)
+		return -EFAULT;
+
+	if (!data)
+		return -EFAULT;
+
+	do {
+		ret = mei_cldev_send(mei->cldev, (u8 *)data, length);
+	} while (ret == -EINTR);
+
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+int ipts_mei_init(struct ipts_mei *mei, struct mei_cl_device *cldev)
+{
+	if (!mei)
+		return -EFAULT;
+
+	if (!cldev)
+		return -EFAULT;
+
+	mei->cldev = cldev;
+
+	INIT_LIST_HEAD(&mei->messages);
+	init_waitqueue_head(&mei->message_queue);
+	init_rwsem(&mei->message_lock);
+
+	mei_cldev_register_rx_cb(cldev, ipts_mei_incoming);
+
+	return 0;
+}
diff --git a/drivers/hid/ipts/mei.h b/drivers/hid/ipts/mei.h
new file mode 100644
index 0000000000000..eadacae54c400
--- /dev/null
+++ b/drivers/hid/ipts/mei.h
@@ -0,0 +1,67 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_MEI_H
+#define IPTS_MEI_H
+
+#include <linux/list.h>
+#include <linux/mei_cl_bus.h>
+#include <linux/rwsem.h>
+#include <linux/types.h>
+#include <linux/wait.h>
+
+#include "spec-device.h"
+
+struct ipts_mei_message {
+	struct list_head list;
+	struct ipts_response rsp;
+};
+
+struct ipts_mei {
+	struct mei_cl_device *cldev;
+
+	struct list_head messages;
+
+	wait_queue_head_t message_queue;
+	struct rw_semaphore message_lock;
+};
+
+/*
+ * ipts_mei_recv() - Receive data from a MEI device.
+ * @mei: The IPTS MEI device context.
+ * @code: The IPTS command code to look for.
+ * @rsp: The address that the received data will be copied to.
+ * @timeout: How many milliseconds the function will wait at most.
+ *
+ * A negative timeout means to wait forever.
+ *
+ * Returns: 0 on success, <0 on error, -EAGAIN if no response has been received.
+ */
+int ipts_mei_recv(struct ipts_mei *mei, enum ipts_command_code code, struct ipts_response *rsp,
+		  u64 timeout);
+
+/*
+ * ipts_mei_send() - Send data to a MEI device.
+ * @ipts: The IPTS MEI device context.
+ * @data: The data to send.
+ * @size: The size of the data.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_mei_send(struct ipts_mei *mei, void *data, size_t length);
+
+/*
+ * ipts_mei_init() - Initialize the MEI device context.
+ * @mei: The MEI device context to initialize.
+ * @cldev: The MEI device the context will be bound to.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_mei_init(struct ipts_mei *mei, struct mei_cl_device *cldev);
+
+#endif /* IPTS_MEI_H */
diff --git a/drivers/hid/ipts/receiver.c b/drivers/hid/ipts/receiver.c
new file mode 100644
index 0000000000000..77234f9e0e178
--- /dev/null
+++ b/drivers/hid/ipts/receiver.c
@@ -0,0 +1,249 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/kthread.h>
+#include <linux/time64.h>
+#include <linux/timekeeping.h>
+#include <linux/types.h>
+
+#include "cmd.h"
+#include "context.h"
+#include "control.h"
+#include "hid.h"
+#include "resources.h"
+#include "spec-device.h"
+#include "thread.h"
+
+static void ipts_receiver_next_doorbell(struct ipts_context *ipts)
+{
+	u32 *doorbell = (u32 *)ipts->resources.doorbell.address;
+	*doorbell = *doorbell + 1;
+}
+
+static u32 ipts_receiver_current_doorbell(struct ipts_context *ipts)
+{
+	u32 *doorbell = (u32 *)ipts->resources.doorbell.address;
+	return *doorbell;
+}
+
+static void ipts_receiver_backoff(time64_t last, u32 n)
+{
+	/*
+	 * If the last change was less than n seconds ago,
+	 * sleep for a shorter period so that new data can be
+	 * processed quickly. If there was no change for more than
+	 * n seconds, sleep longer to avoid wasting CPU cycles.
+	 */
+	if (last + n > ktime_get_seconds())
+		msleep(20);
+	else
+		msleep(200);
+}
+
+static int ipts_receiver_event_loop(struct ipts_thread *thread)
+{
+	int ret = 0;
+	u32 buffer = 0;
+
+	struct ipts_context *ipts = NULL;
+	time64_t last = ktime_get_seconds();
+
+	if (!thread)
+		return -EFAULT;
+
+	ipts = thread->data;
+
+	if (!ipts)
+		return -EFAULT;
+
+	dev_info(ipts->dev, "IPTS running in event mode\n");
+
+	while (!ipts_thread_should_stop(thread)) {
+		for (int i = 0; i < IPTS_BUFFERS; i++) {
+			ret = ipts_control_wait_data(ipts, false);
+			if (ret == -EAGAIN)
+				break;
+
+			if (ret) {
+				dev_err(ipts->dev, "Failed to wait for data: %d\n", ret);
+				continue;
+			}
+
+			buffer = ipts_receiver_current_doorbell(ipts) % IPTS_BUFFERS;
+			ipts_receiver_next_doorbell(ipts);
+
+			ret = ipts_hid_input_data(ipts, buffer);
+			if (ret)
+				dev_err(ipts->dev, "Failed to process buffer: %d\n", ret);
+
+			ret = ipts_control_refill_buffer(ipts, buffer);
+			if (ret)
+				dev_err(ipts->dev, "Failed to send feedback: %d\n", ret);
+
+			ret = ipts_control_request_data(ipts);
+			if (ret)
+				dev_err(ipts->dev, "Failed to request data: %d\n", ret);
+
+			last = ktime_get_seconds();
+		}
+
+		ipts_receiver_backoff(last, 5);
+	}
+
+	ret = ipts_control_request_flush(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to request flush: %d\n", ret);
+		return ret;
+	}
+
+	ret = ipts_control_wait_data(ipts, true);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to wait for data: %d\n", ret);
+
+		if (ret != -EAGAIN)
+			return ret;
+		else
+			return 0;
+	}
+
+	ret = ipts_control_wait_flush(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to wait for flush: %d\n", ret);
+
+		if (ret != -EAGAIN)
+			return ret;
+		else
+			return 0;
+	}
+
+	return 0;
+}
+
+static int ipts_receiver_doorbell_loop(struct ipts_thread *thread)
+{
+	int ret = 0;
+	u32 buffer = 0;
+
+	u32 doorbell = 0;
+	u32 lastdb = 0;
+
+	struct ipts_context *ipts = NULL;
+	time64_t last = ktime_get_seconds();
+
+	if (!thread)
+		return -EFAULT;
+
+	ipts = thread->data;
+
+	if (!ipts)
+		return -EFAULT;
+
+	dev_info(ipts->dev, "IPTS running in doorbell mode\n");
+
+	while (true) {
+		if (ipts_thread_should_stop(thread)) {
+			ret = ipts_control_request_flush(ipts);
+			if (ret) {
+				dev_err(ipts->dev, "Failed to request flush: %d\n", ret);
+				return ret;
+			}
+		}
+
+		doorbell = ipts_receiver_current_doorbell(ipts);
+
+		/*
+		 * After filling up one of the data buffers, IPTS will increment
+		 * the doorbell. The value of the doorbell stands for the *next*
+		 * buffer that IPTS is going to fill.
+		 */
+		while (lastdb != doorbell) {
+			buffer = lastdb % IPTS_BUFFERS;
+
+			ret = ipts_hid_input_data(ipts, buffer);
+			if (ret)
+				dev_err(ipts->dev, "Failed to process buffer: %d\n", ret);
+
+			ret = ipts_control_refill_buffer(ipts, buffer);
+			if (ret)
+				dev_err(ipts->dev, "Failed to send feedback: %d\n", ret);
+
+			last = ktime_get_seconds();
+			lastdb++;
+		}
+
+		if (ipts_thread_should_stop(thread))
+			break;
+
+		ipts_receiver_backoff(last, 5);
+	}
+
+	ret = ipts_control_wait_data(ipts, true);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to wait for data: %d\n", ret);
+
+		if (ret != -EAGAIN)
+			return ret;
+		else
+			return 0;
+	}
+
+	ret = ipts_control_wait_flush(ipts);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to wait for flush: %d\n", ret);
+
+		if (ret != -EAGAIN)
+			return ret;
+		else
+			return 0;
+	}
+
+	return 0;
+}
+
+int ipts_receiver_start(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	if (ipts->mode == IPTS_MODE_EVENT) {
+		ret = ipts_thread_start(&ipts->receiver_loop, ipts_receiver_event_loop, ipts,
+					"ipts_event");
+	} else if (ipts->mode == IPTS_MODE_DOORBELL) {
+		ret = ipts_thread_start(&ipts->receiver_loop, ipts_receiver_doorbell_loop, ipts,
+					"ipts_doorbell");
+	} else {
+		ret = -EINVAL;
+	}
+
+	if (ret) {
+		dev_err(ipts->dev, "Failed to start receiver loop: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+int ipts_receiver_stop(struct ipts_context *ipts)
+{
+	int ret = 0;
+
+	if (!ipts)
+		return -EFAULT;
+
+	ret = ipts_thread_stop(&ipts->receiver_loop);
+	if (ret) {
+		dev_err(ipts->dev, "Failed to stop receiver loop: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
diff --git a/drivers/hid/ipts/receiver.h b/drivers/hid/ipts/receiver.h
new file mode 100644
index 0000000000000..96070f34fbcaa
--- /dev/null
+++ b/drivers/hid/ipts/receiver.h
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_RECEIVER_H
+#define IPTS_RECEIVER_H
+
+#include "context.h"
+
+int ipts_receiver_start(struct ipts_context *ipts);
+int ipts_receiver_stop(struct ipts_context *ipts);
+
+#endif /* IPTS_RECEIVER_H */
diff --git a/drivers/hid/ipts/resources.c b/drivers/hid/ipts/resources.c
new file mode 100644
index 0000000000000..80ba5885bb55d
--- /dev/null
+++ b/drivers/hid/ipts/resources.c
@@ -0,0 +1,108 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/types.h>
+
+#include "resources.h"
+#include "spec-device.h"
+
+static int ipts_resources_alloc_buffer(struct ipts_buffer *buffer, struct device *dev, size_t size)
+{
+	if (!buffer)
+		return -EFAULT;
+
+	if (buffer->address)
+		return 0;
+
+	buffer->address = dma_alloc_coherent(dev, size, &buffer->dma_address, GFP_KERNEL);
+
+	if (!buffer->address)
+		return -ENOMEM;
+
+	buffer->size = size;
+	buffer->device = dev;
+
+	return 0;
+}
+
+static void ipts_resources_free_buffer(struct ipts_buffer *buffer)
+{
+	if (!buffer->address)
+		return;
+
+	dma_free_coherent(buffer->device, buffer->size, buffer->address, buffer->dma_address);
+
+	buffer->address = NULL;
+	buffer->size = 0;
+
+	buffer->dma_address = 0;
+	buffer->device = NULL;
+}
+
+int ipts_resources_init(struct ipts_resources *res, struct device *dev, size_t ds, size_t fs)
+{
+	int ret = 0;
+
+	if (!res)
+		return -EFAULT;
+
+	for (int i = 0; i < IPTS_BUFFERS; i++) {
+		ret = ipts_resources_alloc_buffer(&res->data[i], dev, ds);
+		if (ret)
+			goto err;
+	}
+
+	for (int i = 0; i < IPTS_BUFFERS; i++) {
+		ret = ipts_resources_alloc_buffer(&res->feedback[i], dev, fs);
+		if (ret)
+			goto err;
+	}
+
+	ret = ipts_resources_alloc_buffer(&res->doorbell, dev, sizeof(u32));
+	if (ret)
+		goto err;
+
+	ret = ipts_resources_alloc_buffer(&res->workqueue, dev, sizeof(u32));
+	if (ret)
+		goto err;
+
+	ret = ipts_resources_alloc_buffer(&res->hid2me, dev, fs);
+	if (ret)
+		goto err;
+
+	ret = ipts_resources_alloc_buffer(&res->descriptor, dev, ds + 8);
+	if (ret)
+		goto err;
+
+	return 0;
+
+err:
+
+	ipts_resources_free(res);
+	return ret;
+}
+
+int ipts_resources_free(struct ipts_resources *res)
+{
+	if (!res)
+		return -EFAULT;
+
+	for (int i = 0; i < IPTS_BUFFERS; i++)
+		ipts_resources_free_buffer(&res->data[i]);
+
+	for (int i = 0; i < IPTS_BUFFERS; i++)
+		ipts_resources_free_buffer(&res->feedback[i]);
+
+	ipts_resources_free_buffer(&res->doorbell);
+	ipts_resources_free_buffer(&res->workqueue);
+	ipts_resources_free_buffer(&res->hid2me);
+	ipts_resources_free_buffer(&res->descriptor);
+
+	return 0;
+}
diff --git a/drivers/hid/ipts/resources.h b/drivers/hid/ipts/resources.h
new file mode 100644
index 0000000000000..6cbb24a8a0543
--- /dev/null
+++ b/drivers/hid/ipts/resources.h
@@ -0,0 +1,39 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_RESOURCES_H
+#define IPTS_RESOURCES_H
+
+#include <linux/device.h>
+#include <linux/types.h>
+
+#include "spec-device.h"
+
+struct ipts_buffer {
+	u8 *address;
+	size_t size;
+
+	dma_addr_t dma_address;
+	struct device *device;
+};
+
+struct ipts_resources {
+	struct ipts_buffer data[IPTS_BUFFERS];
+	struct ipts_buffer feedback[IPTS_BUFFERS];
+
+	struct ipts_buffer doorbell;
+	struct ipts_buffer workqueue;
+	struct ipts_buffer hid2me;
+
+	struct ipts_buffer descriptor;
+};
+
+int ipts_resources_init(struct ipts_resources *res, struct device *dev, size_t ds, size_t fs);
+int ipts_resources_free(struct ipts_resources *res);
+
+#endif /* IPTS_RESOURCES_H */
diff --git a/drivers/hid/ipts/spec-data.h b/drivers/hid/ipts/spec-data.h
new file mode 100644
index 0000000000000..e8dd98895a7ee
--- /dev/null
+++ b/drivers/hid/ipts/spec-data.h
@@ -0,0 +1,100 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_SPEC_DATA_H
+#define IPTS_SPEC_DATA_H
+
+#include <linux/build_bug.h>
+#include <linux/types.h>
+
+/**
+ * enum ipts_feedback_cmd_type - Commands that can be executed on the sensor through feedback.
+ */
+enum ipts_feedback_cmd_type {
+	IPTS_FEEDBACK_CMD_TYPE_NONE = 0,
+	IPTS_FEEDBACK_CMD_TYPE_SOFT_RESET = 1,
+	IPTS_FEEDBACK_CMD_TYPE_GOTO_ARMED = 2,
+	IPTS_FEEDBACK_CMD_TYPE_GOTO_SENSING = 3,
+	IPTS_FEEDBACK_CMD_TYPE_GOTO_SLEEP = 4,
+	IPTS_FEEDBACK_CMD_TYPE_GOTO_DOZE = 5,
+	IPTS_FEEDBACK_CMD_TYPE_HARD_RESET = 6,
+};
+
+/**
+ * enum ipts_feedback_data_type - Defines what data a feedback buffer contains.
+ * @IPTS_FEEDBACK_DATA_TYPE_VENDOR:        The buffer contains vendor specific feedback.
+ * @IPTS_FEEDBACK_DATA_TYPE_SET_FEATURES:  The buffer contains a HID set features report.
+ * @IPTS_FEEDBACK_DATA_TYPE_GET_FEATURES:  The buffer contains a HID get features report.
+ * @IPTS_FEEDBACK_DATA_TYPE_OUTPUT_REPORT: The buffer contains a HID output report.
+ * @IPTS_FEEDBACK_DATA_TYPE_STORE_DATA:    The buffer contains calibration data for the sensor.
+ */
+enum ipts_feedback_data_type {
+	IPTS_FEEDBACK_DATA_TYPE_VENDOR = 0,
+	IPTS_FEEDBACK_DATA_TYPE_SET_FEATURES = 1,
+	IPTS_FEEDBACK_DATA_TYPE_GET_FEATURES = 2,
+	IPTS_FEEDBACK_DATA_TYPE_OUTPUT_REPORT = 3,
+	IPTS_FEEDBACK_DATA_TYPE_STORE_DATA = 4,
+};
+
+/**
+ * struct ipts_feedback_header - Header that is prefixed to the data in a feedback buffer.
+ * @cmd_type:   A command that should be executed on the sensor.
+ * @size:       The size of the payload to be written.
+ * @buffer:     The ID of the buffer that contains this feedback data.
+ * @protocol:   The protocol version of the EDS.
+ * @data_type:  The type of data that the buffer contains.
+ * @spi_offset: The offset at which to write the payload data to the sensor.
+ * @payload:    Payload for the feedback command, or 0 if no payload is sent.
+ */
+struct ipts_feedback_header {
+	enum ipts_feedback_cmd_type cmd_type;
+	u32 size;
+	u32 buffer;
+	u32 protocol;
+	enum ipts_feedback_data_type data_type;
+	u32 spi_offset;
+	u8 reserved[40];
+	u8 payload[];
+} __packed;
+
+static_assert(sizeof(struct ipts_feedback_header) == 64);
+
+/**
+ * enum ipts_data_type - Defines what type of data a buffer contains.
+ * @IPTS_DATA_TYPE_FRAME:        Raw data frame.
+ * @IPTS_DATA_TYPE_ERROR:        Error data.
+ * @IPTS_DATA_TYPE_VENDOR:       Vendor specific data.
+ * @IPTS_DATA_TYPE_HID:          A HID report.
+ * @IPTS_DATA_TYPE_GET_FEATURES: The response to a GET_FEATURES HID2ME command.
+ */
+enum ipts_data_type {
+	IPTS_DATA_TYPE_FRAME = 0x00,
+	IPTS_DATA_TYPE_ERROR = 0x01,
+	IPTS_DATA_TYPE_VENDOR = 0x02,
+	IPTS_DATA_TYPE_HID = 0x03,
+	IPTS_DATA_TYPE_GET_FEATURES = 0x04,
+	IPTS_DATA_TYPE_DESCRIPTOR = 0x05,
+};
+
+/**
+ * struct ipts_data_header - Header that is prefixed to the data in a data buffer.
+ * @type: What data the buffer contains.
+ * @size: How much data the buffer contains.
+ * @buffer: Which buffer the data is in.
+ */
+struct ipts_data_header {
+	enum ipts_data_type type;
+	u32 size;
+	u32 buffer;
+	u8 reserved[52];
+	u8 data[];
+} __packed;
+
+static_assert(sizeof(struct ipts_data_header) == 64);
+
+#endif /* IPTS_SPEC_DATA_H */
diff --git a/drivers/hid/ipts/spec-device.h b/drivers/hid/ipts/spec-device.h
new file mode 100644
index 0000000000000..93f673d981f7f
--- /dev/null
+++ b/drivers/hid/ipts/spec-device.h
@@ -0,0 +1,285 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_SPEC_DEVICE_H
+#define IPTS_SPEC_DEVICE_H
+
+#include <linux/build_bug.h>
+#include <linux/types.h>
+
+/*
+ * The amount of buffers that IPTS can use for data transfer.
+ */
+#define IPTS_BUFFERS 16
+
+/*
+ * The buffer ID that is used for HID2ME feedback
+ */
+#define IPTS_HID2ME_BUFFER IPTS_BUFFERS
+
+/**
+ * enum ipts_command - Commands that can be sent to the IPTS hardware.
+ * @IPTS_CMD_GET_DEVICE_INFO:  Retrieves vendor information from the device.
+ * @IPTS_CMD_SET_MODE:         Changes the mode that the device will operate in.
+ * @IPTS_CMD_SET_MEM_WINDOW:   Configures memory buffers for passing data between device and driver.
+ * @IPTS_CMD_QUIESCE_IO:       Stops the data flow from the device to the driver.
+ * @IPTS_CMD_READY_FOR_DATA:   Informs the device that the driver is ready to receive data.
+ * @IPTS_CMD_FEEDBACK:         Informs the device that a buffer was processed and can be refilled.
+ * @IPTS_CMD_CLEAR_MEM_WINDOW: Stops the data flow and clears the buffer addresses on the device.
+ * @IPTS_CMD_RESET_SENSOR:     Resets the sensor to its default state.
+ * @IPTS_CMD_GET_DESCRIPTOR:   Retrieves the HID descriptor of the device.
+ */
+enum ipts_command_code {
+	IPTS_CMD_GET_DEVICE_INFO = 0x01,
+	IPTS_CMD_SET_MODE = 0x02,
+	IPTS_CMD_SET_MEM_WINDOW = 0x03,
+	IPTS_CMD_QUIESCE_IO = 0x04,
+	IPTS_CMD_READY_FOR_DATA = 0x05,
+	IPTS_CMD_FEEDBACK = 0x06,
+	IPTS_CMD_CLEAR_MEM_WINDOW = 0x07,
+	IPTS_CMD_RESET_SENSOR = 0x0B,
+	IPTS_CMD_GET_DESCRIPTOR = 0x0F,
+};
+
+/**
+ * enum ipts_status - Possible status codes returned by the IPTS device.
+ * @IPTS_STATUS_SUCCESS:                 Operation completed successfully.
+ * @IPTS_STATUS_INVALID_PARAMS:          Command contained an invalid payload.
+ * @IPTS_STATUS_ACCESS_DENIED:           ME could not validate a buffer address.
+ * @IPTS_STATUS_CMD_SIZE_ERROR:          Command contains an invalid payload.
+ * @IPTS_STATUS_NOT_READY:               Buffer addresses have not been set.
+ * @IPTS_STATUS_REQUEST_OUTSTANDING:     There is an outstanding command of the same type.
+ * @IPTS_STATUS_NO_SENSOR_FOUND:         No sensor could be found.
+ * @IPTS_STATUS_OUT_OF_MEMORY:           Not enough free memory for requested operation.
+ * @IPTS_STATUS_INTERNAL_ERROR:          An unexpected error occurred.
+ * @IPTS_STATUS_SENSOR_DISABLED:         The sensor has been disabled and must be reinitialized.
+ * @IPTS_STATUS_COMPAT_CHECK_FAIL:       Compatibility revision check between sensor and ME failed.
+ *                                       The host can ignore this error and attempt to continue.
+ * @IPTS_STATUS_SENSOR_EXPECTED_RESET:   The sensor went through a reset initiated by the driver.
+ * @IPTS_STATUS_SENSOR_UNEXPECTED_RESET: The sensor went through an unexpected reset.
+ * @IPTS_STATUS_RESET_FAILED:            Requested sensor reset failed to complete.
+ * @IPTS_STATUS_TIMEOUT:                 The operation timed out.
+ * @IPTS_STATUS_TEST_MODE_FAIL:          Test mode pattern did not match expected values.
+ * @IPTS_STATUS_SENSOR_FAIL_FATAL:       The sensor reported an error during reset sequence.
+ *                                       Further progress is not possible.
+ * @IPTS_STATUS_SENSOR_FAIL_NONFATAL:    The sensor reported an error during reset sequence.
+ *                                       The driver can attempt to continue.
+ * @IPTS_STATUS_INVALID_DEVICE_CAPS:     The device reported invalid capabilities.
+ * @IPTS_STATUS_QUIESCE_IO_IN_PROGRESS:  Command cannot be completed until Quiesce IO is done.
+ */
+enum ipts_status {
+	IPTS_STATUS_SUCCESS = 0x00,
+	IPTS_STATUS_INVALID_PARAMS = 0x01,
+	IPTS_STATUS_ACCESS_DENIED = 0x02,
+	IPTS_STATUS_CMD_SIZE_ERROR = 0x03,
+	IPTS_STATUS_NOT_READY = 0x04,
+	IPTS_STATUS_REQUEST_OUTSTANDING = 0x05,
+	IPTS_STATUS_NO_SENSOR_FOUND = 0x06,
+	IPTS_STATUS_OUT_OF_MEMORY = 0x07,
+	IPTS_STATUS_INTERNAL_ERROR = 0x08,
+	IPTS_STATUS_SENSOR_DISABLED = 0x09,
+	IPTS_STATUS_COMPAT_CHECK_FAIL = 0x0A,
+	IPTS_STATUS_SENSOR_EXPECTED_RESET = 0x0B,
+	IPTS_STATUS_SENSOR_UNEXPECTED_RESET = 0x0C,
+	IPTS_STATUS_RESET_FAILED = 0x0D,
+	IPTS_STATUS_TIMEOUT = 0x0E,
+	IPTS_STATUS_TEST_MODE_FAIL = 0x0F,
+	IPTS_STATUS_SENSOR_FAIL_FATAL = 0x10,
+	IPTS_STATUS_SENSOR_FAIL_NONFATAL = 0x11,
+	IPTS_STATUS_INVALID_DEVICE_CAPS = 0x12,
+	IPTS_STATUS_QUIESCE_IO_IN_PROGRESS = 0x13,
+};
+
+/**
+ * struct ipts_command - Message that is sent to the device for calling a command.
+ * @cmd:     The command that will be called.
+ * @payload: Payload containing parameters for the called command.
+ */
+struct ipts_command {
+	enum ipts_command_code cmd;
+	u8 payload[320];
+} __packed;
+
+static_assert(sizeof(struct ipts_command) == 324);
+
+/**
+ * enum ipts_mode - Configures what data the device produces and how its sent.
+ * @IPTS_MODE_EVENT:    The device will send an event once a buffer was filled.
+ *                      Older devices will return singletouch data in this mode.
+ * @IPTS_MODE_DOORBELL: The device will notify the driver by incrementing the doorbell value.
+ *                      Older devices will return multitouch data in this mode.
+ */
+enum ipts_mode {
+	IPTS_MODE_EVENT = 0x00,
+	IPTS_MODE_DOORBELL = 0x01,
+};
+
+/**
+ * struct ipts_set_mode - Payload for the SET_MODE command.
+ * @mode: Changes the mode that IPTS will operate in.
+ */
+struct ipts_set_mode {
+	enum ipts_mode mode;
+	u8 reserved[12];
+} __packed;
+
+static_assert(sizeof(struct ipts_set_mode) == 16);
+
+#define IPTS_WORKQUEUE_SIZE	 8192
+#define IPTS_WORKQUEUE_ITEM_SIZE 16
+
+/**
+ * struct ipts_mem_window - Payload for the SET_MEM_WINDOW command.
+ * @data_addr_lower:      Lower 32 bits of the data buffer addresses.
+ * @data_addr_upper:      Upper 32 bits of the data buffer addresses.
+ * @workqueue_addr_lower: Lower 32 bits of the workqueue buffer address.
+ * @workqueue_addr_upper: Upper 32 bits of the workqueue buffer address.
+ * @doorbell_addr_lower:  Lower 32 bits of the doorbell buffer address.
+ * @doorbell_addr_upper:  Upper 32 bits of the doorbell buffer address.
+ * @feedbackaddr_lower:   Lower 32 bits of the feedback buffer addresses.
+ * @feedbackaddr_upper:   Upper 32 bits of the feedback buffer addresses.
+ * @hid2me_addr_lower:    Lower 32 bits of the hid2me buffer address.
+ * @hid2me_addr_upper:    Upper 32 bits of the hid2me buffer address.
+ * @hid2me_size:          Size of the hid2me feedback buffer.
+ * @workqueue_item_size:  Magic value. Must be 16.
+ * @workqueue_size:       Magic value. Must be 8192.
+ *
+ * The workqueue related items in this struct are required for using
+ * GuC submission with binary processing firmware. Since this driver does
+ * not use GuC submission and instead exports raw data to userspace, these
+ * items are not actually used, but they need to be allocated and passed
+ * to the device, otherwise initialization will fail.
+ */
+struct ipts_mem_window {
+	u32 data_addr_lower[IPTS_BUFFERS];
+	u32 data_addr_upper[IPTS_BUFFERS];
+	u32 workqueue_addr_lower;
+	u32 workqueue_addr_upper;
+	u32 doorbell_addr_lower;
+	u32 doorbell_addr_upper;
+	u32 feedback_addr_lower[IPTS_BUFFERS];
+	u32 feedback_addr_upper[IPTS_BUFFERS];
+	u32 hid2me_addr_lower;
+	u32 hid2me_addr_upper;
+	u32 hid2me_size;
+	u8 reserved1;
+	u8 workqueue_item_size;
+	u16 workqueue_size;
+	u8 reserved[32];
+} __packed;
+
+static_assert(sizeof(struct ipts_mem_window) == 320);
+
+/**
+ * struct ipts_quiesce_io - Payload for the QUIESCE_IO command.
+ */
+struct ipts_quiesce_io {
+	u8 reserved[12];
+} __packed;
+
+static_assert(sizeof(struct ipts_quiesce_io) == 12);
+
+/**
+ * struct ipts_feedback - Payload for the FEEDBACK command.
+ * @buffer: The buffer that the device should refill.
+ */
+struct ipts_feedback {
+	u32 buffer;
+	u8 reserved[12];
+} __packed;
+
+static_assert(sizeof(struct ipts_feedback) == 16);
+
+/**
+ * enum ipts_reset_type - Possible ways of resetting the device.
+ * @IPTS_RESET_TYPE_HARD: Perform hardware reset using GPIO pin.
+ * @IPTS_RESET_TYPE_SOFT: Perform software reset using SPI command.
+ */
+enum ipts_reset_type {
+	IPTS_RESET_TYPE_HARD = 0x00,
+	IPTS_RESET_TYPE_SOFT = 0x01,
+};
+
+/**
+ * struct ipts_reset - Payload for the RESET_SENSOR command.
+ * @type: How the device should get reset.
+ */
+struct ipts_reset_sensor {
+	enum ipts_reset_type type;
+	u8 reserved[4];
+} __packed;
+
+static_assert(sizeof(struct ipts_reset_sensor) == 8);
+
+/**
+ * struct ipts_get_descriptor - Payload for the GET_DESCRIPTOR command.
+ * @addr_lower: The lower 32 bits of the descriptor buffer address.
+ * @addr_upper: The upper 32 bits of the descriptor buffer address.
+ * @magic:      A magic value. Must be 8.
+ */
+struct ipts_get_descriptor {
+	u32 addr_lower;
+	u32 addr_upper;
+	u32 magic;
+	u8 reserved[12];
+} __packed;
+
+static_assert(sizeof(struct ipts_get_descriptor) == 24);
+
+/*
+ * The type of a response is indicated by a
+ * command code, with the most significant bit flipped to 1.
+ */
+#define IPTS_RSP_BIT BIT(31)
+
+/**
+ * struct ipts_response - Data returned from the device in response to a command.
+ * @cmd:     The command that this response answers (IPTS_RSP_BIT will be 1).
+ * @status:  The return code of the command.
+ * @payload: The data that was produced by the command.
+ */
+struct ipts_response {
+	enum ipts_command_code cmd;
+	enum ipts_status status;
+	u8 payload[80];
+} __packed;
+
+static_assert(sizeof(struct ipts_response) == 88);
+
+/**
+ * struct ipts_device_info - Vendor information of the IPTS device.
+ * @vendor:        Vendor ID of this device.
+ * @product:       Product ID of this device.
+ * @hw_version:    Hardware revision of this device.
+ * @fw_version:    Firmware revision of this device.
+ * @data_size:     Requested size for a data buffer.
+ * @feedback_size: Requested size for a feedback buffer.
+ * @mode:          Mode that the device currently operates in.
+ * @max_contacts:  Maximum amount of concurrent touches the sensor can process.
+ */
+struct ipts_device_info {
+	u16 vendor;
+	u16 product;
+	u32 hw_version;
+	u32 fw_version;
+	u32 data_size;
+	u32 feedback_size;
+	enum ipts_mode mode;
+	u8 max_contacts;
+	u8 reserved1[3];
+	u8 sensor_min_eds;
+	u8 sensor_maj_eds;
+	u8 me_min_eds;
+	u8 me_maj_eds;
+	u8 intf_eds;
+	u8 reserved2[11];
+} __packed;
+
+static_assert(sizeof(struct ipts_device_info) == 44);
+
+#endif /* IPTS_SPEC_DEVICE_H */
diff --git a/drivers/hid/ipts/spec-hid.h b/drivers/hid/ipts/spec-hid.h
new file mode 100644
index 0000000000000..ea70f29ff00cb
--- /dev/null
+++ b/drivers/hid/ipts/spec-hid.h
@@ -0,0 +1,35 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2020-2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_SPEC_HID_H
+#define IPTS_SPEC_HID_H
+
+#include <linux/build_bug.h>
+#include <linux/types.h>
+
+/*
+ * Made-up type for passing raw IPTS data in a HID report.
+ */
+#define IPTS_HID_FRAME_TYPE_RAW 0xEE
+
+/**
+ * struct ipts_hid_frame - Header that is prefixed to raw IPTS data wrapped in a HID report.
+ * @size: Size of the data inside the report, including this header.
+ * @type: What type of data does this report contain.
+ */
+struct ipts_hid_header {
+	u32 size;
+	u8 reserved1;
+	u8 type;
+	u8 reserved2;
+	u8 data[];
+} __packed;
+
+static_assert(sizeof(struct ipts_hid_header) == 7);
+
+#endif /* IPTS_SPEC_HID_H */
diff --git a/drivers/hid/ipts/thread.c b/drivers/hid/ipts/thread.c
new file mode 100644
index 0000000000000..8b46f775c1070
--- /dev/null
+++ b/drivers/hid/ipts/thread.c
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#include <linux/completion.h>
+#include <linux/err.h>
+#include <linux/kthread.h>
+#include <linux/mutex.h>
+
+#include "thread.h"
+
+bool ipts_thread_should_stop(struct ipts_thread *thread)
+{
+	if (!thread)
+		return false;
+
+	return READ_ONCE(thread->should_stop);
+}
+
+static int ipts_thread_runner(void *data)
+{
+	int ret = 0;
+	struct ipts_thread *thread = data;
+
+	if (!thread)
+		return -EFAULT;
+
+	if (!thread->threadfn)
+		return -EFAULT;
+
+	ret = thread->threadfn(thread);
+	complete_all(&thread->done);
+
+	return ret;
+}
+
+int ipts_thread_start(struct ipts_thread *thread, int (*threadfn)(struct ipts_thread *thread),
+		      void *data, const char *name)
+{
+	if (!thread)
+		return -EFAULT;
+
+	if (!threadfn)
+		return -EFAULT;
+
+	init_completion(&thread->done);
+
+	thread->data = data;
+	thread->should_stop = false;
+	thread->threadfn = threadfn;
+
+	thread->thread = kthread_run(ipts_thread_runner, thread, name);
+	return PTR_ERR_OR_ZERO(thread->thread);
+}
+
+int ipts_thread_stop(struct ipts_thread *thread)
+{
+	int ret = 0;
+
+	if (!thread)
+		return -EFAULT;
+
+	if (!thread->thread)
+		return 0;
+
+	WRITE_ONCE(thread->should_stop, true);
+
+	/*
+	 * Make sure that the write has gone through before waiting.
+	 */
+	wmb();
+
+	wait_for_completion(&thread->done);
+	ret = kthread_stop(thread->thread);
+
+	thread->thread = NULL;
+	thread->data = NULL;
+	thread->threadfn = NULL;
+
+	return ret;
+}
diff --git a/drivers/hid/ipts/thread.h b/drivers/hid/ipts/thread.h
new file mode 100644
index 0000000000000..a314843599fc3
--- /dev/null
+++ b/drivers/hid/ipts/thread.h
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (c) 2016 Intel Corporation
+ * Copyright (c) 2023 Dorian Stoll
+ *
+ * Linux driver for Intel Precise Touch & Stylus
+ */
+
+#ifndef IPTS_THREAD_H
+#define IPTS_THREAD_H
+
+#include <linux/completion.h>
+#include <linux/mutex.h>
+#include <linux/sched.h>
+
+/*
+ * This wrapper over kthread is necessary, because calling kthread_stop makes it impossible
+ * to issue MEI commands from that thread while it shuts itself down. By using a custom
+ * boolean variable and a completion object, we can call kthread_stop only when the thread
+ * already finished all of its work and has returned.
+ */
+struct ipts_thread {
+	struct task_struct *thread;
+
+	bool should_stop;
+	struct completion done;
+
+	void *data;
+	int (*threadfn)(struct ipts_thread *thread);
+};
+
+/*
+ * ipts_thread_should_stop() - Returns true if the thread is asked to terminate.
+ * @thread: The current thread.
+ *
+ * Returns: true if the thread should stop, false if not.
+ */
+bool ipts_thread_should_stop(struct ipts_thread *thread);
+
+/*
+ * ipts_thread_start() - Starts an IPTS thread.
+ * @thread: The thread to initialize and start.
+ * @threadfn: The function to execute.
+ * @data: An argument that will be passed to threadfn.
+ * @name: The name of the new thread.
+ *
+ * Returns: 0 on success, <0 on error.
+ */
+int ipts_thread_start(struct ipts_thread *thread, int (*threadfn)(struct ipts_thread *thread),
+		      void *data, const char name[]);
+
+/*
+ * ipts_thread_stop() - Asks the thread to terminate and waits until it has finished.
+ * @thread: The thread that should stop.
+ *
+ * Returns: The return value of the thread function.
+ */
+int ipts_thread_stop(struct ipts_thread *thread);
+
+#endif /* IPTS_THREAD_H */
-- 
2.39.2

From 0dbbe2f1542b0926f6524910ca180166ab9bb778 Mon Sep 17 00:00:00 2001
From: Dorian Stoll <dorian.stoll@tmsp.io>
Date: Sun, 11 Dec 2022 12:03:38 +0100
Subject: [PATCH] iommu: intel: Disable source id verification for ITHC

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>
Patchset: ithc
---
 drivers/iommu/intel/irq_remapping.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/drivers/iommu/intel/irq_remapping.c b/drivers/iommu/intel/irq_remapping.c
index 5962bb5027d06..86fa129d10d40 100644
--- a/drivers/iommu/intel/irq_remapping.c
+++ b/drivers/iommu/intel/irq_remapping.c
@@ -396,6 +396,22 @@ static int set_msi_sid(struct irte *irte, struct pci_dev *dev)
 	data.busmatch_count = 0;
 	pci_for_each_dma_alias(dev, set_msi_sid_cb, &data);
 
+	/*
+	 * The Intel Touch Host Controller is at 00:10.6, but for some reason
+	 * the MSI interrupts have request id 01:05.0.
+	 * Disable id verification to work around this.
+	 * FIXME Find proper fix or turn this into a quirk.
+	 */
+	if (dev->vendor == PCI_VENDOR_ID_INTEL && (dev->class >> 8) == PCI_CLASS_INPUT_PEN) {
+		switch(dev->device) {
+		case 0x98d0: case 0x98d1: // LKF
+		case 0xa0d0: case 0xa0d1: // TGL LP
+		case 0x43d0: case 0x43d1: // TGL H
+			set_irte_sid(irte, SVT_NO_VERIFY, SQ_ALL_16, 0);
+			return 0;
+		}
+	}
+
 	/*
 	 * DMA alias provides us with a PCI device and alias.  The only case
 	 * where the it will return an alias on a different bus than the
-- 
2.39.2

From 5bde25a4abb63731d4a4d42a84725d0043ef0215 Mon Sep 17 00:00:00 2001
From: Dorian Stoll <dorian.stoll@tmsp.io>
Date: Sun, 11 Dec 2022 12:10:54 +0100
Subject: [PATCH] hid: Add support for Intel Touch Host Controller

Based on quo/ithc-linux@55803a2

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>
Patchset: ithc
---
 drivers/hid/Kconfig           |   2 +
 drivers/hid/Makefile          |   1 +
 drivers/hid/ithc/Kbuild       |   6 +
 drivers/hid/ithc/Kconfig      |  12 +
 drivers/hid/ithc/ithc-debug.c |  96 ++++++
 drivers/hid/ithc/ithc-dma.c   | 258 ++++++++++++++++
 drivers/hid/ithc/ithc-dma.h   |  67 +++++
 drivers/hid/ithc/ithc-main.c  | 534 ++++++++++++++++++++++++++++++++++
 drivers/hid/ithc/ithc-regs.c  |  64 ++++
 drivers/hid/ithc/ithc-regs.h  | 186 ++++++++++++
 drivers/hid/ithc/ithc.h       |  60 ++++
 11 files changed, 1286 insertions(+)
 create mode 100644 drivers/hid/ithc/Kbuild
 create mode 100644 drivers/hid/ithc/Kconfig
 create mode 100644 drivers/hid/ithc/ithc-debug.c
 create mode 100644 drivers/hid/ithc/ithc-dma.c
 create mode 100644 drivers/hid/ithc/ithc-dma.h
 create mode 100644 drivers/hid/ithc/ithc-main.c
 create mode 100644 drivers/hid/ithc/ithc-regs.c
 create mode 100644 drivers/hid/ithc/ithc-regs.h
 create mode 100644 drivers/hid/ithc/ithc.h

diff --git a/drivers/hid/Kconfig b/drivers/hid/Kconfig
index 1523ccdf73b51..b7030071ac0c0 100644
--- a/drivers/hid/Kconfig
+++ b/drivers/hid/Kconfig
@@ -1292,4 +1292,6 @@ source "drivers/hid/surface-hid/Kconfig"
 
 source "drivers/hid/ipts/Kconfig"
 
+source "drivers/hid/ithc/Kconfig"
+
 endmenu
diff --git a/drivers/hid/Makefile b/drivers/hid/Makefile
index e48300bcea9be..8ef7308b0f9d7 100644
--- a/drivers/hid/Makefile
+++ b/drivers/hid/Makefile
@@ -166,3 +166,4 @@ obj-$(CONFIG_AMD_SFH_HID)       += amd-sfh-hid/
 obj-$(CONFIG_SURFACE_HID_CORE)  += surface-hid/
 
 obj-$(CONFIG_HID_IPTS)          += ipts/
+obj-$(CONFIG_HID_ITHC)          += ithc/
diff --git a/drivers/hid/ithc/Kbuild b/drivers/hid/ithc/Kbuild
new file mode 100644
index 0000000000000..aea83f2ac07b4
--- /dev/null
+++ b/drivers/hid/ithc/Kbuild
@@ -0,0 +1,6 @@
+obj-$(CONFIG_HID_ITHC) := ithc.o
+
+ithc-objs := ithc-main.o ithc-regs.o ithc-dma.o ithc-debug.o
+
+ccflags-y := -std=gnu11 -Wno-declaration-after-statement
+
diff --git a/drivers/hid/ithc/Kconfig b/drivers/hid/ithc/Kconfig
new file mode 100644
index 0000000000000..ede7130236096
--- /dev/null
+++ b/drivers/hid/ithc/Kconfig
@@ -0,0 +1,12 @@
+config HID_ITHC
+	tristate "Intel Touch Host Controller"
+	depends on PCI
+	depends on HID
+	help
+	  Say Y here if your system has a touchscreen using Intels
+	  Touch Host Controller (ITHC / IPTS) technology.
+
+	  If unsure say N.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called ithc.
diff --git a/drivers/hid/ithc/ithc-debug.c b/drivers/hid/ithc/ithc-debug.c
new file mode 100644
index 0000000000000..57bf125c45bd5
--- /dev/null
+++ b/drivers/hid/ithc/ithc-debug.c
@@ -0,0 +1,96 @@
+#include "ithc.h"
+
+void ithc_log_regs(struct ithc *ithc) {
+	if (!ithc->prev_regs) return;
+	u32 __iomem *cur = (__iomem void*)ithc->regs;
+	u32 *prev = (void*)ithc->prev_regs;
+	for (int i = 1024; i < sizeof *ithc->regs / 4; i++) {
+		u32 x = readl(cur + i);
+		if (x != prev[i]) {
+			pci_info(ithc->pci, "reg %04x: %08x -> %08x\n", i * 4, prev[i], x);
+			prev[i] = x;
+		}
+	}
+}
+
+static ssize_t ithc_debugfs_cmd_write(struct file *f, const char __user *buf, size_t len, loff_t *offset) {
+	struct ithc *ithc = file_inode(f)->i_private;
+	char cmd[256];
+	if (!ithc || !ithc->pci) return -ENODEV;
+	if (!len) return -EINVAL;
+	if (len >= sizeof cmd) return -EINVAL;
+	if (copy_from_user(cmd, buf, len)) return -EFAULT;
+	cmd[len] = 0;
+	if (cmd[len-1] == '\n') cmd[len-1] = 0;
+	pci_info(ithc->pci, "debug command: %s\n", cmd);
+	u32 n = 0;
+	const char *s = cmd + 1;
+	u32 a[32];
+	while (*s && *s != '\n') {
+		if (n >= ARRAY_SIZE(a)) return -EINVAL;
+		if (*s++ != ' ') return -EINVAL;
+		char *e;
+		a[n++] = simple_strtoul(s, &e, 0);
+		if (e == s) return -EINVAL;
+		s = e;
+	}
+	ithc_log_regs(ithc);
+	switch(cmd[0]) {
+	case 'x': // reset
+		ithc_reset(ithc);
+		break;
+	case 'w': // write register: offset mask value
+		if (n != 3 || (a[0] & 3)) return -EINVAL;
+		pci_info(ithc->pci, "debug write 0x%04x = 0x%08x (mask 0x%08x)\n", a[0], a[2], a[1]);
+		bitsl(((__iomem u32 *)ithc->regs) + a[0] / 4, a[1], a[2]);
+		break;
+	case 'r': // read register: offset
+		if (n != 1 || (a[0] & 3)) return -EINVAL;
+		pci_info(ithc->pci, "debug read 0x%04x = 0x%08x\n", a[0], readl(((__iomem u32 *)ithc->regs) + a[0] / 4));
+		break;
+	case 's': // spi command: cmd offset len data...
+		// read config: s 4 0 64 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
+		// set touch cfg: s 6 12 4 XX
+		if (n < 3 || a[2] > (n - 3) * 4) return -EINVAL;
+		pci_info(ithc->pci, "debug spi command %u with %u bytes of data\n", a[0], a[2]);
+		if (!CHECK(ithc_spi_command, ithc, a[0], a[1], a[2], a + 3))
+			for (u32 i = 0; i < (a[2] + 3) / 4; i++) pci_info(ithc->pci, "resp %u = 0x%08x\n", i, a[3+i]);
+		break;
+	case 'd': // dma command: cmd len data...
+		// get report descriptor: d 7 8 0 0
+		// enable multitouch: d 3 2 0x0105
+		if (n < 2 || a[1] > (n - 2) * 4) return -EINVAL;
+		pci_info(ithc->pci, "debug dma command %u with %u bytes of data\n", a[0], a[1]);
+		if (ithc_dma_tx(ithc, a[0], a[1], a + 2)) pci_err(ithc->pci, "dma tx failed\n");
+		break;
+	default:
+		return -EINVAL;
+	}
+	ithc_log_regs(ithc);
+	return len;
+}
+
+static const struct file_operations ithc_debugfops_cmd = {
+	.owner = THIS_MODULE,
+	.write = ithc_debugfs_cmd_write,
+};
+
+static void ithc_debugfs_devres_release(struct device *dev, void *res) {
+	struct dentry **dbgm = res;
+	if (*dbgm) debugfs_remove_recursive(*dbgm);
+}
+
+int ithc_debug_init(struct ithc *ithc) {
+	struct dentry **dbgm = devres_alloc(ithc_debugfs_devres_release, sizeof *dbgm, GFP_KERNEL);
+	if (!dbgm) return -ENOMEM;
+	devres_add(&ithc->pci->dev, dbgm);
+	struct dentry *dbg = debugfs_create_dir(DEVNAME, NULL);
+	if (IS_ERR(dbg)) return PTR_ERR(dbg);
+	*dbgm = dbg;
+
+	struct dentry *cmd = debugfs_create_file("cmd", 0220, dbg, ithc, &ithc_debugfops_cmd);
+	if (IS_ERR(cmd)) return PTR_ERR(cmd);
+
+	return 0;
+}
+
diff --git a/drivers/hid/ithc/ithc-dma.c b/drivers/hid/ithc/ithc-dma.c
new file mode 100644
index 0000000000000..7e89b3496918d
--- /dev/null
+++ b/drivers/hid/ithc/ithc-dma.c
@@ -0,0 +1,258 @@
+#include "ithc.h"
+
+static int ithc_dma_prd_alloc(struct ithc *ithc, struct ithc_dma_prd_buffer *p, unsigned num_buffers, unsigned num_pages, enum dma_data_direction dir) {
+	p->num_pages = num_pages;
+	p->dir = dir;
+	p->size = round_up(num_buffers * num_pages * sizeof(struct ithc_phys_region_desc), PAGE_SIZE);
+	p->addr = dmam_alloc_coherent(&ithc->pci->dev, p->size, &p->dma_addr, GFP_KERNEL);
+	if (!p->addr) return -ENOMEM;
+	if (p->dma_addr & (PAGE_SIZE - 1)) return -EFAULT;
+	return 0;
+}
+
+struct ithc_sg_table {
+	void *addr;
+	struct sg_table sgt;
+	enum dma_data_direction dir;
+};
+static void ithc_dma_sgtable_free(struct sg_table *sgt) {
+	struct scatterlist *sg;
+	int i;
+	for_each_sgtable_sg(sgt, sg, i) {
+		struct page *p = sg_page(sg);
+		if (p) __free_page(p);
+	}
+	sg_free_table(sgt);
+}
+static void ithc_dma_data_devres_release(struct device *dev, void *res) {
+	struct ithc_sg_table *sgt = res;
+	if (sgt->addr) vunmap(sgt->addr);
+	dma_unmap_sgtable(dev, &sgt->sgt, sgt->dir, 0);
+	ithc_dma_sgtable_free(&sgt->sgt);
+}
+
+static int ithc_dma_data_alloc(struct ithc* ithc, struct ithc_dma_prd_buffer *prds, struct ithc_dma_data_buffer *b) {
+	// We don't use dma_alloc_coherent for data buffers, because they don't have to be contiguous (we can use one PRD per page) or coherent (they are unidirectional).
+	// Instead we use an sg_table of individually allocated pages (5.13 has dma_alloc_noncontiguous for this, but we'd like to support 5.10 for now).
+	struct page *pages[16];
+	if (prds->num_pages == 0 || prds->num_pages > ARRAY_SIZE(pages)) return -EINVAL;
+	b->active_idx = -1;
+	struct ithc_sg_table *sgt = devres_alloc(ithc_dma_data_devres_release, sizeof *sgt, GFP_KERNEL);
+	if (!sgt) return -ENOMEM;
+	sgt->dir = prds->dir;
+	if (!sg_alloc_table(&sgt->sgt, prds->num_pages, GFP_KERNEL)) {
+		struct scatterlist *sg;
+		int i;
+		bool ok = true;
+		for_each_sgtable_sg(&sgt->sgt, sg, i) {
+			struct page *p = pages[i] = alloc_page(GFP_KERNEL | __GFP_ZERO); // don't need __GFP_DMA for PCI DMA
+			if (!p) { ok = false; break; }
+			sg_set_page(sg, p, PAGE_SIZE, 0);
+		}
+		if (ok && !dma_map_sgtable(&ithc->pci->dev, &sgt->sgt, prds->dir, 0)) {
+			devres_add(&ithc->pci->dev, sgt);
+			b->sgt = &sgt->sgt;
+			b->addr = sgt->addr = vmap(pages, prds->num_pages, 0, PAGE_KERNEL);
+			if (!b->addr) return -ENOMEM;
+			return 0;
+		}
+		ithc_dma_sgtable_free(&sgt->sgt);
+	}
+	devres_free(sgt);
+	return -ENOMEM;
+}
+
+static int ithc_dma_data_buffer_put(struct ithc *ithc, struct ithc_dma_prd_buffer *prds, struct ithc_dma_data_buffer *b, unsigned idx) {
+	struct ithc_phys_region_desc *prd = prds->addr;
+	prd += idx * prds->num_pages;
+	if (b->active_idx >= 0) { pci_err(ithc->pci, "buffer already active\n"); return -EINVAL; }
+	b->active_idx = idx;
+	if (prds->dir == DMA_TO_DEVICE) {
+		if (b->data_size > PAGE_SIZE) return -EINVAL;
+		prd->addr = sg_dma_address(b->sgt->sgl) >> 10;
+		prd->size = b->data_size | PRD_FLAG_END;
+		flush_kernel_vmap_range(b->addr, b->data_size);
+	} else if (prds->dir == DMA_FROM_DEVICE) {
+		struct scatterlist *sg;
+		int i;
+		for_each_sgtable_dma_sg(b->sgt, sg, i) {
+			prd->addr = sg_dma_address(sg) >> 10;
+			prd->size = sg_dma_len(sg);
+			prd++;
+		}
+		prd[-1].size |= PRD_FLAG_END;
+	}
+	dma_wmb(); // for the prds
+	dma_sync_sgtable_for_device(&ithc->pci->dev, b->sgt, prds->dir);
+	return 0;
+}
+
+static int ithc_dma_data_buffer_get(struct ithc *ithc, struct ithc_dma_prd_buffer *prds, struct ithc_dma_data_buffer *b, unsigned idx) {
+	struct ithc_phys_region_desc *prd = prds->addr;
+	prd += idx * prds->num_pages;
+	if (b->active_idx != idx) { pci_err(ithc->pci, "wrong buffer index\n"); return -EINVAL; }
+	b->active_idx = -1;
+	if (prds->dir == DMA_FROM_DEVICE) {
+		dma_rmb(); // for the prds
+		b->data_size = 0;
+		struct scatterlist *sg;
+		int i;
+		for_each_sgtable_dma_sg(b->sgt, sg, i) {
+			unsigned size = prd->size;
+			b->data_size += size & PRD_SIZE_MASK;
+			if (size & PRD_FLAG_END) break;
+			if ((size & PRD_SIZE_MASK) != sg_dma_len(sg)) { pci_err(ithc->pci, "truncated prd\n"); break; }
+			prd++;
+		}
+		invalidate_kernel_vmap_range(b->addr, b->data_size);
+	}
+	dma_sync_sgtable_for_cpu(&ithc->pci->dev, b->sgt, prds->dir);
+	return 0;
+}
+
+int ithc_dma_rx_init(struct ithc *ithc, u8 channel, const char *devname) {
+	struct ithc_dma_rx *rx = &ithc->dma_rx[channel];
+	mutex_init(&rx->mutex);
+	u32 buf_size = DEVCFG_DMA_RX_SIZE(ithc->config.dma_buf_sizes);
+	unsigned num_pages = (buf_size + PAGE_SIZE - 1) / PAGE_SIZE;
+	pci_dbg(ithc->pci, "allocating rx buffers: num = %u, size = %u, pages = %u\n", NUM_RX_BUF, buf_size, num_pages);
+	CHECK_RET(ithc_dma_prd_alloc, ithc, &rx->prds, NUM_RX_BUF, num_pages, DMA_FROM_DEVICE);
+	for (unsigned i = 0; i < NUM_RX_BUF; i++)
+		CHECK_RET(ithc_dma_data_alloc, ithc, &rx->prds, &rx->bufs[i]);
+	writeb(DMA_RX_CONTROL2_RESET, &ithc->regs->dma_rx[channel].control2);
+	lo_hi_writeq(rx->prds.dma_addr, &ithc->regs->dma_rx[channel].addr);
+	writeb(NUM_RX_BUF - 1, &ithc->regs->dma_rx[channel].num_bufs);
+	writeb(num_pages - 1, &ithc->regs->dma_rx[channel].num_prds);
+	u8 head = readb(&ithc->regs->dma_rx[channel].head);
+	if (head) { pci_err(ithc->pci, "head is nonzero (%u)\n", head); return -EIO; }
+	for (unsigned i = 0; i < NUM_RX_BUF; i++)
+		CHECK_RET(ithc_dma_data_buffer_put, ithc, &rx->prds, &rx->bufs[i], i);
+	writeb(head ^ DMA_RX_WRAP_FLAG, &ithc->regs->dma_rx[channel].tail);
+	return 0;
+}
+void ithc_dma_rx_enable(struct ithc *ithc, u8 channel) {
+	bitsb_set(&ithc->regs->dma_rx[channel].control, DMA_RX_CONTROL_ENABLE | DMA_RX_CONTROL_IRQ_ERROR | DMA_RX_CONTROL_IRQ_DATA);
+	CHECK(waitl, ithc, &ithc->regs->dma_rx[1].status, DMA_RX_STATUS_ENABLED, DMA_RX_STATUS_ENABLED);
+}
+
+int ithc_dma_tx_init(struct ithc *ithc) {
+	struct ithc_dma_tx *tx = &ithc->dma_tx;
+	mutex_init(&tx->mutex);
+	tx->max_size = DEVCFG_DMA_TX_SIZE(ithc->config.dma_buf_sizes);
+	unsigned num_pages = (tx->max_size + PAGE_SIZE - 1) / PAGE_SIZE;
+	pci_dbg(ithc->pci, "allocating tx buffers: size = %u, pages = %u\n", tx->max_size, num_pages);
+	CHECK_RET(ithc_dma_prd_alloc, ithc, &tx->prds, 1, num_pages, DMA_TO_DEVICE);
+	CHECK_RET(ithc_dma_data_alloc, ithc, &tx->prds, &tx->buf);
+	lo_hi_writeq(tx->prds.dma_addr, &ithc->regs->dma_tx.addr);
+	writeb(num_pages - 1, &ithc->regs->dma_tx.num_prds);
+	CHECK_RET(ithc_dma_data_buffer_put, ithc, &ithc->dma_tx.prds, &ithc->dma_tx.buf, 0);
+	return 0;
+}
+
+static int ithc_dma_rx_process_buf(struct ithc *ithc, struct ithc_dma_data_buffer *data, u8 channel, u8 buf) {
+	if (buf >= NUM_RX_BUF) {
+		pci_err(ithc->pci, "invalid dma ringbuffer index\n");
+		return -EINVAL;
+	}
+	ithc_set_active(ithc);
+	u32 len = data->data_size;
+	struct ithc_dma_rx_header *hdr = data->addr;
+	u8 *hiddata = (void *)(hdr + 1);
+	if (len >= sizeof *hdr && hdr->code == DMA_RX_CODE_RESET) {
+		CHECK(ithc_reset, ithc);
+	} else if (len < sizeof *hdr || len != sizeof *hdr + hdr->data_size) {
+		if (hdr->code == DMA_RX_CODE_INPUT_REPORT) {
+			// When the CPU enters a low power state during DMA, we can get truncated messages.
+			// Typically this will be a single touch HID report that is only 1 byte, or a multitouch report that is 257 bytes.
+			// See also ithc_set_active().
+		} else {
+			pci_err(ithc->pci, "invalid dma rx data! channel %u, buffer %u, size %u, code %u, data size %u\n", channel, buf, len, hdr->code, hdr->data_size);
+			print_hex_dump_debug(DEVNAME " data: ", DUMP_PREFIX_OFFSET, 32, 1, hdr, min(len, 0x400u), 0);
+		}
+	} else if (hdr->code == DMA_RX_CODE_REPORT_DESCRIPTOR && hdr->data_size > 8) {
+		CHECK(hid_parse_report, ithc->hid, hiddata + 8, hdr->data_size - 8);
+		WRITE_ONCE(ithc->hid_parse_done, true);
+		wake_up(&ithc->wait_hid_parse);
+	} else if (hdr->code == DMA_RX_CODE_INPUT_REPORT) {
+		CHECK(hid_input_report, ithc->hid, HID_INPUT_REPORT, hiddata, hdr->data_size, 1);
+	} else if (hdr->code == DMA_RX_CODE_FEATURE_REPORT) {
+		bool done = false;
+		mutex_lock(&ithc->hid_get_feature_mutex);
+		if (ithc->hid_get_feature_buf) {
+			if (hdr->data_size < ithc->hid_get_feature_size) ithc->hid_get_feature_size = hdr->data_size;
+			memcpy(ithc->hid_get_feature_buf, hiddata, ithc->hid_get_feature_size);
+			ithc->hid_get_feature_buf = NULL;
+			done = true;
+		}
+		mutex_unlock(&ithc->hid_get_feature_mutex);
+		if (done) wake_up(&ithc->wait_hid_get_feature);
+		else CHECK(hid_input_report, ithc->hid, HID_FEATURE_REPORT, hiddata, hdr->data_size, 1);
+	} else {
+		pci_dbg(ithc->pci, "unhandled dma rx data! channel %u, buffer %u, size %u, code %u\n", channel, buf, len, hdr->code);
+		print_hex_dump_debug(DEVNAME " data: ", DUMP_PREFIX_OFFSET, 32, 1, hdr, min(len, 0x400u), 0);
+	}
+	return 0;
+}
+
+static int ithc_dma_rx_unlocked(struct ithc *ithc, u8 channel) {
+	struct ithc_dma_rx *rx = &ithc->dma_rx[channel];
+	unsigned n = rx->num_received;
+	u8 head_wrap = readb(&ithc->regs->dma_rx[channel].head);
+	while (1) {
+		u8 tail = n % NUM_RX_BUF;
+		u8 tail_wrap = tail | ((n / NUM_RX_BUF) & 1 ? 0 : DMA_RX_WRAP_FLAG);
+		writeb(tail_wrap, &ithc->regs->dma_rx[channel].tail);
+		// ringbuffer is full if tail_wrap == head_wrap
+		// ringbuffer is empty if tail_wrap == head_wrap ^ WRAP_FLAG
+		if (tail_wrap == (head_wrap ^ DMA_RX_WRAP_FLAG)) return 0;
+
+		// take the buffer that the device just filled
+		struct ithc_dma_data_buffer *b = &rx->bufs[n % NUM_RX_BUF];
+		CHECK_RET(ithc_dma_data_buffer_get, ithc, &rx->prds, b, tail);
+		rx->num_received = ++n;
+
+		// process data
+		CHECK(ithc_dma_rx_process_buf, ithc, b, channel, tail);
+
+		// give the buffer back to the device
+		CHECK_RET(ithc_dma_data_buffer_put, ithc, &rx->prds, b, tail);
+	}
+}
+int ithc_dma_rx(struct ithc *ithc, u8 channel) {
+	struct ithc_dma_rx *rx = &ithc->dma_rx[channel];
+	mutex_lock(&rx->mutex);
+	int ret = ithc_dma_rx_unlocked(ithc, channel);
+	mutex_unlock(&rx->mutex);
+	return ret;
+}
+
+static int ithc_dma_tx_unlocked(struct ithc *ithc, u32 cmdcode, u32 datasize, void *data) {
+	pci_dbg(ithc->pci, "dma tx command %u, size %u\n", cmdcode, datasize);
+	struct ithc_dma_tx_header *hdr;
+	u8 padding = datasize & 3 ? 4 - (datasize & 3) : 0;
+	unsigned fullsize = sizeof *hdr + datasize + padding;
+	if (fullsize > ithc->dma_tx.max_size || fullsize > PAGE_SIZE) return -EINVAL;
+	CHECK_RET(ithc_dma_data_buffer_get, ithc, &ithc->dma_tx.prds, &ithc->dma_tx.buf, 0);
+
+	ithc->dma_tx.buf.data_size = fullsize;
+	hdr = ithc->dma_tx.buf.addr;
+	hdr->code = cmdcode;
+	hdr->data_size = datasize;
+	u8 *dest = (void *)(hdr + 1);
+	memcpy(dest, data, datasize);
+	dest += datasize;
+	for (u8 p = 0; p < padding; p++) *dest++ = 0;
+	CHECK_RET(ithc_dma_data_buffer_put, ithc, &ithc->dma_tx.prds, &ithc->dma_tx.buf, 0);
+
+	bitsb_set(&ithc->regs->dma_tx.control, DMA_TX_CONTROL_SEND);
+	CHECK_RET(waitb, ithc, &ithc->regs->dma_tx.control, DMA_TX_CONTROL_SEND, 0);
+	writel(DMA_TX_STATUS_DONE, &ithc->regs->dma_tx.status);
+	return 0;
+}
+int ithc_dma_tx(struct ithc *ithc, u32 cmdcode, u32 datasize, void *data) {
+	mutex_lock(&ithc->dma_tx.mutex);
+	int ret = ithc_dma_tx_unlocked(ithc, cmdcode, datasize, data);
+	mutex_unlock(&ithc->dma_tx.mutex);
+	return ret;
+}
+
diff --git a/drivers/hid/ithc/ithc-dma.h b/drivers/hid/ithc/ithc-dma.h
new file mode 100644
index 0000000000000..d9f2c19a13f3a
--- /dev/null
+++ b/drivers/hid/ithc/ithc-dma.h
@@ -0,0 +1,67 @@
+#define PRD_SIZE_MASK            0xffffff
+#define PRD_FLAG_END             0x1000000
+#define PRD_FLAG_SUCCESS         0x2000000
+#define PRD_FLAG_ERROR           0x4000000
+
+struct ithc_phys_region_desc {
+	u64 addr; // physical addr/1024
+	u32 size; // num bytes, PRD_FLAG_END marks last prd for data split over multiple prds
+	u32 unused;
+};
+
+#define DMA_RX_CODE_INPUT_REPORT          3
+#define DMA_RX_CODE_FEATURE_REPORT        4
+#define DMA_RX_CODE_REPORT_DESCRIPTOR     5
+#define DMA_RX_CODE_RESET                 7
+
+struct ithc_dma_rx_header {
+	u32 code;
+	u32 data_size;
+	u32 _unknown[14];
+};
+
+#define DMA_TX_CODE_SET_FEATURE           3
+#define DMA_TX_CODE_GET_FEATURE           4
+#define DMA_TX_CODE_OUTPUT_REPORT         5
+#define DMA_TX_CODE_GET_REPORT_DESCRIPTOR 7
+
+struct ithc_dma_tx_header {
+	u32 code;
+	u32 data_size;
+};
+
+struct ithc_dma_prd_buffer {
+	void *addr;
+	dma_addr_t dma_addr;
+	u32 size;
+	u32 num_pages; // per data buffer
+	enum dma_data_direction dir;
+};
+
+struct ithc_dma_data_buffer {
+	void *addr;
+	struct sg_table *sgt;
+	int active_idx;
+	u32 data_size;
+};
+
+struct ithc_dma_tx {
+	struct mutex mutex;
+	u32 max_size;
+	struct ithc_dma_prd_buffer prds;
+	struct ithc_dma_data_buffer buf;
+};
+
+struct ithc_dma_rx {
+	struct mutex mutex;
+	u32 num_received;
+	struct ithc_dma_prd_buffer prds;
+	struct ithc_dma_data_buffer bufs[NUM_RX_BUF];
+};
+
+int ithc_dma_rx_init(struct ithc *ithc, u8 channel, const char *devname);
+void ithc_dma_rx_enable(struct ithc *ithc, u8 channel);
+int ithc_dma_tx_init(struct ithc *ithc);
+int ithc_dma_rx(struct ithc *ithc, u8 channel);
+int ithc_dma_tx(struct ithc *ithc, u32 cmdcode, u32 datasize, void *cmddata);
+
diff --git a/drivers/hid/ithc/ithc-main.c b/drivers/hid/ithc/ithc-main.c
new file mode 100644
index 0000000000000..09512b9cb4d31
--- /dev/null
+++ b/drivers/hid/ithc/ithc-main.c
@@ -0,0 +1,534 @@
+#include "ithc.h"
+
+MODULE_DESCRIPTION("Intel Touch Host Controller driver");
+MODULE_LICENSE("Dual BSD/GPL");
+
+// Lakefield
+#define PCI_DEVICE_ID_INTEL_THC_LKF_PORT1    0x98d0
+#define PCI_DEVICE_ID_INTEL_THC_LKF_PORT2    0x98d1
+// Tiger Lake
+#define PCI_DEVICE_ID_INTEL_THC_TGL_LP_PORT1 0xa0d0
+#define PCI_DEVICE_ID_INTEL_THC_TGL_LP_PORT2 0xa0d1
+#define PCI_DEVICE_ID_INTEL_THC_TGL_H_PORT1  0x43d0
+#define PCI_DEVICE_ID_INTEL_THC_TGL_H_PORT2  0x43d1
+// Alder Lake
+#define PCI_DEVICE_ID_INTEL_THC_ADL_S_PORT1  0x7ad8
+#define PCI_DEVICE_ID_INTEL_THC_ADL_S_PORT2  0x7ad9
+#define PCI_DEVICE_ID_INTEL_THC_ADL_P_PORT1  0x51d0
+#define PCI_DEVICE_ID_INTEL_THC_ADL_P_PORT2  0x51d1
+#define PCI_DEVICE_ID_INTEL_THC_ADL_M_PORT1  0x54d0
+#define PCI_DEVICE_ID_INTEL_THC_ADL_M_PORT2  0x54d1
+// Raptor Lake
+#define PCI_DEVICE_ID_INTEL_THC_RPL_S_PORT1  0x7a58
+#define PCI_DEVICE_ID_INTEL_THC_RPL_S_PORT2  0x7a59
+// Meteor Lake
+#define PCI_DEVICE_ID_INTEL_THC_MTL_PORT1    0x7e48
+#define PCI_DEVICE_ID_INTEL_THC_MTL_PORT2    0x7e4a
+
+static const struct pci_device_id ithc_pci_tbl[] = {
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_LKF_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_LKF_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_TGL_LP_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_TGL_LP_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_TGL_H_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_TGL_H_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_S_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_S_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_P_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_P_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_M_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_ADL_M_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_RPL_S_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_RPL_S_PORT2) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_MTL_PORT1) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_THC_MTL_PORT2) },
+	{}
+};
+MODULE_DEVICE_TABLE(pci, ithc_pci_tbl);
+
+// Module parameters
+
+static bool ithc_use_polling = false;
+module_param_named(poll, ithc_use_polling, bool, 0);
+MODULE_PARM_DESC(poll, "Use polling instead of interrupts");
+
+static bool ithc_use_rx0 = false;
+module_param_named(rx0, ithc_use_rx0, bool, 0);
+MODULE_PARM_DESC(rx0, "Use DMA RX channel 0");
+
+static bool ithc_use_rx1 = true;
+module_param_named(rx1, ithc_use_rx1, bool, 0);
+MODULE_PARM_DESC(rx1, "Use DMA RX channel 1");
+
+static bool ithc_log_regs_enabled = false;
+module_param_named(logregs, ithc_log_regs_enabled, bool, 0);
+MODULE_PARM_DESC(logregs, "Log changes in register values (for debugging)");
+
+// Sysfs attributes
+
+static bool ithc_is_config_valid(struct ithc *ithc) {
+	return ithc->config.device_id == DEVCFG_DEVICE_ID_TIC;
+}
+
+static ssize_t vendor_show(struct device *dev, struct device_attribute *attr, char *buf) {
+	struct ithc *ithc = dev_get_drvdata(dev);
+	if (!ithc || !ithc_is_config_valid(ithc)) return -ENODEV;
+	return sprintf(buf, "0x%04x", ithc->config.vendor_id);
+}
+static DEVICE_ATTR_RO(vendor);
+static ssize_t product_show(struct device *dev, struct device_attribute *attr, char *buf) {
+	struct ithc *ithc = dev_get_drvdata(dev);
+	if (!ithc || !ithc_is_config_valid(ithc)) return -ENODEV;
+	return sprintf(buf, "0x%04x", ithc->config.product_id);
+}
+static DEVICE_ATTR_RO(product);
+static ssize_t revision_show(struct device *dev, struct device_attribute *attr, char *buf) {
+	struct ithc *ithc = dev_get_drvdata(dev);
+	if (!ithc || !ithc_is_config_valid(ithc)) return -ENODEV;
+	return sprintf(buf, "%u", ithc->config.revision);
+}
+static DEVICE_ATTR_RO(revision);
+static ssize_t fw_version_show(struct device *dev, struct device_attribute *attr, char *buf) {
+	struct ithc *ithc = dev_get_drvdata(dev);
+	if (!ithc || !ithc_is_config_valid(ithc)) return -ENODEV;
+	u32 v = ithc->config.fw_version;
+	return sprintf(buf, "%i.%i.%i.%i", v >> 24, v >> 16 & 0xff, v >> 8 & 0xff, v & 0xff);
+}
+static DEVICE_ATTR_RO(fw_version);
+
+static const struct attribute_group *ithc_attribute_groups[] = {
+	&(const struct attribute_group){
+		.name = DEVNAME,
+		.attrs = (struct attribute *[]){
+			&dev_attr_vendor.attr,
+			&dev_attr_product.attr,
+			&dev_attr_revision.attr,
+			&dev_attr_fw_version.attr,
+			NULL
+		},
+	},
+	NULL
+};
+
+// HID setup
+
+static int ithc_hid_start(struct hid_device *hdev) { return 0; }
+static void ithc_hid_stop(struct hid_device *hdev) { }
+static int ithc_hid_open(struct hid_device *hdev) { return 0; }
+static void ithc_hid_close(struct hid_device *hdev) { }
+
+static int ithc_hid_parse(struct hid_device *hdev) {
+	struct ithc *ithc = hdev->driver_data;
+	u64 val = 0;
+	WRITE_ONCE(ithc->hid_parse_done, false);
+	CHECK_RET(ithc_dma_tx, ithc, DMA_TX_CODE_GET_REPORT_DESCRIPTOR, sizeof val, &val);
+	if (!wait_event_timeout(ithc->wait_hid_parse, READ_ONCE(ithc->hid_parse_done), msecs_to_jiffies(1000))) return -ETIMEDOUT;
+	return 0;
+}
+
+static int ithc_hid_raw_request(struct hid_device *hdev, unsigned char reportnum, __u8 *buf, size_t len, unsigned char rtype, int reqtype) {
+	struct ithc *ithc = hdev->driver_data;
+	if (!buf || !len) return -EINVAL;
+	u32 code;
+	if (rtype == HID_OUTPUT_REPORT && reqtype == HID_REQ_SET_REPORT) code = DMA_TX_CODE_OUTPUT_REPORT;
+	else if (rtype == HID_FEATURE_REPORT && reqtype == HID_REQ_SET_REPORT) code = DMA_TX_CODE_SET_FEATURE;
+	else if (rtype == HID_FEATURE_REPORT && reqtype == HID_REQ_GET_REPORT) code = DMA_TX_CODE_GET_FEATURE;
+	else {
+		pci_err(ithc->pci, "unhandled hid request %i %i for report id %i\n", rtype, reqtype, reportnum);
+		return -EINVAL;
+	}
+	buf[0] = reportnum;
+	if (reqtype == HID_REQ_GET_REPORT) {
+		mutex_lock(&ithc->hid_get_feature_mutex);
+		ithc->hid_get_feature_buf = buf;
+		ithc->hid_get_feature_size = len;
+		mutex_unlock(&ithc->hid_get_feature_mutex);
+		int r = CHECK(ithc_dma_tx, ithc, code, 1, buf);
+		if (!r) {
+			r = wait_event_interruptible_timeout(ithc->wait_hid_get_feature, !ithc->hid_get_feature_buf, msecs_to_jiffies(1000));
+			if (!r) r = -ETIMEDOUT;
+			else if (r < 0) r = -EINTR;
+			else r = 0;
+		}
+		mutex_lock(&ithc->hid_get_feature_mutex);
+		ithc->hid_get_feature_buf = NULL;
+		if (!r) r = ithc->hid_get_feature_size;
+		mutex_unlock(&ithc->hid_get_feature_mutex);
+		return r;
+	}
+	CHECK_RET(ithc_dma_tx, ithc, code, len, buf);
+	return 0;
+}
+
+static struct hid_ll_driver ithc_ll_driver = {
+	.start = ithc_hid_start,
+	.stop = ithc_hid_stop,
+	.open = ithc_hid_open,
+	.close = ithc_hid_close,
+	.parse = ithc_hid_parse,
+	.raw_request = ithc_hid_raw_request,
+};
+
+static void ithc_hid_devres_release(struct device *dev, void *res) {
+	struct hid_device **hidm = res;
+	if (*hidm) hid_destroy_device(*hidm);
+}
+
+static int ithc_hid_init(struct ithc *ithc) {
+	struct hid_device **hidm = devres_alloc(ithc_hid_devres_release, sizeof *hidm, GFP_KERNEL);
+	if (!hidm) return -ENOMEM;
+	devres_add(&ithc->pci->dev, hidm);
+	struct hid_device *hid = hid_allocate_device();
+	if (IS_ERR(hid)) return PTR_ERR(hid);
+	*hidm = hid;
+
+	strscpy(hid->name, DEVFULLNAME, sizeof(hid->name));
+	strscpy(hid->phys, ithc->phys, sizeof(hid->phys));
+	hid->ll_driver = &ithc_ll_driver;
+	hid->bus = BUS_PCI;
+	hid->vendor = ithc->config.vendor_id;
+	hid->product = ithc->config.product_id;
+	hid->version = 0x100;
+	hid->dev.parent = &ithc->pci->dev;
+	hid->driver_data = ithc;
+
+	ithc->hid = hid;
+	return 0;
+}
+
+// Interrupts/polling
+
+static void ithc_activity_timer_callback(struct timer_list *t) {
+	struct ithc *ithc = container_of(t, struct ithc, activity_timer);
+	cpu_latency_qos_update_request(&ithc->activity_qos, PM_QOS_DEFAULT_VALUE);
+}
+
+void ithc_set_active(struct ithc *ithc) {
+	// When CPU usage is very low, the CPU can enter various low power states (C2-C10).
+	// This disrupts DMA, causing truncated DMA messages. ERROR_FLAG_DMA_UNKNOWN_12 will be set when this happens.
+	// The amount of truncated messages can become very high, resulting in user-visible effects (laggy/stuttering cursor).
+	// To avoid this, we use a CPU latency QoS request to prevent the CPU from entering low power states during touch interactions.
+	cpu_latency_qos_update_request(&ithc->activity_qos, 0);
+	mod_timer(&ithc->activity_timer, jiffies + msecs_to_jiffies(1000));
+}
+
+static int ithc_set_device_enabled(struct ithc *ithc, bool enable) {
+	u32 x = ithc->config.touch_cfg = (ithc->config.touch_cfg & ~(u32)DEVCFG_TOUCH_MASK) | DEVCFG_TOUCH_UNKNOWN_2
+		| (enable ? DEVCFG_TOUCH_ENABLE | DEVCFG_TOUCH_UNKNOWN_3 | DEVCFG_TOUCH_UNKNOWN_4 : 0);
+	return ithc_spi_command(ithc, SPI_CMD_CODE_WRITE, offsetof(struct ithc_device_config, touch_cfg), sizeof x, &x);
+}
+
+static void ithc_disable_interrupts(struct ithc *ithc) {
+	writel(0, &ithc->regs->error_control);
+	bitsb(&ithc->regs->spi_cmd.control, SPI_CMD_CONTROL_IRQ, 0);
+	bitsb(&ithc->regs->dma_rx[0].control, DMA_RX_CONTROL_IRQ_UNKNOWN_1 | DMA_RX_CONTROL_IRQ_ERROR | DMA_RX_CONTROL_IRQ_UNKNOWN_4 | DMA_RX_CONTROL_IRQ_DATA, 0);
+	bitsb(&ithc->regs->dma_rx[1].control, DMA_RX_CONTROL_IRQ_UNKNOWN_1 | DMA_RX_CONTROL_IRQ_ERROR | DMA_RX_CONTROL_IRQ_UNKNOWN_4 | DMA_RX_CONTROL_IRQ_DATA, 0);
+	bitsb(&ithc->regs->dma_tx.control, DMA_TX_CONTROL_IRQ, 0);
+}
+
+static void ithc_clear_dma_rx_interrupts(struct ithc *ithc, unsigned channel) {
+	writel(DMA_RX_STATUS_ERROR | DMA_RX_STATUS_UNKNOWN_4 | DMA_RX_STATUS_HAVE_DATA, &ithc->regs->dma_rx[channel].status);
+}
+
+static void ithc_clear_interrupts(struct ithc *ithc) {
+	writel(0xffffffff, &ithc->regs->error_flags);
+	writel(ERROR_STATUS_DMA | ERROR_STATUS_SPI, &ithc->regs->error_status);
+	writel(SPI_CMD_STATUS_DONE | SPI_CMD_STATUS_ERROR, &ithc->regs->spi_cmd.status);
+	ithc_clear_dma_rx_interrupts(ithc, 0);
+	ithc_clear_dma_rx_interrupts(ithc, 1);
+	writel(DMA_TX_STATUS_DONE | DMA_TX_STATUS_ERROR | DMA_TX_STATUS_UNKNOWN_2, &ithc->regs->dma_tx.status);
+}
+
+static void ithc_process(struct ithc *ithc) {
+	ithc_log_regs(ithc);
+
+	// read and clear error bits
+	u32 err = readl(&ithc->regs->error_flags);
+	if (err) {
+		if (err & ~ERROR_FLAG_DMA_UNKNOWN_12) pci_err(ithc->pci, "error flags: 0x%08x\n", err);
+		writel(err, &ithc->regs->error_flags);
+	}
+
+	// process DMA rx
+	if (ithc_use_rx0) {
+		ithc_clear_dma_rx_interrupts(ithc, 0);
+		ithc_dma_rx(ithc, 0);
+	}
+	if (ithc_use_rx1) {
+		ithc_clear_dma_rx_interrupts(ithc, 1);
+		ithc_dma_rx(ithc, 1);
+	}
+
+	ithc_log_regs(ithc);
+}
+
+static irqreturn_t ithc_interrupt_thread(int irq, void *arg) {
+	struct ithc *ithc = arg;
+	pci_dbg(ithc->pci, "IRQ! err=%08x/%08x/%08x, cmd=%02x/%08x, rx0=%02x/%08x, rx1=%02x/%08x, tx=%02x/%08x\n",
+		readl(&ithc->regs->error_control), readl(&ithc->regs->error_status), readl(&ithc->regs->error_flags),
+		readb(&ithc->regs->spi_cmd.control), readl(&ithc->regs->spi_cmd.status),
+		readb(&ithc->regs->dma_rx[0].control), readl(&ithc->regs->dma_rx[0].status),
+		readb(&ithc->regs->dma_rx[1].control), readl(&ithc->regs->dma_rx[1].status),
+		readb(&ithc->regs->dma_tx.control), readl(&ithc->regs->dma_tx.status));
+	ithc_process(ithc);
+	return IRQ_HANDLED;
+}
+
+static int ithc_poll_thread(void *arg) {
+	struct ithc *ithc = arg;
+	unsigned sleep = 100;
+	while (!kthread_should_stop()) {
+		u32 n = ithc->dma_rx[1].num_received;
+		ithc_process(ithc);
+		if (n != ithc->dma_rx[1].num_received) sleep = 20;
+		else sleep = min(200u, sleep + (sleep >> 4) + 1);
+		msleep_interruptible(sleep);
+	}
+	return 0;
+}
+
+// Device initialization and shutdown
+
+static void ithc_disable(struct ithc *ithc) {
+	bitsl_set(&ithc->regs->control_bits, CONTROL_QUIESCE);
+	CHECK(waitl, ithc, &ithc->regs->control_bits, CONTROL_IS_QUIESCED, CONTROL_IS_QUIESCED);
+	bitsl(&ithc->regs->control_bits, CONTROL_NRESET, 0);
+	bitsb(&ithc->regs->spi_cmd.control, SPI_CMD_CONTROL_SEND, 0);
+	bitsb(&ithc->regs->dma_tx.control, DMA_TX_CONTROL_SEND, 0);
+	bitsb(&ithc->regs->dma_rx[0].control, DMA_RX_CONTROL_ENABLE, 0);
+	bitsb(&ithc->regs->dma_rx[1].control, DMA_RX_CONTROL_ENABLE, 0);
+	ithc_disable_interrupts(ithc);
+	ithc_clear_interrupts(ithc);
+}
+
+static int ithc_init_device(struct ithc *ithc) {
+	ithc_log_regs(ithc);
+	bool was_enabled = (readl(&ithc->regs->control_bits) & CONTROL_NRESET) != 0;
+	ithc_disable(ithc);
+	CHECK_RET(waitl, ithc, &ithc->regs->control_bits, CONTROL_READY, CONTROL_READY);
+	ithc_set_spi_config(ithc, 10, 0);
+	bitsl_set(&ithc->regs->dma_rx[0].unknown_init_bits, 0x80000000); // seems to help with reading config
+
+	if (was_enabled) if (msleep_interruptible(100)) return -EINTR;
+	bitsl(&ithc->regs->control_bits, CONTROL_QUIESCE, 0);
+	CHECK_RET(waitl, ithc, &ithc->regs->control_bits, CONTROL_IS_QUIESCED, 0);
+	for (int retries = 0; ; retries++) {
+		ithc_log_regs(ithc);
+		bitsl_set(&ithc->regs->control_bits, CONTROL_NRESET);
+		if (!waitl(ithc, &ithc->regs->state, 0xf, 2)) break;
+		if (retries > 5) {
+			pci_err(ithc->pci, "too many retries, failed to reset device\n");
+			return -ETIMEDOUT;
+		}
+		pci_err(ithc->pci, "invalid state, retrying reset\n");
+		bitsl(&ithc->regs->control_bits, CONTROL_NRESET, 0);
+		if (msleep_interruptible(1000)) return -EINTR;
+	}
+	ithc_log_regs(ithc);
+
+	CHECK(waitl, ithc, &ithc->regs->dma_rx[0].status, DMA_RX_STATUS_UNKNOWN_4, DMA_RX_STATUS_UNKNOWN_4);
+
+	// read config
+	for (int retries = 0; ; retries++) {
+		ithc_log_regs(ithc);
+		memset(&ithc->config, 0, sizeof ithc->config);
+		CHECK_RET(ithc_spi_command, ithc, SPI_CMD_CODE_READ, 0, sizeof ithc->config, &ithc->config);
+		u32 *p = (void *)&ithc->config;
+		pci_info(ithc->pci, "config: %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
+			p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7], p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
+		if (ithc_is_config_valid(ithc)) break;
+		if (retries > 10) {
+			pci_err(ithc->pci, "failed to read config, unknown device ID 0x%08x\n", ithc->config.device_id);
+			return -EIO;
+		}
+		pci_err(ithc->pci, "failed to read config, retrying\n");
+		if (msleep_interruptible(100)) return -EINTR;
+	}
+	ithc_log_regs(ithc);
+
+	CHECK_RET(ithc_set_spi_config, ithc, DEVCFG_SPI_MAX_FREQ(ithc->config.spi_config), DEVCFG_SPI_MODE(ithc->config.spi_config));
+	CHECK_RET(ithc_set_device_enabled, ithc, true);
+	ithc_log_regs(ithc);
+	return 0;
+}
+
+int ithc_reset(struct ithc *ithc) {
+	// FIXME This should probably do devres_release_group()+ithc_start(). But because this is called during DMA
+	// processing, that would have to be done asynchronously (schedule_work()?). And with extra locking?
+	pci_err(ithc->pci, "reset\n");
+	CHECK(ithc_init_device, ithc);
+	if (ithc_use_rx0) ithc_dma_rx_enable(ithc, 0);
+	if (ithc_use_rx1) ithc_dma_rx_enable(ithc, 1);
+	ithc_log_regs(ithc);
+	pci_dbg(ithc->pci, "reset completed\n");
+	return 0;
+}
+
+static void ithc_stop(void *res) {
+	struct ithc *ithc = res;
+	pci_dbg(ithc->pci, "stopping\n");
+	ithc_log_regs(ithc);
+	if (ithc->poll_thread) CHECK(kthread_stop, ithc->poll_thread);
+	if (ithc->irq >= 0) disable_irq(ithc->irq);
+	CHECK(ithc_set_device_enabled, ithc, false);
+	ithc_disable(ithc);
+	del_timer_sync(&ithc->activity_timer);
+	cpu_latency_qos_remove_request(&ithc->activity_qos);
+	// clear dma config
+	for(unsigned i = 0; i < 2; i++) {
+		CHECK(waitl, ithc, &ithc->regs->dma_rx[i].status, DMA_RX_STATUS_ENABLED, 0);
+		lo_hi_writeq(0, &ithc->regs->dma_rx[i].addr);
+		writeb(0, &ithc->regs->dma_rx[i].num_bufs);
+		writeb(0, &ithc->regs->dma_rx[i].num_prds);
+	}
+	lo_hi_writeq(0, &ithc->regs->dma_tx.addr);
+	writeb(0, &ithc->regs->dma_tx.num_prds);
+	ithc_log_regs(ithc);
+	pci_dbg(ithc->pci, "stopped\n");
+}
+
+static void ithc_clear_drvdata(void *res) {
+	struct pci_dev *pci = res;
+	pci_set_drvdata(pci, NULL);
+}
+
+static int ithc_start(struct pci_dev *pci) {
+	pci_dbg(pci, "starting\n");
+	if (pci_get_drvdata(pci)) {
+		pci_err(pci, "device already initialized\n");
+		return -EINVAL;
+	}
+	if (!devres_open_group(&pci->dev, ithc_start, GFP_KERNEL)) return -ENOMEM;
+
+	struct ithc *ithc = devm_kzalloc(&pci->dev, sizeof *ithc, GFP_KERNEL);
+	if (!ithc) return -ENOMEM;
+	ithc->irq = -1;
+	ithc->pci = pci;
+	snprintf(ithc->phys, sizeof ithc->phys, "pci-%s/" DEVNAME, pci_name(pci));
+	init_waitqueue_head(&ithc->wait_hid_parse);
+	init_waitqueue_head(&ithc->wait_hid_get_feature);
+	mutex_init(&ithc->hid_get_feature_mutex);
+	pci_set_drvdata(pci, ithc);
+	CHECK_RET(devm_add_action_or_reset, &pci->dev, ithc_clear_drvdata, pci);
+	if (ithc_log_regs_enabled) ithc->prev_regs = devm_kzalloc(&pci->dev, sizeof *ithc->prev_regs, GFP_KERNEL);
+
+	CHECK_RET(pcim_enable_device, pci);
+	pci_set_master(pci);
+	CHECK_RET(pcim_iomap_regions, pci, BIT(0), DEVNAME " regs");
+	CHECK_RET(dma_set_mask_and_coherent, &pci->dev, DMA_BIT_MASK(64));
+	CHECK_RET(pci_set_power_state, pci, PCI_D0);
+	ithc->regs = pcim_iomap_table(pci)[0];
+
+	if (!ithc_use_polling) {
+		CHECK_RET(pci_alloc_irq_vectors, pci, 1, 1, PCI_IRQ_MSI | PCI_IRQ_MSIX);
+		ithc->irq = CHECK(pci_irq_vector, pci, 0);
+		if (ithc->irq < 0) return ithc->irq;
+	}
+
+	CHECK_RET(ithc_init_device, ithc);
+	CHECK(devm_device_add_groups, &pci->dev, ithc_attribute_groups);
+	if (ithc_use_rx0) CHECK_RET(ithc_dma_rx_init, ithc, 0, ithc_use_rx1 ? DEVNAME "0" : DEVNAME);
+	if (ithc_use_rx1) CHECK_RET(ithc_dma_rx_init, ithc, 1, ithc_use_rx0 ? DEVNAME "1" : DEVNAME);
+	CHECK_RET(ithc_dma_tx_init, ithc);
+
+	CHECK_RET(ithc_hid_init, ithc);
+
+	cpu_latency_qos_add_request(&ithc->activity_qos, PM_QOS_DEFAULT_VALUE);
+	timer_setup(&ithc->activity_timer, ithc_activity_timer_callback, 0);
+
+	// add ithc_stop callback AFTER setting up DMA buffers, so that polling/irqs/DMA are disabled BEFORE the buffers are freed
+	CHECK_RET(devm_add_action_or_reset, &pci->dev, ithc_stop, ithc);
+
+	if (ithc_use_polling) {
+		pci_info(pci, "using polling instead of irq\n");
+		// use a thread instead of simple timer because we want to be able to sleep
+		ithc->poll_thread = kthread_run(ithc_poll_thread, ithc, DEVNAME "poll");
+		if (IS_ERR(ithc->poll_thread)) {
+			int err = PTR_ERR(ithc->poll_thread);
+			ithc->poll_thread = NULL;
+			return err;
+		}
+	} else {
+		CHECK_RET(devm_request_threaded_irq, &pci->dev, ithc->irq, NULL, ithc_interrupt_thread, IRQF_TRIGGER_HIGH | IRQF_ONESHOT, DEVNAME, ithc);
+	}
+
+	if (ithc_use_rx0) ithc_dma_rx_enable(ithc, 0);
+	if (ithc_use_rx1) ithc_dma_rx_enable(ithc, 1);
+
+	// hid_add_device can only be called after irq/polling is started and DMA is enabled, because it calls ithc_hid_parse which reads the report descriptor via DMA
+	CHECK_RET(hid_add_device, ithc->hid);
+
+	CHECK(ithc_debug_init, ithc);
+
+	pci_dbg(pci, "started\n");
+	return 0;
+}
+
+static int ithc_probe(struct pci_dev *pci, const struct pci_device_id *id) {
+	pci_dbg(pci, "device probe\n");
+	return ithc_start(pci);
+}
+
+static void ithc_remove(struct pci_dev *pci) {
+	pci_dbg(pci, "device remove\n");
+	// all cleanup is handled by devres
+}
+
+static int ithc_suspend(struct device *dev) {
+	struct pci_dev *pci = to_pci_dev(dev);
+	pci_dbg(pci, "pm suspend\n");
+	devres_release_group(dev, ithc_start);
+	return 0;
+}
+
+static int ithc_resume(struct device *dev) {
+	struct pci_dev *pci = to_pci_dev(dev);
+	pci_dbg(pci, "pm resume\n");
+	return ithc_start(pci);
+}
+
+static int ithc_freeze(struct device *dev) {
+	struct pci_dev *pci = to_pci_dev(dev);
+	pci_dbg(pci, "pm freeze\n");
+	devres_release_group(dev, ithc_start);
+	return 0;
+}
+
+static int ithc_thaw(struct device *dev) {
+	struct pci_dev *pci = to_pci_dev(dev);
+	pci_dbg(pci, "pm thaw\n");
+	return ithc_start(pci);
+}
+
+static int ithc_restore(struct device *dev) {
+	struct pci_dev *pci = to_pci_dev(dev);
+	pci_dbg(pci, "pm restore\n");
+	return ithc_start(pci);
+}
+
+static struct pci_driver ithc_driver = {
+	.name = DEVNAME,
+	.id_table = ithc_pci_tbl,
+	.probe = ithc_probe,
+	.remove = ithc_remove,
+	.driver.pm = &(const struct dev_pm_ops) {
+		.suspend = ithc_suspend,
+		.resume = ithc_resume,
+		.freeze = ithc_freeze,
+		.thaw = ithc_thaw,
+		.restore = ithc_restore,
+	},
+	//.dev_groups = ithc_attribute_groups, // could use this (since 5.14), however the attributes won't have valid values until config has been read anyway
+};
+
+static int __init ithc_init(void) {
+	return pci_register_driver(&ithc_driver);
+}
+
+static void __exit ithc_exit(void) {
+	pci_unregister_driver(&ithc_driver);
+}
+
+module_init(ithc_init);
+module_exit(ithc_exit);
+
diff --git a/drivers/hid/ithc/ithc-regs.c b/drivers/hid/ithc/ithc-regs.c
new file mode 100644
index 0000000000000..85d567b05761f
--- /dev/null
+++ b/drivers/hid/ithc/ithc-regs.c
@@ -0,0 +1,64 @@
+#include "ithc.h"
+
+#define reg_num(r) (0x1fff & (u16)(__force u64)(r))
+
+void bitsl(__iomem u32 *reg, u32 mask, u32 val) {
+	if (val & ~mask) pr_err("register 0x%x: invalid value 0x%x for bitmask 0x%x\n", reg_num(reg), val, mask);
+	writel((readl(reg) & ~mask) | (val & mask), reg);
+}
+
+void bitsb(__iomem u8 *reg, u8 mask, u8 val) {
+	if (val & ~mask) pr_err("register 0x%x: invalid value 0x%x for bitmask 0x%x\n", reg_num(reg), val, mask);
+	writeb((readb(reg) & ~mask) | (val & mask), reg);
+}
+
+int waitl(struct ithc *ithc, __iomem u32 *reg, u32 mask, u32 val) {
+	pci_dbg(ithc->pci, "waiting for reg 0x%04x mask 0x%08x val 0x%08x\n", reg_num(reg), mask, val);
+	u32 x;
+	if (readl_poll_timeout(reg, x, (x & mask) == val, 200, 1000*1000)) {
+		pci_err(ithc->pci, "timed out waiting for reg 0x%04x mask 0x%08x val 0x%08x\n", reg_num(reg), mask, val);
+		return -ETIMEDOUT;
+	}
+	pci_dbg(ithc->pci, "done waiting\n");
+	return 0;
+}
+
+int waitb(struct ithc *ithc, __iomem u8 *reg, u8 mask, u8 val) {
+	pci_dbg(ithc->pci, "waiting for reg 0x%04x mask 0x%02x val 0x%02x\n", reg_num(reg), mask, val);
+	u8 x;
+	if (readb_poll_timeout(reg, x, (x & mask) == val, 200, 1000*1000)) {
+		pci_err(ithc->pci, "timed out waiting for reg 0x%04x mask 0x%02x val 0x%02x\n", reg_num(reg), mask, val);
+		return -ETIMEDOUT;
+	}
+	pci_dbg(ithc->pci, "done waiting\n");
+	return 0;
+}
+
+int ithc_set_spi_config(struct ithc *ithc, u8 speed, u8 mode) {
+	pci_dbg(ithc->pci, "setting SPI speed to %i, mode %i\n", speed, mode);
+	if (mode == 3) mode = 2;
+	bitsl(&ithc->regs->spi_config,
+		SPI_CONFIG_MODE(0xff) | SPI_CONFIG_SPEED(0xff) | SPI_CONFIG_UNKNOWN_18(0xff) | SPI_CONFIG_SPEED2(0xff),
+		SPI_CONFIG_MODE(mode) | SPI_CONFIG_SPEED(speed) | SPI_CONFIG_UNKNOWN_18(0) | SPI_CONFIG_SPEED2(speed));
+	return 0;
+}
+
+int ithc_spi_command(struct ithc *ithc, u8 command, u32 offset, u32 size, void *data) {
+	pci_dbg(ithc->pci, "SPI command %u, size %u, offset %u\n", command, size, offset);
+	if (size > sizeof ithc->regs->spi_cmd.data) return -EINVAL;
+	CHECK_RET(waitl, ithc, &ithc->regs->spi_cmd.status, SPI_CMD_STATUS_BUSY, 0);
+	writel(SPI_CMD_STATUS_DONE | SPI_CMD_STATUS_ERROR, &ithc->regs->spi_cmd.status);
+	writeb(command, &ithc->regs->spi_cmd.code);
+	writew(size, &ithc->regs->spi_cmd.size);
+	writel(offset, &ithc->regs->spi_cmd.offset);
+	u32 *p = data, n = (size + 3) / 4;
+	for (u32 i = 0; i < n; i++) writel(p[i], &ithc->regs->spi_cmd.data[i]);
+	bitsb_set(&ithc->regs->spi_cmd.control, SPI_CMD_CONTROL_SEND);
+	CHECK_RET(waitl, ithc, &ithc->regs->spi_cmd.status, SPI_CMD_STATUS_BUSY, 0);
+	if ((readl(&ithc->regs->spi_cmd.status) & (SPI_CMD_STATUS_DONE | SPI_CMD_STATUS_ERROR)) != SPI_CMD_STATUS_DONE) return -EIO;
+	if (readw(&ithc->regs->spi_cmd.size) != size) return -EMSGSIZE;
+	for (u32 i = 0; i < n; i++) p[i] = readl(&ithc->regs->spi_cmd.data[i]);
+	writel(SPI_CMD_STATUS_DONE | SPI_CMD_STATUS_ERROR, &ithc->regs->spi_cmd.status);
+	return 0;
+}
+
diff --git a/drivers/hid/ithc/ithc-regs.h b/drivers/hid/ithc/ithc-regs.h
new file mode 100644
index 0000000000000..1a96092ed7eed
--- /dev/null
+++ b/drivers/hid/ithc/ithc-regs.h
@@ -0,0 +1,186 @@
+#define CONTROL_QUIESCE                     BIT(1)
+#define CONTROL_IS_QUIESCED                 BIT(2)
+#define CONTROL_NRESET                      BIT(3)
+#define CONTROL_READY                       BIT(29)
+
+#define SPI_CONFIG_MODE(x)                  (((x) & 3) << 2)
+#define SPI_CONFIG_SPEED(x)                 (((x) & 7) << 4)
+#define SPI_CONFIG_UNKNOWN_18(x)            (((x) & 3) << 18)
+#define SPI_CONFIG_SPEED2(x)                (((x) & 0xf) << 20) // high bit = high speed mode?
+
+#define ERROR_CONTROL_UNKNOWN_0             BIT(0)
+#define ERROR_CONTROL_DISABLE_DMA           BIT(1) // clears DMA_RX_CONTROL_ENABLE when a DMA error occurs
+#define ERROR_CONTROL_UNKNOWN_2             BIT(2)
+#define ERROR_CONTROL_UNKNOWN_3             BIT(3)
+#define ERROR_CONTROL_IRQ_DMA_UNKNOWN_9     BIT(9)
+#define ERROR_CONTROL_IRQ_DMA_UNKNOWN_10    BIT(10)
+#define ERROR_CONTROL_IRQ_DMA_UNKNOWN_12    BIT(12)
+#define ERROR_CONTROL_IRQ_DMA_UNKNOWN_13    BIT(13)
+#define ERROR_CONTROL_UNKNOWN_16(x)         (((x) & 0xff) << 16) // spi error code irq?
+#define ERROR_CONTROL_SET_DMA_STATUS        BIT(29) // sets DMA_RX_STATUS_ERROR when a DMA error occurs
+
+#define ERROR_STATUS_DMA                    BIT(28)
+#define ERROR_STATUS_SPI                    BIT(30)
+
+#define ERROR_FLAG_DMA_UNKNOWN_9            BIT(9)
+#define ERROR_FLAG_DMA_UNKNOWN_10           BIT(10)
+#define ERROR_FLAG_DMA_UNKNOWN_12           BIT(12) // set when we receive a truncated DMA message
+#define ERROR_FLAG_DMA_UNKNOWN_13           BIT(13)
+#define ERROR_FLAG_SPI_BUS_TURNAROUND       BIT(16)
+#define ERROR_FLAG_SPI_RESPONSE_TIMEOUT     BIT(17)
+#define ERROR_FLAG_SPI_INTRA_PACKET_TIMEOUT BIT(18)
+#define ERROR_FLAG_SPI_INVALID_RESPONSE     BIT(19)
+#define ERROR_FLAG_SPI_HS_RX_TIMEOUT        BIT(20)
+#define ERROR_FLAG_SPI_TOUCH_IC_INIT        BIT(21)
+
+#define SPI_CMD_CONTROL_SEND                BIT(0) // cleared by device when sending is complete
+#define SPI_CMD_CONTROL_IRQ                 BIT(1)
+
+#define SPI_CMD_CODE_READ                   4
+#define SPI_CMD_CODE_WRITE                  6
+
+#define SPI_CMD_STATUS_DONE                 BIT(0)
+#define SPI_CMD_STATUS_ERROR                BIT(1)
+#define SPI_CMD_STATUS_BUSY                 BIT(3)
+
+#define DMA_TX_CONTROL_SEND                 BIT(0) // cleared by device when sending is complete
+#define DMA_TX_CONTROL_IRQ                  BIT(3)
+
+#define DMA_TX_STATUS_DONE                  BIT(0)
+#define DMA_TX_STATUS_ERROR                 BIT(1)
+#define DMA_TX_STATUS_UNKNOWN_2             BIT(2)
+#define DMA_TX_STATUS_UNKNOWN_3             BIT(3) // busy?
+
+#define DMA_RX_CONTROL_ENABLE               BIT(0)
+#define DMA_RX_CONTROL_IRQ_UNKNOWN_1        BIT(1) // rx1 only?
+#define DMA_RX_CONTROL_IRQ_ERROR            BIT(3) // rx1 only?
+#define DMA_RX_CONTROL_IRQ_UNKNOWN_4        BIT(4) // rx0 only?
+#define DMA_RX_CONTROL_IRQ_DATA             BIT(5)
+
+#define DMA_RX_CONTROL2_UNKNOWN_5           BIT(5) // rx0 only?
+#define DMA_RX_CONTROL2_RESET               BIT(7) // resets ringbuffer indices
+
+#define DMA_RX_WRAP_FLAG                    BIT(7)
+
+#define DMA_RX_STATUS_ERROR                 BIT(3)
+#define DMA_RX_STATUS_UNKNOWN_4             BIT(4) // set in rx0 after using CONTROL_NRESET when it becomes possible to read config (can take >100ms)
+#define DMA_RX_STATUS_HAVE_DATA             BIT(5)
+#define DMA_RX_STATUS_ENABLED               BIT(8)
+
+#define COUNTER_RESET                       BIT(31)
+
+struct ithc_registers {
+	/* 0000 */ u32 _unknown_0000[1024];
+	/* 1000 */ u32 _unknown_1000;
+	/* 1004 */ u32 _unknown_1004;
+	/* 1008 */ u32 control_bits;
+	/* 100c */ u32 _unknown_100c;
+	/* 1010 */ u32 spi_config;
+	/* 1014 */ u32 _unknown_1014[3];
+	/* 1020 */ u32 error_control;
+	/* 1024 */ u32 error_status; // write to clear
+	/* 1028 */ u32 error_flags; // write to clear
+	/* 102c */ u32 _unknown_102c[5];
+	struct {
+		/* 1040 */ u8 control;
+		/* 1041 */ u8 code;
+		/* 1042 */ u16 size;
+		/* 1044 */ u32 status; // write to clear
+		/* 1048 */ u32 offset;
+		/* 104c */ u32 data[16];
+		/* 108c */ u32 _unknown_108c;
+	} spi_cmd;
+	struct {
+		/* 1090 */ u64 addr; // cannot be written with writeq(), must use lo_hi_writeq()
+		/* 1098 */ u8 control;
+		/* 1099 */ u8 _unknown_1099;
+		/* 109a */ u8 _unknown_109a;
+		/* 109b */ u8 num_prds;
+		/* 109c */ u32 status; // write to clear
+	} dma_tx;
+	/* 10a0 */ u32 _unknown_10a0[7];
+	/* 10bc */ u32 state; // is 0xe0000402 (dev config val 0) after CONTROL_NRESET, 0xe0000461 after first touch, 0xe0000401 after DMA_RX_CODE_RESET
+	/* 10c0 */ u32 _unknown_10c0[8];
+	/* 10e0 */ u32 _unknown_10e0_counters[3];
+	/* 10ec */ u32 _unknown_10ec[5];
+	struct {
+		/* 1100/1200 */ u64 addr; // cannot be written with writeq(), must use lo_hi_writeq()
+		/* 1108/1208 */ u8 num_bufs;
+		/* 1109/1209 */ u8 num_prds;
+		/* 110a/120a */ u16 _unknown_110a;
+		/* 110c/120c */ u8 control;
+		/* 110d/120d */ u8 head;
+		/* 110e/120e */ u8 tail;
+		/* 110f/120f */ u8 control2;
+		/* 1110/1210 */ u32 status; // write to clear
+		/* 1114/1214 */ u32 _unknown_1114;
+		/* 1118/1218 */ u64 _unknown_1118_guc_addr;
+		/* 1120/1220 */ u32 _unknown_1120_guc;
+		/* 1124/1224 */ u32 _unknown_1124_guc;
+		/* 1128/1228 */ u32 unknown_init_bits; // bit 2 = guc related, bit 3 = rx1 related, bit 4 = guc related
+		/* 112c/122c */ u32 _unknown_112c;
+		/* 1130/1230 */ u64 _unknown_1130_guc_addr;
+		/* 1138/1238 */ u32 _unknown_1138_guc;
+		/* 113c/123c */ u32 _unknown_113c;
+		/* 1140/1240 */ u32 _unknown_1140_guc;
+		/* 1144/1244 */ u32 _unknown_1144[23];
+		/* 11a0/12a0 */ u32 _unknown_11a0_counters[6];
+		/* 11b8/12b8 */ u32 _unknown_11b8[18];
+	} dma_rx[2];
+};
+static_assert(sizeof(struct ithc_registers) == 0x1300);
+
+#define DEVCFG_DMA_RX_SIZE(x)          ((((x) & 0x3fff) + 1) << 6)
+#define DEVCFG_DMA_TX_SIZE(x)          (((((x) >> 14) & 0x3ff) + 1) << 6)
+
+#define DEVCFG_TOUCH_MASK              0x3f
+#define DEVCFG_TOUCH_ENABLE            BIT(0)
+#define DEVCFG_TOUCH_UNKNOWN_1         BIT(1)
+#define DEVCFG_TOUCH_UNKNOWN_2         BIT(2)
+#define DEVCFG_TOUCH_UNKNOWN_3         BIT(3)
+#define DEVCFG_TOUCH_UNKNOWN_4         BIT(4)
+#define DEVCFG_TOUCH_UNKNOWN_5         BIT(5)
+#define DEVCFG_TOUCH_UNKNOWN_6         BIT(6)
+
+#define DEVCFG_DEVICE_ID_TIC           0x43495424 // "$TIC"
+
+#define DEVCFG_SPI_MAX_FREQ(x)         (((x) >> 1) & 0xf) // high bit = use high speed mode?
+#define DEVCFG_SPI_MODE(x)             (((x) >> 6) & 3)
+#define DEVCFG_SPI_UNKNOWN_8(x)        (((x) >> 8) & 0x3f)
+#define DEVCFG_SPI_NEEDS_HEARTBEAT     BIT(20)
+#define DEVCFG_SPI_HEARTBEAT_INTERVAL  (((x) >> 21) & 7)
+#define DEVCFG_SPI_UNKNOWN_25          BIT(25)
+#define DEVCFG_SPI_UNKNOWN_26          BIT(26)
+#define DEVCFG_SPI_UNKNOWN_27          BIT(27)
+#define DEVCFG_SPI_DELAY               (((x) >> 28) & 7)
+#define DEVCFG_SPI_USE_EXT_READ_CFG    BIT(31)
+
+struct ithc_device_config {
+	u32 _unknown_00;      // 00 = 0xe0000402 (0xe0000401 after DMA_RX_CODE_RESET)
+	u32 _unknown_04;      // 04 = 0x00000000
+	u32 dma_buf_sizes;    // 08 = 0x000a00ff
+	u32 touch_cfg;        // 0c = 0x0000001c
+	u32 _unknown_10;      // 10 = 0x0000001c
+	u32 device_id;        // 14 = 0x43495424 = "$TIC"
+	u32 spi_config;       // 18 = 0xfda00a2e
+	u16 vendor_id;        // 1c = 0x045e = Microsoft Corp.
+	u16 product_id;       // 1e = 0x0c1a
+	u32 revision;         // 20 = 0x00000001
+	u32 fw_version;       // 24 = 0x05008a8b = 5.0.138.139
+	u32 _unknown_28;      // 28 = 0x00000000
+	u32 fw_mode;          // 2c = 0x00000000
+	u32 _unknown_30;      // 30 = 0x00000000
+	u32 _unknown_34;      // 34 = 0x0404035e (u8,u8,u8,u8 = version?)
+	u32 _unknown_38;      // 38 = 0x000001c0 (0x000001c1 after DMA_RX_CODE_RESET)
+	u32 _unknown_3c;      // 3c = 0x00000002
+};
+
+void bitsl(__iomem u32 *reg, u32 mask, u32 val);
+void bitsb(__iomem u8 *reg, u8 mask, u8 val);
+#define bitsl_set(reg, x) bitsl(reg, x, x)
+#define bitsb_set(reg, x) bitsb(reg, x, x)
+int waitl(struct ithc *ithc, __iomem u32 *reg, u32 mask, u32 val);
+int waitb(struct ithc *ithc, __iomem u8 *reg, u8 mask, u8 val);
+int ithc_set_spi_config(struct ithc *ithc, u8 speed, u8 mode);
+int ithc_spi_command(struct ithc *ithc, u8 command, u32 offset, u32 size, void *data);
+
diff --git a/drivers/hid/ithc/ithc.h b/drivers/hid/ithc/ithc.h
new file mode 100644
index 0000000000000..6a9b0d480bc15
--- /dev/null
+++ b/drivers/hid/ithc/ithc.h
@@ -0,0 +1,60 @@
+#include <linux/module.h>
+#include <linux/input.h>
+#include <linux/hid.h>
+#include <linux/dma-mapping.h>
+#include <linux/highmem.h>
+#include <linux/pci.h>
+#include <linux/io-64-nonatomic-lo-hi.h>
+#include <linux/iopoll.h>
+#include <linux/delay.h>
+#include <linux/kthread.h>
+#include <linux/miscdevice.h>
+#include <linux/debugfs.h>
+#include <linux/poll.h>
+#include <linux/timer.h>
+#include <linux/pm_qos.h>
+
+#define DEVNAME "ithc"
+#define DEVFULLNAME "Intel Touch Host Controller"
+
+#undef pr_fmt
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#define CHECK(fn, ...) ({ int r = fn(__VA_ARGS__); if (r < 0) pci_err(ithc->pci, "%s: %s failed with %i\n", __func__, #fn, r); r; })
+#define CHECK_RET(...) do { int r = CHECK(__VA_ARGS__); if (r < 0) return r; } while(0)
+
+#define NUM_RX_BUF 16
+
+struct ithc;
+
+#include "ithc-regs.h"
+#include "ithc-dma.h"
+
+struct ithc {
+	char phys[32];
+	struct pci_dev *pci;
+	int irq;
+	struct task_struct *poll_thread;
+	struct pm_qos_request activity_qos;
+	struct timer_list activity_timer;
+
+	struct hid_device *hid;
+	bool hid_parse_done;
+	wait_queue_head_t wait_hid_parse;
+	wait_queue_head_t wait_hid_get_feature;
+	struct mutex hid_get_feature_mutex;
+	void *hid_get_feature_buf;
+	size_t hid_get_feature_size;
+
+	struct ithc_registers __iomem *regs;
+	struct ithc_registers *prev_regs; // for debugging
+	struct ithc_device_config config;
+	struct ithc_dma_rx dma_rx[2];
+	struct ithc_dma_tx dma_tx;
+};
+
+int ithc_reset(struct ithc *ithc);
+void ithc_set_active(struct ithc *ithc);
+int ithc_debug_init(struct ithc *ithc);
+void ithc_log_regs(struct ithc *ithc);
+
-- 
2.39.2

From 95a60c5bc443f1658c987e837722d0dfdbe1e9a4 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:20 +0100
Subject: [PATCH] platform/surface: aggregator: Improve documentation and
 handling of message target and source IDs

The `tid_in` and `tid_out` fields of the serial hub protocol command
struct (struct ssh_command) are actually source and target IDs,
indicating the peer from which the message originated and the peer for
which it is intended.

Change the naming of those fields accordingly and improve the protocol
documentation. Additionally, introduce an enum containing all currently
known peers, i.e. targets and sources.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-3-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 .../driver-api/surface_aggregator/client.rst  |  4 +-
 .../driver-api/surface_aggregator/ssh.rst     | 36 +++++++++--------
 .../platform/surface/aggregator/controller.c  | 12 +++---
 .../platform/surface/aggregator/ssh_msgb.h    |  4 +-
 .../surface/aggregator/ssh_request_layer.c    | 11 ++---
 include/linux/surface_aggregator/controller.h |  4 +-
 include/linux/surface_aggregator/serial_hub.h | 40 +++++++++++++------
 7 files changed, 64 insertions(+), 47 deletions(-)

diff --git a/Documentation/driver-api/surface_aggregator/client.rst b/Documentation/driver-api/surface_aggregator/client.rst
index 27f95abdbe997..9d7411223a848 100644
--- a/Documentation/driver-api/surface_aggregator/client.rst
+++ b/Documentation/driver-api/surface_aggregator/client.rst
@@ -191,7 +191,7 @@ data received from it is converted from little-endian to host endianness.
             *       they do not correspond to an actual SAM/EC request.
             */
            rqst.target_category = SSAM_SSH_TC_SAM;
-           rqst.target_id = 0x01;
+           rqst.target_id = SSAM_SSH_TID_SAM;
            rqst.command_id = 0x02;
            rqst.instance_id = 0x03;
            rqst.flags = SSAM_REQUEST_HAS_RESPONSE;
@@ -241,7 +241,7 @@ one of the generator macros, for example via:
 
    SSAM_DEFINE_SYNC_REQUEST_W(__ssam_tmp_perf_mode_set, __le32, {
            .target_category = SSAM_SSH_TC_TMP,
-           .target_id       = 0x01,
+           .target_id       = SSAM_SSH_TID_SAM,
            .command_id      = 0x03,
            .instance_id     = 0x00,
    });
diff --git a/Documentation/driver-api/surface_aggregator/ssh.rst b/Documentation/driver-api/surface_aggregator/ssh.rst
index bf007d6c98732..18fd0f0aee84b 100644
--- a/Documentation/driver-api/surface_aggregator/ssh.rst
+++ b/Documentation/driver-api/surface_aggregator/ssh.rst
@@ -13,6 +13,7 @@
 .. |DATA_NSQ| replace:: ``DATA_NSQ``
 .. |TC| replace:: ``TC``
 .. |TID| replace:: ``TID``
+.. |SID| replace:: ``SID``
 .. |IID| replace:: ``IID``
 .. |RQID| replace:: ``RQID``
 .. |CID| replace:: ``CID``
@@ -219,13 +220,13 @@ following fields, packed together and in order:
      - |u8|
      - Target category.
 
-   * - |TID| (out)
+   * - |TID|
      - |u8|
-     - Target ID for outgoing (host to EC) commands.
+     - Target ID for commands/messages.
 
-   * - |TID| (in)
+   * - |SID|
      - |u8|
-     - Target ID for incoming (EC to host) commands.
+     - Source ID for commands/messages.
 
    * - |IID|
      - |u8|
@@ -286,19 +287,20 @@ general, however, a single target category should map to a single reserved
 event request ID.
 
 Furthermore, requests, responses, and events have an associated target ID
-(``TID``). This target ID is split into output (host to EC) and input (EC to
-host) fields, with the respecting other field (e.g. output field on incoming
-messages) set to zero. Two ``TID`` values are known: Primary (``0x01``) and
-secondary (``0x02``). In general, the response to a request should have the
-same ``TID`` value, however, the field (output vs. input) should be used in
-accordance to the direction in which the response is sent (i.e. on the input
-field, as responses are generally sent from the EC to the host).
-
-Note that, even though requests and events should be uniquely identifiable
-by target category and command ID alone, the EC may require specific
-target ID and instance ID values to accept a command. A command that is
-accepted for ``TID=1``, for example, may not be accepted for ``TID=2``
-and vice versa.
+(``TID``) and source ID (``SID``). These two fields indicate where a message
+originates from (``SID``) and what the intended target of the message is
+(``TID``). Note that a response to a specific request therefore has the source
+and target IDs swapped when compared to the original request (i.e. the request
+target is the response source and the request source is the response target).
+See (:c:type:`enum ssh_request_id <ssh_request_id>`) for possible values of
+both.
+
+Note that, even though requests and events should be uniquely identifiable by
+target category and command ID alone, the EC may require specific target ID and
+instance ID values to accept a command. A command that is accepted for
+``TID=1``, for example, may not be accepted for ``TID=2`` and vice versa. While
+this may not always hold in reality, you can think of different target/source
+IDs indicating different physical ECs with potentially different feature sets.
 
 
 Limitations and Observations
diff --git a/drivers/platform/surface/aggregator/controller.c b/drivers/platform/surface/aggregator/controller.c
index c6537a1b3a2ec..2c99f51ccd4ec 100644
--- a/drivers/platform/surface/aggregator/controller.c
+++ b/drivers/platform/surface/aggregator/controller.c
@@ -994,7 +994,7 @@ static void ssam_handle_event(struct ssh_rtl *rtl,
 
 	item->rqid = get_unaligned_le16(&cmd->rqid);
 	item->event.target_category = cmd->tc;
-	item->event.target_id = cmd->tid_in;
+	item->event.target_id = cmd->sid;
 	item->event.command_id = cmd->cid;
 	item->event.instance_id = cmd->iid;
 	memcpy(&item->event.data[0], data->ptr, data->len);
@@ -1779,35 +1779,35 @@ EXPORT_SYMBOL_GPL(ssam_request_sync_with_buffer);
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_ssh_get_firmware_version, __le32, {
 	.target_category = SSAM_SSH_TC_SAM,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x13,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_ssh_notif_display_off, u8, {
 	.target_category = SSAM_SSH_TC_SAM,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x15,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_ssh_notif_display_on, u8, {
 	.target_category = SSAM_SSH_TC_SAM,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x16,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_ssh_notif_d0_exit, u8, {
 	.target_category = SSAM_SSH_TC_SAM,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x33,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_ssh_notif_d0_entry, u8, {
 	.target_category = SSAM_SSH_TC_SAM,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x34,
 	.instance_id     = 0x00,
 });
diff --git a/drivers/platform/surface/aggregator/ssh_msgb.h b/drivers/platform/surface/aggregator/ssh_msgb.h
index f3ecad92eefd8..438873e060986 100644
--- a/drivers/platform/surface/aggregator/ssh_msgb.h
+++ b/drivers/platform/surface/aggregator/ssh_msgb.h
@@ -189,8 +189,8 @@ static inline void msgb_push_cmd(struct msgbuf *msgb, u8 seq, u16 rqid,
 
 	__msgb_push_u8(msgb, SSH_PLD_TYPE_CMD);		/* Payload type. */
 	__msgb_push_u8(msgb, rqst->target_category);	/* Target category. */
-	__msgb_push_u8(msgb, rqst->target_id);		/* Target ID (out). */
-	__msgb_push_u8(msgb, 0x00);			/* Target ID (in). */
+	__msgb_push_u8(msgb, rqst->target_id);		/* Target ID. */
+	__msgb_push_u8(msgb, SSAM_SSH_TID_HOST);	/* Source ID. */
 	__msgb_push_u8(msgb, rqst->instance_id);	/* Instance ID. */
 	__msgb_push_u16(msgb, rqid);			/* Request ID. */
 	__msgb_push_u8(msgb, rqst->command_id);		/* Command ID. */
diff --git a/drivers/platform/surface/aggregator/ssh_request_layer.c b/drivers/platform/surface/aggregator/ssh_request_layer.c
index 69132976d297e..90634dcacabf2 100644
--- a/drivers/platform/surface/aggregator/ssh_request_layer.c
+++ b/drivers/platform/surface/aggregator/ssh_request_layer.c
@@ -920,13 +920,14 @@ static void ssh_rtl_rx_command(struct ssh_ptl *p, const struct ssam_span *data)
 	 * Check if the message was intended for us. If not, drop it.
 	 *
 	 * Note: We will need to change this to handle debug messages. On newer
-	 * generation devices, these seem to be sent to tid_out=0x03. We as
-	 * host can still receive them as they can be forwarded via an override
-	 * option on SAM, but doing so does not change tid_out=0x00.
+	 * generation devices, these seem to be sent to SSAM_SSH_TID_DEBUG. We
+	 * as host can still receive them as they can be forwarded via an
+	 * override option on SAM, but doing so does not change the target ID
+	 * to SSAM_SSH_TID_HOST.
 	 */
-	if (command->tid_out != 0x00) {
+	if (command->tid != SSAM_SSH_TID_HOST) {
 		rtl_warn(rtl, "rtl: dropping message not intended for us (tid = %#04x)\n",
-			 command->tid_out);
+			 command->tid);
 		return;
 	}
 
diff --git a/include/linux/surface_aggregator/controller.h b/include/linux/surface_aggregator/controller.h
index d11a1c6e3186a..8932bc0bae187 100644
--- a/include/linux/surface_aggregator/controller.h
+++ b/include/linux/surface_aggregator/controller.h
@@ -912,10 +912,10 @@ enum ssam_event_mask {
 	})
 
 #define SSAM_EVENT_REGISTRY_SAM	\
-	SSAM_EVENT_REGISTRY(SSAM_SSH_TC_SAM, 0x01, 0x0b, 0x0c)
+	SSAM_EVENT_REGISTRY(SSAM_SSH_TC_SAM, SSAM_SSH_TID_SAM, 0x0b, 0x0c)
 
 #define SSAM_EVENT_REGISTRY_KIP	\
-	SSAM_EVENT_REGISTRY(SSAM_SSH_TC_KIP, 0x02, 0x27, 0x28)
+	SSAM_EVENT_REGISTRY(SSAM_SSH_TC_KIP, SSAM_SSH_TID_KIP, 0x27, 0x28)
 
 #define SSAM_EVENT_REGISTRY_REG(tid)\
 	SSAM_EVENT_REGISTRY(SSAM_SSH_TC_REG, tid, 0x01, 0x02)
diff --git a/include/linux/surface_aggregator/serial_hub.h b/include/linux/surface_aggregator/serial_hub.h
index 45501b6e54e8a..5c4ae1a261831 100644
--- a/include/linux/surface_aggregator/serial_hub.h
+++ b/include/linux/surface_aggregator/serial_hub.h
@@ -83,23 +83,21 @@ enum ssh_payload_type {
 
 /**
  * struct ssh_command - Payload of a command-type frame.
- * @type:    The type of the payload. See &enum ssh_payload_type. Should be
- *           SSH_PLD_TYPE_CMD for this struct.
- * @tc:      Command target category.
- * @tid_out: Output target ID. Should be zero if this an incoming (EC to host)
- *           message.
- * @tid_in:  Input target ID. Should be zero if this is an outgoing (host to
- *           EC) message.
- * @iid:     Instance ID.
- * @rqid:    Request ID. Used to match requests with responses and differentiate
- *           between responses and events.
- * @cid:     Command ID.
+ * @type: The type of the payload. See &enum ssh_payload_type. Should be
+ *        SSH_PLD_TYPE_CMD for this struct.
+ * @tc:   Command target category.
+ * @tid:  Target ID. Indicates the target of the message.
+ * @sid:  Source ID. Indicates the source of the message.
+ * @iid:  Instance ID.
+ * @rqid: Request ID. Used to match requests with responses and differentiate
+ *        between responses and events.
+ * @cid:  Command ID.
  */
 struct ssh_command {
 	u8 type;
 	u8 tc;
-	u8 tid_out;
-	u8 tid_in;
+	u8 tid;
+	u8 sid;
 	u8 iid;
 	__le16 rqid;
 	u8 cid;
@@ -280,6 +278,22 @@ struct ssam_span {
 	size_t len;
 };
 
+/**
+ * enum ssam_ssh_tid - Target/source IDs for Serial Hub messages.
+ * @SSAM_SSH_TID_HOST:     We as the kernel Serial Hub driver.
+ * @SSAM_SSH_TID_SAM:      The Surface Aggregator EC.
+ * @SSAM_SSH_TID_KIP:      Keyboard and perihperal controller.
+ * @SSAM_SSH_TID_DEBUG:    Debug connector.
+ * @SSAM_SSH_TID_SURFLINK: SurfLink connector.
+ */
+enum ssam_ssh_tid {
+	SSAM_SSH_TID_HOST     = 0x00,
+	SSAM_SSH_TID_SAM      = 0x01,
+	SSAM_SSH_TID_KIP      = 0x02,
+	SSAM_SSH_TID_DEBUG    = 0x03,
+	SSAM_SSH_TID_SURFLINK = 0x04,
+};
+
 /*
  * Known SSH/EC target categories.
  *
-- 
2.39.2

From 5542a89eaf6142f28befee2602ebe89fb6afccf4 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:21 +0100
Subject: [PATCH] platform/surface: aggregator: Add target and source IDs to
 command trace events

Add command source and target IDs to trace events.

Tracing support for the Surface Aggregator driver was originally
implemented at a time when only two peers were known: Host and SAM. We
now know that there are at least five, with three actively being used
(Host, SAM, KIP; four with Debug if you want to count manually enabling
that interface). So it makes sense to also explicitly name the peers
involved when tracing.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-4-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/aggregator/trace.h | 73 +++++++++++++++++++--
 1 file changed, 67 insertions(+), 6 deletions(-)

diff --git a/drivers/platform/surface/aggregator/trace.h b/drivers/platform/surface/aggregator/trace.h
index 2a2c17771d014..55cc61bba1da6 100644
--- a/drivers/platform/surface/aggregator/trace.h
+++ b/drivers/platform/surface/aggregator/trace.h
@@ -96,6 +96,7 @@ TRACE_DEFINE_ENUM(SSAM_SSH_TC_POS);
 #define SSAM_SEQ_NOT_APPLICABLE		((u16)-1)
 #define SSAM_RQID_NOT_APPLICABLE	((u32)-1)
 #define SSAM_SSH_TC_NOT_APPLICABLE	0
+#define SSAM_SSH_TID_NOT_APPLICABLE	((u8)-1)
 
 #ifndef _SURFACE_AGGREGATOR_TRACE_HELPERS
 #define _SURFACE_AGGREGATOR_TRACE_HELPERS
@@ -150,12 +151,44 @@ static inline u32 ssam_trace_get_request_id(const struct ssh_packet *p)
 	return get_unaligned_le16(&p->data.ptr[SSH_MSGOFFSET_COMMAND(rqid)]);
 }
 
+/**
+ * ssam_trace_get_request_tid() - Read the packet's request target ID.
+ * @p: The packet.
+ *
+ * Return: Returns the packet's request target ID (TID) field if the packet
+ * represents a request with command data, or %SSAM_SSH_TID_NOT_APPLICABLE
+ * if not (e.g. flush request, control packet).
+ */
+static inline u32 ssam_trace_get_request_tid(const struct ssh_packet *p)
+{
+	if (!p->data.ptr || p->data.len < SSH_COMMAND_MESSAGE_LENGTH(0))
+		return SSAM_SSH_TID_NOT_APPLICABLE;
+
+	return get_unaligned_le16(&p->data.ptr[SSH_MSGOFFSET_COMMAND(tid)]);
+}
+
+/**
+ * ssam_trace_get_request_sid() - Read the packet's request source ID.
+ * @p: The packet.
+ *
+ * Return: Returns the packet's request source ID (SID) field if the packet
+ * represents a request with command data, or %SSAM_SSH_TID_NOT_APPLICABLE
+ * if not (e.g. flush request, control packet).
+ */
+static inline u32 ssam_trace_get_request_sid(const struct ssh_packet *p)
+{
+	if (!p->data.ptr || p->data.len < SSH_COMMAND_MESSAGE_LENGTH(0))
+		return SSAM_SSH_TID_NOT_APPLICABLE;
+
+	return get_unaligned_le16(&p->data.ptr[SSH_MSGOFFSET_COMMAND(sid)]);
+}
+
 /**
  * ssam_trace_get_request_tc() - Read the packet's request target category.
  * @p: The packet.
  *
  * Return: Returns the packet's request target category (TC) field if the
- * packet represents a request with command data, or %SSAM_TC_NOT_APPLICABLE
+ * packet represents a request with command data, or %SSAM_SSH_TC_NOT_APPLICABLE
  * if not (e.g. flush request, control packet).
  */
 static inline u32 ssam_trace_get_request_tc(const struct ssh_packet *p)
@@ -232,8 +265,18 @@ static inline u32 ssam_trace_get_request_tc(const struct ssh_packet *p)
 		{ SSAM_RQID_NOT_APPLICABLE,		"N/A" }		\
 	)
 
-#define ssam_show_ssh_tc(rqid)						\
-	__print_symbolic(rqid,						\
+#define ssam_show_ssh_tid(tid)						\
+	__print_symbolic(tid,						\
+		{ SSAM_SSH_TID_NOT_APPLICABLE,		"N/A"      },	\
+		{ SSAM_SSH_TID_HOST,			"Host"     },	\
+		{ SSAM_SSH_TID_SAM,			"SAM"      },	\
+		{ SSAM_SSH_TID_KIP,			"KIP"      },	\
+		{ SSAM_SSH_TID_DEBUG,			"Debug"    },	\
+		{ SSAM_SSH_TID_SURFLINK,		"SurfLink" }	\
+	)
+
+#define ssam_show_ssh_tc(tc)						\
+	__print_symbolic(tc,						\
 		{ SSAM_SSH_TC_NOT_APPLICABLE,		"N/A"  },	\
 		{ SSAM_SSH_TC_SAM,			"SAM"  },	\
 		{ SSAM_SSH_TC_BAT,			"BAT"  },	\
@@ -313,6 +356,8 @@ DECLARE_EVENT_CLASS(ssam_command_class,
 	TP_STRUCT__entry(
 		__field(u16, rqid)
 		__field(u16, len)
+		__field(u8, tid)
+		__field(u8, sid)
 		__field(u8, tc)
 		__field(u8, cid)
 		__field(u8, iid)
@@ -320,14 +365,18 @@ DECLARE_EVENT_CLASS(ssam_command_class,
 
 	TP_fast_assign(
 		__entry->rqid = get_unaligned_le16(&cmd->rqid);
+		__entry->tid = cmd->tid;
+		__entry->sid = cmd->sid;
 		__entry->tc = cmd->tc;
 		__entry->cid = cmd->cid;
 		__entry->iid = cmd->iid;
 		__entry->len = len;
 	),
 
-	TP_printk("rqid=%#06x, tc=%s, cid=%#04x, iid=%#04x, len=%u",
+	TP_printk("rqid=%#06x, tid=%s, sid=%s, tc=%s, cid=%#04x, iid=%#04x, len=%u",
 		__entry->rqid,
+		ssam_show_ssh_tid(__entry->tid),
+		ssam_show_ssh_tid(__entry->sid),
 		ssam_show_ssh_tc(__entry->tc),
 		__entry->cid,
 		__entry->iid,
@@ -430,6 +479,8 @@ DECLARE_EVENT_CLASS(ssam_request_class,
 		__field(u8, tc)
 		__field(u16, cid)
 		__field(u16, iid)
+		__field(u8, tid)
+		__field(u8, sid)
 	),
 
 	TP_fast_assign(
@@ -439,16 +490,20 @@ DECLARE_EVENT_CLASS(ssam_request_class,
 		__entry->state = READ_ONCE(request->state);
 		__entry->rqid = ssam_trace_get_request_id(p);
 		ssam_trace_ptr_uid(p, __entry->uid);
+		__entry->tid = ssam_trace_get_request_tid(p);
+		__entry->sid = ssam_trace_get_request_sid(p);
 		__entry->tc = ssam_trace_get_request_tc(p);
 		__entry->cid = ssam_trace_get_command_field_u8(p, cid);
 		__entry->iid = ssam_trace_get_command_field_u8(p, iid);
 	),
 
-	TP_printk("uid=%s, rqid=%s, ty=%s, sta=%s, tc=%s, cid=%s, iid=%s",
+	TP_printk("uid=%s, rqid=%s, ty=%s, sta=%s, tid=%s, sid=%s, tc=%s, cid=%s, iid=%s",
 		__entry->uid,
 		ssam_show_request_id(__entry->rqid),
 		ssam_show_request_type(__entry->state),
 		ssam_show_request_state(__entry->state),
+		ssam_show_ssh_tid(__entry->tid),
+		ssam_show_ssh_tid(__entry->sid),
 		ssam_show_ssh_tc(__entry->tc),
 		ssam_show_generic_u8_field(__entry->cid),
 		ssam_show_generic_u8_field(__entry->iid)
@@ -474,6 +529,8 @@ DECLARE_EVENT_CLASS(ssam_request_status_class,
 		__field(u8, tc)
 		__field(u16, cid)
 		__field(u16, iid)
+		__field(u8, tid)
+		__field(u8, sid)
 	),
 
 	TP_fast_assign(
@@ -484,16 +541,20 @@ DECLARE_EVENT_CLASS(ssam_request_status_class,
 		__entry->rqid = ssam_trace_get_request_id(p);
 		__entry->status = status;
 		ssam_trace_ptr_uid(p, __entry->uid);
+		__entry->tid = ssam_trace_get_request_tid(p);
+		__entry->sid = ssam_trace_get_request_sid(p);
 		__entry->tc = ssam_trace_get_request_tc(p);
 		__entry->cid = ssam_trace_get_command_field_u8(p, cid);
 		__entry->iid = ssam_trace_get_command_field_u8(p, iid);
 	),
 
-	TP_printk("uid=%s, rqid=%s, ty=%s, sta=%s, tc=%s, cid=%s, iid=%s, status=%d",
+	TP_printk("uid=%s, rqid=%s, ty=%s, sta=%s, tid=%s, sid=%s, tc=%s, cid=%s, iid=%s, status=%d",
 		__entry->uid,
 		ssam_show_request_id(__entry->rqid),
 		ssam_show_request_type(__entry->state),
 		ssam_show_request_state(__entry->state),
+		ssam_show_ssh_tid(__entry->tid),
+		ssam_show_ssh_tid(__entry->sid),
 		ssam_show_ssh_tc(__entry->tc),
 		ssam_show_generic_u8_field(__entry->cid),
 		ssam_show_generic_u8_field(__entry->iid),
-- 
2.39.2

From 45414ba7fa9d6ce1889626710e24f71193c4e4d5 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:22 +0100
Subject: [PATCH] platform/surface: aggregator_hub: Use target-ID enum instead
 of hard-coding values

Instead of hard-coding the target ID, use the respective enum
ssam_ssh_tid value.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-5-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_aggregator_hub.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/platform/surface/surface_aggregator_hub.c b/drivers/platform/surface/surface_aggregator_hub.c
index 43061514be382..62f27cdb6ca8f 100644
--- a/drivers/platform/surface/surface_aggregator_hub.c
+++ b/drivers/platform/surface/surface_aggregator_hub.c
@@ -214,7 +214,7 @@ static void ssam_hub_remove(struct ssam_device *sdev)
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_bas_query_opmode, u8, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x0d,
 	.instance_id     = 0x00,
 });
@@ -292,7 +292,7 @@ static const struct ssam_hub_desc base_hub = {
 
 SSAM_DEFINE_SYNC_REQUEST_R(__ssam_kip_query_state, u8, {
 	.target_category = SSAM_SSH_TC_KIP,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x2c,
 	.instance_id     = 0x00,
 });
-- 
2.39.2

From e436e81159bc3941a3d744143b400a859afbef11 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:23 +0100
Subject: [PATCH] platform/surface: aggregator_tabletsw: Use target-ID enum
 instead of hard-coding values

Instead of hard-coding the target ID, use the respective enum
ssam_ssh_tid value.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-6-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_aggregator_tabletsw.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/platform/surface/surface_aggregator_tabletsw.c b/drivers/platform/surface/surface_aggregator_tabletsw.c
index 27d95a6a78513..bd8cd453c393a 100644
--- a/drivers/platform/surface/surface_aggregator_tabletsw.c
+++ b/drivers/platform/surface/surface_aggregator_tabletsw.c
@@ -247,7 +247,7 @@ static bool ssam_kip_cover_state_is_tablet_mode(struct ssam_tablet_sw *sw, u32 s
 
 SSAM_DEFINE_SYNC_REQUEST_R(__ssam_kip_get_cover_state, u8, {
 	.target_category = SSAM_SSH_TC_KIP,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x1d,
 	.instance_id     = 0x00,
 });
@@ -371,7 +371,7 @@ static int ssam_pos_get_sources_list(struct ssam_tablet_sw *sw, struct ssam_sour
 	int status;
 
 	rqst.target_category = SSAM_SSH_TC_POS;
-	rqst.target_id = 0x01;
+	rqst.target_id = SSAM_SSH_TID_SAM;
 	rqst.command_id = 0x01;
 	rqst.instance_id = 0x00;
 	rqst.flags = SSAM_REQUEST_HAS_RESPONSE;
@@ -430,7 +430,7 @@ static int ssam_pos_get_source(struct ssam_tablet_sw *sw, u32 *source_id)
 
 SSAM_DEFINE_SYNC_REQUEST_WR(__ssam_pos_get_posture_for_source, __le32, __le32, {
 	.target_category = SSAM_SSH_TC_POS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x02,
 	.instance_id     = 0x00,
 });
-- 
2.39.2

From d1b11ceedce2dd2ed1b4e3601d2ccaa873d9b4ba Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:24 +0100
Subject: [PATCH] platform/surface: dtx: Use target-ID enum instead of
 hard-coding values

Instead of hard-coding the target ID, use the respective enum
ssam_ssh_tid value.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-7-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_dtx.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/platform/surface/surface_dtx.c b/drivers/platform/surface/surface_dtx.c
index ed36944467f9f..0de76a784a35f 100644
--- a/drivers/platform/surface/surface_dtx.c
+++ b/drivers/platform/surface/surface_dtx.c
@@ -71,63 +71,63 @@ static_assert(sizeof(struct ssam_bas_base_info) == 2);
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_lock, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x06,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_unlock, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x07,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_request, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x08,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_confirm, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x09,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_heartbeat, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x0a,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_N(ssam_bas_latch_cancel, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x0b,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_bas_get_base, struct ssam_bas_base_info, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x0c,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_bas_get_device_mode, u8, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x0d,
 	.instance_id     = 0x00,
 });
 
 SSAM_DEFINE_SYNC_REQUEST_R(ssam_bas_get_latch_status, u8, {
 	.target_category = SSAM_SSH_TC_BAS,
-	.target_id       = 0x01,
+	.target_id       = SSAM_SSH_TID_SAM,
 	.command_id      = 0x11,
 	.instance_id     = 0x00,
 });
-- 
2.39.2

From 392ce2f40092dfc98134a81d30185e3679489d27 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:25 +0100
Subject: [PATCH] HID: surface-hid: Use target-ID enum instead of hard-coding
 values

Instead of hard-coding the target ID, use the respective enum
ssam_ssh_tid value.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-8-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/hid/surface-hid/surface_kbd.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/hid/surface-hid/surface_kbd.c b/drivers/hid/surface-hid/surface_kbd.c
index 0635341bc5174..42933bf3e925f 100644
--- a/drivers/hid/surface-hid/surface_kbd.c
+++ b/drivers/hid/surface-hid/surface_kbd.c
@@ -250,7 +250,7 @@ static int surface_kbd_probe(struct platform_device *pdev)
 
 	shid->uid.domain = SSAM_DOMAIN_SERIALHUB;
 	shid->uid.category = SSAM_SSH_TC_KBD;
-	shid->uid.target = 2;
+	shid->uid.target = SSAM_SSH_TID_KIP;
 	shid->uid.instance = 0;
 	shid->uid.function = 0;
 
-- 
2.39.2

From 5d06854e48e40547414d309f9b298ca8d5390fe6 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:26 +0100
Subject: [PATCH] platform/surface: aggregator: Enforce use of target-ID enum
 in device ID macros

Similar to the target category (TC), the target ID (TID) can be one
value out of a small number of choices, given in enum ssam_ssh_tid.

In the device ID macros, SSAM_SDEV() and SSAM_VDEV() we already use text
expansion to, both, remove some textual clutter for the target category
values and enforce that the value belongs to the known set. Now that we
know the names for the target IDs, use the same trick for them as well.

Also rename the SSAM_ANY_x macros to SSAM_SSH_x_ANY to better fit in.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Acked-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Link: https://lore.kernel.org/r/20221202223327.690880-9-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/hid/surface-hid/surface_hid.c         |  2 +-
 .../platform/surface/surface_aggregator_hub.c |  4 +-
 .../surface/surface_aggregator_tabletsw.c     |  4 +-
 drivers/platform/surface/surface_dtx.c        |  2 +-
 .../surface/surface_platform_profile.c        |  2 +-
 drivers/power/supply/surface_battery.c        |  4 +-
 drivers/power/supply/surface_charger.c        |  2 +-
 include/linux/surface_aggregator/device.h     | 50 +++++++++----------
 8 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/drivers/hid/surface-hid/surface_hid.c b/drivers/hid/surface-hid/surface_hid.c
index d4aa8c81903ae..aa80d83a83d1b 100644
--- a/drivers/hid/surface-hid/surface_hid.c
+++ b/drivers/hid/surface-hid/surface_hid.c
@@ -230,7 +230,7 @@ static void surface_hid_remove(struct ssam_device *sdev)
 }
 
 static const struct ssam_device_id surface_hid_match[] = {
-	{ SSAM_SDEV(HID, SSAM_ANY_TID, SSAM_ANY_IID, 0x00) },
+	{ SSAM_SDEV(HID, ANY, SSAM_SSH_IID_ANY, 0x00) },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, surface_hid_match);
diff --git a/drivers/platform/surface/surface_aggregator_hub.c b/drivers/platform/surface/surface_aggregator_hub.c
index 62f27cdb6ca8f..6abd1efe20883 100644
--- a/drivers/platform/surface/surface_aggregator_hub.c
+++ b/drivers/platform/surface/surface_aggregator_hub.c
@@ -348,8 +348,8 @@ static const struct ssam_hub_desc kip_hub = {
 /* -- Driver registration. -------------------------------------------------- */
 
 static const struct ssam_device_id ssam_hub_match[] = {
-	{ SSAM_VDEV(HUB, 0x01, SSAM_SSH_TC_KIP, 0x00), (unsigned long)&kip_hub  },
-	{ SSAM_VDEV(HUB, 0x02, SSAM_SSH_TC_BAS, 0x00), (unsigned long)&base_hub },
+	{ SSAM_VDEV(HUB, SAM, SSAM_SSH_TC_KIP, 0x00), (unsigned long)&kip_hub  },
+	{ SSAM_VDEV(HUB, KIP, SSAM_SSH_TC_BAS, 0x00), (unsigned long)&base_hub },
 	{ }
 };
 MODULE_DEVICE_TABLE(ssam, ssam_hub_match);
diff --git a/drivers/platform/surface/surface_aggregator_tabletsw.c b/drivers/platform/surface/surface_aggregator_tabletsw.c
index bd8cd453c393a..6147aa8879391 100644
--- a/drivers/platform/surface/surface_aggregator_tabletsw.c
+++ b/drivers/platform/surface/surface_aggregator_tabletsw.c
@@ -510,8 +510,8 @@ static const struct ssam_tablet_sw_desc ssam_pos_sw_desc = {
 /* -- Driver registration. -------------------------------------------------- */
 
 static const struct ssam_device_id ssam_tablet_sw_match[] = {
-	{ SSAM_SDEV(KIP, 0x01, 0x00, 0x01), (unsigned long)&ssam_kip_sw_desc },
-	{ SSAM_SDEV(POS, 0x01, 0x00, 0x01), (unsigned long)&ssam_pos_sw_desc },
+	{ SSAM_SDEV(KIP, SAM, 0x00, 0x01), (unsigned long)&ssam_kip_sw_desc },
+	{ SSAM_SDEV(POS, SAM, 0x00, 0x01), (unsigned long)&ssam_pos_sw_desc },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, ssam_tablet_sw_match);
diff --git a/drivers/platform/surface/surface_dtx.c b/drivers/platform/surface/surface_dtx.c
index 0de76a784a35f..30cbde278c599 100644
--- a/drivers/platform/surface/surface_dtx.c
+++ b/drivers/platform/surface/surface_dtx.c
@@ -1214,7 +1214,7 @@ static void surface_dtx_ssam_remove(struct ssam_device *sdev)
 }
 
 static const struct ssam_device_id surface_dtx_ssam_match[] = {
-	{ SSAM_SDEV(BAS, 0x01, 0x00, 0x00) },
+	{ SSAM_SDEV(BAS, SAM, 0x00, 0x00) },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, surface_dtx_ssam_match);
diff --git a/drivers/platform/surface/surface_platform_profile.c b/drivers/platform/surface/surface_platform_profile.c
index fbf2e11fd6ce7..f433a13c3689a 100644
--- a/drivers/platform/surface/surface_platform_profile.c
+++ b/drivers/platform/surface/surface_platform_profile.c
@@ -169,7 +169,7 @@ static void surface_platform_profile_remove(struct ssam_device *sdev)
 }
 
 static const struct ssam_device_id ssam_platform_profile_match[] = {
-	{ SSAM_SDEV(TMP, 0x01, 0x00, 0x01) },
+	{ SSAM_SDEV(TMP, SAM, 0x00, 0x01) },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, ssam_platform_profile_match);
diff --git a/drivers/power/supply/surface_battery.c b/drivers/power/supply/surface_battery.c
index 540707882bb0a..19d2f8834e56d 100644
--- a/drivers/power/supply/surface_battery.c
+++ b/drivers/power/supply/surface_battery.c
@@ -852,8 +852,8 @@ static const struct spwr_psy_properties spwr_psy_props_bat2_sb3 = {
 };
 
 static const struct ssam_device_id surface_battery_match[] = {
-	{ SSAM_SDEV(BAT, 0x01, 0x01, 0x00), (unsigned long)&spwr_psy_props_bat1     },
-	{ SSAM_SDEV(BAT, 0x02, 0x01, 0x00), (unsigned long)&spwr_psy_props_bat2_sb3 },
+	{ SSAM_SDEV(BAT, SAM, 0x01, 0x00), (unsigned long)&spwr_psy_props_bat1     },
+	{ SSAM_SDEV(BAT, KIP, 0x01, 0x00), (unsigned long)&spwr_psy_props_bat2_sb3 },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, surface_battery_match);
diff --git a/drivers/power/supply/surface_charger.c b/drivers/power/supply/surface_charger.c
index 59182d55742d0..cabdd8da12d08 100644
--- a/drivers/power/supply/surface_charger.c
+++ b/drivers/power/supply/surface_charger.c
@@ -260,7 +260,7 @@ static const struct spwr_psy_properties spwr_psy_props_adp1 = {
 };
 
 static const struct ssam_device_id surface_ac_match[] = {
-	{ SSAM_SDEV(BAT, 0x01, 0x01, 0x01), (unsigned long)&spwr_psy_props_adp1 },
+	{ SSAM_SDEV(BAT, SAM, 0x01, 0x01), (unsigned long)&spwr_psy_props_adp1 },
 	{ },
 };
 MODULE_DEVICE_TABLE(ssam, surface_ac_match);
diff --git a/include/linux/surface_aggregator/device.h b/include/linux/surface_aggregator/device.h
index 46c45d1b63682..4da20b7a0ee5e 100644
--- a/include/linux/surface_aggregator/device.h
+++ b/include/linux/surface_aggregator/device.h
@@ -68,9 +68,9 @@ struct ssam_device_uid {
  * match_flags member of the device ID structure. Do not use them directly
  * with struct ssam_device_id or struct ssam_device_uid.
  */
-#define SSAM_ANY_TID		0xffff
-#define SSAM_ANY_IID		0xffff
-#define SSAM_ANY_FUN		0xffff
+#define SSAM_SSH_TID_ANY	0xffff
+#define SSAM_SSH_IID_ANY	0xffff
+#define SSAM_SSH_FUN_ANY	0xffff
 
 /**
  * SSAM_DEVICE() - Initialize a &struct ssam_device_id with the given
@@ -83,25 +83,25 @@ struct ssam_device_uid {
  *
  * Initializes a &struct ssam_device_id with the given parameters. See &struct
  * ssam_device_uid for details regarding the parameters. The special values
- * %SSAM_ANY_TID, %SSAM_ANY_IID, and %SSAM_ANY_FUN can be used to specify that
+ * %SSAM_SSH_TID_ANY, %SSAM_SSH_IID_ANY, and %SSAM_SSH_FUN_ANY can be used to specify that
  * matching should ignore target ID, instance ID, and/or sub-function,
  * respectively. This macro initializes the ``match_flags`` field based on the
  * given parameters.
  *
  * Note: The parameters @d and @cat must be valid &u8 values, the parameters
- * @tid, @iid, and @fun must be either valid &u8 values or %SSAM_ANY_TID,
- * %SSAM_ANY_IID, or %SSAM_ANY_FUN, respectively. Other non-&u8 values are not
+ * @tid, @iid, and @fun must be either valid &u8 values or %SSAM_SSH_TID_ANY,
+ * %SSAM_SSH_IID_ANY, or %SSAM_SSH_FUN_ANY, respectively. Other non-&u8 values are not
  * allowed.
  */
 #define SSAM_DEVICE(d, cat, tid, iid, fun)					\
-	.match_flags = (((tid) != SSAM_ANY_TID) ? SSAM_MATCH_TARGET : 0)	\
-		     | (((iid) != SSAM_ANY_IID) ? SSAM_MATCH_INSTANCE : 0)	\
-		     | (((fun) != SSAM_ANY_FUN) ? SSAM_MATCH_FUNCTION : 0),	\
+	.match_flags = (((tid) != SSAM_SSH_TID_ANY) ? SSAM_MATCH_TARGET : 0)	\
+		     | (((iid) != SSAM_SSH_IID_ANY) ? SSAM_MATCH_INSTANCE : 0)	\
+		     | (((fun) != SSAM_SSH_FUN_ANY) ? SSAM_MATCH_FUNCTION : 0),	\
 	.domain   = d,								\
 	.category = cat,							\
-	.target   = __builtin_choose_expr((tid) != SSAM_ANY_TID, (tid), 0),	\
-	.instance = __builtin_choose_expr((iid) != SSAM_ANY_IID, (iid), 0),	\
-	.function = __builtin_choose_expr((fun) != SSAM_ANY_FUN, (fun), 0)
+	.target   = __builtin_choose_expr((tid) != SSAM_SSH_TID_ANY, (tid), 0),	\
+	.instance = __builtin_choose_expr((iid) != SSAM_SSH_IID_ANY, (iid), 0),	\
+	.function = __builtin_choose_expr((fun) != SSAM_SSH_FUN_ANY, (fun), 0)
 
 /**
  * SSAM_VDEV() - Initialize a &struct ssam_device_id as virtual device with
@@ -113,18 +113,18 @@ struct ssam_device_uid {
  *
  * Initializes a &struct ssam_device_id with the given parameters in the
  * virtual domain. See &struct ssam_device_uid for details regarding the
- * parameters. The special values %SSAM_ANY_TID, %SSAM_ANY_IID, and
- * %SSAM_ANY_FUN can be used to specify that matching should ignore target ID,
+ * parameters. The special values %SSAM_SSH_TID_ANY, %SSAM_SSH_IID_ANY, and
+ * %SSAM_SSH_FUN_ANY can be used to specify that matching should ignore target ID,
  * instance ID, and/or sub-function, respectively. This macro initializes the
  * ``match_flags`` field based on the given parameters.
  *
  * Note: The parameter @cat must be a valid &u8 value, the parameters @tid,
- * @iid, and @fun must be either valid &u8 values or %SSAM_ANY_TID,
- * %SSAM_ANY_IID, or %SSAM_ANY_FUN, respectively. Other non-&u8 values are not
+ * @iid, and @fun must be either valid &u8 values or %SSAM_SSH_TID_ANY,
+ * %SSAM_SSH_IID_ANY, or %SSAM_SSH_FUN_ANY, respectively. Other non-&u8 values are not
  * allowed.
  */
 #define SSAM_VDEV(cat, tid, iid, fun) \
-	SSAM_DEVICE(SSAM_DOMAIN_VIRTUAL, SSAM_VIRTUAL_TC_##cat, tid, iid, fun)
+	SSAM_DEVICE(SSAM_DOMAIN_VIRTUAL, SSAM_VIRTUAL_TC_##cat, SSAM_SSH_TID_##tid, iid, fun)
 
 /**
  * SSAM_SDEV() - Initialize a &struct ssam_device_id as physical SSH device
@@ -136,18 +136,18 @@ struct ssam_device_uid {
  *
  * Initializes a &struct ssam_device_id with the given parameters in the SSH
  * domain. See &struct ssam_device_uid for details regarding the parameters.
- * The special values %SSAM_ANY_TID, %SSAM_ANY_IID, and %SSAM_ANY_FUN can be
- * used to specify that matching should ignore target ID, instance ID, and/or
- * sub-function, respectively. This macro initializes the ``match_flags``
- * field based on the given parameters.
+ * The special values %SSAM_SSH_TID_ANY, %SSAM_SSH_IID_ANY, and
+ * %SSAM_SSH_FUN_ANY can be used to specify that matching should ignore target
+ * ID, instance ID, and/or sub-function, respectively. This macro initializes
+ * the ``match_flags`` field based on the given parameters.
  *
  * Note: The parameter @cat must be a valid &u8 value, the parameters @tid,
- * @iid, and @fun must be either valid &u8 values or %SSAM_ANY_TID,
- * %SSAM_ANY_IID, or %SSAM_ANY_FUN, respectively. Other non-&u8 values are not
- * allowed.
+ * @iid, and @fun must be either valid &u8 values or %SSAM_SSH_TID_ANY,
+ * %SSAM_SSH_IID_ANY, or %SSAM_SSH_FUN_ANY, respectively. Other non-&u8 values
+ * are not allowed.
  */
 #define SSAM_SDEV(cat, tid, iid, fun) \
-	SSAM_DEVICE(SSAM_DOMAIN_SERIALHUB, SSAM_SSH_TC_##cat, tid, iid, fun)
+	SSAM_DEVICE(SSAM_DOMAIN_SERIALHUB, SSAM_SSH_TC_##cat, SSAM_SSH_TID_##tid, iid, fun)
 
 /*
  * enum ssam_device_flags - Flags for SSAM client devices.
-- 
2.39.2

From c2d159b625d24c5df3438a7e1080bc13cb68c713 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 2 Dec 2022 23:33:27 +0100
Subject: [PATCH] platform/surface: aggregator_registry: Fix target-ID of
 base-hub

The target ID of the base hub is currently set to KIP (keyboard/
peripherals). However, even though it manages such devices with the KIP
target ID, the base hub itself is actually accessed via the SAM target
ID. So set it accordingly.

Note that the target ID of the hub can be chosen arbitrarily and does
not directly correspond to any physical or virtual component of the EC.
This change is only a code improvement intended for consistency and
clarity, it does not fix an actual bug.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221202223327.690880-10-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_aggregator_hub.c      | 2 +-
 drivers/platform/surface/surface_aggregator_registry.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/platform/surface/surface_aggregator_hub.c b/drivers/platform/surface/surface_aggregator_hub.c
index 6abd1efe20883..8b8b80228c147 100644
--- a/drivers/platform/surface/surface_aggregator_hub.c
+++ b/drivers/platform/surface/surface_aggregator_hub.c
@@ -349,7 +349,7 @@ static const struct ssam_hub_desc kip_hub = {
 
 static const struct ssam_device_id ssam_hub_match[] = {
 	{ SSAM_VDEV(HUB, SAM, SSAM_SSH_TC_KIP, 0x00), (unsigned long)&kip_hub  },
-	{ SSAM_VDEV(HUB, KIP, SSAM_SSH_TC_BAS, 0x00), (unsigned long)&base_hub },
+	{ SSAM_VDEV(HUB, SAM, SSAM_SSH_TC_BAS, 0x00), (unsigned long)&base_hub },
 	{ }
 };
 MODULE_DEVICE_TABLE(ssam, ssam_hub_match);
diff --git a/drivers/platform/surface/surface_aggregator_registry.c b/drivers/platform/surface/surface_aggregator_registry.c
index 023f126121d7d..296f72d52e6a6 100644
--- a/drivers/platform/surface/surface_aggregator_registry.c
+++ b/drivers/platform/surface/surface_aggregator_registry.c
@@ -46,7 +46,7 @@ static const struct software_node ssam_node_hub_kip = {
 
 /* Base device hub (devices attached to Surface Book 3 base). */
 static const struct software_node ssam_node_hub_base = {
-	.name = "ssam:00:00:02:11:00",
+	.name = "ssam:00:00:01:11:00",
 	.parent = &ssam_node_root,
 };
 
-- 
2.39.2

From d37d589914fbc1c54ae9e28fb6261e16346b474e Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Tue, 20 Dec 2022 18:56:08 +0100
Subject: [PATCH] platform/surface: aggregator: Rename top-level request
 functions to avoid ambiguities

We currently have a struct ssam_request_sync and a function
ssam_request_sync(). While this is valid C, there are some downsides to
it.

One of these is that current Sphinx versions (>= 3.0) cannot
disambiguate between the two (see disucssion and pull request linked
below). It instead emits a "WARNING: Duplicate C declaration" and links
for the struct and function in the resulting documentation link to the
same entry (i.e. both to either function or struct documentation)
instead of their respective own entries.

While we could just ignore that and wait for a fix, there's also a point
to be made that the current naming can be somewhat confusing when
searching (e.g. via grep) or trying to understand the levels of
abstraction at play:

We currently have struct ssam_request_sync and associated functions
ssam_request_sync_[alloc|free|init|wait|...]() operating on this struct.
However, function ssam_request_sync() is one abstraction level above
this. Similarly, ssam_request_sync_with_buffer() is not a function
operating on struct ssam_request_sync, but rather a sibling to
ssam_request_sync(), both using the struct under the hood.

Therefore, rename the top level request functions:

  ssam_request_sync() -> ssam_request_do_sync()
  ssam_request_sync_with_buffer() -> ssam_request_do_sync_with_buffer()
  ssam_request_sync_onstack() -> ssam_request_do_sync_onstack()

Link: https://lore.kernel.org/all/085e0ada65c11da9303d07e70c510dc45f21315b.1656756450.git.mchehab@kernel.org/
Link: https://github.com/sphinx-doc/sphinx/pull/8313
Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20221220175608.1436273-2-luzmaximilian@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 .../driver-api/surface_aggregator/client.rst  |  8 +--
 drivers/hid/surface-hid/surface_hid.c         |  6 +-
 drivers/hid/surface-hid/surface_kbd.c         |  6 +-
 drivers/platform/surface/aggregator/bus.c     |  6 +-
 .../platform/surface/aggregator/controller.c  | 32 +++++------
 .../platform/surface/surface_acpi_notify.c    |  2 +-
 .../surface/surface_aggregator_cdev.c         |  6 +-
 .../surface/surface_aggregator_tabletsw.c     |  2 +-
 include/linux/surface_aggregator/controller.h | 56 +++++++++----------
 include/linux/surface_aggregator/device.h     |  8 +--
 10 files changed, 66 insertions(+), 66 deletions(-)

diff --git a/Documentation/driver-api/surface_aggregator/client.rst b/Documentation/driver-api/surface_aggregator/client.rst
index 9d7411223a848..e100ab0a24cc4 100644
--- a/Documentation/driver-api/surface_aggregator/client.rst
+++ b/Documentation/driver-api/surface_aggregator/client.rst
@@ -19,7 +19,7 @@
 .. |ssam_notifier_unregister| replace:: :c:func:`ssam_notifier_unregister`
 .. |ssam_device_notifier_register| replace:: :c:func:`ssam_device_notifier_register`
 .. |ssam_device_notifier_unregister| replace:: :c:func:`ssam_device_notifier_unregister`
-.. |ssam_request_sync| replace:: :c:func:`ssam_request_sync`
+.. |ssam_request_do_sync| replace:: :c:func:`ssam_request_do_sync`
 .. |ssam_event_mask| replace:: :c:type:`enum ssam_event_mask <ssam_event_mask>`
 
 
@@ -209,12 +209,12 @@ data received from it is converted from little-endian to host endianness.
             * with the SSAM_REQUEST_HAS_RESPONSE flag set in the specification
             * above.
             */
-           status = ssam_request_sync(ctrl, &rqst, &resp);
+           status = ssam_request_do_sync(ctrl, &rqst, &resp);
 
            /*
             * Alternatively use
             *
-            *   ssam_request_sync_onstack(ctrl, &rqst, &resp, sizeof(arg_le));
+            *   ssam_request_do_sync_onstack(ctrl, &rqst, &resp, sizeof(arg_le));
             *
             * to perform the request, allocating the message buffer directly
             * on the stack as opposed to allocation via kzalloc().
@@ -230,7 +230,7 @@ data received from it is converted from little-endian to host endianness.
            return status;
    }
 
-Note that |ssam_request_sync| in its essence is a wrapper over lower-level
+Note that |ssam_request_do_sync| in its essence is a wrapper over lower-level
 request primitives, which may also be used to perform requests. Refer to its
 implementation and documentation for more details.
 
diff --git a/drivers/hid/surface-hid/surface_hid.c b/drivers/hid/surface-hid/surface_hid.c
index aa80d83a83d1b..61e5814b0ad7d 100644
--- a/drivers/hid/surface-hid/surface_hid.c
+++ b/drivers/hid/surface-hid/surface_hid.c
@@ -80,7 +80,7 @@ static int ssam_hid_get_descriptor(struct surface_hid_device *shid, u8 entry, u8
 
 		rsp.length = 0;
 
-		status = ssam_retry(ssam_request_sync_onstack, shid->ctrl, &rqst, &rsp,
+		status = ssam_retry(ssam_request_do_sync_onstack, shid->ctrl, &rqst, &rsp,
 				    sizeof(*slice));
 		if (status)
 			return status;
@@ -131,7 +131,7 @@ static int ssam_hid_set_raw_report(struct surface_hid_device *shid, u8 rprt_id,
 
 	buf[0] = rprt_id;
 
-	return ssam_retry(ssam_request_sync, shid->ctrl, &rqst, NULL);
+	return ssam_retry(ssam_request_do_sync, shid->ctrl, &rqst, NULL);
 }
 
 static int ssam_hid_get_raw_report(struct surface_hid_device *shid, u8 rprt_id, u8 *buf, size_t len)
@@ -151,7 +151,7 @@ static int ssam_hid_get_raw_report(struct surface_hid_device *shid, u8 rprt_id,
 	rsp.length = 0;
 	rsp.pointer = buf;
 
-	return ssam_retry(ssam_request_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(rprt_id));
+	return ssam_retry(ssam_request_do_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(rprt_id));
 }
 
 static u32 ssam_hid_event_fn(struct ssam_event_notifier *nf, const struct ssam_event *event)
diff --git a/drivers/hid/surface-hid/surface_kbd.c b/drivers/hid/surface-hid/surface_kbd.c
index 42933bf3e925f..4fbce201db6a1 100644
--- a/drivers/hid/surface-hid/surface_kbd.c
+++ b/drivers/hid/surface-hid/surface_kbd.c
@@ -49,7 +49,7 @@ static int ssam_kbd_get_descriptor(struct surface_hid_device *shid, u8 entry, u8
 	rsp.length = 0;
 	rsp.pointer = buf;
 
-	status = ssam_retry(ssam_request_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(entry));
+	status = ssam_retry(ssam_request_do_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(entry));
 	if (status)
 		return status;
 
@@ -75,7 +75,7 @@ static int ssam_kbd_set_caps_led(struct surface_hid_device *shid, bool value)
 	rqst.length = sizeof(value_u8);
 	rqst.payload = &value_u8;
 
-	return ssam_retry(ssam_request_sync_onstack, shid->ctrl, &rqst, NULL, sizeof(value_u8));
+	return ssam_retry(ssam_request_do_sync_onstack, shid->ctrl, &rqst, NULL, sizeof(value_u8));
 }
 
 static int ssam_kbd_get_feature_report(struct surface_hid_device *shid, u8 *buf, size_t len)
@@ -97,7 +97,7 @@ static int ssam_kbd_get_feature_report(struct surface_hid_device *shid, u8 *buf,
 	rsp.length = 0;
 	rsp.pointer = buf;
 
-	status = ssam_retry(ssam_request_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(payload));
+	status = ssam_retry(ssam_request_do_sync_onstack, shid->ctrl, &rqst, &rsp, sizeof(payload));
 	if (status)
 		return status;
 
diff --git a/drivers/platform/surface/aggregator/bus.c b/drivers/platform/surface/aggregator/bus.c
index de539938896e2..7004eb4a63a28 100644
--- a/drivers/platform/surface/aggregator/bus.c
+++ b/drivers/platform/surface/aggregator/bus.c
@@ -136,9 +136,9 @@ int ssam_device_add(struct ssam_device *sdev)
 	 * is always valid and can be used for requests as long as the client
 	 * device we add here is registered as child under it. This essentially
 	 * guarantees that the client driver can always expect the preconditions
-	 * for functions like ssam_request_sync (controller has to be started
-	 * and is not suspended) to hold and thus does not have to check for
-	 * them.
+	 * for functions like ssam_request_do_sync() (controller has to be
+	 * started and is not suspended) to hold and thus does not have to check
+	 * for them.
 	 *
 	 * Note that for this to work, the controller has to be a parent device.
 	 * If it is not a direct parent, care has to be taken that the device is
diff --git a/drivers/platform/surface/aggregator/controller.c b/drivers/platform/surface/aggregator/controller.c
index 2c99f51ccd4ec..535581c0471c5 100644
--- a/drivers/platform/surface/aggregator/controller.c
+++ b/drivers/platform/surface/aggregator/controller.c
@@ -1674,7 +1674,7 @@ int ssam_request_sync_submit(struct ssam_controller *ctrl,
 EXPORT_SYMBOL_GPL(ssam_request_sync_submit);
 
 /**
- * ssam_request_sync() - Execute a synchronous request.
+ * ssam_request_do_sync() - Execute a synchronous request.
  * @ctrl: The controller via which the request will be submitted.
  * @spec: The request specification and payload.
  * @rsp:  The response buffer.
@@ -1686,9 +1686,9 @@ EXPORT_SYMBOL_GPL(ssam_request_sync_submit);
  *
  * Return: Returns the status of the request or any failure during setup.
  */
-int ssam_request_sync(struct ssam_controller *ctrl,
-		      const struct ssam_request *spec,
-		      struct ssam_response *rsp)
+int ssam_request_do_sync(struct ssam_controller *ctrl,
+			 const struct ssam_request *spec,
+			 struct ssam_response *rsp)
 {
 	struct ssam_request_sync *rqst;
 	struct ssam_span buf;
@@ -1722,10 +1722,10 @@ int ssam_request_sync(struct ssam_controller *ctrl,
 	ssam_request_sync_free(rqst);
 	return status;
 }
-EXPORT_SYMBOL_GPL(ssam_request_sync);
+EXPORT_SYMBOL_GPL(ssam_request_do_sync);
 
 /**
- * ssam_request_sync_with_buffer() - Execute a synchronous request with the
+ * ssam_request_do_sync_with_buffer() - Execute a synchronous request with the
  * provided buffer as back-end for the message buffer.
  * @ctrl: The controller via which the request will be submitted.
  * @spec: The request specification and payload.
@@ -1738,17 +1738,17 @@ EXPORT_SYMBOL_GPL(ssam_request_sync);
  * SSH_COMMAND_MESSAGE_LENGTH() macro can be used to compute the required
  * message buffer size.
  *
- * This function does essentially the same as ssam_request_sync(), but instead
- * of dynamically allocating the request and message data buffer, it uses the
- * provided message data buffer and stores the (small) request struct on the
- * heap.
+ * This function does essentially the same as ssam_request_do_sync(), but
+ * instead of dynamically allocating the request and message data buffer, it
+ * uses the provided message data buffer and stores the (small) request struct
+ * on the heap.
  *
  * Return: Returns the status of the request or any failure during setup.
  */
-int ssam_request_sync_with_buffer(struct ssam_controller *ctrl,
-				  const struct ssam_request *spec,
-				  struct ssam_response *rsp,
-				  struct ssam_span *buf)
+int ssam_request_do_sync_with_buffer(struct ssam_controller *ctrl,
+				     const struct ssam_request *spec,
+				     struct ssam_response *rsp,
+				     struct ssam_span *buf)
 {
 	struct ssam_request_sync rqst;
 	ssize_t len;
@@ -1772,7 +1772,7 @@ int ssam_request_sync_with_buffer(struct ssam_controller *ctrl,
 
 	return status;
 }
-EXPORT_SYMBOL_GPL(ssam_request_sync_with_buffer);
+EXPORT_SYMBOL_GPL(ssam_request_do_sync_with_buffer);
 
 
 /* -- Internal SAM requests. ------------------------------------------------ */
@@ -1864,7 +1864,7 @@ static int __ssam_ssh_event_request(struct ssam_controller *ctrl,
 	result.length = 0;
 	result.pointer = &buf;
 
-	status = ssam_retry(ssam_request_sync_onstack, ctrl, &rqst, &result,
+	status = ssam_retry(ssam_request_do_sync_onstack, ctrl, &rqst, &result,
 			    sizeof(params));
 
 	return status < 0 ? status : buf;
diff --git a/drivers/platform/surface/surface_acpi_notify.c b/drivers/platform/surface/surface_acpi_notify.c
index 50500e562963d..897cdd9c3aae8 100644
--- a/drivers/platform/surface/surface_acpi_notify.c
+++ b/drivers/platform/surface/surface_acpi_notify.c
@@ -590,7 +590,7 @@ static acpi_status san_rqst(struct san_data *d, struct gsb_buffer *buffer)
 		return san_rqst_fixup_suspended(d, &rqst, buffer);
 	}
 
-	status = __ssam_retry(ssam_request_sync_onstack, SAN_REQUEST_NUM_TRIES,
+	status = __ssam_retry(ssam_request_do_sync_onstack, SAN_REQUEST_NUM_TRIES,
 			      d->ctrl, &rqst, &rsp, SAN_GSB_MAX_RQSX_PAYLOAD);
 
 	if (!status) {
diff --git a/drivers/platform/surface/surface_aggregator_cdev.c b/drivers/platform/surface/surface_aggregator_cdev.c
index 492c82e691827..07f0ed658369b 100644
--- a/drivers/platform/surface/surface_aggregator_cdev.c
+++ b/drivers/platform/surface/surface_aggregator_cdev.c
@@ -302,8 +302,8 @@ static long ssam_cdev_request(struct ssam_cdev_client *client, struct ssam_cdev_
 		 * theoretical maximum (SSH_COMMAND_MAX_PAYLOAD_SIZE) of the
 		 * underlying protocol (note that nothing remotely this size
 		 * should ever be allocated in any normal case). This size is
-		 * validated later in ssam_request_sync(), for allocation the
-		 * bound imposed by u16 should be enough.
+		 * validated later in ssam_request_do_sync(), for allocation
+		 * the bound imposed by u16 should be enough.
 		 */
 		spec.payload = kzalloc(spec.length, GFP_KERNEL);
 		if (!spec.payload) {
@@ -342,7 +342,7 @@ static long ssam_cdev_request(struct ssam_cdev_client *client, struct ssam_cdev_
 	}
 
 	/* Perform request. */
-	status = ssam_request_sync(client->cdev->ctrl, &spec, &rsp);
+	status = ssam_request_do_sync(client->cdev->ctrl, &spec, &rsp);
 	if (status)
 		goto out;
 
diff --git a/drivers/platform/surface/surface_aggregator_tabletsw.c b/drivers/platform/surface/surface_aggregator_tabletsw.c
index 6147aa8879391..9fed800c7cc09 100644
--- a/drivers/platform/surface/surface_aggregator_tabletsw.c
+++ b/drivers/platform/surface/surface_aggregator_tabletsw.c
@@ -382,7 +382,7 @@ static int ssam_pos_get_sources_list(struct ssam_tablet_sw *sw, struct ssam_sour
 	rsp.length = 0;
 	rsp.pointer = (u8 *)sources;
 
-	status = ssam_retry(ssam_request_sync_onstack, sw->sdev->ctrl, &rqst, &rsp, 0);
+	status = ssam_retry(ssam_request_do_sync_onstack, sw->sdev->ctrl, &rqst, &rsp, 0);
 	if (status)
 		return status;
 
diff --git a/include/linux/surface_aggregator/controller.h b/include/linux/surface_aggregator/controller.h
index 8932bc0bae187..cb7980805920a 100644
--- a/include/linux/surface_aggregator/controller.h
+++ b/include/linux/surface_aggregator/controller.h
@@ -207,17 +207,17 @@ static inline int ssam_request_sync_wait(struct ssam_request_sync *rqst)
 	return rqst->status;
 }
 
-int ssam_request_sync(struct ssam_controller *ctrl,
-		      const struct ssam_request *spec,
-		      struct ssam_response *rsp);
+int ssam_request_do_sync(struct ssam_controller *ctrl,
+			 const struct ssam_request *spec,
+			 struct ssam_response *rsp);
 
-int ssam_request_sync_with_buffer(struct ssam_controller *ctrl,
-				  const struct ssam_request *spec,
-				  struct ssam_response *rsp,
-				  struct ssam_span *buf);
+int ssam_request_do_sync_with_buffer(struct ssam_controller *ctrl,
+				     const struct ssam_request *spec,
+				     struct ssam_response *rsp,
+				     struct ssam_span *buf);
 
 /**
- * ssam_request_sync_onstack - Execute a synchronous request on the stack.
+ * ssam_request_do_sync_onstack - Execute a synchronous request on the stack.
  * @ctrl: The controller via which the request is submitted.
  * @rqst: The request specification.
  * @rsp:  The response buffer.
@@ -227,7 +227,7 @@ int ssam_request_sync_with_buffer(struct ssam_controller *ctrl,
  * fully initializes it via the provided request specification, submits it,
  * and finally waits for its completion before returning its status. This
  * helper macro essentially allocates the request message buffer on the stack
- * and then calls ssam_request_sync_with_buffer().
+ * and then calls ssam_request_do_sync_with_buffer().
  *
  * Note: The @payload_len parameter specifies the maximum payload length, used
  * for buffer allocation. The actual payload length may be smaller.
@@ -235,12 +235,12 @@ int ssam_request_sync_with_buffer(struct ssam_controller *ctrl,
  * Return: Returns the status of the request or any failure during setup, i.e.
  * zero on success and a negative value on failure.
  */
-#define ssam_request_sync_onstack(ctrl, rqst, rsp, payload_len)			\
+#define ssam_request_do_sync_onstack(ctrl, rqst, rsp, payload_len)		\
 	({									\
 		u8 __data[SSH_COMMAND_MESSAGE_LENGTH(payload_len)];		\
 		struct ssam_span __buf = { &__data[0], ARRAY_SIZE(__data) };	\
 										\
-		ssam_request_sync_with_buffer(ctrl, rqst, rsp, &__buf);		\
+		ssam_request_do_sync_with_buffer(ctrl, rqst, rsp, &__buf);	\
 	})
 
 /**
@@ -349,7 +349,7 @@ struct ssam_request_spec_md {
  * zero on success and negative on failure. The ``ctrl`` parameter is the
  * controller via which the request is being sent.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_N(name, spec...)				\
@@ -366,7 +366,7 @@ struct ssam_request_spec_md {
 		rqst.length = 0;						\
 		rqst.payload = NULL;						\
 										\
-		return ssam_request_sync_onstack(ctrl, &rqst, NULL, 0);		\
+		return ssam_request_do_sync_onstack(ctrl, &rqst, NULL, 0);	\
 	}
 
 /**
@@ -389,7 +389,7 @@ struct ssam_request_spec_md {
  * parameter is the controller via which the request is sent. The request
  * argument is specified via the ``arg`` pointer.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_W(name, atype, spec...)			\
@@ -406,8 +406,8 @@ struct ssam_request_spec_md {
 		rqst.length = sizeof(atype);					\
 		rqst.payload = (u8 *)arg;					\
 										\
-		return ssam_request_sync_onstack(ctrl, &rqst, NULL,		\
-						 sizeof(atype));		\
+		return ssam_request_do_sync_onstack(ctrl, &rqst, NULL,		\
+						    sizeof(atype));		\
 	}
 
 /**
@@ -430,7 +430,7 @@ struct ssam_request_spec_md {
  * the controller via which the request is sent. The request's return value is
  * written to the memory pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_R(name, rtype, spec...)			\
@@ -453,7 +453,7 @@ struct ssam_request_spec_md {
 		rsp.length = 0;							\
 		rsp.pointer = (u8 *)ret;					\
 										\
-		status = ssam_request_sync_onstack(ctrl, &rqst, &rsp, 0);	\
+		status = ssam_request_do_sync_onstack(ctrl, &rqst, &rsp, 0);	\
 		if (status)							\
 			return status;						\
 										\
@@ -491,7 +491,7 @@ struct ssam_request_spec_md {
  * request argument is specified via the ``arg`` pointer. The request's return
  * value is written to the memory pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_WR(name, atype, rtype, spec...)		\
@@ -514,7 +514,7 @@ struct ssam_request_spec_md {
 		rsp.length = 0;							\
 		rsp.pointer = (u8 *)ret;					\
 										\
-		status = ssam_request_sync_onstack(ctrl, &rqst, &rsp, sizeof(atype)); \
+		status = ssam_request_do_sync_onstack(ctrl, &rqst, &rsp, sizeof(atype)); \
 		if (status)							\
 			return status;						\
 										\
@@ -550,7 +550,7 @@ struct ssam_request_spec_md {
  * parameter is the controller via which the request is sent, ``tid`` the
  * target ID for the request, and ``iid`` the instance ID.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_MD_N(name, spec...)				\
@@ -567,7 +567,7 @@ struct ssam_request_spec_md {
 		rqst.length = 0;						\
 		rqst.payload = NULL;						\
 										\
-		return ssam_request_sync_onstack(ctrl, &rqst, NULL, 0);		\
+		return ssam_request_do_sync_onstack(ctrl, &rqst, NULL, 0);	\
 	}
 
 /**
@@ -592,7 +592,7 @@ struct ssam_request_spec_md {
  * ``tid`` the target ID for the request, and ``iid`` the instance ID. The
  * request argument is specified via the ``arg`` pointer.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_MD_W(name, atype, spec...)			\
@@ -609,7 +609,7 @@ struct ssam_request_spec_md {
 		rqst.length = sizeof(atype);					\
 		rqst.payload = (u8 *)arg;					\
 										\
-		return ssam_request_sync_onstack(ctrl, &rqst, NULL,		\
+		return ssam_request_do_sync_onstack(ctrl, &rqst, NULL,		\
 						 sizeof(atype));		\
 	}
 
@@ -635,7 +635,7 @@ struct ssam_request_spec_md {
  * the target ID for the request, and ``iid`` the instance ID. The request's
  * return value is written to the memory pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_MD_R(name, rtype, spec...)			\
@@ -658,7 +658,7 @@ struct ssam_request_spec_md {
 		rsp.length = 0;							\
 		rsp.pointer = (u8 *)ret;					\
 										\
-		status = ssam_request_sync_onstack(ctrl, &rqst, &rsp, 0);	\
+		status = ssam_request_do_sync_onstack(ctrl, &rqst, &rsp, 0);	\
 		if (status)							\
 			return status;						\
 										\
@@ -698,7 +698,7 @@ struct ssam_request_spec_md {
  * The request argument is specified via the ``arg`` pointer. The request's
  * return value is written to the memory pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_MD_WR(name, atype, rtype, spec...)		\
@@ -722,7 +722,7 @@ struct ssam_request_spec_md {
 		rsp.length = 0;							\
 		rsp.pointer = (u8 *)ret;					\
 										\
-		status = ssam_request_sync_onstack(ctrl, &rqst, &rsp, sizeof(atype)); \
+		status = ssam_request_do_sync_onstack(ctrl, &rqst, &rsp, sizeof(atype)); \
 		if (status)							\
 			return status;						\
 										\
diff --git a/include/linux/surface_aggregator/device.h b/include/linux/surface_aggregator/device.h
index 4da20b7a0ee5e..1545e5567b152 100644
--- a/include/linux/surface_aggregator/device.h
+++ b/include/linux/surface_aggregator/device.h
@@ -456,7 +456,7 @@ static inline int ssam_device_register_clients(struct ssam_device *sdev)
  * device of the request and by association the controller via which the
  * request is sent.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_CL_N(name, spec...)			\
@@ -490,7 +490,7 @@ static inline int ssam_device_register_clients(struct ssam_device *sdev)
  * which the request is sent. The request's argument is specified via the
  * ``arg`` pointer.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_CL_W(name, atype, spec...)		\
@@ -524,7 +524,7 @@ static inline int ssam_device_register_clients(struct ssam_device *sdev)
  * the request is sent. The request's return value is written to the memory
  * pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_CL_R(name, rtype, spec...)		\
@@ -560,7 +560,7 @@ static inline int ssam_device_register_clients(struct ssam_device *sdev)
  * specified via the ``arg`` pointer. The request's return value is written to
  * the memory pointed to by the ``ret`` parameter.
  *
- * Refer to ssam_request_sync_onstack() for more details on the behavior of
+ * Refer to ssam_request_do_sync_onstack() for more details on the behavior of
  * the generated function.
  */
 #define SSAM_DEFINE_SYNC_REQUEST_CL_WR(name, atype, rtype, spec...)		\
-- 
2.39.2

From 92a5456ed57b0ab00122d10085387a5b515ec238 Mon Sep 17 00:00:00 2001
From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Date: Wed, 18 Jan 2023 11:38:23 +0200
Subject: [PATCH] platform/surface: Switch to use acpi_evaluate_dsm_typed()

The acpi_evaluate_dsm_typed() provides a way to check the type of the
object evaluated by _DSM call. Use it instead of open coded variant.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20230118093823.39679-1-andriy.shevchenko@linux.intel.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_hotplug.c | 13 +++----------
 1 file changed, 3 insertions(+), 10 deletions(-)

diff --git a/drivers/platform/surface/surface_hotplug.c b/drivers/platform/surface/surface_hotplug.c
index f004a24952013..7b6d887dccdbf 100644
--- a/drivers/platform/surface/surface_hotplug.c
+++ b/drivers/platform/surface/surface_hotplug.c
@@ -101,18 +101,12 @@ static void shps_dsm_notify_irq(struct platform_device *pdev, enum shps_irq_type
 	param.type = ACPI_TYPE_INTEGER;
 	param.integer.value = value;
 
-	result = acpi_evaluate_dsm(handle, &shps_dsm_guid, SHPS_DSM_REVISION,
-				   shps_dsm_fn_for_irq(type), &param);
-
+	result = acpi_evaluate_dsm_typed(handle, &shps_dsm_guid, SHPS_DSM_REVISION,
+					 shps_dsm_fn_for_irq(type), &param, ACPI_TYPE_BUFFER);
 	if (!result) {
 		dev_err(&pdev->dev, "IRQ notification via DSM failed (irq=%d, gpio=%d)\n",
 			type, value);
 
-	} else if (result->type != ACPI_TYPE_BUFFER) {
-		dev_err(&pdev->dev,
-			"IRQ notification via DSM failed: unexpected result type (irq=%d, gpio=%d)\n",
-			type, value);
-
 	} else if (result->buffer.length != 1 || result->buffer.pointer[0] != 0) {
 		dev_err(&pdev->dev,
 			"IRQ notification via DSM failed: unexpected result value (irq=%d, gpio=%d)\n",
@@ -121,8 +115,7 @@ static void shps_dsm_notify_irq(struct platform_device *pdev, enum shps_irq_type
 
 	mutex_unlock(&sdev->lock[type]);
 
-	if (result)
-		ACPI_FREE(result);
+	ACPI_FREE(result);
 }
 
 static irqreturn_t shps_handle_irq(int irq, void *data)
-- 
2.39.2

From 1bf772afae5fed7cd220aa56888e69d4ef0ea31c Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sat, 4 Mar 2023 20:09:36 +0100
Subject: [PATCH] platform/surface: aggregator_tabletsw: Properly handle
 different posture source IDs

The device posture subsystem (POS) can provide different posture
sources. Different sources can provide different posture states and
sources can be identified by their ID.

For example, screen posture of the Surface Laptop Studio (SLS), which is
currently the only supported source, uses a source ID of 0x03. The
Surface Pro 9 uses the same subsystem for its Type-Cover, however,
provides different states for that under the ID 0x00.

To eventually support the Surface Pro 9 and potential future devices, we
need to properly disambiguate between source IDs. Therefore, add the
source ID to the state we carry and determine the tablet-mode state (as
well as state names) based on that.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-sam
---
 .../surface/surface_aggregator_tabletsw.c     | 123 ++++++++++++------
 1 file changed, 84 insertions(+), 39 deletions(-)

diff --git a/drivers/platform/surface/surface_aggregator_tabletsw.c b/drivers/platform/surface/surface_aggregator_tabletsw.c
index 9fed800c7cc09..e8682f52558f3 100644
--- a/drivers/platform/surface/surface_aggregator_tabletsw.c
+++ b/drivers/platform/surface/surface_aggregator_tabletsw.c
@@ -20,16 +20,23 @@
 
 struct ssam_tablet_sw;
 
+struct ssam_tablet_sw_state {
+	u32 source;
+	u32 state;
+};
+
 struct ssam_tablet_sw_ops {
-	int (*get_state)(struct ssam_tablet_sw *sw, u32 *state);
-	const char *(*state_name)(struct ssam_tablet_sw *sw, u32 state);
-	bool (*state_is_tablet_mode)(struct ssam_tablet_sw *sw, u32 state);
+	int (*get_state)(struct ssam_tablet_sw *sw, struct ssam_tablet_sw_state *state);
+	const char *(*state_name)(struct ssam_tablet_sw *sw,
+				  const struct ssam_tablet_sw_state *state);
+	bool (*state_is_tablet_mode)(struct ssam_tablet_sw *sw,
+				     const struct ssam_tablet_sw_state *state);
 };
 
 struct ssam_tablet_sw {
 	struct ssam_device *sdev;
 
-	u32 state;
+	struct ssam_tablet_sw_state state;
 	struct work_struct update_work;
 	struct input_dev *mode_switch;
 
@@ -45,9 +52,11 @@ struct ssam_tablet_sw_desc {
 
 	struct {
 		u32 (*notify)(struct ssam_event_notifier *nf, const struct ssam_event *event);
-		int (*get_state)(struct ssam_tablet_sw *sw, u32 *state);
-		const char *(*state_name)(struct ssam_tablet_sw *sw, u32 state);
-		bool (*state_is_tablet_mode)(struct ssam_tablet_sw *sw, u32 state);
+		int (*get_state)(struct ssam_tablet_sw *sw, struct ssam_tablet_sw_state *state);
+		const char *(*state_name)(struct ssam_tablet_sw *sw,
+					  const struct ssam_tablet_sw_state *state);
+		bool (*state_is_tablet_mode)(struct ssam_tablet_sw *sw,
+					     const struct ssam_tablet_sw_state *state);
 	} ops;
 
 	struct {
@@ -61,7 +70,7 @@ struct ssam_tablet_sw_desc {
 static ssize_t state_show(struct device *dev, struct device_attribute *attr, char *buf)
 {
 	struct ssam_tablet_sw *sw = dev_get_drvdata(dev);
-	const char *state = sw->ops.state_name(sw, sw->state);
+	const char *state = sw->ops.state_name(sw, &sw->state);
 
 	return sysfs_emit(buf, "%s\n", state);
 }
@@ -79,19 +88,19 @@ static const struct attribute_group ssam_tablet_sw_group = {
 static void ssam_tablet_sw_update_workfn(struct work_struct *work)
 {
 	struct ssam_tablet_sw *sw = container_of(work, struct ssam_tablet_sw, update_work);
+	struct ssam_tablet_sw_state state;
 	int tablet, status;
-	u32 state;
 
 	status = sw->ops.get_state(sw, &state);
 	if (status)
 		return;
 
-	if (sw->state == state)
+	if (sw->state.source == state.source && sw->state.state == state.state)
 		return;
 	sw->state = state;
 
 	/* Send SW_TABLET_MODE event. */
-	tablet = sw->ops.state_is_tablet_mode(sw, state);
+	tablet = sw->ops.state_is_tablet_mode(sw, &state);
 	input_report_switch(sw->mode_switch, SW_TABLET_MODE, tablet);
 	input_sync(sw->mode_switch);
 }
@@ -146,7 +155,7 @@ static int ssam_tablet_sw_probe(struct ssam_device *sdev)
 	sw->mode_switch->id.bustype = BUS_HOST;
 	sw->mode_switch->dev.parent = &sdev->dev;
 
-	tablet = sw->ops.state_is_tablet_mode(sw, sw->state);
+	tablet = sw->ops.state_is_tablet_mode(sw, &sw->state);
 	input_set_capability(sw->mode_switch, EV_SW, SW_TABLET_MODE);
 	input_report_switch(sw->mode_switch, SW_TABLET_MODE, tablet);
 
@@ -203,9 +212,10 @@ enum ssam_kip_cover_state {
 	SSAM_KIP_COVER_STATE_FOLDED_BACK   = 0x05,
 };
 
-static const char *ssam_kip_cover_state_name(struct ssam_tablet_sw *sw, u32 state)
+static const char *ssam_kip_cover_state_name(struct ssam_tablet_sw *sw,
+					     const struct ssam_tablet_sw_state *state)
 {
-	switch (state) {
+	switch (state->state) {
 	case SSAM_KIP_COVER_STATE_DISCONNECTED:
 		return "disconnected";
 
@@ -222,14 +232,15 @@ static const char *ssam_kip_cover_state_name(struct ssam_tablet_sw *sw, u32 stat
 		return "folded-back";
 
 	default:
-		dev_warn(&sw->sdev->dev, "unknown KIP cover state: %u\n", state);
+		dev_warn(&sw->sdev->dev, "unknown KIP cover state: %u\n", state->state);
 		return "<unknown>";
 	}
 }
 
-static bool ssam_kip_cover_state_is_tablet_mode(struct ssam_tablet_sw *sw, u32 state)
+static bool ssam_kip_cover_state_is_tablet_mode(struct ssam_tablet_sw *sw,
+						const struct ssam_tablet_sw_state *state)
 {
-	switch (state) {
+	switch (state->state) {
 	case SSAM_KIP_COVER_STATE_DISCONNECTED:
 	case SSAM_KIP_COVER_STATE_FOLDED_CANVAS:
 	case SSAM_KIP_COVER_STATE_FOLDED_BACK:
@@ -240,7 +251,7 @@ static bool ssam_kip_cover_state_is_tablet_mode(struct ssam_tablet_sw *sw, u32 s
 		return false;
 
 	default:
-		dev_warn(&sw->sdev->dev, "unknown KIP cover state: %d\n", sw->state);
+		dev_warn(&sw->sdev->dev, "unknown KIP cover state: %d\n", state->state);
 		return true;
 	}
 }
@@ -252,7 +263,7 @@ SSAM_DEFINE_SYNC_REQUEST_R(__ssam_kip_get_cover_state, u8, {
 	.instance_id     = 0x00,
 });
 
-static int ssam_kip_get_cover_state(struct ssam_tablet_sw *sw, u32 *state)
+static int ssam_kip_get_cover_state(struct ssam_tablet_sw *sw, struct ssam_tablet_sw_state *state)
 {
 	int status;
 	u8 raw;
@@ -263,7 +274,8 @@ static int ssam_kip_get_cover_state(struct ssam_tablet_sw *sw, u32 *state)
 		return status;
 	}
 
-	*state = raw;
+	state->source = 0;	/* Unused for KIP switch. */
+	state->state = raw;
 	return 0;
 }
 
@@ -312,11 +324,15 @@ MODULE_PARM_DESC(tablet_mode_in_slate_state, "Enable tablet mode in slate device
 #define SSAM_EVENT_POS_CID_POSTURE_CHANGED	0x03
 #define SSAM_POS_MAX_SOURCES			4
 
-enum ssam_pos_state {
-	SSAM_POS_POSTURE_LID_CLOSED = 0x00,
-	SSAM_POS_POSTURE_LAPTOP     = 0x01,
-	SSAM_POS_POSTURE_SLATE      = 0x02,
-	SSAM_POS_POSTURE_TABLET     = 0x03,
+enum ssam_pos_source_id {
+	SSAM_POS_SOURCE_SLS   = 0x03,
+};
+
+enum ssam_pos_state_sls {
+	SSAM_POS_SLS_LID_CLOSED = 0x00,
+	SSAM_POS_SLS_LAPTOP     = 0x01,
+	SSAM_POS_SLS_SLATE      = 0x02,
+	SSAM_POS_SLS_TABLET     = 0x03,
 };
 
 struct ssam_sources_list {
@@ -324,42 +340,68 @@ struct ssam_sources_list {
 	__le32 id[SSAM_POS_MAX_SOURCES];
 } __packed;
 
-static const char *ssam_pos_state_name(struct ssam_tablet_sw *sw, u32 state)
+static const char *ssam_pos_state_name_sls(struct ssam_tablet_sw *sw, u32 state)
 {
 	switch (state) {
-	case SSAM_POS_POSTURE_LID_CLOSED:
+	case SSAM_POS_SLS_LID_CLOSED:
 		return "closed";
 
-	case SSAM_POS_POSTURE_LAPTOP:
+	case SSAM_POS_SLS_LAPTOP:
 		return "laptop";
 
-	case SSAM_POS_POSTURE_SLATE:
+	case SSAM_POS_SLS_SLATE:
 		return "slate";
 
-	case SSAM_POS_POSTURE_TABLET:
+	case SSAM_POS_SLS_TABLET:
 		return "tablet";
 
 	default:
-		dev_warn(&sw->sdev->dev, "unknown device posture: %u\n", state);
+		dev_warn(&sw->sdev->dev, "unknown device posture for SLS: %u\n", state);
 		return "<unknown>";
 	}
 }
 
-static bool ssam_pos_state_is_tablet_mode(struct ssam_tablet_sw *sw, u32 state)
+static const char *ssam_pos_state_name(struct ssam_tablet_sw *sw,
+				       const struct ssam_tablet_sw_state *state)
+{
+	switch (state->source) {
+	case SSAM_POS_SOURCE_SLS:
+		return ssam_pos_state_name_sls(sw, state->state);
+
+	default:
+		dev_warn(&sw->sdev->dev, "unknown device posture source: %u\n", state->source);
+		return "<unknown>";
+	}
+}
+
+static bool ssam_pos_state_is_tablet_mode_sls(struct ssam_tablet_sw *sw, u32 state)
 {
 	switch (state) {
-	case SSAM_POS_POSTURE_LAPTOP:
-	case SSAM_POS_POSTURE_LID_CLOSED:
+	case SSAM_POS_SLS_LAPTOP:
+	case SSAM_POS_SLS_LID_CLOSED:
 		return false;
 
-	case SSAM_POS_POSTURE_SLATE:
+	case SSAM_POS_SLS_SLATE:
 		return tablet_mode_in_slate_state;
 
-	case SSAM_POS_POSTURE_TABLET:
+	case SSAM_POS_SLS_TABLET:
 		return true;
 
 	default:
-		dev_warn(&sw->sdev->dev, "unknown device posture: %u\n", state);
+		dev_warn(&sw->sdev->dev, "unknown device posture for SLS: %u\n", state);
+		return true;
+	}
+}
+
+static bool ssam_pos_state_is_tablet_mode(struct ssam_tablet_sw *sw,
+					  const struct ssam_tablet_sw_state *state)
+{
+	switch (state->source) {
+	case SSAM_POS_SOURCE_SLS:
+		return ssam_pos_state_is_tablet_mode_sls(sw, state->state);
+
+	default:
+		dev_warn(&sw->sdev->dev, "unknown device posture source: %u\n", state->source);
 		return true;
 	}
 }
@@ -450,9 +492,10 @@ static int ssam_pos_get_posture_for_source(struct ssam_tablet_sw *sw, u32 source
 	return 0;
 }
 
-static int ssam_pos_get_posture(struct ssam_tablet_sw *sw, u32 *state)
+static int ssam_pos_get_posture(struct ssam_tablet_sw *sw, struct ssam_tablet_sw_state *state)
 {
 	u32 source_id;
+	u32 source_state;
 	int status;
 
 	status = ssam_pos_get_source(sw, &source_id);
@@ -461,13 +504,15 @@ static int ssam_pos_get_posture(struct ssam_tablet_sw *sw, u32 *state)
 		return status;
 	}
 
-	status = ssam_pos_get_posture_for_source(sw, source_id, state);
+	status = ssam_pos_get_posture_for_source(sw, source_id, &source_state);
 	if (status) {
 		dev_err(&sw->sdev->dev, "failed to get posture value for source %u: %d\n",
 			source_id, status);
 		return status;
 	}
 
+	state->source = source_id;
+	state->state = source_state;
 	return 0;
 }
 
-- 
2.39.2

From 64197727498531561cb89c910e119e70533d13fd Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sun, 19 Feb 2023 23:33:43 +0100
Subject: [PATCH] platform/surface: aggregator_tabletsw: Add support for
 Type-Cover posture source

Implement support for the Type-Cover posture source (ID 0x00), found on
the Surface Pro 9.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-sam
---
 .../surface/surface_aggregator_tabletsw.c     | 57 +++++++++++++++++++
 1 file changed, 57 insertions(+)

diff --git a/drivers/platform/surface/surface_aggregator_tabletsw.c b/drivers/platform/surface/surface_aggregator_tabletsw.c
index e8682f52558f3..8f52b62d1c195 100644
--- a/drivers/platform/surface/surface_aggregator_tabletsw.c
+++ b/drivers/platform/surface/surface_aggregator_tabletsw.c
@@ -325,9 +325,18 @@ MODULE_PARM_DESC(tablet_mode_in_slate_state, "Enable tablet mode in slate device
 #define SSAM_POS_MAX_SOURCES			4
 
 enum ssam_pos_source_id {
+	SSAM_POS_SOURCE_COVER = 0x00,
 	SSAM_POS_SOURCE_SLS   = 0x03,
 };
 
+enum ssam_pos_state_cover {
+	SSAM_POS_COVER_DISCONNECTED  = 0x01,
+	SSAM_POS_COVER_CLOSED        = 0x02,
+	SSAM_POS_COVER_LAPTOP        = 0x03,
+	SSAM_POS_COVER_FOLDED_CANVAS = 0x04,
+	SSAM_POS_COVER_FOLDED_BACK   = 0x05,
+};
+
 enum ssam_pos_state_sls {
 	SSAM_POS_SLS_LID_CLOSED = 0x00,
 	SSAM_POS_SLS_LAPTOP     = 0x01,
@@ -340,6 +349,30 @@ struct ssam_sources_list {
 	__le32 id[SSAM_POS_MAX_SOURCES];
 } __packed;
 
+static const char *ssam_pos_state_name_cover(struct ssam_tablet_sw *sw, u32 state)
+{
+	switch (state) {
+	case SSAM_POS_COVER_DISCONNECTED:
+		return "disconnected";
+
+	case SSAM_POS_COVER_CLOSED:
+		return "closed";
+
+	case SSAM_POS_COVER_LAPTOP:
+		return "laptop";
+
+	case SSAM_POS_COVER_FOLDED_CANVAS:
+		return "folded-canvas";
+
+	case SSAM_POS_COVER_FOLDED_BACK:
+		return "folded-back";
+
+	default:
+		dev_warn(&sw->sdev->dev, "unknown device posture for type-cover: %u\n", state);
+		return "<unknown>";
+	}
+}
+
 static const char *ssam_pos_state_name_sls(struct ssam_tablet_sw *sw, u32 state)
 {
 	switch (state) {
@@ -365,6 +398,9 @@ static const char *ssam_pos_state_name(struct ssam_tablet_sw *sw,
 				       const struct ssam_tablet_sw_state *state)
 {
 	switch (state->source) {
+	case SSAM_POS_SOURCE_COVER:
+		return ssam_pos_state_name_cover(sw, state->state);
+
 	case SSAM_POS_SOURCE_SLS:
 		return ssam_pos_state_name_sls(sw, state->state);
 
@@ -374,6 +410,24 @@ static const char *ssam_pos_state_name(struct ssam_tablet_sw *sw,
 	}
 }
 
+static bool ssam_pos_state_is_tablet_mode_cover(struct ssam_tablet_sw *sw, u32 state)
+{
+	switch (state) {
+	case SSAM_POS_COVER_DISCONNECTED:
+	case SSAM_POS_COVER_FOLDED_CANVAS:
+	case SSAM_POS_COVER_FOLDED_BACK:
+		return true;
+
+	case SSAM_POS_COVER_CLOSED:
+	case SSAM_POS_COVER_LAPTOP:
+		return false;
+
+	default:
+		dev_warn(&sw->sdev->dev, "unknown device posture for type-cover: %u\n", state);
+		return true;
+	}
+}
+
 static bool ssam_pos_state_is_tablet_mode_sls(struct ssam_tablet_sw *sw, u32 state)
 {
 	switch (state) {
@@ -397,6 +451,9 @@ static bool ssam_pos_state_is_tablet_mode(struct ssam_tablet_sw *sw,
 					  const struct ssam_tablet_sw_state *state)
 {
 	switch (state->source) {
+	case SSAM_POS_SOURCE_COVER:
+		return ssam_pos_state_is_tablet_mode_cover(sw, state->state);
+
 	case SSAM_POS_SOURCE_SLS:
 		return ssam_pos_state_is_tablet_mode_sls(sw, state->state);
 
-- 
2.39.2

From 8357939c324e363bfad4813d881b34ead8e4c243 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sun, 19 Feb 2023 23:41:18 +0100
Subject: [PATCH] platform/surface: aggregator_registry: Add support for
 tablet-mode switch on Surface Pro 9

Add support for the POS-subsystem tablet-mode switch used on the Surface
Pro 9.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-sam
---
 drivers/platform/surface/surface_aggregator_registry.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/platform/surface/surface_aggregator_registry.c b/drivers/platform/surface/surface_aggregator_registry.c
index 296f72d52e6a6..0fe5be5396525 100644
--- a/drivers/platform/surface/surface_aggregator_registry.c
+++ b/drivers/platform/surface/surface_aggregator_registry.c
@@ -305,7 +305,7 @@ static const struct software_node *ssam_node_group_sp9[] = {
 	&ssam_node_bat_ac,
 	&ssam_node_bat_main,
 	&ssam_node_tmp_pprof,
-	/* TODO: Tablet mode switch (via POS subsystem) */
+	&ssam_node_pos_tablet_switch,
 	&ssam_node_hid_kip_keyboard,
 	&ssam_node_hid_kip_penstash,
 	&ssam_node_hid_kip_touchpad,
-- 
2.39.2

From 6a5e1c3f0a2a55dd2cb55db7b8b8b0c1369443b6 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sat, 25 Jul 2020 17:19:53 +0200
Subject: [PATCH] i2c: acpi: Implement RawBytes read access

Microsoft Surface Pro 4 and Book 1 devices access the MSHW0030 I2C
device via a generic serial bus operation region and RawBytes read
access. On the Surface Book 1, this access is required to turn on (and
off) the discrete GPU.

Multiple things are to note here:

a) The RawBytes access is device/driver dependent. The ACPI
   specification states:

   > Raw accesses assume that the writer has knowledge of the bus that
   > the access is made over and the device that is being accessed. The
   > protocol may only ensure that the buffer is transmitted to the
   > appropriate driver, but the driver must be able to interpret the
   > buffer to communicate to a register.

   Thus this implementation may likely not work on other devices
   accessing I2C via the RawBytes accessor type.

b) The MSHW0030 I2C device is an HID-over-I2C device which seems to
   serve multiple functions:

   1. It is the main access point for the legacy-type Surface Aggregator
      Module (also referred to as SAM-over-HID, as opposed to the newer
      SAM-over-SSH/UART). It has currently not been determined on how
      support for the legacy SAM should be implemented. Likely via a
      custom HID driver.

   2. It seems to serve as the HID device for the Integrated Sensor Hub.
      This might complicate matters with regards to implementing a
      SAM-over-HID driver required by legacy SAM.

In light of this, the simplest approach has been chosen for now.
However, it may make more sense regarding breakage and compatibility to
either provide functionality for replacing or enhancing the default
operation region handler via some additional API functions, or even to
completely blacklist MSHW0030 from the I2C core and provide a custom
driver for it.

Replacing/enhancing the default operation region handler would, however,
either require some sort of secondary driver and access point for it,
from which the new API functions would be called and the new handler
(part) would be installed, or hard-coding them via some sort of
quirk-like interface into the I2C core.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-sam-over-hid
---
 drivers/i2c/i2c-core-acpi.c | 35 +++++++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)

diff --git a/drivers/i2c/i2c-core-acpi.c b/drivers/i2c/i2c-core-acpi.c
index 4dd777cc0c89f..b2338618163ad 100644
--- a/drivers/i2c/i2c-core-acpi.c
+++ b/drivers/i2c/i2c-core-acpi.c
@@ -639,6 +639,28 @@ static int acpi_gsb_i2c_write_bytes(struct i2c_client *client,
 	return (ret == 1) ? 0 : -EIO;
 }
 
+static int acpi_gsb_i2c_write_raw_bytes(struct i2c_client *client,
+		u8 *data, u8 data_len)
+{
+	struct i2c_msg msgs[1];
+	int ret = AE_OK;
+
+	msgs[0].addr = client->addr;
+	msgs[0].flags = client->flags;
+	msgs[0].len = data_len + 1;
+	msgs[0].buf = data;
+
+	ret = i2c_transfer(client->adapter, msgs, ARRAY_SIZE(msgs));
+
+	if (ret < 0) {
+		dev_err(&client->adapter->dev, "i2c write failed: %d\n", ret);
+		return ret;
+	}
+
+	/* 1 transfer must have completed successfully */
+	return (ret == 1) ? 0 : -EIO;
+}
+
 static acpi_status
 i2c_acpi_space_handler(u32 function, acpi_physical_address command,
 			u32 bits, u64 *value64,
@@ -740,6 +762,19 @@ i2c_acpi_space_handler(u32 function, acpi_physical_address command,
 		}
 		break;
 
+	case ACPI_GSB_ACCESS_ATTRIB_RAW_BYTES:
+		if (action == ACPI_READ) {
+			dev_warn(&adapter->dev,
+				 "protocol 0x%02x not supported for client 0x%02x\n",
+				 accessor_type, client->addr);
+			ret = AE_BAD_PARAMETER;
+			goto err;
+		} else {
+			status = acpi_gsb_i2c_write_raw_bytes(client,
+					gsb->data, info->access_length);
+		}
+		break;
+
 	default:
 		dev_warn(&adapter->dev, "protocol 0x%02x not supported for client 0x%02x\n",
 			 accessor_type, client->addr);
-- 
2.39.2

From 2a6f08f97a8f645dabd0538a09d3692c88474066 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sat, 13 Feb 2021 16:41:18 +0100
Subject: [PATCH] platform/surface: Add driver for Surface Book 1 dGPU switch

Add driver exposing the discrete GPU power-switch of the  Microsoft
Surface Book 1 to user-space.

On the Surface Book 1, the dGPU power is controlled via the Surface
System Aggregator Module (SAM). The specific SAM-over-HID command for
this is exposed via ACPI. This module provides a simple driver exposing
the ACPI call via a sysfs parameter to user-space, so that users can
easily power-on/-off the dGPU.

Patchset: surface-sam-over-hid
---
 drivers/platform/surface/Kconfig              |   7 +
 drivers/platform/surface/Makefile             |   1 +
 .../surface/surfacebook1_dgpu_switch.c        | 162 ++++++++++++++++++
 3 files changed, 170 insertions(+)
 create mode 100644 drivers/platform/surface/surfacebook1_dgpu_switch.c

diff --git a/drivers/platform/surface/Kconfig b/drivers/platform/surface/Kconfig
index b629e82af97c0..68656e8f309ed 100644
--- a/drivers/platform/surface/Kconfig
+++ b/drivers/platform/surface/Kconfig
@@ -149,6 +149,13 @@ config SURFACE_AGGREGATOR_TABLET_SWITCH
 	  Select M or Y here, if you want to provide tablet-mode switch input
 	  events on the Surface Pro 8, Surface Pro X, and Surface Laptop Studio.
 
+config SURFACE_BOOK1_DGPU_SWITCH
+	tristate "Surface Book 1 dGPU Switch Driver"
+	depends on SYSFS
+	help
+	  This driver provides a sysfs switch to set the power-state of the
+	  discrete GPU found on the Microsoft Surface Book 1.
+
 config SURFACE_DTX
 	tristate "Surface DTX (Detachment System) Driver"
 	depends on SURFACE_AGGREGATOR
diff --git a/drivers/platform/surface/Makefile b/drivers/platform/surface/Makefile
index 53344330939bf..7efcd0cdb5329 100644
--- a/drivers/platform/surface/Makefile
+++ b/drivers/platform/surface/Makefile
@@ -12,6 +12,7 @@ obj-$(CONFIG_SURFACE_AGGREGATOR_CDEV)	+= surface_aggregator_cdev.o
 obj-$(CONFIG_SURFACE_AGGREGATOR_HUB)	+= surface_aggregator_hub.o
 obj-$(CONFIG_SURFACE_AGGREGATOR_REGISTRY) += surface_aggregator_registry.o
 obj-$(CONFIG_SURFACE_AGGREGATOR_TABLET_SWITCH) += surface_aggregator_tabletsw.o
+obj-$(CONFIG_SURFACE_BOOK1_DGPU_SWITCH) += surfacebook1_dgpu_switch.o
 obj-$(CONFIG_SURFACE_DTX)		+= surface_dtx.o
 obj-$(CONFIG_SURFACE_GPE)		+= surface_gpe.o
 obj-$(CONFIG_SURFACE_HOTPLUG)		+= surface_hotplug.o
diff --git a/drivers/platform/surface/surfacebook1_dgpu_switch.c b/drivers/platform/surface/surfacebook1_dgpu_switch.c
new file mode 100644
index 0000000000000..8b816ed8f35c6
--- /dev/null
+++ b/drivers/platform/surface/surfacebook1_dgpu_switch.c
@@ -0,0 +1,162 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/acpi.h>
+#include <linux/platform_device.h>
+
+
+#ifdef pr_fmt
+#undef pr_fmt
+#endif
+#define pr_fmt(fmt) "%s:%s: " fmt, KBUILD_MODNAME, __func__
+
+
+static const guid_t dgpu_sw_guid = GUID_INIT(0x6fd05c69, 0xcde3, 0x49f4,
+	0x95, 0xed, 0xab, 0x16, 0x65, 0x49, 0x80, 0x35);
+
+#define DGPUSW_ACPI_PATH_DSM	"\\_SB_.PCI0.LPCB.EC0_.VGBI"
+#define DGPUSW_ACPI_PATH_HGON	"\\_SB_.PCI0.RP05.HGON"
+#define DGPUSW_ACPI_PATH_HGOF	"\\_SB_.PCI0.RP05.HGOF"
+
+
+static int sb1_dgpu_sw_dsmcall(void)
+{
+	union acpi_object *ret;
+	acpi_handle handle;
+	acpi_status status;
+
+	status = acpi_get_handle(NULL, DGPUSW_ACPI_PATH_DSM, &handle);
+	if (status)
+		return -EINVAL;
+
+	ret = acpi_evaluate_dsm_typed(handle, &dgpu_sw_guid, 1, 1, NULL, ACPI_TYPE_BUFFER);
+	if (!ret)
+		return -EINVAL;
+
+	ACPI_FREE(ret);
+	return 0;
+}
+
+static int sb1_dgpu_sw_hgon(void)
+{
+	struct acpi_buffer buf = {ACPI_ALLOCATE_BUFFER, NULL};
+	acpi_status status;
+
+	status = acpi_evaluate_object(NULL, DGPUSW_ACPI_PATH_HGON, NULL, &buf);
+	if (status) {
+		pr_err("failed to run HGON: %d\n", status);
+		return -EINVAL;
+	}
+
+	if (buf.pointer)
+		ACPI_FREE(buf.pointer);
+
+	pr_info("turned-on dGPU via HGON\n");
+	return 0;
+}
+
+static int sb1_dgpu_sw_hgof(void)
+{
+	struct acpi_buffer buf = {ACPI_ALLOCATE_BUFFER, NULL};
+	acpi_status status;
+
+	status = acpi_evaluate_object(NULL, DGPUSW_ACPI_PATH_HGOF, NULL, &buf);
+	if (status) {
+		pr_err("failed to run HGOF: %d\n", status);
+		return -EINVAL;
+	}
+
+	if (buf.pointer)
+		ACPI_FREE(buf.pointer);
+
+	pr_info("turned-off dGPU via HGOF\n");
+	return 0;
+}
+
+
+static ssize_t dgpu_dsmcall_store(struct device *dev, struct device_attribute *attr,
+				  const char *buf, size_t len)
+{
+	int status, value;
+
+	status = kstrtoint(buf, 0, &value);
+	if (status < 0)
+		return status;
+
+	if (value != 1)
+		return -EINVAL;
+
+	status = sb1_dgpu_sw_dsmcall();
+
+	return status < 0 ? status : len;
+}
+
+static ssize_t dgpu_power_store(struct device *dev, struct device_attribute *attr,
+				const char *buf, size_t len)
+{
+	bool power;
+	int status;
+
+	status = kstrtobool(buf, &power);
+	if (status < 0)
+		return status;
+
+	if (power)
+		status = sb1_dgpu_sw_hgon();
+	else
+		status = sb1_dgpu_sw_hgof();
+
+	return status < 0 ? status : len;
+}
+
+static DEVICE_ATTR_WO(dgpu_dsmcall);
+static DEVICE_ATTR_WO(dgpu_power);
+
+static struct attribute *sb1_dgpu_sw_attrs[] = {
+	&dev_attr_dgpu_dsmcall.attr,
+	&dev_attr_dgpu_power.attr,
+	NULL,
+};
+
+static const struct attribute_group sb1_dgpu_sw_attr_group = {
+	.attrs = sb1_dgpu_sw_attrs,
+};
+
+
+static int sb1_dgpu_sw_probe(struct platform_device *pdev)
+{
+	return sysfs_create_group(&pdev->dev.kobj, &sb1_dgpu_sw_attr_group);
+}
+
+static int sb1_dgpu_sw_remove(struct platform_device *pdev)
+{
+	sysfs_remove_group(&pdev->dev.kobj, &sb1_dgpu_sw_attr_group);
+	return 0;
+}
+
+/*
+ * The dGPU power seems to be actually handled by MSHW0040. However, that is
+ * also the power-/volume-button device with a mainline driver. So let's use
+ * MSHW0041 instead for now, which seems to be the LTCH (latch/DTX) device.
+ */
+static const struct acpi_device_id sb1_dgpu_sw_match[] = {
+	{ "MSHW0041", },
+	{ },
+};
+MODULE_DEVICE_TABLE(acpi, sb1_dgpu_sw_match);
+
+static struct platform_driver sb1_dgpu_sw = {
+	.probe = sb1_dgpu_sw_probe,
+	.remove = sb1_dgpu_sw_remove,
+	.driver = {
+		.name = "surfacebook1_dgpu_switch",
+		.acpi_match_table = sb1_dgpu_sw_match,
+		.probe_type = PROBE_PREFER_ASYNCHRONOUS,
+	},
+};
+module_platform_driver(sb1_dgpu_sw);
+
+MODULE_AUTHOR("Maximilian Luz <luzmaximilian@gmail.com>");
+MODULE_DESCRIPTION("Discrete GPU Power-Switch for Surface Book 1");
+MODULE_LICENSE("GPL");
-- 
2.39.2

From c158663450515d39c6ed38d8d2654761b1aad0a7 Mon Sep 17 00:00:00 2001
From: Sachi King <nakato@nakato.io>
Date: Tue, 5 Oct 2021 00:05:09 +1100
Subject: [PATCH] Input: soc_button_array - support AMD variant Surface devices

The power button on the AMD variant of the Surface Laptop uses the
same MSHW0040 device ID as the 5th and later generation of Surface
devices, however they report 0 for their OEM platform revision.  As the
_DSM does not exist on the devices requiring special casing, check for
the existance of the _DSM to determine if soc_button_array should be
loaded.

Fixes: c394159310d0 ("Input: soc_button_array - add support for newer surface devices")
Co-developed-by: Maximilian Luz <luzmaximilian@gmail.com>

Signed-off-by: Sachi King <nakato@nakato.io>
Patchset: surface-button
---
 drivers/input/misc/soc_button_array.c | 33 +++++++--------------------
 1 file changed, 8 insertions(+), 25 deletions(-)

diff --git a/drivers/input/misc/soc_button_array.c b/drivers/input/misc/soc_button_array.c
index 09489380afda7..0f02411a60f1c 100644
--- a/drivers/input/misc/soc_button_array.c
+++ b/drivers/input/misc/soc_button_array.c
@@ -507,8 +507,8 @@ static const struct soc_device_data soc_device_MSHW0028 = {
  * Both, the Surface Pro 4 (surfacepro3_button.c) and the above mentioned
  * devices use MSHW0040 for power and volume buttons, however the way they
  * have to be addressed differs. Make sure that we only load this drivers
- * for the correct devices by checking the OEM Platform Revision provided by
- * the _DSM method.
+ * for the correct devices by checking if the OEM Platform Revision DSM call
+ * exists.
  */
 #define MSHW0040_DSM_REVISION		0x01
 #define MSHW0040_DSM_GET_OMPR		0x02	// get OEM Platform Revision
@@ -519,31 +519,14 @@ static const guid_t MSHW0040_DSM_UUID =
 static int soc_device_check_MSHW0040(struct device *dev)
 {
 	acpi_handle handle = ACPI_HANDLE(dev);
-	union acpi_object *result;
-	u64 oem_platform_rev = 0;	// valid revisions are nonzero
-
-	// get OEM platform revision
-	result = acpi_evaluate_dsm_typed(handle, &MSHW0040_DSM_UUID,
-					 MSHW0040_DSM_REVISION,
-					 MSHW0040_DSM_GET_OMPR, NULL,
-					 ACPI_TYPE_INTEGER);
-
-	if (result) {
-		oem_platform_rev = result->integer.value;
-		ACPI_FREE(result);
-	}
-
-	/*
-	 * If the revision is zero here, the _DSM evaluation has failed. This
-	 * indicates that we have a Pro 4 or Book 1 and this driver should not
-	 * be used.
-	 */
-	if (oem_platform_rev == 0)
-		return -ENODEV;
+	bool exists;
 
-	dev_dbg(dev, "OEM Platform Revision %llu\n", oem_platform_rev);
+	// check if OEM platform revision DSM call exists
+	exists = acpi_check_dsm(handle, &MSHW0040_DSM_UUID,
+				MSHW0040_DSM_REVISION,
+				BIT(MSHW0040_DSM_GET_OMPR));
 
-	return 0;
+	return exists ? 0 : -ENODEV;
 }
 
 /*
-- 
2.39.2

From b11c73499f55ce84a9a06220d365ef5c9a23f80b Mon Sep 17 00:00:00 2001
From: Sachi King <nakato@nakato.io>
Date: Tue, 5 Oct 2021 00:22:57 +1100
Subject: [PATCH] platform/surface: surfacepro3_button: don't load on amd
 variant

The AMD variant of the Surface Laptop report 0 for their OEM platform
revision.  The Surface devices that require the surfacepro3_button
driver do not have the _DSM that gets the OEM platform revision.  If the
method does not exist, load surfacepro3_button.

Fixes: 64dd243d7356 ("platform/x86: surfacepro3_button: Fix device check")
Co-developed-by: Maximilian Luz <luzmaximilian@gmail.com>

Signed-off-by: Sachi King <nakato@nakato.io>
Patchset: surface-button
---
 drivers/platform/surface/surfacepro3_button.c | 30 ++++---------------
 1 file changed, 6 insertions(+), 24 deletions(-)

diff --git a/drivers/platform/surface/surfacepro3_button.c b/drivers/platform/surface/surfacepro3_button.c
index 242fb690dcaf7..30eea54dbb477 100644
--- a/drivers/platform/surface/surfacepro3_button.c
+++ b/drivers/platform/surface/surfacepro3_button.c
@@ -149,7 +149,8 @@ static int surface_button_resume(struct device *dev)
 /*
  * Surface Pro 4 and Surface Book 2 / Surface Pro 2017 use the same device
  * ID (MSHW0040) for the power/volume buttons. Make sure this is the right
- * device by checking for the _DSM method and OEM Platform Revision.
+ * device by checking for the _DSM method and OEM Platform Revision DSM
+ * function.
  *
  * Returns true if the driver should bind to this device, i.e. the device is
  * either MSWH0028 (Pro 3) or MSHW0040 on a Pro 4 or Book 1.
@@ -157,30 +158,11 @@ static int surface_button_resume(struct device *dev)
 static bool surface_button_check_MSHW0040(struct acpi_device *dev)
 {
 	acpi_handle handle = dev->handle;
-	union acpi_object *result;
-	u64 oem_platform_rev = 0;	// valid revisions are nonzero
-
-	// get OEM platform revision
-	result = acpi_evaluate_dsm_typed(handle, &MSHW0040_DSM_UUID,
-					 MSHW0040_DSM_REVISION,
-					 MSHW0040_DSM_GET_OMPR,
-					 NULL, ACPI_TYPE_INTEGER);
-
-	/*
-	 * If evaluating the _DSM fails, the method is not present. This means
-	 * that we have either MSHW0028 or MSHW0040 on Pro 4 or Book 1, so we
-	 * should use this driver. We use revision 0 indicating it is
-	 * unavailable.
-	 */
-
-	if (result) {
-		oem_platform_rev = result->integer.value;
-		ACPI_FREE(result);
-	}
-
-	dev_dbg(&dev->dev, "OEM Platform Revision %llu\n", oem_platform_rev);
 
-	return oem_platform_rev == 0;
+	// make sure that OEM platform revision DSM call does not exist
+	return !acpi_check_dsm(handle, &MSHW0040_DSM_UUID,
+			       MSHW0040_DSM_REVISION,
+			       BIT(MSHW0040_DSM_GET_OMPR));
 }
 
 
-- 
2.39.2

From 6a4abbc116311e4882a0a4643737bed04b36a907 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sat, 18 Feb 2023 01:02:49 +0100
Subject: [PATCH] USB: quirks: Add USB_QUIRK_DELAY_INIT for Surface Go 3
 Type-Cover

The touchpad on the Type-Cover of the Surface Go 3 is sometimes not
being initialized properly. Apply USB_QUIRK_DELAY_INIT to fix this
issue.

More specifically, the device in question is a fairly standard modern
touchpad with pointer and touchpad input modes. During setup, the device
needs to be switched from pointer- to touchpad-mode (which is done in
hid-multitouch) to fully utilize it as intended. Unfortunately, however,
this seems to occasionally fail silently, leaving the device in
pointer-mode. Applying USB_QUIRK_DELAY_INIT seems to fix this.

Link: https://github.com/linux-surface/linux-surface/issues/1059
Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-typecover
---
 drivers/usb/core/quirks.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/usb/core/quirks.c b/drivers/usb/core/quirks.c
index 934b3d997702e..2c6604c6e8e12 100644
--- a/drivers/usb/core/quirks.c
+++ b/drivers/usb/core/quirks.c
@@ -220,6 +220,9 @@ static const struct usb_device_id usb_quirk_list[] = {
 	/* Microsoft Surface Dock Ethernet (RTL8153 GigE) */
 	{ USB_DEVICE(0x045e, 0x07c6), .driver_info = USB_QUIRK_NO_LPM },
 
+	/* Microsoft Surface Go 3 Type-Cover */
+	{ USB_DEVICE(0x045e, 0x09b5), .driver_info = USB_QUIRK_DELAY_INIT },
+
 	/* Cherry Stream G230 2.0 (G85-231) and 3.0 (G85-232) */
 	{ USB_DEVICE(0x046a, 0x0023), .driver_info = USB_QUIRK_RESET_RESUME },
 
-- 
2.39.2

From 7b335ce8bcb89dab6a3dc2dbd00a68828c991130 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jonas=20Dre=C3=9Fler?= <verdre@v0yd.nl>
Date: Thu, 5 Nov 2020 13:09:45 +0100
Subject: [PATCH] hid/multitouch: Turn off Type Cover keyboard backlight when
 suspending

The Type Cover for Microsoft Surface devices supports a special usb
control request to disable or enable the built-in keyboard backlight.
On Windows, this request happens when putting the device into suspend or
resuming it, without it the backlight of the Type Cover will remain
enabled for some time even though the computer is suspended, which looks
weird to the user.

So add support for this special usb control request to hid-multitouch,
which is the driver that's handling the Type Cover.

The reason we have to use a pm_notifier for this instead of the usual
suspend/resume methods is that those won't get called in case the usb
device is already autosuspended.

Also, if the device is autosuspended, we have to briefly autoresume it
in order to send the request. Doing that should be fine, the usb-core
driver does something similar during suspend inside choose_wakeup().

To make sure we don't send that request to every device but only to
devices which support it, add a new quirk
MT_CLS_WIN_8_MS_SURFACE_TYPE_COVER to hid-multitouch. For now this quirk
is only enabled for the usb id of the Surface Pro 2017 Type Cover, which
is where I confirmed that it's working.

Patchset: surface-typecover
---
 drivers/hid/hid-multitouch.c | 100 ++++++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 2 deletions(-)

diff --git a/drivers/hid/hid-multitouch.c b/drivers/hid/hid-multitouch.c
index e31be0cb8b850..63fd042aba6ba 100644
--- a/drivers/hid/hid-multitouch.c
+++ b/drivers/hid/hid-multitouch.c
@@ -34,7 +34,10 @@
 #include <linux/device.h>
 #include <linux/hid.h>
 #include <linux/module.h>
+#include <linux/pm_runtime.h>
 #include <linux/slab.h>
+#include <linux/suspend.h>
+#include <linux/usb.h>
 #include <linux/input/mt.h>
 #include <linux/jiffies.h>
 #include <linux/string.h>
@@ -47,6 +50,7 @@ MODULE_DESCRIPTION("HID multitouch panels");
 MODULE_LICENSE("GPL");
 
 #include "hid-ids.h"
+#include "usbhid/usbhid.h"
 
 /* quirks to control the device */
 #define MT_QUIRK_NOT_SEEN_MEANS_UP	BIT(0)
@@ -72,12 +76,15 @@ MODULE_LICENSE("GPL");
 #define MT_QUIRK_FORCE_MULTI_INPUT	BIT(20)
 #define MT_QUIRK_DISABLE_WAKEUP		BIT(21)
 #define MT_QUIRK_ORIENTATION_INVERT	BIT(22)
+#define MT_QUIRK_HAS_TYPE_COVER_BACKLIGHT	BIT(23)
 
 #define MT_INPUTMODE_TOUCHSCREEN	0x02
 #define MT_INPUTMODE_TOUCHPAD		0x03
 
 #define MT_BUTTONTYPE_CLICKPAD		0
 
+#define MS_TYPE_COVER_FEATURE_REPORT_USAGE	0xff050086
+
 enum latency_mode {
 	HID_LATENCY_NORMAL = 0,
 	HID_LATENCY_HIGH = 1,
@@ -169,6 +176,8 @@ struct mt_device {
 
 	struct list_head applications;
 	struct list_head reports;
+
+	struct notifier_block pm_notifier;
 };
 
 static void mt_post_parse_default_settings(struct mt_device *td,
@@ -213,6 +222,7 @@ static void mt_post_parse(struct mt_device *td, struct mt_application *app);
 #define MT_CLS_GOOGLE				0x0111
 #define MT_CLS_RAZER_BLADE_STEALTH		0x0112
 #define MT_CLS_SMART_TECH			0x0113
+#define MT_CLS_WIN_8_MS_SURFACE_TYPE_COVER	0x0114
 
 #define MT_DEFAULT_MAXCONTACT	10
 #define MT_MAX_MAXCONTACT	250
@@ -397,6 +407,16 @@ static const struct mt_class mt_classes[] = {
 			MT_QUIRK_CONTACT_CNT_ACCURATE |
 			MT_QUIRK_SEPARATE_APP_REPORT,
 	},
+	{ .name = MT_CLS_WIN_8_MS_SURFACE_TYPE_COVER,
+		.quirks = MT_QUIRK_HAS_TYPE_COVER_BACKLIGHT |
+			MT_QUIRK_ALWAYS_VALID |
+			MT_QUIRK_IGNORE_DUPLICATES |
+			MT_QUIRK_HOVERING |
+			MT_QUIRK_CONTACT_CNT_ACCURATE |
+			MT_QUIRK_STICKY_FINGERS |
+			MT_QUIRK_WIN8_PTP_BUTTONS,
+		.export_all_inputs = true
+	},
 	{ }
 };
 
@@ -1728,6 +1748,69 @@ static void mt_expired_timeout(struct timer_list *t)
 	clear_bit_unlock(MT_IO_FLAGS_RUNNING, &td->mt_io_flags);
 }
 
+static void get_type_cover_backlight_field(struct hid_device *hdev,
+					   struct hid_field **field)
+{
+	struct hid_report_enum *rep_enum;
+	struct hid_report *rep;
+	struct hid_field *cur_field;
+	int i, j;
+
+	rep_enum = &hdev->report_enum[HID_FEATURE_REPORT];
+	list_for_each_entry(rep, &rep_enum->report_list, list) {
+		for (i = 0; i < rep->maxfield; i++) {
+			cur_field = rep->field[i];
+
+			for (j = 0; j < cur_field->maxusage; j++) {
+				if (cur_field->usage[j].hid
+				    == MS_TYPE_COVER_FEATURE_REPORT_USAGE) {
+					*field = cur_field;
+					return;
+				}
+			}
+		}
+	}
+}
+
+static void update_keyboard_backlight(struct hid_device *hdev, bool enabled)
+{
+	struct usb_device *udev = hid_to_usb_dev(hdev);
+	struct hid_field *field = NULL;
+
+	/* Wake up the device in case it's already suspended */
+	pm_runtime_get_sync(&udev->dev);
+
+	get_type_cover_backlight_field(hdev, &field);
+	if (!field) {
+		hid_err(hdev, "couldn't find backlight field\n");
+		goto out;
+	}
+
+	field->value[field->index] = enabled ? 0x01ff00ff : 0x00ff00ff;
+	hid_hw_request(hdev, field->report, HID_REQ_SET_REPORT);
+
+out:
+	pm_runtime_put_sync(&udev->dev);
+}
+
+static int mt_pm_notifier(struct notifier_block *notifier,
+			  unsigned long pm_event,
+			  void *unused)
+{
+	struct mt_device *td =
+		container_of(notifier, struct mt_device, pm_notifier);
+	struct hid_device *hdev = td->hdev;
+
+	if (td->mtclass.quirks & MT_QUIRK_HAS_TYPE_COVER_BACKLIGHT) {
+		if (pm_event == PM_SUSPEND_PREPARE)
+			update_keyboard_backlight(hdev, 0);
+		else if (pm_event == PM_POST_SUSPEND)
+			update_keyboard_backlight(hdev, 1);
+	}
+
+	return NOTIFY_DONE;
+}
+
 static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id)
 {
 	int ret, i;
@@ -1751,6 +1834,9 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id)
 	td->inputmode_value = MT_INPUTMODE_TOUCHSCREEN;
 	hid_set_drvdata(hdev, td);
 
+	td->pm_notifier.notifier_call = mt_pm_notifier;
+	register_pm_notifier(&td->pm_notifier);
+
 	INIT_LIST_HEAD(&td->applications);
 	INIT_LIST_HEAD(&td->reports);
 
@@ -1789,15 +1875,19 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id)
 	timer_setup(&td->release_timer, mt_expired_timeout, 0);
 
 	ret = hid_parse(hdev);
-	if (ret != 0)
+	if (ret != 0) {
+		unregister_pm_notifier(&td->pm_notifier);
 		return ret;
+	}
 
 	if (mtclass->quirks & MT_QUIRK_FIX_CONST_CONTACT_ID)
 		mt_fix_const_fields(hdev, HID_DG_CONTACTID);
 
 	ret = hid_hw_start(hdev, HID_CONNECT_DEFAULT);
-	if (ret)
+	if (ret) {
+		unregister_pm_notifier(&td->pm_notifier);
 		return ret;
+	}
 
 	ret = sysfs_create_group(&hdev->dev.kobj, &mt_attribute_group);
 	if (ret)
@@ -1849,6 +1939,7 @@ static void mt_remove(struct hid_device *hdev)
 {
 	struct mt_device *td = hid_get_drvdata(hdev);
 
+	unregister_pm_notifier(&td->pm_notifier);
 	del_timer_sync(&td->release_timer);
 
 	sysfs_remove_group(&hdev->dev.kobj, &mt_attribute_group);
@@ -2226,6 +2317,11 @@ static const struct hid_device_id mt_devices[] = {
 		MT_USB_DEVICE(USB_VENDOR_ID_XIROKU,
 			USB_DEVICE_ID_XIROKU_CSR2) },
 
+	/* Microsoft Surface type cover */
+	{ .driver_data = MT_CLS_WIN_8_MS_SURFACE_TYPE_COVER,
+		HID_DEVICE(HID_BUS_ANY, HID_GROUP_ANY,
+			USB_VENDOR_ID_MICROSOFT, 0x09c0) },
+
 	/* Google MT devices */
 	{ .driver_data = MT_CLS_GOOGLE,
 		HID_DEVICE(HID_BUS_ANY, HID_GROUP_ANY, USB_VENDOR_ID_GOOGLE,
-- 
2.39.2

From 1ef242e34b7ed5aeda8072d566365a5484592ebf Mon Sep 17 00:00:00 2001
From: PJungkamp <p.jungkamp@gmail.com>
Date: Fri, 25 Feb 2022 12:04:25 +0100
Subject: [PATCH] hid/multitouch: Add support for surface pro type cover tablet
 switch

The Surface Pro Type Cover has several non standard HID usages in it's
hid report descriptor.
I noticed that, upon folding the typecover back, a vendor specific range
of 4 32 bit integer hid usages is transmitted.
Only the first byte of the message seems to convey reliable information
about the keyboard state.

0x22 => Normal (keys enabled)
0x33 => Folded back (keys disabled)
0x53 => Rotated left/right side up (keys disabled)
0x13 => Cover closed (keys disabled)
0x43 => Folded back and Tablet upside down (keys disabled)
This list may not be exhaustive.

The tablet mode switch will be disabled for a value of 0x22 and enabled
on any other value.

Patchset: surface-typecover
---
 drivers/hid/hid-multitouch.c | 148 +++++++++++++++++++++++++++++------
 1 file changed, 122 insertions(+), 26 deletions(-)

diff --git a/drivers/hid/hid-multitouch.c b/drivers/hid/hid-multitouch.c
index 63fd042aba6ba..508a250ff4bf1 100644
--- a/drivers/hid/hid-multitouch.c
+++ b/drivers/hid/hid-multitouch.c
@@ -77,6 +77,7 @@ MODULE_LICENSE("GPL");
 #define MT_QUIRK_DISABLE_WAKEUP		BIT(21)
 #define MT_QUIRK_ORIENTATION_INVERT	BIT(22)
 #define MT_QUIRK_HAS_TYPE_COVER_BACKLIGHT	BIT(23)
+#define MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH	BIT(24)
 
 #define MT_INPUTMODE_TOUCHSCREEN	0x02
 #define MT_INPUTMODE_TOUCHPAD		0x03
@@ -84,6 +85,8 @@ MODULE_LICENSE("GPL");
 #define MT_BUTTONTYPE_CLICKPAD		0
 
 #define MS_TYPE_COVER_FEATURE_REPORT_USAGE	0xff050086
+#define MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE	0xff050072
+#define MS_TYPE_COVER_APPLICATION	0xff050050
 
 enum latency_mode {
 	HID_LATENCY_NORMAL = 0,
@@ -409,6 +412,7 @@ static const struct mt_class mt_classes[] = {
 	},
 	{ .name = MT_CLS_WIN_8_MS_SURFACE_TYPE_COVER,
 		.quirks = MT_QUIRK_HAS_TYPE_COVER_BACKLIGHT |
+			MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH |
 			MT_QUIRK_ALWAYS_VALID |
 			MT_QUIRK_IGNORE_DUPLICATES |
 			MT_QUIRK_HOVERING |
@@ -1390,6 +1394,9 @@ static int mt_input_mapping(struct hid_device *hdev, struct hid_input *hi,
 	    field->application != HID_CP_CONSUMER_CONTROL &&
 	    field->application != HID_GD_WIRELESS_RADIO_CTLS &&
 	    field->application != HID_GD_SYSTEM_MULTIAXIS &&
+	    !(field->application == MS_TYPE_COVER_APPLICATION &&
+	      application->quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH &&
+	      usage->hid == MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE) &&
 	    !(field->application == HID_VD_ASUS_CUSTOM_MEDIA_KEYS &&
 	      application->quirks & MT_QUIRK_ASUS_CUSTOM_UP))
 		return -1;
@@ -1417,6 +1424,21 @@ static int mt_input_mapping(struct hid_device *hdev, struct hid_input *hi,
 		return 1;
 	}
 
+	/*
+	 * The Microsoft Surface Pro Typecover has a non-standard HID
+	 * tablet mode switch on a vendor specific usage page with vendor
+	 * specific usage.
+	 */
+	if (field->application == MS_TYPE_COVER_APPLICATION &&
+	    application->quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH &&
+	    usage->hid == MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE) {
+		usage->type = EV_SW;
+		usage->code = SW_TABLET_MODE;
+		*max = SW_MAX;
+		*bit = hi->input->swbit;
+		return 1;
+	}
+
 	if (rdata->is_mt_collection)
 		return mt_touch_input_mapping(hdev, hi, field, usage, bit, max,
 					      application);
@@ -1438,6 +1460,7 @@ static int mt_input_mapped(struct hid_device *hdev, struct hid_input *hi,
 {
 	struct mt_device *td = hid_get_drvdata(hdev);
 	struct mt_report_data *rdata;
+	struct input_dev *input;
 
 	rdata = mt_find_report_data(td, field->report);
 	if (rdata && rdata->is_mt_collection) {
@@ -1445,6 +1468,19 @@ static int mt_input_mapped(struct hid_device *hdev, struct hid_input *hi,
 		return -1;
 	}
 
+	/*
+	 * We own an input device which acts as a tablet mode switch for
+	 * the Surface Pro Typecover.
+	 */
+	if (field->application == MS_TYPE_COVER_APPLICATION &&
+	    rdata->application->quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH &&
+	    usage->hid == MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE) {
+		input = hi->input;
+		input_set_capability(input, EV_SW, SW_TABLET_MODE);
+		input_report_switch(input, SW_TABLET_MODE, 0);
+		return -1;
+	}
+
 	/* let hid-core decide for the others */
 	return 0;
 }
@@ -1454,11 +1490,21 @@ static int mt_event(struct hid_device *hid, struct hid_field *field,
 {
 	struct mt_device *td = hid_get_drvdata(hid);
 	struct mt_report_data *rdata;
+	struct input_dev *input;
 
 	rdata = mt_find_report_data(td, field->report);
 	if (rdata && rdata->is_mt_collection)
 		return mt_touch_event(hid, field, usage, value);
 
+	if (field->application == MS_TYPE_COVER_APPLICATION &&
+	    rdata->application->quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH &&
+	    usage->hid == MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE) {
+		input = field->hidinput->input;
+		input_report_switch(input, SW_TABLET_MODE, (value & 0xFF) != 0x22);
+		input_sync(input);
+		return 1;
+	}
+
 	return 0;
 }
 
@@ -1611,6 +1657,42 @@ static void mt_post_parse(struct mt_device *td, struct mt_application *app)
 		app->quirks &= ~MT_QUIRK_CONTACT_CNT_ACCURATE;
 }
 
+static int get_type_cover_field(struct hid_report_enum *rep_enum,
+				struct hid_field **field, int usage)
+{
+	struct hid_report *rep;
+	struct hid_field *cur_field;
+	int i, j;
+
+	list_for_each_entry(rep, &rep_enum->report_list, list) {
+		for (i = 0; i < rep->maxfield; i++) {
+			cur_field = rep->field[i];
+			if (cur_field->application != MS_TYPE_COVER_APPLICATION)
+				continue;
+			for (j = 0; j < cur_field->maxusage; j++) {
+				if (cur_field->usage[j].hid == usage) {
+					*field = cur_field;
+					return true;
+				}
+			}
+		}
+	}
+	return false;
+}
+
+static void request_type_cover_tablet_mode_switch(struct hid_device *hdev)
+{
+	struct hid_field *field;
+
+	if (get_type_cover_field(&hdev->report_enum[HID_INPUT_REPORT],
+				 &field,
+				 MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE)) {
+		hid_hw_request(hdev, field->report, HID_REQ_GET_REPORT);
+	} else {
+		hid_err(hdev, "couldn't find tablet mode field\n");
+	}
+}
+
 static int mt_input_configured(struct hid_device *hdev, struct hid_input *hi)
 {
 	struct mt_device *td = hid_get_drvdata(hdev);
@@ -1660,6 +1742,13 @@ static int mt_input_configured(struct hid_device *hdev, struct hid_input *hi)
 		/* force BTN_STYLUS to allow tablet matching in udev */
 		__set_bit(BTN_STYLUS, hi->input->keybit);
 		break;
+	case MS_TYPE_COVER_APPLICATION:
+		if (td->mtclass.quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH) {
+			suffix = "Tablet Mode Switch";
+			request_type_cover_tablet_mode_switch(hdev);
+			break;
+		}
+		fallthrough;
 	default:
 		suffix = "UNKNOWN";
 		break;
@@ -1748,30 +1837,6 @@ static void mt_expired_timeout(struct timer_list *t)
 	clear_bit_unlock(MT_IO_FLAGS_RUNNING, &td->mt_io_flags);
 }
 
-static void get_type_cover_backlight_field(struct hid_device *hdev,
-					   struct hid_field **field)
-{
-	struct hid_report_enum *rep_enum;
-	struct hid_report *rep;
-	struct hid_field *cur_field;
-	int i, j;
-
-	rep_enum = &hdev->report_enum[HID_FEATURE_REPORT];
-	list_for_each_entry(rep, &rep_enum->report_list, list) {
-		for (i = 0; i < rep->maxfield; i++) {
-			cur_field = rep->field[i];
-
-			for (j = 0; j < cur_field->maxusage; j++) {
-				if (cur_field->usage[j].hid
-				    == MS_TYPE_COVER_FEATURE_REPORT_USAGE) {
-					*field = cur_field;
-					return;
-				}
-			}
-		}
-	}
-}
-
 static void update_keyboard_backlight(struct hid_device *hdev, bool enabled)
 {
 	struct usb_device *udev = hid_to_usb_dev(hdev);
@@ -1780,8 +1845,9 @@ static void update_keyboard_backlight(struct hid_device *hdev, bool enabled)
 	/* Wake up the device in case it's already suspended */
 	pm_runtime_get_sync(&udev->dev);
 
-	get_type_cover_backlight_field(hdev, &field);
-	if (!field) {
+	if (!get_type_cover_field(&hdev->report_enum[HID_FEATURE_REPORT],
+				  &field,
+				  MS_TYPE_COVER_FEATURE_REPORT_USAGE)) {
 		hid_err(hdev, "couldn't find backlight field\n");
 		goto out;
 	}
@@ -1916,13 +1982,24 @@ static int mt_suspend(struct hid_device *hdev, pm_message_t state)
 
 static int mt_reset_resume(struct hid_device *hdev)
 {
+	struct mt_device *td = hid_get_drvdata(hdev);
+
 	mt_release_contacts(hdev);
 	mt_set_modes(hdev, HID_LATENCY_NORMAL, true, true);
+
+	/* Request an update on the typecover folding state on resume
+	 * after reset.
+	 */
+	if (td->mtclass.quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH)
+		request_type_cover_tablet_mode_switch(hdev);
+
 	return 0;
 }
 
 static int mt_resume(struct hid_device *hdev)
 {
+	struct mt_device *td = hid_get_drvdata(hdev);
+
 	/* Some Elan legacy devices require SET_IDLE to be set on resume.
 	 * It should be safe to send it to other devices too.
 	 * Tested on 3M, Stantum, Cypress, Zytronic, eGalax, and Elan panels. */
@@ -1931,6 +2008,10 @@ static int mt_resume(struct hid_device *hdev)
 
 	mt_set_modes(hdev, HID_LATENCY_NORMAL, true, true);
 
+	/* Request an update on the typecover folding state on resume. */
+	if (td->mtclass.quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH)
+		request_type_cover_tablet_mode_switch(hdev);
+
 	return 0;
 }
 #endif
@@ -1938,6 +2019,21 @@ static int mt_resume(struct hid_device *hdev)
 static void mt_remove(struct hid_device *hdev)
 {
 	struct mt_device *td = hid_get_drvdata(hdev);
+	struct hid_field *field;
+	struct input_dev *input;
+
+	/* Reset tablet mode switch on disconnect. */
+	if (td->mtclass.quirks & MT_QUIRK_HAS_TYPE_COVER_TABLET_MODE_SWITCH) {
+		if (get_type_cover_field(&hdev->report_enum[HID_INPUT_REPORT],
+					 &field,
+					 MS_TYPE_COVER_TABLET_MODE_SWITCH_USAGE)) {
+			input = field->hidinput->input;
+			input_report_switch(input, SW_TABLET_MODE, 0);
+			input_sync(input);
+		} else {
+			hid_err(hdev, "couldn't find tablet mode field\n");
+		}
+	}
 
 	unregister_pm_notifier(&td->pm_notifier);
 	del_timer_sync(&td->release_timer);
-- 
2.39.2

From a1d8dbed4b776b6b979deeb0f8c3010bd19369fe Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sun, 19 Feb 2023 22:12:24 +0100
Subject: [PATCH] PCI: Add quirk to prevent calling shutdown mehtod

Work around buggy EFI firmware: On some Microsoft Surface devices
(Surface Pro 9 and Surface Laptop 5) the EFI ResetSystem call with
EFI_RESET_SHUTDOWN doesn't function properly. Instead of shutting the
system down, it returns and the system stays on.

It turns out that this only happens after PCI shutdown callbacks ran for
specific devices. Excluding those devices from the shutdown process
makes the ResetSystem call work as expected.

TODO: Maybe we can find a better way or the root cause of this?

Not-Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-shutdown
---
 drivers/pci/pci-driver.c |  3 +++
 drivers/pci/quirks.c     | 36 ++++++++++++++++++++++++++++++++++++
 include/linux/pci.h      |  1 +
 3 files changed, 40 insertions(+)

diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index f47a3b10bf504..9ea4352aac56c 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -507,6 +507,9 @@ static void pci_device_shutdown(struct device *dev)
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct pci_driver *drv = pci_dev->driver;
 
+	if (pci_dev->no_shutdown)
+		return;
+
 	pm_runtime_resume(dev);
 
 	if (drv && drv->shutdown)
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 494fa46f57671..106fb2ff855b2 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -6015,3 +6015,39 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x9a2d, dpc_log_size);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x9a2f, dpc_log_size);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x9a31, dpc_log_size);
 #endif
+
+static const struct dmi_system_id no_shutdown_dmi_table[] = {
+	/*
+	 * Systems on which some devices should not be touched during shutdown.
+	 */
+	{
+		.ident = "Microsoft Surface Pro 9",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Surface Pro 9"),
+		},
+	},
+	{
+		.ident = "Microsoft Surface Laptop 5",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Surface Laptop 5"),
+		},
+	},
+	{}
+};
+
+static void quirk_no_shutdown(struct pci_dev *dev)
+{
+	if (!dmi_check_system(no_shutdown_dmi_table))
+		return;
+
+	dev->no_shutdown = 1;
+	pci_info(dev, "disabling shutdown ops for [%04x:%04x]\n",
+		 dev->vendor, dev->device);
+}
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, 0x461e, quirk_no_shutdown);  // Thunderbolt 4 USB Controller
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, 0x461f, quirk_no_shutdown);  // Thunderbolt 4 PCI Express Root Port
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, 0x462f, quirk_no_shutdown);  // Thunderbolt 4 PCI Express Root Port
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, 0x466d, quirk_no_shutdown);  // Thunderbolt 4 NHI
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, 0x46a8, quirk_no_shutdown);  // GPU
diff --git a/include/linux/pci.h b/include/linux/pci.h
index cb538bc579710..409925d303ed7 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -462,6 +462,7 @@ struct pci_dev {
 	unsigned int	no_vf_scan:1;		/* Don't scan for VFs after IOV enablement */
 	unsigned int	no_command_memory:1;	/* No PCI_COMMAND_MEMORY */
 	unsigned int	rom_bar_overlap:1;	/* ROM BAR disable broken */
+	unsigned int	no_shutdown:1;		/* Do not touch device on shutdown */
 	pci_dev_flags_t dev_flags;
 	atomic_t	enable_cnt;	/* pci_enable_device has been called */
 
-- 
2.39.2

From 23fb72a959d8cb8842ea87ee7705427e3c1d6ab9 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Sun, 12 Mar 2023 01:41:57 +0100
Subject: [PATCH] platform/surface: gpe: Add support for Surface Pro 9

Add the lid GPE used by the Surface Pro 9.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: surface-gpe
---
 drivers/platform/surface/surface_gpe.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/drivers/platform/surface/surface_gpe.c b/drivers/platform/surface/surface_gpe.c
index c219b840d491a..69c4352e8406b 100644
--- a/drivers/platform/surface/surface_gpe.c
+++ b/drivers/platform/surface/surface_gpe.c
@@ -41,6 +41,11 @@ static const struct property_entry lid_device_props_l4F[] = {
 	{},
 };
 
+static const struct property_entry lid_device_props_l52[] = {
+	PROPERTY_ENTRY_U32("gpe", 0x52),
+	{},
+};
+
 static const struct property_entry lid_device_props_l57[] = {
 	PROPERTY_ENTRY_U32("gpe", 0x57),
 	{},
@@ -107,6 +112,18 @@ static const struct dmi_system_id dmi_lid_device_table[] = {
 		},
 		.driver_data = (void *)lid_device_props_l4B,
 	},
+	{
+		/*
+		 * We match for SKU here due to product name clash with the ARM
+		 * version.
+		 */
+		.ident = "Surface Pro 9",
+		.matches = {
+			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
+			DMI_EXACT_MATCH(DMI_PRODUCT_SKU, "Surface_Pro_9_2038"),
+		},
+		.driver_data = (void *)lid_device_props_l52,
+	},
 	{
 		.ident = "Surface Book 1",
 		.matches = {
-- 
2.39.2

From 033ef8c5b7ad4308e3ac076e35256488491046e3 Mon Sep 17 00:00:00 2001
From: Hans de Goede <hdegoede@redhat.com>
Date: Sun, 10 Oct 2021 20:56:57 +0200
Subject: [PATCH] ACPI: delay enumeration of devices with a _DEP pointing to an
 INT3472 device

The clk and regulator frameworks expect clk/regulator consumer-devices
to have info about the consumed clks/regulators described in the device's
fw_node.

To work around cases where this info is not present in the firmware tables,
which is often the case on x86/ACPI devices, both frameworks allow the
provider-driver to attach info about consumers to the clks/regulators
when registering these.

This causes problems with the probe ordering wrt drivers for consumers
of these clks/regulators. Since the lookups are only registered when the
provider-driver binds, trying to get these clks/regulators before then
results in a -ENOENT error for clks and a dummy regulator for regulators.

One case where we hit this issue is camera sensors such as e.g. the OV8865
sensor found on the Microsoft Surface Go. The sensor uses clks, regulators
and GPIOs provided by a TPS68470 PMIC which is described in an INT3472
ACPI device. There is special platform code handling this and setting
platform_data with the necessary consumer info on the MFD cells
instantiated for the PMIC under: drivers/platform/x86/intel/int3472.

For this to work properly the ov8865 driver must not bind to the I2C-client
for the OV8865 sensor until after the TPS68470 PMIC gpio, regulator and
clk MFD cells have all been fully setup.

The OV8865 on the Microsoft Surface Go is just one example, all X86
devices using the Intel IPU3 camera block found on recent Intel SoCs
have similar issues where there is an INT3472 HID ACPI-device, which
describes the clks and regulators, and the driver for this INT3472 device
must be fully initialized before the sensor driver (any sensor driver)
binds for things to work properly.

On these devices the ACPI nodes describing the sensors all have a _DEP
dependency on the matching INT3472 ACPI device (there is one per sensor).

This allows solving the probe-ordering problem by delaying the enumeration
(instantiation of the I2C-client in the ov8865 example) of ACPI-devices
which have a _DEP dependency on an INT3472 device.

The new acpi_dev_ready_for_enumeration() helper used for this is also
exported because for devices, which have the enumeration_by_parent flag
set, the parent-driver will do its own scan of child ACPI devices and
it will try to enumerate those during its probe(). Code doing this such
as e.g. the i2c-core-acpi.c code must call this new helper to ensure
that it too delays the enumeration until all the _DEP dependencies are
met on devices which have the new honor_deps flag set.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Patchset: cameras
---
 drivers/acpi/scan.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
index dbfa58e799e28..ccbd3f8b523e8 100644
--- a/drivers/acpi/scan.c
+++ b/drivers/acpi/scan.c
@@ -2106,6 +2106,9 @@ static acpi_status acpi_bus_check_add_2(acpi_handle handle, u32 lvl_not_used,
 
 static void acpi_default_enumeration(struct acpi_device *device)
 {
+	if (!acpi_dev_ready_for_enumeration(device))
+		return;
+
 	/*
 	 * Do not enumerate devices with enumeration_by_parent flag set as
 	 * they will be enumerated by their respective parents.
-- 
2.39.2

From 0b559e61d232d1fd2dfd6ebcf3a1626a1f434531 Mon Sep 17 00:00:00 2001
From: zouxiaoh <xiaohong.zou@intel.com>
Date: Fri, 25 Jun 2021 08:52:59 +0800
Subject: [PATCH] iommu: intel-ipu: use IOMMU passthrough mode for Intel IPUs

Intel IPU(Image Processing Unit) has its own (IO)MMU hardware,
The IPU driver allocates its own page table that is not mapped
via the DMA, and thus the Intel IOMMU driver blocks access giving
this error: DMAR: DRHD: handling fault status reg 3 DMAR:
[DMA Read] Request device [00:05.0] PASID ffffffff
fault addr 76406000 [fault reason 06] PTE Read access is not set
As IPU is not an external facing device which is not risky, so use
IOMMU passthrough mode for Intel IPUs.

Change-Id: I6dcccdadac308cf42e20a18e1b593381391e3e6b
Depends-On: Iacd67578e8c6a9b9ac73285f52b4081b72fb68a6
Tracked-On: #JIITL8-411
Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: zouxiaoh <xiaohong.zou@intel.com>
Signed-off-by: Xu Chongyang <chongyang.xu@intel.com>
Patchset: cameras
---
 drivers/iommu/intel/iommu.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 3965979c7bd41..509cfb9f66954 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -37,6 +37,12 @@
 #define IS_GFX_DEVICE(pdev) ((pdev->class >> 16) == PCI_BASE_CLASS_DISPLAY)
 #define IS_USB_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_SERIAL_USB)
 #define IS_ISA_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA)
+#define IS_INTEL_IPU(pdev) ((pdev)->vendor == PCI_VENDOR_ID_INTEL &&	\
+			   ((pdev)->device == 0x9a19 ||		\
+			    (pdev)->device == 0x9a39 ||		\
+			    (pdev)->device == 0x4e19 ||		\
+			    (pdev)->device == 0x465d ||		\
+			    (pdev)->device == 0x1919))
 #define IS_IPTS(pdev) ((pdev)->vendor == PCI_VENDOR_ID_INTEL &&	\
 			    ((pdev)->device == 0x9d3e))
 #define IS_AZALIA(pdev) ((pdev)->vendor == 0x8086 && (pdev)->device == 0x3a3e)
@@ -289,12 +295,14 @@ EXPORT_SYMBOL_GPL(intel_iommu_enabled);
 
 static int dmar_map_gfx = 1;
 static int dmar_map_ipts = 1;
+static int dmar_map_ipu = 1;
 static int intel_iommu_superpage = 1;
 static int iommu_identity_mapping;
 static int iommu_skip_te_disable;
 
 #define IDENTMAP_GFX		2
 #define IDENTMAP_AZALIA		4
+#define IDENTMAP_IPU		8
 #define IDENTMAP_IPTS		16
 
 const struct iommu_ops intel_iommu_ops;
@@ -2635,6 +2643,9 @@ static int device_def_domain_type(struct device *dev)
 		if ((iommu_identity_mapping & IDENTMAP_GFX) && IS_GFX_DEVICE(pdev))
 			return IOMMU_DOMAIN_IDENTITY;
 
+		if ((iommu_identity_mapping & IDENTMAP_IPU) && IS_INTEL_IPU(pdev))
+			return IOMMU_DOMAIN_IDENTITY;
+
 		if ((iommu_identity_mapping & IDENTMAP_IPTS) && IS_IPTS(pdev))
 			return IOMMU_DOMAIN_IDENTITY;
 	}
@@ -3026,6 +3037,9 @@ static int __init init_dmars(void)
 	if (!dmar_map_gfx)
 		iommu_identity_mapping |= IDENTMAP_GFX;
 
+	if (!dmar_map_ipu)
+		iommu_identity_mapping |= IDENTMAP_IPU;
+
 	if (!dmar_map_ipts)
 		iommu_identity_mapping |= IDENTMAP_IPTS;
 
@@ -4798,6 +4812,18 @@ static void quirk_iommu_igfx(struct pci_dev *dev)
 	dmar_map_gfx = 0;
 }
 
+static void quirk_iommu_ipu(struct pci_dev *dev)
+{
+	if (!IS_INTEL_IPU(dev))
+		return;
+
+	if (risky_device(dev))
+		return;
+
+	pci_info(dev, "Passthrough IOMMU for integrated Intel IPU\n");
+	dmar_map_ipu = 0;
+}
+
 static void quirk_iommu_ipts(struct pci_dev *dev)
 {
 	if (!IS_IPTS(dev))
@@ -4809,6 +4835,7 @@ static void quirk_iommu_ipts(struct pci_dev *dev)
 	pci_info(dev, "Passthrough IOMMU for IPTS\n");
 	dmar_map_ipts = 0;
 }
+
 /* G4x/GM45 integrated gfx dmar support is totally busted. */
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2a40, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2e00, quirk_iommu_igfx);
@@ -4844,6 +4871,9 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x1632, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x163A, quirk_iommu_igfx);
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x163D, quirk_iommu_igfx);
 
+/* disable IPU dmar support */
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_ANY_ID, quirk_iommu_ipu);
+
 /* disable IPTS dmar support */
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x9D3E, quirk_iommu_ipts);
 
-- 
2.39.2

From 26b6001d2f0738dc69212b714b4cfe5ad5c769da Mon Sep 17 00:00:00 2001
From: Daniel Scally <djrscally@gmail.com>
Date: Sun, 10 Oct 2021 20:57:02 +0200
Subject: [PATCH] platform/x86: int3472: Enable I2c daisy chain

The TPS68470 PMIC has an I2C passthrough mode through which I2C traffic
can be forwarded to a device connected to the PMIC as though it were
connected directly to the system bus. Enable this mode when the chip
is initialised.

Signed-off-by: Daniel Scally <djrscally@gmail.com>
Patchset: cameras
---
 drivers/platform/x86/intel/int3472/tps68470.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/platform/x86/intel/int3472/tps68470.c b/drivers/platform/x86/intel/int3472/tps68470.c
index 5b8d1a9620a5d..6a0ff035cf209 100644
--- a/drivers/platform/x86/intel/int3472/tps68470.c
+++ b/drivers/platform/x86/intel/int3472/tps68470.c
@@ -46,6 +46,13 @@ static int tps68470_chip_init(struct device *dev, struct regmap *regmap)
 		return ret;
 	}
 
+	/* Enable I2C daisy chain */
+	ret = regmap_write(regmap, TPS68470_REG_S_I2C_CTL, 0x03);
+	if (ret) {
+		dev_err(dev, "Failed to enable i2c daisy chain\n");
+		return ret;
+	}
+
 	dev_info(dev, "TPS68470 REVID: 0x%02x\n", version);
 
 	return 0;
-- 
2.39.2

From b71265fab43356eada571af1390deb3ac5fc238a Mon Sep 17 00:00:00 2001
From: Daniel Scally <djrscally@gmail.com>
Date: Thu, 28 Oct 2021 21:55:16 +0100
Subject: [PATCH] media: i2c: Add driver for DW9719 VCM

Add a driver for the DW9719 VCM. The driver creates a v4l2 subdevice
and registers a control to set the desired focus.

Signed-off-by: Daniel Scally <djrscally@gmail.com>
Patchset: cameras
---
 MAINTAINERS                |   7 +
 drivers/media/i2c/Kconfig  |  11 +
 drivers/media/i2c/Makefile |   1 +
 drivers/media/i2c/dw9719.c | 425 +++++++++++++++++++++++++++++++++++++
 4 files changed, 444 insertions(+)
 create mode 100644 drivers/media/i2c/dw9719.c

diff --git a/MAINTAINERS b/MAINTAINERS
index 350d7e3ba94f9..ce7d8d9c83cb1 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -6286,6 +6286,13 @@ T:	git git://linuxtv.org/media_tree.git
 F:	Documentation/devicetree/bindings/media/i2c/dongwoon,dw9714.yaml
 F:	drivers/media/i2c/dw9714.c
 
+DONGWOON DW9719 LENS VOICE COIL DRIVER
+M:	Daniel Scally <djrscally@gmail.com>
+L:	linux-media@vger.kernel.org
+S:	Maintained
+T:	git git://linuxtv.org/media_tree.git
+F:	drivers/media/i2c/dw9719.c
+
 DONGWOON DW9768 LENS VOICE COIL DRIVER
 M:	Dongchun Zhu <dongchun.zhu@mediatek.com>
 L:	linux-media@vger.kernel.org
diff --git a/drivers/media/i2c/Kconfig b/drivers/media/i2c/Kconfig
index 7806d4b81716e..98d081efeef7f 100644
--- a/drivers/media/i2c/Kconfig
+++ b/drivers/media/i2c/Kconfig
@@ -821,6 +821,17 @@ config VIDEO_DW9714
 	  capability. This is designed for linear control of
 	  voice coil motors, controlled via I2C serial interface.
 
+config VIDEO_DW9719
+	tristate "DW9719 lens voice coil support"
+	depends on I2C && VIDEO_V4L2
+	select MEDIA_CONTROLLER
+	select VIDEO_V4L2_SUBDEV_API
+	select V4L2_ASYNC
+	help
+	  This is a driver for the DW9719 camera lens voice coil.
+	  This is designed for linear control of  voice coil motors,
+	  controlled via I2C serial interface.
+
 config VIDEO_DW9768
 	tristate "DW9768 lens voice coil support"
 	depends on I2C && VIDEO_DEV
diff --git a/drivers/media/i2c/Makefile b/drivers/media/i2c/Makefile
index 0a2933103dd92..b82a07c763880 100644
--- a/drivers/media/i2c/Makefile
+++ b/drivers/media/i2c/Makefile
@@ -30,6 +30,7 @@ obj-$(CONFIG_VIDEO_CS5345) += cs5345.o
 obj-$(CONFIG_VIDEO_CS53L32A) += cs53l32a.o
 obj-$(CONFIG_VIDEO_CX25840) += cx25840/
 obj-$(CONFIG_VIDEO_DW9714) += dw9714.o
+obj-$(CONFIG_VIDEO_DW9719) += dw9719.o
 obj-$(CONFIG_VIDEO_DW9768) += dw9768.o
 obj-$(CONFIG_VIDEO_DW9807_VCM) += dw9807-vcm.o
 obj-$(CONFIG_VIDEO_ET8EK8) += et8ek8/
diff --git a/drivers/media/i2c/dw9719.c b/drivers/media/i2c/dw9719.c
new file mode 100644
index 0000000000000..180b04d2a6b3a
--- /dev/null
+++ b/drivers/media/i2c/dw9719.c
@@ -0,0 +1,425 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2012 Intel Corporation
+
+/*
+ * Based on linux/modules/camera/drivers/media/i2c/imx/dw9719.c in this repo:
+ * https://github.com/ZenfoneArea/android_kernel_asus_zenfone5
+ */
+
+#include <asm/unaligned.h>
+
+#include <linux/delay.h>
+#include <linux/i2c.h>
+#include <linux/pm_runtime.h>
+#include <linux/regulator/consumer.h>
+#include <linux/types.h>
+
+#include <media/v4l2-common.h>
+#include <media/v4l2-ctrls.h>
+#include <media/v4l2-subdev.h>
+
+#define DW9719_MAX_FOCUS_POS	1023
+#define DW9719_CTRL_STEPS	16
+#define DW9719_CTRL_DELAY_US	1000
+#define DELAY_MAX_PER_STEP_NS	(1000000 * 1023)
+
+#define DW9719_INFO			0
+#define DW9719_ID			0xF1
+#define DW9719_CONTROL			2
+#define DW9719_VCM_CURRENT		3
+
+#define DW9719_MODE			6
+#define DW9719_VCM_FREQ			7
+
+#define DW9719_MODE_SAC3		0x40
+#define DW9719_DEFAULT_VCM_FREQ		0x60
+#define DW9719_ENABLE_RINGING		0x02
+
+#define NUM_REGULATORS			2
+
+#define to_dw9719_device(x) container_of(x, struct dw9719_device, sd)
+
+struct dw9719_device {
+	struct device *dev;
+	struct i2c_client *client;
+	struct regulator_bulk_data regulators[NUM_REGULATORS];
+	struct v4l2_subdev sd;
+
+	struct dw9719_v4l2_ctrls {
+		struct v4l2_ctrl_handler handler;
+		struct v4l2_ctrl *focus;
+	} ctrls;
+};
+
+static int dw9719_i2c_rd8(struct i2c_client *client, u8 reg, u8 *val)
+{
+	struct i2c_msg msg[2];
+	u8 buf[2] = { reg };
+	int ret;
+
+	msg[0].addr = client->addr;
+	msg[0].flags = 0;
+	msg[0].len = 1;
+	msg[0].buf = buf;
+
+	msg[1].addr = client->addr;
+	msg[1].flags = I2C_M_RD;
+	msg[1].len = 1;
+	msg[1].buf = &buf[1];
+	*val = 0;
+
+	ret = i2c_transfer(client->adapter, msg, 2);
+	if (ret < 0)
+		return ret;
+
+	*val = buf[1];
+
+	return 0;
+}
+
+static int dw9719_i2c_wr8(struct i2c_client *client, u8 reg, u8 val)
+{
+	struct i2c_msg msg;
+	int ret;
+
+	u8 buf[2] = { reg, val };
+
+	msg.addr = client->addr;
+	msg.flags = 0;
+	msg.len = sizeof(buf);
+	msg.buf = buf;
+
+	ret = i2c_transfer(client->adapter, &msg, 1);
+
+	return ret < 0 ? ret : 0;
+}
+
+static int dw9719_i2c_wr16(struct i2c_client *client, u8 reg, u16 val)
+{
+	struct i2c_msg msg;
+	u8 buf[3] = { reg };
+	int ret;
+
+	put_unaligned_be16(val, buf + 1);
+
+	msg.addr = client->addr;
+	msg.flags = 0;
+	msg.len = sizeof(buf);
+	msg.buf = buf;
+
+	ret = i2c_transfer(client->adapter, &msg, 1);
+
+	return ret < 0 ? ret : 0;
+}
+
+static int dw9719_detect(struct dw9719_device *dw9719)
+{
+	int ret;
+	u8 val;
+
+	ret = dw9719_i2c_rd8(dw9719->client, DW9719_INFO, &val);
+	if (ret < 0)
+		return ret;
+
+	if (val != DW9719_ID) {
+		dev_err(dw9719->dev, "Failed to detect correct id\n");
+		ret = -ENXIO;
+	}
+
+	return 0;
+}
+
+static int dw9719_power_down(struct dw9719_device *dw9719)
+{
+	return regulator_bulk_disable(NUM_REGULATORS, dw9719->regulators);
+}
+
+static int dw9719_power_up(struct dw9719_device *dw9719)
+{
+	int ret;
+
+	ret = regulator_bulk_enable(NUM_REGULATORS, dw9719->regulators);
+	if (ret)
+		return ret;
+
+	/* Jiggle SCL pin to wake up device */
+	ret = dw9719_i2c_wr8(dw9719->client, DW9719_CONTROL, 1);
+
+	/* Need 100us to transit from SHUTDOWN to STANDBY*/
+	usleep_range(100, 1000);
+
+	ret = dw9719_i2c_wr8(dw9719->client, DW9719_CONTROL,
+			     DW9719_ENABLE_RINGING);
+	if (ret < 0)
+		goto fail_powerdown;
+
+	ret = dw9719_i2c_wr8(dw9719->client, DW9719_MODE, DW9719_MODE_SAC3);
+	if (ret < 0)
+		goto fail_powerdown;
+
+	ret = dw9719_i2c_wr8(dw9719->client, DW9719_VCM_FREQ,
+			     DW9719_DEFAULT_VCM_FREQ);
+	if (ret < 0)
+		goto fail_powerdown;
+
+	return 0;
+
+fail_powerdown:
+	dw9719_power_down(dw9719);
+	return ret;
+}
+
+static int dw9719_t_focus_abs(struct dw9719_device *dw9719, s32 value)
+{
+	int ret;
+
+	value = clamp(value, 0, DW9719_MAX_FOCUS_POS);
+	ret = dw9719_i2c_wr16(dw9719->client, DW9719_VCM_CURRENT, value);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static int dw9719_set_ctrl(struct v4l2_ctrl *ctrl)
+{
+	struct dw9719_device *dw9719 = container_of(ctrl->handler,
+						    struct dw9719_device,
+						    ctrls.handler);
+	int ret;
+
+	/* Only apply changes to the controls if the device is powered up */
+	if (!pm_runtime_get_if_in_use(dw9719->dev))
+		return 0;
+
+	switch (ctrl->id) {
+	case V4L2_CID_FOCUS_ABSOLUTE:
+		ret = dw9719_t_focus_abs(dw9719, ctrl->val);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	pm_runtime_put(dw9719->dev);
+
+	return ret;
+}
+
+static const struct v4l2_ctrl_ops dw9719_ctrl_ops = {
+	.s_ctrl = dw9719_set_ctrl,
+};
+
+static int __maybe_unused dw9719_suspend(struct device *dev)
+{
+	struct v4l2_subdev *sd = dev_get_drvdata(dev);
+	struct dw9719_device *dw9719 = to_dw9719_device(sd);
+	int ret;
+	int val;
+
+	for (val = dw9719->ctrls.focus->val; val >= 0;
+	     val -= DW9719_CTRL_STEPS) {
+		ret = dw9719_t_focus_abs(dw9719, val);
+		if (ret)
+			return ret;
+
+		usleep_range(DW9719_CTRL_DELAY_US, DW9719_CTRL_DELAY_US + 10);
+	}
+
+	return dw9719_power_down(dw9719);
+}
+
+static int __maybe_unused dw9719_resume(struct device *dev)
+{
+	struct v4l2_subdev *sd = dev_get_drvdata(dev);
+	struct dw9719_device *dw9719 = to_dw9719_device(sd);
+	int current_focus = dw9719->ctrls.focus->val;
+	int ret;
+	int val;
+
+	ret = dw9719_power_up(dw9719);
+	if (ret)
+		return ret;
+
+	for (val = current_focus % DW9719_CTRL_STEPS; val < current_focus;
+	     val += DW9719_CTRL_STEPS) {
+		ret = dw9719_t_focus_abs(dw9719, val);
+		if (ret)
+			goto err_power_down;
+
+		usleep_range(DW9719_CTRL_DELAY_US, DW9719_CTRL_DELAY_US + 10);
+	}
+
+	return 0;
+
+err_power_down:
+	dw9719_power_down(dw9719);
+	return ret;
+}
+
+static int dw9719_open(struct v4l2_subdev *sd, struct v4l2_subdev_fh *fh)
+{
+	return pm_runtime_resume_and_get(sd->dev);
+}
+
+static int dw9719_close(struct v4l2_subdev *sd, struct v4l2_subdev_fh *fh)
+{
+	pm_runtime_put(sd->dev);
+
+	return 0;
+}
+
+static const struct v4l2_subdev_internal_ops dw9719_internal_ops = {
+	.open = dw9719_open,
+	.close = dw9719_close,
+};
+
+static int dw9719_init_controls(struct dw9719_device *dw9719)
+{
+	const struct v4l2_ctrl_ops *ops = &dw9719_ctrl_ops;
+	int ret;
+
+	ret = v4l2_ctrl_handler_init(&dw9719->ctrls.handler, 1);
+	if (ret)
+		return ret;
+
+	dw9719->ctrls.focus = v4l2_ctrl_new_std(&dw9719->ctrls.handler, ops,
+						V4L2_CID_FOCUS_ABSOLUTE, 0,
+						DW9719_MAX_FOCUS_POS, 1, 0);
+
+	if (dw9719->ctrls.handler.error) {
+		dev_err(dw9719->dev, "Error initialising v4l2 ctrls\n");
+		ret = dw9719->ctrls.handler.error;
+		goto err_free_handler;
+	}
+
+	dw9719->sd.ctrl_handler = &dw9719->ctrls.handler;
+
+	return ret;
+
+err_free_handler:
+	v4l2_ctrl_handler_free(&dw9719->ctrls.handler);
+	return ret;
+}
+
+static const struct v4l2_subdev_ops dw9719_ops = { };
+
+static int dw9719_probe(struct i2c_client *client)
+{
+	struct dw9719_device *dw9719;
+	int ret;
+
+	dw9719 = devm_kzalloc(&client->dev, sizeof(*dw9719), GFP_KERNEL);
+	if (!dw9719)
+		return -ENOMEM;
+
+	dw9719->client = client;
+	dw9719->dev = &client->dev;
+
+	dw9719->regulators[0].supply = "vdd";
+	/*
+	 * The DW9719 has only the 1 VDD voltage input, but some PMICs such as
+	 * the TPS68470 PMIC have I2C passthrough capability, to disconnect the
+	 * sensor's I2C pins from the I2C bus when the sensors VSIO (Sensor-IO)
+	 * is off, because some sensors then short these pins to ground;
+	 * and the DW9719 might sit behind this passthrough, this it needs to
+	 * enable VSIO as that will also enable the I2C passthrough.
+	 */
+	dw9719->regulators[1].supply = "vsio";
+
+	ret = devm_regulator_bulk_get(&client->dev, NUM_REGULATORS,
+				      dw9719->regulators);
+	if (ret)
+		return dev_err_probe(&client->dev, ret, "getting regulators\n");
+
+	v4l2_i2c_subdev_init(&dw9719->sd, client, &dw9719_ops);
+	dw9719->sd.flags |= V4L2_SUBDEV_FL_HAS_DEVNODE;
+	dw9719->sd.internal_ops = &dw9719_internal_ops;
+
+	ret = dw9719_init_controls(dw9719);
+	if (ret)
+		return ret;
+
+	ret = media_entity_pads_init(&dw9719->sd.entity, 0, NULL);
+	if (ret < 0)
+		goto err_free_ctrl_handler;
+
+	dw9719->sd.entity.function = MEDIA_ENT_F_LENS;
+
+	/*
+	 * We need the driver to work in the event that pm runtime is disable in
+	 * the kernel, so power up and verify the chip now. In the event that
+	 * runtime pm is disabled this will leave the chip on, so that the lens
+	 * will work.
+	 */
+
+	ret = dw9719_power_up(dw9719);
+	if (ret)
+		goto err_cleanup_media;
+
+	ret = dw9719_detect(dw9719);
+	if (ret)
+		goto err_powerdown;
+
+	pm_runtime_set_active(&client->dev);
+	pm_runtime_get_noresume(&client->dev);
+	pm_runtime_enable(&client->dev);
+
+	ret = v4l2_async_register_subdev(&dw9719->sd);
+	if (ret < 0)
+		goto err_pm_runtime;
+
+	pm_runtime_set_autosuspend_delay(&client->dev, 1000);
+	pm_runtime_use_autosuspend(&client->dev);
+	pm_runtime_put_autosuspend(&client->dev);
+
+	return ret;
+
+err_pm_runtime:
+	pm_runtime_disable(&client->dev);
+	pm_runtime_put_noidle(&client->dev);
+err_powerdown:
+	dw9719_power_down(dw9719);
+err_cleanup_media:
+	media_entity_cleanup(&dw9719->sd.entity);
+err_free_ctrl_handler:
+	v4l2_ctrl_handler_free(&dw9719->ctrls.handler);
+
+	return ret;
+}
+
+static void dw9719_remove(struct i2c_client *client)
+{
+	struct v4l2_subdev *sd = i2c_get_clientdata(client);
+	struct dw9719_device *dw9719 = container_of(sd, struct dw9719_device,
+						    sd);
+
+	pm_runtime_disable(&client->dev);
+	v4l2_async_unregister_subdev(sd);
+	v4l2_ctrl_handler_free(&dw9719->ctrls.handler);
+	media_entity_cleanup(&dw9719->sd.entity);
+}
+
+static const struct i2c_device_id dw9719_id_table[] = {
+	{ "dw9719" },
+	{ }
+};
+MODULE_DEVICE_TABLE(i2c, dw9719_id_table);
+
+static const struct dev_pm_ops dw9719_pm_ops = {
+	SET_RUNTIME_PM_OPS(dw9719_suspend, dw9719_resume, NULL)
+};
+
+static struct i2c_driver dw9719_i2c_driver = {
+	.driver = {
+		.name = "dw9719",
+		.pm = &dw9719_pm_ops,
+	},
+	.probe_new = dw9719_probe,
+	.remove = dw9719_remove,
+	.id_table = dw9719_id_table,
+};
+module_i2c_driver(dw9719_i2c_driver);
+
+MODULE_AUTHOR("Daniel Scally <djrscally@gmail.com>");
+MODULE_DESCRIPTION("DW9719 VCM Driver");
+MODULE_LICENSE("GPL");
-- 
2.39.2

From 0ee02b001c6e806cca3e7eeef5bf0608e7580436 Mon Sep 17 00:00:00 2001
From: Daniel Scally <djrscally@gmail.com>
Date: Wed, 4 May 2022 23:21:45 +0100
Subject: [PATCH] media: ipu3-cio2: Move functionality from .complete() to
 .bound()

Creating links and registering subdev nodes during the .complete()
callback has the unfortunate effect of preventing all cameras that
connect to a notifier from working if any one of their drivers fails
to probe. Moving the functionality from .complete() to .bound() allows
those camera sensor drivers that did probe correctly to work regardless.

Signed-off-by: Daniel Scally <djrscally@gmail.com>
Patchset: cameras
---
 drivers/media/pci/intel/ipu3/ipu3-cio2-main.c | 65 +++++++------------
 1 file changed, 23 insertions(+), 42 deletions(-)

diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
index 3b76a9d0383a8..38f9f4da1922e 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
@@ -1383,7 +1383,10 @@ static int cio2_notifier_bound(struct v4l2_async_notifier *notifier,
 {
 	struct cio2_device *cio2 = to_cio2_device(notifier);
 	struct sensor_async_subdev *s_asd = to_sensor_asd(asd);
+	struct device *dev = &cio2->pci_dev->dev;
 	struct cio2_queue *q;
+	unsigned int pad;
+	int ret;
 
 	if (cio2->queue[s_asd->csi2.port].sensor)
 		return -EBUSY;
@@ -1394,7 +1397,26 @@ static int cio2_notifier_bound(struct v4l2_async_notifier *notifier,
 	q->sensor = sd;
 	q->csi_rx_base = cio2->base + CIO2_REG_PIPE_BASE(q->csi2.port);
 
-	return 0;
+	for (pad = 0; pad < q->sensor->entity.num_pads; pad++)
+		if (q->sensor->entity.pads[pad].flags &
+					MEDIA_PAD_FL_SOURCE)
+			break;
+
+	if (pad == q->sensor->entity.num_pads) {
+		dev_err(dev, "failed to find src pad for %s\n",
+			q->sensor->name);
+		return -ENXIO;
+	}
+
+	ret = media_create_pad_link(&q->sensor->entity, pad, &q->subdev.entity,
+				    CIO2_PAD_SINK, 0);
+	if (ret) {
+		dev_err(dev, "failed to create link for %s\n",
+			q->sensor->name);
+		return ret;
+	}
+
+	return v4l2_device_register_subdev_nodes(&cio2->v4l2_dev);
 }
 
 /* The .unbind callback */
@@ -1408,50 +1430,9 @@ static void cio2_notifier_unbind(struct v4l2_async_notifier *notifier,
 	cio2->queue[s_asd->csi2.port].sensor = NULL;
 }
 
-/* .complete() is called after all subdevices have been located */
-static int cio2_notifier_complete(struct v4l2_async_notifier *notifier)
-{
-	struct cio2_device *cio2 = to_cio2_device(notifier);
-	struct device *dev = &cio2->pci_dev->dev;
-	struct sensor_async_subdev *s_asd;
-	struct v4l2_async_subdev *asd;
-	struct cio2_queue *q;
-	unsigned int pad;
-	int ret;
-
-	list_for_each_entry(asd, &cio2->notifier.asd_list, asd_list) {
-		s_asd = to_sensor_asd(asd);
-		q = &cio2->queue[s_asd->csi2.port];
-
-		for (pad = 0; pad < q->sensor->entity.num_pads; pad++)
-			if (q->sensor->entity.pads[pad].flags &
-						MEDIA_PAD_FL_SOURCE)
-				break;
-
-		if (pad == q->sensor->entity.num_pads) {
-			dev_err(dev, "failed to find src pad for %s\n",
-				q->sensor->name);
-			return -ENXIO;
-		}
-
-		ret = media_create_pad_link(
-				&q->sensor->entity, pad,
-				&q->subdev.entity, CIO2_PAD_SINK,
-				0);
-		if (ret) {
-			dev_err(dev, "failed to create link for %s\n",
-				q->sensor->name);
-			return ret;
-		}
-	}
-
-	return v4l2_device_register_subdev_nodes(&cio2->v4l2_dev);
-}
-
 static const struct v4l2_async_notifier_operations cio2_async_ops = {
 	.bound = cio2_notifier_bound,
 	.unbind = cio2_notifier_unbind,
-	.complete = cio2_notifier_complete,
 };
 
 static int cio2_parse_firmware(struct cio2_device *cio2)
-- 
2.39.2

From a7a074d5023e76983ca0325eb8dedca224a858b8 Mon Sep 17 00:00:00 2001
From: Daniel Scally <djrscally@gmail.com>
Date: Thu, 2 Jun 2022 22:15:56 +0100
Subject: [PATCH] media: ipu3-cio2: Re-add .complete() to ipu3-cio2

Removing the .complete() callback had some unintended consequences.
Because the VCM driver is not directly linked to the ipu3-cio2
driver .bound() never gets called for it, which means its devnode
is never created if it probes late. Because .complete() waits for
any sub-notifiers to also be complete it is captured in that call.

Signed-off-by: Daniel Scally <djrscally@gmail.com>
Patchset: cameras
---
 drivers/media/pci/intel/ipu3/ipu3-cio2-main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
index 38f9f4da1922e..82681df7d794f 100644
--- a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
+++ b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c
@@ -1430,9 +1430,18 @@ static void cio2_notifier_unbind(struct v4l2_async_notifier *notifier,
 	cio2->queue[s_asd->csi2.port].sensor = NULL;
 }
 
+/* .complete() is called after all subdevices have been located */
+static int cio2_notifier_complete(struct v4l2_async_notifier *notifier)
+{
+	struct cio2_device *cio2 = to_cio2_device(notifier);
+
+	return v4l2_device_register_subdev_nodes(&cio2->v4l2_dev);
+}
+
 static const struct v4l2_async_notifier_operations cio2_async_ops = {
 	.bound = cio2_notifier_bound,
 	.unbind = cio2_notifier_unbind,
+	.complete = cio2_notifier_complete,
 };
 
 static int cio2_parse_firmware(struct cio2_device *cio2)
-- 
2.39.2

From c4a0681b762a438e7ab1df93fe1ba0bb5f14d799 Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Fri, 15 Jul 2022 23:48:00 +0200
Subject: [PATCH] drivers/media/i2c: Fix DW9719 dependencies

It should depend on VIDEO_DEV instead of VIDEO_V4L2.

Signed-off-by: Maximilian Luz <luzmaximilian@gmail.com>
Patchset: cameras
---
 drivers/media/i2c/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/media/i2c/Kconfig b/drivers/media/i2c/Kconfig
index 98d081efeef7f..c67966cafe109 100644
--- a/drivers/media/i2c/Kconfig
+++ b/drivers/media/i2c/Kconfig
@@ -823,7 +823,7 @@ config VIDEO_DW9714
 
 config VIDEO_DW9719
 	tristate "DW9719 lens voice coil support"
-	depends on I2C && VIDEO_V4L2
+	depends on I2C && VIDEO_DEV
 	select MEDIA_CONTROLLER
 	select VIDEO_V4L2_SUBDEV_API
 	select V4L2_ASYNC
-- 
2.39.2

From d3903b5c3021fb0874d2313cd23097931b08b801 Mon Sep 17 00:00:00 2001
From: Daniel Scally <dan.scally@ideasonboard.com>
Date: Fri, 24 Feb 2023 14:03:41 +0000
Subject: [PATCH] platform/x86: int3472: Add GPIOs to Surface Go 3 Board data

Add the INT347E GPIO lookup table to the board data for the Surface
Go 3. This is necessary to allow the ov7251 IR camera to probe
properly on that platform.

Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Patchset: cameras
---
 drivers/platform/x86/intel/int3472/tps68470_board_data.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/platform/x86/intel/int3472/tps68470_board_data.c b/drivers/platform/x86/intel/int3472/tps68470_board_data.c
index 309eab9c05588..322237e056f32 100644
--- a/drivers/platform/x86/intel/int3472/tps68470_board_data.c
+++ b/drivers/platform/x86/intel/int3472/tps68470_board_data.c
@@ -159,9 +159,10 @@ static const struct int3472_tps68470_board_data surface_go_tps68470_board_data =
 static const struct int3472_tps68470_board_data surface_go3_tps68470_board_data = {
 	.dev_name = "i2c-INT3472:01",
 	.tps68470_regulator_pdata = &surface_go_tps68470_pdata,
-	.n_gpiod_lookups = 1,
+	.n_gpiod_lookups = 2,
 	.tps68470_gpio_lookup_tables = {
-		&surface_go_int347a_gpios
+		&surface_go_int347a_gpios,
+		&surface_go_int347e_gpios,
 	},
 };
 
-- 
2.39.2

From 0fc904dd6658824c7624a27618cf280c7afeeeb4 Mon Sep 17 00:00:00 2001
From: Daniel Scally <dan.scally@ideasonboard.com>
Date: Thu, 2 Mar 2023 12:59:39 +0000
Subject: [PATCH] platform/x86: int3472: Remap reset GPIO for INT347E

ACPI _HID INT347E represents the OmniVision 7251 camera sensor. The
driver for this sensor expects a single pin named "enable", but on
some Microsoft Surface platforms the sensor is assigned a single
GPIO who's type flag is INT3472_GPIO_TYPE_RESET.

Remap the GPIO pin's function from "reset" to "enable". This is done
outside of the existing remap table since it is a more widespread
discrepancy than that method is designed for. Additionally swap the
polarity of the pin to match the driver's expectation.

Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Patchset: cameras
---
 drivers/platform/x86/intel/int3472/discrete.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/platform/x86/intel/int3472/discrete.c b/drivers/platform/x86/intel/int3472/discrete.c
index c42c3faa2c32d..6f4b8e24eb56c 100644
--- a/drivers/platform/x86/intel/int3472/discrete.c
+++ b/drivers/platform/x86/intel/int3472/discrete.c
@@ -108,6 +108,9 @@ static int skl_int3472_map_gpio_to_sensor(struct int3472_discrete_device *int347
 {
 	const struct int3472_sensor_config *sensor_config;
 	char *path = agpio->resource_source.string_ptr;
+	const struct acpi_device_id ov7251_ids[] = {
+		{ "INT347E" },
+	};
 	struct gpiod_lookup *table_entry;
 	struct acpi_device *adev;
 	acpi_handle handle;
@@ -130,6 +133,17 @@ static int skl_int3472_map_gpio_to_sensor(struct int3472_discrete_device *int347
 		}
 	}
 
+	/*
+	 * In addition to the function remap table we need to bulk remap the
+	 * "reset" GPIO for the OmniVision 7251 sensor, as the driver for that
+	 * expects its only GPIO pin to be called "enable" (and to have the
+	 * opposite polarity).
+	 */
+	if (!strcmp(func, "reset") && !acpi_match_device_ids(int3472->sensor, ov7251_ids)) {
+		func = "enable";
+		polarity = GPIO_ACTIVE_HIGH;
+	}
+
 	/* Functions mapped to NULL should not be mapped to the sensor */
 	if (!func)
 		return 0;
-- 
2.39.2

From 7739b393ccba091cac6d21776f65aa2490d45a91 Mon Sep 17 00:00:00 2001
From: Sachi King <nakato@nakato.io>
Date: Sat, 29 May 2021 17:47:38 +1000
Subject: [PATCH] ACPI: Add quirk for Surface Laptop 4 AMD missing irq 7
 override

This patch is the work of Thomas Gleixner <tglx@linutronix.de> and is
copied from:
https://lore.kernel.org/lkml/87lf8ddjqx.ffs@nanos.tec.linutronix.de/

This patch adds a quirk to the ACPI setup to patch in the the irq 7 pin
setup that is missing in the laptops ACPI table.

This patch was used for validation of the issue, and is not a proper
fix, but is probably a better temporary hack than continuing to probe
the Legacy PIC and run with the PIC in an unknown state.

Patchset: amd-gpio
---
 arch/x86/kernel/acpi/boot.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 518bda50068cb..7706380d825a3 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -22,6 +22,7 @@
 #include <linux/efi-bgrt.h>
 #include <linux/serial_core.h>
 #include <linux/pgtable.h>
+#include <linux/dmi.h>
 
 #include <asm/e820/api.h>
 #include <asm/irqdomain.h>
@@ -1247,6 +1248,17 @@ static void __init mp_config_acpi_legacy_irqs(void)
 	}
 }
 
+static const struct dmi_system_id surface_quirk[] __initconst = {
+	{
+		.ident = "Microsoft Surface Laptop 4 (AMD)",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
+			DMI_MATCH(DMI_PRODUCT_SKU, "Surface_Laptop_4_1952:1953")
+		},
+	},
+	{}
+};
+
 /*
  * Parse IOAPIC related entries in MADT
  * returns 0 on success, < 0 on error
@@ -1302,6 +1314,11 @@ static int __init acpi_parse_madt_ioapic_entries(void)
 		acpi_sci_ioapic_setup(acpi_gbl_FADT.sci_interrupt, 0, 0,
 				      acpi_gbl_FADT.sci_interrupt);
 
+	if (dmi_check_system(surface_quirk)) {
+		pr_warn("Surface hack: Override irq 7\n");
+		mp_override_legacy_irq(7, 3, 3, 7);
+	}
+
 	/* Fill in identity legacy mappings where no override */
 	mp_config_acpi_legacy_irqs();
 
-- 
2.39.2

From 09ed2994012b6aafb79e8e48080f8ad9f3a4b17e Mon Sep 17 00:00:00 2001
From: Maximilian Luz <luzmaximilian@gmail.com>
Date: Thu, 3 Jun 2021 14:04:26 +0200
Subject: [PATCH] ACPI: Add AMD 13" Surface Laptop 4 model to irq 7 override
 quirk

The 13" version of the Surface Laptop 4 has the same problem as the 15"
version, but uses a different SKU. Add that SKU to the quirk as well.

Patchset: amd-gpio
---
 arch/x86/kernel/acpi/boot.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 7706380d825a3..e26c38ddc601d 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -1250,12 +1250,19 @@ static void __init mp_config_acpi_legacy_irqs(void)
 
 static const struct dmi_system_id surface_quirk[] __initconst = {
 	{
-		.ident = "Microsoft Surface Laptop 4 (AMD)",
+		.ident = "Microsoft Surface Laptop 4 (AMD 15\")",
 		.matches = {
 			DMI_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
 			DMI_MATCH(DMI_PRODUCT_SKU, "Surface_Laptop_4_1952:1953")
 		},
 	},
+	{
+		.ident = "Microsoft Surface Laptop 4 (AMD 13\")",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Microsoft Corporation"),
+			DMI_MATCH(DMI_PRODUCT_SKU, "Surface_Laptop_4_1958:1959")
+		},
+	},
 	{}
 };
 
-- 
2.39.2

From 8ddc376b24ac53d198df35cbbc0034e3a7e9bcf1 Mon Sep 17 00:00:00 2001
From: "Bart Groeneveld | GPX Solutions B.V" <bart@gpxbv.nl>
Date: Mon, 5 Dec 2022 16:08:46 +0100
Subject: [PATCH] acpi: allow usage of acpi_tad on HW-reduced platforms

The specification [1] allows so-called HW-reduced platforms,
which do not implement everything, especially the wakeup related stuff.

In that case, it is still usable as a RTC. This is helpful for [2]
and [3], which is about a device with no other working RTC,
but it does have an HW-reduced TAD, which can be used as a RTC instead.

[1]: https://uefi.org/specs/ACPI/6.5/09_ACPI_Defined_Devices_and_Device_Specific_Objects.html#time-and-alarm-device
[2]: https://bugzilla.kernel.org/show_bug.cgi?id=212313
[3]: https://github.com/linux-surface/linux-surface/issues/415

Signed-off-by: Bart Groeneveld | GPX Solutions B.V. <bart@gpxbv.nl>
Patchset: rtc
---
 drivers/acpi/acpi_tad.c | 35 ++++++++++++++++++++++++-----------
 1 file changed, 24 insertions(+), 11 deletions(-)

diff --git a/drivers/acpi/acpi_tad.c b/drivers/acpi/acpi_tad.c
index e9b8e8305e23e..944276934e7ec 100644
--- a/drivers/acpi/acpi_tad.c
+++ b/drivers/acpi/acpi_tad.c
@@ -432,6 +432,14 @@ static ssize_t caps_show(struct device *dev, struct device_attribute *attr,
 
 static DEVICE_ATTR_RO(caps);
 
+static struct attribute *acpi_tad_attrs[] = {
+	&dev_attr_caps.attr,
+	NULL,
+};
+static const struct attribute_group acpi_tad_attr_group = {
+	.attrs	= acpi_tad_attrs,
+};
+
 static ssize_t ac_alarm_store(struct device *dev, struct device_attribute *attr,
 			      const char *buf, size_t count)
 {
@@ -480,15 +488,14 @@ static ssize_t ac_status_show(struct device *dev, struct device_attribute *attr,
 
 static DEVICE_ATTR_RW(ac_status);
 
-static struct attribute *acpi_tad_attrs[] = {
-	&dev_attr_caps.attr,
+static struct attribute *acpi_tad_ac_attrs[] = {
 	&dev_attr_ac_alarm.attr,
 	&dev_attr_ac_policy.attr,
 	&dev_attr_ac_status.attr,
 	NULL,
 };
-static const struct attribute_group acpi_tad_attr_group = {
-	.attrs	= acpi_tad_attrs,
+static const struct attribute_group acpi_tad_ac_attr_group = {
+	.attrs	= acpi_tad_ac_attrs,
 };
 
 static ssize_t dc_alarm_store(struct device *dev, struct device_attribute *attr,
@@ -563,13 +570,18 @@ static int acpi_tad_remove(struct platform_device *pdev)
 
 	pm_runtime_get_sync(dev);
 
+	if (dd->capabilities & ACPI_TAD_AC_WAKE)
+		sysfs_remove_group(&dev->kobj, &acpi_tad_ac_attr_group);
+
 	if (dd->capabilities & ACPI_TAD_DC_WAKE)
 		sysfs_remove_group(&dev->kobj, &acpi_tad_dc_attr_group);
 
 	sysfs_remove_group(&dev->kobj, &acpi_tad_attr_group);
 
-	acpi_tad_disable_timer(dev, ACPI_TAD_AC_TIMER);
-	acpi_tad_clear_status(dev, ACPI_TAD_AC_TIMER);
+	if (dd->capabilities & ACPI_TAD_AC_WAKE) {
+		acpi_tad_disable_timer(dev, ACPI_TAD_AC_TIMER);
+		acpi_tad_clear_status(dev, ACPI_TAD_AC_TIMER);
+	}
 	if (dd->capabilities & ACPI_TAD_DC_WAKE) {
 		acpi_tad_disable_timer(dev, ACPI_TAD_DC_TIMER);
 		acpi_tad_clear_status(dev, ACPI_TAD_DC_TIMER);
@@ -604,11 +616,6 @@ static int acpi_tad_probe(struct platform_device *pdev)
 		return -ENODEV;
 	}
 
-	if (!acpi_has_method(handle, "_PRW")) {
-		dev_info(dev, "Missing _PRW\n");
-		return -ENODEV;
-	}
-
 	dd = devm_kzalloc(dev, sizeof(*dd), GFP_KERNEL);
 	if (!dd)
 		return -ENOMEM;
@@ -637,6 +644,12 @@ static int acpi_tad_probe(struct platform_device *pdev)
 	if (ret)
 		goto fail;
 
+	if (caps & ACPI_TAD_AC_WAKE) {
+		ret = sysfs_create_group(&dev->kobj, &acpi_tad_ac_attr_group);
+		if (ret)
+			goto fail;
+	}
+
 	if (caps & ACPI_TAD_DC_WAKE) {
 		ret = sysfs_create_group(&dev->kobj, &acpi_tad_dc_attr_group);
 		if (ret)
-- 
2.39.2

From 8cb6f5bf4f355b0169e3f60c559a5e2203c1a988 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Tue, 18 Oct 2022 04:22:40 -0700
Subject: [PATCH] thermal: intel: hfi: Improve the type of
 hfi_features::nr_table_pages

A Coverity static code scan raised a potential overflow_before_widen
warning when hfi_features::nr_table_pages is used as an argument to
memcpy in intel_hfi_process_event().

Even though the overflow can never happen (the maximum number of pages of
the HFI table is 0x10 and 0x10 << PAGE_SHIFT = 0x10000), using size_t as
the data type of hfi_features::nr_table_pages makes Coverity happy and
matches the data type of the argument 'size' of memcpy().

Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Patchset: intel-thread-director
---
 drivers/thermal/intel/intel_hfi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index a0640f762dc5d..239afe02e5182 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -137,7 +137,7 @@ struct hfi_instance {
  * Parameters and supported features that are common to all HFI instances
  */
 struct hfi_features {
-	unsigned int	nr_table_pages;
+	size_t		nr_table_pages;
 	unsigned int	cpu_stride;
 	unsigned int	hdr_size;
 };
-- 
2.39.2

From 8f590ad883a1330a99f0bbd18d76c4631d7fddbb Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:29 -0800
Subject: [PATCH] sched/fair: Generalize asym_packing logic for SMT cores

When doing asym_packing load balancing between cores, all we care is that
the destination core is fully idle (including SMT siblings, if any) and
that the busiest candidate scheduling group has exactly one busy CPU. It is
irrelevant whether the candidate busiest core is non-SMT, SMT2, SMT4, SMT8,
etc.

Do not handle the candidate busiest non-SMT vs SMT cases separately. Simply
do the two checks described above. Let find_busiest_group() handle bigger
imbalances in the number of idle CPUs.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Reviewed-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 41 ++++++++++++++---------------------------
 1 file changed, 14 insertions(+), 27 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 2c3d0d49c80ea..8b5fc8e86addb 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9042,13 +9042,11 @@ group_type group_classify(unsigned int imbalance_pct,
  * the SMT siblings of @sg are busy. If only one CPU in @sg is busy, pull tasks
  * only if @dst_cpu has higher priority.
  *
- * If both @dst_cpu and @sg have SMT siblings, and @sg has exactly one more
- * busy CPU than @sds::local, let @dst_cpu pull tasks if it has higher priority.
- * Bigger imbalances in the number of busy CPUs will be dealt with in
- * update_sd_pick_busiest().
- *
- * If @sg does not have SMT siblings, only pull tasks if all of the SMT siblings
- * of @dst_cpu are idle and @sg has lower priority.
+ * If @dst_cpu has SMT siblings, check if there are no running tasks in
+ * @sds::local. In such case, decide based on the priority of @sg. Do it only
+ * if @sg has exactly one busy CPU (i.e., one more than @sds::local). Bigger
+ * imbalances in the number of busy CPUs will be dealt with in
+ * find_busiest_group().
  *
  * Return: true if @dst_cpu can pull tasks, false otherwise.
  */
@@ -9057,12 +9055,10 @@ static bool asym_smt_can_pull_tasks(int dst_cpu, struct sd_lb_stats *sds,
 				    struct sched_group *sg)
 {
 #ifdef CONFIG_SCHED_SMT
-	bool local_is_smt, sg_is_smt;
+	bool local_is_smt;
 	int sg_busy_cpus;
 
 	local_is_smt = sds->local->flags & SD_SHARE_CPUCAPACITY;
-	sg_is_smt = sg->flags & SD_SHARE_CPUCAPACITY;
-
 	sg_busy_cpus = sgs->group_weight - sgs->idle_cpus;
 
 	if (!local_is_smt) {
@@ -9083,25 +9079,16 @@ static bool asym_smt_can_pull_tasks(int dst_cpu, struct sd_lb_stats *sds,
 		return sched_asym_prefer(dst_cpu, sg->asym_prefer_cpu);
 	}
 
-	/* @dst_cpu has SMT siblings. */
-
-	if (sg_is_smt) {
-		int local_busy_cpus = sds->local->group_weight -
-				      sds->local_stat.idle_cpus;
-		int busy_cpus_delta = sg_busy_cpus - local_busy_cpus;
-
-		if (busy_cpus_delta == 1)
-			return sched_asym_prefer(dst_cpu, sg->asym_prefer_cpu);
-
-		return false;
-	}
-
 	/*
-	 * @sg does not have SMT siblings. Ensure that @sds::local does not end
-	 * up with more than one busy SMT sibling and only pull tasks if there
-	 * are not busy CPUs (i.e., no CPU has running tasks).
+	 * @dst_cpu has SMT siblings. Do asym_packing load balancing only if
+	 * all its siblings are idle (moving tasks between physical cores in
+	 * which some SMT siblings are busy results in the same throughput).
+	 *
+	 * If the difference in the number of busy CPUs is two or more, let
+	 * find_busiest_group() take care of it. We only care if @sg has
+	 * exactly one busy CPU. This covers SMT and non-SMT sched groups.
 	 */
-	if (!sds->local_stat.sum_nr_running)
+	if (sg_busy_cpus == 1 && !sds->local_stat.sum_nr_running)
 		return sched_asym_prefer(dst_cpu, sg->asym_prefer_cpu);
 
 	return false;
-- 
2.39.2

From 5a3b5eb5f79e51634f5fd173c0949c5293c93566 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:30 -0800
Subject: [PATCH] sched/fair: Move is_core_idle() out of CONFIG_NUMA

asym_packing needs this function to determine whether an SMT core is a
suitable destination for load balancing.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 34 +++++++++++++++++-----------------
 1 file changed, 17 insertions(+), 17 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 8b5fc8e86addb..98c64f1db20e0 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1049,6 +1049,23 @@ update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
  * Scheduling class queueing methods:
  */
 
+static inline bool is_core_idle(int cpu)
+{
+#ifdef CONFIG_SCHED_SMT
+	int sibling;
+
+	for_each_cpu(sibling, cpu_smt_mask(cpu)) {
+		if (cpu == sibling)
+			continue;
+
+		if (!idle_cpu(sibling))
+			return false;
+	}
+#endif
+
+	return true;
+}
+
 #ifdef CONFIG_NUMA
 #define NUMA_IMBALANCE_MIN 2
 
@@ -1688,23 +1705,6 @@ struct numa_stats {
 	int idle_cpu;
 };
 
-static inline bool is_core_idle(int cpu)
-{
-#ifdef CONFIG_SCHED_SMT
-	int sibling;
-
-	for_each_cpu(sibling, cpu_smt_mask(cpu)) {
-		if (cpu == sibling)
-			continue;
-
-		if (!idle_cpu(sibling))
-			return false;
-	}
-#endif
-
-	return true;
-}
-
 struct task_numa_env {
 	struct task_struct *p;
 
-- 
2.39.2

From d4ba60a8be784dc7ed866fb52ff94519eb9d1586 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:31 -0800
Subject: [PATCH] sched/fair: Only do asym_packing load balancing from fully
 idle SMT cores

When balancing load between cores, all the SMT siblings of the destination
CPU, if any, must be idle. Otherwise, pulling new tasks degrades the
throughput of the busy SMT siblings. The overall throughput of the system
remains the same.

When balancing load within an SMT core this consideration is not relevant
relevant. Follow the priorities that hardware indicates.

Using is_core_idle() renders checking !sds->local_stat.sum_nr_running
redundant. Remove it.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Suggested-by: Valentin Schneider <vschneid@redhat.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 34 +++++++++++++++++++++++++---------
 1 file changed, 25 insertions(+), 9 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 98c64f1db20e0..f74777fc78d7d 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9038,12 +9038,14 @@ group_type group_classify(unsigned int imbalance_pct,
  * Check the state of the SMT siblings of both @sds::local and @sg and decide
  * if @dst_cpu can pull tasks.
  *
+ * This function must be called only if all the SMT siblings of @dst_cpu are
+ * idle, if any.
+ *
  * If @dst_cpu does not have SMT siblings, it can pull tasks if two or more of
  * the SMT siblings of @sg are busy. If only one CPU in @sg is busy, pull tasks
  * only if @dst_cpu has higher priority.
  *
- * If @dst_cpu has SMT siblings, check if there are no running tasks in
- * @sds::local. In such case, decide based on the priority of @sg. Do it only
+ * If @dst_cpu has SMT siblings, decide based on the priority of @sg. Do it only
  * if @sg has exactly one busy CPU (i.e., one more than @sds::local). Bigger
  * imbalances in the number of busy CPUs will be dealt with in
  * find_busiest_group().
@@ -9080,15 +9082,13 @@ static bool asym_smt_can_pull_tasks(int dst_cpu, struct sd_lb_stats *sds,
 	}
 
 	/*
-	 * @dst_cpu has SMT siblings. Do asym_packing load balancing only if
-	 * all its siblings are idle (moving tasks between physical cores in
-	 * which some SMT siblings are busy results in the same throughput).
+	 * @dst_cpu has SMT siblings and are also idle.
 	 *
 	 * If the difference in the number of busy CPUs is two or more, let
 	 * find_busiest_group() take care of it. We only care if @sg has
 	 * exactly one busy CPU. This covers SMT and non-SMT sched groups.
 	 */
-	if (sg_busy_cpus == 1 && !sds->local_stat.sum_nr_running)
+	if (sg_busy_cpus == 1)
 		return sched_asym_prefer(dst_cpu, sg->asym_prefer_cpu);
 
 	return false;
@@ -9102,7 +9102,14 @@ static inline bool
 sched_asym(struct lb_env *env, struct sd_lb_stats *sds,  struct sg_lb_stats *sgs,
 	   struct sched_group *group)
 {
-	/* Only do SMT checks if either local or candidate have SMT siblings */
+	/*
+	 * If the destination CPU has SMT siblings, env->idle != CPU_NOT_IDLE
+	 * is not sufficient. We need to make sure the whole core is idle.
+	 */
+	if (sds->local->flags & SD_SHARE_CPUCAPACITY && !is_core_idle(env->dst_cpu))
+		return false;
+
+	/* Only do SMT checks if either local or candidate have SMT siblings. */
 	if ((sds->local->flags & SD_SHARE_CPUCAPACITY) ||
 	    (group->flags & SD_SHARE_CPUCAPACITY))
 		return asym_smt_can_pull_tasks(env->dst_cpu, sds, sgs, group);
@@ -11049,8 +11056,17 @@ static void nohz_balancer_kick(struct rq *rq)
 		 */
 		for_each_cpu_and(i, sched_domain_span(sd), nohz.idle_cpus_mask) {
 			if (sched_asym_prefer(i, cpu)) {
-				flags = NOHZ_STATS_KICK | NOHZ_BALANCE_KICK;
-				goto unlock;
+				/*
+				 * Always do ASYM_PACKING balance in the SMT
+				 * domain. In upper domains, the core must be
+				 * fully idle.
+				 */
+				if (sd->flags & SD_SHARE_CPUCAPACITY ||
+				    (!(sd->flags & SD_SHARE_CPUCAPACITY) &&
+				     is_core_idle(i))) {
+					flags = NOHZ_STATS_KICK | NOHZ_BALANCE_KICK;
+					goto unlock;
+				}
 			}
 		}
 	}
-- 
2.39.2

From 20641917bc6bad6f81bafe1bac213c1c3e70ed09 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:32 -0800
Subject: [PATCH] sched/fair: Let low-priority cores help high-priority busy
 SMT cores

Using asym_packing priorities within an SMT core is straightforward. Just
follow the priorities that hardware indicates.

When balancing load from an SMT core, also consider the idle of its
siblings. Priorities do not reflect that an SMT core divides its throughput
among all its busy siblings. They only makes sense when exactly one sibling
is busy.

Indicate that active balance is needed if the destination CPU has lower
priority than the source CPU but the latter has busy SMT siblings.

Make find_busiest_queue() not skip higher-priority SMT cores with more than
busy sibling.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Suggested-by: Valentin Schneider <vschneid@redhat.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 31 ++++++++++++++++++++++++++-----
 1 file changed, 26 insertions(+), 5 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index f74777fc78d7d..24183e3eb3d47 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -10224,11 +10224,20 @@ static struct rq *find_busiest_queue(struct lb_env *env,
 		    nr_running == 1)
 			continue;
 
-		/* Make sure we only pull tasks from a CPU of lower priority */
+		/*
+		 * Make sure we only pull tasks from a CPU of lower priority
+		 * when balancing between SMT siblings.
+		 *
+		 * If balancing between cores, let lower priority CPUs help
+		 * SMT cores with more than one busy sibling.
+		 */
 		if ((env->sd->flags & SD_ASYM_PACKING) &&
 		    sched_asym_prefer(i, env->dst_cpu) &&
-		    nr_running == 1)
-			continue;
+		    nr_running == 1) {
+			if (env->sd->flags & SD_SHARE_CPUCAPACITY ||
+			    (!(env->sd->flags & SD_SHARE_CPUCAPACITY) && is_core_idle(i)))
+				continue;
+		}
 
 		switch (env->migration_type) {
 		case migrate_load:
@@ -10318,8 +10327,20 @@ asym_active_balance(struct lb_env *env)
 	 * lower priority CPUs in order to pack all tasks in the
 	 * highest priority CPUs.
 	 */
-	return env->idle != CPU_NOT_IDLE && (env->sd->flags & SD_ASYM_PACKING) &&
-	       sched_asym_prefer(env->dst_cpu, env->src_cpu);
+	if (env->idle != CPU_NOT_IDLE && (env->sd->flags & SD_ASYM_PACKING)) {
+		/* Always obey priorities between SMT siblings. */
+		if (env->sd->flags & SD_SHARE_CPUCAPACITY)
+			return sched_asym_prefer(env->dst_cpu, env->src_cpu);
+
+		/*
+		 * A lower priority CPU can help an SMT core with more than one
+		 * busy sibling.
+		 */
+		return sched_asym_prefer(env->dst_cpu, env->src_cpu) ||
+		       !is_core_idle(env->src_cpu);
+	}
+
+	return false;
 }
 
 static inline bool
-- 
2.39.2

From c1e77e8eea55b2d16b80c6dde3bcb3cf7e232aa5 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:33 -0800
Subject: [PATCH] sched/fair: Keep a fully_busy SMT sched group as busiest

When comparing two fully_busy scheduling groups, keep the current busiest
group if it represents an SMT core. Tasks in such scheduling group share
CPU resources and need more help than tasks in a non-SMT fully_busy group.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 24183e3eb3d47..30b0e8476d1c6 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9302,10 +9302,22 @@ static bool update_sd_pick_busiest(struct lb_env *env,
 		 * contention when accessing shared HW resources.
 		 *
 		 * XXX for now avg_load is not computed and always 0 so we
-		 * select the 1st one.
+		 * select the 1st one, except if @sg is composed of SMT
+		 * siblings.
 		 */
-		if (sgs->avg_load <= busiest->avg_load)
+
+		if (sgs->avg_load < busiest->avg_load)
 			return false;
+
+		if (sgs->avg_load == busiest->avg_load) {
+			/*
+			 * SMT sched groups need more help than non-SMT groups.
+			 * If @sg happens to also be SMT, either choice is good.
+			 */
+			if (sds->busiest->flags & SD_SHARE_CPUCAPACITY)
+				return false;
+		}
+
 		break;
 
 	case group_has_spare:
-- 
2.39.2

From a6d2b260d711873add514001a4eca293ff40c860 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:34 -0800
Subject: [PATCH] sched/fair: Use the prefer_sibling flag of the current sched
 domain

SD_PREFER_SIBLING is set from the SMT scheduling domain up to the first
non-NUMA domain (the exception is systems with SD_ASYM_CPUCAPACITY).

Above the SMT sched domain, all domains have a child. The SD_PREFER_
SIBLING is honored always regardless of the scheduling domain at which the
load balance takes place.

There are cases, however, in which the busiest CPU's sched domain has
child but the destination CPU's does not. Consider, for instance a non-SMT
core (or an SMT core with only one online sibling) doing load balance with
an SMT core at the MC level. SD_PREFER_SIBLING will not be honored. We are
left with a fully busy SMT core and an idle non-SMT core.

Avoid inconsistent behavior. Use the prefer_sibling behavior at the current
scheduling domain, not its child.

The NUMA sched domain does not have the SD_PREFER_SIBLING flag. Thus, we
will not spread load among NUMA sched groups, as desired.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Suggested-by: Valentin Schneider <vschneid@redhat.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 30b0e8476d1c6..9e98cfcf1e48b 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9792,7 +9792,6 @@ static void update_idle_cpu_scan(struct lb_env *env,
 
 static inline void update_sd_lb_stats(struct lb_env *env, struct sd_lb_stats *sds)
 {
-	struct sched_domain *child = env->sd->child;
 	struct sched_group *sg = env->sd->groups;
 	struct sg_lb_stats *local = &sds->local_stat;
 	struct sg_lb_stats tmp_sgs;
@@ -9833,9 +9832,11 @@ static inline void update_sd_lb_stats(struct lb_env *env, struct sd_lb_stats *sd
 		sg = sg->next;
 	} while (sg != env->sd->groups);
 
-	/* Tag domain that child domain prefers tasks go to siblings first */
-	sds->prefer_sibling = child && child->flags & SD_PREFER_SIBLING;
-
+	/*
+	 * Tag domain that @env::sd prefers to spread excess tasks among
+	 * sibling sched groups.
+	 */
+	sds->prefer_sibling = env->sd->flags & SD_PREFER_SIBLING;
 
 	if (env->sd->flags & SD_NUMA)
 		env->fbq_type = fbq_classify_group(&sds->busiest_stat);
@@ -10134,7 +10135,6 @@ static struct sched_group *find_busiest_group(struct lb_env *env)
 			goto out_balanced;
 	}
 
-	/* Try to move all excess tasks to child's sibling domain */
 	if (sds.prefer_sibling && local->group_type == group_has_spare &&
 	    busiest->sum_nr_running > local->sum_nr_running + 1)
 		goto force_balance;
-- 
2.39.2

From 08ff98c145516a1013cc5cc4f1d9b0d36388ec90 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:35 -0800
Subject: [PATCH] sched/fair: Do not even the number of busy CPUs via
 asym_packing

Now that find_busiest_group() triggers load balancing between a fully_
busy SMT2 core and an idle non-SMT core, it is no longer needed to force
balancing via asym_packing. Use asym_packing only as intended: when there
is high-priority CPU that is idle.

After this change, the same logic apply to SMT and non-SMT local groups.
Simplify asym_smt_can_pull_tasks() accordingly.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 37 +++++--------------------------------
 1 file changed, 5 insertions(+), 32 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 9e98cfcf1e48b..635e8b41a87c9 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9035,20 +9035,15 @@ group_type group_classify(unsigned int imbalance_pct,
  * @sgs:	Load-balancing statistics of the candidate busiest group
  * @sg:		The candidate busiest group
  *
- * Check the state of the SMT siblings of both @sds::local and @sg and decide
- * if @dst_cpu can pull tasks.
+ * Check the state of the SMT siblings of @sg and decide if @dst_cpu can pull
+ * tasks.
  *
  * This function must be called only if all the SMT siblings of @dst_cpu are
  * idle, if any.
  *
- * If @dst_cpu does not have SMT siblings, it can pull tasks if two or more of
- * the SMT siblings of @sg are busy. If only one CPU in @sg is busy, pull tasks
- * only if @dst_cpu has higher priority.
- *
- * If @dst_cpu has SMT siblings, decide based on the priority of @sg. Do it only
- * if @sg has exactly one busy CPU (i.e., one more than @sds::local). Bigger
- * imbalances in the number of busy CPUs will be dealt with in
- * find_busiest_group().
+ * @dst_cpu can pull tasks if @sg has exactly one busy CPU (i.e., one more than
+ * @sds::local) and has lower group priority than @sds::local. Bigger imbalances
+ * in the number of busy CPUs will be dealt with in find_busiest_group().
  *
  * Return: true if @dst_cpu can pull tasks, false otherwise.
  */
@@ -9057,33 +9052,11 @@ static bool asym_smt_can_pull_tasks(int dst_cpu, struct sd_lb_stats *sds,
 				    struct sched_group *sg)
 {
 #ifdef CONFIG_SCHED_SMT
-	bool local_is_smt;
 	int sg_busy_cpus;
 
-	local_is_smt = sds->local->flags & SD_SHARE_CPUCAPACITY;
 	sg_busy_cpus = sgs->group_weight - sgs->idle_cpus;
 
-	if (!local_is_smt) {
-		/*
-		 * If we are here, @dst_cpu is idle and does not have SMT
-		 * siblings. Pull tasks if candidate group has two or more
-		 * busy CPUs.
-		 */
-		if (sg_busy_cpus >= 2) /* implies sg_is_smt */
-			return true;
-
-		/*
-		 * @dst_cpu does not have SMT siblings. @sg may have SMT
-		 * siblings and only one is busy. In such case, @dst_cpu
-		 * can help if it has higher priority and is idle (i.e.,
-		 * it has no running tasks).
-		 */
-		return sched_asym_prefer(dst_cpu, sg->asym_prefer_cpu);
-	}
-
 	/*
-	 * @dst_cpu has SMT siblings and are also idle.
-	 *
 	 * If the difference in the number of busy CPUs is two or more, let
 	 * find_busiest_group() take care of it. We only care if @sg has
 	 * exactly one busy CPU. This covers SMT and non-SMT sched groups.
-- 
2.39.2

From f70e8d703a3703327fea540de41d6b4df922a85d Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:36 -0800
Subject: [PATCH] sched/topology: Remove SHARED_CHILD from ASYM_PACKING

Only x86 and Power7 use ASYM_PACKING. They use it differently.

Power7 has cores of equal priority, but the SMT siblings of a core have
different priorities. Parent scheduling domains do not need (nor have) the
ASYM_PACKING flag. SHARED_CHILD is not needed. Using SHARED_PARENT would
cause the topology debug code to complain.

X86 has cores of different priority, but all the SMT siblings of the core
have equal priority. It needs ASYM_PACKING at the MC level, but not at the
SMT level (it also needs it at upper levels if they have scheduling groups
of different priority). Removing ASYM_PACKING from the SMT domain causes
the topology debug code to complain.

Remove SHARED_CHILD for now. We still need a topology check that satisfies
both architectures.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Suggested-by: Valentin Schneider <vschneid@redhat.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 include/linux/sched/sd_flags.h | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/include/linux/sched/sd_flags.h b/include/linux/sched/sd_flags.h
index 57bde66d95f7a..800238854ba54 100644
--- a/include/linux/sched/sd_flags.h
+++ b/include/linux/sched/sd_flags.h
@@ -132,12 +132,9 @@ SD_FLAG(SD_SERIALIZE, SDF_SHARED_PARENT | SDF_NEEDS_GROUPS)
 /*
  * Place busy tasks earlier in the domain
  *
- * SHARED_CHILD: Usually set on the SMT level. Technically could be set further
- *               up, but currently assumed to be set from the base domain
- *               upwards (see update_top_cache_domain()).
  * NEEDS_GROUPS: Load balancing flag.
  */
-SD_FLAG(SD_ASYM_PACKING, SDF_SHARED_CHILD | SDF_NEEDS_GROUPS)
+SD_FLAG(SD_ASYM_PACKING,  SDF_NEEDS_GROUPS)
 
 /*
  * Prefer to place tasks in a sibling domain
-- 
2.39.2

From fa1585eecfadd43f866f50951bbe106c09e1f79f Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:37 -0800
Subject: [PATCH] x86/sched: Remove SD_ASYM_PACKING from the SMT domain flags

There is no difference between any of the SMT siblings of a physical core.
Do not do asym_packing load balancing at this level.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 arch/x86/kernel/smpboot.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 3f3ea0287f694..c3de98224cb4f 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -545,7 +545,7 @@ static int x86_core_flags(void)
 #ifdef CONFIG_SCHED_SMT
 static int x86_smt_flags(void)
 {
-	return cpu_smt_flags() | x86_sched_itmt_flags();
+	return cpu_smt_flags();
 }
 #endif
 #ifdef CONFIG_SCHED_CLUSTER
-- 
2.39.2

From 5f0150f15f343e543bc257bca1db30f3481ed474 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 20:58:38 -0800
Subject: [PATCH] x86/sched/itmt: Give all SMT siblings of a core the same
 priority

X86 does not have the SD_ASYM_PACKING flag in the SMT domain. The scheduler
knows how to handle SMT and non-SMT cores of different priority. There is
no reason for SMT siblings of a core to have different priorities.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Reviewed-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Patchset: intel-thread-director
---
 arch/x86/kernel/itmt.c | 23 +++++------------------
 1 file changed, 5 insertions(+), 18 deletions(-)

diff --git a/arch/x86/kernel/itmt.c b/arch/x86/kernel/itmt.c
index 9ff480e94511b..6510883c5e817 100644
--- a/arch/x86/kernel/itmt.c
+++ b/arch/x86/kernel/itmt.c
@@ -174,32 +174,19 @@ int arch_asym_cpu_priority(int cpu)
 
 /**
  * sched_set_itmt_core_prio() - Set CPU priority based on ITMT
- * @prio:	Priority of cpu core
- * @core_cpu:	The cpu number associated with the core
+ * @prio:	Priority of @cpu
+ * @cpu:	The CPU number
  *
  * The pstate driver will find out the max boost frequency
  * and call this function to set a priority proportional
- * to the max boost frequency. CPU with higher boost
+ * to the max boost frequency. CPUs with higher boost
  * frequency will receive higher priority.
  *
  * No need to rebuild sched domain after updating
  * the CPU priorities. The sched domains have no
  * dependency on CPU priorities.
  */
-void sched_set_itmt_core_prio(int prio, int core_cpu)
+void sched_set_itmt_core_prio(int prio, int cpu)
 {
-	int cpu, i = 1;
-
-	for_each_cpu(cpu, topology_sibling_cpumask(core_cpu)) {
-		int smt_prio;
-
-		/*
-		 * Ensure that the siblings are moved to the end
-		 * of the priority chain and only used when
-		 * all other high priority cpus are out of capacity.
-		 */
-		smt_prio = prio * smp_num_siblings / (i * i);
-		per_cpu(sched_core_priority, cpu) = smt_prio;
-		i++;
-	}
+	per_cpu(sched_core_priority, cpu) = prio;
 }
-- 
2.39.2

From 59c32bf713f2d1c52fedb38ceeef07375736dbbc Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:42 -0800
Subject: [PATCH] sched/task_struct: Introduce IPC classes of tasks

On hybrid processors, the architecture differences between the types of
CPUs lead to different instructions-per-cycle (IPC) on each type of CPU.
IPCs may differ further by the type of instructions. Instructions can be
grouped into classes of similar IPCs.

Hence, tasks can be classified into groups based on the type of
instructions they execute.

Add a new member task_struct::ipcc to associate a particular task to
an IPC class that depends on the instructions it executes.

The scheduler may use the IPC class of a task and data about the
performance among CPUs of a given IPC class to improve throughput. It
may, for instance, place certain classes of tasks on CPUs of higher
performance.

The methods to determine the classification of a task and its relative
IPC score are specific to each CPU architecture.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 include/linux/sched.h | 10 ++++++++++
 init/Kconfig          | 12 ++++++++++++
 2 files changed, 22 insertions(+)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index ffb6eb55cd135..ca0c32bf796fb 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -127,6 +127,8 @@ struct task_group;
 					 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
 					 TASK_PARKED)
 
+#define IPC_CLASS_UNCLASSIFIED		0
+
 #define task_is_running(task)		(READ_ONCE((task)->__state) == TASK_RUNNING)
 
 #define task_is_traced(task)		((READ_ONCE(task->jobctl) & JOBCTL_TRACED) != 0)
@@ -1528,6 +1530,14 @@ struct task_struct {
 	union rv_task_monitor		rv[RV_PER_TASK_MONITORS];
 #endif
 
+#ifdef CONFIG_IPC_CLASSES
+	/*
+	 * A hardware-defined classification of task that reflects but is
+	 * not identical to the number of instructions per cycle.
+	 */
+	unsigned short			ipcc;
+#endif
+
 	/*
 	 * New fields for task_struct should be added above here, so that
 	 * they are included in the randomized portion of task_struct.
diff --git a/init/Kconfig b/init/Kconfig
index 0c214af99085d..0ddda55fde6a6 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -865,6 +865,18 @@ config UCLAMP_BUCKETS_COUNT
 
 	  If in doubt, use the default value.
 
+config IPC_CLASSES
+	bool "IPC classes of tasks"
+	depends on SMP
+	help
+	  If selected, each task is assigned a classification value that
+	  reflects the type of instructions that the task executes. This
+	  classification reflects but is not equal to the number of
+	  instructions retired per cycle.
+
+	  The scheduler uses the classification value to improve the placement
+	  of tasks.
+
 endmenu
 
 #
-- 
2.39.2

From c4738d1d773b0e4066f30875c6393adb77d55837 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:43 -0800
Subject: [PATCH] sched: Add interfaces for IPC classes

Add the interfaces that architectures shall implement to convey the data
to support IPC classes.

arch_update_ipcc() updates the IPC classification of the current task as
given by hardware.

arch_get_ipcc_score() provides a performance score for a given IPC class
when placed on a specific CPU. Higher scores indicate higher performance.

When a driver or equivalent enablement code has configured the necessary
hardware to support IPC classes, it should call sched_enable_ipc_classes()
to notify the scheduler that it can start using IPC classes data.

The number of classes and the score of each class of task are determined
by hardware.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 include/linux/sched/topology.h |  6 ++++
 kernel/sched/sched.h           | 66 ++++++++++++++++++++++++++++++++++
 kernel/sched/topology.c        |  9 +++++
 3 files changed, 81 insertions(+)

diff --git a/include/linux/sched/topology.h b/include/linux/sched/topology.h
index 816df6cc444e1..5b084d3c9ad12 100644
--- a/include/linux/sched/topology.h
+++ b/include/linux/sched/topology.h
@@ -280,4 +280,10 @@ static inline int task_node(const struct task_struct *p)
 	return cpu_to_node(task_cpu(p));
 }
 
+#ifdef CONFIG_IPC_CLASSES
+extern void sched_enable_ipc_classes(void);
+#else
+static inline void sched_enable_ipc_classes(void) { }
+#endif
+
 #endif /* _LINUX_SCHED_TOPOLOGY_H */
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index d6d488e8eb554..a3b2b66e077d9 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2511,6 +2511,72 @@ void arch_scale_freq_tick(void)
 }
 #endif
 
+#ifdef CONFIG_IPC_CLASSES
+DECLARE_STATIC_KEY_FALSE(sched_ipcc);
+
+static inline bool sched_ipcc_enabled(void)
+{
+	return static_branch_unlikely(&sched_ipcc);
+}
+
+#ifndef arch_update_ipcc
+/**
+ * arch_update_ipcc() - Update the IPC class of the current task
+ * @curr:		The current task
+ *
+ * Request that the IPC classification of @curr is updated.
+ *
+ * Returns: none
+ */
+static __always_inline
+void arch_update_ipcc(struct task_struct *curr)
+{
+}
+#endif
+
+#ifndef arch_get_ipcc_score
+
+#define SCHED_IPCC_SCORE_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
+/**
+ * arch_get_ipcc_score() - Get the IPC score of a class of task
+ * @ipcc:	The IPC class
+ * @cpu:	A CPU number
+ *
+ * The IPC performance scores reflects (but it is not identical to) the number
+ * of instructions retired per cycle for a given IPC class. It is a linear and
+ * abstract metric. Higher scores reflect better performance.
+ *
+ * The IPC score can be normalized with respect to the class, i, with the
+ * highest IPC score on the CPU, c, with highest performance:
+ *
+ *            IPC(i, c)
+ *  ------------------------------------ * SCHED_IPCC_SCORE_SCALE
+ *     max(IPC(i, c) : (i, c))
+ *
+ * Scheduling schemes that want to use the IPC score along with other
+ * normalized metrics for scheduling (e.g., CPU capacity) may need to normalize
+ * it.
+ *
+ * Other scheduling schemes (e.g., asym_packing) do not need normalization.
+ *
+ * Returns the performance score of an IPC class, @ipcc, when running on @cpu.
+ * Error when either @ipcc or @cpu are invalid.
+ */
+static __always_inline
+unsigned long arch_get_ipcc_score(unsigned short ipcc, int cpu)
+{
+	return SCHED_IPCC_SCORE_SCALE;
+}
+#endif
+#else /* CONFIG_IPC_CLASSES */
+
+#define arch_get_ipcc_score(ipcc, cpu) (-EINVAL)
+#define arch_update_ipcc(curr)
+
+static inline bool sched_ipcc_enabled(void) { return false; }
+
+#endif /* CONFIG_IPC_CLASSES */
+
 #ifndef arch_scale_freq_capacity
 /**
  * arch_scale_freq_capacity - get the frequency scale factor of a given CPU.
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 8739c2a5a54ea..60e03d15f58ca 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -670,6 +670,15 @@ DEFINE_PER_CPU(struct sched_domain __rcu *, sd_asym_packing);
 DEFINE_PER_CPU(struct sched_domain __rcu *, sd_asym_cpucapacity);
 DEFINE_STATIC_KEY_FALSE(sched_asym_cpucapacity);
 
+#ifdef CONFIG_IPC_CLASSES
+DEFINE_STATIC_KEY_FALSE(sched_ipcc);
+
+void sched_enable_ipc_classes(void)
+{
+	static_branch_enable_cpuslocked(&sched_ipcc);
+}
+#endif
+
 static void update_top_cache_domain(int cpu)
 {
 	struct sched_domain_shared *sds = NULL;
-- 
2.39.2

From 021e8196d4ef87aa80fae8024b8055a6bca5b9c4 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:44 -0800
Subject: [PATCH] sched/core: Initialize the IPC class of a new task

New tasks shall start life as unclassified. They will be classified by
hardware when they run.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index f730b6fe94a7f..7b63bf90430bb 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4377,6 +4377,9 @@ static void __sched_fork(unsigned long clone_flags, struct task_struct *p)
 	p->se.prev_sum_exec_runtime	= 0;
 	p->se.nr_migrations		= 0;
 	p->se.vruntime			= 0;
+#ifdef CONFIG_IPC_CLASSES
+	p->ipcc				= IPC_CLASS_UNCLASSIFIED;
+#endif
 	INIT_LIST_HEAD(&p->se.group_node);
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
-- 
2.39.2

From ef293bd0404885577a038daf0c011a57eeece3de Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:45 -0800
Subject: [PATCH] sched/core: Add user_tick as argument to scheduler_tick()

Differentiate between user and kernel ticks so that the scheduler updates
the IPC class of the current task during the former.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 include/linux/sched.h | 2 +-
 kernel/sched/core.c   | 2 +-
 kernel/time/timer.c   | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index ca0c32bf796fb..e58dc7503864c 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -293,7 +293,7 @@ enum {
 	TASK_COMM_LEN = 16,
 };
 
-extern void scheduler_tick(void);
+extern void scheduler_tick(bool user_tick);
 
 #define	MAX_SCHEDULE_TIMEOUT		LONG_MAX
 
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 7b63bf90430bb..0a8558421eba2 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5492,7 +5492,7 @@ static inline u64 cpu_resched_latency(struct rq *rq) { return 0; }
  * This function gets called by the timer code, with HZ frequency.
  * We call it with interrupts disabled.
  */
-void scheduler_tick(void)
+void scheduler_tick(bool user_tick)
 {
 	int cpu = smp_processor_id();
 	struct rq *rq = cpu_rq(cpu);
diff --git a/kernel/time/timer.c b/kernel/time/timer.c
index 717fcb9fb14aa..b444b6f5f585b 100644
--- a/kernel/time/timer.c
+++ b/kernel/time/timer.c
@@ -1841,7 +1841,7 @@ void update_process_times(int user_tick)
 	if (in_irq())
 		irq_work_tick();
 #endif
-	scheduler_tick();
+	scheduler_tick(user_tick);
 	if (IS_ENABLED(CONFIG_POSIX_TIMERS))
 		run_posix_cpu_timers();
 }
-- 
2.39.2

From 45a0511c0378f765cc26fb1faaddc3080ab2e179 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:46 -0800
Subject: [PATCH] sched/core: Update the IPC class of the current task

When supported, hardware monitors the instruction stream to classify the
current task. Hence, at userspace tick, we are ready to read the most
recent classification result for the current task.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 0a8558421eba2..4782b1359eb89 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5504,6 +5504,9 @@ void scheduler_tick(bool user_tick)
 	if (housekeeping_cpu(cpu, HK_TYPE_TICK))
 		arch_scale_freq_tick();
 
+	if (sched_ipcc_enabled() && user_tick)
+		arch_update_ipcc(curr);
+
 	sched_clock_tick();
 
 	rq_lock(rq, &rf);
-- 
2.39.2

From b88b78dd46cf545386f984768597179451f2bdf0 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:47 -0800
Subject: [PATCH] sched/fair: Collect load-balancing stats for IPC classes

When selecting a busiest scheduling group, the IPC class of the current
task can be used to select between two scheduling groups of types asym_
packing or fully_busy that are otherwise identical.

Compute the IPC class performance score for a scheduling group. It
is the sum of the scores of the current tasks of all the runqueues.

Also, keep track of the class of the task with the lowest IPC class score
in the scheduling group.

These two metrics will be used during idle load balancing to compute the
current and the prospective IPC class score of a scheduling group.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 61 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 61 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 635e8b41a87c9..86f779d9d2a90 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8751,6 +8751,11 @@ struct sg_lb_stats {
 	unsigned int nr_numa_running;
 	unsigned int nr_preferred_running;
 #endif
+#ifdef CONFIG_IPC_CLASSES
+	unsigned long min_score; /* Min(score(rq->curr->ipcc)) */
+	unsigned short min_ipcc; /* Class of the task with the minimum IPCC score in the rq */
+	unsigned long sum_score; /* Sum(score(rq->curr->ipcc)) */
+#endif
 };
 
 /*
@@ -9028,6 +9033,59 @@ group_type group_classify(unsigned int imbalance_pct,
 	return group_has_spare;
 }
 
+#ifdef CONFIG_IPC_CLASSES
+static void init_rq_ipcc_stats(struct sg_lb_stats *sgs)
+{
+	/* All IPCC stats have been set to zero in update_sg_lb_stats(). */
+	sgs->min_score = ULONG_MAX;
+}
+
+/* Called only if cpu_of(@rq) is not idle and has tasks running. */
+static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
+				    struct rq *rq)
+{
+	struct task_struct *curr;
+	unsigned short ipcc;
+	unsigned long score;
+
+	if (!sched_ipcc_enabled())
+		return;
+
+	curr = rcu_dereference(rq->curr);
+	if (!curr || (curr->flags & PF_EXITING) || is_idle_task(curr) ||
+	    task_is_realtime(curr) ||
+	    !cpumask_test_cpu(dst_cpu, curr->cpus_ptr))
+		return;
+
+	ipcc = curr->ipcc;
+	score = arch_get_ipcc_score(ipcc, cpu_of(rq));
+
+	/*
+	 * Ignore tasks with invalid scores. When finding the busiest group, we
+	 * prefer those with higher sum_score. This group will not be selected.
+	 */
+	if (IS_ERR_VALUE(score))
+		return;
+
+	sgs->sum_score += score;
+
+	if (score < sgs->min_score) {
+		sgs->min_score = score;
+		sgs->min_ipcc = ipcc;
+	}
+}
+
+#else /* CONFIG_IPC_CLASSES */
+static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
+				    struct rq *rq)
+{
+}
+
+static void init_rq_ipcc_stats(struct sg_lb_stats *sgs)
+{
+}
+#endif /* CONFIG_IPC_CLASSES */
+
 /**
  * asym_smt_can_pull_tasks - Check whether the load balancing CPU can pull tasks
  * @dst_cpu:	Destination CPU of the load balancing
@@ -9120,6 +9178,7 @@ static inline void update_sg_lb_stats(struct lb_env *env,
 	int i, nr_running, local_group;
 
 	memset(sgs, 0, sizeof(*sgs));
+	init_rq_ipcc_stats(sgs);
 
 	local_group = group == sds->local;
 
@@ -9169,6 +9228,8 @@ static inline void update_sg_lb_stats(struct lb_env *env,
 			if (sgs->group_misfit_task_load < load)
 				sgs->group_misfit_task_load = load;
 		}
+
+		update_sg_lb_ipcc_stats(env->dst_cpu, sgs, rq);
 	}
 
 	sgs->group_capacity = group->sgc->capacity;
-- 
2.39.2

From ba3c46567c032ee843d8ab5f576ddc78df1a56bf Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:48 -0800
Subject: [PATCH] sched/fair: Compute IPC class scores for load balancing

Compute the joint total (both current and prospective) IPC class score of
a scheduling group and the local scheduling group.

These IPCC statistics are used during idle load balancing. The candidate
scheduling group will have one fewer busy CPU after load balancing. This
observation is important for cores with SMT support.

The IPCC score of scheduling groups composed of SMT siblings needs to
consider that the siblings share CPU resources. When computing the total
IPCC score of the scheduling group, divide score of each sibling by the
number of busy siblings.

Collect IPCC statistics for asym_packing and fully_busy scheduling groups.
When picking a busiest group, they are used to break ties between otherwise
identical groups.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 68 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 68 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 86f779d9d2a90..3b84fb72891bc 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8755,6 +8755,8 @@ struct sg_lb_stats {
 	unsigned long min_score; /* Min(score(rq->curr->ipcc)) */
 	unsigned short min_ipcc; /* Class of the task with the minimum IPCC score in the rq */
 	unsigned long sum_score; /* Sum(score(rq->curr->ipcc)) */
+	long ipcc_score_after; /* Prospective IPCC score after load balancing */
+	unsigned long ipcc_score_before; /* IPCC score before load balancing */
 #endif
 };
 
@@ -9075,6 +9077,62 @@ static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
 	}
 }
 
+static void update_sg_lb_stats_scores(struct sg_lb_stats *sgs,
+				      struct sched_group *sg,
+				      struct lb_env *env)
+{
+	unsigned long score_on_dst_cpu, before;
+	int busy_cpus;
+	long after;
+
+	if (!sched_ipcc_enabled())
+		return;
+
+	/*
+	 * IPCC scores are only useful during idle load balancing. For now,
+	 * only asym_packing uses IPCC scores.
+	 */
+	if (!(env->sd->flags & SD_ASYM_PACKING) ||
+	    env->idle == CPU_NOT_IDLE)
+		return;
+
+	/*
+	 * IPCC scores are used to break ties only between these types of
+	 * groups.
+	 */
+	if (sgs->group_type != group_fully_busy &&
+	    sgs->group_type != group_asym_packing)
+		return;
+
+	busy_cpus = sgs->group_weight - sgs->idle_cpus;
+
+	/* No busy CPUs in the group. No tasks to move. */
+	if (!busy_cpus)
+		return;
+
+	score_on_dst_cpu = arch_get_ipcc_score(sgs->min_ipcc, env->dst_cpu);
+
+	/*
+	 * Do not use IPC scores. sgs::ipcc_score_{after, before} will be zero
+	 * and not used.
+	 */
+	if (IS_ERR_VALUE(score_on_dst_cpu))
+		return;
+
+	before = sgs->sum_score;
+	after = before - sgs->min_score;
+
+	/* SMT siblings share throughput. */
+	if (busy_cpus > 1 && sg->flags & SD_SHARE_CPUCAPACITY) {
+		before /= busy_cpus;
+		/* One sibling will become idle after load balance. */
+		after /= busy_cpus - 1;
+	}
+
+	sgs->ipcc_score_after = after + score_on_dst_cpu;
+	sgs->ipcc_score_before = before;
+}
+
 #else /* CONFIG_IPC_CLASSES */
 static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
 				    struct rq *rq)
@@ -9084,6 +9142,13 @@ static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
 static void init_rq_ipcc_stats(struct sg_lb_stats *sgs)
 {
 }
+
+static void update_sg_lb_stats_scores(struct sg_lb_stats *sgs,
+				      struct sched_group *sg,
+				      struct lb_env *env)
+{
+}
+
 #endif /* CONFIG_IPC_CLASSES */
 
 /**
@@ -9245,6 +9310,9 @@ static inline void update_sg_lb_stats(struct lb_env *env,
 
 	sgs->group_type = group_classify(env->sd->imbalance_pct, group, sgs);
 
+	if (!local_group)
+		update_sg_lb_stats_scores(sgs, group, env);
+
 	/* Computing avg_load makes sense only when group is overloaded */
 	if (sgs->group_type == group_overloaded)
 		sgs->avg_load = (sgs->group_load * SCHED_CAPACITY_SCALE) /
-- 
2.39.2

From 98664eb2ae97aaf9b290077ef9e6629a7c1d7d79 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:49 -0800
Subject: [PATCH] sched/fair: Use IPCC stats to break ties between asym_packing
 sched groups

As it iterates, update_sd_pick_busiest() keeps on selecting as busiest
sched groups of identical priority. Since both groups have the same
priority, either group is a good choice. The IPCC statistics provide a
measure of the throughput before and after load balance. Use them to
pick a busiest scheduling group from otherwise identical asym_packing
scheduling groups.

Pick as busiest the scheduling group that yields a higher IPCC score
after load balancing.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 72 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 3b84fb72891bc..89a13ae0185e6 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9133,6 +9133,60 @@ static void update_sg_lb_stats_scores(struct sg_lb_stats *sgs,
 	sgs->ipcc_score_before = before;
 }
 
+/**
+ * sched_asym_ipcc_prefer - Select a sched group based on its IPCC score
+ * @a:	Load balancing statistics of a sched group
+ * @b:	Load balancing statistics of a second sched group
+ *
+ * Returns: true if @a has a higher IPCC score than @b after load balance.
+ * False otherwise.
+ */
+static bool sched_asym_ipcc_prefer(struct sg_lb_stats *a,
+				   struct sg_lb_stats *b)
+{
+	if (!sched_ipcc_enabled())
+		return false;
+
+	/* @a increases overall throughput after load balance. */
+	if (a->ipcc_score_after > b->ipcc_score_after)
+		return true;
+
+	/*
+	 * If @a and @b yield the same overall throughput, pick @a if
+	 * its current throughput is lower than that of @b.
+	 */
+	if (a->ipcc_score_after == b->ipcc_score_after)
+		return a->ipcc_score_before < b->ipcc_score_before;
+
+	return false;
+}
+
+/**
+ * sched_asym_ipcc_pick - Select a sched group based on its IPCC score
+ * @a:		A scheduling group
+ * @b:		A second scheduling group
+ * @a_stats:	Load balancing statistics of @a
+ * @b_stats:	Load balancing statistics of @b
+ *
+ * Returns: true if @a has the same priority and @a has tasks with IPC classes
+ * that yield higher overall throughput after load balance. False otherwise.
+ */
+static bool sched_asym_ipcc_pick(struct sched_group *a,
+				 struct sched_group *b,
+				 struct sg_lb_stats *a_stats,
+				 struct sg_lb_stats *b_stats)
+{
+	/*
+	 * Only use the class-specific preference selection if both sched
+	 * groups have the same priority.
+	 */
+	if (arch_asym_cpu_priority(a->asym_prefer_cpu) !=
+	    arch_asym_cpu_priority(b->asym_prefer_cpu))
+		return false;
+
+	return sched_asym_ipcc_prefer(a_stats, b_stats);
+}
+
 #else /* CONFIG_IPC_CLASSES */
 static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
 				    struct rq *rq)
@@ -9149,6 +9203,14 @@ static void update_sg_lb_stats_scores(struct sg_lb_stats *sgs,
 {
 }
 
+static bool sched_asym_ipcc_pick(struct sched_group *a,
+				 struct sched_group *b,
+				 struct sg_lb_stats *a_stats,
+				 struct sg_lb_stats *b_stats)
+{
+	return false;
+}
+
 #endif /* CONFIG_IPC_CLASSES */
 
 /**
@@ -9384,6 +9446,16 @@ static bool update_sd_pick_busiest(struct lb_env *env,
 		/* Prefer to move from lowest priority CPU's work */
 		if (sched_asym_prefer(sg->asym_prefer_cpu, sds->busiest->asym_prefer_cpu))
 			return false;
+
+		/*
+		 * Unlike other callers of sched_asym_prefer(), here both @sg
+		 * and @sds::busiest have tasks running. When they have equal
+		 * priority, their IPC class scores can be used to select a
+		 * better busiest.
+		 */
+		if (sched_asym_ipcc_pick(sds->busiest, sg, &sds->busiest_stat, sgs))
+			return false;
+
 		break;
 
 	case group_misfit_task:
-- 
2.39.2

From a4b0646c23237cc67703ed40bdde243d78bb7910 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:50 -0800
Subject: [PATCH] sched/fair: Use IPCC stats to break ties between fully_busy
 SMT groups

IPCC statistics are used during idle load balancing. After balancing one
of the siblings of an SMT core will become idle. The rest of the busy
siblings will enjoy increased throughput. The IPCC statistics provide
a measure of the increased throughput. Use them to pick a busiest group
from otherwise identical fully_busy scheduling groups (of which the
avg_load is equal - and zero).

Using IPCC scores to break ties with non-SMT fully_busy sched groups
is not necessary. SMT sched groups always need more help.

Add a stub sched_asym_ipcc_prefer() for !CONFIG_IPC_CLASSES.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 23 ++++++++++++++++++++---
 1 file changed, 20 insertions(+), 3 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 89a13ae0185e6..4f75e9964e8ca 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9203,6 +9203,12 @@ static void update_sg_lb_stats_scores(struct sg_lb_stats *sgs,
 {
 }
 
+static bool sched_asym_ipcc_prefer(struct sg_lb_stats *a,
+				   struct sg_lb_stats *b)
+{
+	return false;
+}
+
 static bool sched_asym_ipcc_pick(struct sched_group *a,
 				 struct sched_group *b,
 				 struct sg_lb_stats *a_stats,
@@ -9486,10 +9492,21 @@ static bool update_sd_pick_busiest(struct lb_env *env,
 		if (sgs->avg_load == busiest->avg_load) {
 			/*
 			 * SMT sched groups need more help than non-SMT groups.
-			 * If @sg happens to also be SMT, either choice is good.
 			 */
-			if (sds->busiest->flags & SD_SHARE_CPUCAPACITY)
-				return false;
+			if (sds->busiest->flags & SD_SHARE_CPUCAPACITY) {
+				if (!(sg->flags & SD_SHARE_CPUCAPACITY))
+					return false;
+
+				/*
+				 * Between two SMT groups, use IPCC scores to pick the
+				 * one that would improve throughput the most (only
+				 * asym_packing uses IPCC scores for now).
+				 */
+				if (sched_ipcc_enabled() &&
+				    env->sd->flags & SD_ASYM_PACKING &&
+				    sched_asym_ipcc_prefer(busiest, sgs))
+					return false;
+			}
 		}
 
 		break;
-- 
2.39.2

From cc96548bd32f891b0cdf162cdde5f8ca2ba98404 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:51 -0800
Subject: [PATCH] sched/fair: Use IPCC scores to select a busiest runqueue

For two runqueues of equal priority and equal number of running of tasks,
select the one whose current task would have the highest IPC class score
if placed on the destination CPU.

For now, use IPCC scores only for scheduling domains with the
SD_ASYM_PACKING flag.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/fair.c | 64 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 64 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 4f75e9964e8ca..fc42b58f1ba42 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -9187,6 +9187,37 @@ static bool sched_asym_ipcc_pick(struct sched_group *a,
 	return sched_asym_ipcc_prefer(a_stats, b_stats);
 }
 
+/**
+ * ipcc_score_delta - Get the IPCC score delta wrt the load balance's dst_cpu
+ * @p:		A task
+ * @env:	Load balancing environment
+ *
+ * Returns: The IPCC score delta that @p would get if placed in the destination
+ * CPU of @env. LONG_MIN to indicate that the delta should not be used.
+ */
+static long ipcc_score_delta(struct task_struct *p, struct lb_env *env)
+{
+	unsigned long score_src, score_dst;
+	unsigned short ipcc = p->ipcc;
+
+	if (!sched_ipcc_enabled())
+		return LONG_MIN;
+
+	/* Only asym_packing uses IPCC scores at the moment. */
+	if (!(env->sd->flags & SD_ASYM_PACKING))
+		return LONG_MIN;
+
+	score_dst = arch_get_ipcc_score(ipcc, env->dst_cpu);
+	if (IS_ERR_VALUE(score_dst))
+		return LONG_MIN;
+
+	score_src = arch_get_ipcc_score(ipcc, task_cpu(p));
+	if (IS_ERR_VALUE(score_src))
+		return LONG_MIN;
+
+	return score_dst - score_src;
+}
+
 #else /* CONFIG_IPC_CLASSES */
 static void update_sg_lb_ipcc_stats(int dst_cpu, struct sg_lb_stats *sgs,
 				    struct rq *rq)
@@ -9217,6 +9248,11 @@ static bool sched_asym_ipcc_pick(struct sched_group *a,
 	return false;
 }
 
+static long ipcc_score_delta(struct task_struct *p, struct lb_env *env)
+{
+	return LONG_MIN;
+}
+
 #endif /* CONFIG_IPC_CLASSES */
 
 /**
@@ -10377,6 +10413,7 @@ static struct rq *find_busiest_queue(struct lb_env *env,
 {
 	struct rq *busiest = NULL, *rq;
 	unsigned long busiest_util = 0, busiest_load = 0, busiest_capacity = 1;
+	long busiest_ipcc_delta = LONG_MIN;
 	unsigned int busiest_nr = 0;
 	int i;
 
@@ -10493,8 +10530,35 @@ static struct rq *find_busiest_queue(struct lb_env *env,
 
 		case migrate_task:
 			if (busiest_nr < nr_running) {
+				struct task_struct *curr;
+
 				busiest_nr = nr_running;
 				busiest = rq;
+
+				/*
+				 * Remember the IPCC score delta of busiest::curr.
+				 * We may need it to break a tie with other queues
+				 * with equal nr_running.
+				 */
+				curr = rcu_dereference(busiest->curr);
+				busiest_ipcc_delta = ipcc_score_delta(curr, env);
+			/*
+			 * If rq and busiest have the same number of running
+			 * tasks and IPC classes are supported, pick rq if doing
+			 * so would give rq::curr a bigger IPC boost on dst_cpu.
+			 */
+			} else if (busiest_nr == nr_running) {
+				struct task_struct *curr;
+				long delta;
+
+				curr = rcu_dereference(rq->curr);
+				delta = ipcc_score_delta(curr, env);
+
+				if (busiest_ipcc_delta < delta) {
+					busiest_ipcc_delta = delta;
+					busiest_nr = nr_running;
+					busiest = rq;
+				}
 			}
 			break;
 
-- 
2.39.2

From 8435406d47c8fb7c349f7615ebb1cd0a812afc90 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:52 -0800
Subject: [PATCH] thermal: intel: hfi: Introduce Intel Thread Director classes

On Intel hybrid parts, each type of CPU has specific performance and
energy efficiency capabilities. The Intel Thread Director technology
extends the Hardware Feedback Interface (HFI) to provide performance and
energy efficiency data for advanced classes of instructions.

Add support to parse per-class capabilities.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 drivers/thermal/intel/intel_hfi.c | 30 ++++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)

diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index 239afe02e5182..ae7eec197f680 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -79,7 +79,7 @@ union cpuid6_edx {
  * @ee_cap:		Energy efficiency capability
  *
  * Capabilities of a logical processor in the HFI table. These capabilities are
- * unitless.
+ * unitless and specific to each HFI class.
  */
 struct hfi_cpu_data {
 	u8	perf_cap;
@@ -91,7 +91,8 @@ struct hfi_cpu_data {
  * @perf_updated:	Hardware updated performance capabilities
  * @ee_updated:		Hardware updated energy efficiency capabilities
  *
- * Properties of the data in an HFI table.
+ * Properties of the data in an HFI table. There exists one header per each
+ * HFI class.
  */
 struct hfi_hdr {
 	u8	perf_updated;
@@ -129,16 +130,21 @@ struct hfi_instance {
 
 /**
  * struct hfi_features - Supported HFI features
+ * @nr_classes:		Number of classes supported
  * @nr_table_pages:	Size of the HFI table in 4KB pages
  * @cpu_stride:		Stride size to locate the capability data of a logical
  *			processor within the table (i.e., row stride)
+ * @class_stride:	Stride size to locate a class within the capability
+ *			data of a logical processor or the HFI table header
  * @hdr_size:		Size of the table header
  *
  * Parameters and supported features that are common to all HFI instances
  */
 struct hfi_features {
+	unsigned int	nr_classes;
 	size_t		nr_table_pages;
 	unsigned int	cpu_stride;
+	unsigned int	class_stride;
 	unsigned int	hdr_size;
 };
 
@@ -325,8 +331,8 @@ static void init_hfi_cpu_index(struct hfi_cpu_info *info)
 }
 
 /*
- * The format of the HFI table depends on the number of capabilities that the
- * hardware supports. Keep a data structure to navigate the table.
+ * The format of the HFI table depends on the number of capabilities and classes
+ * that the hardware supports. Keep a data structure to navigate the table.
  */
 static void init_hfi_instance(struct hfi_instance *hfi_instance)
 {
@@ -507,18 +513,30 @@ static __init int hfi_parse_features(void)
 	/* The number of 4KB pages required by the table */
 	hfi_features.nr_table_pages = edx.split.table_pages + 1;
 
+	/*
+	 * Capability fields of an HFI class are grouped together. Classes are
+	 * contiguous in memory.  Hence, use the number of supported features to
+	 * locate a specific class.
+	 */
+	hfi_features.class_stride = nr_capabilities;
+
+	/* For now, use only one class of the HFI table */
+	hfi_features.nr_classes = 1;
+
 	/*
 	 * The header contains change indications for each supported feature.
 	 * The size of the table header is rounded up to be a multiple of 8
 	 * bytes.
 	 */
-	hfi_features.hdr_size = DIV_ROUND_UP(nr_capabilities, 8) * 8;
+	hfi_features.hdr_size = DIV_ROUND_UP(nr_capabilities *
+					     hfi_features.nr_classes, 8) * 8;
 
 	/*
 	 * Data of each logical processor is also rounded up to be a multiple
 	 * of 8 bytes.
 	 */
-	hfi_features.cpu_stride = DIV_ROUND_UP(nr_capabilities, 8) * 8;
+	hfi_features.cpu_stride = DIV_ROUND_UP(nr_capabilities *
+					       hfi_features.nr_classes, 8) * 8;
 
 	return 0;
 }
-- 
2.39.2

From 78b14aa7d839c69d3073e92c54445a085b22df4f Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:53 -0800
Subject: [PATCH] x86/cpufeatures: Add the Intel Thread Director feature
 definitions

Intel Thread Director (ITD) provides hardware resources to classify
the current task. The classification reflects the type of instructions that
a task currently executes.

ITD extends the Hardware Feedback Interface table to provide performance
and energy efficiency capabilities for each of the supported classes of
tasks.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/cpufeatures.h       | 1 +
 arch/x86/include/asm/disabled-features.h | 8 +++++++-
 arch/x86/kernel/cpu/cpuid-deps.c         | 1 +
 3 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
index 92729c38853d1..02f10cd5c7536 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -342,6 +342,7 @@
 #define X86_FEATURE_HWP_EPP		(14*32+10) /* HWP Energy Perf. Preference */
 #define X86_FEATURE_HWP_PKG_REQ		(14*32+11) /* HWP Package Level Request */
 #define X86_FEATURE_HFI			(14*32+19) /* Hardware Feedback Interface */
+#define X86_FEATURE_ITD			(14*32+23) /* Intel Thread Director */
 
 /* AMD SVM Feature Identification, CPUID level 0x8000000a (EDX), word 15 */
 #define X86_FEATURE_NPT			(15*32+ 0) /* Nested Page Table support */
diff --git a/arch/x86/include/asm/disabled-features.h b/arch/x86/include/asm/disabled-features.h
index 33d2cd04d2544..7a668f6d0502d 100644
--- a/arch/x86/include/asm/disabled-features.h
+++ b/arch/x86/include/asm/disabled-features.h
@@ -87,6 +87,12 @@
 # define DISABLE_TDX_GUEST	(1 << (X86_FEATURE_TDX_GUEST & 31))
 #endif
 
+#ifdef CONFIG_IPC_CLASSES
+# define DISABLE_ITD	0
+#else
+# define DISABLE_ITD	(1 << (X86_FEATURE_ITD & 31))
+#endif
+
 /*
  * Make sure to add features to the correct mask
  */
@@ -104,7 +110,7 @@
 #define DISABLED_MASK11	(DISABLE_RETPOLINE|DISABLE_RETHUNK|DISABLE_UNRET)
 #define DISABLED_MASK12	0
 #define DISABLED_MASK13	0
-#define DISABLED_MASK14	0
+#define DISABLED_MASK14	(DISABLE_ITD)
 #define DISABLED_MASK15	0
 #define DISABLED_MASK16	(DISABLE_PKU|DISABLE_OSPKE|DISABLE_LA57|DISABLE_UMIP| \
 			 DISABLE_ENQCMD)
diff --git a/arch/x86/kernel/cpu/cpuid-deps.c b/arch/x86/kernel/cpu/cpuid-deps.c
index c881bcafba7d7..f6f8a3cd4f2ce 100644
--- a/arch/x86/kernel/cpu/cpuid-deps.c
+++ b/arch/x86/kernel/cpu/cpuid-deps.c
@@ -78,6 +78,7 @@ static const struct cpuid_dep cpuid_deps[] = {
 	{ X86_FEATURE_XFD,			X86_FEATURE_XSAVES    },
 	{ X86_FEATURE_XFD,			X86_FEATURE_XGETBV1   },
 	{ X86_FEATURE_AMX_TILE,			X86_FEATURE_XFD       },
+	{ X86_FEATURE_ITD,			X86_FEATURE_HFI       },
 	{}
 };
 
-- 
2.39.2

From b4ce6b73c21395712049b4eeb71e55bf617be6f2 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:54 -0800
Subject: [PATCH] thermal: intel: hfi: Store per-CPU IPCC scores

The scheduler reads the IPCC scores when balancing load. These reads can
be quite frequent. Hardware can also update the HFI table frequently.
Concurrent access may cause a lot of lock contention. It gets worse as the
number of CPUs increases.

Instead, create separate per-CPU IPCC scores that the scheduler can read
without the HFI table lock.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 drivers/thermal/intel/intel_hfi.c | 46 +++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index ae7eec197f680..e84dcfbef0dd9 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -29,6 +29,7 @@
 #include <linux/kernel.h>
 #include <linux/math.h>
 #include <linux/mutex.h>
+#include <linux/percpu.h>
 #include <linux/percpu-defs.h>
 #include <linux/printk.h>
 #include <linux/processor.h>
@@ -172,6 +173,43 @@ static struct workqueue_struct *hfi_updates_wq;
 #define HFI_UPDATE_INTERVAL		HZ
 #define HFI_MAX_THERM_NOTIFY_COUNT	16
 
+#ifdef CONFIG_IPC_CLASSES
+static int __percpu *hfi_ipcc_scores;
+
+static int alloc_hfi_ipcc_scores(void)
+{
+	if (!cpu_feature_enabled(X86_FEATURE_ITD))
+		return 0;
+
+	hfi_ipcc_scores = __alloc_percpu(sizeof(*hfi_ipcc_scores) *
+					 hfi_features.nr_classes,
+					 sizeof(*hfi_ipcc_scores));
+
+	return !hfi_ipcc_scores;
+}
+
+static void set_hfi_ipcc_score(void *caps, int cpu)
+{
+	int i, *hfi_class;
+
+	if (!cpu_feature_enabled(X86_FEATURE_ITD))
+		return;
+
+	hfi_class = per_cpu_ptr(hfi_ipcc_scores, cpu);
+
+	for (i = 0;  i < hfi_features.nr_classes; i++) {
+		struct hfi_cpu_data *class_caps;
+
+		class_caps = caps + i * hfi_features.class_stride;
+		WRITE_ONCE(hfi_class[i], class_caps->perf_cap);
+	}
+}
+
+#else
+static int alloc_hfi_ipcc_scores(void) { return 0; }
+static void set_hfi_ipcc_score(void *caps, int cpu) { }
+#endif /* CONFIG_IPC_CLASSES */
+
 static void get_hfi_caps(struct hfi_instance *hfi_instance,
 			 struct thermal_genl_cpu_caps *cpu_caps)
 {
@@ -194,6 +232,8 @@ static void get_hfi_caps(struct hfi_instance *hfi_instance,
 		cpu_caps[i].efficiency = caps->ee_cap << 2;
 
 		++i;
+
+		set_hfi_ipcc_score(caps, cpu);
 	}
 	raw_spin_unlock_irq(&hfi_instance->table_lock);
 }
@@ -572,8 +612,14 @@ void __init intel_hfi_init(void)
 	if (!hfi_updates_wq)
 		goto err_nomem;
 
+	if (alloc_hfi_ipcc_scores())
+		goto err_ipcc;
+
 	return;
 
+err_ipcc:
+	destroy_workqueue(hfi_updates_wq);
+
 err_nomem:
 	for (j = 0; j < i; ++j) {
 		hfi_instance = &hfi_instances[j];
-- 
2.39.2

From 8826c8ac9227c660003590b333f80a0d144cba2e Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:55 -0800
Subject: [PATCH] thermal: intel: hfi: Update the IPC class of the current task

Use Intel Thread Director classification to update the IPC class of a
task. Implement the arch_update_ipcc() interface of the scheduler.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/topology.h   |  6 ++++++
 drivers/thermal/intel/intel_hfi.c | 32 +++++++++++++++++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index 458c891a82736..ffcdac3f398f0 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -227,4 +227,10 @@ void init_freq_invariance_cppc(void);
 #define arch_init_invariance_cppc init_freq_invariance_cppc
 #endif
 
+#if defined(CONFIG_IPC_CLASSES) && defined(CONFIG_INTEL_HFI_THERMAL)
+void intel_hfi_update_ipcc(struct task_struct *curr);
+
+#define arch_update_ipcc intel_hfi_update_ipcc
+#endif /* defined(CONFIG_IPC_CLASSES) && defined(CONFIG_INTEL_HFI_THERMAL) */
+
 #endif /* _ASM_X86_TOPOLOGY_H */
diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index e84dcfbef0dd9..f2de597b3118a 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -74,6 +74,17 @@ union cpuid6_edx {
 	u32 full;
 };
 
+#ifdef CONFIG_IPC_CLASSES
+union hfi_thread_feedback_char_msr {
+	struct {
+		u64	classid : 8;
+		u64	__reserved : 55;
+		u64	valid : 1;
+	} split;
+	u64 full;
+};
+#endif
+
 /**
  * struct hfi_cpu_data - HFI capabilities per CPU
  * @perf_cap:		Performance capability
@@ -176,6 +187,27 @@ static struct workqueue_struct *hfi_updates_wq;
 #ifdef CONFIG_IPC_CLASSES
 static int __percpu *hfi_ipcc_scores;
 
+void intel_hfi_update_ipcc(struct task_struct *curr)
+{
+	union hfi_thread_feedback_char_msr msr;
+
+	/* We should not be here if ITD is not supported. */
+	if (!cpu_feature_enabled(X86_FEATURE_ITD)) {
+		pr_warn_once("task classification requested but not supported!");
+		return;
+	}
+
+	rdmsrl(MSR_IA32_HW_FEEDBACK_CHAR, msr.full);
+	if (!msr.split.valid)
+		return;
+
+	/*
+	 * 0 is a valid classification for Intel Thread Director. A scheduler
+	 * IPCC class of 0 means that the task is unclassified. Adjust.
+	 */
+	curr->ipcc = msr.split.classid + 1;
+}
+
 static int alloc_hfi_ipcc_scores(void)
 {
 	if (!cpu_feature_enabled(X86_FEATURE_ITD))
-- 
2.39.2

From e6c5f3a62bb1fc563887cca4df785b49f77ef0af Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:56 -0800
Subject: [PATCH] thermal: intel: hfi: Report the IPC class score of a CPU

Implement the arch_get_ipcc_score() interface of the scheduler. Use the
performance capabilities of the extended Hardware Feedback Interface table
as the IPC score.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/topology.h   |  2 ++
 drivers/thermal/intel/intel_hfi.c | 27 +++++++++++++++++++++++++++
 2 files changed, 29 insertions(+)

diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index ffcdac3f398f0..c4fcd9c3c634f 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -229,8 +229,10 @@ void init_freq_invariance_cppc(void);
 
 #if defined(CONFIG_IPC_CLASSES) && defined(CONFIG_INTEL_HFI_THERMAL)
 void intel_hfi_update_ipcc(struct task_struct *curr);
+unsigned long intel_hfi_get_ipcc_score(unsigned short ipcc, int cpu);
 
 #define arch_update_ipcc intel_hfi_update_ipcc
+#define arch_get_ipcc_score intel_hfi_get_ipcc_score
 #endif /* defined(CONFIG_IPC_CLASSES) && defined(CONFIG_INTEL_HFI_THERMAL) */
 
 #endif /* _ASM_X86_TOPOLOGY_H */
diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index f2de597b3118a..e8ad2be47e82f 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -208,6 +208,33 @@ void intel_hfi_update_ipcc(struct task_struct *curr)
 	curr->ipcc = msr.split.classid + 1;
 }
 
+unsigned long intel_hfi_get_ipcc_score(unsigned short ipcc, int cpu)
+{
+	unsigned short hfi_class;
+	int *scores;
+
+	if (cpu < 0 || cpu >= nr_cpu_ids)
+		return -EINVAL;
+
+	if (ipcc == IPC_CLASS_UNCLASSIFIED)
+		return -EINVAL;
+
+	/*
+	 * Scheduler IPC classes start at 1. HFI classes start at 0.
+	 * See note intel_hfi_update_ipcc().
+	 */
+	hfi_class = ipcc - 1;
+
+	if (hfi_class >= hfi_features.nr_classes)
+		return -EINVAL;
+
+	scores = per_cpu_ptr(hfi_ipcc_scores, cpu);
+	if (!scores)
+		return -ENODEV;
+
+	return READ_ONCE(scores[hfi_class]);
+}
+
 static int alloc_hfi_ipcc_scores(void)
 {
 	if (!cpu_feature_enabled(X86_FEATURE_ITD))
-- 
2.39.2

From 502fea0f97180e7e5a9fb1a88d31659d30813348 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:57 -0800
Subject: [PATCH] thermal: intel: hfi: Define a default class for unclassified
 tasks

A task may be unclassified if it has been recently created, spend most of
its lifetime sleeping, or hardware has not provided a classification.

Most tasks will be eventually classified as scheduler's IPC class 1
(HFI class 0). This class corresponds to the capabilities in the legacy,
classless, HFI table.

IPC class 1 is a reasonable choice until hardware provides an actual
classification. Meanwhile, the scheduler will place classes of tasks with
higher IPC scores on higher-performance CPUs.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 drivers/thermal/intel/intel_hfi.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index e8ad2be47e82f..0f6935705e559 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -187,6 +187,19 @@ static struct workqueue_struct *hfi_updates_wq;
 #ifdef CONFIG_IPC_CLASSES
 static int __percpu *hfi_ipcc_scores;
 
+/*
+ * A task may be unclassified if it has been recently created, spend most of
+ * its lifetime sleeping, or hardware has not provided a classification.
+ *
+ * Most tasks will be classified as scheduler's IPC class 1 (HFI class 0)
+ * eventually. Meanwhile, the scheduler will place classes of tasks with higher
+ * IPC scores on higher-performance CPUs.
+ *
+ * IPC class 1 is a reasonable choice. It matches the performance capability
+ * of the legacy, classless, HFI table.
+ */
+#define HFI_UNCLASSIFIED_DEFAULT 1
+
 void intel_hfi_update_ipcc(struct task_struct *curr)
 {
 	union hfi_thread_feedback_char_msr msr;
@@ -217,7 +230,7 @@ unsigned long intel_hfi_get_ipcc_score(unsigned short ipcc, int cpu)
 		return -EINVAL;
 
 	if (ipcc == IPC_CLASS_UNCLASSIFIED)
-		return -EINVAL;
+		ipcc = HFI_UNCLASSIFIED_DEFAULT;
 
 	/*
 	 * Scheduler IPC classes start at 1. HFI classes start at 0.
-- 
2.39.2

From 5c8a8a9dc0da32f5cb1321e5cd86e68c7221af74 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:58 -0800
Subject: [PATCH] thermal: intel: hfi: Enable the Intel Thread Director

Enable Intel Thread Director from the CPU hotplug callback: globally from
CPU0 and then enable the thread-classification hardware in each logical
processor individually.

Also, initialize the number of classes supported.

Let the scheduler know that it can start using IPC classes.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/msr-index.h  |  2 ++
 drivers/thermal/intel/intel_hfi.c | 40 +++++++++++++++++++++++++++++--
 2 files changed, 40 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
index 117e4e977b55d..b4a5de303b88a 100644
--- a/arch/x86/include/asm/msr-index.h
+++ b/arch/x86/include/asm/msr-index.h
@@ -1089,6 +1089,8 @@
 /* Hardware Feedback Interface */
 #define MSR_IA32_HW_FEEDBACK_PTR        0x17d0
 #define MSR_IA32_HW_FEEDBACK_CONFIG     0x17d1
+#define MSR_IA32_HW_FEEDBACK_THREAD_CONFIG 0x17d4
+#define MSR_IA32_HW_FEEDBACK_CHAR	0x17d2
 
 /* x2APIC locked status */
 #define MSR_IA32_XAPIC_DISABLE_STATUS	0xBD
diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index 0f6935705e559..21a0d246ca501 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -50,6 +50,8 @@
 /* Hardware Feedback Interface MSR configuration bits */
 #define HW_FEEDBACK_PTR_VALID_BIT		BIT(0)
 #define HW_FEEDBACK_CONFIG_HFI_ENABLE_BIT	BIT(0)
+#define HW_FEEDBACK_CONFIG_ITD_ENABLE_BIT	BIT(1)
+#define HW_FEEDBACK_THREAD_CONFIG_ENABLE_BIT	BIT(0)
 
 /* CPUID detection and enumeration definitions for HFI */
 
@@ -74,6 +76,15 @@ union cpuid6_edx {
 	u32 full;
 };
 
+union cpuid6_ecx {
+	struct {
+		u32	dont_care0:8;
+		u32	nr_classes:8;
+		u32	dont_care1:16;
+	} split;
+	u32 full;
+};
+
 #ifdef CONFIG_IPC_CLASSES
 union hfi_thread_feedback_char_msr {
 	struct {
@@ -498,6 +509,11 @@ void intel_hfi_online(unsigned int cpu)
 
 	init_hfi_cpu_index(info);
 
+	if (cpu_feature_enabled(X86_FEATURE_ITD)) {
+		msr_val = HW_FEEDBACK_THREAD_CONFIG_ENABLE_BIT;
+		wrmsrl(MSR_IA32_HW_FEEDBACK_THREAD_CONFIG, msr_val);
+	}
+
 	/*
 	 * Now check if the HFI instance of the package/die of @cpu has been
 	 * initialized (by checking its header). In such case, all we have to
@@ -553,8 +569,22 @@ void intel_hfi_online(unsigned int cpu)
 	 */
 	rdmsrl(MSR_IA32_HW_FEEDBACK_CONFIG, msr_val);
 	msr_val |= HW_FEEDBACK_CONFIG_HFI_ENABLE_BIT;
+
+	if (cpu_feature_enabled(X86_FEATURE_ITD))
+		msr_val |= HW_FEEDBACK_CONFIG_ITD_ENABLE_BIT;
+
 	wrmsrl(MSR_IA32_HW_FEEDBACK_CONFIG, msr_val);
 
+	/*
+	 * We have all we need to support IPC classes. Task classification is
+	 * now working.
+	 *
+	 * All class scores are zero until after the first HFI update. That is
+	 * OK. The scheduler queries these scores at every load balance.
+	 */
+	if (cpu_feature_enabled(X86_FEATURE_ITD))
+		sched_enable_ipc_classes();
+
 unlock:
 	mutex_unlock(&hfi_instance_lock);
 	return;
@@ -632,8 +662,14 @@ static __init int hfi_parse_features(void)
 	 */
 	hfi_features.class_stride = nr_capabilities;
 
-	/* For now, use only one class of the HFI table */
-	hfi_features.nr_classes = 1;
+	if (cpu_feature_enabled(X86_FEATURE_ITD)) {
+		union cpuid6_ecx ecx;
+
+		ecx.full = cpuid_ecx(CPUID_HFI_LEAF);
+		hfi_features.nr_classes = ecx.split.nr_classes;
+	} else {
+		hfi_features.nr_classes = 1;
+	}
 
 	/*
 	 * The header contains change indications for each supported feature.
-- 
2.39.2

From 721e6a04633384f5f0a837ee1347e99ba9f2357d Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:10:59 -0800
Subject: [PATCH] sched/task_struct: Add helpers for IPC classification

The unprocessed classification that hardware provides for a task may not
be usable by the scheduler: the classification may change too frequently or
architectures may want to consider extra factors. For instance, some
processors with Intel Thread Director need to consider the state of the SMT
siblings of a core.

Provide per-task helper variables that architectures can use to post-
process the classification that hardware provides.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 include/linux/sched.h | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index e58dc7503864c..63c2f88f0168a 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1535,7 +1535,17 @@ struct task_struct {
 	 * A hardware-defined classification of task that reflects but is
 	 * not identical to the number of instructions per cycle.
 	 */
-	unsigned short			ipcc;
+	unsigned int			ipcc : 9;
+	/*
+	 * A candidate classification that arch-specific implementations
+	 * qualify for correctness.
+	 */
+	unsigned int			ipcc_tmp : 9;
+	/*
+	 * Counter to filter out transient candidate classifications
+	 * of a task.
+	 */
+	unsigned int			ipcc_cntr : 14;
 #endif
 
 	/*
-- 
2.39.2

From afb386225a0c507949d96ea5122003edce813bb0 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:00 -0800
Subject: [PATCH] sched/core: Initialize helpers of task classification

Just as tasks start life unclassified, initialize the classification
auxiliar variables.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 kernel/sched/core.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 4782b1359eb89..d9a026845d51c 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4379,6 +4379,8 @@ static void __sched_fork(unsigned long clone_flags, struct task_struct *p)
 	p->se.vruntime			= 0;
 #ifdef CONFIG_IPC_CLASSES
 	p->ipcc				= IPC_CLASS_UNCLASSIFIED;
+	p->ipcc_tmp			= IPC_CLASS_UNCLASSIFIED;
+	p->ipcc_cntr			= 0;
 #endif
 	INIT_LIST_HEAD(&p->se.group_node);
 
-- 
2.39.2

From 20d6343c567733d5a2704e1f4bb437acd115683e Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:01 -0800
Subject: [PATCH] sched/fair: Introduce sched_smt_siblings_idle()

X86 needs to know the idle state of the SMT siblings of a CPU to improve
the accuracy of IPCC classification. X86 implements support for IPC classes
in the thermal HFI driver.

Rename is_core_idle() as sched_smt_siblings_idle() and make it available
outside the scheduler code.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 include/linux/sched.h |  2 ++
 kernel/sched/fair.c   | 21 +++++++++++++++------
 2 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 63c2f88f0168a..b049584f4c1a5 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -2446,4 +2446,6 @@ static inline void sched_core_fork(struct task_struct *p) { }
 
 extern void sched_set_stop_task(int cpu, struct task_struct *stop);
 
+extern bool sched_smt_siblings_idle(int cpu);
+
 #endif
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index fc42b58f1ba42..ff1fd953258bd 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1049,7 +1049,14 @@ update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
  * Scheduling class queueing methods:
  */
 
-static inline bool is_core_idle(int cpu)
+/**
+ * sched_smt_siblings_idle - Check whether SMT siblings of a CPU are idle
+ * @cpu:	The CPU to check
+ *
+ * Returns true if all the SMT siblings of @cpu are idle or @cpu does not have
+ * SMT siblings. The idle state of @cpu is not considered.
+ */
+bool sched_smt_siblings_idle(int cpu)
 {
 #ifdef CONFIG_SCHED_SMT
 	int sibling;
@@ -1755,7 +1762,7 @@ static inline int numa_idle_core(int idle_core, int cpu)
 	 * Prefer cores instead of packing HT siblings
 	 * and triggering future load balancing.
 	 */
-	if (is_core_idle(cpu))
+	if (sched_smt_siblings_idle(cpu))
 		idle_core = cpu;
 
 	return idle_core;
@@ -9306,7 +9313,8 @@ sched_asym(struct lb_env *env, struct sd_lb_stats *sds,  struct sg_lb_stats *sgs
 	 * If the destination CPU has SMT siblings, env->idle != CPU_NOT_IDLE
 	 * is not sufficient. We need to make sure the whole core is idle.
 	 */
-	if (sds->local->flags & SD_SHARE_CPUCAPACITY && !is_core_idle(env->dst_cpu))
+	if (sds->local->flags & SD_SHARE_CPUCAPACITY &&
+	    !sched_smt_siblings_idle(env->dst_cpu))
 		return false;
 
 	/* Only do SMT checks if either local or candidate have SMT siblings. */
@@ -10475,7 +10483,8 @@ static struct rq *find_busiest_queue(struct lb_env *env,
 		    sched_asym_prefer(i, env->dst_cpu) &&
 		    nr_running == 1) {
 			if (env->sd->flags & SD_SHARE_CPUCAPACITY ||
-			    (!(env->sd->flags & SD_SHARE_CPUCAPACITY) && is_core_idle(i)))
+			    (!(env->sd->flags & SD_SHARE_CPUCAPACITY) &&
+			     sched_smt_siblings_idle(i)))
 				continue;
 		}
 
@@ -10604,7 +10613,7 @@ asym_active_balance(struct lb_env *env)
 		 * busy sibling.
 		 */
 		return sched_asym_prefer(env->dst_cpu, env->src_cpu) ||
-		       !is_core_idle(env->src_cpu);
+		       !sched_smt_siblings_idle(env->src_cpu);
 	}
 
 	return false;
@@ -11351,7 +11360,7 @@ static void nohz_balancer_kick(struct rq *rq)
 				 */
 				if (sd->flags & SD_SHARE_CPUCAPACITY ||
 				    (!(sd->flags & SD_SHARE_CPUCAPACITY) &&
-				     is_core_idle(i))) {
+				     sched_smt_siblings_idle(i))) {
 					flags = NOHZ_STATS_KICK | NOHZ_BALANCE_KICK;
 					goto unlock;
 				}
-- 
2.39.2

From 0a346e6136d9ca6c52ef6061e5d24b231629cd67 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:02 -0800
Subject: [PATCH] thermal: intel: hfi: Implement model-specific checks for task
 classification

In Alder Lake and Raptor Lake, the result of thread classification is more
accurate when only one SMT sibling is busy. Classification results for
class 2 and 3 are always reliable.

To avoid unnecessary migrations, only update the class of a task if it has
been the same during 4 consecutive user ticks.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 drivers/thermal/intel/intel_hfi.c | 60 ++++++++++++++++++++++++++++++-
 1 file changed, 59 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/intel/intel_hfi.c b/drivers/thermal/intel/intel_hfi.c
index 21a0d246ca501..751b84b6b8fda 100644
--- a/drivers/thermal/intel/intel_hfi.c
+++ b/drivers/thermal/intel/intel_hfi.c
@@ -40,6 +40,7 @@
 #include <linux/workqueue.h>
 
 #include <asm/msr.h>
+#include <asm/intel-family.h>
 
 #include "../thermal_core.h"
 #include "intel_hfi.h"
@@ -211,9 +212,64 @@ static int __percpu *hfi_ipcc_scores;
  */
 #define HFI_UNCLASSIFIED_DEFAULT 1
 
+#define CLASS_DEBOUNCER_SKIPS 4
+
+/**
+ * debounce_and_update_class() - Process and update a task's classification
+ *
+ * @p:		The task of which the classification will be updated
+ * @new_ipcc:	The new IPC classification
+ *
+ * Update the classification of @p with the new value that hardware provides.
+ * Only update the classification of @p if it has been the same during
+ * CLASS_DEBOUNCER_SKIPS consecutive ticks.
+ */
+static void debounce_and_update_class(struct task_struct *p, u8 new_ipcc)
+{
+	u16 debounce_skip;
+
+	/* The class of @p changed. Only restart the debounce counter. */
+	if (p->ipcc_tmp != new_ipcc) {
+		p->ipcc_cntr = 1;
+		goto out;
+	}
+
+	/*
+	 * The class of @p did not change. Update it if it has been the same
+	 * for CLASS_DEBOUNCER_SKIPS user ticks.
+	 */
+	debounce_skip = p->ipcc_cntr + 1;
+	if (debounce_skip < CLASS_DEBOUNCER_SKIPS)
+		p->ipcc_cntr++;
+	else
+		p->ipcc = new_ipcc;
+
+out:
+	p->ipcc_tmp = new_ipcc;
+}
+
+static bool classification_is_accurate(u8 hfi_class, bool smt_siblings_idle)
+{
+	switch (boot_cpu_data.x86_model) {
+	case INTEL_FAM6_ALDERLAKE:
+	case INTEL_FAM6_ALDERLAKE_L:
+	case INTEL_FAM6_RAPTORLAKE:
+	case INTEL_FAM6_RAPTORLAKE_P:
+	case INTEL_FAM6_RAPTORLAKE_S:
+		if (hfi_class == 3 || hfi_class == 2 || smt_siblings_idle)
+			return true;
+
+		return false;
+
+	default:
+		return true;
+	}
+}
+
 void intel_hfi_update_ipcc(struct task_struct *curr)
 {
 	union hfi_thread_feedback_char_msr msr;
+	bool idle;
 
 	/* We should not be here if ITD is not supported. */
 	if (!cpu_feature_enabled(X86_FEATURE_ITD)) {
@@ -229,7 +285,9 @@ void intel_hfi_update_ipcc(struct task_struct *curr)
 	 * 0 is a valid classification for Intel Thread Director. A scheduler
 	 * IPCC class of 0 means that the task is unclassified. Adjust.
 	 */
-	curr->ipcc = msr.split.classid + 1;
+	idle = sched_smt_siblings_idle(task_cpu(curr));
+	if (classification_is_accurate(msr.split.classid, idle))
+		debounce_and_update_class(curr, msr.split.classid + 1);
 }
 
 unsigned long intel_hfi_get_ipcc_score(unsigned short ipcc, int cpu)
-- 
2.39.2

From 0ae9abd0c6e2109c538599229c213539fb1386ca Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:03 -0800
Subject: [PATCH] x86/cpufeatures: Add feature bit for HRESET

The HRESET instruction prevents the classification of the current task
from influencing the classification of the next task when running serially
on the same logical processor.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/cpufeatures.h | 1 +
 arch/x86/include/asm/msr-index.h   | 4 +++-
 arch/x86/kernel/cpu/scattered.c    | 1 +
 3 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
index 02f10cd5c7536..2fc261302f5cc 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -307,6 +307,7 @@
 
 
 #define X86_FEATURE_MSR_TSX_CTRL	(11*32+20) /* "" MSR IA32_TSX_CTRL (Intel) implemented */
+#define X86_FEATURE_HRESET		(11*32+23) /* Hardware history reset instruction */
 
 /* Intel-defined CPU features, CPUID level 0x00000007:1 (EAX), word 12 */
 #define X86_FEATURE_AVX_VNNI		(12*32+ 4) /* AVX VNNI instructions */
diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
index b4a5de303b88a..4ef01ce8eadb5 100644
--- a/arch/x86/include/asm/msr-index.h
+++ b/arch/x86/include/asm/msr-index.h
@@ -1092,6 +1092,9 @@
 #define MSR_IA32_HW_FEEDBACK_THREAD_CONFIG 0x17d4
 #define MSR_IA32_HW_FEEDBACK_CHAR	0x17d2
 
+/* Hardware History Reset  */
+#define MSR_IA32_HW_HRESET_ENABLE	0x17da
+
 /* x2APIC locked status */
 #define MSR_IA32_XAPIC_DISABLE_STATUS	0xBD
 #define LEGACY_XAPIC_DISABLED		BIT(0) /*
@@ -1099,5 +1102,4 @@
 						* disabling x2APIC will cause
 						* a #GP
 						*/
-
 #endif /* _ASM_X86_MSR_INDEX_H */
diff --git a/arch/x86/kernel/cpu/scattered.c b/arch/x86/kernel/cpu/scattered.c
index fc01f81f6e2a3..b8ad312d36949 100644
--- a/arch/x86/kernel/cpu/scattered.c
+++ b/arch/x86/kernel/cpu/scattered.c
@@ -28,6 +28,7 @@ static const struct cpuid_bit cpuid_bits[] = {
 	{ X86_FEATURE_EPB,		CPUID_ECX,  3, 0x00000006, 0 },
 	{ X86_FEATURE_INTEL_PPIN,	CPUID_EBX,  0, 0x00000007, 1 },
 	{ X86_FEATURE_RRSBA_CTRL,	CPUID_EDX,  2, 0x00000007, 2 },
+	{ X86_FEATURE_HRESET,		CPUID_EAX, 22, 0x00000007, 1 },
 	{ X86_FEATURE_CQM_LLC,		CPUID_EDX,  1, 0x0000000f, 0 },
 	{ X86_FEATURE_CQM_OCCUP_LLC,	CPUID_EDX,  0, 0x0000000f, 1 },
 	{ X86_FEATURE_CQM_MBM_TOTAL,	CPUID_EDX,  1, 0x0000000f, 1 },
-- 
2.39.2

From dcf0f4ab904b1ef3090f2de141f55ff42bf9b0f2 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:04 -0800
Subject: [PATCH] x86/hreset: Configure history reset

Configure the MSR that controls the behavior of HRESET on each logical
processor.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/kernel/cpu/common.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index c34bdba57993a..98b48cf80ec21 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -411,6 +411,26 @@ static __always_inline void setup_umip(struct cpuinfo_x86 *c)
 	cr4_clear_bits(X86_CR4_UMIP);
 }
 
+static u32 hardware_history_features __ro_after_init;
+
+static __always_inline void setup_hreset(struct cpuinfo_x86 *c)
+{
+	if (!cpu_feature_enabled(X86_FEATURE_HRESET))
+		return;
+
+	/*
+	 * Use on all CPUs the hardware history features that the boot
+	 * CPU supports.
+	 */
+	if (c == &boot_cpu_data)
+		hardware_history_features = cpuid_ebx(0x20);
+
+	if (!hardware_history_features)
+		return;
+
+	wrmsrl(MSR_IA32_HW_HRESET_ENABLE, hardware_history_features);
+}
+
 /* These bits should not change their value after CPU init is finished. */
 static const unsigned long cr4_pinned_mask =
 	X86_CR4_SMEP | X86_CR4_SMAP | X86_CR4_UMIP |
@@ -1828,10 +1848,11 @@ static void identify_cpu(struct cpuinfo_x86 *c)
 	/* Disable the PN if appropriate */
 	squash_the_stupid_serial_number(c);
 
-	/* Set up SMEP/SMAP/UMIP */
+	/* Set up SMEP/SMAP/UMIP/HRESET */
 	setup_smep(c);
 	setup_smap(c);
 	setup_umip(c);
+	setup_hreset(c);
 
 	/* Enable FSGSBASE instructions if available. */
 	if (cpu_has(c, X86_FEATURE_FSGSBASE)) {
-- 
2.39.2

From d562f8856447bb5019dad771dab745821b383939 Mon Sep 17 00:00:00 2001
From: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Date: Mon, 6 Feb 2023 21:11:05 -0800
Subject: [PATCH] x86/process: Reset hardware history in context switch

Reset the classification history of the current task when switching to the
next task. Hardware will start the classification of the next task from
scratch.

Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ionela Voinescu <ionela.voinescu@arm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tim C. Chen <tim.c.chen@intel.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: x86@kernel.org
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Patchset: intel-thread-director
---
 arch/x86/include/asm/hreset.h | 30 ++++++++++++++++++++++++++++++
 arch/x86/kernel/cpu/common.c  |  7 +++++++
 arch/x86/kernel/process_32.c  |  3 +++
 arch/x86/kernel/process_64.c  |  3 +++
 4 files changed, 43 insertions(+)
 create mode 100644 arch/x86/include/asm/hreset.h

diff --git a/arch/x86/include/asm/hreset.h b/arch/x86/include/asm/hreset.h
new file mode 100644
index 0000000000000..d68ca2fb8642b
--- /dev/null
+++ b/arch/x86/include/asm/hreset.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_X86_HRESET_H
+
+/**
+ * HRESET - History reset. Available since binutils v2.36.
+ *
+ * Request the processor to reset the history of task classification on the
+ * current logical processor. The history components to be
+ * reset are specified in %eax. Only bits specified in CPUID(0x20).EBX
+ * and enabled in the IA32_HRESET_ENABLE MSR can be selected.
+ *
+ * The assembly code looks like:
+ *
+ *	hreset %eax
+ *
+ * The corresponding machine code looks like:
+ *
+ *	F3 0F 3A F0 ModRM Imm
+ *
+ * The value of ModRM is 0xc0 to specify %eax register addressing.
+ * The ignored immediate operand is set to 0.
+ *
+ * The instruction is documented in the Intel SDM.
+ */
+
+#define __ASM_HRESET  ".byte 0xf3, 0xf, 0x3a, 0xf0, 0xc0, 0x0"
+
+void reset_hardware_history(void);
+
+#endif /* _ASM_X86_HRESET_H */
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 98b48cf80ec21..67bf6fbae082a 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -52,6 +52,7 @@
 #include <asm/cpu.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
+#include <asm/hreset.h>
 #include <asm/memtype.h>
 #include <asm/microcode.h>
 #include <asm/microcode_intel.h>
@@ -413,6 +414,12 @@ static __always_inline void setup_umip(struct cpuinfo_x86 *c)
 
 static u32 hardware_history_features __ro_after_init;
 
+void reset_hardware_history(void)
+{
+	asm_inline volatile (ALTERNATIVE("", __ASM_HRESET, X86_FEATURE_HRESET)
+			     : : "a" (hardware_history_features) : "memory");
+}
+
 static __always_inline void setup_hreset(struct cpuinfo_x86 *c)
 {
 	if (!cpu_feature_enabled(X86_FEATURE_HRESET))
diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c
index ceab14b6118f7..888cdb7624dcd 100644
--- a/arch/x86/kernel/process_32.c
+++ b/arch/x86/kernel/process_32.c
@@ -52,6 +52,7 @@
 #include <asm/switch_to.h>
 #include <asm/vm86.h>
 #include <asm/resctrl.h>
+#include <asm/hreset.h>
 #include <asm/proto.h>
 
 #include "process.h"
@@ -214,6 +215,8 @@
 	/* Load the Intel cache allocation PQR MSR. */
 	resctrl_sched_in();
 
+	reset_hardware_history();
+
 	return prev_p;
 }
 
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 7f94dbbc397b7..d0d3c33237475 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -53,6 +53,7 @@
 #include <asm/xen/hypervisor.h>
 #include <asm/vdso.h>
 #include <asm/resctrl.h>
+#include <asm/hreset.h>
 #include <asm/unistd.h>
 #include <asm/fsgsbase.h>
 #ifdef CONFIG_IA32_EMULATION
@@ -658,6 +659,8 @@
 	/* Load the Intel cache allocation PQR MSR. */
 	resctrl_sched_in();
 
+	reset_hardware_history();
+
 	return prev_p;
 }
 
-- 
2.39.2