Skip to contentSkip to conversation

thermal: intel_pch_thermal: Support skylake PCH 100 series thermal

PATCH Superseded OGAWA Hirofumi 3 participants 9 messages Oct 17, 2016 – Nov 15, 2016 Thermal

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 thermal: intel_pch_thermal: Support skylake PCH 100 series thermal

Using registers are compatible with current driver.

---

 drivers/thermal/intel_pch_thermal.c |   48 +++++++++++++++++++++--------------
 1 file changed, 30 insertions(+), 18 deletions(-)
drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +30 -18
diff -puN drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup drivers/thermal/intel_pch_thermal.c
@@ -220,32 +220,39 @@ static struct thermal_zone_device_ops tz
 	.get_trip_temp = pch_get_trip_temp,
 };
 
+enum board_ids {
+	board_wpt,
+	board_skl,
+};
+
+static const struct board_info {
+	const char *name;
+	const struct pch_dev_ops *ops;
+} board_info[] = {
+	[board_wpt] = {
+		.name = "pch_wildcat_point",
+		.ops = &pch_dev_ops_wpt,
+	},
+	[board_skl] = {
+		.name = "pch_skylake",
+		.ops = &pch_dev_ops_wpt,
+	},
+};
 
 static int intel_pch_thermal_probe(struct pci_dev *pdev,
 				   const struct pci_device_id *id)
 {
+	enum board_ids board_id = id->driver_data;
+	const struct board_info *bi = &board_info[board_id];
 	struct pch_thermal_device *ptd;
 	int err;
 	int nr_trips;
-	char *dev_name;
 
 	ptd = devm_kzalloc(&pdev->dev, sizeof(*ptd), GFP_KERNEL);
 	if (!ptd)
 		return -ENOMEM;
 
-	switch (pdev->device) {
-	case PCH_THERMAL_DID_WPT:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_wildcat_point";
-		break;
-	case PCH_THERMAL_DID_SKL:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_skylake";
-		break;
-	default:
-		dev_err(&pdev->dev, "unknown pch thermal device\n");
-		return -ENODEV;
-	}
+	ptd->ops = bi->ops;
 
 	pci_set_drvdata(pdev, ptd);
 	ptd->pdev = pdev;
@@ -273,11 +280,11 @@ static int intel_pch_thermal_probe(struc
 	if (err)
 		goto error_cleanup;
 
-	ptd->tzd = thermal_zone_device_register(dev_name, nr_trips, 0, ptd,
+	ptd->tzd = thermal_zone_device_register(bi->name, nr_trips, 0, ptd,
 						&tzd_ops, NULL, 0, 0);
 	if (IS_ERR(ptd->tzd)) {
 		dev_err(&pdev->dev, "Failed to register thermal zone %s\n",
-			dev_name);
+			bi->name);
 		err = PTR_ERR(ptd->tzd);
 		goto error_cleanup;
 	}
@@ -322,8 +329,13 @@ static int intel_pch_thermal_resume(stru
 }
 
 static struct pci_device_id intel_pch_thermal_id[] = {
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT) },
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	/* skylake PCH 100 series */
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0xa131),
+	  .driver_data = board_skl, },
 	{ 0, },
 };
 MODULE_DEVICE_TABLE(pci, intel_pch_thermal_id);

_

Signature · 1 line
OGAWA Hirofumi <[email protected]>
#1 lore raw
Quoted 1 line · OGAWA Hirofumi

Using registers are compatible with current driver.

This description is not enough. Without going through the whole code, I was not able to find out what this patch is doing.

Basically you want to tie the themal_zone name and device operations to a board, rather than to PCI devices ids. In this way multiple PCI ids for the same board can be accounted for.

Also need to consider another patch sent to mailing list for supporting Haswell PCH.

One comment below in the code.

[...]

drivers/thermal/intel_pch_thermal.c static int intel_pch_thermal_resume(stru · line 337

Quoted 9 lines

 static struct pci_device_id intel_pch_thermal_id[] = {

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +6 -2
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT) },
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	/* skylake PCH 100 series */
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0xa131),

Change a131 to a define (PCH_THERMAL_DID_SKL_H)

Thanks,
Srinivas

#2 lore raw
Quoted 8 lines · Srinivas
Quoted 1 line · OGAWA Hirofumi

Using registers are compatible with current driver.

This description is not enough. Without going through the whole code, I was not able to find out what this patch is doing.

Basically you want to tie the themal_zone name and device operations to a board, rather than to PCI devices ids. In this way multiple PCI ids for the same board can be accounted for.

Exactly. I thought the patch is enough easy to read, and changelog would be better to not explain what code itself, and wrote note of code (registers are same with current).

Well, ok. I will explain what patch does. Your explain looks like good, so I will borrow it :)

Quoted 2 lines

Also need to consider another patch sent to mailing list for supporting Haswell PCH.

Sounds good. I think the patch makes it easier.

Quoted 4 lines

One comment below in the code.

[...]

Quoted 10 lines

drivers/thermal/intel_pch_thermal.c static int intel_pch_thermal_resume(stru · line 337

Quoted 9 lines

 static struct pci_device_id intel_pch_thermal_id[] = {

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +6 -2
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT) },
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	/* skylake PCH 100 series */
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0xa131),

Change a131 to a define (PCH_THERMAL_DID_SKL_H)

I don't think it helps anything actually, because no users of define anymore. Well, anyway, ok, I don't care either case.

Thanks.

Signature · 1 line
OGAWA Hirofumi <[email protected]>
#3 lore raw
Quoted 16 lines · OGAWA Hirofumi
Quoted 8 lines · Srinivas
Quoted 2 lines · OGAWA Hirofumi

Using registers are compatible with current driver.

This description is not enough. Without going through the whole code, I
was not able to find out what this patch is doing.

Quoted 5 lines

Basically you want to tie the themal_zone name and device operations to
a board, rather than to PCI devices ids. In this way multiple PCI ids
for the same board can be accounted for.

Quoted 4 lines

Exactly. I thought the patch is enough easy to read, and changelog would
be better to not explain what code itself, and wrote note of code (registers are same with current).

Reviewers will look at the description before to look at the code.

[...]

Quoted 13 lines
Quoted 4 lines
Quoted 4 lines

 static struct pci_device_id intel_pch_thermal_id[] = {

(quoted diff) +0 -1
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT)

},

(quoted diff) +0 -1
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL)
Quoted 8 lines

drivers/thermal/intel_pch_thermal.c static int intel_pch_thermal_resume(stru · line 337

Quoted 7 lines

},

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +6 -0
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	/* skylake PCH 100 series */
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0xa131),

Change a131 to a define (PCH_THERMAL_DID_SKL_H)

Quoted 2 lines

I don't think it helps anything actually, because no users of define anymore. Well, anyway, ok, I don't care either case.

Only for the sake consistency with rest of the ids defined here.

Thanks,
Srinivas

#4 lore raw
Quoted 6 lines · Srinivas
Quoted 4 lines

Exactly. I thought the patch is enough easy to read, and changelog would
be better to not explain what code itself, and wrote note of code (registers are same with current).

Reviewers will look at the description before to look at the code.

Oops, I was thinking I sent updated patch before this.

Signature · 1 line
OGAWA Hirofumi <[email protected]>
#5 lore raw

PATCH v2 1/1 thermal: intel_pch_thermal: Support skylake PCH 100 series thermal

This patch uses .driver_data and board_info[] to make per pci device behavior table (name and ops), instead of adding the code for each pci device in switch-case. This will make easier to add new pci device ids.

Then this adds new device id actually for skylake PCH 100 series (using registers are compatible with currently driver, so no need to change except adding device id to table).

---

 drivers/thermal/intel_pch_thermal.c |   48 +++++++++++++++++++++--------------
 1 file changed, 30 insertions(+), 18 deletions(-)
drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +30 -18
diff -puN drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup drivers/thermal/intel_pch_thermal.c
@@ -26,6 +26,7 @@
 /* Intel PCH thermal Device IDs */
 #define PCH_THERMAL_DID_WPT	0x9CA4 /* Wildcat Point */
 #define PCH_THERMAL_DID_SKL	0x9D31 /* Skylake PCH */
+#define PCH_THERMAL_DID_SKL_H	0xA131 /* Skylake PCH 100 series */
 
 /* Wildcat Point-LP  PCH Thermal registers */
 #define WPT_TEMP	0x0000	/* Temperature */
@@ -220,32 +221,39 @@ static struct thermal_zone_device_ops tz
 	.get_trip_temp = pch_get_trip_temp,
 };
 
+enum board_ids {
+	board_wpt,
+	board_skl,
+};
+
+static const struct board_info {
+	const char *name;
+	const struct pch_dev_ops *ops;
+} board_info[] = {
+	[board_wpt] = {
+		.name = "pch_wildcat_point",
+		.ops = &pch_dev_ops_wpt,
+	},
+	[board_skl] = {
+		.name = "pch_skylake",
+		.ops = &pch_dev_ops_wpt,
+	},
+};
 
 static int intel_pch_thermal_probe(struct pci_dev *pdev,
 				   const struct pci_device_id *id)
 {
+	enum board_ids board_id = id->driver_data;
+	const struct board_info *bi = &board_info[board_id];
 	struct pch_thermal_device *ptd;
 	int err;
 	int nr_trips;
-	char *dev_name;
 
 	ptd = devm_kzalloc(&pdev->dev, sizeof(*ptd), GFP_KERNEL);
 	if (!ptd)
 		return -ENOMEM;
 
-	switch (pdev->device) {
-	case PCH_THERMAL_DID_WPT:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_wildcat_point";
-		break;
-	case PCH_THERMAL_DID_SKL:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_skylake";
-		break;
-	default:
-		dev_err(&pdev->dev, "unknown pch thermal device\n");
-		return -ENODEV;
-	}
+	ptd->ops = bi->ops;
 
 	pci_set_drvdata(pdev, ptd);
 	ptd->pdev = pdev;
@@ -273,11 +281,11 @@ static int intel_pch_thermal_probe(struc
 	if (err)
 		goto error_cleanup;
 
-	ptd->tzd = thermal_zone_device_register(dev_name, nr_trips, 0, ptd,
+	ptd->tzd = thermal_zone_device_register(bi->name, nr_trips, 0, ptd,
 						&tzd_ops, NULL, 0, 0);
 	if (IS_ERR(ptd->tzd)) {
 		dev_err(&pdev->dev, "Failed to register thermal zone %s\n",
-			dev_name);
+			bi->name);
 		err = PTR_ERR(ptd->tzd);
 		goto error_cleanup;
 	}
@@ -322,8 +330,12 @@ static int intel_pch_thermal_resume(stru
 }
 
 static struct pci_device_id intel_pch_thermal_id[] = {
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT) },
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL_H),
+	  .driver_data = board_skl, },
 	{ 0, },
 };
 MODULE_DEVICE_TABLE(pci, intel_pch_thermal_id);

_

Signature · 1 line
OGAWA Hirofumi <[email protected]>
Reviewed-by Srinivas Pandruvada on v2 1/1
#6 lore raw
Quoted 11 lines · OGAWA Hirofumi

This patch uses .driver_data and board_info[] to make per pci device behavior table (name and ops), instead of adding the code for each pci
device in switch-case. This will make easier to add new pci device ids.

Then this adds new device id actually for skylake PCH 100 series (using registers are compatible with currently driver, so no need to change except adding device id to table).

Quoted 110 lines

---

 drivers/thermal/intel_pch_thermal.c |   48 +++++++++++++++++++++--
------------

 1 file changed, 30 insertions(+), 18 deletions(-)

diff -puN drivers/thermal/intel_pch_thermal.c~intel_pch_thermal- cleanup drivers/thermal/intel_pch_thermal.c
--- linux/drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-
cleanup 2016-10-18 06:02:07.469831874 +0900
+++ linux-hirofumi/drivers/thermal/intel_pch_thermal.c 2016-
10-19 05:13:39.028672579 +0900

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +1 -0
@@ -26,6 +26,7 @@
  /* Intel PCH thermal Device IDs */
  #define PCH_THERMAL_DID_WPT	0x9CA4 /* Wildcat Point */
  #define PCH_THERMAL_DID_SKL	0x9D31 /* Skylake PCH */
+#define PCH_THERMAL_DID_SKL_H	0xA131 /* Skylake PCH 100

series */

 /* Wildcat Point-LP  PCH Thermal registers */
 #define WPT_TEMP 0x0000 /* Temperature */

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +0 -0
@@ -220,32 +221,39 @@ static struct thermal_zone_device_ops tz
  	.get_trip_temp = pch_get_trip_temp,
  };

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +18 -0
+enum board_ids {
+	board_wpt,
+	board_skl,
+};
+
+static const struct board_info {
+	const char *name;
+	const struct pch_dev_ops *ops;
+} board_info[] = {
+	[board_wpt] = {
+		.name = "pch_wildcat_point",
+		.ops = &pch_dev_ops_wpt,
+	},
+	[board_skl] = {
+		.name = "pch_skylake",
+		.ops = &pch_dev_ops_wpt,
+	},
+};


 static int intel_pch_thermal_probe(struct pci_dev *pdev, const struct pci_device_id *id) {

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +2 -0
+	enum board_ids board_id = id->driver_data;
+	const struct board_info *bi = &board_info[board_id];

  struct pch_thermal_device *ptd;
  int err;
  int nr_trips;

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +0 -1
-	char *dev_name;


  ptd = devm_kzalloc(&pdev->dev, sizeof(*ptd), GFP_KERNEL); if (!ptd)
  return -ENOMEM;

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +1 -13
-	switch (pdev->device) {
-	case PCH_THERMAL_DID_WPT:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_wildcat_point";
-		break;
-	case PCH_THERMAL_DID_SKL:
-		ptd->ops = &pch_dev_ops_wpt;
-		dev_name = "pch_skylake";
-		break;
-	default:
-		dev_err(&pdev->dev, "unknown pch thermal device\n");
-		return -ENODEV;
-	}
+	ptd->ops = bi->ops;


  pci_set_drvdata(pdev, ptd);
  ptd->pdev = pdev;

(quoted diff) +0 -0
@@ -273,11 +281,11 @@ static int intel_pch_thermal_probe(struc
  	if (err)
  		goto error_cleanup;

(quoted diff) +0 -1
-	ptd->tzd = thermal_zone_device_register(dev_name, nr_trips,

0, ptd,

(quoted diff) +1 -0
+	ptd->tzd = thermal_zone_device_register(bi->name, nr_trips,

0, ptd,
  &tzd_ops, NULL, 0, 0);
  if (IS_ERR(ptd->tzd)) {
  dev_err(&pdev->dev, "Failed to register thermal zone %s\n",

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +1 -1
-			dev_name);
+			bi->name);

  err = PTR_ERR(ptd->tzd);
  goto error_cleanup;
  }

(quoted diff) +0 -0
@@ -322,8 +330,12 @@ static int intel_pch_thermal_resume(stru
  }


 static struct pci_device_id intel_pch_thermal_id[] = {

drivers/thermal/intel_pch_thermal.c renamed from drivers/thermal/intel_pch_thermal.c~intel_pch_thermal-cleanup +6 -2
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT) },
-	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_WPT),
+	  .driver_data = board_wpt, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL),
+	  .driver_data = board_skl, },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCH_THERMAL_DID_SKL_H),
+	  .driver_data = board_skl, },

  { 0, },
 };
 MODULE_DEVICE_TABLE(pci, intel_pch_thermal_id);
_

reviewer
#7 lore raw
Quoted 8 lines · Srinivas
Quoted 7 lines · OGAWA Hirofumi

This patch uses .driver_data and board_info[] to make per pci device behavior table (name and ops), instead of adding the code for each pci
device in switch-case. This will make easier to add new pci device ids.

Quoted 6 lines
Quoted 5 lines

Then this adds new device id actually for skylake PCH 100 series (using registers are compatible with currently driver, so no need to change except adding device id to table).

hmmm, this patch is queued in my tree for some time. However, it conflicts with
commit 33086a9a3071812a829d4b63bed826736a5d8b17
Author: Srinivas Pandruvada <[email protected]> Date:   Mon Oct 3 12:36:02 2016 -0700

    thermal: intel_pch_thermal: Enable Haswell PCH
    Added missing support for Haswell PCH thermal sensor. which has been shipped in v4.9-rc3.

Rebased patch sent, please double check.

thanks,
rui

#8 lore raw
Quoted 6 lines · Zhang Rui

hmmm, this patch is queued in my tree for some time. However, it conflicts with
commit 33086a9a3071812a829d4b63bed826736a5d8b17

Author: Srinivas Pandruvada <[email protected]>
Date:   Mon Oct 3 12:36:02 2016 -0700
Quoted 6 lines

thermal: intel_pch_thermal: Enable Haswell PCH

Added missing support for Haswell PCH thermal sensor.
which has been shipped in v4.9-rc3.

Rebased patch sent, please double check.

"[PATCH v3] thermal: intel_pch_thermal: ..." looks good to me.

[I was thinking I checked latest kernel though, it looks like I was missing something.]

Thanks.

Signature · 1 line
OGAWA Hirofumi <[email protected]>