On 07-09-2024 16:48, Raag Jadav wrote:
On Fri, Sep 06, 2024 at 04:33:12PM +0530, Nilawar, Badal wrote:
On 06-09-2024 11:56, Anshuman Gupta wrote:
On 2024-09-05 at 22:18:17 +0300, Raag Jadav wrote:
On Thu, Sep 05, 2024 at 07:39:31PM +0530, Anshuman Gupta wrote:
On 2024-09-05 at 11:55:23 +0300, Raag Jadav wrote:
On Thu, Sep 05, 2024 at 11:56:15AM +0530, Nilawar, Badal wrote:
On 28-08-2024 10:15, Raag Jadav wrote:
Add hwmon support for temp1_input attribute, which will expose package
temperature in millidegree Celsius. With this in place we can monitor
package temperature using lm-sensors tool.
$ sensors
i915-pci-0300
Adapter: PCI adapter
in0: 990.00 mV
fan1: 1260 RPM
temp1: +45.0°C
power1: N/A (max = 35.00 W)
energy1: 12.62 kJ
Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/11276
Signed-off-by: Raag Jadav <raag.jadav@xxxxxxxxx>
---
.../ABI/testing/sysfs-driver-intel-i915-hwmon | 8 ++++
drivers/gpu/drm/i915/i915_hwmon.c | 39 +++++++++++++++++++
drivers/gpu/drm/i915/intel_mchbar_regs.h | 4 ++
3 files changed, 51 insertions(+)
diff --git a/Documentation/ABI/testing/sysfs-driver-intel-i915-hwmon b/Documentation/ABI/testing/sysfs-driver-intel-i915-hwmon
index be4141a7522f..a885e5316d02 100644
--- a/Documentation/ABI/testing/sysfs-driver-intel-i915-hwmon
+++ b/Documentation/ABI/testing/sysfs-driver-intel-i915-hwmon
@@ -83,3 +83,11 @@ Contact: intel-gfx@xxxxxxxxxxxxxxxxxxxxx
Description: RO. Fan speed of device in RPM.
Only supported for particular Intel i915 graphics platforms.
+
+What: /sys/bus/pci/drivers/i915/.../hwmon/hwmon<i>/temp1_input
+Date: November 2024
+KernelVersion: 6.12
+Contact: intel-gfx@xxxxxxxxxxxxxxxxxxxxx
+Description: RO. GPU package temperature in millidegree Celsius.
+
+ Only supported for particular Intel i915 graphics platforms.
diff --git a/drivers/gpu/drm/i915/i915_hwmon.c b/drivers/gpu/drm/i915/i915_hwmon.c
index 17d30f6b84b0..9f1a2300510b 100644
--- a/drivers/gpu/drm/i915/i915_hwmon.c
+++ b/drivers/gpu/drm/i915/i915_hwmon.c
@@ -7,6 +7,7 @@
#include <linux/hwmon-sysfs.h>
#include <linux/jiffies.h>
#include <linux/types.h>
+#include <linux/units.h>
#include "i915_drv.h"
#include "i915_hwmon.h"
@@ -32,6 +33,7 @@
struct hwm_reg {
i915_reg_t gt_perf_status;
+ i915_reg_t pkg_temp;
i915_reg_t pkg_power_sku_unit;
i915_reg_t pkg_power_sku;
i915_reg_t pkg_rapl_limit;
@@ -280,6 +282,7 @@ static const struct attribute_group *hwm_groups[] = {
};
static const struct hwmon_channel_info * const hwm_info[] = {
+ HWMON_CHANNEL_INFO(temp, HWMON_T_INPUT),
HWMON_CHANNEL_INFO(in, HWMON_I_INPUT),
HWMON_CHANNEL_INFO(power, HWMON_P_MAX | HWMON_P_RATED_MAX | HWMON_P_CRIT),
HWMON_CHANNEL_INFO(energy, HWMON_E_INPUT),
@@ -310,6 +313,36 @@ static int hwm_pcode_write_i1(struct drm_i915_private *i915, u32 uval)
POWER_SETUP_SUBCOMMAND_WRITE_I1, 0, uval);
}
+static umode_t
+hwm_temp_is_visible(const struct hwm_drvdata *ddat, u32 attr)
+{
+ struct i915_hwmon *hwmon = ddat->hwmon;
+
+ if (attr == hwmon_temp_input && i915_mmio_reg_valid(hwmon->rg.pkg_temp))
+ return 0444;
+
+ return 0;
+}
+
+static int
+hwm_temp_read(struct hwm_drvdata *ddat, u32 attr, long *val)
+{
+ struct i915_hwmon *hwmon = ddat->hwmon;
+ intel_wakeref_t wakeref;
+ u32 reg_val;
+
+ if (attr == hwmon_temp_input) {
+ with_intel_runtime_pm(ddat->uncore->rpm, wakeref)
+ reg_val = intel_uncore_read(ddat->uncore, hwmon->rg.pkg_temp);
+
+ /* HW register value is in degrees, convert to millidegrees. */
+ *val = REG_FIELD_GET(TEMP_MASK, reg_val) * MILLIDEGREE_PER_DEGREE;
+ return 0;
+ }
+
+ return -EOPNOTSUPP;
+}
Let's try to have synergy between previous attribute, such as hwm_fan_input,
and this one.
This one's simple enough to be inline IMHO.
Besides, it's already in synergy with hwm_in_read() which has similar
implementation.
Agree this is pretty simple to have an any helper but IMO it would have been cleaner to have a switch
like hwm_in_read() to return -EOPNOTSUPP in default case. i think that was reason switch case was
used in entire file.
Extending on the simplicity argument above, if() makes more sense for a single case.
IMO lets prefer the style which was used in this entire file,
that is more readable along with other attributes.
Idea behind switch was scalable attribute for future.
It is something related to individual preference therefore
let's prefer the symmetry with other hwmon attributes.
I agree with this, but even if this approach is used file-wide, there were
concerns about using a switch case for a single case while implementing the
fan_input attribute.
https://patchwork.freedesktop.org/patch/607642/?series=136036&rev=4";
So I suggested to implement temp_input the way fan_input is implemented, at
least we should follow this approach to maintain symmetry with new
attributes. But in case if there is agreement to use file wide approach then
please follow that approach for fan_input as well.
Let's try to keep it simple and use whatever works for the case.
I'm sure the driver is much easier to maintain with a few less lines.
Let's not break symmetry, whichever approach is being followed please
maintain it file wide.
Regards,
Badal
Raag
