From: "Poosa, Karthik" <karthik.poosa@intel.com>
To: "Purkait, Soham" <soham.purkait@intel.com>,
<intel-xe@lists.freedesktop.org>
Cc: <rodrigo.vivi@intel.com>, <anshuman.gupta@intel.com>,
<badal.nilawar@intel.com>, <raag.jadav@intel.com>,
<riana.tauro@intel.com>, <sk.anirban@intel.com>,
<mallesh.koujalagi@intel.com>,
<ravi.kishore.koppuravuri@intel.com>
Subject: Re: [PATCH v3 4/9] drm/xe/hwmon: expose pwm[1-3]
Date: Thu, 5 Nov 2026 22:39:27 +0530 [thread overview]
Message-ID: <146734d9-e23c-4f0a-bf15-54b97b9f76b0@intel.com> (raw)
In-Reply-To: <f0cb6786-c5af-41d7-ba7a-463eefeb42cc@intel.com>
On 29-10-2026 17:01, Purkait, Soham wrote:
>
> Hi Karthik,
>
> On 16-10-2026 00:37, Karthik Poosa wrote:
>> Expose pwm[1-3] fan duty attributes through hwmon.
>>
>> This enables reading and writing user PWM values (0..255) for each
>> available fan channel.
>>
>> Update Xe hwmon ABI documentation for pwm[1-3].
>>
>> This sysfs node can be used for manual control of fan speed,
>> irrespective of fan curve.
>>
>> v2:
>> - Use xe helpers for dmesg logs.
>> - Avoid using hwmon->num_fans to get fan count in
>> xe_hwmon_pwm_is_visible().
>>
>> v3:
>> - Set PWM directly through the user table; avoid initializing it from the
>> stock table. (Soham)
>> - Move the unused xe_hwmon_activate_user_fan_table() to the subsequent
>> patch.
>> - Fail pwmN reads when no value has been written, rather than activating
>> the user table from the stock table.
>> - Define MIN_FAN_CONTROL_POINTS as 2.
>> - Use MIN_FAN_CONTROL_POINTS for PWM configuration. (Soham)
>> - Use batched writes for configuring fan control points when setting pwmN.
>> - Add and rephrase couple debug logs.
>> - Update the hwmon documentation.
>>
>> Signed-off-by: Karthik Poosa<karthik.poosa@intel.com>
>> Assisted-by: LLM
>> ---
>> .../ABI/testing/sysfs-driver-intel-xe-hwmon | 9 +
>> drivers/gpu/drm/xe/xe_hwmon.c | 170 ++++++++++++++++++
>> drivers/gpu/drm/xe/xe_pcode_api.h | 3 +
>> 3 files changed, 182 insertions(+)
>>
>> diff --git a/Documentation/ABI/testing/sysfs-driver-intel-xe-hwmon b/Documentation/ABI/testing/sysfs-driver-intel-xe-hwmon
>> index 5ee68aefa485..ce17e9ca3ed8 100644
>> --- a/Documentation/ABI/testing/sysfs-driver-intel-xe-hwmon
>> +++ b/Documentation/ABI/testing/sysfs-driver-intel-xe-hwmon
>> @@ -321,3 +321,12 @@ Description: RW. Package burst power limit interval (Tau in PL2/Tau) in
>> milliseconds over which sustained power is averaged.
>>
>> Only supported for particular Intel Xe graphics platforms.
>> +
>> +What: /sys/bus/pci/drivers/xe/.../hwmon/hwmon<i>/pwm[1-3]
>> +Date: October 2026
>> +KernelVersion: 7.3
>> +Contact:intel-xe@lists.freedesktop.org
>> +Description: RW. Target fan PWM duty value in the range 0..255. 0 is zero
>> + fan speed and 255 is 100% fan speed.
>> +
>> + Only supported for particular Intel Xe graphics platforms.
> Use tab indentation after the field labels.
>> diff --git a/drivers/gpu/drm/xe/xe_hwmon.c b/drivers/gpu/drm/xe/xe_hwmon.c
>> index 32016c5d4770..a86ed741feb3 100644
>> --- a/drivers/gpu/drm/xe/xe_hwmon.c
>> +++ b/drivers/gpu/drm/xe/xe_hwmon.c
>> @@ -8,6 +8,7 @@
>> #include <linux/jiffies.h>
>> #include <linux/types.h>
>> #include <linux/units.h>
>> +#include <linux/delay.h>
> Is this header required ?
we can remove this, i think this was part of some temporary debug code.
Not required. This was left over from temporary debug code and can be
removed.
>> #include <drm/drm_managed.h>
>> #include "regs/xe_gt_regs.h"
>> @@ -153,6 +154,9 @@ enum fan_table_type {
>> /* Maximum number of fan control points supported by each fan. */
>> #define MAX_FAN_CONTROL_POINTS (10)
>>
>> +/* Minimum number of points required for a user fan table. */
>> +#define MIN_FAN_CONTROL_POINTS (2)
> Use tab indentation.
ok
>> +
>> /* Fan control point index: bit31 selects fan table type; bits30:0 control point index. */
>> #define FCP_INDEX(FAN_TABLE_TYPE, POINT_NUM) (((FAN_TABLE_TYPE) << 31) | \
>> ((POINT_NUM) & REG_GENMASK(30, 0)))
>> @@ -913,6 +917,7 @@ static const struct hwmon_channel_info * const hwmon_info[] = {
>> HWMON_CHANNEL_INFO(energy, HWMON_E_INPUT | HWMON_E_LABEL, HWMON_E_INPUT | HWMON_E_LABEL),
>> HWMON_CHANNEL_INFO(fan, HWMON_F_INPUT | HWMON_F_MAX, HWMON_F_INPUT | HWMON_F_MAX,
>> HWMON_F_INPUT | HWMON_F_MAX),
>> + HWMON_CHANNEL_INFO(pwm, HWMON_PWM_INPUT, HWMON_PWM_INPUT, HWMON_PWM_INPUT),
>> NULL
>> };
>>
>> @@ -1154,6 +1159,89 @@ static int xe_hwmon_get_fan_enable_mask(struct xe_hwmon *hwmon)
>> return 0;
>> }
>>
>> +static int xe_hwmon_get_fan_point_count(struct xe_hwmon *hwmon, u8 fan, u32 *point_count,
>> + enum fan_table_type table_type)
>> +{
>> + int ret;
>> +
>> + ret = xe_hwmon_pcode_read_fan_control(hwmon,
>> + (table_type == USER_FAN_TABLE) ?
>> + FSC_READ_USER_FAN_CONTROL_POINTS :
>> + FSC_READ_STOCK_FAN_CONTROL_POINTS,
>> + fan, point_count);
>> + if (ret) {
>> + xe_err(hwmon->xe, "failed to read fan %d %s point count, ret=%d\n", fan,
>> + (table_type == USER_FAN_TABLE) ? "user" : "stock", ret);
>> + return ret;
>> + }
>> +
>> + if (*point_count > MAX_FAN_CONTROL_POINTS) {
>> + xe_err(hwmon->xe, "fan %d %s point count %u exceeds max %u\n", fan,
>> + (table_type == USER_FAN_TABLE) ? "user" : "stock", *point_count,
>> + MAX_FAN_CONTROL_POINTS);
>> + return -EINVAL;
>> + }
>> +
>> + return 0;
>> +}
>> +
>> +/* Temperature to set when setting the fan PWM with pwmN. */
>> +#define FIXED_PWM_TEMP (1)
>> +
>> +static int xe_hwmon_set_user_fan_pwm(struct xe_hwmon *hwmon, u8 fan, u8 pwm)
>> +{
>> + struct xe_hwmon_fan_info *fi = &hwmon->fi[fan];
>> + struct xe_pcode_write writes[MIN_FAN_CONTROL_POINTS + 1];
>> + u32 point_count = 0;
>> + u8 clamped_pwm;
>> + int ret, i;
>> +
>> + pwm = DIV_ROUND_CLOSEST(pwm * 100, U8_MAX);
>> + clamped_pwm = max_t(u8, pwm, min_t(u32, fi->min_pwm, U8_MAX));
> Is clamped_pwm and fi->min_pwm in percentage ? If so use 100 *instead
> of* U8_MAX to be the max value.
you are right, will change it.
>> +
>> + /*
>> + * Set the first fan control point in the user table to the clamped PWM value.
>> + * Use only the first fan control point for setting the PWM value.
> It says "only the first fan control point" but the loop end up writing
> 2 points.
will rephrase this, fan control needs atleast 2 fan control points
>> + */
>> +
>> + for (i = 0; i < MIN_FAN_CONTROL_POINTS; i++) {
>> + writes[i].mbox = PCODE_MBOX(FAN_SPEED_CONTROL, FSC_WRITE_FAN_TABLE, fan);
>> + writes[i].data = FIELD_PREP(FAN_CONTROL_POINT_TEMP_MASK, FIXED_PWM_TEMP + (i)) |
>> + FIELD_PREP(FAN_CONTROL_POINT_SPEED_MASK, clamped_pwm);
>> + xe_dbg(hwmon->xe, "prepared fan control point %d with temp=%d, speed=%u\n", i,
>> + FIXED_PWM_TEMP + (i), clamped_pwm);
>> + }
>> +
>> + writes[i].mbox = PCODE_MBOX(FAN_SPEED_CONTROL, FSC_WRITE_NUM_FAN_CONTROL_POINTS, fan);
>> + writes[i].data = MIN_FAN_CONTROL_POINTS;
>> + ret = xe_pcode_write_batch(xe_device_get_root_tile(hwmon->xe), writes, ARRAY_SIZE(writes));
>> + if (ret) {
>> + xe_err(hwmon->xe, "failed to write pwm %d for fan %d, ret=%d\n", pwm, fan, ret);
> "pwm" has been reassigned to a percentage by this point, so either
> print the original 0–255 value (keep it in a separate variable) or
> label this one as a percentage.
ok
>> + return ret;
>> + }
>> + xe_dbg(hwmon->xe, "applied fixed pwm %u for fan %d\n", clamped_pwm, fan);
>> +
>> + /* Verify if the fan points are set */
>> + ret = xe_hwmon_get_fan_point_count(hwmon, fan, &point_count, USER_FAN_TABLE);
>> + if (ret && point_count == 0) {
> On failure point_count is already 0, so && point_count == 0 seems to
> be redundant.
> Instead could this be more usable :
> if (ret || point_count != MIN_FAN_CONTROL_POINTS), if the valid write
> needed to be checked ?
ok
>
>> + xe_err(hwmon->xe, "pwm set verify failed for fan %d, ret=%d, point count=%u\n", fan,
>> + ret, point_count);
>> + return ret;
>> + }
>> +
>> + /* Update the user fan table cache. */
>> + for (i = 0; i < MIN_FAN_CONTROL_POINTS; i++) {
>> + fi->fan_table[USER_FAN_TABLE].fcp[i].temp = FIXED_PWM_TEMP + (i);
> Drop the parentheses around i: FIXED_PWM_TEMP + i.
ok
>> + fi->fan_table[USER_FAN_TABLE].fcp[i].speed = clamped_pwm;
>> + xe_dbg(hwmon->xe, "fan %d user table cache point %d with temp=%d, speed=%u\n", fan,
>> + i, fi->fan_table[USER_FAN_TABLE].fcp[i].temp,
>> + fi->fan_table[USER_FAN_TABLE].fcp[i].speed);
> The arguments don't align under the opening (.
ok
>> + }
>> + fi->fan_table[USER_FAN_TABLE].point_count = MIN_FAN_CONTROL_POINTS;
>> +
>> + return 0;
>> +}
>> +
>> static int xe_hwmon_read_fan_control_info(struct xe_hwmon *hwmon)
>> {
>> struct xe_tile *root_tile = xe_device_get_root_tile(hwmon->xe);
>> @@ -1774,6 +1862,20 @@ xe_hwmon_fan_is_visible(struct xe_hwmon *hwmon, u32 attr, int channel)
>> }
>> }
>>
>> +static umode_t
>> +xe_hwmon_pwm_is_visible(struct xe_hwmon *hwmon, u32 attr, int channel)
>> +{
>> + if (!hwmon->xe->info.has_fan_control || !IS_FAN_ENABLED(channel))
>> + return 0;
>> +
>> + switch (attr) {
>> + case hwmon_pwm_input:
>> + return 0644;
>> + default:
>> + return 0;
>> + }
>> +}
>> +
>> static int
>> xe_hwmon_fan_input_read(struct xe_hwmon *hwmon, int channel, long *val)
>> {
>> @@ -1828,6 +1930,67 @@ xe_hwmon_fan_read(struct xe_hwmon *hwmon, u32 attr, int channel, long *val)
>> }
>> }
>>
>> +static int
>> +xe_hwmon_pwm_read(struct xe_hwmon *hwmon, u32 attr, int channel, long *val)
>> +{
>> + struct xe_hwmon_fan_info *fi;
>> + int ret = 0;
>> +
>> + if (channel < 0 || channel >= FAN_MAX)
> Add a short xe_dbg/xe_warn before these -EINVAL returns for clarity.
this can be removed, as pwm nodes will be created only for available
fans, so there cannot be channel outside the range.
2699 static umode_t
2700 xe_hwmon_pwm_is_visible(struct xe_hwmon *hwmon, u32 attr, int channel)
2701 {
2702 if (!hwmon->xe->info.has_fan_control ||
*!IS_FAN_ENABLED(channel))*
2703 return 0;
>> + return -EINVAL;
>> +
>> + mutex_lock(&hwmon->hwmon_lock);
>> +
>> + fi = &hwmon->fi[channel];
>> +
>> + switch (attr) {
>> + case hwmon_pwm_input:
>> + /* Check if user fan table is set else activate it.*/
> Where does it activate ? Use space before */.
Activation code was there previous version, now we read it from cache
else return error.
I shall remove this comment.
>> + if (!fi->fan_table[USER_FAN_TABLE].point_count) {
>> + xe_warn(hwmon->xe, "pwm not set for fan %d\n", channel);
>> + ret = -ENXIO;
>> + goto unlock;
>> + }
>> + *val = DIV_ROUND_CLOSEST(fi->fan_table[USER_FAN_TABLE].fcp[0].speed * U8_MAX, 100);
> The 0–255 <-> percentage round-trip is lossy for small values; Better
> to add a one-line note since the ABI says a full 0–255 range.
ok.
>> + xe_dbg(hwmon->xe, "pwm read for fan %d: %ld\n", channel, *val);
>> + break;
>> + default:
>> + ret = -EOPNOTSUPP;
>> + break;
>> + }
>> +
>> +unlock:
>> + mutex_unlock(&hwmon->hwmon_lock);
>> + return ret;
>> +}
>> +
>> +static int
>> +xe_hwmon_pwm_write(struct xe_hwmon *hwmon, u32 attr, int channel, long val)
>> +{
>> + int ret = 0;
>> +
>> + if (channel < 0 || channel >= FAN_MAX)
>
> Add a short xe_dbg/xe_warn before these -EINVAL returns for clarity.
>
>
> Regards,
> Soham
same as in xe_hwmon_pwm_read.
>> + return -EINVAL;
>> +
>> + mutex_lock(&hwmon->hwmon_lock);
>> +
>> + switch (attr) {
>> + case hwmon_pwm_input:
>> + if (val < 0 || val > U8_MAX) {
>> + ret = -EINVAL;
>> + break;
>> + }
>> + ret = xe_hwmon_set_user_fan_pwm(hwmon, channel, (u8)val);
>> + break;
>> + default:
>> + ret = -EOPNOTSUPP;
>> + break;
>> + }
>> + mutex_unlock(&hwmon->hwmon_lock);
>> +
>> + return ret;
>> +}
>> +
>> static umode_t
>> xe_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_types type,
>> u32 attr, int channel)
>> @@ -1854,6 +2017,9 @@ xe_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_types type,
>> case hwmon_fan:
>> ret = xe_hwmon_fan_is_visible(hwmon, attr, channel);
>> break;
>> + case hwmon_pwm:
>> + ret = xe_hwmon_pwm_is_visible(hwmon, attr, channel);
>> + break;
>> default:
>> ret = 0;
>> break;
>> @@ -1883,6 +2049,8 @@ xe_hwmon_read(struct device *dev, enum hwmon_sensor_types type, u32 attr,
>> return xe_hwmon_energy_read(hwmon, attr, channel, val);
>> case hwmon_fan:
>> return xe_hwmon_fan_read(hwmon, attr, channel, val);
>> + case hwmon_pwm:
>> + return xe_hwmon_pwm_read(hwmon, attr, channel, val);
>> default:
>> return -EOPNOTSUPP;
>> }
>> @@ -1901,6 +2069,8 @@ xe_hwmon_write(struct device *dev, enum hwmon_sensor_types type, u32 attr,
>> return xe_hwmon_power_write(hwmon, attr, channel, val);
>> case hwmon_curr:
>> return xe_hwmon_curr_write(hwmon, attr, channel, val);
>> + case hwmon_pwm:
>> + return xe_hwmon_pwm_write(hwmon, attr, channel, val);
>> default:
>> return -EOPNOTSUPP;
>> }
>> diff --git a/drivers/gpu/drm/xe/xe_pcode_api.h b/drivers/gpu/drm/xe/xe_pcode_api.h
>> index 32bac74c4281..80642bee0ac8 100644
>> --- a/drivers/gpu/drm/xe/xe_pcode_api.h
>> +++ b/drivers/gpu/drm/xe/xe_pcode_api.h
>> @@ -82,9 +82,12 @@
>> #define PCODE_MBOX_DOMAIN_HBM 0x2
>>
>> #define FAN_SPEED_CONTROL 0x7D
>> +#define FSC_WRITE_NUM_FAN_CONTROL_POINTS 0x0
>> +#define FSC_WRITE_FAN_TABLE 0x1
>> #define FSC_READ_MAX_FAN_RPS 0x3
>> #define FSC_READ_NUM_FANS 0x4
>> #define FSC_READ_STOCK_FAN_CONTROL_POINTS 0x5
>> +#define FSC_READ_USER_FAN_CONTROL_POINTS 0x6
>> #define FSC_READ_FAN_TABLE 0x7
>> #define FAN_CONTROL_POINT_TEMP_MASK REG_GENMASK(7, 0)
>> #define FAN_CONTROL_POINT_SPEED_MASK REG_GENMASK(15, 8)
next prev parent reply other threads:[~2026-11-05 17:10 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-15 19:07 [PATCH v3 0/9] Add fan control support Karthik Poosa
2026-10-15 19:07 ` [PATCH v3 1/9] drm/xe/hwmon: initialize fan-control backend and table cache Karthik Poosa
2026-10-28 18:46 ` Purkait, Soham
2026-10-29 11:34 ` Poosa, Karthik
2026-10-29 12:02 ` Purkait, Soham
2026-10-29 12:04 ` Purkait, Soham
2026-11-05 5:58 ` Purkait, Soham
2026-11-05 16:56 ` Poosa, Karthik
2026-10-15 19:07 ` [PATCH v3 2/9] drm/xe/hwmon: expose fanN_max Karthik Poosa
2026-10-29 4:35 ` Purkait, Soham
2026-10-29 11:39 ` Poosa, Karthik
2026-10-15 19:07 ` [PATCH v3 3/9] drm/xe/pcode: add ordered write batch helper Karthik Poosa
2026-10-29 5:45 ` Purkait, Soham
2026-10-29 11:40 ` Poosa, Karthik
2026-10-15 19:07 ` [PATCH v3 4/9] drm/xe/hwmon: expose pwm[1-3] Karthik Poosa
2026-10-29 11:31 ` Purkait, Soham
2026-11-05 17:09 ` Poosa, Karthik [this message]
2026-10-15 19:07 ` [PATCH v3 5/9] drm/xe/hwmon: expose pwm[1-3]_enable Karthik Poosa
2026-10-29 18:33 ` Purkait, Soham
2026-11-05 17:41 ` Poosa, Karthik
2026-10-15 19:07 ` [PATCH v3 6/9] drm/xe/hwmon: Enable fan curve control support Karthik Poosa
2026-10-30 6:33 ` Purkait, Soham
2026-10-15 19:07 ` [PATCH v3 7/9] drm/xe/hwmon: Add kernel-doc for fan control Karthik Poosa
2026-10-30 13:42 ` Purkait, Soham
2026-10-15 19:07 ` [PATCH v3 8/9] drm/xe/hwmon: preserve fan user table across suspend resume Karthik Poosa
2026-11-02 5:58 ` Purkait, Soham
2026-10-15 19:07 ` [PATCH v3 9/9] drm/xe/hwmon: Update fan info after late binding Karthik Poosa
2026-11-02 6:26 ` Purkait, Soham
2026-10-15 19:46 ` ✗ CI.checkpatch: warning for Add fan control support Patchwork
2026-10-15 19:48 ` ✓ CI.KUnit: success " Patchwork
2026-10-15 21:01 ` ✗ Xe.CI.BAT: failure " Patchwork
2026-10-16 7:56 ` ✗ Xe.CI.FULL: " Patchwork
-- strict thread matches above, loose matches on Subject: below --
2026-10-15 19:01 [PATCH v3 0/9] " Karthik Poosa
2026-10-15 19:01 ` [PATCH v3 4/9] drm/xe/hwmon: expose pwm[1-3] Karthik Poosa
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=146734d9-e23c-4f0a-bf15-54b97b9f76b0@intel.com \
--to=karthik.poosa@intel.com \
--cc=anshuman.gupta@intel.com \
--cc=badal.nilawar@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=mallesh.koujalagi@intel.com \
--cc=raag.jadav@intel.com \
--cc=ravi.kishore.koppuravuri@intel.com \
--cc=riana.tauro@intel.com \
--cc=rodrigo.vivi@intel.com \
--cc=sk.anirban@intel.com \
--cc=soham.purkait@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox