From: sashiko-bot@kernel.org
To: "Vinay Belgaumkar" <vinay.belgaumkar@intel.com>,
intel-xe@lists.freedesktop.org
Subject: Re: [PATCH v5 1/2] drm/xe: Disallow freq changes while mert cap is active
Date: Fri, 20 Nov 2026 02:29:36 +0000 [thread overview]
Message-ID: <sashiko-outbox-224364@kernel.org> (raw)
In-Reply-To: <20261120021606.1360244-3-vinay.belgaumkar@intel.com>
Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] drm/xe: xe_guc_pc_get_min/max_freq_locked() return freq instead
of 0
- [Medium] drm/xe: mert_freq_cap_active is not cleared on GuC PC stop or
restart
--
commit a5ffc89a51c9df2129a3327ea1a0076ff70c2a90
Author: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
drm/xe: Disallow freq changes while mert cap is active
This commit tracks when the Wa_22019338487 MERT frequency cap is active,
defers user min/max frequency changes until the cap is lifted, and lifts
the cap even if GSC firmware load fails.
> diff --git a/drivers/gpu/drm/xe/xe_guc_pc.c b/drivers/gpu/drm/xe/xe_guc_pc.c
> index 9558d7196a51a..7e9700f16bfb6 100644
> --- a/drivers/gpu/drm/xe/xe_guc_pc.c
> +++ b/drivers/gpu/drm/xe/xe_guc_pc.c
[ ... ]
> @@ -612,6 +613,10 @@ static int xe_guc_pc_get_min_freq_locked(struct xe_guc_pc *pc, u32 *freq)
> if (!pc->freq_ready)
> return -EAGAIN;
>
> + /* In the middle of a temporary freq cap, return stashed value */
> + if (XE_GT_WA(pc_to_gt(pc), 22019338487) && pc->mert_freq_cap_active)
> + return pc->stashed_min_freq;
> +
> ret = pc_action_query_task_state(pc);
> if (ret)
> return ret;
[ ... ]
> @@ -681,6 +695,10 @@ static int xe_guc_pc_get_max_freq_locked(struct xe_guc_pc *pc, u32 *freq)
> if (!pc->freq_ready)
> return -EAGAIN;
>
> + /* In the middle of a temporary freq cap, return stashed value */
> + if (XE_GT_WA(pc_to_gt(pc), 22019338487) && pc->mert_freq_cap_active)
> + return pc->stashed_max_freq;
> +
> ret = pc_action_query_task_state(pc);
> if (ret)
> return ret;
[Severity: High]
Should xe_guc_pc_get_min_freq_locked() and xe_guc_pc_get_max_freq_locked()
write the frequency into *freq and return 0 here instead of returning the
positive frequency value as the status code?
Callers such as min_freq_show() and max_freq_show() check if (ret) return
ret;, so returning a positive MHz value leaves *freq unassigned and causes
the sysfs show callbacks to return that positive integer as the byte count
without writing to buf. In addition, if a user sets a new min or max
frequency while the cap is active, it is saved in pc->user_requested_min /
pc->user_requested_max rather than pc->stashed_min_freq /
pc->stashed_max_freq, so *freq should probably be assigned
pc->user_requested_min ?: pc->stashed_min_freq (and similarly for max) to
match xe_guc_pc_restore_stashed_freq().
[ ... ]
> @@ -1094,6 +1121,9 @@ static int pc_set_mert_freq_cap(struct xe_guc_pc *pc)
> if (!ret)
> ret = pc_set_max_freq(pc, min(pc->rp0_freq, pc_max_freq_cap(pc)));
>
> + if (!ret)
> + pc->mert_freq_cap_active = true;
> +
> return ret;
> }
[Severity: Medium]
What happens if GuC PC is stopped and restarted (for example during a GT
reset) while pc->mert_freq_cap_active is still true?
Neither xe_guc_pc_stop() nor pc_set_mert_freq_cap() clears
pc->mert_freq_cap_active. When xe_guc_pc_start() calls
pc_set_mert_freq_cap() after pc_init_freqs(), its calls to
xe_guc_pc_get_min_freq_locked() and xe_guc_pc_get_max_freq_locked() will
hit the pc->mert_freq_cap_active early return instead of querying the
newly initialized SLPC state from GuC.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20261120021606.1360244-3-vinay.belgaumkar@intel.com?part=1
next prev parent reply other threads:[~2026-11-20 2:29 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-20 2:16 [PATCH v5 1/2] drm/xe: Disallow freq changes while mert cap is active Vinay Belgaumkar
2026-11-20 2:16 ` [PATCH v5 2/2] drm/xe: Add sysfs to disable RPe Vinay Belgaumkar
2026-11-20 2:35 ` sashiko-bot
2026-11-20 2:29 ` sashiko-bot [this message]
2026-11-20 4:54 ` ✓ CI.KUnit: success for series starting with [v5,1/2] drm/xe: Disallow freq changes while mert cap is active Patchwork
2026-11-20 5:29 ` ✗ Xe.CI.BAT: failure " Patchwork
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=sashiko-outbox-224364@kernel.org \
--to=sashiko-bot@kernel.org \
--cc=intel-xe@lists.freedesktop.org \
--cc=sashiko-reviews@lists.linux.dev \
--cc=vinay.belgaumkar@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.