From: Matthew Brost <matthew.brost@intel.com>
To: Francois Dugast <francois.dugast@intel.com>
Cc: intel-xe@lists.freedesktop.org,
"Thomas Hellström" <thomas.hellstrom@linux.intel.com>
Subject: Re: [PATCH] drm/xe/hw_engine_group: Add stats for mode switching
Date: Tue, 9 Dec 2025 14:28:07 -0800 [thread overview]
Message-ID: <aTiid7BPIZb/Bvqu@lstrano-desk.jf.intel.com> (raw)
In-Reply-To: <20251209164625.228826-1-francois.dugast@intel.com>
On Tue, Dec 09, 2025 at 05:46:25PM +0100, Francois Dugast wrote:
> The GT stats interface is extended to include counters of how many jobs
> are either interrupted or waited on in the hardware engine groups. This
> can help application debugging.
>
> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> Signed-off-by: Francois Dugast <francois.dugast@intel.com>
> ---
> drivers/gpu/drm/xe/xe_gt_stats.c | 4 ++++
> drivers/gpu/drm/xe/xe_gt_stats_types.h | 2 ++
> drivers/gpu/drm/xe/xe_hw_engine_group.c | 3 +++
> 3 files changed, 9 insertions(+)
>
> diff --git a/drivers/gpu/drm/xe/xe_gt_stats.c b/drivers/gpu/drm/xe/xe_gt_stats.c
> index 5f74706bab81..edb18582e0f4 100644
> --- a/drivers/gpu/drm/xe/xe_gt_stats.c
> +++ b/drivers/gpu/drm/xe/xe_gt_stats.c
> @@ -66,6 +66,10 @@ static const char *const stat_description[__XE_GT_STATS_NUM_IDS] = {
> DEF_STAT_STR(SVM_4K_BIND_US, "svm_4K_bind_us"),
> DEF_STAT_STR(SVM_64K_BIND_US, "svm_64K_bind_us"),
> DEF_STAT_STR(SVM_2M_BIND_US, "svm_2M_bind_us"),
> + DEF_STAT_STR(HW_ENGINE_GROUP_SUSPEND_LR_JOB_COUNT,
> + "hw_engine_group_suspend_lr_job_count"),
> + DEF_STAT_STR(HW_ENGINE_GROUP_WAIT_DMA_JOB_COUNT,
> + "hw_engine_group_wait_dma_job_count"),
This is will be helpful. One nit, this is really isn't a per-job
operation - it is per queue. Can you change the name to reflect that?
Matt
> };
>
> /**
> diff --git a/drivers/gpu/drm/xe/xe_gt_stats_types.h b/drivers/gpu/drm/xe/xe_gt_stats_types.h
> index d8348a8de2e1..826727eac485 100644
> --- a/drivers/gpu/drm/xe/xe_gt_stats_types.h
> +++ b/drivers/gpu/drm/xe/xe_gt_stats_types.h
> @@ -44,6 +44,8 @@ enum xe_gt_stats_id {
> XE_GT_STATS_ID_SVM_4K_BIND_US,
> XE_GT_STATS_ID_SVM_64K_BIND_US,
> XE_GT_STATS_ID_SVM_2M_BIND_US,
> + XE_GT_STATS_ID_HW_ENGINE_GROUP_SUSPEND_LR_JOB_COUNT,
> + XE_GT_STATS_ID_HW_ENGINE_GROUP_WAIT_DMA_JOB_COUNT,
> /* must be the last entry */
> __XE_GT_STATS_NUM_IDS,
> };
> diff --git a/drivers/gpu/drm/xe/xe_hw_engine_group.c b/drivers/gpu/drm/xe/xe_hw_engine_group.c
> index fa4db5f23342..47a9d1af7d81 100644
> --- a/drivers/gpu/drm/xe/xe_hw_engine_group.c
> +++ b/drivers/gpu/drm/xe/xe_hw_engine_group.c
> @@ -9,6 +9,7 @@
> #include "xe_device.h"
> #include "xe_exec_queue.h"
> #include "xe_gt.h"
> +#include "xe_gt_stats.h"
> #include "xe_hw_engine_group.h"
> #include "xe_vm.h"
>
> @@ -203,6 +204,7 @@ static int xe_hw_engine_group_suspend_faulting_lr_jobs(struct xe_hw_engine_group
> if (!xe_vm_in_fault_mode(q->vm))
> continue;
>
> + xe_gt_stats_incr(q->gt, XE_GT_STATS_ID_HW_ENGINE_GROUP_SUSPEND_LR_JOB_COUNT, 1);
> need_resume = true;
> q->ops->suspend(q);
> }
> @@ -244,6 +246,7 @@ static int xe_hw_engine_group_wait_for_dma_fence_jobs(struct xe_hw_engine_group
> if (xe_vm_in_lr_mode(q->vm))
> continue;
>
> + xe_gt_stats_incr(q->gt, XE_GT_STATS_ID_HW_ENGINE_GROUP_WAIT_DMA_JOB_COUNT, 1);
> fence = xe_exec_queue_last_fence_get_for_resume(q, q->vm);
> timeout = dma_fence_wait(fence, false);
> dma_fence_put(fence);
> --
> 2.43.0
>
next prev parent reply other threads:[~2025-12-09 22:28 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-12-09 16:46 [PATCH] drm/xe/hw_engine_group: Add stats for mode switching Francois Dugast
2025-12-09 22:28 ` Matthew Brost [this message]
2025-12-10 2:45 ` ✓ CI.KUnit: success for " Patchwork
2025-12-10 4:49 ` ✓ Xe.CI.BAT: " Patchwork
2025-12-10 7:22 ` ✓ Xe.CI.Full: " 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=aTiid7BPIZb/Bvqu@lstrano-desk.jf.intel.com \
--to=matthew.brost@intel.com \
--cc=francois.dugast@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=thomas.hellstrom@linux.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