Re: [PATCH V1] accel/amdxdna: Add per-process BO memory usage query support

[Ian Rogers]

On Tue, Mar 31, 2026 at 8:53 AM Lizhi Hou <lizhi.hou@xxxxxxx> wrote:
>
>
> On 3/30/26 19:30, Ian Rogers wrote:
> > On Wed, Mar 25, 2026 at 11:47 AM Lizhi Hou <lizhi.hou@xxxxxxx> wrote:
> >> Applied to drm-misc-next.
> >>
> >> On 3/24/26 10:01, Mario Limonciello wrote:
> >>>
> >>> On 3/24/26 11:31, Lizhi Hou wrote:
> >>>> From: Max Zhen <max.zhen@xxxxxxx>
> >>>>
> >>>> Add support for querying per-process buffer object (BO) memory
> >>>> usage through the amdxdna GET_ARRAY UAPI.
> >>>>
> >>>> Introduce a new query type, DRM_AMDXDNA_BO_USAGE, along with
> >>>> struct amdxdna_drm_bo_usage to report BO memory usage statistics,
> >>>> including heap, total, and internal usage.
> >>>>
> >>>> Track BO memory usage on a per-client basis by maintaining counters
> >>>> in GEM open/close and heap allocation/free paths. This ensures the
> >>>> reported statistics reflect the current memory footprint of each
> >>>> process.
> >>>>
> >>>> Wire the new query into the GET_ARRAY implementation to expose
> >>>> the usage information to userspace.
> >>>>
> >>>> Signed-off-by: Max Zhen <max.zhen@xxxxxxx>
> >>>> Reviewed-by: Lizhi Hou <lizhi.hou@xxxxxxx>
> >>>> Signed-off-by: Lizhi Hou <lizhi.hou@xxxxxxx>
> >>> I'm assuming you also have userspace side ready for this too right?
> >>> If you have a link handy can you please include it when committing.
> > Sorry for a naive question, would support in procfs be possible?
>
> Do you mean fdinfo? And yes, the fdinfo is in our upstream stack. I will
> post the patch later.

That's great! I did mean fdinfo. I'd be interested to see the patch,
and possibly the strings matched in perf need updating:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/drm_pmu.c?h=perf-tools-next#n183

Thanks,
Ian

> Lizhi
>
> > There's support in perf for displaying DRM usage stat data from there
> > [1], for example:
> > ```
> > $ perf list drm-
> >
> > List of pre-defined events (to be used in -e or -M):
> >
> >
> > drm:
> >   drm-active-stolen-system0
> >        [Total memory active in one or more engines. Unit: drm_i915]
> >   drm-active-system0
> >        [Total memory active in one or more engines. Unit: drm_i915]
> >   drm-engine-capacity-video
> >        [Engine capacity. Unit: drm_i915]
> >   drm-engine-copy
> >        [Utilization in ns. Unit: drm_i915]
> >   drm-engine-render
> >        [Utilization in ns. Unit: drm_i915]
> >   drm-engine-video
> >        [Utilization in ns. Unit: drm_i915]
> >   drm-engine-video-enhance
> >        [Utilization in ns. Unit: drm_i915]
> >   drm-purgeable-stolen-system0
> >        [Size of resident and purgeable memory buffers. Unit: drm_i915]
> >   drm-purgeable-system0
> >        [Size of resident and purgeable memory buffers. Unit: drm_i915]
> >   drm-resident-stolen-system0
> >        [Size of resident memory buffers. Unit: drm_i915]
> >   drm-resident-system0
> >        [Size of resident memory buffers. Unit: drm_i915]
> >   drm-shared-stolen-system0
> >        [Size of shared memory buffers. Unit: drm_i915]
> >   drm-shared-system0
> >        [Size of shared memory buffers. Unit: drm_i915]
> >   drm-total-stolen-system0
> >        [Size of shared and private memory. Unit: drm_i915]
> >   drm-total-system0
> >        [Size of shared and private memory. Unit: drm_i915]
> >
> >
> > $ perf stat -e drm-engine-render -a sleep 1
> >
> > Performance counter stats for 'system wide':
> >
> >    557,542,732,344 ns   drm-engine-render
> >
> >        1.001575975 seconds time elapsed
> > ```
> >
> > Thanks,
> > Ian
> >
> > [1] https://lore.kernel.org/lkml/20250403202439.57791-1-irogers@xxxxxxxxxx/