Skip to content

drm/xe: Add support for GPU health indicator

Soham Purkait intel-xe 11 versions – Applied

Status from this message: Pushed to drm-xe-next

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v11 Latest1 patch 1 replyApplied
  2. v101 patch 1 replySuperseded
  3. v91 patch 1 replySuperseded
Earlier versions · 8
VersionPostedPatchesRepliesStatus
v81 patch3 repliesApplied
v71 patch7 repliesSuperseded
v61 patch1 replySuperseded
v51 patch3 repliesSuperseded
v41 patch5 repliesSuperseded
v33 patches7 repliesSuperseded
v22 patches10 repliesSuperseded
v12 patches11 repliesSuperseded

Patches in v11 Applied

#Subject and filesSizeRepliesApplied
1/1drm/xe/xe_ras: Add RAS GPU health indicatorDocumentation/ABI/testing/sysfs-driver-intel-xe-ras Documentation/gpu/xe/xe_device.rst drivers/gpu/drm/xe/xe_ras.c +2 more+235 -01 reply

Reviews trailers per patch

The patch has a Reviewed-by or Acked-by.

PatchRodrigo ViviRaag JadavAndi ShytiBadal Nilawar
1/1A-b A-b R-b R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes. Scroll the table sideways for all 4 reviewers.

Read the v11 thread

Cover letter [PATCH v11 0/1] drm/xe: Add support for GPU health indicator

GPUs commonly rely on various reactive health monitoring approaches.
The Xe GPU health indicator is intended to fit into such reactive monitoring flows, where it could be used by management tools to fetch and update GPU health status.

This patch adds Xe GPU health indicator support as a RAS feature. It introduces the health command IDs and request/response structures used by the System Controller mailbox, and integrates the feature into Xe through the gpu_health sysfs interface.

The sysfs file, gpu_health, is created at the device level and provides a simple interface for observing and updating the reported GPU health state. It is exposed as read-only for non admin users while write access is only provided to the admin users.

The sysfs file (gpu_health) is placed at the device level and behaves as follows:

  $ cat /sys/.../device/gpu_health
  ok
  $ echo critical > /sys/.../device/gpu_health
  $ cat /sys/.../device/gpu_health
  critical

Userspace consumer (Level Zero Sysman):
https://github.com/intel/compute-runtime/commit/3d974c6d9b65bede475c118e582761670ef110d2

Soham Purkait (1):
drm/xe/xe_ras: Add RAS GPU health indicator

 .../ABI/testing/sysfs-driver-intel-xe-ras     |  30 ++++
 Documentation/gpu/xe/xe_device.rst            |   7 +
 drivers/gpu/drm/xe/xe_ras.c                   | 153 ++++++++++++++++++
 drivers/gpu/drm/xe/xe_ras_types.h             |  41 +++++
 drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h |   4 +
 5 files changed, 235 insertions(+)
 create mode 100644 Documentation/ABI/testing/sysfs-driver-intel-xe-ras
Signature · 1 line
2.43.0