There is a need to inform user-space clients when a rebind worker
has ran out of memory so that it can react, adjust its working-set
and restart the job. This patch series aims to start a discussion
about the best way to accomplish this.
The series builds on the core "general notification mechanism" or
"watch_queue", and attaches a watch queue to each xe drm file.
The watch_queue is extremely flexible and allows filtering out
events of interest at the kernel level. There can be multiple
listeners.
Patch 1 Implements a restart IOCTL for rebind-workers
paused on OOM.
Patch 2 Adds fault-injection into the rebind worker for
testing.
Patch 3 Adds a DRM_XE_NOTIFY watch_type.
Patch 4 Implements watch_queue event sending from within
xe.
igt series:
Test-with: https://patchwork.freedesktop.org/series/168429/
Compute UMD side is not available yet. Will be available before
final review.
Thomas Hellström (4):
drm/xe: Add DRM_IOCTL_XE_VM_RESTART IOCTL
drm/xe: Add fault injection for rebind worker -ENOSPC
watch_queue: Add a DRM_XE_NOTIFY watch type and export init_watch()
drm/xe: Add watch_queue-based device event notification
MAINTAINERS | 1 +
drivers/gpu/drm/xe/Kconfig | 1 +
drivers/gpu/drm/xe/Makefile | 1 +
drivers/gpu/drm/xe/xe_debugfs.c | 4 +-
drivers/gpu/drm/xe/xe_device.c | 8 ++
drivers/gpu/drm/xe/xe_device_types.h | 6 ++
drivers/gpu/drm/xe/xe_vm.c | 135 ++++++++++++++++++++++++++-
drivers/gpu/drm/xe/xe_vm.h | 13 ++-
drivers/gpu/drm/xe/xe_vm_types.h | 3 +
drivers/gpu/drm/xe/xe_watch_queue.c | 111 ++++++++++++++++++++++
drivers/gpu/drm/xe/xe_watch_queue.h | 20 ++++
include/uapi/drm/xe_drm.h | 91 +++++++++++++++++-
include/uapi/drm/xe_drm_events.h | 62 ++++++++++++
include/uapi/linux/watch_queue.h | 3 +-
kernel/watch_queue.c | 13 ++-
15 files changed, 462 insertions(+), 10 deletions(-)
create mode 100644 drivers/gpu/drm/xe/xe_watch_queue.c
create mode 100644 drivers/gpu/drm/xe/xe_watch_queue.h
create mode 100644 include/uapi/drm/xe_drm_events.h
--
2.54.0
