| Message ID | 20190614173335.30907-1-emil.l.velikov@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/amdgpu: extend AMDGPU_CTX_PRIORITY_NORMAL comment | expand |
Am 14.06.19 um 19:33 schrieb Emil Velikov: > From: Emil Velikov <emil.velikov@collabora.com> > > Currently the AMDGPU_CTX_PRIORITY_* defines are used in both > drm_amdgpu_ctx_in::priority and drm_amdgpu_sched_in::priority. > > Extend the comment to mention the CAP_SYS_NICE or DRM_MASTER requirement > is only applicable with the former. > > Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> > Cc: Christian König <christian.koenig@amd.com> > Cc: Alex Deucher <alexander.deucher@amd.com> > Signed-off-by: Emil Velikov <emil.velikov@collabora.com> > --- > Mildly curious: why didn't one extend ctx_amdgpu_ctx instead of adding > drm_amdgpu_sched? New flag + _u32 fd at the end (for the former) would > have been enough (and tweaking the ioctl permission thingy). The drm_amdgpu_sched is only allowed for DRM_MASTER. Christian. > > Speaking of flags, drm_amdgpu_sched_in lost its so extending it will > be "fun" > --- > include/uapi/drm/amdgpu_drm.h | 7 ++++++- > 1 file changed, 6 insertions(+), 1 deletion(-) > > diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h > index 4788730dbe78..dfb10fba2fe8 100644 > --- a/include/uapi/drm/amdgpu_drm.h > +++ b/include/uapi/drm/amdgpu_drm.h > @@ -219,7 +219,10 @@ union drm_amdgpu_bo_list { > #define AMDGPU_CTX_PRIORITY_VERY_LOW -1023 > #define AMDGPU_CTX_PRIORITY_LOW -512 > #define AMDGPU_CTX_PRIORITY_NORMAL 0 > -/* Selecting a priority above NORMAL requires CAP_SYS_NICE or DRM_MASTER */ > +/* > + * When used in struct drm_amdgpu_ctx_in, a priority above NORMAL requires > + * CAP_SYS_NICE or DRM_MASTER > +*/ > #define AMDGPU_CTX_PRIORITY_HIGH 512 > #define AMDGPU_CTX_PRIORITY_VERY_HIGH 1023 > > @@ -229,6 +232,7 @@ struct drm_amdgpu_ctx_in { > /** For future use, no flags defined so far */ > __u32 flags; > __u32 ctx_id; > + /** AMDGPU_CTX_PRIORITY_* */ > __s32 priority; > }; > > @@ -281,6 +285,7 @@ struct drm_amdgpu_sched_in { > /* AMDGPU_SCHED_OP_* */ > __u32 op; > __u32 fd; > + /** AMDGPU_CTX_PRIORITY_* */ > __s32 priority; > __u32 ctx_id; > };
On Fri, 14 Jun 2019 at 19:02, Koenig, Christian <Christian.Koenig@amd.com> wrote: > > Am 14.06.19 um 19:33 schrieb Emil Velikov: > > From: Emil Velikov <emil.velikov@collabora.com> > > > > Currently the AMDGPU_CTX_PRIORITY_* defines are used in both > > drm_amdgpu_ctx_in::priority and drm_amdgpu_sched_in::priority. > > > > Extend the comment to mention the CAP_SYS_NICE or DRM_MASTER requirement > > is only applicable with the former. > > > > Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> > > Cc: Christian König <christian.koenig@amd.com> > > Cc: Alex Deucher <alexander.deucher@amd.com> > > Signed-off-by: Emil Velikov <emil.velikov@collabora.com> > > --- > > Mildly curious: why didn't one extend ctx_amdgpu_ctx instead of adding > > drm_amdgpu_sched? New flag + _u32 fd at the end (for the former) would > > have been enough (and tweaking the ioctl permission thingy). > > The drm_amdgpu_sched is only allowed for DRM_MASTER. > Fair enough. Is the patch wrong or did it slip through the cracks? I cannot see it in Alex's tree. -Emil
Am 02.07.19 um 19:15 schrieb Emil Velikov: > On Fri, 14 Jun 2019 at 19:02, Koenig, Christian > <Christian.Koenig@amd.com> wrote: >> Am 14.06.19 um 19:33 schrieb Emil Velikov: >>> From: Emil Velikov <emil.velikov@collabora.com> >>> >>> Currently the AMDGPU_CTX_PRIORITY_* defines are used in both >>> drm_amdgpu_ctx_in::priority and drm_amdgpu_sched_in::priority. >>> >>> Extend the comment to mention the CAP_SYS_NICE or DRM_MASTER requirement >>> is only applicable with the former. >>> >>> Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> >>> Cc: Christian König <christian.koenig@amd.com> >>> Cc: Alex Deucher <alexander.deucher@amd.com> >>> Signed-off-by: Emil Velikov <emil.velikov@collabora.com> >>> --- >>> Mildly curious: why didn't one extend ctx_amdgpu_ctx instead of adding >>> drm_amdgpu_sched? New flag + _u32 fd at the end (for the former) would >>> have been enough (and tweaking the ioctl permission thingy). >> The drm_amdgpu_sched is only allowed for DRM_MASTER. >> > Fair enough. > > Is the patch wrong or did it slip through the cracks? I cannot see it > in Alex's tree. Looks like Alex just missed this one and I was on vacation/out of office for a while. I've gone ahead added my rb and just pushed it. Thanks, Christian. > > -Emil > _______________________________________________ > amd-gfx mailing list > amd-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/amd-gfx
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index 4788730dbe78..dfb10fba2fe8 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -219,7 +219,10 @@ union drm_amdgpu_bo_list { #define AMDGPU_CTX_PRIORITY_VERY_LOW -1023 #define AMDGPU_CTX_PRIORITY_LOW -512 #define AMDGPU_CTX_PRIORITY_NORMAL 0 -/* Selecting a priority above NORMAL requires CAP_SYS_NICE or DRM_MASTER */ +/* + * When used in struct drm_amdgpu_ctx_in, a priority above NORMAL requires + * CAP_SYS_NICE or DRM_MASTER +*/ #define AMDGPU_CTX_PRIORITY_HIGH 512 #define AMDGPU_CTX_PRIORITY_VERY_HIGH 1023 @@ -229,6 +232,7 @@ struct drm_amdgpu_ctx_in { /** For future use, no flags defined so far */ __u32 flags; __u32 ctx_id; + /** AMDGPU_CTX_PRIORITY_* */ __s32 priority; }; @@ -281,6 +285,7 @@ struct drm_amdgpu_sched_in { /* AMDGPU_SCHED_OP_* */ __u32 op; __u32 fd; + /** AMDGPU_CTX_PRIORITY_* */ __s32 priority; __u32 ctx_id; };