Message ID | 20210609172902.1937-1-christian.koenig@amd.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | drm/nouveau: init the base GEM fields for internal BOs | expand |
On 6/9/21 8:29 PM, Christian König wrote: > TTMs buffer objects are based on GEM objects for quite a while > and rely on initializing those fields before initializing the TTM BO. > > Noveau now doesn't init the GEM object for internally allocated BOs, Nouveau > so make sure that we at least initialize some necessary fields. > > Signed-off-by: Christian König <christian.koenig@amd.com> > --- > drivers/gpu/drm/nouveau/nouveau_bo.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c > index 520b1ea9d16c..085023624fb0 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_bo.c > +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c > @@ -149,6 +149,8 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) > */ > if (bo->base.dev) > drm_gem_object_release(&bo->base); > + else > + dma_resv_fini(&bo->base._resv); > > kfree(nvbo); > } > @@ -330,6 +332,10 @@ nouveau_bo_new(struct nouveau_cli *cli, u64 size, int align, > if (IS_ERR(nvbo)) > return PTR_ERR(nvbo); > > + nvbo->bo.base.size = size; > + dma_resv_init(&nvbo->bo.base._resv); > + drm_vma_node_reset(&nvbo->bo.base.vma_node); > + > ret = nouveau_bo_init(nvbo, size, align, domain, sg, robj); > if (ret) > return ret; > That works, thanks for the fix! Tested-by: Mikko Perttunen <mperttunen@nvidia.com> Mikko
Am 09.06.21 um 19:45 schrieb Mikko Perttunen: > On 6/9/21 8:29 PM, Christian König wrote: >> TTMs buffer objects are based on GEM objects for quite a while >> and rely on initializing those fields before initializing the TTM BO. >> >> Noveau now doesn't init the GEM object for internally allocated BOs, > > Nouveau > >> so make sure that we at least initialize some necessary fields. >> >> Signed-off-by: Christian König <christian.koenig@amd.com> >> --- >> drivers/gpu/drm/nouveau/nouveau_bo.c | 6 ++++++ >> 1 file changed, 6 insertions(+) >> >> diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c >> b/drivers/gpu/drm/nouveau/nouveau_bo.c >> index 520b1ea9d16c..085023624fb0 100644 >> --- a/drivers/gpu/drm/nouveau/nouveau_bo.c >> +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c >> @@ -149,6 +149,8 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) >> */ >> if (bo->base.dev) >> drm_gem_object_release(&bo->base); >> + else >> + dma_resv_fini(&bo->base._resv); >> kfree(nvbo); >> } >> @@ -330,6 +332,10 @@ nouveau_bo_new(struct nouveau_cli *cli, u64 >> size, int align, >> if (IS_ERR(nvbo)) >> return PTR_ERR(nvbo); >> + nvbo->bo.base.size = size; >> + dma_resv_init(&nvbo->bo.base._resv); >> + drm_vma_node_reset(&nvbo->bo.base.vma_node); >> + >> ret = nouveau_bo_init(nvbo, size, align, domain, sg, robj); >> if (ret) >> return ret; >> > > That works, thanks for the fix! > > Tested-by: Mikko Perttunen <mperttunen@nvidia.com> Thanks. Can anybody give me an rb that I can push this to drm-misc-next before the weekend? Regards, Christian. > > Mikko
On Wed, 9 Jun 2021 at 18:29, Christian König <ckoenig.leichtzumerken@gmail.com> wrote: > > TTMs buffer objects are based on GEM objects for quite a while > and rely on initializing those fields before initializing the TTM BO. > > Noveau now doesn't init the GEM object for internally allocated BOs, > so make sure that we at least initialize some necessary fields. > > Signed-off-by: Christian König <christian.koenig@amd.com> fwiw, Reviewed-by: Matthew Auld <matthew.auld@intel.com> > --- > drivers/gpu/drm/nouveau/nouveau_bo.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c > index 520b1ea9d16c..085023624fb0 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_bo.c > +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c > @@ -149,6 +149,8 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) > */ > if (bo->base.dev) > drm_gem_object_release(&bo->base); > + else > + dma_resv_fini(&bo->base._resv); > > kfree(nvbo); > } > @@ -330,6 +332,10 @@ nouveau_bo_new(struct nouveau_cli *cli, u64 size, int align, > if (IS_ERR(nvbo)) > return PTR_ERR(nvbo); > > + nvbo->bo.base.size = size; > + dma_resv_init(&nvbo->bo.base._resv); > + drm_vma_node_reset(&nvbo->bo.base.vma_node); > + > ret = nouveau_bo_init(nvbo, size, align, domain, sg, robj); > if (ret) > return ret; > -- > 2.25.1 >
On Thu, Jun 10, 2021 at 04:08:40PM +0800, Koenig, Christian wrote: > > > Am 09.06.21 um 19:45 schrieb Mikko Perttunen: > > On 6/9/21 8:29 PM, Christian König wrote: > >> TTMs buffer objects are based on GEM objects for quite a while > >> and rely on initializing those fields before initializing the TTM BO. > >> > >> Noveau now doesn't init the GEM object for internally allocated BOs, > > > > Nouveau > > > >> so make sure that we at least initialize some necessary fields. > >> > >> Signed-off-by: Christian König <christian.koenig@amd.com> > >> --- > >> drivers/gpu/drm/nouveau/nouveau_bo.c | 6 ++++++ > >> 1 file changed, 6 insertions(+) > >> > >> diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c > >> b/drivers/gpu/drm/nouveau/nouveau_bo.c > >> index 520b1ea9d16c..085023624fb0 100644 > >> --- a/drivers/gpu/drm/nouveau/nouveau_bo.c > >> +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c > >> @@ -149,6 +149,8 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) > >> */ > >> if (bo->base.dev) > >> drm_gem_object_release(&bo->base); > >> + else > >> + dma_resv_fini(&bo->base._resv); > >> kfree(nvbo); > >> } > >> @@ -330,6 +332,10 @@ nouveau_bo_new(struct nouveau_cli *cli, u64 > >> size, int align, > >> if (IS_ERR(nvbo)) > >> return PTR_ERR(nvbo); > >> + nvbo->bo.base.size = size; > >> + dma_resv_init(&nvbo->bo.base._resv); > >> + drm_vma_node_reset(&nvbo->bo.base.vma_node); > >> + > >> ret = nouveau_bo_init(nvbo, size, align, domain, sg, robj); > >> if (ret) > >> return ret; > >> > > > > That works, thanks for the fix! > > > > Tested-by: Mikko Perttunen <mperttunen@nvidia.com> Reviewed-by: Huang Rui <ray.huang@amd.com> > > Thanks. Can anybody give me an rb that I can push this to drm-misc-next > before the weekend? > > Regards, > Christian. > > > > > Mikko >
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 520b1ea9d16c..085023624fb0 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -149,6 +149,8 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) */ if (bo->base.dev) drm_gem_object_release(&bo->base); + else + dma_resv_fini(&bo->base._resv); kfree(nvbo); } @@ -330,6 +332,10 @@ nouveau_bo_new(struct nouveau_cli *cli, u64 size, int align, if (IS_ERR(nvbo)) return PTR_ERR(nvbo); + nvbo->bo.base.size = size; + dma_resv_init(&nvbo->bo.base._resv); + drm_vma_node_reset(&nvbo->bo.base.vma_node); + ret = nouveau_bo_init(nvbo, size, align, domain, sg, robj); if (ret) return ret;
TTMs buffer objects are based on GEM objects for quite a while and rely on initializing those fields before initializing the TTM BO. Noveau now doesn't init the GEM object for internally allocated BOs, so make sure that we at least initialize some necessary fields. Signed-off-by: Christian König <christian.koenig@amd.com> --- drivers/gpu/drm/nouveau/nouveau_bo.c | 6 ++++++ 1 file changed, 6 insertions(+)