similar to: [PATCH v3] drm/virtio: fix missing dma_fence_put() in virtio_gpu_execbuffer_ioctl()

On Tue, Jul 21, 2020 at 06:16:47PM +0800, Xin He wrote: > From: Qi Liu <liuqi.16 at bytedance.com> > > We should put the reference count of the fence after calling > virtio_gpu_cmd_submit(). So add the missing dma_fence_put(). > virtio_gpu_cmd_submit(vgdev, buf, exbuf->size, > vfpriv->ctx_id, buflist, out_fence); > +

From: Xin He <hexin.op at bytedance.com> [ Upstream commit 836b194d65782aaec4485a07d2aab52d3f698505 ] Before setting shmem->pages to NULL, kfree() should be called. Signed-off-by: Xin He <hexin.op at bytedance.com> Reviewed-by: Qi Liu <liuqi.16 at bytedance.com> Link: http://patchwork.freedesktop.org/patch/msgid/20200722051851.72662-1-hexin.op at bytedance.com

From: Xin He <hexin.op at bytedance.com> [ Upstream commit 836b194d65782aaec4485a07d2aab52d3f698505 ] Before setting shmem->pages to NULL, kfree() should be called. Signed-off-by: Xin He <hexin.op at bytedance.com> Reviewed-by: Qi Liu <liuqi.16 at bytedance.com> Link: http://patchwork.freedesktop.org/patch/msgid/20200722051851.72662-1-hexin.op at bytedance.com

On Mon, Aug 07, 2023 at 07:09:31PM +0800, Qi Zheng wrote: > Currently, the synchronize_shrinkers() is only used by TTM pool. It only > requires that no shrinkers run in parallel. > > After we use RCU+refcount method to implement the lockless slab shrink, > we can not use shrinker_rwsem or synchronize_rcu() to guarantee that all > shrinker invocations have seen an update before

On Tue, Jun 18, 2019 at 03:58:15PM +0200, Gerd Hoffmann wrote: > Use gem reservation helpers and direct reservation_object_* calls > instead of ttm. > > Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> > --- > drivers/gpu/drm/virtio/virtgpu_ioctl.c | 36 ++++++++++++-------------- > 1 file changed, 17 insertions(+), 19 deletions(-) > > diff --git

Lockdep says we can't call vmemdup() while having objects reserved because it needs the mmap semaphore. So reorder the calls reserve the objects later. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c

From: Gustavo Padovan <gustavo.padovan at collabora.co.uk> Instead of wrapping virtio_gpu_execbuffer() to execute the ioctl just execute it directly. Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.co.uk> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 24 ++++++++---------------- 1 file changed, 8 insertions(+), 16 deletions(-) diff --git

From: Gustavo Padovan <gustavo.padovan at collabora.co.uk> Instead of wrapping virtio_gpu_execbuffer() to execute the ioctl just execute it directly. Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.co.uk> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 24 ++++++++---------------- 1 file changed, 8 insertions(+), 16 deletions(-) diff --git

Hi, > Also, I strongly recommend you do a very basic igt to exercise this, i.e. > allocate some buffers, submit them in a dummby op, then close the entire > drmfd. The old version should at least have tripped over kasan, maybe even > oopses somewhere. Hmm, I suspect I have to extend igt for that (adding support for virtio ioctls), right? A quick and dirty test (run webgl demo in

On Thu, Jul 4, 2019 at 4:25 AM Gerd Hoffmann <kraxel at redhat.com> wrote: > > Hi, > > > > if (fence) > > > virtio_gpu_fence_emit(vgdev, hdr, fence); > > > + if (vbuf->objs) { > > > + virtio_gpu_array_add_fence(vbuf->objs, &fence->f); > > > +

Rework fencing workflow, starting with virtio_gpu_execbuffer_ioctl. Stop using ttm helpers, use the virtio_gpu_array_* helpers (which work on the reservation objects directly) instead. Also store the object array in struct virtio_gpu_vbuffer, so we explicitly keep a reference of all buffers used instead of depending on ttm_bo_put() checking whenever the object is actually idle before releasing

On Tue, Jul 2, 2019 at 7:19 AM Gerd Hoffmann <kraxel at redhat.com> wrote: > > Rework fencing workflow, starting with virtio_gpu_execbuffer_ioctl. > Stop using ttm helpers, use the virtio_gpu_array_* helpers (which work > on the reservation objects directly) instead. > > New workflow: > > (1) All gem objects needed by a command are added to a >

On Fri, Jun 28, 2019 at 5:13 AM Gerd Hoffmann <kraxel at redhat.com> wrote: > > Use gem reservation helpers and direct reservation_object_* calls > instead of ttm. > > v5: fix fencing (Chia-I Wu). > v3: Also attach the array of gem objects to the virtio command buffer, > so we can drop the object references in the completion callback. Needed > because ttm fence

On Wed, Jun 19, 2019 at 11:08 PM Gerd Hoffmann <kraxel at redhat.com> wrote: > > Use gem reservation helpers and direct reservation_object_* calls > instead of ttm. > > v3: Also attach the array of gem objects to the virtio command buffer, > so we can drop the object references in the completion callback. Needed > because ttm fence helpers grab a reference for us, but

Hi, > > if (fence) > > virtio_gpu_fence_emit(vgdev, hdr, fence); > > + if (vbuf->objs) { > > + virtio_gpu_array_add_fence(vbuf->objs, &fence->f); > > + virtio_gpu_array_unlock_resv(vbuf->objs); > > + } > This is with the spinlock held. Maybe we should move the >

Hi, > > if (fence) > > virtio_gpu_fence_emit(vgdev, hdr, fence); > > + if (vbuf->objs) { > > + virtio_gpu_array_add_fence(vbuf->objs, &fence->f); > > + virtio_gpu_array_unlock_resv(vbuf->objs); > > + } > This is with the spinlock held. Maybe we should move the >

On Wed, Jun 19, 2019 at 11:04:16AM +0200, Gerd Hoffmann wrote: > Use gem reservation helpers and direct reservation_object_* calls > instead of ttm. > > v3: Also attach the array of gem objects to the virtio command buffer, > so we can drop the object references in the completion callback. Needed > because ttm fence helpers grab a reference for us, but gem helpers >

Hi, > > --- a/drivers/gpu/drm/virtio/virtgpu_drv.h > > +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h > > @@ -120,9 +120,9 @@ struct virtio_gpu_vbuffer { > > > > char *resp_buf; > > int resp_size; > > - > > virtio_gpu_resp_cb resp_cb; > > > > + struct virtio_gpu_object_array *objs; > This can use a comment

Use gem reservation helpers and direct reservation_object_* calls instead of ttm. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 36 ++++++++++++-------------- 1 file changed, 17 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 5cffd2e54c04..6db6a6e92dde

Use gem reservation helpers and direct reservation_object_* calls instead of ttm. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 36 ++++++++++++-------------- 1 file changed, 17 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 5cffd2e54c04..6db6a6e92dde