Reviewed-by: Lyude Paul lyude@redhat.com

Let me know when this has been pushed somewhere

On Mon, 2021-02-01 at 11:01 +1000, Dave Airlie wrote:

From: Dave Airlie airlied@redhat.com

Since I wrote the below patch if you run a debug kernel you can a dma debug warning like: nouveau 0000:1f:00.0: DMA-API: device driver tries to sync DMA memory it has not allocated [device address=0x000000016e012000] [size=4096 bytes]

The old nouveau code wasn't consolidate the pages like the ttm code, but the dma-debug expects the sync code to give it the same base/range pairs as the allocator.

Fix the nouveau sync code to consolidate pages before calling the sync code.

Fixes: bd549d35b4be0 ("nouveau: use ttm populate mapping functions. (v2)") Reported-by: Lyude Paul lyude@redhat.com Reviewed-by: Ben Skeggs bskeggs@redhat.com Signed-off-by: Dave Airlie airlied@redhat.com

---

drivers/gpu/drm/nouveau/nouveau_bo.c | 35 +++++++++++++++++++++++-----  1 file changed, 29 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 33dc886d1d6d..2375711877cf 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -547,7 +547,7 @@ nouveau_bo_sync_for_device(struct nouveau_bo *nvbo)  {         struct nouveau_drm *drm = nouveau_bdev(nvbo->bo.bdev);         struct ttm_tt *ttm_dma = (struct ttm_tt *)nvbo->bo.ttm; -       int i; +       int i, j;           if (!ttm_dma)                 return; @@ -556,10 +556,21 @@ nouveau_bo_sync_for_device(struct nouveau_bo *nvbo)         if (nvbo->force_coherent)                 return;   -       for (i = 0; i < ttm_dma->num_pages; i++) +       for (i = 0; i < ttm_dma->num_pages; ++i) { +               struct page *p = ttm_dma->pages[i]; +               size_t num_pages = 1;

+               for (j = i + 1; j < ttm_dma->num_pages; ++j) { +                       if (++p != ttm_dma->pages[j]) +                               break;

+                       ++num_pages; +               }                 dma_sync_single_for_device(drm->dev->dev,                                            ttm_dma->dma_address[i], -                                          PAGE_SIZE, DMA_TO_DEVICE); +                                          num_pages * PAGE_SIZE, DMA_TO_DEVICE); +               i += num_pages; +       }  }    void @@ -567,7 +578,7 @@ nouveau_bo_sync_for_cpu(struct nouveau_bo *nvbo)  {         struct nouveau_drm *drm = nouveau_bdev(nvbo->bo.bdev);         struct ttm_tt *ttm_dma = (struct ttm_tt *)nvbo->bo.ttm; -       int i; +       int i, j;           if (!ttm_dma)                 return; @@ -576,9 +587,21 @@ nouveau_bo_sync_for_cpu(struct nouveau_bo *nvbo)         if (nvbo->force_coherent)                 return;   -       for (i = 0; i < ttm_dma->num_pages; i++) +       for (i = 0; i < ttm_dma->num_pages; ++i) { +               struct page *p = ttm_dma->pages[i]; +               size_t num_pages = 1;

+               for (j = i + 1; j < ttm_dma->num_pages; ++j) { +                       if (++p != ttm_dma->pages[j]) +                               break;

+                       ++num_pages; +               }

dma_sync_single_for_cpu(drm->dev->dev, ttm_dma-

...

dma_address[i],

-                                       PAGE_SIZE, DMA_FROM_DEVICE); +                                       num_pages * PAGE_SIZE, DMA_FROM_DEVICE); +               i += num_pages; +       }  }    void nouveau_bo_add_io_reserve_lru(struct ttm_buffer_object *bo)