[PATCH 10/10] drm/ttm: flip the switch for driver allocated resources v2

[Thomas Hellström (Intel)]

On 6/7/21 12:37 PM, Christian König wrote:
> Am 07.06.21 um 12:15 schrieb Thomas Hellström (Intel):
>>
>> On 6/2/21 12:09 PM, Christian König wrote:
>>> Instead of both driver and TTM allocating memory finalize embedding the
>>> ttm_resource object as base into the driver backends.
>>>
>>> v2: fix typo in vmwgfx grid mgr and double init in amdgpu_vram_mgr.c
>>>
>>> Signed-off-by: Christian König <christian.koenig at amd.com>
>>> Reviewed-by: Matthew Auld <matthew.auld at intel.com>
>>> ---
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c   | 44 ++++++--------
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_object.c    |  2 +-
>>>   .../gpu/drm/amd/amdgpu/amdgpu_res_cursor.h    |  5 +-
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c  | 60 
>>> +++++++++----------
>>>   drivers/gpu/drm/drm_gem_vram_helper.c         |  3 +-
>>>   drivers/gpu/drm/nouveau/nouveau_bo.c          |  8 +--
>>>   drivers/gpu/drm/nouveau/nouveau_mem.c         | 11 ++--
>>>   drivers/gpu/drm/nouveau/nouveau_mem.h         | 14 ++---
>>>   drivers/gpu/drm/nouveau/nouveau_ttm.c         | 32 +++++-----
>>>   drivers/gpu/drm/ttm/ttm_range_manager.c       | 23 +++----
>>>   drivers/gpu/drm/ttm/ttm_resource.c            | 18 +-----
>>>   drivers/gpu/drm/ttm/ttm_sys_manager.c         | 12 ++--
>>>   drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c | 24 ++++----
>>>   drivers/gpu/drm/vmwgfx/vmwgfx_thp.c           | 27 ++++-----
>>>   include/drm/ttm/ttm_range_manager.h           |  3 +-
>>>   include/drm/ttm/ttm_resource.h                | 43 ++++++-------
>>>   16 files changed, 140 insertions(+), 189 deletions(-)
>> ...
>>>   diff --git a/drivers/gpu/drm/ttm/ttm_range_manager.c 
>>> b/drivers/gpu/drm/ttm/ttm_range_manager.c
>>> index ce5d07ca384c..c32e1aee2481 100644
>>> --- a/drivers/gpu/drm/ttm/ttm_range_manager.c
>>> +++ b/drivers/gpu/drm/ttm/ttm_range_manager.c
>>> @@ -58,7 +58,7 @@ to_range_manager(struct ttm_resource_manager *man)
>>>   static int ttm_range_man_alloc(struct ttm_resource_manager *man,
>>>                      struct ttm_buffer_object *bo,
>>>                      const struct ttm_place *place,
>>> -                   struct ttm_resource *mem)
>>> +                   struct ttm_resource **res)
>>>   {
>>>       struct ttm_range_manager *rman = to_range_manager(man);
>>>       struct ttm_range_mgr_node *node;
>>> @@ -83,37 +83,30 @@ static int ttm_range_man_alloc(struct 
>>> ttm_resource_manager *man,
>>>         spin_lock(&rman->lock);
>>>       ret = drm_mm_insert_node_in_range(mm, &node->mm_nodes[0],
>>> -                      mem->num_pages, bo->page_alignment, 0,
>>> +                      node->base.num_pages,
>>> +                      bo->page_alignment, 0,
>>>                         place->fpfn, lpfn, mode);
>>>       spin_unlock(&rman->lock);
>>>   -    if (unlikely(ret)) {
>>> +    if (unlikely(ret))
>>>           kfree(node);
>>> -    } else {
>>> -        mem->mm_node = &node->mm_nodes[0];
>>> -        mem->start = node->mm_nodes[0].start;
>>> -    }
>>> +    else
>>> +        node->base.start = node->mm_nodes[0].start;
>>>         return ret;
>>>   }
>>
>> Looks like this patch forgets to assign *@res. Null pointer derefs 
>> when testing i915.
>
> I should really CC the Intel list for TTM patches as well. The CI 
> system should have spotted that.

Unfortunately, the dg1 system is not participating in CI yet AFAICT, but 
moving forward I think it's a good idea.

/Thomas