Re: [PATCH 1/3] drm/radeon: fix VMID use tracking

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Jan 24, 2014 at 10:52 AM, Christian König
<deathsimple@xxxxxxxxxxx> wrote:
> Am 23.01.2014 18:46, schrieb Alex Deucher:
>
>> On Thu, Jan 23, 2014 at 8:24 AM, Christian König
>> <deathsimple@xxxxxxxxxxx> wrote:
>>>
>>> From: Christian König <christian.koenig@xxxxxxx>
>>>
>>> Otherwise we allocate a new VMID on nearly every submit.
>>
>> I wonder if this fix would allow us to fix up this:
>>
>> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=466476dfdcafbb4286ffa232a3a792731b9dc852
>
>
> It's probably a step into the right direction, but not necessary the
> complete solution.
>
> For now I would say we should keep the extra flushing as it is.

Patches applied.  thanks!

Alex

>
> Christian.
>
>
>> Alex
>>
>>> Signed-off-by: Christian König <christian.koenig@xxxxxxx>
>>> ---
>>>   drivers/gpu/drm/radeon/radeon.h      | 2 ++
>>>   drivers/gpu/drm/radeon/radeon_gart.c | 8 +++++++-
>>>   2 files changed, 9 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/radeon/radeon.h
>>> b/drivers/gpu/drm/radeon/radeon.h
>>> index c5519ca..4a8ac1c 100644
>>> --- a/drivers/gpu/drm/radeon/radeon.h
>>> +++ b/drivers/gpu/drm/radeon/radeon.h
>>> @@ -867,6 +867,8 @@ struct radeon_vm {
>>>          struct radeon_fence             *fence;
>>>          /* last flush or NULL if we still need to flush */
>>>          struct radeon_fence             *last_flush;
>>> +       /* last use of vmid */
>>> +       struct radeon_fence             *last_id_use;
>>>   };
>>>
>>>   struct radeon_vm_manager {
>>> diff --git a/drivers/gpu/drm/radeon/radeon_gart.c
>>> b/drivers/gpu/drm/radeon/radeon_gart.c
>>> index 96e4400..0e9143b 100644
>>> --- a/drivers/gpu/drm/radeon/radeon_gart.c
>>> +++ b/drivers/gpu/drm/radeon/radeon_gart.c
>>> @@ -713,7 +713,7 @@ struct radeon_fence *radeon_vm_grab_id(struct
>>> radeon_device *rdev,
>>>          unsigned i;
>>>
>>>          /* check if the id is still valid */
>>> -       if (vm->fence && vm->fence == rdev->vm_manager.active[vm->id])
>>> +       if (vm->last_id_use && vm->last_id_use ==
>>> rdev->vm_manager.active[vm->id])
>>>                  return NULL;
>>>
>>>          /* we definately need to flush */
>>> @@ -769,6 +769,9 @@ void radeon_vm_fence(struct radeon_device *rdev,
>>>
>>>          radeon_fence_unref(&vm->fence);
>>>          vm->fence = radeon_fence_ref(fence);
>>> +
>>> +       radeon_fence_unref(&vm->last_id_use);
>>> +       vm->last_id_use = radeon_fence_ref(fence);
>>>   }
>>>
>>>   /**
>>> @@ -1303,6 +1306,8 @@ void radeon_vm_init(struct radeon_device *rdev,
>>> struct radeon_vm *vm)
>>>   {
>>>          vm->id = 0;
>>>          vm->fence = NULL;
>>> +       vm->last_flush = NULL;
>>> +       vm->last_id_use = NULL;
>>>          mutex_init(&vm->mutex);
>>>          INIT_LIST_HEAD(&vm->list);
>>>          INIT_LIST_HEAD(&vm->va);
>>> @@ -1341,5 +1346,6 @@ void radeon_vm_fini(struct radeon_device *rdev,
>>> struct radeon_vm *vm)
>>>          }
>>>          radeon_fence_unref(&vm->fence);
>>>          radeon_fence_unref(&vm->last_flush);
>>> +       radeon_fence_unref(&vm->last_id_use);
>>>          mutex_unlock(&vm->mutex);
>>>   }
>>> --
>>> 1.8.1.2
>>>
>
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/dri-devel





[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux