[PATCH] drm/radeon: Always flush VM again on < CIK

[Christian König]

The GFX CP is split up into two different engines - the PFP and the ME 
(starting with SI you additionally get the CE as well).

The PFP is responsible for reading commands out of memory and forwarding 
it to the ME (or the CE). Some commands can be executed on the PFP as 
well, like simple register writes, but most commands can only run on the ME.

The PFP and the ME are connected through a 8 entry ring buffer (IIRC), 
so when you do something on the ME the PFP depends on you need to block 
the PFP for the ME to finish it's operation.

It strongly depends on what we want to do if we should use the PFP or 
the ME, but in most cases (like writing to memory) it's only the ME that 
can do the operation anyway.

Regards,
Christian.

Am 07.08.2014 um 17:38 schrieb Marek Olšák:
> So what's difference between WRITE_DATA with PFP vs ME? Would it also
> be preferable for DMA_DATA and COPY_DATA?
>
> Marek
>
> On Thu, Aug 7, 2014 at 3:59 PM, Alex Deucher <alexdeucher at gmail.com> wrote:
>> On Thu, Aug 7, 2014 at 3:46 AM, Michel Dänzer <michel at daenzer.net> wrote:
>>> From: Michel Dänzer <michel.daenzer at amd.com>
>>>
>>> Not doing this causes piglit hangs[0] on my Cape Verde card. No issues on
>>> Bonaire and Kaveri though.
>>>
>>> [0] Same symptoms as those fixed on CIK by 'drm/radeon: set VM base addr
>>> using the PFP v2'.
>>>
>>> Signed-off-by: Michel Dänzer <michel.daenzer at amd.com>
>> We should be using PFP as much as possible.  Does the attached patch help?
>>
>> Alex
>>
>>> ---
>>>   drivers/gpu/drm/radeon/radeon_vm.c | 4 +++-
>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c
>>> index ccae4d9..898cbb7 100644
>>> --- a/drivers/gpu/drm/radeon/radeon_vm.c
>>> +++ b/drivers/gpu/drm/radeon/radeon_vm.c
>>> @@ -238,7 +238,9 @@ void radeon_vm_flush(struct radeon_device *rdev,
>>>          uint64_t pd_addr = radeon_bo_gpu_offset(vm->page_directory);
>>>
>>>          /* if we can't remember our last VM flush then flush now! */
>>> -       if (!vm->last_flush || pd_addr != vm->pd_gpu_addr) {
>>> +       /* XXX figure out why we have to flush all the time before CIK */
>>> +       if (rdev->family < CHIP_BONAIRE ||
>>> +           !vm->last_flush || pd_addr != vm->pd_gpu_addr) {
>>>                  trace_radeon_vm_flush(pd_addr, ring, vm->id);
>>>                  vm->pd_gpu_addr = pd_addr;
>>>                  radeon_ring_vm_flush(rdev, ring, vm);
>>> --
>>> 2.0.1
>>>
>>> _______________________________________________
>>> dri-devel mailing list
>>> dri-devel at lists.freedesktop.org
>>> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>>
> _______________________________________________
> dri-devel mailing list
> dri-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel