[PATCH v1 1/1] drm/xe/guc: Do not assert CTB state while sending MMIO

[Michal Wajdeczko]

On 08.11.2024 04:04, Matthew Brost wrote:
> On Thu, Nov 07, 2024 at 05:56:09PM +0100, Michal Wajdeczko wrote:
>>
>>
>> On 07.11.2024 16:13, Tomasz Lis wrote:
>>> During VF post-migration recovery, MMIO communication channel to GuC
>>> is used, despite CTB channel being enabled. This behavior is rooted
>>> in the save-restore architecture specification.
>>>
>>> Therefore, a VF driver cannot assert that CTB is disabled while sending
>>> MMIO messages to GuC. Such assertion needs to be PF only, or be removed.
>>
>> + Matt
>>
> 
> IIRC there was a big arch push to avoid using MMIO communication
> entirely but that was several years ago and lost track of GuC interface
> specifics since then.
> 
>>>
>>> This patch simply removes the assertion.
>>>
>>> Signed-off-by: Tomasz Lis <tomasz.lis at intel.com>
>>> Suggested-by: Michał Wajdeczko <michal.wajdeczko at intel.com>
>>
>> in fact I was already trying to fix that [1], but I had to wait until we
>> have some S/R code in place, so I'm fine with it as-is
>>
>> Reviewed-by: Michal Wajdeczko <michal.wajdeczko at intel.com>
>>
> 
> I'll defer to Michal here as I'm sure you are more well versed expected
> GuC interaction at this point that me.

thanks, pushed as-is to drm-xe-next

> 
> Matt
> 
>> [1] https://patchwork.freedesktop.org/series/128721/#rev1
>>
>>> ---
>>>  drivers/gpu/drm/xe/xe_guc.c | 1 -
>>>  1 file changed, 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/xe/xe_guc.c b/drivers/gpu/drm/xe/xe_guc.c
>>> index 7224593c9ce9..df1ba94cf4ca 100644
>>> --- a/drivers/gpu/drm/xe/xe_guc.c
>>> +++ b/drivers/gpu/drm/xe/xe_guc.c
>>> @@ -945,7 +945,6 @@ int xe_guc_mmio_send_recv(struct xe_guc *guc, const u32 *request,
>>>  
>>>  	BUILD_BUG_ON(VF_SW_FLAG_COUNT != MED_VF_SW_FLAG_COUNT);
>>>  
>>> -	xe_assert(xe, !xe_guc_ct_enabled(&guc->ct));
>>>  	xe_assert(xe, len);
>>>  	xe_assert(xe, len <= VF_SW_FLAG_COUNT);
>>>  	xe_assert(xe, len <= MED_VF_SW_FLAG_COUNT);
>>