regression on raven (gmc9)

regression on raven (gmc9)

[Tom St Denis]

This commit

[root@raven linux]# git bisect bad
c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
Author: Alex Deucher <alexander.deucher@amd.com>
Date:   Tue Mar 27 17:10:56 2018 -0500

     drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu tlb 
flush

     Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support it,
     it provides a write and wait in a single packet which avoids a missed
     ack if a world switch happens between the request and waiting for the
     ack.

     Reviewed-by: Huang Rui <ray.huang@amd.com>
     Reviewed-by: Christian König <christian.koenig@amd.com>
     Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

:040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8 
521c8a18b0c63e185e416a267619a92822d7f268 M      drivers

Causes my raven1 to lockup on starting X with a gfx ring stall.

Tom
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: regression on raven (gmc9)

[Alex Deucher]

On Mon, Apr 2, 2018 at 7:17 AM, Tom St Denis <tstdenis@amd.com> wrote:
> This commit
>
> [root@raven linux]# git bisect bad
> c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
> commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
> Author: Alex Deucher <alexander.deucher@amd.com>
> Date:   Tue Mar 27 17:10:56 2018 -0500
>
>     drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu tlb
> flush
>
>     Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support it,
>     it provides a write and wait in a single packet which avoids a missed
>     ack if a world switch happens between the request and waiting for the
>     ack.
>
>     Reviewed-by: Huang Rui <ray.huang@amd.com>
>     Reviewed-by: Christian König <christian.koenig@amd.com>
>     Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
>
> :040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8
> 521c8a18b0c63e185e416a267619a92822d7f268 M      drivers
>
> Causes my raven1 to lockup on starting X with a gfx ring stall.

I've reverted the patches.

Alex

>
> Tom
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: regression on raven (gmc9)

[Christian König]

Am 02.04.2018 um 17:00 schrieb Alex Deucher:
> On Mon, Apr 2, 2018 at 7:17 AM, Tom St Denis <tstdenis@amd.com> wrote:
>> This commit
>>
>> [root@raven linux]# git bisect bad
>> c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
>> commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
>> Author: Alex Deucher <alexander.deucher@amd.com>
>> Date:   Tue Mar 27 17:10:56 2018 -0500
>>
>>      drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu tlb
>> flush
>>
>>      Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support it,
>>      it provides a write and wait in a single packet which avoids a missed
>>      ack if a world switch happens between the request and waiting for the
>>      ack.
>>
>>      Reviewed-by: Huang Rui <ray.huang@amd.com>
>>      Reviewed-by: Christian König <christian.koenig@amd.com>
>>      Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
>>
>> :040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8
>> 521c8a18b0c63e185e416a267619a92822d7f268 M      drivers
>>
>> Causes my raven1 to lockup on starting X with a gfx ring stall.
> I've reverted the patches.

Yeah, you need a specialized firmware version for this which Raven 
probably doesn't have.

Anyway the root cause of the issue which should be fixed by this patch 
set was found in the meantime, so the whole patch set shouldn't be 
necessary any more.

Christian.

>
> Alex
>
>> Tom
>> _______________________________________________
>> amd-gfx mailing list
>> amd-gfx@lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

RE: regression on raven (gmc9)

[Deng, Emily]

Hi Alex and Christian,
    Those patches are important for vega10 SRIOV, without this, vega10 run many games will hang on SRIOV, 
how about to add  "if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_VEGA10)" to separate the 
bare metal and SRIOV?

Best Wishes,
Emily Deng




> -----Original Message-----
> From: amd-gfx [mailto:amd-gfx-bounces@lists.freedesktop.org] On Behalf
> Of Christian K?nig
> Sent: Monday, April 02, 2018 11:23 PM
> To: Alex Deucher <alexdeucher@gmail.com>; StDenis, Tom
> <Tom.StDenis@amd.com>
> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Huang, Ray
> <Ray.Huang@amd.com>; amd-gfx@lists.freedesktop.org
> Subject: Re: regression on raven (gmc9)
> 
> Am 02.04.2018 um 17:00 schrieb Alex Deucher:
> > On Mon, Apr 2, 2018 at 7:17 AM, Tom St Denis <tstdenis@amd.com>
> wrote:
> >> This commit
> >>
> >> [root@raven linux]# git bisect bad
> >> c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
> >> commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
> >> Author: Alex Deucher <alexander.deucher@amd.com>
> >> Date:   Tue Mar 27 17:10:56 2018 -0500
> >>
> >>      drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu
> >> tlb flush
> >>
> >>      Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support
> it,
> >>      it provides a write and wait in a single packet which avoids a missed
> >>      ack if a world switch happens between the request and waiting for the
> >>      ack.
> >>
> >>      Reviewed-by: Huang Rui <ray.huang@amd.com>
> >>      Reviewed-by: Christian König <christian.koenig@amd.com>
> >>      Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> >>
> >> :040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8
> >> 521c8a18b0c63e185e416a267619a92822d7f268 M      drivers
> >>
> >> Causes my raven1 to lockup on starting X with a gfx ring stall.
> > I've reverted the patches.
> 
> Yeah, you need a specialized firmware version for this which Raven probably
> doesn't have.
> 
> Anyway the root cause of the issue which should be fixed by this patch set
> was found in the meantime, so the whole patch set shouldn't be necessary
> any more.
> 
> Christian.
> 
> >
> > Alex
> >
> >> Tom
> >> _______________________________________________
> >> amd-gfx mailing list
> >> amd-gfx@lists.freedesktop.org
> >> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
> 
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: regression on raven (gmc9)

[Christian König]

Hi Emily,

the root cause of the hang was found and as far as I know the plan is to 
fix this with new RLC firmware.

Please see the internal thread with the hardware guys.

Thanks,
Christian.

Am 03.04.2018 um 03:54 schrieb Deng, Emily:
> Hi Alex and Christian,
>      Those patches are important for vega10 SRIOV, without this, vega10 run many games will hang on SRIOV,
> how about to add  "if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_VEGA10)" to separate the
> bare metal and SRIOV?
>
> Best Wishes,
> Emily Deng
>
>
>
>
>> -----Original Message-----
>> From: amd-gfx [mailto:amd-gfx-bounces@lists.freedesktop.org] On Behalf
>> Of Christian K?nig
>> Sent: Monday, April 02, 2018 11:23 PM
>> To: Alex Deucher <alexdeucher@gmail.com>; StDenis, Tom
>> <Tom.StDenis@amd.com>
>> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Huang, Ray
>> <Ray.Huang@amd.com>; amd-gfx@lists.freedesktop.org
>> Subject: Re: regression on raven (gmc9)
>>
>> Am 02.04.2018 um 17:00 schrieb Alex Deucher:
>>> On Mon, Apr 2, 2018 at 7:17 AM, Tom St Denis <tstdenis@amd.com>
>> wrote:
>>>> This commit
>>>>
>>>> [root@raven linux]# git bisect bad
>>>> c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
>>>> commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
>>>> Author: Alex Deucher <alexander.deucher@amd.com>
>>>> Date:   Tue Mar 27 17:10:56 2018 -0500
>>>>
>>>>       drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu
>>>> tlb flush
>>>>
>>>>       Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support
>> it,
>>>>       it provides a write and wait in a single packet which avoids a missed
>>>>       ack if a world switch happens between the request and waiting for the
>>>>       ack.
>>>>
>>>>       Reviewed-by: Huang Rui <ray.huang@amd.com>
>>>>       Reviewed-by: Christian König <christian.koenig@amd.com>
>>>>       Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
>>>>
>>>> :040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8
>>>> 521c8a18b0c63e185e416a267619a92822d7f268 M      drivers
>>>>
>>>> Causes my raven1 to lockup on starting X with a gfx ring stall.
>>> I've reverted the patches.
>> Yeah, you need a specialized firmware version for this which Raven probably
>> doesn't have.
>>
>> Anyway the root cause of the issue which should be fixed by this patch set
>> was found in the meantime, so the whole patch set shouldn't be necessary
>> any more.
>>
>> Christian.
>>
>>> Alex
>>>
>>>> Tom
>>>> _______________________________________________
>>>> amd-gfx mailing list
>>>> amd-gfx@lists.freedesktop.org
>>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>> _______________________________________________
>> amd-gfx mailing list
>> amd-gfx@lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

RE: regression on raven (gmc9)

[Deng, Emily]

Hi Christian,
     I know this, but seems it will take long time to do this, and this issue has blocked vega10 sriov for long time. 
But we want the staging to be stable on vega10 sriov  as soon as possible.

Best Wishes,
Emily Deng

> -----Original Message-----
> From: Koenig, Christian
> Sent: Tuesday, April 03, 2018 5:22 PM
> To: Deng, Emily <Emily.Deng@amd.com>; Alex Deucher
> <alexdeucher@gmail.com>; StDenis, Tom <Tom.StDenis@amd.com>
> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Huang, Ray
> <Ray.Huang@amd.com>; amd-gfx@lists.freedesktop.org
> Subject: Re: regression on raven (gmc9)
> 
> Hi Emily,
> 
> the root cause of the hang was found and as far as I know the plan is to fix
> this with new RLC firmware.
> 
> Please see the internal thread with the hardware guys.
> 
> Thanks,
> Christian.
> 
> Am 03.04.2018 um 03:54 schrieb Deng, Emily:
> > Hi Alex and Christian,
> >      Those patches are important for vega10 SRIOV, without this,
> > vega10 run many games will hang on SRIOV, how about to add  "if
> > (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_VEGA10)" to
> separate the bare metal and SRIOV?
> >
> > Best Wishes,
> > Emily Deng
> >
> >
> >
> >
> >> -----Original Message-----
> >> From: amd-gfx [mailto:amd-gfx-bounces@lists.freedesktop.org] On
> >> Behalf Of Christian K?nig
> >> Sent: Monday, April 02, 2018 11:23 PM
> >> To: Alex Deucher <alexdeucher@gmail.com>; StDenis, Tom
> >> <Tom.StDenis@amd.com>
> >> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Huang, Ray
> >> <Ray.Huang@amd.com>; amd-gfx@lists.freedesktop.org
> >> Subject: Re: regression on raven (gmc9)
> >>
> >> Am 02.04.2018 um 17:00 schrieb Alex Deucher:
> >>> On Mon, Apr 2, 2018 at 7:17 AM, Tom St Denis <tstdenis@amd.com>
> >> wrote:
> >>>> This commit
> >>>>
> >>>> [root@raven linux]# git bisect bad
> >>>> c704d7a15d292e6e222b7e990ed7e9f41617cdd8 is the first bad commit
> >>>> commit c704d7a15d292e6e222b7e990ed7e9f41617cdd8
> >>>> Author: Alex Deucher <alexander.deucher@amd.com>
> >>>> Date:   Tue Mar 27 17:10:56 2018 -0500
> >>>>
> >>>>       drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in
> >>>> gpu tlb flush
> >>>>
> >>>>       Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that
> >>>> support
> >> it,
> >>>>       it provides a write and wait in a single packet which avoids a missed
> >>>>       ack if a world switch happens between the request and waiting for
> the
> >>>>       ack.
> >>>>
> >>>>       Reviewed-by: Huang Rui <ray.huang@amd.com>
> >>>>       Reviewed-by: Christian König <christian.koenig@amd.com>
> >>>>       Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> >>>>
> >>>> :040000 040000 a08a12a316d09ea231337136cc7aa076d4da44a8
> >>>> 521c8a18b0c63e185e416a267619a92822d7f268 M      drivers
> >>>>
> >>>> Causes my raven1 to lockup on starting X with a gfx ring stall.
> >>> I've reverted the patches.
> >> Yeah, you need a specialized firmware version for this which Raven
> >> probably doesn't have.
> >>
> >> Anyway the root cause of the issue which should be fixed by this
> >> patch set was found in the meantime, so the whole patch set shouldn't
> >> be necessary any more.
> >>
> >> Christian.
> >>
> >>> Alex
> >>>
> >>>> Tom
> >>>> _______________________________________________
> >>>> amd-gfx mailing list
> >>>> amd-gfx@lists.freedesktop.org
> >>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
> >> _______________________________________________
> >> amd-gfx mailing list
> >> amd-gfx@lists.freedesktop.org
> >> https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: regression on raven (gmc9)

[Dawson Dias]

[-- Attachment #1.1: Type: text/plain, Size: 56 bytes --]

Thanks. This was locking up my Raven too.

-Dawson Dias

[-- Attachment #1.2: Type: text/html, Size: 290 bytes --]

[-- Attachment #2: Type: text/plain, Size: 154 bytes --]

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx