Reviewed-by: Xiangliang Yu
-Original Message-
From: Lou, Wentao
Sent: Tuesday, December 18, 2018 3:52 PM
To: amd-gfx@lists.freedesktop.org
Cc: Lou, Wentao
Subject: [PATCH] drm/amdgpu: psp_ring_destroy cause psp->km_ring.ring_mem NULL
psp_ring_destroy inside psp_load_fw cause psp->km
psp_ring_destroy inside psp_load_fw cause psp->km_ring.ring_mem NULL.
Call Trace occurred when psp_cmd_submit.
should be psp_ring_stop instead.
Change-Id: Ib332004b3b9edc9e002adc532b2d45cdad929b05
Signed-off-by: Wentao Lou
---
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 2 +-
1 file changed, 1 ins
effect asics: VEGA10 and VEGA12
Change-Id: Idfac315920a4fae771b080ff836bd4c5c980348e
Signed-off-by: Jim Qu
Acked-by: Alex Deucher
---
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
b/drivers/gp
Hi Xiangliang,
Could you add more comment on the session_id in the commit message? Like using
the session_id to distinguish each VF v.s. PF etc.
With that added,Reviewed-by: Feifei Xu
Thanks
Feifei
-Original Message-
From: amd-gfx On Behalf Of Xiangliang Yu
Sent: Thursday, December 13
Reviewed-by: Feifei Xu
-Original Message-
From: amd-gfx On Behalf Of Xiangliang Yu
Sent: Thursday, December 13, 2018 3:42 PM
To: amd-gfx@lists.freedesktop.org
Cc: Yu, Xiangliang
Subject: [PATCH 2/2] drm/amdgpu/psp: Fix can't detect psp INVOKE command failed
There isn't ucode when execu
Anyone to help reviewing?
Thanks!
-Original Message-
From: amd-gfx On Behalf Of Xiangliang Yu
Sent: Thursday, December 13, 2018 3:42 PM
To: amd-gfx@lists.freedesktop.org
Cc: Yu, Xiangliang
Subject: [PATCH 2/2] drm/amdgpu/psp: Fix can't detect psp INVOKE command failed
There isn't ucode
Different ASIC has different sdma doorbell range. Add
a per device sdma_doorbell_range field and initialize
it.
Change-Id: Idd980db1a72cfb373e24ac23ba3e48bb329ed4ad
Signed-off-by: Oak Zeng
---
drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h | 2 ++
drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c | 1 +
Different ASIC has different SDMA queues so different
SDMA doorbell range. Introduce an extra parameter
to sdma_doorbell_range function and set sdma doorbell
range correctly.
Change-Id: I9b8d75b04f5a47ef1c6fd7cc1caaefd98dd2ff2b
Signed-off-by: Oak Zeng
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h|
Yeah, make sense.
Let me give another patch: [PATCH] drm/amdgpu: print process info when job
timeout
-Original Message-
From: Koenig, Christian
Sent: Monday, December 17, 2018 9:39 PM
To: Liu, Monk ; Huang, Trigger ;
amd-gfx@lists.freedesktop.org
Cc: Grodzovsky, Andrey ; Qu, Jim
Subje
When a job is timeout, try to print the related process information
for debugging
Signed-off-by: Trigger Huang
---
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 4
1 file changed, 4 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
in
On Mon, Dec 17, 2018 at 1:05 PM Zeng, Oak wrote:
>
> Use sdma_engine[8] array instead of sdma_engine0~7 so it is easier
> to program.
>
> Change-Id: I04873ddfd82e613c28908ae81665c6b02836d093
> Signed-off-by: Oak Zeng
Reviewed-by: Alex Deucher
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
On 2018-12-14 3:47 a.m., Daniel Vetter wrote:
> On Thu, Dec 13, 2018 at 08:25:31PM -0500, Lyude Paul wrote:
>> There should be no functional changes here
>
> Would be good to explain what you did refactor here, instead of me trying
> to reconstruct it from the patch. Especially pre-coffee that hel
On 2018-12-14 3:42 a.m., Daniel Vetter wrote:
> On Thu, Dec 13, 2018 at 08:25:30PM -0500, Lyude Paul wrote:
>> There's no reason we need this, it's just confusing looking.
>>
>> Signed-off-by: Lyude Paul
>> Cc: Juston Li
>> ---
>> drivers/gpu/drm/drm_dp_mst_topology.c | 4 +---
>> 1 file chang
Decauple sched threads stop and start and ring mirror
list handling from the policy of what to do about the
guilty jobs.
When stoppping the sched thread and detaching sched fences
from non signaled HW fenes wait for all signaled HW fences
to complete before rerunning the jobs.
v2: Fix resubmission
Expedite job deletion from ring mirror list to the HW fence signal
callback instead from finish_work, together with waiting for all
such fences to signal in drm_sched_stop we garantee that
already signaled job will not be processed twice.
Remove the sched finish fence callback and just submit finis
It's not a core function, and the matching atomic functions are also
not in the core. Plus the suspend/resume helper is also already there.
Needs a tiny bit of open-coding, but less midlayer beats that I think.
v2: Rebase onto ast (which gained a new user).
Cc: Sam Bobroff
Reviewed-by: Alex Deu
On 2018-12-05 12:08 p.m., Nicholas Kazlauskas wrote:
> [Why]
> When the flip-rate is below the minimum supported variable refresh rate
> range for the monitor the front porch wait will timeout and be
> frequently misaligned resulting in stuttering and/or flickering.
>
> The FreeSync module can sti
On 12/17/2018 01:51 PM, Wentland, Harry wrote:
> On 2018-12-15 4:42 a.m., Mikhail Gavrilov wrote:
>> On Sat, 15 Dec 2018 at 00:36, Wentland, Harry wrote:
>>> Looks like there's an error before this happens that might get us into this
>>> mess:
>>>
>>> [ 229.741741] [drm:amdgpu_job_timedout [am
On 2018-12-15 4:42 a.m., Mikhail Gavrilov wrote:
> On Sat, 15 Dec 2018 at 00:36, Wentland, Harry wrote:
>>
>> Looks like there's an error before this happens that might get us into this
>> mess:
>>
>> [ 229.741741] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx timeout,
>> signaled seq=286
Am 17.12.18 um 17:57 schrieb Grodzovsky, Andrey:
>
> On 12/17/2018 10:27 AM, Christian König wrote:
>> Am 10.12.18 um 22:43 schrieb Andrey Grodzovsky:
>>> Decauple sched threads stop and start and ring mirror
>>> list handling from the policy of what to do about the
>>> guilty jobs.
>>> When stoppp
Use sdma_engine[8] array instead of sdma_engine0~7 so it is easier
to program.
Change-Id: I04873ddfd82e613c28908ae81665c6b02836d093
Signed-off-by: Oak Zeng
---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 8
drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h | 9 +
drivers/gpu/drm
Hi Chunming Zhou,
On Fri, 2 Nov 2018 at 08:27, Chunming Zhou wrote:
>
> Signed-off-by: Chunming Zhou
> ---
> include/drm/drm.h | 38 ++
Please read through include/drm/README about how include/drm/ should be updated.
Thanks
Emil
_
On 12/17/2018 10:27 AM, Christian König wrote:
> Am 10.12.18 um 22:43 schrieb Andrey Grodzovsky:
>> Decauple sched threads stop and start and ring mirror
>> list handling from the policy of what to do about the
>> guilty jobs.
>> When stoppping the sched thread and detaching sched fences
>> from
From: Young Xiao
When looking up the connector type make sure the index
is valid. Avoids a later crash if we read past the end
of the array.
See commit e1718d97aa88 ("drm/amdgpu: avoid a possible
array overflow") for detail.
Signed-off-by: Young Xiao
---
drivers/gpu/drm/radeon/radeon_atombio
Am 10.12.18 um 22:43 schrieb Andrey Grodzovsky:
Decauple sched threads stop and start and ring mirror
list handling from the policy of what to do about the
guilty jobs.
When stoppping the sched thread and detaching sched fences
from non signaled HW fenes wait for all signaled HW fences
to complet
Reviewed-by: Alex Deucher
From: amd-gfx on behalf of Evan Quan
Sent: Monday, December 17, 2018 4:59:02 AM
To: amd-gfx@lists.freedesktop.org
Cc: Quan, Evan
Subject: [PATCH] drm/amdgpu: correct the return value for error case
It should not return 0 for error cas
On 12/17/2018 04:53 AM, Michel Dänzer wrote:
> On 2018-12-15 6:25 a.m., Grodzovsky, Andrey wrote:
>> On 12/14/2018 02:17 PM, Kazlauskas, Nicholas wrote:
>>> On 12/14/18 2:06 PM, Grodzovsky, Andrey wrote:
In general I agree with Michel that DRM solution is required to
properly address t
If register value is updating, reset timeout counter.
It improves robustness of SOC15_WAIT_ON_RREG.
Signed-off-by: James Zhu
---
drivers/gpu/drm/amd/amdgpu/soc15_common.h | 9 +++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h
b/dri
If monitor register value is updating, reset timeout counter.
It improves robustness of SOC15_WAIT_ON_RREG.
Signed-off-by: James Zhu
---
drivers/gpu/drm/amd/amdgpu/soc15_common.h | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h
Hi guys,
We could also print shaders, IBs or textures to figure out what's going
wrong, but that would be overkill as well.
The log is to notice that something is wrong and not to a detailed crash
report.
The PID is perfectly sufficient to identify the process which triggered
an issue and whe
It should not return 0 for error case as '0' is actually
a special value for index.
Change-Id: Iced8b4427d4403f86826a7c8e3c9d1da3394246c
Signed-off-by: Evan Quan
---
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h | 12 ++--
drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 15 +--
drivers
On 2018-12-15 6:25 a.m., Grodzovsky, Andrey wrote:
> On 12/14/2018 02:17 PM, Kazlauskas, Nicholas wrote:
>> On 12/14/18 2:06 PM, Grodzovsky, Andrey wrote:
>>> In general I agree with Michel that DRM solution is required to
>>> properly address this but since now it's not really obvious what is the
Hi Christian,
I think for some SRIOV customers they need this rich information,
Maybe we can use an kernel option to let user select if rich or simple
information should be printed upon job TDR ?
In SRIOV branch we can set it enable by default while set by default disable
for drm-next
/Monk
Hi Christian,
Yes, if the test machine is still there for debugging, we can login it and
check a lot of things, such as ' ps -p 1 -o args ' as you suggested.
But sometimes, the system is not alive anymore, and we only got some log files
(such as kern.log ) from QA or customers.
And at this tim
34 matches
Mail list logo
