Signed-off-by: Junwei Zhang
---
amdgpu/amdgpu_bo.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 71cd666..8fe3cfd 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -952,7 +952,10 @@ int amdgpu_bo_va_op_raw(amdgpu_de
Based on latest PRT support by Nicolai
Junwei Zhang (2):
amdgpu: add new VA operations CLEAR and REPLACE
amdgpu: add REPLACE and CLEAR checking for VA op
amdgpu/amdgpu_bo.c | 5 -
include/drm/amdgpu_drm.h | 2 ++
2 files changed, 6 insertions(+), 1 deletion(-)
--
1.9.1
_
Signed-off-by: Junwei Zhang
---
include/drm/amdgpu_drm.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/include/drm/amdgpu_drm.h b/include/drm/amdgpu_drm.h
index 7002dc4..a73b40a 100644
--- a/include/drm/amdgpu_drm.h
+++ b/include/drm/amdgpu_drm.h
@@ -410,6 +410,8 @@ struct drm_amdgpu_gem_
Change-Id: I730eed1bca04725696cd086e9c6ba1a31bafd8a7
Signed-off-by: Evan Quan
---
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index d7f286d..d9f036b 100644
--- a/driv
> -Original Message-
> From: Michel Dänzer [mailto:mic...@daenzer.net]
> Sent: Friday, March 17, 2017 11:32
> To: Zhang, Jerry
> Cc: x...@lists.x.org; amd-gfx@lists.freedesktop.org
> Subject: Re: [ANNOUNCE] xf86-video-amdgpu 1.3.0
>
> On 17/03/17 11:32 AM, Zhang, Jerry wrote:
> > Hi Michel
On 17/03/17 11:32 AM, Zhang, Jerry wrote:
> Hi Michel,
>
>> * Use libdrm_amdgpu functionality to determine the GPU marketing name,
>> remove corresponding tables from this driver.
>
> Could you elaborate it?
> I'd like to know how DDX(amdgpu) get the marketing name.
It uses libdrm_amdgpu's amd
On 17/03/17 11:35 AM, Zhou, David(ChunMing) wrote:
> Hi Christian, Andres,
>
> I understand your mean, especially high priority queue from Andres is
> for VR, which is liking an engine to himself, but our scheduler is a
> generic thing, which is used by multiple case. From scheduler
> perspective,
Hi Christian, Andres,
I understand your mean, especially high priority queue from Andres is for VR,
which is liking an engine to himself, but our scheduler is a generic thing,
which is used by multiple case.
From scheduler perspective, scheduling policy is popular, e.g. CPU scheduler,
it provid
Hi Michel,
> * Use libdrm_amdgpu functionality to determine the GPU marketing name,
> remove corresponding tables from this driver.
Could you elaborate it?
I'd like to know how DDX(amdgpu) get the marketing name.
I hope we go the same way in hybrid stack as well.
Regards,
Jerry (Junwei Zhang)
On 17/03/17 09:16 AM, Zhang, Jerry wrote:
>
> Thanks for your reminder.
>
> I always keep it in mind.
[...]
> *From:*Koenig, Christian
>
>> fell free to use this patch internally, but the upstream copy of
>> include/drm/amdgpu_drm.h should be updated when the kernel change land.
>>
>> Emil us
Reviewed-by: Edward O'Callaghan
Alex, as a matter of procedure could you perhaps have the actual version
bump as part of the final patch in a given series? This would mitigate
the issue of forgetting to bump the version for a given relevant changeset.
Kindly,
Edward.
On 03/17/2017 01:46 AM, Ale
> -Original Message-
> From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of
> Alex Deucher
> Sent: Thursday, March 16, 2017 22:47
> To: amd-gfx@lists.freedesktop.org
> Cc: Deucher, Alexander
> Subject: [PATCH] drm/amdgpu: bump version for PRT support
>
> Signed-off-by:
Hi Christian,
Thanks for your reminder.
I always keep it in mind.
Regards,
Jerry (Junwei Zhang)
Linux Base Graphics
SRDC Software Development
_
From: Koenig, Christian
Sent: Thursday, March 16, 2017 22:21
To: Zhang, Jerry; amd-gfx@lists.freedesktop.org
Subjec
On Thu, 2017-03-16 at 10:59 +0100, poma wrote:
> There are two applicable patches within Fedora:
> https://src.fedoraproject.org/cgit/rpms/xorg-x11-drv-ati.git/commit/fix-dri-removal.patch?id=d5f48e7d5b6c
An artifact of Fedora trying to not build DRI1. It's a fix in the wrong
place, xserver shoul
On Thu, Mar 16, 2017 at 2:25 PM, Adam Jackson wrote:
> On Thu, 2017-03-16 at 10:59 +0100, poma wrote:
>
>> There are two applicable patches within Fedora:
>> https://src.fedoraproject.org/cgit/rpms/xorg-x11-drv-ati.git/commit/fix-dri-removal.patch?id=d5f48e7d5b6c
>
> An artifact of Fedora trying t
On 2017-03-16 12:01 PM, Christian König wrote:
Am 16.03.2017 um 16:31 schrieb Andres Rodriguez:
On 2017-03-16 11:13 AM, Andres Rodriguez wrote:
On 2017-03-16 05:15 AM, zhoucm1 wrote:
On 2017年03月16日 17:10, Christian König wrote:
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high pr
On 16/03/17 12:24 PM, Alex Deucher wrote:
On Tue, Mar 14, 2017 at 3:32 PM, Alex Deucher wrote:
Even if we disable clockgating, we still need to make sure the
cp/rlc interrupts are enabled for powergating which might still
be enabled.
Signed-off-by: Alex Deucher
Anyone?
Haven't had a chanc
On Tue, Mar 14, 2017 at 3:32 PM, Alex Deucher wrote:
> Even if we disable clockgating, we still need to make sure the
> cp/rlc interrupts are enabled for powergating which might still
> be enabled.
>
> Signed-off-by: Alex Deucher
Anyone?
> ---
> drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 2 ++
>
Am 16.03.2017 um 16:31 schrieb Andres Rodriguez:
On 2017-03-16 11:13 AM, Andres Rodriguez wrote:
On 2017-03-16 05:15 AM, zhoucm1 wrote:
On 2017年03月16日 17:10, Christian König wrote:
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high priority rq is full, then process with low priority
On 2017-03-16 11:13 AM, Andres Rodriguez wrote:
On 2017-03-16 05:15 AM, zhoucm1 wrote:
On 2017年03月16日 17:10, Christian König wrote:
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high priority rq is full, then process with low priority could be
starve.
Add policy for this problem, the
On 2017-03-16 05:15 AM, zhoucm1 wrote:
On 2017年03月16日 17:10, Christian König wrote:
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high priority rq is full, then process with low priority could be
starve.
Add policy for this problem, the high proiority can ahead of next
priority queue,
th
From: David Panariti
KIQ is the Kernel Interface Queue for managing the MEC. Rather than setting
up rings via direct MMIO of ring registers, the rings are configured via
special packets sent to the KIQ. The allows the MEC to better manage shared
resources and certain power events.
Signed-off-b
Signed-off-by: Alex Deucher
---
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index c4dacd5..c75f48e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dr
Hi Jerry,
fell free to use this patch internally, but the upstream copy of
include/drm/amdgpu_drm.h should be updated when the kernel change land.
Emil usually points out how to do this correctly whenever we try to add
changes manually.
Christian.
Am 16.03.2017 um 15:05 schrieb Zhang, Jerr
Hi Christian,
May I get a RB for this patch?
Regards,
Jerry(Junwei Zhang)
From: amd-gfx on behalf of Junwei Zhang
Sent: Thursday, March 16, 2017 11:56
To: amd-gfx@lists.freedesktop.org
Cc: Zhang, Jerry
Subject: [PATCH 1/4] amdgpu: add new VA operations
Hi Michel,
that's really help me :)
thanks.
Regards,
Jerry(Junwei Zhang)
From: Michel Dänzer
Sent: Thursday, March 16, 2017 5:42:53 PM
To: Christian König; Zhang, Jerry
Cc: amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH 0/4] [libdrm] support PRT and r
Hi Christian,
Thanks for your info.
I missed them completely :(
the va op raw func looks similar as my va op2.
I will re-use them in hybrid libdrm and provide for UMD development.
Regards,
Jerry(Junwei Zhang)
From: Christian König
Sent: Thursday, March
On 16.03.2017 09:16, Michel Dänzer wrote:
>
> I'm pleased to announce the 7.9.0 release of xf86-video-ati, the Xorg
> driver for ATI/AMD Radeon GPUs supported by the radeon kernel driver.
> This release supports xserver versions 1.10-1.19.
>
> NOTE for packagers: Please ship the new 10-radeon.con
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high priority rq is full, then process with low priority could be starve.
Add policy for this problem, the high proiority can ahead of next priority
queue,
the ratio is 2 : 1.
Change-Id: I58f4a6b9cdce8689b18dd8e83dd6e2cf5f99d5fb
Signed-off-by: Ch
On 16/03/17 06:28 PM, Christian König wrote:
> Am 16.03.2017 um 10:14 schrieb Zhang, Jerry:
>>
>> So I prepare the patch 2 to use the input flags:
>> * [PATCH 2/4] amdgpu: set va flag with input parameter
>
> Nicolai handled both issues by creating a new function where the BO is
> optional an
Am 16.03.2017 um 10:14 schrieb Zhang, Jerry:
Hi Christian,
Thanks for your info.
Sorry to miss them...
Actually this a serial of messed function patch, I should send them separately
next time.
If we support PRT with bo=NULL, we may need a new VA op.
So please take a look at patch 4:
* [PA
Alright.
Regards,
Jerry (Junwei Zhang)
Linux Base Graphics
SRDC Software Development
_
> -Original Message-
> From: Christian König [mailto:deathsim...@vodafone.de]
> Sent: Thursday, March 16, 2017 17:27
> To: Zhang, Jerry; amd-gfx@lists.freedesktop.o
Am 16.03.2017 um 10:16 schrieb Zhang, Jerry:
Hi Christian,
Thanks for your review.
BTW, Have you pushed the previous patches in all-open branch?
I didn't find them. Or any concern?
No, unfortunately had a few very busy days this week and didn't found time.
Feel free to push them with your rb
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
big number is to high priority.
Change-Id: I1e94b3403d0cfd41a418d2bd741736cf7edc5d77
Signed-off-by: Chunming Zhou
Reviewed-by: Christian König
---
drivers/gpu/drm/amd/scheduler/gpu_scheduler.c | 4 ++--
drivers/gpu/drm/amd/scheduler/gpu_sche
On 2017年03月16日 17:10, Christian König wrote:
Am 16.03.2017 um 10:00 schrieb Chunming Zhou:
if high priority rq is full, then process with low priority could be
starve.
Add policy for this problem, the high proiority can ahead of next
priority queue,
the ratio is 2 : 1.
Change-Id: I58f4a6b9c
big number is to high priority.
Change-Id: I1e94b3403d0cfd41a418d2bd741736cf7edc5d77
Signed-off-by: Chunming Zhou
---
drivers/gpu/drm/amd/scheduler/gpu_scheduler.c | 4 ++--
drivers/gpu/drm/amd/scheduler/gpu_scheduler.h | 9 +
2 files changed, 7 insertions(+), 6 deletions(-)
diff --git
Am 16.03.2017 um 03:06 schrieb Alex Deucher:
GFX_MGLS was added twice.
Noticed-by: David Binderman
Signed-off-by: Alex Deucher
Reviewed-by: Christian König for both.
---
drivers/gpu/drm/amd/amdgpu/vi.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c
Hi Christian,
Thanks for your review.
BTW, Have you pushed the previous patches in all-open branch?
I didn't find them. Or any concern?
So we may use these patch on topic branch only for now.
Regards,
Jerry (Junwei Zhang)
Linux Base Graphics
SRDC Software Development
__
Hi Christian,
Thanks for your info.
Sorry to miss them...
Actually this a serial of messed function patch, I should send them separately
next time.
If we support PRT with bo=NULL, we may need a new VA op.
So please take a look at patch 4:
* [PATCH 4/4] amdgpu: add a new bo va function to sup
Am 16.03.2017 um 09:13 schrieb Junwei Zhang:
If the before mapping is 1 page size, so its start and last will be same.
Thus below condition will become false, then to free the before mapping.
> if (before->it.start != before->it.last)
But in this case, we need the before mapping of 1 page siz
if high priority rq is full, then process with low priority could be starve.
Add policy for this problem, the high proiority can ahead of next priority
queue,
the ratio is 2 : 1.
Change-Id: I58f4a6b9cdce8689b18dd8e83dd6e2cf5f99d5fb
Signed-off-by: Chunming Zhou
---
drivers/gpu/drm/amd/scheduler/
Andres added high priority queue for userspace, which got me to think more about
scheduling policy, if high priority queue is always full and busy, low priority
queue could starve.
Chunming Zhou (2):
drm/amd/sched: revise priority number
drm/amd/sched: add schuduling policy
drivers/gpu/drm
Am 16.03.2017 um 07:22 schrieb Ayyappa Ch:
Hello,
Can amdgpu driver supports if we enable CONFIG_PREEMPT in configuration?
It certainly should. If you find issues please open a bug report.
Christian.
Thanks,
Ayyappa
___
amd-gfx mailing list
amd-g
Nicolai already proposed patches for PRT support in libdrm (except the
CLEAR/REPLACE addition).
They look a bit cleaner and where posted first, I think we should use
them instead.
Christian.
Am 16.03.2017 um 04:56 schrieb Junwei Zhang:
* add PRT flag and new VA op
* fix flag setting for cur
I'm pleased to announce the 1.3.0 release of xf86-video-amdgpu, the Xorg
driver for AMD Radeon GPUs supported by the amdgpu kernel driver.
This release supports xserver versions 1.10-1.19.
Highlights:
* Allow TearFree to be toggled at runtime via an RandR output property
"TearFree". The xorg.c
> -Original Message-
> From: Christian König [mailto:deathsim...@vodafone.de]
> Sent: Thursday, March 16, 2017 15:59
> To: Zhang, Jerry; amd-gfx@lists.freedesktop.org
> Subject: Re: [PATCH 1/2] drm/amdgpu: insert partial mappings before and after
> directly
>
> Am 16.03.2017 um 08:46 schri
I'm pleased to announce the 7.9.0 release of xf86-video-ati, the Xorg
driver for ATI/AMD Radeon GPUs supported by the radeon kernel driver.
This release supports xserver versions 1.10-1.19.
NOTE for packagers: Please ship the new 10-radeon.conf file in the same
package as radeon_drv.so.
Highligh
If the before mapping is 1 page size, so its start and last will be same.
Thus below condition will become false, then to free the before mapping.
> if (before->it.start != before->it.last)
But in this case, we need the before mapping of 1 page size.
So does after mapping.
Signed-off-by: Junwei
> -Original Message-
> From: Christian König [mailto:deathsim...@vodafone.de]
> Sent: Thursday, March 16, 2017 15:33
> To: Zhang, Jerry; amd-gfx@lists.freedesktop.org
> Subject: Re: [PATCH 1/2] drm/amdgpu: insert partial mappings before and after
> directly
>
> Am 16.03.2017 um 04:44 schri
Am 16.03.2017 um 08:46 schrieb Zhang, Jerry:
-Original Message-
From: Christian König [mailto:deathsim...@vodafone.de]
Sent: Thursday, March 16, 2017 15:33
To: Zhang, Jerry; amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH 1/2] drm/amdgpu: insert partial mappings before and after
directl
Am 16.03.2017 um 02:34 schrieb Alex Deucher:
On Wed, Mar 15, 2017 at 5:05 PM, Andres Rodriguez wrote:
Hey Alex, Christian,
On a slightly unrelated note.
Have you also considered using system_highpri_wq instead of system_wq
for the delayed interrupt work?
There is a potential for multi-ms lat
Am 16.03.2017 um 04:44 schrieb Junwei Zhang:
Currently it may miss one page before or after the target mapping
I don't think that this will work correctly. The interval tree still
contains the old mapping at this point and as far as I know inserting
overlapping mappings is not allowed here.
52 matches
Mail list logo
