On Fri, Mar 16, 2018 at 11:07:08AM -0500, Alex Deucher wrote:
> Need to be able to query the VCN firmware version from
> userspace to determine supported features, etc.
>
Reviewed-by: Huang Rui
> Signed-off-by: Alex Deucher
> ---
>
On 2018年03月19日 18:03, Christian König wrote:
Am 19.03.2018 um 11:00 schrieb Chunming Zhou:
v2: add sanity checking
v3: make code open
Change-Id: I2cf672ad36b8b4cc1a6b2e704f786bf6a155d9ce
Signed-off-by: Chunming Zhou
Reviewed-by: Christian König
Hi Christian,
The messages prompted on timeout are Errors not just Warnings although we did
not see any real problem(for the dgemm special case). That's why we say it
confusing.
And i suppose you want a fix like my previous patch(see attachment).
Regards,
Evan
> -Original Message-
>
Thanks, John!
On Sat, Mar 17, 2018 at 4:17 PM, Bridgman, John wrote:
>
>>-Original Message-
>>From: Ming Yang [mailto:minos.fut...@gmail.com]
>>Sent: Saturday, March 17, 2018 12:35 PM
>>To: Kuehling, Felix; Bridgman, John
>>Cc: amd-gfx@lists.freedesktop.org
In theory, Mesa doesn't have to do anything. It can continue setting VRAM
and if the kernel has to put a display buffer into GTT, it doesn't matter
(for Mesa). Whether the VRAM placement is really used is largely determined
by BO priorities.
The way I understand scather/gather is that it only
Reviewed-by: Alex Deucher
From: amd-gfx on behalf of Andrey
Grodzovsky
Sent: Monday, March 19, 2018 4:55 PM
To: amd-gfx@lists.freedesktop.org
Cc: Grodzovsky, Andrey;
On Mon, Mar 19, 2018 at 4:29 PM, Li, Samuel wrote:
>>to my earlier point, there may be cases where it is advantageous to put
>> display buffers in vram even if s/g display is supported
>
> Agreed. That is also why the patch has the options to let user select where
> to put
Signed-off-by: Andrey Grodzovsky
---
drivers/gpu/drm/amd/powerplay/amd_powerplay.c | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
index
On Mon, Mar 19, 2018 at 4:53 PM, Alex Deucher wrote:
> On Mon, Mar 19, 2018 at 4:26 PM, Marek Olšák wrote:
> > When Mesa wants a buffer in VRAM, it always sets VRAM. It relies on BO
> move
> > throttling to prevent unnecessary BO moves.
> >
> > My
On Mon, Mar 19, 2018 at 4:26 PM, Marek Olšák wrote:
> When Mesa wants a buffer in VRAM, it always sets VRAM. It relies on BO move
> throttling to prevent unnecessary BO moves.
>
> My questions are:
> - what should Mesa do differently for tiny VRAM?
> - what is a tiny VRAM?
> -
>to my earlier point, there may be cases where it is advantageous to put
>display buffers in vram even if s/g display is supported
Agreed. That is also why the patch has the options to let user select where to
put display buffers.
As whether to put the option in Mesa or kernel, it seems the
When Mesa wants a buffer in VRAM, it always sets VRAM. It relies on BO move
throttling to prevent unnecessary BO moves.
My questions are:
- what should Mesa do differently for tiny VRAM?
- what is a tiny VRAM?
- if VRAM is tiny, which allocations should we put there?
Marek
On Mon, Mar 19, 2018
s/not/now/. I meant to say, "we have NOW excluded the possibility of ever
setting displays anywhere else without a kernel update".
Alex
From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of
Deucher, Alexander
Sent: Monday, March 19, 2018 4:13 PM
To: Li, Samuel
I'm not sure what you mean by the 3 scenarios. Generally userspace selects
what domains it wants a buffer to be in, vram, gtt, or both (don't care). I'd
rather not have the kernel second guess the UMDs if we can help it. I'd rather
leave the kernel for cases where we have to force things due
Hi Sam,
no that is seriously a no-go. The placement decision is done my
userspace, not the kernel.
The only time the kernel overrides that decision is a) hardware
requirements (for example old UVD on AGP systems) and b) memory pressure.
We are have actually pushed quite hard on avoiding
Alex,
I assume you are talking the three scenarios here, 1)VRAM, 2)GTT, 3)VRAM/GTT.
But kernel will need the decision too(amdgpufb). I think it shall be better to
do it in kernel, instead of different clients(mesa, ddx, kernel ...)
Regards,
Samuel Li
From: Deucher, Alexander
Sent: Monday,
My personal preference is still to plumb this through to mesa rather than
forcing it in the kernel.
Alex
From: amd-gfx on behalf of Li, Samuel
Sent: Monday, March 19, 2018 3:50:34 PM
To: Koenig,
Christian,
You misunderstood Alex’s comments,
>Regardless of which scenarios we need to support, I think we also need
>to really plumb this through to mesa however since user space is who
>ultimately requests the location. Overriding it in the kernel gets
>tricky and can lead to
Quoting Alex:
Regardless of which scenarios we need to support, I think we also need
to really plumb this through to mesa however since user space is who
ultimately requests the location. Overriding it in the kernel gets
tricky and can lead to ping-ponging as others have noted. Better to
have
Am 19.03.2018 um 20:39 schrieb Marek Olšák:
On Mon, Mar 19, 2018 at 3:27 PM, Christian König
> wrote:
I think that the consensus with Alex and me is that we should
avoid exactly that.
Overriding the preferred domain in the
Agreed.
>I think that the consensus with Alex and me is that we should avoid exactly
>that.
Christian, Alex’s concern is about ping-pong, not about the preferred domain.
Regards,
Samuel Li
From: Marek Olšák [mailto:mar...@gmail.com]
Sent: Monday, March 19, 2018 3:39 PM
To: Koenig, Christian
On Mon, Mar 19, 2018 at 3:27 PM, Christian König
wrote:
> I think that the consensus with Alex and me is that we should avoid
> exactly that.
>
> Overriding the preferred domain in the kernel is a no-go for that patch
> set, so please implement the discussed changes in
I think that the consensus with Alex and me is that we should avoid
exactly that.
Overriding the preferred domain in the kernel is a no-go for that patch
set, so please implement the discussed changes in Mesa.
Regards,
Christian.
Am 19.03.2018 um 20:22 schrieb Li, Samuel:
I agree with
Am 19.03.2018 um 19:27 schrieb Alex Deucher:
On Fri, Mar 16, 2018 at 4:19 PM, Alex Deucher wrote:
Add kernel doc for the functions in amdgpu_device.c
Signed-off-by: Alex Deucher
Ping?
Acked-by: Christian König
I agree with Marek/Michel: since kernel sets the domain before scanning out, it
shall update the preferred domain here too.
Regards,
Samuel Li
> -Original Message-
> From: Koenig, Christian
> Sent: Thursday, March 08, 2018 4:07 AM
> To: Michel Dänzer ; Li, Samuel
>
On 2018-03-19 12:39 PM, Christian König wrote:
> So coming back to this series once more.
>
> Patch #1, #3 are Reviewed-by: Christian König .
>
> Patch #2, #4 - #13 and #18-#19 are Acked-by: Christian König
> .
>
> Patch #14: What's the
Acked-by: Leo Liu
On 03/19/2018 02:26 PM, Alex Deucher wrote:
On Fri, Mar 16, 2018 at 12:07 PM, Alex Deucher wrote:
Need to be able to query the VCN firmware version from
userspace to determine supported features, etc.
Signed-off-by: Alex Deucher
4.9-stable review patch. If anyone has any objections, please let me know.
--
From: Christopher James Halse Rogers
[ Upstream commit a294043b2fbd8de69d161457ed0c7a4026bbfa5a ]
Any use of the framebuffer will migrate it to VRAM, which
3.18-stable review patch. If anyone has any objections, please let me know.
--
From: Christopher James Halse Rogers
[ Upstream commit a294043b2fbd8de69d161457ed0c7a4026bbfa5a ]
Any use of the framebuffer will migrate it to VRAM, which
4.9-stable review patch. If anyone has any objections, please let me know.
--
From: Christopher James Halse Rogers
[ Upstream commit 1769152ac64b0b07583f696b621624df2ca4c840 ]
Any use of the framebuffer will migrate it to VRAM, which
4.4-stable review patch. If anyone has any objections, please let me know.
--
From: Christopher James Halse Rogers
[ Upstream commit 1769152ac64b0b07583f696b621624df2ca4c840 ]
Any use of the framebuffer will migrate it to VRAM, which
On Fri, Mar 16, 2018 at 4:19 PM, Alex Deucher wrote:
> Add kernel doc for the functions in amdgpu_device.c
>
> Signed-off-by: Alex Deucher
Ping?
Alex
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 439
> -
>
On Fri, Mar 16, 2018 at 12:07 PM, Alex Deucher wrote:
> Need to be able to query the VCN firmware version from
> userspace to determine supported features, etc.
>
> Signed-off-by: Alex Deucher
Ping?
Alex
> ---
>
On 2018-03-19 01:45 PM, Harry Wentland wrote:
> We'd like to use them for reading DCN debug status.
>
> Signed-off-by: Harry Wentland
> ---
>
> See patch for where we use them http://git.amd.com:8080/#/c/137151/1
>
> Tony can comment on details.
>
Whoops, didn't mean
So coming back to this series once more.
Patch #1, #3 are Reviewed-by: Christian König .
Patch #2, #4 - #13 and #18-#19 are Acked-by: Christian König
.
Patch #14: What's the difference to setting vramlimit=$size_of_bar ?
Patch #15 & #20:
Am 19.03.2018 um 16:53 schrieb Chris Wilson:
Quoting Christian König (2018-03-16 14:22:32)
[snip, probably lost too must context]
This allows for full grown pipelining, e.g. the exporter can say I need
to move the buffer for some operation. Then let the move operation wait
for all existing
Quoting Christian König (2018-03-16 14:22:32)
[snip, probably lost too must context]
> This allows for full grown pipelining, e.g. the exporter can say I need
> to move the buffer for some operation. Then let the move operation wait
> for all existing fences in the reservation object and install
> -Original Message-
> From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf
> Of Vijendar Mukunda
> Sent: Monday, March 19, 2018 2:47 AM
> To: amd-gfx@lists.freedesktop.org
> Cc: Deucher, Alexander ; Mukunda,
> Vijendar
On Fri, Mar 16, 2018 at 02:20:44PM +0100, Christian König wrote:
> Hi everybody,
>
> since I've got positive feedback from Daniel I continued working on this
> approach.
>
> A few issues are still open:
> 1. Daniel suggested that I make the invalidate_mappings callback a parameter
> of
On Fri, Mar 16, 2018 at 02:20:45PM +0100, Christian König wrote:
> Each importer can now provide an invalidate_mappings callback.
>
> This allows the exporter to provide the mappings without the need to pin
> the backing store.
>
> v2: don't try to invalidate mappings when the callback is NULL,
On Mon, Mar 19, 2018 at 5:32 AM, Hawking Zhang wrote:
> PCIE_INDEX2/DATA2 pair will be used for smn register accessing since from
> vega.
> PCIE_INDEX/DATA pair should be reserved for smu
>
> Change-Id: Ie597d89001e706225521c94161d2b40443ec3c48
> Signed-off-by: Hawking
No one looked why Vulkan driver start crashing?
Thread 1 "vulkandriverque" received signal SIGSEGV, Segmentation fault.
0x in ?? ()
(gdb) thread apply all bt
Thread 3 (Thread 0x7fffef06c700 (LWP 15668)):
#0 0x76e4f82d in futex_wait_cancelable (private=,
expected=0,
On Mon, 19 Mar 2018, Christian König wrote:
> Mhm, actually that patch isn't correct. What we grab get here is the next
> entry, not the first one.
>
> We don't have an alias list_next_entry for list_first_entry?
As compared to the semantic patch I proposed earlier today, it would seem
that
Am 19.03.2018 um 12:06 schrieb Julia Lawall:
On Mon, 19 Mar 2018, Christian König wrote:
Mhm, actually that patch isn't correct. What we grab get here is the next
entry, not the first one.
We don't have an alias list_next_entry for list_first_entry?
As compared to the semantic patch I
If a per VM BO ends up in a allowed domain it never moves back into the
prefered domain.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 15 +--
1 file changed, 13 insertions(+), 2 deletions(-)
diff --git
Am 19.03.2018 um 11:00 schrieb Chunming Zhou:
issue:
Game F1 performance drops 13% when per vm bo is enabled.
root cause:
if some BOs are fallback to allowed domain, they will never be validated if no
eviction happens,
that means they always exist in allowed domain.
Fix:
maintain a per vm
Am 19.03.2018 um 11:00 schrieb Chunming Zhou:
v2: add sanity checking
v3: make code open
Change-Id: I2cf672ad36b8b4cc1a6b2e704f786bf6a155d9ce
Signed-off-by: Chunming Zhou
Reviewed-by: Christian König
---
issue:
Game F1 performance drops 13% when per vm bo is enabled.
root cause:
if some BOs are fallback to allowed domain, they will never be validated if no
eviction happens,
that means they always exist in allowed domain.
Fix:
maintain a per vm allowed domain BOs list, then try to validated them
v2: add sanity checking
v3: make code open
Change-Id: I2cf672ad36b8b4cc1a6b2e704f786bf6a155d9ce
Signed-off-by: Chunming Zhou
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 5 -
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 16
2 files changed, 12
Am 19.03.2018 um 09:19 schrieb Chunming Zhou:
v2: add sanity checking
Change-Id: I2cf672ad36b8b4cc1a6b2e704f786bf6a155d9ce
Signed-off-by: Chunming Zhou
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 5 -
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 15
Am 19.03.2018 um 07:08 schrieb Evan Quan:
Since under some heavy computing environment(dgemm test), it takes
the asic over 10+ seconds to finish the dispatched single job
which will trigger the timeout. It's quite confusing although it
does not seem to bring any real problems.
As a quick
Mhm, actually that patch isn't correct. What we grab get here is the
next entry, not the first one.
We don't have an alias list_next_entry for list_first_entry?
Regards,
Christian.
Am 18.03.2018 um 22:51 schrieb Arushi Singhal:
This patch replaces list_entry with list_first_entry as it makes
PCIE_INDEX2/DATA2 pair will be used for smn register accessing since from vega.
PCIE_INDEX/DATA pair should be reserved for smu
Change-Id: Ie597d89001e706225521c94161d2b40443ec3c48
Signed-off-by: Hawking Zhang
---
drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c | 4 ++--
1 file
issue:
Game F1 performance drops 13% when per vm bo is enabled.
root cause:
if some BOs are fallback to allowed domain, they will never be validated if no
eviction happens,
that means they always exist in allowed domain.
Fix:
maintain a per vm allowed domain BOs list, then try to validated them
v2: add sanity checking
Change-Id: I2cf672ad36b8b4cc1a6b2e704f786bf6a155d9ce
Signed-off-by: Chunming Zhou
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 5 -
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 15 +--
2 files changed, 13 insertions(+), 7
i2s instance value is passed as platform data to dwc driver.
this parameter will be useful to distinguish current i2s
instance value when multiple i2s controller instances are created.
Signed-off-by: Vijendar Mukunda
---
drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c | 5
Hi Alex,
INT_MAX is used instead of MAX_SCHEDULE_TIMEOUT(which we discussed in another
mail thread) since the amdgpu_lockup_timeout is with data type int.
Using MAX_SCHEDULE_TIMEOUT(data type:long) will get compile warnings.
Regards,
Evan
-Original Message-
From: Evan Quan
Since under some heavy computing environment(dgemm test), it takes
the asic over 10+ seconds to finish the dispatched single job
which will trigger the timeout. It's quite confusing although it
does not seem to bring any real problems.
As a quick workround, we choose to disable timeout when GPU
58 matches
Mail list logo