Am 05.08.24 um 09:34 schrieb Nikita Zhandarovich:
On 7/30/24 23:56, Christian König wrote:
Am 30.07.24 um 19:36 schrieb Nikita Zhandarovich:
On 7/29/24 11:12, Christian König wrote:
Am 29.07.24 um 20:04 schrieb Christian König:
Am 29.07.24 um 19:26 schrieb Nikita Zhandarovich:
Hi,
On 7/29
Am 05.08.24 um 16:01 schrieb Arunpravin Paneer Selvam:
We require this flag AMDGPU_GEM_CREATE_GFX12_DCC or any other
kernel level GFX12 DCC flag to differentiate the DCC buffers and other
pinned display buffers(which has TTM_PL_FLAG_CONTIGUOUS enabled).
That's a pretty bad idea, the DCC flag is
Am 06.08.24 um 18:00 schrieb David (Ming Qiang) Wu:
Add JPEG IB command parser to ensure registers
in the command are within the JPEG IP block.
You should probably re-order the patches so that the cleanup comes first
and then the new functionality.
Apart from that looks good to me.
Regards,
Am 19.08.24 um 09:21 schrieb Friedrich Vock:
In Vulkan, it is the application's responsibility to perform adequate
synchronization before a sparse unmap, replace or BO destroy operation.
This adds an option to AMDGPU_VA_OPs to disable redundant implicit sync
that happens on sparse unmap or replac
to wait another year for the fix to
reach users.
>However, I understand that back-porting UAPI is problematic
and I'll
>leave that decision up to the amdgpu maintainers. Michel
suggested that
>a new CTX_OP would make more sense if we wan
Am 08.08.24 um 19:48 schrieb Victor Skvortsov:
Register access from userspace should be blocked until
reset is complete.
Signed-off-by: Victor Skvortsov
---
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 44 ++---
1 file changed, 32 insertions(+), 12 deletions(-)
diff --git a
This patch tries to solve the basic problem we also need to sync to
the KFD fences of the BO because otherwise it can be that we clear
PTEs while the KFD queues are still running.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c | 30
drivers
Rework how VM operations synchronize to submissions. Provide an
amdgpu_sync container to the backends instead of an reservation
object and fill in the amdgpu_sync object in the higher layers
of the code.
No intended functional change, just prepares for upcomming changes.
Signed-off-by: Christian
Requested by both Bas and Friedrich. Mapping PTEs as PRT doesn't need to
sync for anything.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 +-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
b/driver
Am 15.08.24 um 02:04 schrieb Alex Deucher:
From: Srinivasan Shanmugam
The cleaner shader is used by the CP firmware to clean LDS and GPRs
between processes on the CUs.
This adds an internal API for GFX IP code to allocate and initialize the
cleaner shader.
Cc: Christian König
Cc: Alex
butes.
v2: fix error handling (Alex)
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 134
1 file changed, 134 insertions(+)
diff --git a/drivers/gpu/drm/amd/a
.
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
Signed-off-by: Alex Deucher
Suggested-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
b/drivers/gpu/drm
Reviewed-by: Christian König for patches #1,
#3-#5,
Acked-by: Christian König for patches #6,
#8-#10, #12, #13, #14, #17.
Some nit picks in patches #2, use of // for comments in patch #11 and
question on patch #15/#16
Really big bug in patch #7, that needs to be addressed.
Regards
Am 21.08.24 um 22:01 schrieb Felix Kuehling:
On 2024-08-21 08:03, Christian König wrote:
This patch tries to solve the basic problem we also need to sync to
the KFD fences of the BO because otherwise it can be that we clear
PTEs while the KFD queues are still running.
This is going to trigger
Link: https://www.spinics.net/lists/dri-devel/msg457005.html
Signed-off-by: Alex Deucher
Cc: Zhu Lingshan
Cc: Christian König
Reviewed-by: Christian König
---
include/drm/ttm/ttm_bo.h | 4
1 file changed, 4 deletions(-)
diff --git a/include/drm/ttm/ttm_bo.h b/include/drm/ttm/ttm_bo.h
me of the bootup time constrains we have.
In addition this fixes the high CPU spikes seen when doing
wipe on release.
Mhm, interesting.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3528
Reviewed-by: Michel Dänzer
Signed-off-by: Alex Deucher
Cc: Arunpravin Paneer Selvam
Cc: Chri
We haven't used the functionality to pin BOs in a certain range at all
while the driver existed. Just nuke it.
Signed-off-by: Christian König
---
.../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c| 56 ++-
drivers/gpu/dr
recoverable in any way when VRAM is
lost.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 4 -
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 87 +
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 67 +---
drivers/gpu/drm/amd/amdgpu
Instead of having that in the amdgpu_bo_pin() function applied for all
pinned BOs.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 1 -
drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
Am 30.07.24 um 06:36 schrieb Yifan Zhang:
avoid using SDMA if it is unavailable.
That is clearly not something we should do. The SDMA is mandatory for
the driver to work.
We could do this for emulation and bringup, but in those cases the
engineer should probably enabled CPU based updates ma
23335f9577e0b509c20ad8d65d9fdedd14545b55.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 3464a7a880f0..f0ccc560fd5c 100644
--- a/drivers/gpu/drm/amd/amdgpu
Am 29.08.24 um 00:40 schrieb Felix Kuehling:
On 2024-08-22 05:07, Christian König wrote:
Am 21.08.24 um 22:01 schrieb Felix Kuehling:
On 2024-08-21 08:03, Christian König wrote:
This patch tries to solve the basic problem we also need to sync to
the KFD fences of the BO because otherwise it
Am 30.08.24 um 03:22 schrieb Li Zetao:
When it needs to get a value within a certain interval, using clamp()
makes the code easier to understand than min(max()).
Signed-off-by: Li Zetao
This patch and #1 is a nice cleanup and Reviewed-by: Christian König
But as Alex also pointed out
like "CPU based updates doesn't produce a fence."
With that done Reviewed-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
b/drivers/gpu/drm/amd/amdgpu/
ger
is finilized, drain the workqueue in ttm_device.
Fixes:d99fbd9aab62 ("drm/ttm: Always take the bo delayed cleanup path for imported
bos")
Acked-by: Christian König
The correct tag is Suggested-by.
Signed-off-by: Asher Song
---
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 1 +
1
t;drm/amdgpu: Add sysfs interface for running cleaner
shader")
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
Suggested-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 32 -
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h | 1 +
ger is
finilized, drain the workqueue in ttm_device.
v2: move drain_workqueue to amdgpu_ttm.c
Fixes:d99fbd9aab62 ("drm/ttm: Always take the bo delayed cleanup path for imported
bos")
Suggested-by: Christian König
Signed-off-by: Asher Song
Reviewed-by: Christian König
---
dri
t;drm/amdgpu: Add sysfs interface for running cleaner
shader")
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
Suggested-by: Christian König
Reviewed-by: Christian König
---
v2:
- Dropped default assignment of f. (Christian)
- Moved drm_sched entity; to
shader uses two kernels launched
separately to clean VGPRs, LDS, and lower SGPRs, and to clean remaining
SGPRs.
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
Acked-by: Christian König
---
.../amd/amdgpu/gfx_v9_4_3_cleaner_shader.asm | 139 +-
1
, and lower SGPRs, and to clean remaining
SGPRs.
Fixes: 3b721dfb2c95 ("drm/amdgpu/gfx9: Add cleaner shader for GFX9.4.3")
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
---
v2:
- Corrected typo for iteraions
- Added fixes tag
.../amd/amdgpu/gfx_v9_4_3_cle
Am 02.09.24 um 09:34 schrieb Lijo Lazar:
There are case where a device needs to be reset first before it is fully
initialized. An example case is a driver reinstallation with a different version
of PSP TOS. In such a case, if a device supports reset in which PSP TOS is
unloaded, then driver needs
member 'ts' not described in 'amdgpu_vm_handle_fault'
Cc: Xiaogang.Chen
Cc: Christian König
Cc: Alex Deucher
Reported-by: kernel test robot
Closes:
https://lore.kernel.org/oe-kbuild-all/202408251419.vgzhg3gv-...@intel.com/
Signed-off-by: Srinivasan Shanmugam
Reviewed-b
u.c:50: warning: Excess function
parameter 'sync_mode' description in 'amdgpu_vm_cpu_prepare'
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions
L.
V9: (Christian)
- Store all the fence_drv ref to other drivers and not ourself.
- Remove the userq fence xa implementation and replace with
kvmalloc_array.
Signed-off-by: Arunpravin Paneer Selvam
Suggested-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h
call within the xa_lock
as there is a chance that fence_drv might be freed.
Signed-off-by: Arunpravin Paneer Selvam
Acked-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c| 2 ++
.../gpu/drm/amd/amdgpu/amdgpu_userq_fence.c | 6 +
drivers/gpu/drm/amd/a
Am 05.09.24 um 10:58 schrieb Dejan Andjelkovic:
When using SDMA for PT updates, a TLB fence hooked to a buffer
objects dma resv object with usage declared as BOOKKEEP leaves a
chance for TLB flush to occur prematurely. This will lead to a page
fault. Raising the usage from BOOKKEEP to KERNEL remo
Am 30.08.24 um 20:43 schrieb Arunpravin Paneer Selvam:
Screen freeze and userq fence driver crash while playing Xonotic
Signed-off-by: Arunpravin Paneer Selvam
---
drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c | 12 +++-
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a
Am 30.08.24 um 20:43 schrieb Arunpravin Paneer Selvam:
Remove MES self test as this conflicts the userqueue fence
interrupts.
v2:(Christian)
- remove the amdgpu_mes_self_test() function and any now unused code.
Signed-off-by: Arunpravin Paneer Selvam
Acked-by: Christian König
Am 30.08.24 um 20:43 schrieb Arunpravin Paneer Selvam:
Add few optimizations to userq fence driver.
v1:(Christian):
- Remove unnecessary comments.
- In drm_exec_init call give num_bo_handles as last parameter it would
making allocation of the array more efficient
- Handle return va
Well that explanation doesn't seem to make much sense either.
What do you mean with TLB flush is occurring prematurely?
Regards,
Christian.
Am 05.09.24 um 14:38 schrieb Andjelkovic, Dejan:
[AMD Official Use Only - AMD Internal Distribution Only]
Hi there. We're running into a page fault iss
Am 19.08.24 um 19:59 schrieb Jonathan Kim:
Currently multiple partitions will incorrectly overwrite the VM lookup
table since the table is indexed by PASID and multiple partitions can
register different VM objects on the same PASID.
That's a rather bad idea. Why do we have the same PASID for di
Well that's the whole reason I'm asking :)
Why do you think it should be added as dependency in
amdgpu_vm_sdma_update? As far as I can see that is complete nonsense.
Page table updates never depend on TLB flushes, it's the TLB flush which
depends on the page table update.
Regards,
Christian
ting the two.
Then audit the code if all users of rq and priority actually hold the
correct locks while reading and writing them.
Regards,
Christian.
P.
Signed-off-by: Tvrtko Ursulin
Fixes: b37aced31eb0 ("drm/scheduler: implement a function to modify
sched list")
Cc: Christ
: b37aced31eb0 ("drm/scheduler: implement a function to modify sched list")
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: David Airlie
Cc: Daniel Vetter
Cc: dri-de...@lists.freedesktop.org
Cc: # v5.7+
---
drivers/gpu/drm/scheduler/sched_entity.c | 8 ++-
ueues")
Good catch, that might explain some of the odd behavior we have seen for
load balancing.
Reviewed-by: Christian König
Cc: Nirmoy Das
Cc: Christian König
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: David Airlie
Cc: Daniel Vetter
Cc: dri-de...@lists.freedesktop.org
C
Am 09.09.24 um 14:13 schrieb Philipp Stanner:
On Mon, 2024-09-09 at 13:29 +0200, Christian König wrote:
Am 09.09.24 um 11:44 schrieb Philipp Stanner:
On Fri, 2024-09-06 at 19:06 +0100, Tvrtko Ursulin wrote:
From: Tvrtko Ursulin
Without the locking amdgpu currently can race
nt different hw priorities.
Christian.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 4
1 file changed, 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
ind
tko
Cheers,
P.
Regards,
Tvrtko
[...]
P.
Then audit the code if all users of rq and priority actually
hold the
correct locks while reading and writing them.
Regards,
Christian.
P.
Signed-off-by: Tvrtko Ursulin
Fixes: b37aced31eb0 ("drm/scheduler: implement a function
Am 09.09.24 um 18:02 schrieb Kim, Jonathan:
[Public]
-Original Message-
From: Christian König
Sent: Thursday, September 5, 2024 10:24 AM
To: Kim, Jonathan ; amd-gfx@lists.freedesktop.org
Cc: Kuehling, Felix ; Deucher, Alexander
; Joshi, Mukul
Subject: Re: [PATCH] drm/amdkfd: fix vm
and entity->num_sched_list pair.
v2:
* Improve commit message. (Philipp)
Signed-off-by: Tvrtko Ursulin
Fixes: b37aced31eb0 ("drm/scheduler: implement a function to modify sched list")
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: David Airlie
Cc: D
ent a function to modify sched list")
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: David Airlie
Cc: Daniel Vetter
Cc: Philipp Stanner
Cc: dri-de...@lists.freedesktop.org
Cc: # v5.7+
---
drivers/gpu/drm/scheduler/sched_entity.c | 10 --
1 file
Am 09.09.24 um 19:19 schrieb Tvrtko Ursulin:
From: Tvrtko Ursulin
In FIFO mode We can avoid dropping the lock only to immediately re-acquire
by adding a new drm_sched_rq_update_fifo_locked() helper.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc
to restart from the first entity, instead of continuing in RR mode from
where FIFO left it, and that sounds completely fine.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: Philipp Stanner
Acked-by: Christian König
---
drive
Am 09.09.24 um 19:19 schrieb Tvrtko Ursulin:
From: Tvrtko Ursulin
Lets re-order the members to make it clear which are protected by the lock
and at the same time document it via kerneldoc.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew
: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: Philipp Stanner
Reviewed-by: Christian König
---
drivers/gpu/drm/scheduler/sched_entity.c | 24
drivers/gpu/drm/scheduler/sched_main.c | 6 +++---
include/drm/gpu_scheduler.h
same expectation to
drm_sched_rq_update_fifo_locked().
For more stream-lining we also add the run-queue as an explicit parameter
to drm_sched_rq_remove_fifo_locked() to avoid both callers and callee
having to dereference entity->rq.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: A
Am 10.09.24 um 11:46 schrieb Tvrtko Ursulin:
On 10/09/2024 10:08, Christian König wrote:
Am 09.09.24 um 19:19 schrieb Tvrtko Ursulin:
From: Tvrtko Ursulin
Having removed one re-lock cycle on the entity->lock in a patch titled
"drm/sched: Optimise drm_sched_entity_push_job", wit
e add "without holding the appropriate lock".
Christian.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: Philipp Stanner
---
drivers/gpu/drm/scheduler/sched_entity.c | 7 ++--
drivers/gpu/d
Am 24.06.24 um 16:08 schrieb Yunxiang Li:
Before, every time fdinfo is queried we try to lock all the BOs in the
VM and calculate memory usage from scratch. This works okay if the
fdinfo is rarely read and the VMs don't have a ton of BOs. If either of
these conditions is not true, we get a massiv
Am 10.09.24 um 19:40 schrieb Li, Yunxiang (Teddy):
[Public]
Ok that looks extremely ugly. Please just add a separate function and call that
from the TTM move function.
Should I still remove the adev argument? It is never used and causes a few call
sites having to find an adev unnecessarily.
#x27; with 'disable' in the comment &
replace 'Initiailize' with 'Initialize' in the comment &
replace 'disble' with 'disable' in the comment &
replace 'Disbale' with 'Disable' in the comment &
replace 'enogh'
Am 11.09.24 um 07:00 schrieb Paneer Selvam, Arunpravin:
Hi Christian,
On 9/5/2024 4:50 PM, Christian König wrote:
Am 30.08.24 um 20:43 schrieb Arunpravin Paneer Selvam:
Screen freeze and userq fence driver crash while playing Xonotic
Signed-off-by: Arunpravin Paneer Selvam
---
drivers
Li
Reviewed-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c | 6 --
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 11 +--
drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 14 --
3 files changed, 5 insertions(+), 26 deletions(-)
diff --git a/drivers
Am 11.09.24 um 17:13 schrieb Yunxiang Li:
amdgpu_vm_bo_invalidate doesn't use the adev parameter and not all
callers have a reference to adev handy, so remove it for cleanliness.
Signed-off-by: Yunxiang Li
Reviewed-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc: Matthew Brost
Cc: Philipp Stanner
Reviewed-by: Christian König
---
drivers/gpu/drm/scheduler/sched_entity.c | 13 +
drivers/gpu/drm/scheduler/sched_main.c | 6 +++---
include/drm/gpu_scheduler.h
_update_fifo_locked().
Finally, to align drm_sched_rq_update_fifo_locked(),
drm_sched_rq_add_entity_locked() and
drm_sched_rq_remove_fifo_locked() function signatures, we add rq as a
parameter to the latter.
Signed-off-by: Tvrtko Ursulin
Cc: Christian König
Cc: Alex Deucher
Cc: Luben Tuikov
Cc
V8: Rebase
V9: Changed the function names from gfx_v11* to mes_v11*
V10: Remove unused adev (Harish)
Cc: Alex Deucher
Cc: Christian Koenig
Signed-off-by: Shashank Sharma
Signed-off-by: Arvind Yadav
Reviewed-by: Christian König
---
.../gpu/drm/amd/amdgpu/mes_v11_0_userqueue.c |
Am 09.09.24 um 22:06 schrieb Shashank Sharma:
This patch adds basic eviction fence framework for the gfx buffers.
The idea is to:
- One eviction fence is created per gfx process, at kms_open.
- This fence is attached to all the gem buffers created
by this process.
- This fence is detached to a
("drm/amdgpu: Fix with right return code '-EIO' in
'amdgpu_gmc_vram_checking()'")
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Srinivasan Shanmugam
Reviewed-by: Christian König
---
v3:
- Changed from 'return r;' to 'return 0' (Christian)
drivers/gpu/d
Am 30.01.24 um 12:16 schrieb Daniel Vetter:
On Tue, Jan 30, 2024 at 12:10:31PM +0100, Daniel Vetter wrote:
On Mon, Jan 29, 2024 at 06:31:19PM +0800, Julia Zhang wrote:
As vram objects don't have backing pages and thus can't implement
drm_gem_object_funcs.get_sg_table callback. This removes drm
virtual address mappings as
seen in /proc/sys/vm/mmap_min_addr.
Signed-off-by: Felix Kuehling
Reviewed-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
b/drivers
Am 30.01.24 um 21:08 schrieb Felix Kuehling:
The TBA and TMA, along with an unused IB allocation, reside at low
addresses in the VM address space. A stray VM fault which hits these
pages must be serviced by making their page table entries invalid.
The scheduler depends upon these pages being r
Am 31.01.24 um 11:20 schrieb Zhang, Julia:
On 2024/1/30 22:23, Christian König wrote:
Am 30.01.24 um 12:16 schrieb Daniel Vetter:
On Tue, Jan 30, 2024 at 12:10:31PM +0100, Daniel Vetter wrote:
[SNIP]
Hi Sima, Christian,
Yeah, that is really just speculative. All importers need to set the
structure to save vm
The idea is to delay freeing of page table objects until we have the
respective TLB entries flushed.
V2: rebase
Cc: Christian König
Cc: Alex Deucher
Cc: Felix Kuehling
Cc: Rajneesh Bhardwaj
Signed-off-by: Shashank Sharma
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
Am 11.01.24 um 16:58 schrieb Alex Deucher:
Missing space.
Signed-off-by: Alex Deucher
Reviewed-by: Christian König
And sorry that this took so long. I'm still trying to catch up to my mails.
Christian.
---
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 +-
1 file changed, 1 inse
. :P
Thanks
- Joshie 🐸✨
On 1/23/24 12:49, Christian König wrote:
Am 23.01.24 um 12:35 schrieb Friedrich Vock:
On 23.01.24 10:36, Christian König wrote:
Am 22.01.24 um 23:39 schrieb Joshua Ashton:
[SNIP]
Most work submissions in practice submit more waves than the
number of
wave slots
Am 02.02.24 um 23:25 schrieb Hamza Mahfooz:
Currently, drivers have no mechanism to block requests to unbind
devices. However, this can cause resource leaks and leave the device in
an inconsistent state, such that rebinding the device may cause a hang
or otherwise prevent the device from being re
Am 04.02.24 um 07:12 schrieb Emily Deng:
When a vf has been reset, the pf wants to get notification to remove the vf
out of schedule.
Solution:
Add the callback function in pci_driver sriov_vf_reset_notification. When
vf reset happens, then call this callback function.
Well that doesn't make m
ll.
sorry, but please, just fix your driver.
fwiw Christian König from amd already rejected this too, I have no idea
why this was submitted
Well that was my fault.
I commented on an internal bug tracker that when sysfs bind/undbind is a
different code path from PCI remove/re-scan we could try
Am 01.02.24 um 17:50 schrieb Philip Yang:
SVM migration unmap pages from GPU and then update mapping to GPU to
recover page fault. Currently unmap clears the PDE entry for range
length >= huge page and free PTB bo, update mapping to alloc new PT bo.
There is race bug that the freed entry bo maybe
Am 08.02.24 um 06:52 schrieb Mario Limonciello:
amdgpu_acpi_is_s0ix_active() and amdgpu_acpi_is_s0ix_active() aren't
needed to be checked multiple times in a suspend cycle. Checking and
setting up policy one time in the prepare() callback is sufficient.
Mhm, looking at amdgpu_acpi_is_s3_active
allocation")
Signed-off-by: Arunpravin Paneer Selvam
Tested-by: Mario Limonciello
Acked-by: Christian König
CC: stable.. ?
---
drivers/gpu/drm/drm_buddy.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c
index f57e6d74fb0
Am 08.02.24 um 16:04 schrieb Mario Limonciello:
On 2/8/2024 00:54, Christian König wrote:
Am 08.02.24 um 06:52 schrieb Mario Limonciello:
amdgpu_acpi_is_s0ix_active() and amdgpu_acpi_is_s0ix_active() aren't
needed to be checked multiple times in a suspend cycle. Checking and
setti
changed the patch authorship here.
Going to fix this if I get tasked with pushing this to drm-misc-fixes.
But I still have hope that Arun will figure out how to do this himself.
Christian.
Cc: Arunpravin Paneer Selvam
Cc: Limonciello
Cc: Christian König
Signed-off-by: Arunpravin Paneer
Am 13.02.24 um 16:50 schrieb Pierre-Eric Pelloux-Prayer:
This new event can be used to trace where a given dma_fence is added
as a dependency of some other work.
I plan to use it in amdgpu.
Signed-off-by: Pierre-Eric Pelloux-Prayer
---
drivers/dma-buf/dma-fence.c | 1 +
include/trace/
Am 13.02.24 um 16:50 schrieb Pierre-Eric Pelloux-Prayer:
Useful to identify why sdma jobs are submitted.
Signed-off-by: Pierre-Eric Pelloux-Prayer
---
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h | 16
2 files changed, 18 in
Am 13.02.24 um 16:50 schrieb Pierre-Eric Pelloux-Prayer:
amdgpu_cs_ioctl already exists but serves a different
purpose.
amdgpu_cs_ioctl2 marks the beginning of the kernel processing of
the ioctl which is useful for tools to map which events belong to
the same submission (without this, the first
Am 13.02.24 um 16:50 schrieb Pierre-Eric Pelloux-Prayer:
This makes it possible to understand the dependencies between jobs.
Possible usage of this trace:
* stuttering issues like Mesa !9189
* incorrect synchronization: I don't have a link for this one, but having
these events was very usef
Am 14.02.24 um 17:38 schrieb Pierre-Eric Pelloux-Prayer:
Le 14/02/2024 à 13:09, Christian König a écrit :
Am 13.02.24 um 16:50 schrieb Pierre-Eric Pelloux-Prayer:
amdgpu_cs_ioctl already exists but serves a different
purpose.
amdgpu_cs_ioctl2 marks the beginning of the kernel processing of
That looks like an incorrect error handling to me.
The invalid address is rejected and because of this we free up the data
structures, but probably in the wrong order or something like that.
Going to take a look.
Thanks,
Christian.
Am 15.02.24 um 03:34 schrieb 정준교:
Hello,
We would like to
Well using this is in sysfs is a bug to begin with. This would prevent
starting new applications and crashing applications which don't expect
to get an -EPERM in return here.
If we need to make operations mutual exclusive with resets then we need
to take the appropriate locks and *not* work ar
istency.
v4: switch drm_gem_object_is_shared_for_memory_stats() to an inline function
I'm still not sure if looking at the actual handle count is the right
approach, but it's certainly better than before.
So Reviewed-by: Christian König for the
entire series.
Should I take this through drm-misc-next?
Regards,
: Christian Koenig
Signed-off-by: Jay Cornwall
Signed-off-by: Felix Kuehling
Can't full judge the KFD VI changes, but the rest looks good to me.
Reviewed-by: Christian König
Regards,
Christian.
---
drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c | 3 +-
drivers/gpu/drm/amd/a
Am 15.02.24 um 15:36 schrieb Alex Deucher:
On Thu, Feb 15, 2024 at 2:53 AM Christian König
wrote:
Well using this is in sysfs is a bug to begin with. This would prevent
starting new applications and crashing applications which don't expect
to get an -EPERM in return here.
If we need to
Am 15.02.24 um 15:43 schrieb Christian König:
Am 15.02.24 um 15:36 schrieb Alex Deucher:
On Thu, Feb 15, 2024 at 2:53 AM Christian König
wrote:
Well using this is in sysfs is a bug to begin with. This would prevent
starting new applications and crashing applications which don't expe
drm/amd/amdgpu/ih_v7_0.c:458: warning: Function parameter or member
'ih' not described in 'ih_v7_0_set_rptr'
Fixes: b6ba7a165b13 ("drm/amdgpu: Add ih v7_0 ip block support")
Cc: Likun Gao
Cc: Hawking Zhang
Cc: Christian König
Cc: Alex Deucher
Signed-off-by: Sriniva
Can you bisect where exactly between 6.6.0 and 6.7.4 the problems started?
Thanks,
Christian.
Am 15.02.24 um 16:59 schrieb Michael Zimmermann:
I have a Framework 13 with a 7840U and started having massive GPU
driver issues a few weeks ago (including system freezes).
Unfortunately the informatio
Am 02.02.24 um 16:28 schrieb Hamza Mahfooz:
We want programs besides the compositor to be able to enable or disable
panel power saving features.
Well I don't know the full background, but that is usually a no-go.
However, since they are currently only
configurable through DRM properties, that
Am 16.02.24 um 12:46 schrieb Arunpravin Paneer Selvam:
On 2/16/2024 4:41 PM, Matthew Auld wrote:
On 16/02/2024 10:00, Arunpravin Paneer Selvam wrote:
Remove the duplicate list_splice_tail call when the
total_allocated < size condition is true.
Cc: # 6.7+
Fixes: 8746c6c9dfa3 ("drm/buddy:
boot_time_tmr flag
drm/amdgpu: add psp_timeout to limit PSP related operation
drm/amdgpu: support psp ip block discovery for psp v14
Acked-by: Christian König for the series.
drivers/gpu/drm/amd/amdgpu/Makefile | 3 +-
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1
1 - 100 of 10978 matches
Mail list logo
