Since the PD BOs are assigned at initialization,
the placement stage can obtain the PD physical address
(PA) directly from the PD BOs without additional lookups.
Signed-off-by: Prike Liang
---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Loading and unloading the amdgpu module in TTY mode is successful at
first glance, but attempting to turn off the system after this action
results in a kernel panic. Sometimes, after trying to run load/unload
multiple times, the user may also encounter other issues. The easiest
way to reproduce thi
On 03.09.25 04:52, Liang, Prike wrote:
>> + ret = amdgpu_vm_update_pdes(adev, vm, false);
>> + if (ret)
>> + goto unlock_all;
>
> Do we need to create a sync point for syncing the VM PD/PT/PET update, and
> ensure
> all the updates done before attach the eviction fence?
Oh, g
[AMD Official Use Only - AMD Internal Distribution Only]
-Original Message-
From: Limonciello, Mario
Sent: Wednesday, September 3, 2025 2:43 PM
To: Nirujogi, Pratap ; amd-gfx@lists.freedesktop.org;
Deucher, Alexander ; Koenig, Christian
; Limonciello, Mario
Cc: Chan, Benjamin (Koon Pan
When we backup ring contents to reemit after a queue reset,
we don't backup ring contents from the bad context. When
we signal the fences, we should set an error on those
fences as well.
Fixes: 77cc0da39c7c ("drm/amdgpu: track ring state associated with a fence")
Signed-off-by: Alex Deucher
---
This patch adds robust reset handling for user queues (userq) to improve
recovery from queue failures. The key components include:
1. Queue detection and reset logic:
- amdgpu_userq_detect_and_reset_queues() identifies failed queues
- Per-IP detect_and_reset callbacks for targeted recovery
Normally resources are evicted on dGPUs at suspend or hibernate and
on APUs at hibernate. These steps are unnecessary when using the S4
callbacks to put the system into S5.
Cc: AceLan Kao
Cc: Kai-Heng Feng
Cc: Mark Pearson
Cc: Denis Benato
Cc: Merthan Karakaş
Tested-by: Eric Naim
Acked-by:
Add a UAPI flag indicating if gang submit is supported or not.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 4
include/uapi/drm/amdgpu_drm.h | 9 +
2 files changed, 9 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amd
On 9/4/2025 8:03 AM, Przemysław Kopa wrote:
I'm attaching two new dmesg logs.
Sorry, I attached wrong files to my previous message. The proper ones
are attached to this one.
I'm guessing what's going on is a race between the HDMI device going to
D3 through snd-hda-intel (which provides HDMI
PCI devices can be programmed as a wakeup source from low power states
by sysfs. However when using the S4 flow to go into S5 these wakeup
sources should be disabled to avoid what users would perceive as
spurious wakeup events.
Tested-by: Eric Naim
Signed-off-by: Mario Limonciello (AMD)
---
dr
From: "Mario Limonciello"
[Why]
dm_prepare_suspend() was added in commit 50e0bae34fa6b
("drm/amd/display: Add and use new dm_prepare_suspend() callback")
to allow display to turn off earlier in the suspend sequence.
This caused a regression that HDMI audio sometimes didn't work
properly after re
Applied. Thanks!
Alex
On Thu, Sep 4, 2025 at 8:54 AM Qianfeng Rong wrote:
>
> Use negative error code -EINVAL instead of positive EINVAL in the default
> case of svm_ioctl() to conform to Linux kernel error code conventions.
>
> Fixes: 42de677f7999 ("drm/amdkfd: register svm range")
> Signed-of
Move isp firmware load from generic amdgpu_isp driver to isp
version specific driver modules isp_v4_1_0 and isp_v4_1_1.
Signed-off-by: Pratap Nirujogi
---
drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c | 22 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h | 2 ++
drivers/gpu/drm/amd/amd
Devices with no subordinate should be put into D3 during hibernate, but
devices that have bridge_d3 set should also be put to sleep during
hibernate. Adjust the check in pci_pm_poweroff_noirq() to use
pci_power_manageable() to cover those as well.
Tested-by: Eric Naim
Signed-off-by: Mario Limonci
Applied. Thanks!
On Thu, Sep 4, 2025 at 3:29 AM Liao Yuanhong wrote:
>
> For ternary operators in the form of "a ? true : false" or
> "a ? false : true", if 'a' itself returns a boolean result, the ternary
> operator can be omitted. Remove redundant ternary operators to clean up the
> code.
>
>
Alex already send exactly that patch yesterday.
Please review that one instead.
Thanks,
Christian.
On 05.09.25 04:34, Lin.Cao wrote:
> commit 4220d2c7c41b ("drm/amdgpu: remove is_mes_queue flag") set
> ring->adev->ring[ring-idx] as NULL at the end of function amdgpu_ring_fini()
> which will caus
Applied the series. Thanks!
Alex
On Thu, Sep 4, 2025 at 7:49 AM Liao Yuanhong wrote:
>
> For ternary operators in the form of "a ? true : false", if 'a' itself
> returns a boolean result, the ternary operator can be omitted. Remove
> redundant ternary operators to clean up the code.
>
> Signed-
Ping ...
On 2025-08-22 15:38, James Zhu wrote:
This reverts commit cab1cec78c8fd52e014546739875a81150f11080.
migrate_vma_pages can fail if a CPU thread faults on the same page.
However, the page table is locked and only one of the new pages will
be inserted. The device driver will see that the
Gang submission means that the kernel driver guarantees that multiple
submissions are executed on the HW at the same time on different engines.
Background is that those submissions then depend on each other and each
can't finish stand alone.
SRIOV now uses world switch to preempt submissions on t
commit 4220d2c7c41b ("drm/amdgpu: remove is_mes_queue flag") set
ring->adev->ring[ring-idx] as NULL at the end of function amdgpu_ring_fini()
which will cause function amdgpu_fence_driver_sw_fini() skip
drm_sched_fini() and free fence_drv.fence then cause memory leak.
Remove set rings[ring->idx] a
Allow setting power cap for smu_v13_0_6 in 1vf mode
Signed-off-by: Asad Kamal
---
drivers/gpu/drm/amd/pm/amdgpu_pm.c | 7 ---
drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 6 +-
2 files changed, 9 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/
The copy_to_user() function returns the number of bytes that it wasn't
able to copy, but we should return -EFAULT to the user.
Fixes: 4d82724f7f2b ("drm/amdgpu: Add mapping info option for GEM_OP ioctl")
Fixes: f9db1fc52ceb ("drm/amdgpu: Add ioctl to get all gem handles for a
process")
Signed-off
Applied. Thanks!
On Fri, Sep 5, 2025 at 8:49 AM Wang, Yang(Kevin) wrote:
>
> [AMD Official Use Only - AMD Internal Distribution Only]
>
> Reviewed-by: Yang Wang
>
> Best Regards,
> Kevin
>
> -Original Message-
> From: John Olender
> Sent: Friday, September 5, 2025 6:11 PM
> To: Wang, Y
On Fri, Sep 5, 2025 at 1:41 PM Mario Limonciello
wrote:
>
> From: "Mario Limonciello"
>
> [Why]
> dm_prepare_suspend() was added in commit 50e0bae34fa6b
> ("drm/amd/display: Add and use new dm_prepare_suspend() callback")
> to allow display to turn off earlier in the suspend sequence.
>
> This ca
Disable dpcd probe quirk to native aux.
Cc: # 6.16.y: 5281cbe0b55a
Cc: # 6.16.y: 0b4aa85e8981
Cc: # 6.16.y: b87ed522b364
Cc: # 6.16.y
Signed-off-by: Fangzhi Zuo
Reviewed-by: Imre Deak
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 1 +
1 file changed, 1 insertion(+)
diff
In order to unify suspend and hibernate codepaths without code duplication
the common code should be in common helpers. Move it from
pci_pm_suspend_noirq() into a helper. No intended functional changes.
Tested-by: Eric Naim
Signed-off-by: Mario Limonciello (AMD)
---
drivers/pci/pci-driver.c |
26 matches
Mail list logo
