Documentation/process/deprecated.rst recommends against the use of kmalloc
with dynamic size calculations due to the risk of overflow and smaller
allocation being made than the caller was expecting. This could lead to
buffer overflow in code similar to the memcpy in
amdgpu_dm_plane_add_modifier().
Avoid constant register reloads while emitting IBs by using a local write
pointer and only updating the size at the end of each helper.
Signed-off-by: Tvrtko Ursulin
---
drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 108 ++-
1 file changed, 65 insertions(+), 43 deletions(-)
dif
smu_hw_init() and smu_hw_fini() call smu_dpm_set_vpe_enable() for
APUs as part of startup and teardown. These calls however are
not necessary because vpe_hw_init()/vpe_hw_fini() will call at
init/fini:
```
vpe_hw_init() / vpe_hw_fini()
amdgpu_device_ip_set_powergating_state()
vpe_set_powerg
If the drm device is used for displaying the console then writing
to the console could potentially cause a hang on the system if DCN
is in IPS. This is more likely to happen after commit 6ea9a1781c70a
("Flush console log from kernel_power_off()") which has caused the
console to get flushed at shutd
On Fri, Sep 12, 2025 at 10:06 AM Christian König
wrote:
>
> On 12.09.25 15:02, Alex Deucher wrote:
> > On Fri, Sep 12, 2025 at 7:17 AM Christian König
> > wrote:
> >>
> >> On 12.09.25 11:31, Jesse.Zhang wrote:
> >>> The current user queue implementation uses a dual-mutex scheme with
> >>> both pe
On 12.09.25 15:02, Alex Deucher wrote:
> On Fri, Sep 12, 2025 at 7:17 AM Christian König
> wrote:
>>
>> On 12.09.25 11:31, Jesse.Zhang wrote:
>>> The current user queue implementation uses a dual-mutex scheme with
>>> both per-device (adev->userq_mutex) and per-process (uq_mgr->userq_mutex)
>>> lo
Hi Alex,
On 9. Sep 2025, at 17:35, Alex Deucher wrote:
> Applied. Thanks!
>
> On Tue, Sep 9, 2025 at 11:29 AM Thorsten Blum wrote:
>>
>> Replace kmalloc() followed by copy_from_user() with memdup_user() to
>> improve and simplify kfd_criu_restore_queue().
>>
>> No functional changes intended.
[Public]
late_init is the right place to apply any late WA (after that there is no other
opportunity). There are already multiple callbacks done during hw_init. Any
other SOC specific thing can be done during one of them.
Thanks,
Lijo
-Original Message-
From: Wang, Yang(Kevin)
Sent: Fr
[AMD Official Use Only - AMD Internal Distribution Only]
Series is
Reviewed-by: Asad Kamal
Thanks & Regards
Asad
-Original Message-
From: Lazar, Lijo
Sent: Tuesday, September 9, 2025 11:18 AM
To: amd-gfx@lists.freedesktop.org
Cc: Zhang, Hawking ; Deucher, Alexander
; Kamal, Asad
Sub
On 12.09.25 11:31, Jesse.Zhang wrote:
> The current user queue implementation uses a dual-mutex scheme with
> both per-device (adev->userq_mutex) and per-process (uq_mgr->userq_mutex)
> locking. This overcomplicated design creates potential deadlock scenarios
> and makes the code harder to maintain
MMIO_REMAP (HDP flush page) exposes a hardware MMIO register window via a PCI
BAR.
Handle it as fixed I/O:
- map(): if MMIO_REMAP, require P2P, compute the BAR address (bus_addr + page
offset), and build a 1-entry sg_table with dma_map_resource().
- unmap(): if MMIO_REMAP, call dma_unmap_resour
MMIO_REMAP (HDP flush page) is a hardware I/O window exposed via a PCI
BAR. It must not migrate or be evicted.
Allocate a single 4 KB GEM BO in AMDGPU_GEM_DOMAIN_MMIO_REMAP during TTM
initialization when the hardware exposes a remap bus address and the
host page size is <= 4 KiB. Reserve the BO a
[Public]
>> There needs to be a specific callback to do SOC specific things during
>> late_init stage. Presently post_init is providing that, I don't think that
>> should be moved.
@post_init: Helper function for asic specific workarounds.
In the previous code, the "post_init" was mainly used
On Wed, 2025-09-03 at 11:18 +0100, Tvrtko Ursulin wrote:
> Move the code dealing with entities entering and exiting run queues to
> helpers to logically separate it from jobs entering and exiting entities.
Sorry if I've asked this before, but does this strictly depend on the
preceding patches or c
Hi,
On 9/11/25 12:55 AM, Ilpo Järvinen wrote:
> Fix the copy-pasted errors in the Resizable BAR handling functions
> kernel doc and generally improve wording choices.
>
> Fix the formatting errors of the Return: line.
>
> Signed-off-by: Ilpo Järvinen
> ---
> drivers/pci/rebar.c | 29 ++
On 01.09.25 12:10, Borislav Petkov wrote:
> On Mon, Sep 01, 2025 at 11:27:01AM +0200, Michel Dänzer wrote:
>> use some kind of debug output API which doesn't hit dmesg by default
>
> You still want to be enabled by default so that normal users can see it and
> actually report it.
These messages a
Applied. Thanks!
Alex
On Tue, Sep 9, 2025 at 11:29 AM Thorsten Blum wrote:
>
> Replace kmalloc() followed by copy_from_user() with memdup_user() to
> improve and simplify kfd_criu_restore_queue().
>
> No functional changes intended.
>
> Signed-off-by: Thorsten Blum
> ---
> .../amd/amdkfd/kfd_
Hi Christian,
On 9/9/2025 9:55 PM, Christian König wrote:
On 09.09.25 16:05, Peter Zijlstra wrote:
On Tue, Sep 09, 2025 at 02:04:30PM +0200, Christian König wrote:
Hi Arun,
On 09.09.25 11:56, Arunpravin Paneer Selvam wrote:
[SNIP]
+/**
+ * rbtree_for_each_entry_safe - iterate in-order over
On 09/09/2025 12:00, Harry Wentland wrote:
On 2025-09-01 17:33, Melissa Wen wrote:
Don't update DC stream color components during atomic check. The driver
will continue validating the new CRTC color state but will not change DC
stream color components. The DC stream color state will only be
it is more reasonable to move smu post_init() from late_init() to hw_init()
stage,
beacuse the smu specific hw init should be done before call other ip block
late_init funcs.
Signed-off-by: Yang Wang
---
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 12 ++--
1 file changed, 6 insertions(+
Fix the copy-pasted errors in the Resizable BAR handling functions
kernel doc and generally improve wording choices.
Fix the formatting errors of the Return: line.
Signed-off-by: Ilpo Järvinen
---
drivers/pci/rebar.c | 29 ++---
1 file changed, 18 insertions(+), 11 delet
userptrs could be changed by the user at any time and
hence while locking all the bos before GPU start processing
validate all the userptr bos.
Signed-off-by: Sunil Khatri
---
drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 89 +++
1 file changed, 89 insertions(+)
diff --git a/d
This keeps MMIO_REMAP fixed for its whole lifetime.
Other buffers (VRAM/GTT) are unchanged.
This change pins it when we export the dma-buf.
If the export fails, we undo the pin.
When the dma-buf is released, we unpin it.
MMIO_REMAP (HDP flush page) is a hardware I/O window, not normal RAM.
It sho
[Public]
Regards,
Prike
> -Original Message-
> From: Christian König
> Sent: Thursday, September 11, 2025 8:10 PM
> To: alexdeuc...@gmail.com; Khatri, Sunil ; Yang, Philip
> ; Liang, Prike
> Cc: amd-gfx@lists.freedesktop.org
> Subject: [PATCH 1/4] drm/amdgpu: fix userq VM validati
24 matches
Mail list logo
