Reviewed-by: Lizhi Hou
On 9/25/25 00:42, Karol Wachowski wrote:
Split firmware boot parameters (4KB) and FW version (4KB) into dedicated
buffer objects, separating them from the FW runtime memory buffer. This
creates three distinct buffers with independent allocation control.
This enables
Applied to drm-misc-next
On 9/23/25 08:29, Karol Wachowski wrote:
Reviewed-by: Karol Wachowski
On 9/23/2025 5:22 PM, Lizhi Hou wrote:
Currently, pm_runtime_resume_and_get() is invoked in the driver's open
callback, and pm_runtime_put_autosuspend() is called in the close
callback. As a r
On 9/23/25 04:18, Karol Wachowski wrote:
Split firmware boot parameters (4KB) and FW version (4KB) into dedicated
buffer objects, separating them from the FW runtime memory buffer. This
creates three distinct buffers with independent allocation control.
This enables future modifications, parti
Move the runtime PM calls to the AIE2 callbacks that actually interact
with the hardware. The device will automatically suspend after 5 seconds
of inactivity (no hardware accesses and no pending commands), and it will
be resumed on the next hardware access.
Signed-off-by: Lizhi Hou
---
drivers/
On 9/22/25 01:40, Karol Wachowski wrote:
On 9/18/2025 9:51 PM, Lizhi Hou wrote:
Currently, pm_runtime_resume_and_get() is invoked in the driver's open
callback, and pm_runtime_put_autosuspend() is called in the close
callback. As a result, the device remains active whenever an applic
Move the runtime PM calls to the AIE2 callbacks that actually interact
with the hardware. The device will automatically suspend after 5 seconds
of inactivity (no hardware accesses and no pending commands), and it will
be resumed on the next hardware access.
Signed-off-by: Lizhi Hou
---
drivers/
On 9/18/25 11:23, Mario Limonciello wrote:
On 9/18/2025 1:05 PM, Lizhi Hou wrote:
On 9/18/25 10:43, Mario Limonciello wrote:
On 9/18/2025 12:41 PM, Lizhi Hou wrote:
On 9/18/25 10:31, Mario Limonciello wrote:
On 9/18/2025 12:24 PM, Lizhi Hou wrote:
Currently
On 9/18/25 10:43, Mario Limonciello wrote:
On 9/18/2025 12:41 PM, Lizhi Hou wrote:
On 9/18/25 10:31, Mario Limonciello wrote:
On 9/18/2025 12:24 PM, Lizhi Hou wrote:
Currently, pm_runtime_resume_and_get() is invoked in the driver's open
callback, and pm_runtime_put_autosuspend
On 9/18/25 10:31, Mario Limonciello wrote:
On 9/18/2025 12:24 PM, Lizhi Hou wrote:
Currently, pm_runtime_resume_and_get() is invoked in the driver's open
callback, and pm_runtime_put_autosuspend() is called in the close
callback. As a result, the device remains active whenev
6a42c ("accel/amdxdna: Add ioctl DRM_IOCTL_AMDXDNA_GET_ARRAY")
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index 87c425e3d2b9..6e39c769bb6d 100
Move the runtime PM calls to the AIE2 callbacks that actually interact
with the hardware. The device will automatically suspend after 5 seconds
of inactivity (no hardware accesses and no pending commands), and it will
be resumed on the next hardware access.
Signed-off-by: Lizhi Hou
---
drivers/
On 9/16/25 08:55, Falkowski, Maciej wrote:
On 9/15/2025 6:10 PM, Lizhi Hou wrote:
In amdxdna_gem_obj_vmap(), calling dma_buf_vmap() triggers a kernel
warning if LOCKDEP is enabled. So for imported object, use
dma_buf_vmap_unlocked(). Then, use drm_gem_vmap() for other objects.
The similar
Applied to drm-misc-next
On 9/17/25 07:15, Falkowski, Maciej wrote:
Reviewed-by: Maciej Falkowski
On 9/16/2025 7:48 PM, Lizhi Hou wrote:
In amdxdna_gem_obj_vmap(), calling dma_buf_vmap() triggers a kernel
warning if LOCKDEP is enabled. So for imported object, use
dma_buf_vmap_unlocked
On 9/17/25 00:24, Karol Wachowski wrote:
On 9/16/2025 5:29 PM, Lizhi Hou wrote:
On 9/16/25 01:25, Karol Wachowski wrote:
On 9/15/2025 10:33 PM, Lizhi Hou wrote:
On 9/15/25 03:34, Karol Wachowski wrote:
From: Andrzej Kacprowski
Allow user mode drivers to manage preemption buffers
space allocated buffer")
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/amdxdna_gem.c | 47 -
1 file changed, 20 insertions(+), 27 deletions(-)
diff --git a/drivers/accel/amdxdna/amdxdna_gem.c
b/drivers/accel/amdxdna/amdxdna_gem.c
index d407a36eb412..7f91863c3
On 9/16/25 01:25, Karol Wachowski wrote:
On 9/15/2025 10:33 PM, Lizhi Hou wrote:
On 9/15/25 03:34, Karol Wachowski wrote:
From: Andrzej Kacprowski
Allow user mode drivers to manage preemption buffers, enabling
memory savings by sharing a single buffer across multiple
command queues within
Reviewed-by: Lizhi Hou
On 9/16/25 01:41, Karol Wachowski wrote:
Synchronize the JSM API header file with the latest 3.32.5 version
to reflect all changes introduced in the new firmware API
Signed-off-by: Karol Wachowski
---
v0 -> v1:
Corrected commit message wording to better reflect
Reviewed-by: Lizhi Hou
On 9/16/25 01:48, Karol Wachowski wrote:
Previously, aborting work could return early after engine reset or resume
failure, skipping the necessary runtime_put cleanup leaving the device
with incorrect reference count breaking runtime power management state.
Replace
On 9/15/25 03:35, Karol Wachowski wrote:
Update firmware JSM API header to version 3.32.5
The changes seem more than updating header version?
Lizhi
Signed-off-by: Karol Wachowski
---
drivers/accel/ivpu/vpu_jsm_api.h | 513 ---
1 file changed, 326 insertion
Reviewed-by: Lizhi Hou
On 9/15/25 03:34, Karol Wachowski wrote:
From: Jacek Lawrynowicz
Reduce code duplication and improve the overall readability of the debugfs
output for job scheduling priority bands.
Additionally fix clang-tidy warning about missing default case in the
switch statement
Reviewed-by: Lizhi Hou
On 9/15/25 03:34, Karol Wachowski wrote:
Ensure that cmdq->db_id is reset to 0 if ivpu_jsm_register_db fails,
preventing potential reuse of invalid command queue with
unregistered doorbell.
Signed-off-by: Karol Wachowski
---
drivers/accel/ivpu/ivpu_job.c
On 9/15/25 03:34, Karol Wachowski wrote:
From: Andrzej Kacprowski
Allow user mode drivers to manage preemption buffers, enabling
memory savings by sharing a single buffer across multiple
command queues within the same memory context.
Introduce DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE to report the
This looks a bug fix. Is adding 'Fixes:' tag better?
On 9/15/25 03:35, Karol Wachowski wrote:
Previously, aborting work could return early after engine reset or resume
failure, skipping the necessary runtime_put cleanup leaving the device
with incorrect reference count breaking runtime power man
Reviewed-by: Lizhi Hou
On 9/15/25 03:35, Karol Wachowski wrote:
From: Andrzej Kacprowski
Remove references to firmware boot parameters that were never used
by any production version of device firmware.
Signed-off-by: Andrzej Kacprowski
Signed-off-by: Karol Wachowski
---
drivers/accel
space allocated buffer")
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/amdxdna_gem.c | 38 +++--
1 file changed, 14 insertions(+), 24 deletions(-)
diff --git a/drivers/accel/amdxdna/amdxdna_gem.c
b/drivers/accel/amdxdna/amdxdna_gem.c
index d407a36eb412..50950be18
Applied to drm-misc-next
On 9/11/25 07:56, Falkowski, Maciej wrote:
Reviewed-by: Maciej Falkowski
On 9/9/2025 5:45 PM, Lizhi Hou wrote:
The unpublished smatch static checker reported a warning.
drivers/accel/amdxdna/aie2_pci.c:904 aie2_query_ctx_status_array()
warn: potential user
On 9/8/25 13:13, Dan Carpenter wrote:
On Mon, Sep 08, 2025 at 11:19:33AM -0700, Lizhi Hou wrote:
On 9/7/25 23:40, Dan Carpenter wrote:
Hello Lizhi Hou,
Commit 2f509fe6a42c ("accel/amdxdna: Add ioctl
DRM_IOCTL_AMDXDNA_GET_ARRAY") from Sep 2, 2025 (linux-next), leads to
the
On 9/7/25 23:40, Dan Carpenter wrote:
Hello Lizhi Hou,
Commit 2f509fe6a42c ("accel/amdxdna: Add ioctl
DRM_IOCTL_AMDXDNA_GET_ARRAY") from Sep 2, 2025 (linux-next), leads to
the following (UNPUBLISHED) Smatch static checker warning:
drivers/accel/amdxdna/aie2
space allocated buffer")
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/amdxdna_gem.c | 38 +++--
1 file changed, 14 insertions(+), 24 deletions(-)
diff --git a/drivers/accel/amdxdna/amdxdna_gem.c
b/drivers/accel/amdxdna/amdxdna_gem.c
index d407a36eb412..50950be18
Thanks. Applied to drm-misc-next
On 9/4/25 10:50, Lucas De Marchi wrote:
On Thu, Sep 04, 2025 at 10:15:42AM -0700, Lizhi Hou wrote:
Hi Lucas
I got a dim error when applying this patch. Could you take a look?
Maybe there is not a maintainer for MAINTAINERS? :)
Yes. A little bit worse than
i
On 9/2/25 12:47, Lizhi Hou wrote:
Reviewed-by: Lizhi Hou
On 8/30/25 17:12, Min Ma wrote:
I recently left AMD and would like to continue participating in
the review and maintenance of the XDNA driver using my personal
email address.
Signed-off-by: Min Ma
---
MAINTAINERS | 2 +-
1
Applied to drm-misc-next
On 9/2/25 23:54, Falkowski, Maciej wrote:
Reviewed-by: Maciej Falkowski
On 9/3/2025 7:34 AM, Lizhi Hou wrote:
Add interface for applications to get information array. The application
provides a buffer pointer along with information type, maximum number of
entries and
On 9/2/25 12:48, Alex Deucher wrote:
On Tue, Sep 2, 2025 at 2:09 PM Lizhi Hou wrote:
Add interface for applications to get information array. The application
provides a buffer pointer along with information type, maximum number of
entries and maximum size of each entry. The buffer may also
actual number of entries and entry size are returned. (see [1], used by
driver runtime library)
[1]
https://github.com/amd/xdna-driver/blob/main/src/shim/host/platform_host.cpp#L337
Reviewed-by: Mario Limonciello (AMD)
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c| 116
Reviewed-by: Lizhi Hou
On 8/30/25 17:12, Min Ma wrote:
I recently left AMD and would like to continue participating in
the review and maintenance of the XDNA driver using my personal
email address.
Signed-off-by: Min Ma
---
MAINTAINERS | 2 +-
1 file changed, 1 insertion(+), 1 deletion
On 9/2/25 13:15, Alex Deucher wrote:
On Tue, Sep 2, 2025 at 3:58 PM Lizhi Hou wrote:
On 9/2/25 12:48, Alex Deucher wrote:
On Tue, Sep 2, 2025 at 2:09 PM Lizhi Hou wrote:
Add interface for applications to get information array. The application
provides a buffer pointer along with
actual number of entries and entry size are returned.
Signed-off-by: Lizhi Hou
Link: https://lore.kernel.org/r/20250827203031.1512508-1-lizhi@amd.com
---
drivers/accel/amdxdna/aie2_pci.c| 116 ++--
drivers/accel/amdxdna/amdxdna_pci_drv.c | 30 ++
drivers/accel
On 9/2/25 10:02, Falkowski, Maciej wrote:
On 8/27/2025 10:30 PM, Lizhi Hou wrote:
Add interface for applications to get information array. The application
provides a buffer pointer along with information type, maximum number of
entries and maximum size of each entry. The buffer may also
Applied to drm-misc-next.
On 8/28/25 13:23, Lizhi Hou wrote:
Reviewed-by: Lizhi Hou
On 8/27/25 20:39, Qianfeng Rong wrote:
Change the 'ret' variable from u32 to int to store -EINVAL. Storing the
negative error codes in unsigned type, doesn't cause an issue at runtime
but it&
(+), 1 deletion(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 10850512c118..6eefa494000c 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1231,7 +1231,7 @@ F:drivers/spi/spi-amd.c
F:drivers/spi/spi-amd.h
AMD XDNA DRIVER
-M: Min Ma
+M: Min Ma
M:Lizhi Hou
L:dri
Reviewed-by: Lizhi Hou
On 8/27/25 20:39, Qianfeng Rong wrote:
Change the 'ret' variable from u32 to int to store -EINVAL. Storing the
negative error codes in unsigned type, doesn't cause an issue at runtime
but it's ugly as pants.
Additionally, assigning -EINVAL to u3
actual number of entries and entry size are returned.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c| 116 ++--
drivers/accel/amdxdna/amdxdna_pci_drv.c | 30 ++
drivers/accel/amdxdna/amdxdna_pci_drv.h | 1 +
include/uapi/drm/amdxdna_accel.h
On 8/27/25 11:09, Mario Limonciello wrote:
On 8/27/2025 11:41 AM, Lizhi Hou wrote:
On 8/26/25 17:31, Mario Limonciello wrote:
On 8/26/2025 1:10 PM, Lizhi Hou wrote:
On 8/26/25 10:58, Mario Limonciello wrote:
On 8/26/2025 12:55 PM, Lizhi Hou wrote:
On 8/26/25 10:18, Mario Limonciello
On 8/26/25 19:15, Qianfeng Rong wrote:
在 2025/8/27 0:31, Lizhi Hou 写道:
On 8/26/25 00:29, Qianfeng Rong wrote:
Change the 'ret' variable from u32 to int to store -EINVAL, reducing
potential risks such as incorrect results when comparing 'ret' with
error codes.
Sounds t
On 8/26/25 17:31, Mario Limonciello wrote:
On 8/26/2025 1:10 PM, Lizhi Hou wrote:
On 8/26/25 10:58, Mario Limonciello wrote:
On 8/26/2025 12:55 PM, Lizhi Hou wrote:
On 8/26/25 10:18, Mario Limonciello wrote:
On 8/25/2025 11:48 PM, Lizhi Hou wrote:
On 8/25/25 14:28, Mario Limonciello
On 8/26/25 10:58, Mario Limonciello wrote:
On 8/26/2025 12:55 PM, Lizhi Hou wrote:
On 8/26/25 10:18, Mario Limonciello wrote:
On 8/25/2025 11:48 PM, Lizhi Hou wrote:
On 8/25/25 14:28, Mario Limonciello wrote:
On 8/22/2025 12:23 PM, Lizhi Hou wrote:
Add interface for applications to get
On 8/26/25 10:18, Mario Limonciello wrote:
On 8/25/2025 11:48 PM, Lizhi Hou wrote:
On 8/25/25 14:28, Mario Limonciello wrote:
On 8/22/2025 12:23 PM, Lizhi Hou wrote:
Add interface for applications to get information array. The
application
provides a buffer pointer along with information
5.0b9effl6-...@intel.com/
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index 16ac0cab4f44..7a3449541107 100644
--- a/drivers/accel/amdxdna/aie2_p
On 8/26/25 09:50, Mario Limonciello wrote:
On 8/26/2025 11:03 AM, Lizhi Hou wrote:
drivers/accel/amdxdna/aie2_pci.c:794:13: sparse: sparse: incorrect
type in assignment (different address spaces)
Fixes: c8cea4371e5e ("accel/amdxdna: Add a function to walk hardware
contexts")
R
On 8/26/25 00:29, Qianfeng Rong wrote:
Change the 'ret' variable from u32 to int to store -EINVAL, reducing
potential risks such as incorrect results when comparing 'ret' with
error codes.
Sounds this fixes code issue. Could you add "Fixes" tag?
Thanks,
Lizhi
Signed-off-by: Qianfeng Rong
5.0b9effl6-...@intel.com/
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index 16ac0cab4f44..2885a3c17e89 100644
--- a/drivers/accel/amdxdna/aie2_p
On 8/25/25 14:28, Mario Limonciello wrote:
On 8/22/2025 12:23 PM, Lizhi Hou wrote:
Add interface for applications to get information array. The application
provides a buffer pointer along with information type, maximum number of
entries and maximum size of each entry. The buffer may also
actual number of entries and entry size are returned.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_pci.c| 114 ++--
drivers/accel/amdxdna/amdxdna_pci_drv.c | 21 +
drivers/accel/amdxdna/amdxdna_pci_drv.h | 1 +
include/uapi/drm/amdxdna_accel.h| 109
INERS
index 10850512c118..6eefa494000c 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1231,7 +1231,7 @@ F: drivers/spi/spi-amd.c
F: drivers/spi/spi-amd.h
AMD XDNA DRIVER
-M: Min Ma
+M: Min Ma
M: Lizhi Hou
L: dri-devel@lists.freedesktop.org
S: Supported
--
2.43.0
Applied to drm-misc-next.
Lizhi
On 8/21/25 05:52, Alex Deucher wrote:
On Wed, Aug 20, 2025 at 8:03 PM Lizhi Hou wrote:
The compiler information is outdated. Update it to the latest.
Signed-off-by: Lizhi Hou
Reviewed-by: Alex Deucher
---
Documentation/accel/amdxdna/amdnpu.rst | 10
The compiler information is outdated. Update it to the latest.
Signed-off-by: Lizhi Hou
---
Documentation/accel/amdxdna/amdnpu.rst | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/Documentation/accel/amdxdna/amdnpu.rst
b/Documentation/accel/amdxdna/amdnpu.rst
Applied to drm-misc-next
Thanks,
Lizhi
On 8/15/25 13:43, Limonciello, Mario wrote:
On 8/15/25 12:16 PM, Lizhi Hou wrote:
Walking hardware contexts created by a process is duplicated in multiple
spots. Add a function, amdxdna_hwctx_walk(), and replace all spots.
hwctx_srcu and dev_lock are
Walking hardware contexts created by a process is duplicated in multiple
spots. Add a function, amdxdna_hwctx_walk(), and replace all spots.
hwctx_srcu and dev_lock are good enough to protect hardware context list.
Remove hwctx_lock.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna
On 8/14/25 13:43, Limonciello, Mario wrote:
On 8/14/25 3:29 PM, Lizhi Hou wrote:
Walking hardware contexts created by a process is duplicated in multiple
spots. Add a function, amdxdna_hwctx_walk(), and replace all spots.
hwctx_srcu and dev_lock are good enough to protect hardware context
Walking hardware contexts created by a process is duplicated in multiple
spots. Add a function, amdxdna_hwctx_walk(), and replace all spots.
hwctx_srcu and dev_lock are good enough to protect hardware context list.
Remove hwctx_lock.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna
ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u32
dct_percent);
+void ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u32
active_percent);
Reviewed-by: Lizhi Hou
u32 ivpu_hw_btrs_telemetry_offset_get(struct ivpu_device *vdev);
u32 ivpu_hw_btrs_telemetry_size_get(struct
= 0;
} else {
wp->target = hw->pll.pn_ratio;
- wp->cfg = enable ? PLL_CONFIG_DEFAULT : 0;
+ wp->cfg = 0;
Reviewed-by: Lizhi Hou
wp->cdyn = enable ? PLL_CDYN_DEFAULT : 0;
wp->epp = enable ? PLL_EPP_DEFAULT : 0;
}
;
void ivpu_pm_enable(struct ivpu_device *vdev);
void ivpu_pm_disable(struct ivpu_device *vdev);
-void ivpu_pm_cancel_recovery(struct ivpu_device *vdev);
+void ivpu_pm_disable_recovery(struct ivpu_device *vdev);
Reviewed-by: Lizhi Hou
int ivpu_pm_suspend_cb(struct device *dev);
int ivpu_pm_resume_cb(struct device *dev);
Applied to drm-misc-next
On 8/5/25 11:55, Mario Limonciello wrote:
On 8/3/25 2:14 PM, Lizhi Hou wrote:
The suspend and resume callbacks for pm and runtime pm should be same.
During suspending, it needs to stop all hardware contexts first. And
the hardware contexts will be restarted after the
The suspend and resume callbacks for pm and runtime pm should be same.
During suspending, it needs to stop all hardware contexts first. And
the hardware contexts will be restarted after the device is resumed.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_ctx.c| 59
On 8/1/25 05:36, Mario Limonciello wrote:
On 7/31/2025 10:05 PM, Lizhi Hou wrote:
The suspend and resume callbacks for pm and runtime pm should be same.
During suspending, it needs to stop all hardware contexts first. And
the hardware contexts will be restarted after the device is resumed
On 8/1/25 07:48, Falkowski, Maciej wrote:
On 7/31/2025 6:35 PM, Lizhi Hou wrote:
The suspend and resume callbacks for pm and runtime pm should be same.
During suspending, it needs to stop all hardware contexts first. And
the hardware contexts will be restarted after the device is resumed
The suspend and resume callbacks for pm and runtime pm should be same.
During suspending, it needs to stop all hardware contexts first. And
the hardware contexts will be restarted after the device is resumed.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_ctx.c| 32
Pushed to drm-misc-next. Thanks.
Lizhi
On 7/22/25 00:55, Jacek Lawrynowicz wrote:
Reviewed-by: Jacek Lawrynowicz
On 7/19/2025 8:33 AM, Salah Triki wrote:
The device is managed so pci_free_irq_vectors() is called automatically
no need to do it manually.
Signed-off-by: Salah Triki
---
dri
Pushed to drm-misc-next. Thanks.
Lizhi
On 7/22/25 01:22, Jacek Lawrynowicz wrote:
Reviewed-by: Jacek Lawrynowicz
On 7/16/2025 6:44 PM, Lizhi Hou wrote:
Enhance DRM_IOCTL_AMDXDNA_CREATE_BO to accept user space allocated
buffer pointer. The buffer pages will be pinned in memory. Unless
the
Enhance DRM_IOCTL_AMDXDNA_CREATE_BO to accept user space allocated
buffer pointer. The buffer pages will be pinned in memory. Unless
the CAP_IPC_LOCK is enabled for the application process, the total
pinned memory can not beyond rlimit_memlock.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna
On 7/14/25 03:33, Jacek Lawrynowicz wrote:
Hi,
This looks very interesting. It's a clever way of adding userptr on top of
existing gem code.
On 7/7/2025 9:38 PM, Lizhi Hou wrote:
Enhance DRM_IOCTL_AMDXDNA_CREATE_BO to accept user space allocated
buffer pointer. The buffer pages wi
Enhance DRM_IOCTL_AMDXDNA_CREATE_BO to accept user space allocated
buffer pointer. The buffer pages will be pinned in memory. Unless
the CAP_IPC_LOCK is enabled for the application process, the total
pinned memory can not beyond rlimit_memlock.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna
Reviewed-by: Lizhi Hou
On 7/4/25 00:53, Sakari Ailus wrote:
pm_runtime_put_autosuspend(), pm_runtime_put_sync_autosuspend(),
pm_runtime_autosuspend() and pm_request_autosuspend() now include a call
to pm_runtime_mark_last_busy(). Remove the now-reduntant explicit call to
On 7/2/25 02:14, Marcello Sylvester Bauer wrote:
Greetings,
I have some questions about making the AMD XDNA NPU accessible in a
virtualized environment. I tried using VFIO on QEMU, but I could not
get it to work.
Here is a brief rundown of the resulting issues:
Simply adding the device to
Pushed to drm-misc-next
On 6/16/25 04:30, Jacek Lawrynowicz wrote:
Hi,
On 6/16/2025 11:14 AM, Lizhi Hou wrote:
The device bo is allocated from the device heap memory. (a trunk of
memory dedicated to device)
Rename amdxdna_gem_insert_node_locked to amdxdna_gem_heap_alloc
and move related
address of device bo by the device
heap memory address and offset.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_ctx.c| 2 +-
drivers/accel/amdxdna/amdxdna_gem.c | 193 +++-
drivers/accel/amdxdna/amdxdna_gem.h | 3 +-
3 files changed, 106 insertions
Pushed to drm-misc-fixes
On 6/9/25 06:35, Alex Deucher wrote:
On Wed, Jun 4, 2025 at 8:12 PM Lizhi Hou wrote:
On 6/4/25 07:51, Alex Deucher wrote:
On Wed, Jun 4, 2025 at 10:42 AM Lizhi Hou wrote:
The incorrect PSP firmware size is used for initializing. It may
cause error for newer
On 6/2/25 05:58, Jacek Lawrynowicz wrote:
Hi,
On 5/28/2025 7:50 PM, Lizhi Hou wrote:
On 5/28/25 08:43, Jacek Lawrynowicz wrote:
This fixes a potential race conditions in:
- ivpu_bo_unbind_locked() where we modified the shmem->sgt without
holding the dma_resv_l
On 6/2/25 06:05, Jacek Lawrynowicz wrote:
Hi,
On 5/28/2025 7:53 PM, Lizhi Hou wrote:
On 5/28/25 08:42, Jacek Lawrynowicz wrote:
From: Karol Wachowski
Trigger full device recovery when the driver fails to restore device state
via engine reset and resume operations. This is necessary
On 6/4/25 07:51, Alex Deucher wrote:
On Wed, Jun 4, 2025 at 10:42 AM Lizhi Hou wrote:
The incorrect PSP firmware size is used for initializing. It may
cause error for newer version firmware.
Fixes: 8c9ff1b181ba ("accel/amdxdna: Add a new driver for AMD AI Engine")
Signed-off-by:
The incorrect PSP firmware size is used for initializing. It may
cause error for newer version firmware.
Fixes: 8c9ff1b181ba ("accel/amdxdna: Add a new driver for AMD AI Engine")
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/aie2_psp.c | 4 ++--
1 file changed, 2 insertions(+), 2
On 5/28/25 08:42, Jacek Lawrynowicz wrote:
From: Karol Wachowski
Trigger full device recovery when the driver fails to restore device state
via engine reset and resume operations. This is necessary because, even if
submissions from a faulty context are blocked, the NPU may still process
previ
On 5/28/25 08:43, Jacek Lawrynowicz wrote:
This fixes a potential race conditions in:
- ivpu_bo_unbind_locked() where we modified the shmem->sgt without
holding the dma_resv_lock().
- ivpu_bo_print_info() where we read the shmem->pages without
holding the dma_resv_lock().
Using dma
m, !dma_resv_test_signaled(obj->resv, DMA_RESV_USAGE_READ));
+ drm_WARN_ON(&vdev->drm, !drm_gem_is_imported(&bo->base.base) &&
+ !dma_resv_test_signaled(obj->resv, DMA_RESV_USAGE_READ));
Reviewed-by: Lizhi Hou
drm_WARN_ON(&vdev->drm, ivpu_bo_size(bo) == 0);
drm_WARN_ON(&vdev->drm, bo->base.vaddr);
On 5/28/25 08:42, Jacek Lawrynowicz wrote:
Don't WARN if imported buffers are in use in ivpu_gem_bo_free() as they
can be indeed used in the original context/driver.
Fixes: 647371a6609d ("accel/ivpu: Add GEM buffer object management")
Cc: # v6.3
Signed-off-by: Jacek Lawrynowicz
---
drivers
Merged to drm-misc-next
On 5/7/25 10:31, Lizhi Hou wrote:
On 5/7/25 09:29, Mario Limonciello wrote:
On 5/7/2025 11:15 AM, Lizhi Hou wrote:
The latest userspace runtime allows generating commands which do not
have any argument. Remove the corresponding check in driver IOCTL to
enable this use
On 5/7/25 09:29, Mario Limonciello wrote:
On 5/7/2025 11:15 AM, Lizhi Hou wrote:
The latest userspace runtime allows generating commands which do not
have any argument. Remove the corresponding check in driver IOCTL to
enable this use case.
Signed-off-by: Lizhi Hou
Can the userspace
The latest userspace runtime allows generating commands which do not
have any argument. Remove the corresponding check in driver IOCTL to
enable this use case.
Signed-off-by: Lizhi Hou
---
drivers/accel/amdxdna/amdxdna_ctx.c | 22 --
1 file changed, 12 insertions(+), 10
On 5/6/25 23:59, Jacek Lawrynowicz wrote:
Hi,
On 5/6/2025 5:41 PM, Lizhi Hou wrote:
On 5/6/25 02:20, Jacek Lawrynowicz wrote:
Use FW names from linux-firmware repo instead of deprecated ones.
Fixes: c140244f0cfb ("accel/ivpu: Add initial Panther Lake support")
Cc: # v6.13+
Sig
On 5/6/25 02:20, Jacek Lawrynowicz wrote:
Use FW names from linux-firmware repo instead of deprecated ones.
Fixes: c140244f0cfb ("accel/ivpu: Add initial Panther Lake support")
Cc: # v6.13+
Signed-off-by: Jacek Lawrynowicz
---
drivers/accel/ivpu/ivpu_fw.c | 12 ++--
1 file changed
ad(&bo->base.base.refcount));
if (bo->base.pages)
diff --git a/drivers/accel/ivpu/ivpu_gem.h b/drivers/accel/ivpu/ivpu_gem.h
index a222a9ec9d611..0c93118c85bd3 100644
--- a/drivers/accel/ivpu/ivpu_gem.h
+++ b/drivers/accel/ivpu/ivpu_gem.h
@@ -21,6 +21,7 @@ struct ivpu_bo {
u64 vpu_addr;
u32 flags;
u32 job_status; /* Valid only for command buffer */
+ u32 ctx_id;
Reviewed-by: Lizhi Hou
bool mmu_mapped;
};
On 4/10/25 00:27, Falkowski, Maciej wrote:
On 4/9/2025 11:00 PM, Lizhi Hou wrote:
When multiple ERT_START_NPU commands are combined in one buffer, the
buffer size calculation is incorrect. Also, the condition to make sure
the buffer size is not beyond 4K is also fixed.
Fixes: aac243092b70
When multiple ERT_START_NPU commands are combined in one buffer, the
buffer size calculation is incorrect. Also, the condition to make sure
the buffer size is not beyond 4K is also fixed.
Fixes: aac243092b70 ("accel/amdxdna: Add command execution")
Signed-off-by: Lizhi Hou
---
dri
ount %u\n", file_priv->ctx.id, buffer_count);
+ ivpu_dbg(vdev, JOB, "Submit ioctl: ctx %u cmdq_id %u buf_count %u\n",
+file_priv->ctx.id, cmdq_id, buffer_count);
Reviewed-by: Lizhi Hou
job = ivpu_job_create(file_priv, engine, buffer_count);
if (!job) {
On 4/1/25 08:58, Maciej Falkowski wrote:
From: Karol Wachowski
This commit bumps FW Boot API to 3.28.3.
Bump FW Boot API ...
Please see:
https://www.kernel.org/doc/html/v6.14/process/submitting-patches.html
Lizhi
Use new preemption buffer size fields from FW header added to
firmware
flush_work(&vdev->irq_ipc_work);
+ flush_work(&vdev->irq_dct_work);
+ flush_work(&vdev->context_abort_work);
Reviewed-by: Lizhi Hou
ivpu_ipc_disable(vdev);
ivpu_mmu_disable(vdev);
}
On 3/28/25 09:16, Jeff Hugo wrote:
On 3/25/2025 2:01 PM, Lizhi Hou wrote:
Add amdxdna_gem_prime_export() and amdxdna_gem_prime_import() for BO
import and export. Register mmu notifier for imported BO as well. When
MMU_NOTIFIER_UNMAP event is received, queue work to remove the notifier.
The
k);
ms = get_instance_by_mask(file_priv, args->metric_group_mask);
@@ -218,6 +235,7 @@ int ivpu_ms_stop_ioctl(struct drm_device *dev, void *data,
struct drm_file *file
mutex_unlock(&file_priv->ms_lock);
+ ivpu_rpm_put(vdev);
Reviewed-by: Lizhi Hou
return ms ? 0 : -EINVAL;
}
On 3/28/25 01:23, Jacek Lawrynowicz wrote:
On 3/27/2025 6:38 PM, Lizhi Hou wrote:
On 3/26/25 01:06, Jacek Lawrynowicz wrote:
Hi,
On 3/25/2025 9:50 PM, Lizhi Hou wrote:
On 3/25/25 04:43, Maciej Falkowski wrote:
From: Jacek Lawrynowicz
Fix deadlock in ivpu_ms_cleanup() by preventing
1 - 100 of 338 matches
Mail list logo
