RE: [PATCH v4 2/4] drm/amdgpu: Add reset_context flag for host FLR
[AMD Official Use Only - General]
Reviewed-by: Emily Deng
Emily Deng
Best Wishes
>-Original Message-
>From: Li, Yunxiang (Teddy)
>Sent: Saturday, April 27, 2024 2:27 AM
>To: amd-gfx@lists.freedesktop.org
>Cc: Deucher, Alexander ; Koenig, Christian
>; Lazar, Lijo ; Kuehling,
>Felix ; Deng, Emily ; Li,
>Yunxiang (Teddy)
>Subject: [PATCH v4 2/4] drm/amdgpu: Add reset_context flag for host FLR
>
>There are other reset sources that pass NULL as the job pointer, such as
>amdgpu_amdkfd_reset_work. Therefore, using the job pointer to check if the
>FLR comes from the host does not work.
>
>Add a flag in reset_context to explicitly mark host triggered reset, and set
>this flag when we receive host reset notification.
>
>Signed-off-by: Yunxiang Li
>---
>v2: fix typo
>v3: pass reset_context directly
>v4: clear the flag in case we retry
>
> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 13 -
>drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h | 1 +
> drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c | 1 +
> drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c | 1 +
> drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c | 1 +
> 5 files changed, 12 insertions(+), 5 deletions(-)
>
>diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>index 8befd10bf007..33c889c027a5 100644
>--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>@@ -5055,13 +5055,13 @@ static int amdgpu_device_recover_vram(struct
>amdgpu_device *adev)
> * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
> *
> * @adev: amdgpu_device pointer
>- * @from_hypervisor: request from hypervisor
>+ * @reset_context: amdgpu reset context pointer
> *
> * do VF FLR and reinitialize Asic
> * return 0 means succeeded otherwise failed
> */
> static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
>- bool from_hypervisor)
>+ struct amdgpu_reset_context
>*reset_context)
> {
> int r;
> struct amdgpu_hive_info *hive = NULL;
>@@ -5070,12 +5070,15 @@ static int amdgpu_device_reset_sriov(struct
>amdgpu_device *adev,
> retry:
> amdgpu_amdkfd_pre_reset(adev);
>
>- if (from_hypervisor)
>+ if (test_bit(AMDGPU_HOST_FLR, &reset_context->flags)) {
>+ clear_bit(AMDGPU_HOST_FLR, &reset_context->flags);
> r = amdgpu_virt_request_full_gpu(adev, true);
>- else
>+ } else {
> r = amdgpu_virt_reset_gpu(adev);
>+ }
> if (r)
> return r;
>+
> amdgpu_ras_set_fed(adev, false);
> amdgpu_irq_gpu_reset_resume_helper(adev);
>
>@@ -5826,7 +5829,7 @@ int amdgpu_device_gpu_recover(struct
>amdgpu_device *adev,
> /* Actual ASIC resets if needed.*/
> /* Host driver will handle XGMI hive reset for SRIOV */
> if (amdgpu_sriov_vf(adev)) {
>- r = amdgpu_device_reset_sriov(adev, job ? false : true);
>+ r = amdgpu_device_reset_sriov(adev, reset_context);
> if (r)
> adev->asic_reset_res = r;
>
>diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
>b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
>index b11d190ece53..5a9cc043b858 100644
>--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
>+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
>@@ -33,6 +33,7 @@ enum AMDGPU_RESET_FLAGS {
> AMDGPU_NEED_FULL_RESET = 0,
> AMDGPU_SKIP_HW_RESET = 1,
> AMDGPU_SKIP_COREDUMP = 2,
>+ AMDGPU_HOST_FLR = 3,
> };
>
> struct amdgpu_reset_context {
>diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
>b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
>index c5ba9c4757a8..f4c47492e0cd 100644
>--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
>+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
>@@ -292,6 +292,7 @@ static void xgpu_ai_mailbox_flr_work(struct
>work_struct *work)
> reset_context.method = AMD_RESET_METHOD_NONE;
> reset_context.reset_req_dev = adev;
> clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
>+ set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
>
> amdgpu_device_gpu_recover(adev, NULL, &reset_context);
> }
>diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
>b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
>index fa9d1b02f391..14cc7910e5cf 100644
>--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
>+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
>@@ -328,6 +328,7 @@ static void xgpu_nv_mailbox_flr_work(struct
>work_struct *work)
> reset_context.method = AMD_RESET_METHOD_NONE;
> reset_context.reset_req_dev = adev;
> clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
>+ set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
>
> amdgpu_device_gpu_recover(adev, NULL, &reset_context);
> }
>diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
>b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
>index 14a065516ae4..78cd07744ebe 100644
>--- a/driver
RE: [PATCH v4 2/4] drm/amdgpu: Add reset_context flag for host FLR
[AMD Official Use Only - General]
Reviewed-by: Zhigang Luo
-Original Message-
From: amd-gfx On Behalf Of Yunxiang Li
Sent: Friday, April 26, 2024 2:27 PM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander ; Koenig, Christian
; Lazar, Lijo ; Kuehling, Felix
; Deng, Emily ; Li, Yunxiang
(Teddy)
Subject: [PATCH v4 2/4] drm/amdgpu: Add reset_context flag for host FLR
There are other reset sources that pass NULL as the job pointer, such as
amdgpu_amdkfd_reset_work. Therefore, using the job pointer to check if the FLR
comes from the host does not work.
Add a flag in reset_context to explicitly mark host triggered reset, and set
this flag when we receive host reset notification.
Signed-off-by: Yunxiang Li
---
v2: fix typo
v3: pass reset_context directly
v4: clear the flag in case we retry
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 13 -
drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h | 1 +
drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c | 1 +
drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c | 1 +
drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c | 1 +
5 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 8befd10bf007..33c889c027a5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -5055,13 +5055,13 @@ static int amdgpu_device_recover_vram(struct
amdgpu_device *adev)
* amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
*
* @adev: amdgpu_device pointer
- * @from_hypervisor: request from hypervisor
+ * @reset_context: amdgpu reset context pointer
*
* do VF FLR and reinitialize Asic
* return 0 means succeeded otherwise failed
*/
static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
-bool from_hypervisor)
+struct amdgpu_reset_context *reset_context)
{
int r;
struct amdgpu_hive_info *hive = NULL;
@@ -5070,12 +5070,15 @@ static int amdgpu_device_reset_sriov(struct
amdgpu_device *adev,
retry:
amdgpu_amdkfd_pre_reset(adev);
- if (from_hypervisor)
+ if (test_bit(AMDGPU_HOST_FLR, &reset_context->flags)) {
+ clear_bit(AMDGPU_HOST_FLR, &reset_context->flags);
r = amdgpu_virt_request_full_gpu(adev, true);
- else
+ } else {
r = amdgpu_virt_reset_gpu(adev);
+ }
if (r)
return r;
+
amdgpu_ras_set_fed(adev, false);
amdgpu_irq_gpu_reset_resume_helper(adev);
@@ -5826,7 +5829,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
/* Actual ASIC resets if needed.*/
/* Host driver will handle XGMI hive reset for SRIOV */
if (amdgpu_sriov_vf(adev)) {
- r = amdgpu_device_reset_sriov(adev, job ? false : true);
+ r = amdgpu_device_reset_sriov(adev, reset_context);
if (r)
adev->asic_reset_res = r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
index b11d190ece53..5a9cc043b858 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -33,6 +33,7 @@ enum AMDGPU_RESET_FLAGS {
AMDGPU_NEED_FULL_RESET = 0,
AMDGPU_SKIP_HW_RESET = 1,
AMDGPU_SKIP_COREDUMP = 2,
+ AMDGPU_HOST_FLR = 3,
};
struct amdgpu_reset_context {
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
index c5ba9c4757a8..f4c47492e0cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
@@ -292,6 +292,7 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct
*work)
reset_context.method = AMD_RESET_METHOD_NONE;
reset_context.reset_req_dev = adev;
clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+ set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
amdgpu_device_gpu_recover(adev, NULL, &reset_context);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
index fa9d1b02f391..14cc7910e5cf 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
@@ -328,6 +328,7 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct
*work)
reset_context.method = AMD_RESET_METHOD_NONE;
reset_context.reset_req_dev = adev;
clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+ set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
amdgpu_device_gpu_recover(adev, NULL, &reset_context);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
index 14a065516ae4..78cd07744ebe 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
@@ -529,6
