Separate RAS poison consumption handling from the instance irq, and
register dedicated ras_poison_irq src and funcs for UVD_POISON. Fix
the amdgpu_irq_put call trace in vcn_v4_0_hw_fini.
[ 50.497562] RIP: 0010:amdgpu_irq_put+0xa4/0xc0 [amdgpu]
[ 50.497619] RSP: 0018:ffffaa2400fcfcb0 EFLAGS: 00010246
[ 50.497620] RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000000
[ 50.497621] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
[ 50.497621] RBP: ffffaa2400fcfcd0 R08: 0000000000000000 R09: 0000000000000000
[ 50.497622] R10: 0000000000000000 R11: 0000000000000000 R12: ffff99b2105242d8
[ 50.497622] R13: 0000000000000000 R14: ffff99b210500000 R15: ffff99b210500000
[ 50.497623] FS: 0000000000000000(0000) GS:ffff99b518480000(0000) knlGS:0000000000000000
[ 50.497623] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 50.497624] CR2: 00007f9d32aa91e8 CR3: 00000001ba210000 CR4: 0000000000750ee0
[ 50.497624] PKRU: 55555554
[ 50.497625] Call Trace:
[ 50.497625] <TASK>
[ 50.497627] jpeg_v4_0_hw_fini+0x43/0xc0 [amdgpu]
[ 50.497693] jpeg_v4_0_suspend+0x13/0x30 [amdgpu]
[ 50.497751] amdgpu_device_ip_suspend_phase2+0x240/0x470 [amdgpu]
[ 50.497802] amdgpu_device_ip_suspend+0x41/0x80 [amdgpu]
[ 50.497854] amdgpu_device_pre_asic_reset+0xd9/0x4a0 [amdgpu]
[ 50.497905] amdgpu_device_gpu_recover.cold+0x548/0xcf1 [amdgpu]
[ 50.498005] amdgpu_debugfs_reset_work+0x4c/0x80 [amdgpu]
[ 50.498060] process_one_work+0x21f/0x400
[ 50.498063] worker_thread+0x200/0x3f0
[ 50.498064] ? process_one_work+0x400/0x400
[ 50.498065] kthread+0xee/0x120
[ 50.498067] ? kthread_complete_and_exit+0x20/0x20
[ 50.498068] ret_from_fork+0x22/0x30
Suggested-by: Hawking Zhang <Hawking.Zhang@xxxxxxx>
Signed-off-by: Horatio Zhang <Hongkun.Zhang@xxxxxxx>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c | 27 +++++++++++++++++++++++-
drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h | 3 +++
drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c | 27 ++++++++++++++++++------
drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c | 27 ++++++++++++++++++------
4 files changed, 70 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
index b07c000fc8ba..f27078fc67c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -241,6 +241,31 @@ int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev,
return 0;
}
+int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+ int r, i;
+
+ r = amdgpu_ras_block_late_init(adev, ras_block);
+ if (r)
+ return r;
+
+ if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ if (adev->jpeg.harvest_config & (1 << i))
+ continue;
+
+ r = amdgpu_irq_get(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);
+ if (r)
+ goto late_fini;
+ }
+ }
+ return 0;
+
+late_fini:
+ amdgpu_ras_block_late_fini(adev, ras_block);
+ return r;
+}
+
int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev)
{
int err;
@@ -262,7 +287,7 @@ int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev)
adev->jpeg.ras_if = &ras->ras_block.ras_comm;
if (!ras->ras_block.ras_late_init)
- ras->ras_block.ras_late_init = amdgpu_ras_block_late_init;
+ ras->ras_block.ras_late_init = amdgpu_jpeg_ras_late_init;
return 0;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
index 0ca76f0f23e9..172296e9f40c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
@@ -38,6 +38,7 @@ struct amdgpu_jpeg_reg{
struct amdgpu_jpeg_inst {
struct amdgpu_ring ring_dec;
struct amdgpu_irq_src irq;
+ struct amdgpu_irq_src ras_poison_irq;
struct amdgpu_jpeg_reg external;
};
@@ -72,6 +73,8 @@ int amdgpu_jpeg_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout);
int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry);
+int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev,
+ struct ras_common_if *ras_block);
int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev);
#endif /*__AMDGPU_JPEG_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
index b040f51d9aa9..5ee7bc6a9360 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
@@ -102,13 +102,13 @@ static int jpeg_v2_5_sw_init(void *handle)
/* JPEG DJPEG POISON EVENT */
r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
- VCN_2_6__SRCID_DJPEG0_POISON, &adev->jpeg.inst[i].irq);
+ VCN_2_6__SRCID_DJPEG0_POISON, &adev->jpeg.inst[i].ras_poison_irq);
if (r)
return r;
/* JPEG EJPEG POISON EVENT */
r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
- VCN_2_6__SRCID_EJPEG0_POISON, &adev->jpeg.inst[i].irq);
+ VCN_2_6__SRCID_EJPEG0_POISON, &adev->jpeg.inst[i].ras_poison_irq);
if (r)
return r;
}
@@ -221,6 +221,9 @@ static int jpeg_v2_5_hw_fini(void *handle)
if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
RREG32_SOC15(JPEG, i, mmUVD_JRBC_STATUS))
jpeg_v2_5_set_powergating_state(adev, AMD_PG_STATE_GATE);
+
+ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG))
+ amdgpu_irq_put(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);
}
return 0;
@@ -569,6 +572,14 @@ static int jpeg_v2_5_set_interrupt_state(struct amdgpu_device *adev,
return 0;
}
+static int jpeg_v2_6_set_ras_interrupt_state(struct amdgpu_device *adev,
+ struct amdgpu_irq_src *source,
+ unsigned int type,
+ enum amdgpu_interrupt_state state)
+{
+ return 0;
+}
+
static int jpeg_v2_5_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
@@ -593,10 +604,6 @@ static int jpeg_v2_5_process_interrupt(struct amdgpu_device *adev,
case VCN_2_0__SRCID__JPEG_DECODE:
amdgpu_fence_process(&adev->jpeg.inst[ip_instance].ring_dec);
break;
- case VCN_2_6__SRCID_DJPEG0_POISON:
- case VCN_2_6__SRCID_EJPEG0_POISON:
- amdgpu_jpeg_process_poison_irq(adev, source, entry);
- break;
default:
DRM_ERROR("Unhandled interrupt: %d %d\n",
entry->src_id, entry->src_data[0]);
@@ -725,6 +732,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v2_5_irq_funcs = {
.process = jpeg_v2_5_process_interrupt,
};
+static const struct amdgpu_irq_src_funcs jpeg_v2_6_ras_irq_funcs = {
+ .set = jpeg_v2_6_set_ras_interrupt_state,
+ .process = amdgpu_jpeg_process_poison_irq,
+};
+
static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev)
{
int i;
@@ -735,6 +747,9 @@ static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev)
adev->jpeg.inst[i].irq.num_types = 1;
adev->jpeg.inst[i].irq.funcs = &jpeg_v2_5_irq_funcs;
+
+ adev->jpeg.inst[i].ras_poison_irq.num_types = 1;
+ adev->jpeg.inst[i].ras_poison_irq.funcs = &jpeg_v2_6_ras_irq_funcs;
}
}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
index 77e1e64aa1d1..60b9362fcebc 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -87,13 +87,13 @@ static int jpeg_v4_0_sw_init(void *handle)
/* JPEG DJPEG POISON EVENT */
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
- VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->irq);
+ VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
if (r)
return r;
/* JPEG EJPEG POISON EVENT */
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
- VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->irq);
+ VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
if (r)
return r;
@@ -202,7 +202,8 @@ static int jpeg_v4_0_hw_fini(void *handle)
RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS))
jpeg_v4_0_set_powergating_state(adev, AMD_PG_STATE_GATE);
}
- amdgpu_irq_put(adev, &adev->jpeg.inst->irq, 0);
+ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG))
+ amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
return 0;
}
@@ -670,6 +671,14 @@ static int jpeg_v4_0_set_interrupt_state(struct amdgpu_device *adev,
return 0;
}
+static int jpeg_v4_0_set_ras_interrupt_state(struct amdgpu_device *adev,
+ struct amdgpu_irq_src *source,
+ unsigned int type,
+ enum amdgpu_interrupt_state state)
+{
+ return 0;
+}
+
static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
@@ -680,10 +689,6 @@ static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev,
case VCN_4_0__SRCID__JPEG_DECODE:
amdgpu_fence_process(&adev->jpeg.inst->ring_dec);
break;
- case VCN_4_0__SRCID_DJPEG0_POISON:
- case VCN_4_0__SRCID_EJPEG0_POISON:
- amdgpu_jpeg_process_poison_irq(adev, source, entry);
- break;
default:
DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
entry->src_id, entry->src_data[0]);
@@ -753,10 +758,18 @@ static const struct amdgpu_irq_src_funcs jpeg_v4_0_irq_funcs = {
.process = jpeg_v4_0_process_interrupt,
};
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_ras_irq_funcs = {
+ .set = jpeg_v4_0_set_ras_interrupt_state,
+ .process = amdgpu_jpeg_process_poison_irq,
+};
+
static void jpeg_v4_0_set_irq_funcs(struct amdgpu_device *adev)
{
adev->jpeg.inst->irq.num_types = 1;
adev->jpeg.inst->irq.funcs = &jpeg_v4_0_irq_funcs;
+
+ adev->jpeg.inst->ras_poison_irq.num_types = 1;
+ adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_ras_irq_funcs;
}
const struct amdgpu_ip_block_version jpeg_v4_0_ip_block = {
--
2.34.1
