[PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Alex Deucher]

From: Christian König <christian.koenig@amd.com>

We need to grab the lock of the BO or otherwise can run into a crash
when we try to inspect the current location.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Guchun Chen <guchun.chen@amd.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
Cc: stable@vger.kernel.org # 6.3.x
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 69 +++++++++++++++-----------
 1 file changed, 39 insertions(+), 30 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 5b3a70becbdf..a252a206f37b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -920,42 +920,51 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	return r;
 }
 
+static void amdgpu_vm_bo_get_memory(struct amdgpu_bo_va *bo_va,
+				    struct amdgpu_mem_stats *stats)
+{
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+
+	if (!bo)
+		return;
+
+	/*
+	 * For now ignore BOs which are currently locked and potentially
+	 * changing their location.
+	 */
+	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv &&
+	    !dma_resv_trylock(bo->tbo.base.resv))
+		return;
+
+	amdgpu_bo_get_memory(bo, stats);
+	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv)
+	    dma_resv_unlock(bo->tbo.base.resv);
+}
+
 void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
 			  struct amdgpu_mem_stats *stats)
 {
 	struct amdgpu_bo_va *bo_va, *tmp;
 
 	spin_lock(&vm->status_lock);
-	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
+	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
 	spin_unlock(&vm->status_lock);
 }
 
-- 
2.41.0

[PATCH 2/9] drm/amdgpu: make sure that BOs have a backing store

[Alex Deucher]

From: Christian König <christian.koenig@amd.com>

It's perfectly possible that the BO is about to be destroyed and doesn't
have a backing store associated with it.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Guchun Chen <guchun.chen@amd.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit ca0b954a4315ca2228001c439ae1062561c81989)
Cc: stable@vger.kernel.org # 6.3.x
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index a70103ac0026..46557bbbc18a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -1266,8 +1266,12 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 void amdgpu_bo_get_memory(struct amdgpu_bo *bo,
 			  struct amdgpu_mem_stats *stats)
 {
-	unsigned int domain;
 	uint64_t size = amdgpu_bo_size(bo);
+	unsigned int domain;
+
+	/* Abort if the BO doesn't currently have a backing store */
+	if (!bo->tbo.resource)
+		return;
 
 	domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type);
 	switch (domain) {
-- 
2.41.0

[PATCH 3/9] drm/amdgpu: Skip mark offset for high priority rings

[Alex Deucher]

From: Jiadong Zhu <Jiadong.Zhu@amd.com>

Only low priority rings are using chunks to save the offset.
Bypass the mark offset callings from high priority rings.

Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit ef3c36a6e025e9b16ca3321479ba016841fa17a0)
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
index 73516abef662..b779ee4bbaa7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
@@ -423,6 +423,9 @@ void amdgpu_sw_ring_ib_mark_offset(struct amdgpu_ring *ring, enum amdgpu_ring_mu
 	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
 	unsigned offset;
 
+	if (ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT)
+		return;
+
 	offset = ring->wptr & ring->buf_mask;
 
 	amdgpu_ring_mux_ib_mark_offset(mux, ring, offset, type);
-- 
2.41.0

[PATCH 4/9] drm/amd/pm: revise the ASPM settings for thunderbolt attached scenario

[Alex Deucher]

From: Evan Quan <evan.quan@amd.com>

Also, correct the comment for NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT
as 0x0000000E stands for 400ms instead of 4ms.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit fd21987274463a439c074b8f3c93d3b132e4c031)
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c
index aa761ff3a5fa..7ba47fc1917b 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c
@@ -346,7 +346,7 @@ static void nbio_v2_3_init_registers(struct amdgpu_device *adev)
 
 #define NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT		0x00000000 // off by default, no gains over L1
 #define NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT		0x00000009 // 1=1us, 9=1ms
-#define NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT	0x0000000E // 4ms
+#define NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT	0x0000000E // 400ms
 
 static void nbio_v2_3_enable_aspm(struct amdgpu_device *adev,
 				  bool enable)
@@ -479,9 +479,12 @@ static void nbio_v2_3_program_aspm(struct amdgpu_device *adev)
 		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP5, data);
 
 	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
-	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
-	data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
-	data |= 0x1 << PCIE_LC_CNTL__LC_PMI_TO_L1_DIS__SHIFT;
+	data |= NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+	if (pci_is_thunderbolt_attached(adev->pdev))
+		data |= NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT  << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	else
+		data |= NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
 	if (def != data)
 		WREG32_PCIE(smnPCIE_LC_CNTL, data);
 
-- 
2.41.0

[PATCH 5/9] drm/amdgpu/sdma4: set align mask to 255

[Alex Deucher]

The wptr needs to be incremented at at least 64 dword intervals,
use 256 to align with windows.  This should fix potential hangs
with unaligned updates.

Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Aaron Liu <aaron.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit e5df16d9428f5c6d2d0b1eff244d6c330ba9ef3a)
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c   | 4 ++--
 drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index 9295ac7edd56..d35c8a33d06d 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -2306,7 +2306,7 @@ const struct amd_ip_funcs sdma_v4_0_ip_funcs = {
 
 static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
@@ -2338,7 +2338,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 
 static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
index 64dcaa2670dd..ac7aa8631f6a 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -1740,7 +1740,7 @@ const struct amd_ip_funcs sdma_v4_4_2_ip_funcs = {
 
 static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
@@ -1771,7 +1771,7 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {
 
 static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
-- 
2.41.0

[PATCH 6/9] drm/amd/pm: add abnormal fan detection for smu 13.0.0

[Alex Deucher]

From: Kenneth Feng <kenneth.feng@amd.com>

add abnormal fan detection for smu 13.0.0

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 2da0036ea99bccb27f7fe3cf2aa2900860e9be46)
Cc: stable@vger.kernel.org # 6.1.x
---
 drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
index 08577d1b84ec..c42c0c1446f4 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
@@ -1300,6 +1300,7 @@ static int smu_v13_0_0_get_thermal_temperature_range(struct smu_context *smu,
 	range->mem_emergency_max = (pptable->SkuTable.TemperatureLimit[TEMP_MEM] + CTF_OFFSET_MEM)*
 		SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	range->software_shutdown_temp = powerplay_table->software_shutdown_temp;
+	range->software_shutdown_temp_offset = pptable->SkuTable.FanAbnormalTempLimitOffset;
 
 	return 0;
 }
-- 
2.41.0

[PATCH 7/9] drm/amdgpu: check RAS irq existence for VCN/JPEG

[Alex Deucher]

From: Tao Zhou <tao.zhou1@amd.com>

No RAS irq is allowed.

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 4ff96bcc0d40b66bf3ddd6010830e9a4f9b85d53)
Cc: stable@vger.kernel.org # 6.1.x
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c | 3 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c  | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
index 4fa019c8aefc..fb9251d9c899 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -251,7 +251,8 @@ int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *
 
 	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))
+			if (adev->jpeg.harvest_config & (1 << i) ||
+			    !adev->jpeg.inst[i].ras_poison_irq.funcs)
 				continue;
 
 			r = amdgpu_irq_get(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index 2d94f1b63bd6..b46a5771c3ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -1191,7 +1191,8 @@ int amdgpu_vcn_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *r
 
 	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
 		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
-			if (adev->vcn.harvest_config & (1 << i))
+			if (adev->vcn.harvest_config & (1 << i) ||
+			    !adev->vcn.inst[i].ras_poison_irq.funcs)
 				continue;
 
 			r = amdgpu_irq_get(adev, &adev->vcn.inst[i].ras_poison_irq, 0);
-- 
2.41.0

[PATCH 8/9] drm/amdgpu: fix number of fence calculations

[Alex Deucher]

From: Christian König <christian.koenig@amd.com>

Since adding gang submit we need to take the gang size into account
while reserving fences.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 4624459c84d7 ("drm/amdgpu: add gang submit frontend v6")
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 570b295248b00c3cf4cf59e397de5cb2361e10c2)
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 2eb2c66843a8..5612caf77dd6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -133,9 +133,6 @@ static int amdgpu_cs_p1_user_fence(struct amdgpu_cs_parser *p,
 	bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
 	p->uf_entry.priority = 0;
 	p->uf_entry.tv.bo = &bo->tbo;
-	/* One for TTM and two for the CS job */
-	p->uf_entry.tv.num_shared = 3;
-
 	drm_gem_object_put(gobj);
 
 	size = amdgpu_bo_size(bo);
@@ -882,15 +879,19 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	mutex_lock(&p->bo_list->bo_list_mutex);
 
-	/* One for TTM and one for the CS job */
+	/* One for TTM and one for each CS job */
 	amdgpu_bo_list_for_each_entry(e, p->bo_list)
-		e->tv.num_shared = 2;
+		e->tv.num_shared = 1 + p->gang_size;
+	p->uf_entry.tv.num_shared = 1 + p->gang_size;
 
 	amdgpu_bo_list_get_list(p->bo_list, &p->validated);
 
 	INIT_LIST_HEAD(&duplicates);
 	amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
 
+	/* Two for VM updates, one for TTM and one for each CS job */
+	p->vm_pd.tv.num_shared = 3 + p->gang_size;
+
 	if (p->uf_entry.tv.bo && !ttm_to_amdgpu_bo(p->uf_entry.tv.bo)->parent)
 		list_add(&p->uf_entry.tv.head, &p->validated);
 
-- 
2.41.0

[PATCH 9/9] drm/amd: Don't try to enable secure display TA multiple times

[Alex Deucher]

From: Mario Limonciello <mario.limonciello@amd.com>

If the securedisplay TA failed to load the first time, it's unlikely
to work again after a suspend/resume cycle or reset cycle and it appears
to be causing problems in futher attempts.

Fixes: e42dfa66d592 ("drm/amdgpu: Add secure display TA load for Renoir")
Reported-by: Filip Hejsek <filip.hejsek@gmail.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2633
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 5c6d52ff4b61e5267b25be714eb5a9ba2a338199)
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index a150b7a4b4aa..e4757a2807d9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -1947,6 +1947,8 @@ static int psp_securedisplay_initialize(struct psp_context *psp)
 		psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status);
 		dev_err(psp->adev->dev, "SECUREDISPLAY: query securedisplay TA failed. ret 0x%x\n",
 			securedisplay_cmd->securedisplay_out_message.query_ta.query_cmd_ret);
+		/* don't try again */
+		psp->securedisplay_context.context.bin_desc.size_bytes = 0;
 	}
 
 	return 0;
-- 
2.41.0

Re: [PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Mario Limonciello]

On 7/7/23 10:07, Alex Deucher wrote:
> From: Christian König <christian.koenig@amd.com>
> 
> We need to grab the lock of the BO or otherwise can run into a crash
> when we try to inspect the current location.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> Acked-by: Guchun Chen <guchun.chen@amd.com>
> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> (cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
> Cc: stable@vger.kernel.org # 6.3.x
> ---

Greg,

Just want to make sure you saw these 9 commits as you're processing 
queues since they don't stand out as being sent directly to stable.

>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 69 +++++++++++++++-----------
>   1 file changed, 39 insertions(+), 30 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 5b3a70becbdf..a252a206f37b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -920,42 +920,51 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
>   	return r;
>   }
>   
> +static void amdgpu_vm_bo_get_memory(struct amdgpu_bo_va *bo_va,
> +				    struct amdgpu_mem_stats *stats)
> +{
> +	struct amdgpu_vm *vm = bo_va->base.vm;
> +	struct amdgpu_bo *bo = bo_va->base.bo;
> +
> +	if (!bo)
> +		return;
> +
> +	/*
> +	 * For now ignore BOs which are currently locked and potentially
> +	 * changing their location.
> +	 */
> +	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv &&
> +	    !dma_resv_trylock(bo->tbo.base.resv))
> +		return;
> +
> +	amdgpu_bo_get_memory(bo, stats);
> +	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv)
> +	    dma_resv_unlock(bo->tbo.base.resv);
> +}
> +
>   void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
>   			  struct amdgpu_mem_stats *stats)
>   {
>   	struct amdgpu_bo_va *bo_va, *tmp;
>   
>   	spin_lock(&vm->status_lock);
> -	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> -	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> -	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> -	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> -	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> -	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) {
> -		if (!bo_va->base.bo)
> -			continue;
> -		amdgpu_bo_get_memory(bo_va->base.bo, stats);
> -	}
> +	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
> +
> +	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
> +
> +	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
> +
> +	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
> +
> +	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
> +
> +	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status)
> +		amdgpu_vm_bo_get_memory(bo_va, stats);
>   	spin_unlock(&vm->status_lock);
>   }
>   

Re: [PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Greg Kroah-Hartman]

On Tue, Jul 11, 2023 at 04:40:44PM -0500, Mario Limonciello wrote:
> On 7/7/23 10:07, Alex Deucher wrote:
> > From: Christian König <christian.koenig@amd.com>
> > 
> > We need to grab the lock of the BO or otherwise can run into a crash
> > when we try to inspect the current location.
> > 
> > Signed-off-by: Christian König <christian.koenig@amd.com>
> > Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> > Acked-by: Guchun Chen <guchun.chen@amd.com>
> > Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
> > Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> > (cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
> > Cc: stable@vger.kernel.org # 6.3.x
> > ---
> 
> Greg,
> 
> Just want to make sure you saw these 9 commits as you're processing queues
> since they don't stand out as being sent directly to stable.

Thanks for the pointer, no, I had missed them in the flood of stable
patches recently.  I have many hundreds of other patches to still get
to, and these are now in that review queue as well.

greg k-h

Re: [PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Greg KH]

On Fri, Jul 07, 2023 at 11:07:26AM -0400, Alex Deucher wrote:
> From: Christian König <christian.koenig@amd.com>
> 
> We need to grab the lock of the BO or otherwise can run into a crash
> when we try to inspect the current location.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> Acked-by: Guchun Chen <guchun.chen@amd.com>
> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> (cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
> Cc: stable@vger.kernel.org # 6.3.x
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 69 +++++++++++++++-----------
>  1 file changed, 39 insertions(+), 30 deletions(-)

I've applied the first 7 patches here to 6.4.y, which I am guessing is
where you want them applied to, yet you didn't really say?

The last 2 did not apply :(

And some of these should go into 6.1.y also?  Please send a patch series
and give me a hint as to where they should be applied to next time so I
don't have to guess...

thanks,

greg k-h

Re: [PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Mario Limonciello]

On 7/16/23 14:16, Greg KH wrote:
> On Fri, Jul 07, 2023 at 11:07:26AM -0400, Alex Deucher wrote:
>> From: Christian König <christian.koenig@amd.com>
>>
>> We need to grab the lock of the BO or otherwise can run into a crash
>> when we try to inspect the current location.
>>
>> Signed-off-by: Christian König <christian.koenig@amd.com>
>> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
>> Acked-by: Guchun Chen <guchun.chen@amd.com>
>> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
>> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
>> (cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
>> Cc: stable@vger.kernel.org # 6.3.x
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 69 +++++++++++++++-----------
>>   1 file changed, 39 insertions(+), 30 deletions(-)
> 
> I've applied the first 7 patches here to 6.4.y, which I am guessing is
> where you want them applied to, yet you didn't really say?
> 
> The last 2 did not apply :(
> 
> And some of these should go into 6.1.y also?  Please send a patch series
> and give me a hint as to where they should be applied to next time so I
> don't have to guess...
> 
> thanks,
> 
> greg k-h

In this case the individual patches with specific requirements have:

Cc: stable@vger.kernel.org # version

They were sent before 6.3 went EOL, so here is the updated summary from 
them:
6.4.y:
1, 2, 3, 4, 5, 6, 7, 8, 9

6.1.y:
3, 4, 5, 6, 7, 8, 9

3 is particularly important for 6.1.y as we have active regressions 
reported related to it on 6.1.y.

So can you please take 3-7 to 6.1.y and I'll look more closely at what 
is wrong with 8 and 9 on 6.1.y and 6.4.y and resend them?

Re: [PATCH 1/9] drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory

[Greg KH]

On Sun, Jul 16, 2023 at 02:22:36PM -0500, Mario Limonciello wrote:
> On 7/16/23 14:16, Greg KH wrote:
> > On Fri, Jul 07, 2023 at 11:07:26AM -0400, Alex Deucher wrote:
> > > From: Christian König <christian.koenig@amd.com>
> > > 
> > > We need to grab the lock of the BO or otherwise can run into a crash
> > > when we try to inspect the current location.
> > > 
> > > Signed-off-by: Christian König <christian.koenig@amd.com>
> > > Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> > > Acked-by: Guchun Chen <guchun.chen@amd.com>
> > > Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
> > > Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> > > (cherry picked from commit e2ad8e2df432498b1cee2af04df605723f4d75e6)
> > > Cc: stable@vger.kernel.org # 6.3.x
> > > ---
> > >   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 69 +++++++++++++++-----------
> > >   1 file changed, 39 insertions(+), 30 deletions(-)
> > 
> > I've applied the first 7 patches here to 6.4.y, which I am guessing is
> > where you want them applied to, yet you didn't really say?
> > 
> > The last 2 did not apply :(
> > 
> > And some of these should go into 6.1.y also?  Please send a patch series
> > and give me a hint as to where they should be applied to next time so I
> > don't have to guess...
> > 
> > thanks,
> > 
> > greg k-h
> 
> In this case the individual patches with specific requirements have:
> 
> Cc: stable@vger.kernel.org # version
> 
> They were sent before 6.3 went EOL, so here is the updated summary from
> them:
> 6.4.y:
> 1, 2, 3, 4, 5, 6, 7, 8, 9
> 
> 6.1.y:
> 3, 4, 5, 6, 7, 8, 9
> 
> 3 is particularly important for 6.1.y as we have active regressions reported
> related to it on 6.1.y.
> 
> So can you please take 3-7 to 6.1.y and I'll look more closely at what is
> wrong with 8 and 9 on 6.1.y and 6.4.y and resend them?

I can't really pick out these for 6.1 from the larger series as I'm
drowning in patches at the moment.  Please send a backported series and
I'll be glad to queue that up.

thanks,

greg k-h