* [PATCH 0/3] io-pgtable-arm + drm/msm: Extend iova fault debugging
@ 2021-09-22 22:30 ` Rob Clark
0 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: linux-arm-msm, freedreno, Rob Clark, Akhil P Oommen, Dave Airlie,
Dmitry Baryshkov, Douglas Anderson, open list:IOMMU DRIVERS,
Joerg Roedel, Jonathan Marek, Jordan Crouse, Konrad Dybcio,
moderated list:ARM SMMU DRIVERS, open list, Robin Murphy,
Sai Prakash Ranjan, Sharat Masetty, Stephen Boyd,
Suravee Suthikulpanit, Will Deacon, Yangtao Li, Yong Wu
From: Rob Clark <robdclark@chromium.org>
This series extends io-pgtable-arm with a method to retrieve the page
table entries traversed in the process of address translation, and then
beefs up drm/msm gpu devcore dump to include this (and additional info)
in the devcore dump.
The motivation is tracking down an obscure iova fault triggered crash on
the address of the IB1 cmdstream. This is one of the few places where
the GPU address written into the cmdstream is soley under control of the
kernel mode driver, so I don't think it can be a userspace bug. The
logged cmdstream from the devcore's I've looked at look correct, and the
TTBR0 read back from arm-smmu agrees with the kernel emitted cmdstream.
Unfortunately it happens infrequently enough (something like once per
1000hrs of usage, from what I can tell from our telemetry) that actually
reproducing it with an instrumented debug kernel is not an option. So
further spiffying out the devcore dumps and hoping we can spot a clue is
the plan I'm shooting for.
See https://gitlab.freedesktop.org/drm/msm/-/issues/8 for more info on
the issue I'm trying to debug.
Rob Clark (3):
iommu/io-pgtable-arm: Add way to debug pgtable walk
drm/msm: Show all smmu info for iova fault devcore dumps
drm/msm: Extend gpu devcore dumps with pgtbl info
drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 2 +-
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 35 +++++++++++++++++-----
drivers/gpu/drm/msm/msm_gpu.c | 10 +++++++
drivers/gpu/drm/msm/msm_gpu.h | 10 ++++++-
drivers/gpu/drm/msm/msm_iommu.c | 17 +++++++++++
drivers/gpu/drm/msm/msm_mmu.h | 2 ++
drivers/iommu/io-pgtable-arm.c | 40 ++++++++++++++++++++-----
include/linux/io-pgtable.h | 9 ++++++
8 files changed, 107 insertions(+), 18 deletions(-)
--
2.31.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 0/3] io-pgtable-arm + drm/msm: Extend iova fault debugging
@ 2021-09-22 22:30 ` Rob Clark
0 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: Yangtao Li, Konrad Dybcio, Akhil P Oommen, Douglas Anderson,
Will Deacon, Rob Clark, Jonathan Marek,
moderated list:ARM SMMU DRIVERS, Dave Airlie, Joerg Roedel,
linux-arm-msm, Sharat Masetty, Stephen Boyd, Robin Murphy,
open list, open list:IOMMU DRIVERS, Dmitry Baryshkov, freedreno
From: Rob Clark <robdclark@chromium.org>
This series extends io-pgtable-arm with a method to retrieve the page
table entries traversed in the process of address translation, and then
beefs up drm/msm gpu devcore dump to include this (and additional info)
in the devcore dump.
The motivation is tracking down an obscure iova fault triggered crash on
the address of the IB1 cmdstream. This is one of the few places where
the GPU address written into the cmdstream is soley under control of the
kernel mode driver, so I don't think it can be a userspace bug. The
logged cmdstream from the devcore's I've looked at look correct, and the
TTBR0 read back from arm-smmu agrees with the kernel emitted cmdstream.
Unfortunately it happens infrequently enough (something like once per
1000hrs of usage, from what I can tell from our telemetry) that actually
reproducing it with an instrumented debug kernel is not an option. So
further spiffying out the devcore dumps and hoping we can spot a clue is
the plan I'm shooting for.
See https://gitlab.freedesktop.org/drm/msm/-/issues/8 for more info on
the issue I'm trying to debug.
Rob Clark (3):
iommu/io-pgtable-arm: Add way to debug pgtable walk
drm/msm: Show all smmu info for iova fault devcore dumps
drm/msm: Extend gpu devcore dumps with pgtbl info
drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 2 +-
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 35 +++++++++++++++++-----
drivers/gpu/drm/msm/msm_gpu.c | 10 +++++++
drivers/gpu/drm/msm/msm_gpu.h | 10 ++++++-
drivers/gpu/drm/msm/msm_iommu.c | 17 +++++++++++
drivers/gpu/drm/msm/msm_mmu.h | 2 ++
drivers/iommu/io-pgtable-arm.c | 40 ++++++++++++++++++++-----
include/linux/io-pgtable.h | 9 ++++++
8 files changed, 107 insertions(+), 18 deletions(-)
--
2.31.1
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 1/3] iommu/io-pgtable-arm: Add way to debug pgtable walk
2021-09-22 22:30 ` Rob Clark
(?)
@ 2021-09-22 22:30 ` Rob Clark
-1 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: linux-arm-msm, freedreno, Rob Clark, Will Deacon, Robin Murphy,
Joerg Roedel, Sai Prakash Ranjan, Yong Wu, Suravee Suthikulpanit,
moderated list:ARM SMMU DRIVERS, open list:IOMMU DRIVERS,
open list
From: Rob Clark <robdclark@chromium.org>
Add an io-pgtable method to retrieve the raw PTEs that would be
traversed for a given iova access.
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
drivers/iommu/io-pgtable-arm.c | 40 +++++++++++++++++++++++++++-------
include/linux/io-pgtable.h | 9 ++++++++
2 files changed, 41 insertions(+), 8 deletions(-)
diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c
index 87def58e79b5..5571d7203f11 100644
--- a/drivers/iommu/io-pgtable-arm.c
+++ b/drivers/iommu/io-pgtable-arm.c
@@ -638,38 +638,61 @@ static size_t arm_lpae_unmap(struct io_pgtable_ops *ops, unsigned long iova,
return __arm_lpae_unmap(data, gather, iova, size, data->start_level, ptep);
}
-static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
- unsigned long iova)
+static int arm_lpae_pgtable_walk(struct io_pgtable_ops *ops, unsigned long iova,
+ void *_ptes, int *num_ptes)
{
struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
arm_lpae_iopte pte, *ptep = data->pgd;
+ arm_lpae_iopte *ptes = _ptes;
+ int max_ptes = *num_ptes;
int lvl = data->start_level;
+ *num_ptes = 0;
+
do {
+ if (*num_ptes >= max_ptes)
+ return -ENOSPC;
+
/* Valid IOPTE pointer? */
if (!ptep)
- return 0;
+ return -EFAULT;
/* Grab the IOPTE we're interested in */
ptep += ARM_LPAE_LVL_IDX(iova, lvl, data);
pte = READ_ONCE(*ptep);
+ ptes[(*num_ptes)++] = pte;
+
/* Valid entry? */
if (!pte)
- return 0;
+ return -EFAULT;
/* Leaf entry? */
if (iopte_leaf(pte, lvl, data->iop.fmt))
- goto found_translation;
+ return 0;
/* Take it to the next level */
ptep = iopte_deref(pte, data);
} while (++lvl < ARM_LPAE_MAX_LEVELS);
- /* Ran out of page tables to walk */
- return 0;
+ return -EFAULT;
+}
+
+static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
+ unsigned long iova)
+{
+ struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+ arm_lpae_iopte pte, ptes[ARM_LPAE_MAX_LEVELS];
+ int lvl, num_ptes = ARM_LPAE_MAX_LEVELS;
+ int ret;
+
+ ret = arm_lpae_pgtable_walk(ops, iova, ptes, &num_ptes);
+ if (ret)
+ return 0;
+
+ pte = ptes[num_ptes - 1];
+ lvl = num_ptes - 1 + data->start_level;
-found_translation:
iova &= (ARM_LPAE_BLOCK_SIZE(lvl, data) - 1);
return iopte_to_paddr(pte, data) | iova;
}
@@ -752,6 +775,7 @@ arm_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg)
.map = arm_lpae_map,
.unmap = arm_lpae_unmap,
.iova_to_phys = arm_lpae_iova_to_phys,
+ .pgtable_walk = arm_lpae_pgtable_walk,
};
return data;
diff --git a/include/linux/io-pgtable.h b/include/linux/io-pgtable.h
index 4d40dfa75b55..6cba731ed8d3 100644
--- a/include/linux/io-pgtable.h
+++ b/include/linux/io-pgtable.h
@@ -145,6 +145,13 @@ struct io_pgtable_cfg {
* @map: Map a physically contiguous memory region.
* @unmap: Unmap a physically contiguous memory region.
* @iova_to_phys: Translate iova to physical address.
+ * @pgtable_walk: Return details of a page table walk for a given iova.
+ * This returns the array of PTEs in a format that is
+ * specific to the page table format. The number of
+ * PTEs can be format specific. The num_ptes parameter
+ * on input specifies the size of the ptes array, and
+ * on output the number of PTEs filled in (which depends
+ * on the number of PTEs walked to resolve the iova)
*
* These functions map directly onto the iommu_ops member functions with
* the same names.
@@ -156,6 +163,8 @@ struct io_pgtable_ops {
size_t size, struct iommu_iotlb_gather *gather);
phys_addr_t (*iova_to_phys)(struct io_pgtable_ops *ops,
unsigned long iova);
+ int (*pgtable_walk)(struct io_pgtable_ops *ops, unsigned long iova,
+ void *ptes, int *num_ptes);
};
/**
--
2.31.1
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 1/3] iommu/io-pgtable-arm: Add way to debug pgtable walk
@ 2021-09-22 22:30 ` Rob Clark
0 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: linux-arm-msm, freedreno, Rob Clark, Will Deacon, Robin Murphy,
Joerg Roedel, Sai Prakash Ranjan, Yong Wu, Suravee Suthikulpanit,
moderated list:ARM SMMU DRIVERS, open list:IOMMU DRIVERS,
open list
From: Rob Clark <robdclark@chromium.org>
Add an io-pgtable method to retrieve the raw PTEs that would be
traversed for a given iova access.
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
drivers/iommu/io-pgtable-arm.c | 40 +++++++++++++++++++++++++++-------
include/linux/io-pgtable.h | 9 ++++++++
2 files changed, 41 insertions(+), 8 deletions(-)
diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c
index 87def58e79b5..5571d7203f11 100644
--- a/drivers/iommu/io-pgtable-arm.c
+++ b/drivers/iommu/io-pgtable-arm.c
@@ -638,38 +638,61 @@ static size_t arm_lpae_unmap(struct io_pgtable_ops *ops, unsigned long iova,
return __arm_lpae_unmap(data, gather, iova, size, data->start_level, ptep);
}
-static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
- unsigned long iova)
+static int arm_lpae_pgtable_walk(struct io_pgtable_ops *ops, unsigned long iova,
+ void *_ptes, int *num_ptes)
{
struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
arm_lpae_iopte pte, *ptep = data->pgd;
+ arm_lpae_iopte *ptes = _ptes;
+ int max_ptes = *num_ptes;
int lvl = data->start_level;
+ *num_ptes = 0;
+
do {
+ if (*num_ptes >= max_ptes)
+ return -ENOSPC;
+
/* Valid IOPTE pointer? */
if (!ptep)
- return 0;
+ return -EFAULT;
/* Grab the IOPTE we're interested in */
ptep += ARM_LPAE_LVL_IDX(iova, lvl, data);
pte = READ_ONCE(*ptep);
+ ptes[(*num_ptes)++] = pte;
+
/* Valid entry? */
if (!pte)
- return 0;
+ return -EFAULT;
/* Leaf entry? */
if (iopte_leaf(pte, lvl, data->iop.fmt))
- goto found_translation;
+ return 0;
/* Take it to the next level */
ptep = iopte_deref(pte, data);
} while (++lvl < ARM_LPAE_MAX_LEVELS);
- /* Ran out of page tables to walk */
- return 0;
+ return -EFAULT;
+}
+
+static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
+ unsigned long iova)
+{
+ struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+ arm_lpae_iopte pte, ptes[ARM_LPAE_MAX_LEVELS];
+ int lvl, num_ptes = ARM_LPAE_MAX_LEVELS;
+ int ret;
+
+ ret = arm_lpae_pgtable_walk(ops, iova, ptes, &num_ptes);
+ if (ret)
+ return 0;
+
+ pte = ptes[num_ptes - 1];
+ lvl = num_ptes - 1 + data->start_level;
-found_translation:
iova &= (ARM_LPAE_BLOCK_SIZE(lvl, data) - 1);
return iopte_to_paddr(pte, data) | iova;
}
@@ -752,6 +775,7 @@ arm_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg)
.map = arm_lpae_map,
.unmap = arm_lpae_unmap,
.iova_to_phys = arm_lpae_iova_to_phys,
+ .pgtable_walk = arm_lpae_pgtable_walk,
};
return data;
diff --git a/include/linux/io-pgtable.h b/include/linux/io-pgtable.h
index 4d40dfa75b55..6cba731ed8d3 100644
--- a/include/linux/io-pgtable.h
+++ b/include/linux/io-pgtable.h
@@ -145,6 +145,13 @@ struct io_pgtable_cfg {
* @map: Map a physically contiguous memory region.
* @unmap: Unmap a physically contiguous memory region.
* @iova_to_phys: Translate iova to physical address.
+ * @pgtable_walk: Return details of a page table walk for a given iova.
+ * This returns the array of PTEs in a format that is
+ * specific to the page table format. The number of
+ * PTEs can be format specific. The num_ptes parameter
+ * on input specifies the size of the ptes array, and
+ * on output the number of PTEs filled in (which depends
+ * on the number of PTEs walked to resolve the iova)
*
* These functions map directly onto the iommu_ops member functions with
* the same names.
@@ -156,6 +163,8 @@ struct io_pgtable_ops {
size_t size, struct iommu_iotlb_gather *gather);
phys_addr_t (*iova_to_phys)(struct io_pgtable_ops *ops,
unsigned long iova);
+ int (*pgtable_walk)(struct io_pgtable_ops *ops, unsigned long iova,
+ void *ptes, int *num_ptes);
};
/**
--
2.31.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 1/3] iommu/io-pgtable-arm: Add way to debug pgtable walk
@ 2021-09-22 22:30 ` Rob Clark
0 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: Rob Clark, Will Deacon, linux-arm-msm, Robin Murphy, open list,
open list:IOMMU DRIVERS, moderated list:ARM SMMU DRIVERS,
freedreno
From: Rob Clark <robdclark@chromium.org>
Add an io-pgtable method to retrieve the raw PTEs that would be
traversed for a given iova access.
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
drivers/iommu/io-pgtable-arm.c | 40 +++++++++++++++++++++++++++-------
include/linux/io-pgtable.h | 9 ++++++++
2 files changed, 41 insertions(+), 8 deletions(-)
diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c
index 87def58e79b5..5571d7203f11 100644
--- a/drivers/iommu/io-pgtable-arm.c
+++ b/drivers/iommu/io-pgtable-arm.c
@@ -638,38 +638,61 @@ static size_t arm_lpae_unmap(struct io_pgtable_ops *ops, unsigned long iova,
return __arm_lpae_unmap(data, gather, iova, size, data->start_level, ptep);
}
-static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
- unsigned long iova)
+static int arm_lpae_pgtable_walk(struct io_pgtable_ops *ops, unsigned long iova,
+ void *_ptes, int *num_ptes)
{
struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
arm_lpae_iopte pte, *ptep = data->pgd;
+ arm_lpae_iopte *ptes = _ptes;
+ int max_ptes = *num_ptes;
int lvl = data->start_level;
+ *num_ptes = 0;
+
do {
+ if (*num_ptes >= max_ptes)
+ return -ENOSPC;
+
/* Valid IOPTE pointer? */
if (!ptep)
- return 0;
+ return -EFAULT;
/* Grab the IOPTE we're interested in */
ptep += ARM_LPAE_LVL_IDX(iova, lvl, data);
pte = READ_ONCE(*ptep);
+ ptes[(*num_ptes)++] = pte;
+
/* Valid entry? */
if (!pte)
- return 0;
+ return -EFAULT;
/* Leaf entry? */
if (iopte_leaf(pte, lvl, data->iop.fmt))
- goto found_translation;
+ return 0;
/* Take it to the next level */
ptep = iopte_deref(pte, data);
} while (++lvl < ARM_LPAE_MAX_LEVELS);
- /* Ran out of page tables to walk */
- return 0;
+ return -EFAULT;
+}
+
+static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
+ unsigned long iova)
+{
+ struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+ arm_lpae_iopte pte, ptes[ARM_LPAE_MAX_LEVELS];
+ int lvl, num_ptes = ARM_LPAE_MAX_LEVELS;
+ int ret;
+
+ ret = arm_lpae_pgtable_walk(ops, iova, ptes, &num_ptes);
+ if (ret)
+ return 0;
+
+ pte = ptes[num_ptes - 1];
+ lvl = num_ptes - 1 + data->start_level;
-found_translation:
iova &= (ARM_LPAE_BLOCK_SIZE(lvl, data) - 1);
return iopte_to_paddr(pte, data) | iova;
}
@@ -752,6 +775,7 @@ arm_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg)
.map = arm_lpae_map,
.unmap = arm_lpae_unmap,
.iova_to_phys = arm_lpae_iova_to_phys,
+ .pgtable_walk = arm_lpae_pgtable_walk,
};
return data;
diff --git a/include/linux/io-pgtable.h b/include/linux/io-pgtable.h
index 4d40dfa75b55..6cba731ed8d3 100644
--- a/include/linux/io-pgtable.h
+++ b/include/linux/io-pgtable.h
@@ -145,6 +145,13 @@ struct io_pgtable_cfg {
* @map: Map a physically contiguous memory region.
* @unmap: Unmap a physically contiguous memory region.
* @iova_to_phys: Translate iova to physical address.
+ * @pgtable_walk: Return details of a page table walk for a given iova.
+ * This returns the array of PTEs in a format that is
+ * specific to the page table format. The number of
+ * PTEs can be format specific. The num_ptes parameter
+ * on input specifies the size of the ptes array, and
+ * on output the number of PTEs filled in (which depends
+ * on the number of PTEs walked to resolve the iova)
*
* These functions map directly onto the iommu_ops member functions with
* the same names.
@@ -156,6 +163,8 @@ struct io_pgtable_ops {
size_t size, struct iommu_iotlb_gather *gather);
phys_addr_t (*iova_to_phys)(struct io_pgtable_ops *ops,
unsigned long iova);
+ int (*pgtable_walk)(struct io_pgtable_ops *ops, unsigned long iova,
+ void *ptes, int *num_ptes);
};
/**
--
2.31.1
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 2/3] drm/msm: Show all smmu info for iova fault devcore dumps
2021-09-22 22:30 ` Rob Clark
` (2 preceding siblings ...)
(?)
@ 2021-09-22 22:30 ` Rob Clark
-1 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: linux-arm-msm, freedreno, Rob Clark, Rob Clark, Sean Paul,
David Airlie, Daniel Vetter, Jordan Crouse, Akhil P Oommen,
Jonathan Marek, Sai Prakash Ranjan, Sharat Masetty,
Douglas Anderson, Dave Airlie, Stephen Boyd, Konrad Dybcio,
open list
From: Rob Clark <robdclark@chromium.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 2 +-
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 25 +++++++++++++++++--------
drivers/gpu/drm/msm/msm_gpu.h | 2 +-
3 files changed, 19 insertions(+), 10 deletions(-)
diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
index 4ac652c35c43..f6a4dbef796b 100644
--- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
@@ -1269,7 +1269,7 @@ static int a6xx_fault_handler(void *arg, unsigned long iova, int flags, void *da
/* Turn off the hangcheck timer to keep it from bothering us */
del_timer(&gpu->hangcheck_timer);
- gpu->fault_info.ttbr0 = info->ttbr0;
+ gpu->fault_info.smmu_info = *info;
gpu->fault_info.iova = iova;
gpu->fault_info.flags = flags;
gpu->fault_info.type = type;
diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index 748665232d29..42e522a60623 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -685,19 +685,28 @@ void adreno_show(struct msm_gpu *gpu, struct msm_gpu_state *state,
adreno_gpu->rev.major, adreno_gpu->rev.minor,
adreno_gpu->rev.patchid);
/*
- * If this is state collected due to iova fault, so fault related info
+ * If this is state collected due to iova fault, show fault related
+ * info
*
- * TTBR0 would not be zero, so this is a good way to distinguish
+ * TTBR0 would not be zero in this case, so this is a good way to
+ * distinguish
*/
- if (state->fault_info.ttbr0) {
+ if (state->fault_info.smmu_info.ttbr0) {
const struct msm_gpu_fault_info *info = &state->fault_info;
+ const struct adreno_smmu_fault_info *smmu_info = &info->smmu_info;
drm_puts(p, "fault-info:\n");
- drm_printf(p, " - ttbr0=%.16llx\n", info->ttbr0);
- drm_printf(p, " - iova=%.16lx\n", info->iova);
- drm_printf(p, " - dir=%s\n", info->flags & IOMMU_FAULT_WRITE ? "WRITE" : "READ");
- drm_printf(p, " - type=%s\n", info->type);
- drm_printf(p, " - source=%s\n", info->block);
+ drm_printf(p, " - far: %.16llx\n", smmu_info->far);
+ drm_printf(p, " - ttbr0: %.16llx\n", smmu_info->ttbr0);
+ drm_printf(p, " - contextidr: %.8x\n", smmu_info->contextidr);
+ drm_printf(p, " - fsr: %.8x\n", smmu_info->fsr);
+ drm_printf(p, " - fsynr0: %.8x\n", smmu_info->fsynr0);
+ drm_printf(p, " - fsynr1: %.8x\n", smmu_info->fsynr1);
+ drm_printf(p, " - cbfrsynra: %.8x\n", smmu_info->cbfrsynra);
+ drm_printf(p, " - iova: %.16lx\n", info->iova);
+ drm_printf(p, " - dir: %s\n", info->flags & IOMMU_FAULT_WRITE ? "WRITE" : "READ");
+ drm_printf(p, " - type: %s\n", info->type);
+ drm_printf(p, " - source: %s\n", info->block);
}
drm_printf(p, "rbbm-status: 0x%08x\n", state->rbbm_status);
diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h
index e031c9b495ed..a7a5a53536a8 100644
--- a/drivers/gpu/drm/msm/msm_gpu.h
+++ b/drivers/gpu/drm/msm/msm_gpu.h
@@ -73,7 +73,7 @@ struct msm_gpu_funcs {
/* Additional state for iommu faults: */
struct msm_gpu_fault_info {
- u64 ttbr0;
+ struct adreno_smmu_fault_info smmu_info;
unsigned long iova;
int flags;
const char *type;
--
2.31.1
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 3/3] drm/msm: Extend gpu devcore dumps with pgtbl info
2021-09-22 22:30 ` Rob Clark
` (3 preceding siblings ...)
(?)
@ 2021-09-22 22:30 ` Rob Clark
-1 siblings, 0 replies; 8+ messages in thread
From: Rob Clark @ 2021-09-22 22:30 UTC (permalink / raw)
To: dri-devel
Cc: linux-arm-msm, freedreno, Rob Clark, Rob Clark, Sean Paul,
David Airlie, Daniel Vetter, Sai Prakash Ranjan, Jordan Crouse,
Jonathan Marek, Dave Airlie, Sharat Masetty, Akhil P Oommen,
Konrad Dybcio, open list
From: Rob Clark <robdclark@chromium.org>
In the case of iova fault triggered devcore dumps, include additional
debug information based on what we think is the current page tables,
including the TTBR0 value (which should match what we have in
adreno_smmu_fault_info unless things have gone horribly wrong), and
the pagetable entries traversed in the process of resolving the
faulting iova.
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 10 ++++++++++
drivers/gpu/drm/msm/msm_gpu.c | 10 ++++++++++
drivers/gpu/drm/msm/msm_gpu.h | 8 ++++++++
drivers/gpu/drm/msm/msm_iommu.c | 17 +++++++++++++++++
drivers/gpu/drm/msm/msm_mmu.h | 2 ++
5 files changed, 47 insertions(+)
diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index 42e522a60623..d3718982be77 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -707,6 +707,16 @@ void adreno_show(struct msm_gpu *gpu, struct msm_gpu_state *state,
drm_printf(p, " - dir: %s\n", info->flags & IOMMU_FAULT_WRITE ? "WRITE" : "READ");
drm_printf(p, " - type: %s\n", info->type);
drm_printf(p, " - source: %s\n", info->block);
+
+ /* Information extracted from what we think are the current
+ * pgtables. Hopefully the TTBR0 matches what we've extracted
+ * from the SMMU registers in smmu_info!
+ */
+ drm_puts(p, "pgtable-fault-info:\n");
+ drm_printf(p, " - ttbr0: %.16llx\n", info->pgtbl_ttbr0);
+ drm_printf(p, " - asid: %d\n", info->asid);
+ drm_printf(p, " - ptes: %.16llx %.16llx %.16llx %.16llx\n",
+ info->ptes[0], info->ptes[1], info->ptes[2], info->ptes[3]);
}
drm_printf(p, "rbbm-status: 0x%08x\n", state->rbbm_status);
diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c
index 8a3a592da3a4..d1a16642ecd5 100644
--- a/drivers/gpu/drm/msm/msm_gpu.c
+++ b/drivers/gpu/drm/msm/msm_gpu.c
@@ -284,6 +284,16 @@ static void msm_gpu_crashstate_capture(struct msm_gpu *gpu,
if (submit) {
int i, nr = 0;
+ if (state->fault_info.smmu_info.ttbr0) {
+ struct msm_gpu_fault_info *info = &state->fault_info;
+ struct msm_mmu *mmu = submit->aspace->mmu;
+
+ msm_iommu_pagetable_params(mmu, &info->pgtbl_ttbr0,
+ &info->asid);
+ msm_iommu_pagetable_walk(mmu, info->iova, info->ptes,
+ ARRAY_SIZE(info->ptes));
+ }
+
/* count # of buffers to dump: */
for (i = 0; i < submit->nr_bos; i++)
if (should_dump(submit, i))
diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h
index a7a5a53536a8..32a859307e81 100644
--- a/drivers/gpu/drm/msm/msm_gpu.h
+++ b/drivers/gpu/drm/msm/msm_gpu.h
@@ -78,6 +78,14 @@ struct msm_gpu_fault_info {
int flags;
const char *type;
const char *block;
+
+ /* Information about what we think/expect is the current SMMU state,
+ * for example expected_ttbr0 should match smmu_info.ttbr0 which
+ * was read back from SMMU registers.
+ */
+ u64 pgtbl_ttbr0;
+ u64 ptes[4];
+ int asid;
};
/**
diff --git a/drivers/gpu/drm/msm/msm_iommu.c b/drivers/gpu/drm/msm/msm_iommu.c
index eed2a762e9dd..1bd985b56e35 100644
--- a/drivers/gpu/drm/msm/msm_iommu.c
+++ b/drivers/gpu/drm/msm/msm_iommu.c
@@ -116,6 +116,23 @@ int msm_iommu_pagetable_params(struct msm_mmu *mmu,
return 0;
}
+int msm_iommu_pagetable_walk(struct msm_mmu *mmu, unsigned long iova,
+ u64 *ptes, int num_ptes)
+{
+ struct msm_iommu_pagetable *pagetable;
+
+ if (mmu->type != MSM_MMU_IOMMU_PAGETABLE)
+ return -EINVAL;
+
+ pagetable = to_pagetable(mmu);
+
+ if (!pagetable->pgtbl_ops->pgtable_walk)
+ return -EINVAL;
+
+ return pagetable->pgtbl_ops->pgtable_walk(pagetable->pgtbl_ops, iova,
+ ptes, &num_ptes);
+}
+
static const struct msm_mmu_funcs pagetable_funcs = {
.map = msm_iommu_pagetable_map,
.unmap = msm_iommu_pagetable_unmap,
diff --git a/drivers/gpu/drm/msm/msm_mmu.h b/drivers/gpu/drm/msm/msm_mmu.h
index de158e1bf765..519b749c61af 100644
--- a/drivers/gpu/drm/msm/msm_mmu.h
+++ b/drivers/gpu/drm/msm/msm_mmu.h
@@ -58,5 +58,7 @@ void msm_gpummu_params(struct msm_mmu *mmu, dma_addr_t *pt_base,
int msm_iommu_pagetable_params(struct msm_mmu *mmu, phys_addr_t *ttbr,
int *asid);
+int msm_iommu_pagetable_walk(struct msm_mmu *mmu, unsigned long iova,
+ u64 *ptes, int num_ptes);
#endif /* __MSM_MMU_H__ */
--
2.31.1
^ permalink raw reply related [flat|nested] 8+ messages in thread