[PATCH AUTOSEL 4.19 26/38] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index 9c9a5b12f92f..7c6cd00d0fca 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -469,7 +469,7 @@ void debug_dma_dump_mappings(struct device *dev) * At any time debug_dma_assert_idle() can be called to trigger a * warning if any cachelines in the given page are in the active set. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 37/55] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index 4dc3bbfd3e3f..1c133f610f59 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -450,7 +450,7 @@ void debug_dma_dump_mappings(struct device *dev) * At any time debug_dma_assert_idle() can be called to trigger a * warning if any cachelines in the given page are in the active set. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 50/76] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index f8ae54679865..ee7da1f2462f 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -448,7 +448,7 @@ void debug_dma_dump_mappings(struct device *dev) * other hand, consumes a single dma_debug_entry, but inserts 'nents' * entries into the tree. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 01/76] iommu/vt-d: Add RPLS to quirk list to skip TE disabling
From: Tejas Upadhyay [ Upstream commit 0a967f5bfd9134b89681cae58deb222e20840e76 ] The VT-d spec requires (10.4.4 Global Command Register, TE field) that: Hardware implementations supporting DMA draining must drain any in-flight DMA read/write requests queued within the Root-Complex before completing the translation enable command and reflecting the status of the command through the TES field in the Global Status register. Unfortunately, some integrated graphic devices fail to do so after some kind of power state transition. As the result, the system might stuck in iommu_disable_translati on(), waiting for the completion of TE transition. This adds RPLS to a quirk list for those devices and skips TE disabling if the qurik hits. Link: https://gitlab.freedesktop.org/drm/intel/-/issues/4898 Tested-by: Raviteja Goud Talla Cc: Rodrigo Vivi Acked-by: Lu Baolu Signed-off-by: Tejas Upadhyay Reviewed-by: Rodrigo Vivi Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20220302043256.191529-1-tejaskumarx.surendrakumar.upadh...@intel.com Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index 21749859ad45..477dde39823c 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -6296,7 +6296,7 @@ static void quirk_igfx_skip_te_disable(struct pci_dev *dev) ver = (dev->device >> 8) & 0xff; if (ver != 0x45 && ver != 0x46 && ver != 0x4c && ver != 0x4e && ver != 0x8a && ver != 0x98 && - ver != 0x9a) + ver != 0x9a && ver != 0xa7) return; if (risky_device(dev)) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 069/109] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index f8ff598596b8..ac740630c79c 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -448,7 +448,7 @@ void debug_dma_dump_mappings(struct device *dev) * other hand, consumes a single dma_debug_entry, but inserts 'nents' * entries into the tree. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 001/109] iommu/vt-d: Add RPLS to quirk list to skip TE disabling
From: Tejas Upadhyay [ Upstream commit 0a967f5bfd9134b89681cae58deb222e20840e76 ] The VT-d spec requires (10.4.4 Global Command Register, TE field) that: Hardware implementations supporting DMA draining must drain any in-flight DMA read/write requests queued within the Root-Complex before completing the translation enable command and reflecting the status of the command through the TES field in the Global Status register. Unfortunately, some integrated graphic devices fail to do so after some kind of power state transition. As the result, the system might stuck in iommu_disable_translati on(), waiting for the completion of TE transition. This adds RPLS to a quirk list for those devices and skips TE disabling if the qurik hits. Link: https://gitlab.freedesktop.org/drm/intel/-/issues/4898 Tested-by: Raviteja Goud Talla Cc: Rodrigo Vivi Acked-by: Lu Baolu Signed-off-by: Tejas Upadhyay Reviewed-by: Rodrigo Vivi Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20220302043256.191529-1-tejaskumarx.surendrakumar.upadh...@intel.com Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index 91a5c75966f3..a1ffb3d6d901 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -5728,7 +5728,7 @@ static void quirk_igfx_skip_te_disable(struct pci_dev *dev) ver = (dev->device >> 8) & 0xff; if (ver != 0x45 && ver != 0x46 && ver != 0x4c && ver != 0x4e && ver != 0x8a && ver != 0x98 && - ver != 0x9a) + ver != 0x9a && ver != 0xa7) return; if (risky_device(dev)) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 083/135] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index f8ff598596b8..ac740630c79c 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -448,7 +448,7 @@ void debug_dma_dump_mappings(struct device *dev) * other hand, consumes a single dma_debug_entry, but inserts 'nents' * entries into the tree. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 001/135] iommu/vt-d: Add RPLS to quirk list to skip TE disabling
From: Tejas Upadhyay [ Upstream commit 0a967f5bfd9134b89681cae58deb222e20840e76 ] The VT-d spec requires (10.4.4 Global Command Register, TE field) that: Hardware implementations supporting DMA draining must drain any in-flight DMA read/write requests queued within the Root-Complex before completing the translation enable command and reflecting the status of the command through the TES field in the Global Status register. Unfortunately, some integrated graphic devices fail to do so after some kind of power state transition. As the result, the system might stuck in iommu_disable_translati on(), waiting for the completion of TE transition. This adds RPLS to a quirk list for those devices and skips TE disabling if the qurik hits. Link: https://gitlab.freedesktop.org/drm/intel/-/issues/4898 Tested-by: Raviteja Goud Talla Cc: Rodrigo Vivi Acked-by: Lu Baolu Signed-off-by: Tejas Upadhyay Reviewed-by: Rodrigo Vivi Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20220302043256.191529-1-tejaskumarx.surendrakumar.upadh...@intel.com Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index ab2273300346..e3f15e0cae34 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -5764,7 +5764,7 @@ static void quirk_igfx_skip_te_disable(struct pci_dev *dev) ver = (dev->device >> 8) & 0xff; if (ver != 0x45 && ver != 0x46 && ver != 0x4c && ver != 0x4e && ver != 0x8a && ver != 0x98 && - ver != 0x9a) + ver != 0x9a && ver != 0xa7) return; if (risky_device(dev)) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.18 100/159] dma-debug: change allocation mode from GFP_NOWAIT to GFP_ATIOMIC
From: Mikulas Patocka [ Upstream commit 84bc4f1d5f8aa68706a96711dccb28b518e5 ] We observed the error "cacheline tracking ENOMEM, dma-debug disabled" during a light system load (copying some files). The reason for this error is that the dma_active_cacheline radix tree uses GFP_NOWAIT allocation - so it can't access the emergency memory reserves and it fails as soon as anybody reaches the watermark. This patch changes GFP_NOWAIT to GFP_ATOMIC, so that it can access the emergency memory reserves. Signed-off-by: Mikulas Patocka Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index f8ff598596b8..ac740630c79c 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -448,7 +448,7 @@ void debug_dma_dump_mappings(struct device *dev) * other hand, consumes a single dma_debug_entry, but inserts 'nents' * entries into the tree. */ -static RADIX_TREE(dma_active_cacheline, GFP_NOWAIT); +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC); static DEFINE_SPINLOCK(radix_lock); #define ACTIVE_CACHELINE_MAX_OVERLAP ((1 << RADIX_TREE_MAX_TAGS) - 1) #define CACHELINE_PER_PAGE_SHIFT (PAGE_SHIFT - L1_CACHE_SHIFT) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.18 001/159] iommu/vt-d: Add RPLS to quirk list to skip TE disabling
From: Tejas Upadhyay [ Upstream commit 0a967f5bfd9134b89681cae58deb222e20840e76 ] The VT-d spec requires (10.4.4 Global Command Register, TE field) that: Hardware implementations supporting DMA draining must drain any in-flight DMA read/write requests queued within the Root-Complex before completing the translation enable command and reflecting the status of the command through the TES field in the Global Status register. Unfortunately, some integrated graphic devices fail to do so after some kind of power state transition. As the result, the system might stuck in iommu_disable_translati on(), waiting for the completion of TE transition. This adds RPLS to a quirk list for those devices and skips TE disabling if the qurik hits. Link: https://gitlab.freedesktop.org/drm/intel/-/issues/4898 Tested-by: Raviteja Goud Talla Cc: Rodrigo Vivi Acked-by: Lu Baolu Signed-off-by: Tejas Upadhyay Reviewed-by: Rodrigo Vivi Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20220302043256.191529-1-tejaskumarx.surendrakumar.upadh...@intel.com Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index 0ea47e17b379..ba9a63cac47c 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -5031,7 +5031,7 @@ static void quirk_igfx_skip_te_disable(struct pci_dev *dev) ver = (dev->device >> 8) & 0xff; if (ver != 0x45 && ver != 0x46 && ver != 0x4c && ver != 0x4e && ver != 0x8a && ver != 0x98 && - ver != 0x9a) + ver != 0x9a && ver != 0xa7) return; if (risky_device(dev)) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 03/19] iommu: arm-smmu: disable large page mappings for Nvidia arm-smmu
From: Ashish Mhetre [ Upstream commit 4a25f2ea0e030b2fc852c4059a50181bfc5b2f57 ] Tegra194 and Tegra234 SoCs have the erratum that causes walk cache entries to not be invalidated correctly. The problem is that the walk cache index generated for IOVA is not same across translation and invalidation requests. This is leading to page faults when PMD entry is released during unmap and populated with new PTE table during subsequent map request. Disabling large page mappings avoids the release of PMD entry and avoid translations seeing stale PMD entry in walk cache. Fix this by limiting the page mappings to PAGE_SIZE for Tegra194 and Tegra234 devices. This is recommended fix from Tegra hardware design team. Acked-by: Robin Murphy Reviewed-by: Krishna Reddy Co-developed-by: Pritesh Raithatha Signed-off-by: Pritesh Raithatha Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/20220421081504.24678-1-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 30 1 file changed, 30 insertions(+) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c index 01e9b50b10a1..87bf522b9d2e 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c @@ -258,6 +258,34 @@ static void nvidia_smmu_probe_finalize(struct arm_smmu_device *smmu, struct devi dev_name(dev), err); } +static int nvidia_smmu_init_context(struct arm_smmu_domain *smmu_domain, + struct io_pgtable_cfg *pgtbl_cfg, + struct device *dev) +{ + struct arm_smmu_device *smmu = smmu_domain->smmu; + const struct device_node *np = smmu->dev->of_node; + + /* +* Tegra194 and Tegra234 SoCs have the erratum that causes walk cache +* entries to not be invalidated correctly. The problem is that the walk +* cache index generated for IOVA is not same across translation and +* invalidation requests. This is leading to page faults when PMD entry +* is released during unmap and populated with new PTE table during +* subsequent map request. Disabling large page mappings avoids the +* release of PMD entry and avoid translations seeing stale PMD entry in +* walk cache. +* Fix this by limiting the page mappings to PAGE_SIZE on Tegra194 and +* Tegra234. +*/ + if (of_device_is_compatible(np, "nvidia,tegra234-smmu") || + of_device_is_compatible(np, "nvidia,tegra194-smmu")) { + smmu->pgsize_bitmap = PAGE_SIZE; + pgtbl_cfg->pgsize_bitmap = smmu->pgsize_bitmap; + } + + return 0; +} + static const struct arm_smmu_impl nvidia_smmu_impl = { .read_reg = nvidia_smmu_read_reg, .write_reg = nvidia_smmu_write_reg, @@ -268,10 +296,12 @@ static const struct arm_smmu_impl nvidia_smmu_impl = { .global_fault = nvidia_smmu_global_fault, .context_fault = nvidia_smmu_context_fault, .probe_finalize = nvidia_smmu_probe_finalize, + .init_context = nvidia_smmu_init_context, }; static const struct arm_smmu_impl nvidia_smmu_single_impl = { .probe_finalize = nvidia_smmu_probe_finalize, + .init_context = nvidia_smmu_init_context, }; struct arm_smmu_device *nvidia_smmu_impl_init(struct arm_smmu_device *smmu) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 04/21] iommu: arm-smmu: disable large page mappings for Nvidia arm-smmu
From: Ashish Mhetre [ Upstream commit 4a25f2ea0e030b2fc852c4059a50181bfc5b2f57 ] Tegra194 and Tegra234 SoCs have the erratum that causes walk cache entries to not be invalidated correctly. The problem is that the walk cache index generated for IOVA is not same across translation and invalidation requests. This is leading to page faults when PMD entry is released during unmap and populated with new PTE table during subsequent map request. Disabling large page mappings avoids the release of PMD entry and avoid translations seeing stale PMD entry in walk cache. Fix this by limiting the page mappings to PAGE_SIZE for Tegra194 and Tegra234 devices. This is recommended fix from Tegra hardware design team. Acked-by: Robin Murphy Reviewed-by: Krishna Reddy Co-developed-by: Pritesh Raithatha Signed-off-by: Pritesh Raithatha Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/20220421081504.24678-1-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 30 1 file changed, 30 insertions(+) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c index 01e9b50b10a1..87bf522b9d2e 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c @@ -258,6 +258,34 @@ static void nvidia_smmu_probe_finalize(struct arm_smmu_device *smmu, struct devi dev_name(dev), err); } +static int nvidia_smmu_init_context(struct arm_smmu_domain *smmu_domain, + struct io_pgtable_cfg *pgtbl_cfg, + struct device *dev) +{ + struct arm_smmu_device *smmu = smmu_domain->smmu; + const struct device_node *np = smmu->dev->of_node; + + /* +* Tegra194 and Tegra234 SoCs have the erratum that causes walk cache +* entries to not be invalidated correctly. The problem is that the walk +* cache index generated for IOVA is not same across translation and +* invalidation requests. This is leading to page faults when PMD entry +* is released during unmap and populated with new PTE table during +* subsequent map request. Disabling large page mappings avoids the +* release of PMD entry and avoid translations seeing stale PMD entry in +* walk cache. +* Fix this by limiting the page mappings to PAGE_SIZE on Tegra194 and +* Tegra234. +*/ + if (of_device_is_compatible(np, "nvidia,tegra234-smmu") || + of_device_is_compatible(np, "nvidia,tegra194-smmu")) { + smmu->pgsize_bitmap = PAGE_SIZE; + pgtbl_cfg->pgsize_bitmap = smmu->pgsize_bitmap; + } + + return 0; +} + static const struct arm_smmu_impl nvidia_smmu_impl = { .read_reg = nvidia_smmu_read_reg, .write_reg = nvidia_smmu_write_reg, @@ -268,10 +296,12 @@ static const struct arm_smmu_impl nvidia_smmu_impl = { .global_fault = nvidia_smmu_global_fault, .context_fault = nvidia_smmu_context_fault, .probe_finalize = nvidia_smmu_probe_finalize, + .init_context = nvidia_smmu_init_context, }; static const struct arm_smmu_impl nvidia_smmu_single_impl = { .probe_finalize = nvidia_smmu_probe_finalize, + .init_context = nvidia_smmu_init_context, }; struct arm_smmu_device *nvidia_smmu_impl_init(struct arm_smmu_device *smmu) -- 2.35.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.9 06/16] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c index 48d382008788..db40ce599e97 100644 --- a/drivers/iommu/arm-smmu-v3.c +++ b/drivers/iommu/arm-smmu-v3.c @@ -1171,6 +1171,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.14 09/22] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c index 29feafa8007f..878087b9ddfe 100644 --- a/drivers/iommu/arm-smmu-v3.c +++ b/drivers/iommu/arm-smmu-v3.c @@ -1210,6 +1210,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.19 11/29] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c index 6b7664052b5b..9f16f47e7021 100644 --- a/drivers/iommu/arm-smmu-v3.c +++ b/drivers/iommu/arm-smmu-v3.c @@ -1250,6 +1250,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 27/37] iommu/iova: Improve 32-bit free space estimate
From: Robin Murphy [ Upstream commit 5b61343b50590fb04a3f6be2cdc4868091757262 ] For various reasons based on the allocator behaviour and typical use-cases at the time, when the max32_alloc_size optimisation was introduced it seemed reasonable to couple the reset of the tracked size to the update of cached32_node upon freeing a relevant IOVA. However, since subsequent optimisations focused on helping genuine 32-bit devices make best use of even more limited address spaces, it is now a lot more likely for cached32_node to be anywhere in a "full" 32-bit address space, and as such more likely for space to become available from IOVAs below that node being freed. At this point, the short-cut in __cached_rbnode_delete_update() really doesn't hold up any more, and we need to fix the logic to reliably provide the expected behaviour. We still want cached32_node to only move upwards, but we should reset the allocation size if *any* 32-bit space has become available. Reported-by: Yunfei Wang Signed-off-by: Robin Murphy Reviewed-by: Miles Chen Link: https://lore.kernel.org/r/033815732d83ca73b13c11485ac39336f15c3b40.1646318408.git.robin.mur...@arm.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iova.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index 906582a21124..628a586be695 100644 --- a/drivers/iommu/iova.c +++ b/drivers/iommu/iova.c @@ -138,10 +138,11 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free) cached_iova = rb_entry(iovad->cached32_node, struct iova, node); if (free == cached_iova || (free->pfn_hi < iovad->dma_32bit_pfn && -free->pfn_lo >= cached_iova->pfn_lo)) { +free->pfn_lo >= cached_iova->pfn_lo)) iovad->cached32_node = rb_next(&free->node); + + if (free->pfn_lo < iovad->dma_32bit_pfn) iovad->max32_alloc_size = iovad->dma_32bit_pfn; - } cached_iova = rb_entry(iovad->cached_node, struct iova, node); if (free->pfn_lo >= cached_iova->pfn_lo) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 14/37] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c index ef6af714a7e6..02c2fb551f38 100644 --- a/drivers/iommu/arm-smmu-v3.c +++ b/drivers/iommu/arm-smmu-v3.c @@ -1708,6 +1708,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 45/65] iommu/iova: Improve 32-bit free space estimate
From: Robin Murphy [ Upstream commit 5b61343b50590fb04a3f6be2cdc4868091757262 ] For various reasons based on the allocator behaviour and typical use-cases at the time, when the max32_alloc_size optimisation was introduced it seemed reasonable to couple the reset of the tracked size to the update of cached32_node upon freeing a relevant IOVA. However, since subsequent optimisations focused on helping genuine 32-bit devices make best use of even more limited address spaces, it is now a lot more likely for cached32_node to be anywhere in a "full" 32-bit address space, and as such more likely for space to become available from IOVAs below that node being freed. At this point, the short-cut in __cached_rbnode_delete_update() really doesn't hold up any more, and we need to fix the logic to reliably provide the expected behaviour. We still want cached32_node to only move upwards, but we should reset the allocation size if *any* 32-bit space has become available. Reported-by: Yunfei Wang Signed-off-by: Robin Murphy Reviewed-by: Miles Chen Link: https://lore.kernel.org/r/033815732d83ca73b13c11485ac39336f15c3b40.1646318408.git.robin.mur...@arm.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iova.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index 1164d1a42cbc..4600e97acb26 100644 --- a/drivers/iommu/iova.c +++ b/drivers/iommu/iova.c @@ -138,10 +138,11 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free) cached_iova = rb_entry(iovad->cached32_node, struct iova, node); if (free == cached_iova || (free->pfn_hi < iovad->dma_32bit_pfn && -free->pfn_lo >= cached_iova->pfn_lo)) { +free->pfn_lo >= cached_iova->pfn_lo)) iovad->cached32_node = rb_next(&free->node); + + if (free->pfn_lo < iovad->dma_32bit_pfn) iovad->max32_alloc_size = iovad->dma_32bit_pfn; - } cached_iova = rb_entry(iovad->cached_node, struct iova, node); if (free->pfn_lo >= cached_iova->pfn_lo) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 22/65] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c index 7067b7c11626..483c1362cc4a 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c @@ -1368,6 +1368,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 69/98] iommu/iova: Improve 32-bit free space estimate
From: Robin Murphy [ Upstream commit 5b61343b50590fb04a3f6be2cdc4868091757262 ] For various reasons based on the allocator behaviour and typical use-cases at the time, when the max32_alloc_size optimisation was introduced it seemed reasonable to couple the reset of the tracked size to the update of cached32_node upon freeing a relevant IOVA. However, since subsequent optimisations focused on helping genuine 32-bit devices make best use of even more limited address spaces, it is now a lot more likely for cached32_node to be anywhere in a "full" 32-bit address space, and as such more likely for space to become available from IOVAs below that node being freed. At this point, the short-cut in __cached_rbnode_delete_update() really doesn't hold up any more, and we need to fix the logic to reliably provide the expected behaviour. We still want cached32_node to only move upwards, but we should reset the allocation size if *any* 32-bit space has become available. Reported-by: Yunfei Wang Signed-off-by: Robin Murphy Reviewed-by: Miles Chen Link: https://lore.kernel.org/r/033815732d83ca73b13c11485ac39336f15c3b40.1646318408.git.robin.mur...@arm.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iova.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index 920fcc27c9a1..cae5a73ff518 100644 --- a/drivers/iommu/iova.c +++ b/drivers/iommu/iova.c @@ -154,10 +154,11 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free) cached_iova = to_iova(iovad->cached32_node); if (free == cached_iova || (free->pfn_hi < iovad->dma_32bit_pfn && -free->pfn_lo >= cached_iova->pfn_lo)) { +free->pfn_lo >= cached_iova->pfn_lo)) iovad->cached32_node = rb_next(&free->node); + + if (free->pfn_lo < iovad->dma_32bit_pfn) iovad->max32_alloc_size = iovad->dma_32bit_pfn; - } cached_iova = to_iova(iovad->cached_node); if (free->pfn_lo >= cached_iova->pfn_lo) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 33/98] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c index a388e318f86e..430315135cff 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c @@ -1552,6 +1552,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.16 080/109] iommu/iova: Improve 32-bit free space estimate
From: Robin Murphy [ Upstream commit 5b61343b50590fb04a3f6be2cdc4868091757262 ] For various reasons based on the allocator behaviour and typical use-cases at the time, when the max32_alloc_size optimisation was introduced it seemed reasonable to couple the reset of the tracked size to the update of cached32_node upon freeing a relevant IOVA. However, since subsequent optimisations focused on helping genuine 32-bit devices make best use of even more limited address spaces, it is now a lot more likely for cached32_node to be anywhere in a "full" 32-bit address space, and as such more likely for space to become available from IOVAs below that node being freed. At this point, the short-cut in __cached_rbnode_delete_update() really doesn't hold up any more, and we need to fix the logic to reliably provide the expected behaviour. We still want cached32_node to only move upwards, but we should reset the allocation size if *any* 32-bit space has become available. Reported-by: Yunfei Wang Signed-off-by: Robin Murphy Reviewed-by: Miles Chen Link: https://lore.kernel.org/r/033815732d83ca73b13c11485ac39336f15c3b40.1646318408.git.robin.mur...@arm.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iova.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index 920fcc27c9a1..cae5a73ff518 100644 --- a/drivers/iommu/iova.c +++ b/drivers/iommu/iova.c @@ -154,10 +154,11 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free) cached_iova = to_iova(iovad->cached32_node); if (free == cached_iova || (free->pfn_hi < iovad->dma_32bit_pfn && -free->pfn_lo >= cached_iova->pfn_lo)) { +free->pfn_lo >= cached_iova->pfn_lo)) iovad->cached32_node = rb_next(&free->node); + + if (free->pfn_lo < iovad->dma_32bit_pfn) iovad->max32_alloc_size = iovad->dma_32bit_pfn; - } cached_iova = to_iova(iovad->cached_node); if (free->pfn_lo >= cached_iova->pfn_lo) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.16 040/109] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c index f5848b351b19..af00714c8056 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c @@ -1558,6 +1558,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 115/149] iommu/iova: Improve 32-bit free space estimate
From: Robin Murphy [ Upstream commit 5b61343b50590fb04a3f6be2cdc4868091757262 ] For various reasons based on the allocator behaviour and typical use-cases at the time, when the max32_alloc_size optimisation was introduced it seemed reasonable to couple the reset of the tracked size to the update of cached32_node upon freeing a relevant IOVA. However, since subsequent optimisations focused on helping genuine 32-bit devices make best use of even more limited address spaces, it is now a lot more likely for cached32_node to be anywhere in a "full" 32-bit address space, and as such more likely for space to become available from IOVAs below that node being freed. At this point, the short-cut in __cached_rbnode_delete_update() really doesn't hold up any more, and we need to fix the logic to reliably provide the expected behaviour. We still want cached32_node to only move upwards, but we should reset the allocation size if *any* 32-bit space has become available. Reported-by: Yunfei Wang Signed-off-by: Robin Murphy Reviewed-by: Miles Chen Link: https://lore.kernel.org/r/033815732d83ca73b13c11485ac39336f15c3b40.1646318408.git.robin.mur...@arm.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iova.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index b28c9435b898..170e0f33040e 100644 --- a/drivers/iommu/iova.c +++ b/drivers/iommu/iova.c @@ -95,10 +95,11 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free) cached_iova = to_iova(iovad->cached32_node); if (free == cached_iova || (free->pfn_hi < iovad->dma_32bit_pfn && -free->pfn_lo >= cached_iova->pfn_lo)) { +free->pfn_lo >= cached_iova->pfn_lo)) iovad->cached32_node = rb_next(&free->node); + + if (free->pfn_lo < iovad->dma_32bit_pfn) iovad->max32_alloc_size = iovad->dma_32bit_pfn; - } cached_iova = to_iova(iovad->cached_node); if (free->pfn_lo >= cached_iova->pfn_lo) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 057/149] iommu/arm-smmu-v3: fix event handling soft lockup
From: Zhou Guanghui [ Upstream commit 30de2b541af98179780054836b48825fcfba4408 ] During event processing, events are read from the event queue one by one until the queue is empty.If the master device continuously requests address access at the same time and the SMMU generates events, the cyclic processing of the event takes a long time and softlockup warnings may be reported. arm-smmu-v3 arm-smmu-v3.34.auto: event 0x0a received: arm-smmu-v3 arm-smmu-v3.34.auto:0x7f22280a arm-smmu-v3 arm-smmu-v3.34.auto:0x107e arm-smmu-v3 arm-smmu-v3.34.auto:0x034e8670 watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [irq/268-arm-smm:247] Call trace: _dev_info+0x7c/0xa0 arm_smmu_evtq_thread+0x1c0/0x230 irq_thread_fn+0x30/0x80 irq_thread+0x128/0x210 kthread+0x134/0x138 ret_from_fork+0x10/0x1c Kernel panic - not syncing: softlockup: hung tasks Fix this by calling cond_resched() after the event information is printed. Signed-off-by: Zhou Guanghui Link: https://lore.kernel.org/r/20220119070754.26528-1-zhouguangh...@huawei.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c index 6dc6d8b6b368..f60381cdf1c4 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c @@ -1558,6 +1558,7 @@ static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev) dev_info(smmu->dev, "\t0x%016llx\n", (unsigned long long)evt[i]); + cond_resched(); } /* -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 12/50] media: iommu/mediatek: Add device_link between the consumer and the larb devices
From: Yong Wu [ Upstream commit 635319a4a7444ca97124d781cd96deb277ff4d40 ] MediaTek IOMMU-SMI diagram is like below. all the consumer connect with smi-larb, then connect with smi-common. M4U | smi-common | - | |... | | larb1 larb2 | | vdec venc When the consumer works, it should enable the smi-larb's power which also need enable the smi-common's power firstly. Thus, First of all, use the device link connect the consumer and the smi-larbs. then add device link between the smi-larb and smi-common. This patch adds device_link between the consumer and the larbs. When device_link_add, I add the flag DL_FLAG_STATELESS to avoid calling pm_runtime_xx to keep the original status of clocks. It can avoid two issues: 1) Display HW show fastlogo abnormally reported in [1]. At the beggining, all the clocks are enabled before entering kernel, but the clocks for display HW(always in larb0) will be gated after clk_enable and clk_disable called from device_link_add(->pm_runtime_resume) and rpm_idle. The clock operation happened before display driver probe. At that time, the display HW will be abnormal. 2) A deadlock issue reported in [2]. Use DL_FLAG_STATELESS to skip pm_runtime_xx to avoid the deadlock. Corresponding, DL_FLAG_AUTOREMOVE_CONSUMER can't be added, then device_link_removed should be added explicitly. Meanwhile, Currently we don't have a device connect with 2 larbs at the same time. Disallow this case, print the error log. [1] https://lore.kernel.org/linux-mediatek/1564213888.22908.4.camel@mhfsdcap03/ [2] https://lore.kernel.org/patchwork/patch/1086569/ Suggested-by: Tomasz Figa Signed-off-by: Yong Wu Tested-by: Frank Wunderlich # BPI-R2/MT7623 Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 30 ++ drivers/iommu/mtk_iommu_v1.c | 29 - 2 files changed, 58 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 77ae20ff9b35..5971a1168666 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -562,22 +562,52 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct mtk_iommu_data *data; + struct device_link *link; + struct device *larbdev; + unsigned int larbid, larbidx, i; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return ERR_PTR(-ENODEV); /* Not a iommu client device */ data = dev_iommu_priv_get(dev); + /* +* Link the consumer device with the smi-larb device(supplier). +* The device that connects with each a larb is a independent HW. +* All the ports in each a device should be in the same larbs. +*/ + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + for (i = 1; i < fwspec->num_ids; i++) { + larbidx = MTK_M4U_TO_LARB(fwspec->ids[i]); + if (larbid != larbidx) { + dev_err(dev, "Can only use one larb. Fail@larb%d-%d.\n", + larbid, larbidx); + return ERR_PTR(-EINVAL); + } + } + larbdev = data->larb_imu[larbid].dev; + link = device_link_add(dev, larbdev, + DL_FLAG_PM_RUNTIME | DL_FLAG_STATELESS); + if (!link) + dev_err(dev, "Unable to link %s\n", dev_name(larbdev)); return &data->iommu; } static void mtk_iommu_release_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); + struct mtk_iommu_data *data; + struct device *larbdev; + unsigned int larbid; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return; + data = dev_iommu_priv_get(dev); + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + larbdev = data->larb_imu[larbid].dev; + device_link_remove(dev, larbdev); + iommu_fwspec_free(dev); } diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 68bf02f87cfd..bc7ee90b9373 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -423,7 +423,9 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct of_phandle_args iommu_spec; struct mtk_iommu_data *data; - int err, idx = 0; + int err, idx = 0, larbid, larbidx; + struct device_link *link; + struct device *larbdev; /* * In the deferred case, free the existed fwspec. @@ -453,6 +455,23 @@ static struct iommu_device
[PATCH AUTOSEL 5.15 10/50] media: iommu/mediatek-v1: Free the existed fwspec if the master dev already has
From: Yong Wu [ Upstream commit 822a2ed8c606caf6a11b1a180b8e46292bd77d71 ] When the iommu master device enters of_iommu_xlate, the ops may be NULL(iommu dev is defered), then it will initialize the fwspec here: [] (dev_iommu_fwspec_set) from [] (iommu_fwspec_init+0xbc/0xd4) [] (iommu_fwspec_init) from [] (of_iommu_xlate+0x7c/0x12c) [] (of_iommu_xlate) from [] (of_iommu_configure+0x144/0x1e8) BUT the mtk_iommu_v1.c only supports arm32, the probing flow still is a bit weird. We always expect create the fwspec internally. otherwise it will enter here and return fail. static int mtk_iommu_create_mapping(struct device *dev, struct of_phandle_args *args) { ... if (!fwspec) { } else if (dev_iommu_fwspec_get(dev)->ops != &mtk_iommu_ops) { >>>>>>>>>>Enter here. return fail.<<<<<<<<<<<< return -EINVAL; } ... } Thus, Free the existed fwspec if the master device already has fwspec. This issue is reported at: https://lore.kernel.org/linux-mediatek/trinity-7d9ebdc9-4849-4d93-bfb5-429dcb4ee449-1626253158870@3c-app-gmx-bs01/ Reported-by: Frank Wunderlich Tested-by: Frank Wunderlich # BPI-R2/MT7623 Signed-off-by: Yong Wu Acked-by: Joerg Roedel Acked-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu_v1.c | 9 + 1 file changed, 9 insertions(+) diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index be22fcf988ce..1467ba1e4417 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -425,6 +425,15 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct mtk_iommu_data *data; int err, idx = 0; + /* +* In the deferred case, free the existed fwspec. +* Always initialize the fwspec internally. +*/ + if (fwspec) { + iommu_fwspec_free(dev); + fwspec = dev_iommu_fwspec_get(dev); + } + while (!of_parse_phandle_with_args(dev->of_node, "iommus", "#iommu-cells", idx, &iommu_spec)) { -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 11/50] media: iommu/mediatek: Return ENODEV if the device is NULL
From: Yong Wu [ Upstream commit 2fb0feed51085db77606de9b9477c96894328809 ] The platform device is created at: of_platform_default_populate_init: arch_initcall_sync ->of_platform_populate ->of_platform_device_create_pdata When entering our probe, all the devices should be already created. if it is null, means NODEV. Currently we don't get the fail case. It's a minor fix, no need add fixes tags. Signed-off-by: Yong Wu Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 2 +- drivers/iommu/mtk_iommu_v1.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 25b834104790..77ae20ff9b35 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -848,7 +848,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[id].dev = &plarbdev->dev; diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 1467ba1e4417..68bf02f87cfd 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -604,7 +604,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[i].dev = &plarbdev->dev; -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.16 13/59] media: iommu/mediatek: Add device_link between the consumer and the larb devices
From: Yong Wu [ Upstream commit 635319a4a7444ca97124d781cd96deb277ff4d40 ] MediaTek IOMMU-SMI diagram is like below. all the consumer connect with smi-larb, then connect with smi-common. M4U | smi-common | - | |... | | larb1 larb2 | | vdec venc When the consumer works, it should enable the smi-larb's power which also need enable the smi-common's power firstly. Thus, First of all, use the device link connect the consumer and the smi-larbs. then add device link between the smi-larb and smi-common. This patch adds device_link between the consumer and the larbs. When device_link_add, I add the flag DL_FLAG_STATELESS to avoid calling pm_runtime_xx to keep the original status of clocks. It can avoid two issues: 1) Display HW show fastlogo abnormally reported in [1]. At the beggining, all the clocks are enabled before entering kernel, but the clocks for display HW(always in larb0) will be gated after clk_enable and clk_disable called from device_link_add(->pm_runtime_resume) and rpm_idle. The clock operation happened before display driver probe. At that time, the display HW will be abnormal. 2) A deadlock issue reported in [2]. Use DL_FLAG_STATELESS to skip pm_runtime_xx to avoid the deadlock. Corresponding, DL_FLAG_AUTOREMOVE_CONSUMER can't be added, then device_link_removed should be added explicitly. Meanwhile, Currently we don't have a device connect with 2 larbs at the same time. Disallow this case, print the error log. [1] https://lore.kernel.org/linux-mediatek/1564213888.22908.4.camel@mhfsdcap03/ [2] https://lore.kernel.org/patchwork/patch/1086569/ Suggested-by: Tomasz Figa Signed-off-by: Yong Wu Tested-by: Frank Wunderlich # BPI-R2/MT7623 Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 30 ++ drivers/iommu/mtk_iommu_v1.c | 29 - 2 files changed, 58 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 77ae20ff9b35..5971a1168666 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -562,22 +562,52 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct mtk_iommu_data *data; + struct device_link *link; + struct device *larbdev; + unsigned int larbid, larbidx, i; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return ERR_PTR(-ENODEV); /* Not a iommu client device */ data = dev_iommu_priv_get(dev); + /* +* Link the consumer device with the smi-larb device(supplier). +* The device that connects with each a larb is a independent HW. +* All the ports in each a device should be in the same larbs. +*/ + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + for (i = 1; i < fwspec->num_ids; i++) { + larbidx = MTK_M4U_TO_LARB(fwspec->ids[i]); + if (larbid != larbidx) { + dev_err(dev, "Can only use one larb. Fail@larb%d-%d.\n", + larbid, larbidx); + return ERR_PTR(-EINVAL); + } + } + larbdev = data->larb_imu[larbid].dev; + link = device_link_add(dev, larbdev, + DL_FLAG_PM_RUNTIME | DL_FLAG_STATELESS); + if (!link) + dev_err(dev, "Unable to link %s\n", dev_name(larbdev)); return &data->iommu; } static void mtk_iommu_release_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); + struct mtk_iommu_data *data; + struct device *larbdev; + unsigned int larbid; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return; + data = dev_iommu_priv_get(dev); + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + larbdev = data->larb_imu[larbid].dev; + device_link_remove(dev, larbdev); + iommu_fwspec_free(dev); } diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 68bf02f87cfd..bc7ee90b9373 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -423,7 +423,9 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct of_phandle_args iommu_spec; struct mtk_iommu_data *data; - int err, idx = 0; + int err, idx = 0, larbid, larbidx; + struct device_link *link; + struct device *larbdev; /* * In the deferred case, free the existed fwspec. @@ -453,6 +455,23 @@ static struct iommu_device
[PATCH AUTOSEL 5.16 12/59] media: iommu/mediatek: Return ENODEV if the device is NULL
From: Yong Wu [ Upstream commit 2fb0feed51085db77606de9b9477c96894328809 ] The platform device is created at: of_platform_default_populate_init: arch_initcall_sync ->of_platform_populate ->of_platform_device_create_pdata When entering our probe, all the devices should be already created. if it is null, means NODEV. Currently we don't get the fail case. It's a minor fix, no need add fixes tags. Signed-off-by: Yong Wu Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 2 +- drivers/iommu/mtk_iommu_v1.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 25b834104790..77ae20ff9b35 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -848,7 +848,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[id].dev = &plarbdev->dev; diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 1467ba1e4417..68bf02f87cfd 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -604,7 +604,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[i].dev = &plarbdev->dev; -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.16 11/59] media: iommu/mediatek-v1: Free the existed fwspec if the master dev already has
From: Yong Wu [ Upstream commit 822a2ed8c606caf6a11b1a180b8e46292bd77d71 ] When the iommu master device enters of_iommu_xlate, the ops may be NULL(iommu dev is defered), then it will initialize the fwspec here: [] (dev_iommu_fwspec_set) from [] (iommu_fwspec_init+0xbc/0xd4) [] (iommu_fwspec_init) from [] (of_iommu_xlate+0x7c/0x12c) [] (of_iommu_xlate) from [] (of_iommu_configure+0x144/0x1e8) BUT the mtk_iommu_v1.c only supports arm32, the probing flow still is a bit weird. We always expect create the fwspec internally. otherwise it will enter here and return fail. static int mtk_iommu_create_mapping(struct device *dev, struct of_phandle_args *args) { ... if (!fwspec) { } else if (dev_iommu_fwspec_get(dev)->ops != &mtk_iommu_ops) { >>>>>>>>>>Enter here. return fail.<<<<<<<<<<<< return -EINVAL; } ... } Thus, Free the existed fwspec if the master device already has fwspec. This issue is reported at: https://lore.kernel.org/linux-mediatek/trinity-7d9ebdc9-4849-4d93-bfb5-429dcb4ee449-1626253158870@3c-app-gmx-bs01/ Reported-by: Frank Wunderlich Tested-by: Frank Wunderlich # BPI-R2/MT7623 Signed-off-by: Yong Wu Acked-by: Joerg Roedel Acked-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu_v1.c | 9 + 1 file changed, 9 insertions(+) diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index be22fcf988ce..1467ba1e4417 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -425,6 +425,15 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct mtk_iommu_data *data; int err, idx = 0; + /* +* In the deferred case, free the existed fwspec. +* Always initialize the fwspec internally. +*/ + if (fwspec) { + iommu_fwspec_free(dev); + fwspec = dev_iommu_fwspec_get(dev); + } + while (!of_parse_phandle_with_args(dev->of_node, "iommus", "#iommu-cells", idx, &iommu_spec)) { -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 11/66] media: iommu/mediatek-v1: Free the existed fwspec if the master dev already has
From: Yong Wu [ Upstream commit 822a2ed8c606caf6a11b1a180b8e46292bd77d71 ] When the iommu master device enters of_iommu_xlate, the ops may be NULL(iommu dev is defered), then it will initialize the fwspec here: [] (dev_iommu_fwspec_set) from [] (iommu_fwspec_init+0xbc/0xd4) [] (iommu_fwspec_init) from [] (of_iommu_xlate+0x7c/0x12c) [] (of_iommu_xlate) from [] (of_iommu_configure+0x144/0x1e8) BUT the mtk_iommu_v1.c only supports arm32, the probing flow still is a bit weird. We always expect create the fwspec internally. otherwise it will enter here and return fail. static int mtk_iommu_create_mapping(struct device *dev, struct of_phandle_args *args) { ... if (!fwspec) { } else if (dev_iommu_fwspec_get(dev)->ops != &mtk_iommu_ops) { >>>>>>>>>>Enter here. return fail.<<<<<<<<<<<< return -EINVAL; } ... } Thus, Free the existed fwspec if the master device already has fwspec. This issue is reported at: https://lore.kernel.org/linux-mediatek/trinity-7d9ebdc9-4849-4d93-bfb5-429dcb4ee449-1626253158870@3c-app-gmx-bs01/ Reported-by: Frank Wunderlich Tested-by: Frank Wunderlich # BPI-R2/MT7623 Signed-off-by: Yong Wu Acked-by: Joerg Roedel Acked-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu_v1.c | 9 + 1 file changed, 9 insertions(+) diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index be22fcf988ce..1467ba1e4417 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -425,6 +425,15 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct mtk_iommu_data *data; int err, idx = 0; + /* +* In the deferred case, free the existed fwspec. +* Always initialize the fwspec internally. +*/ + if (fwspec) { + iommu_fwspec_free(dev); + fwspec = dev_iommu_fwspec_get(dev); + } + while (!of_parse_phandle_with_args(dev->of_node, "iommus", "#iommu-cells", idx, &iommu_spec)) { -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.17 13/66] media: iommu/mediatek: Add device_link between the consumer and the larb devices
From: Yong Wu [ Upstream commit 635319a4a7444ca97124d781cd96deb277ff4d40 ] MediaTek IOMMU-SMI diagram is like below. all the consumer connect with smi-larb, then connect with smi-common. M4U | smi-common | - | |... | | larb1 larb2 | | vdec venc When the consumer works, it should enable the smi-larb's power which also need enable the smi-common's power firstly. Thus, First of all, use the device link connect the consumer and the smi-larbs. then add device link between the smi-larb and smi-common. This patch adds device_link between the consumer and the larbs. When device_link_add, I add the flag DL_FLAG_STATELESS to avoid calling pm_runtime_xx to keep the original status of clocks. It can avoid two issues: 1) Display HW show fastlogo abnormally reported in [1]. At the beggining, all the clocks are enabled before entering kernel, but the clocks for display HW(always in larb0) will be gated after clk_enable and clk_disable called from device_link_add(->pm_runtime_resume) and rpm_idle. The clock operation happened before display driver probe. At that time, the display HW will be abnormal. 2) A deadlock issue reported in [2]. Use DL_FLAG_STATELESS to skip pm_runtime_xx to avoid the deadlock. Corresponding, DL_FLAG_AUTOREMOVE_CONSUMER can't be added, then device_link_removed should be added explicitly. Meanwhile, Currently we don't have a device connect with 2 larbs at the same time. Disallow this case, print the error log. [1] https://lore.kernel.org/linux-mediatek/1564213888.22908.4.camel@mhfsdcap03/ [2] https://lore.kernel.org/patchwork/patch/1086569/ Suggested-by: Tomasz Figa Signed-off-by: Yong Wu Tested-by: Frank Wunderlich # BPI-R2/MT7623 Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 30 ++ drivers/iommu/mtk_iommu_v1.c | 29 - 2 files changed, 58 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 77ae20ff9b35..5971a1168666 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -562,22 +562,52 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct mtk_iommu_data *data; + struct device_link *link; + struct device *larbdev; + unsigned int larbid, larbidx, i; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return ERR_PTR(-ENODEV); /* Not a iommu client device */ data = dev_iommu_priv_get(dev); + /* +* Link the consumer device with the smi-larb device(supplier). +* The device that connects with each a larb is a independent HW. +* All the ports in each a device should be in the same larbs. +*/ + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + for (i = 1; i < fwspec->num_ids; i++) { + larbidx = MTK_M4U_TO_LARB(fwspec->ids[i]); + if (larbid != larbidx) { + dev_err(dev, "Can only use one larb. Fail@larb%d-%d.\n", + larbid, larbidx); + return ERR_PTR(-EINVAL); + } + } + larbdev = data->larb_imu[larbid].dev; + link = device_link_add(dev, larbdev, + DL_FLAG_PM_RUNTIME | DL_FLAG_STATELESS); + if (!link) + dev_err(dev, "Unable to link %s\n", dev_name(larbdev)); return &data->iommu; } static void mtk_iommu_release_device(struct device *dev) { struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); + struct mtk_iommu_data *data; + struct device *larbdev; + unsigned int larbid; if (!fwspec || fwspec->ops != &mtk_iommu_ops) return; + data = dev_iommu_priv_get(dev); + larbid = MTK_M4U_TO_LARB(fwspec->ids[0]); + larbdev = data->larb_imu[larbid].dev; + device_link_remove(dev, larbdev); + iommu_fwspec_free(dev); } diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 68bf02f87cfd..bc7ee90b9373 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -423,7 +423,9 @@ static struct iommu_device *mtk_iommu_probe_device(struct device *dev) struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); struct of_phandle_args iommu_spec; struct mtk_iommu_data *data; - int err, idx = 0; + int err, idx = 0, larbid, larbidx; + struct device_link *link; + struct device *larbdev; /* * In the deferred case, free the existed fwspec. @@ -453,6 +455,23 @@ static struct iommu_device
[PATCH AUTOSEL 5.17 12/66] media: iommu/mediatek: Return ENODEV if the device is NULL
From: Yong Wu [ Upstream commit 2fb0feed51085db77606de9b9477c96894328809 ] The platform device is created at: of_platform_default_populate_init: arch_initcall_sync ->of_platform_populate ->of_platform_device_create_pdata When entering our probe, all the devices should be already created. if it is null, means NODEV. Currently we don't get the fail case. It's a minor fix, no need add fixes tags. Signed-off-by: Yong Wu Acked-by: Joerg Roedel Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab Signed-off-by: Sasha Levin --- drivers/iommu/mtk_iommu.c| 2 +- drivers/iommu/mtk_iommu_v1.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c index 25b834104790..77ae20ff9b35 100644 --- a/drivers/iommu/mtk_iommu.c +++ b/drivers/iommu/mtk_iommu.c @@ -848,7 +848,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[id].dev = &plarbdev->dev; diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c index 1467ba1e4417..68bf02f87cfd 100644 --- a/drivers/iommu/mtk_iommu_v1.c +++ b/drivers/iommu/mtk_iommu_v1.c @@ -604,7 +604,7 @@ static int mtk_iommu_probe(struct platform_device *pdev) plarbdev = of_find_device_by_node(larbnode); if (!plarbdev) { of_node_put(larbnode); - return -EPROBE_DEFER; + return -ENODEV; } data->larb_imu[i].dev = &plarbdev->dev; -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 02/23] swiotlb: fix info leak with DMA_FROM_DEVICE
From: Halil Pasic [ Upstream commit ddbd89deb7d32b1fbb879f48d68fda1a8ac58e8e ] The problem I'm addressing was discovered by the LTP test covering cve-2018-1000204. A short description of what happens follows: 1) The test case issues a command code 00 (TEST UNIT READY) via the SG_IO interface with: dxfer_len == 524288, dxdfer_dir == SG_DXFER_FROM_DEV and a corresponding dxferp. The peculiar thing about this is that TUR is not reading from the device. 2) In sg_start_req() the invocation of blk_rq_map_user() effectively bounces the user-space buffer. As if the device was to transfer into it. Since commit a45b599ad808 ("scsi: sg: allocate with __GFP_ZERO in sg_build_indirect()") we make sure this first bounce buffer is allocated with GFP_ZERO. 3) For the rest of the story we keep ignoring that we have a TUR, so the device won't touch the buffer we prepare as if the we had a DMA_FROM_DEVICE type of situation. My setup uses a virtio-scsi device and the buffer allocated by SG is mapped by the function virtqueue_add_split() which uses DMA_FROM_DEVICE for the "in" sgs (here scatter-gather and not scsi generics). This mapping involves bouncing via the swiotlb (we need swiotlb to do virtio in protected guest like s390 Secure Execution, or AMD SEV). 4) When the SCSI TUR is done, we first copy back the content of the second (that is swiotlb) bounce buffer (which most likely contains some previous IO data), to the first bounce buffer, which contains all zeros. Then we copy back the content of the first bounce buffer to the user-space buffer. 5) The test case detects that the buffer, which it zero-initialized, ain't all zeros and fails. One can argue that this is an swiotlb problem, because without swiotlb we leak all zeros, and the swiotlb should be transparent in a sense that it does not affect the outcome (if all other participants are well behaved). Copying the content of the original buffer into the swiotlb buffer is the only way I can think of to make swiotlb transparent in such scenarios. So let's do just that if in doubt, but allow the driver to tell us that the whole mapped buffer is going to be overwritten, in which case we can preserve the old behavior and avoid the performance impact of the extra bounce. Signed-off-by: Halil Pasic Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- Documentation/core-api/dma-attributes.rst | 8 include/linux/dma-mapping.h | 8 kernel/dma/swiotlb.c | 3 ++- 3 files changed, 18 insertions(+), 1 deletion(-) diff --git a/Documentation/core-api/dma-attributes.rst b/Documentation/core-api/dma-attributes.rst index 1887d92e8e926..17706dc91ec9f 100644 --- a/Documentation/core-api/dma-attributes.rst +++ b/Documentation/core-api/dma-attributes.rst @@ -130,3 +130,11 @@ accesses to DMA buffers in both privileged "supervisor" and unprivileged subsystem that the buffer is fully accessible at the elevated privilege level (and ideally inaccessible or at least read-only at the lesser-privileged levels). + +DMA_ATTR_OVERWRITE +-- + +This is a hint to the DMA-mapping subsystem that the device is expected to +overwrite the entire mapped size, thus the caller does not require any of the +previous buffer contents to be preserved. This allows bounce-buffering +implementations to optimise DMA_FROM_DEVICE transfers. diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h index dca2b1355bb13..6150d11a607e1 100644 --- a/include/linux/dma-mapping.h +++ b/include/linux/dma-mapping.h @@ -61,6 +61,14 @@ */ #define DMA_ATTR_PRIVILEGED(1UL << 9) +/* + * This is a hint to the DMA-mapping subsystem that the device is expected + * to overwrite the entire mapped size, thus the caller does not require any + * of the previous buffer contents to be preserved. This allows + * bounce-buffering implementations to optimise DMA_FROM_DEVICE transfers. + */ +#define DMA_ATTR_OVERWRITE (1UL << 10) + /* * A dma_addr_t can hold any valid DMA or bus address for the platform. It can * be given to a device to use as a DMA source or target. It is specific to a diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index 87c40517e8227..aca0690550e2f 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -579,7 +579,8 @@ phys_addr_t swiotlb_tbl_map_single(struct device *dev, phys_addr_t orig_addr, mem->slots[index + i].orig_addr = slot_addr(orig_addr, i); tlb_addr = slot_addr(mem->start, index) + offset; if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC) && - (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)) + (!(attrs & DMA_ATTR_OVERWRITE) || dir == DMA_TO_DEVICE || + dir == DMA_BIDIRECTIONAL)) swiotlb_bounce(dev, tlb_addr, mapping_size, DMA_TO_DEVICE); r
[PATCH AUTOSEL 5.16 02/28] swiotlb: fix info leak with DMA_FROM_DEVICE
From: Halil Pasic [ Upstream commit ddbd89deb7d32b1fbb879f48d68fda1a8ac58e8e ] The problem I'm addressing was discovered by the LTP test covering cve-2018-1000204. A short description of what happens follows: 1) The test case issues a command code 00 (TEST UNIT READY) via the SG_IO interface with: dxfer_len == 524288, dxdfer_dir == SG_DXFER_FROM_DEV and a corresponding dxferp. The peculiar thing about this is that TUR is not reading from the device. 2) In sg_start_req() the invocation of blk_rq_map_user() effectively bounces the user-space buffer. As if the device was to transfer into it. Since commit a45b599ad808 ("scsi: sg: allocate with __GFP_ZERO in sg_build_indirect()") we make sure this first bounce buffer is allocated with GFP_ZERO. 3) For the rest of the story we keep ignoring that we have a TUR, so the device won't touch the buffer we prepare as if the we had a DMA_FROM_DEVICE type of situation. My setup uses a virtio-scsi device and the buffer allocated by SG is mapped by the function virtqueue_add_split() which uses DMA_FROM_DEVICE for the "in" sgs (here scatter-gather and not scsi generics). This mapping involves bouncing via the swiotlb (we need swiotlb to do virtio in protected guest like s390 Secure Execution, or AMD SEV). 4) When the SCSI TUR is done, we first copy back the content of the second (that is swiotlb) bounce buffer (which most likely contains some previous IO data), to the first bounce buffer, which contains all zeros. Then we copy back the content of the first bounce buffer to the user-space buffer. 5) The test case detects that the buffer, which it zero-initialized, ain't all zeros and fails. One can argue that this is an swiotlb problem, because without swiotlb we leak all zeros, and the swiotlb should be transparent in a sense that it does not affect the outcome (if all other participants are well behaved). Copying the content of the original buffer into the swiotlb buffer is the only way I can think of to make swiotlb transparent in such scenarios. So let's do just that if in doubt, but allow the driver to tell us that the whole mapped buffer is going to be overwritten, in which case we can preserve the old behavior and avoid the performance impact of the extra bounce. Signed-off-by: Halil Pasic Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- Documentation/core-api/dma-attributes.rst | 8 include/linux/dma-mapping.h | 8 kernel/dma/swiotlb.c | 3 ++- 3 files changed, 18 insertions(+), 1 deletion(-) diff --git a/Documentation/core-api/dma-attributes.rst b/Documentation/core-api/dma-attributes.rst index 1887d92e8e926..17706dc91ec9f 100644 --- a/Documentation/core-api/dma-attributes.rst +++ b/Documentation/core-api/dma-attributes.rst @@ -130,3 +130,11 @@ accesses to DMA buffers in both privileged "supervisor" and unprivileged subsystem that the buffer is fully accessible at the elevated privilege level (and ideally inaccessible or at least read-only at the lesser-privileged levels). + +DMA_ATTR_OVERWRITE +-- + +This is a hint to the DMA-mapping subsystem that the device is expected to +overwrite the entire mapped size, thus the caller does not require any of the +previous buffer contents to be preserved. This allows bounce-buffering +implementations to optimise DMA_FROM_DEVICE transfers. diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h index dca2b1355bb13..6150d11a607e1 100644 --- a/include/linux/dma-mapping.h +++ b/include/linux/dma-mapping.h @@ -61,6 +61,14 @@ */ #define DMA_ATTR_PRIVILEGED(1UL << 9) +/* + * This is a hint to the DMA-mapping subsystem that the device is expected + * to overwrite the entire mapped size, thus the caller does not require any + * of the previous buffer contents to be preserved. This allows + * bounce-buffering implementations to optimise DMA_FROM_DEVICE transfers. + */ +#define DMA_ATTR_OVERWRITE (1UL << 10) + /* * A dma_addr_t can hold any valid DMA or bus address for the platform. It can * be given to a device to use as a DMA source or target. It is specific to a diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index 8e840fbbed7c7..d958b1201092c 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -582,7 +582,8 @@ phys_addr_t swiotlb_tbl_map_single(struct device *dev, phys_addr_t orig_addr, mem->slots[index + i].orig_addr = slot_addr(orig_addr, i); tlb_addr = slot_addr(mem->start, index) + offset; if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC) && - (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)) + (!(attrs & DMA_ATTR_OVERWRITE) || dir == DMA_TO_DEVICE || + dir == DMA_BIDIRECTIONAL)) swiotlb_bounce(dev, tlb_addr, mapping_size, DMA_TO_DEVICE); r
[PATCH AUTOSEL 5.10 17/27] iommu: Fix potential use-after-free during probe
From: Vijayanand Jitta [ Upstream commit b54240ad494300ff0994c4539a531727874381f4 ] Kasan has reported the following use after free on dev->iommu. when a device probe fails and it is in process of freeing dev->iommu in dev_iommu_free function, a deferred_probe_work_func runs in parallel and tries to access dev->iommu->fwspec in of_iommu_configure path thus causing use after free. BUG: KASAN: use-after-free in of_iommu_configure+0xb4/0x4a4 Read of size 8 at addr ff87a2f1acb8 by task kworker/u16:2/153 Workqueue: events_unbound deferred_probe_work_func Call trace: dump_backtrace+0x0/0x33c show_stack+0x18/0x24 dump_stack_lvl+0x16c/0x1e0 print_address_description+0x84/0x39c __kasan_report+0x184/0x308 kasan_report+0x50/0x78 __asan_load8+0xc0/0xc4 of_iommu_configure+0xb4/0x4a4 of_dma_configure_id+0x2fc/0x4d4 platform_dma_configure+0x40/0x5c really_probe+0x1b4/0xb74 driver_probe_device+0x11c/0x228 __device_attach_driver+0x14c/0x304 bus_for_each_drv+0x124/0x1b0 __device_attach+0x25c/0x334 device_initial_probe+0x24/0x34 bus_probe_device+0x78/0x134 deferred_probe_work_func+0x130/0x1a8 process_one_work+0x4c8/0x970 worker_thread+0x5c8/0xaec kthread+0x1f8/0x220 ret_from_fork+0x10/0x18 Allocated by task 1: kasan_kmalloc+0xd4/0x114 __kasan_kmalloc+0x10/0x1c kmem_cache_alloc_trace+0xe4/0x3d4 __iommu_probe_device+0x90/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Freed by task 1: kasan_set_track+0x4c/0x84 kasan_set_free_info+0x28/0x4c kasan_slab_free+0x120/0x15c __kasan_slab_free+0x18/0x28 slab_free_freelist_hook+0x204/0x2fc kfree+0xfc/0x3a4 __iommu_probe_device+0x284/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Fix this by setting dev->iommu to NULL first and then freeing dev_iommu structure in dev_iommu_free function. Suggested-by: Robin Murphy Signed-off-by: Vijayanand Jitta Link: https://lore.kernel.org/r/1643613155-20215-1-git-send-email-quic_vji...@quicinc.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 9 +++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index bcf060b5cf85b..9d65557dfb2ce 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -185,9 +185,14 @@ static struct dev_iommu *dev_iommu_get(struct device *dev) static void dev_iommu_free(struct device *dev) { - iommu_fwspec_free(dev); - kfree(dev->iommu); + struct dev_iommu *param = dev->iommu; + dev->iommu = NULL; + if (param->fwspec) { + fwnode_handle_put(param->fwspec->iommu_fwnode); + kfree(param->fwspec); + } + kfree(param); } static int __iommu_probe_device(struct device *dev, struct list_head *group_list) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 19/36] iommu: Fix potential use-after-free during probe
From: Vijayanand Jitta [ Upstream commit b54240ad494300ff0994c4539a531727874381f4 ] Kasan has reported the following use after free on dev->iommu. when a device probe fails and it is in process of freeing dev->iommu in dev_iommu_free function, a deferred_probe_work_func runs in parallel and tries to access dev->iommu->fwspec in of_iommu_configure path thus causing use after free. BUG: KASAN: use-after-free in of_iommu_configure+0xb4/0x4a4 Read of size 8 at addr ff87a2f1acb8 by task kworker/u16:2/153 Workqueue: events_unbound deferred_probe_work_func Call trace: dump_backtrace+0x0/0x33c show_stack+0x18/0x24 dump_stack_lvl+0x16c/0x1e0 print_address_description+0x84/0x39c __kasan_report+0x184/0x308 kasan_report+0x50/0x78 __asan_load8+0xc0/0xc4 of_iommu_configure+0xb4/0x4a4 of_dma_configure_id+0x2fc/0x4d4 platform_dma_configure+0x40/0x5c really_probe+0x1b4/0xb74 driver_probe_device+0x11c/0x228 __device_attach_driver+0x14c/0x304 bus_for_each_drv+0x124/0x1b0 __device_attach+0x25c/0x334 device_initial_probe+0x24/0x34 bus_probe_device+0x78/0x134 deferred_probe_work_func+0x130/0x1a8 process_one_work+0x4c8/0x970 worker_thread+0x5c8/0xaec kthread+0x1f8/0x220 ret_from_fork+0x10/0x18 Allocated by task 1: kasan_kmalloc+0xd4/0x114 __kasan_kmalloc+0x10/0x1c kmem_cache_alloc_trace+0xe4/0x3d4 __iommu_probe_device+0x90/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Freed by task 1: kasan_set_track+0x4c/0x84 kasan_set_free_info+0x28/0x4c kasan_slab_free+0x120/0x15c __kasan_slab_free+0x18/0x28 slab_free_freelist_hook+0x204/0x2fc kfree+0xfc/0x3a4 __iommu_probe_device+0x284/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Fix this by setting dev->iommu to NULL first and then freeing dev_iommu structure in dev_iommu_free function. Suggested-by: Robin Murphy Signed-off-by: Vijayanand Jitta Link: https://lore.kernel.org/r/1643613155-20215-1-git-send-email-quic_vji...@quicinc.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 9 +++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index f62fb6a58f109..7f409e9eea4b7 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -206,9 +206,14 @@ static struct dev_iommu *dev_iommu_get(struct device *dev) static void dev_iommu_free(struct device *dev) { - iommu_fwspec_free(dev); - kfree(dev->iommu); + struct dev_iommu *param = dev->iommu; + dev->iommu = NULL; + if (param->fwspec) { + fwnode_handle_put(param->fwspec->iommu_fwnode); + kfree(param->fwspec); + } + kfree(param); } static int __iommu_probe_device(struct device *dev, struct list_head *group_list) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.16 24/42] iommu: Fix potential use-after-free during probe
From: Vijayanand Jitta [ Upstream commit b54240ad494300ff0994c4539a531727874381f4 ] Kasan has reported the following use after free on dev->iommu. when a device probe fails and it is in process of freeing dev->iommu in dev_iommu_free function, a deferred_probe_work_func runs in parallel and tries to access dev->iommu->fwspec in of_iommu_configure path thus causing use after free. BUG: KASAN: use-after-free in of_iommu_configure+0xb4/0x4a4 Read of size 8 at addr ff87a2f1acb8 by task kworker/u16:2/153 Workqueue: events_unbound deferred_probe_work_func Call trace: dump_backtrace+0x0/0x33c show_stack+0x18/0x24 dump_stack_lvl+0x16c/0x1e0 print_address_description+0x84/0x39c __kasan_report+0x184/0x308 kasan_report+0x50/0x78 __asan_load8+0xc0/0xc4 of_iommu_configure+0xb4/0x4a4 of_dma_configure_id+0x2fc/0x4d4 platform_dma_configure+0x40/0x5c really_probe+0x1b4/0xb74 driver_probe_device+0x11c/0x228 __device_attach_driver+0x14c/0x304 bus_for_each_drv+0x124/0x1b0 __device_attach+0x25c/0x334 device_initial_probe+0x24/0x34 bus_probe_device+0x78/0x134 deferred_probe_work_func+0x130/0x1a8 process_one_work+0x4c8/0x970 worker_thread+0x5c8/0xaec kthread+0x1f8/0x220 ret_from_fork+0x10/0x18 Allocated by task 1: kasan_kmalloc+0xd4/0x114 __kasan_kmalloc+0x10/0x1c kmem_cache_alloc_trace+0xe4/0x3d4 __iommu_probe_device+0x90/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Freed by task 1: kasan_set_track+0x4c/0x84 kasan_set_free_info+0x28/0x4c kasan_slab_free+0x120/0x15c __kasan_slab_free+0x18/0x28 slab_free_freelist_hook+0x204/0x2fc kfree+0xfc/0x3a4 __iommu_probe_device+0x284/0x394 probe_iommu_group+0x70/0x9c bus_for_each_dev+0x11c/0x19c bus_iommu_probe+0xb8/0x7d4 bus_set_iommu+0xcc/0x13c arm_smmu_bus_init+0x44/0x130 [arm_smmu] arm_smmu_device_probe+0xb88/0xc54 [arm_smmu] platform_drv_probe+0xe4/0x13c really_probe+0x2c8/0xb74 driver_probe_device+0x11c/0x228 device_driver_attach+0xf0/0x16c __driver_attach+0x80/0x320 bus_for_each_dev+0x11c/0x19c driver_attach+0x38/0x48 bus_add_driver+0x1dc/0x3a4 driver_register+0x18c/0x244 __platform_driver_register+0x88/0x9c init_module+0x64/0xff4 [arm_smmu] do_one_initcall+0x17c/0x2f0 do_init_module+0xe8/0x378 load_module+0x3f80/0x4a40 __se_sys_finit_module+0x1a0/0x1e4 __arm64_sys_finit_module+0x44/0x58 el0_svc_common+0x100/0x264 do_el0_svc+0x38/0xa4 el0_svc+0x20/0x30 el0_sync_handler+0x68/0xac el0_sync+0x160/0x180 Fix this by setting dev->iommu to NULL first and then freeing dev_iommu structure in dev_iommu_free function. Suggested-by: Robin Murphy Signed-off-by: Vijayanand Jitta Link: https://lore.kernel.org/r/1643613155-20215-1-git-send-email-quic_vji...@quicinc.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 9 +++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index 8b86406b71627..3632bf8b4031c 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -207,9 +207,14 @@ static struct dev_iommu *dev_iommu_get(struct device *dev) static void dev_iommu_free(struct device *dev) { - iommu_fwspec_free(dev); - kfree(dev->iommu); + struct dev_iommu *param = dev->iommu; + dev->iommu = NULL; + if (param->fwspec) { + fwnode_handle_put(param->fwspec->iommu_fwnode); + kfree(param->fwspec); + } + kfree(param); } static int __iommu_probe_device(struct device *dev, struct list_head *group_list) -- 2.34.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 55/75] iommu/vt-d: Do not falsely log intel_iommu is unsupported kernel option
From: Tvrtko Ursulin [ Upstream commit 5240aed2cd2594fb392239f11b9681e5e1591619 ] Handling of intel_iommu kernel command line option should return "true" to indicate option is valid and so avoid logging it as unknown by the core parsing code. Also log unknown sub-options at the notice level to let user know of potential typos or similar. Reported-by: Eero Tamminen Signed-off-by: Tvrtko Ursulin Link: https://lore.kernel.org/r/20210831112947.310080-1-tvrtko.ursu...@linux.intel.com Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20211014053839.727419-2-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index dd22fc7d51764..663ef69769f2b 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -439,6 +439,7 @@ static int __init intel_iommu_setup(char *str) { if (!str) return -EINVAL; + while (*str) { if (!strncmp(str, "on", 2)) { dmar_disabled = 0; @@ -465,13 +466,16 @@ static int __init intel_iommu_setup(char *str) } else if (!strncmp(str, "tboot_noforce", 13)) { pr_info("Intel-IOMMU: not forcing on after tboot. This could expose security risk for tboot\n"); intel_iommu_tboot_noforce = 1; + } else { + pr_notice("Unknown option - '%s'\n", str); } str += strcspn(str, ","); while (*str == ',') str++; } - return 0; + + return 1; } __setup("intel_iommu=", intel_iommu_setup); -- 2.33.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 79/82] iommu/dart: Initialize DART_STREAMS_ENABLE
From: Sven Peter [ Upstream commit 5a009fc1364170b240a4d351b345e69bb3728b3e ] DART has an additional global register to control which streams are isolated. This register is a bit redundant since DART_TCR can already be used to control isolation and is usually initialized to DART_STREAM_ALL by the time we get control. Some DARTs (namely the one used for the audio controller) however have some streams disabled initially. Make sure those work by initializing DART_STREAMS_ENABLE during reset. Reported-by: Martin Povišer Signed-off-by: Sven Peter Reviewed-by: Hector Martin Link: https://lore.kernel.org/r/20211019162253.45919-1-s...@svenpeter.dev Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/apple-dart.c | 5 + 1 file changed, 5 insertions(+) diff --git a/drivers/iommu/apple-dart.c b/drivers/iommu/apple-dart.c index fdfa39ec2a4d4..ad69eeb5ac5ba 100644 --- a/drivers/iommu/apple-dart.c +++ b/drivers/iommu/apple-dart.c @@ -70,6 +70,8 @@ #define DART_ERROR_ADDR_HI 0x54 #define DART_ERROR_ADDR_LO 0x50 +#define DART_STREAMS_ENABLE 0xfc + #define DART_TCR(sid) (0x100 + 4 * (sid)) #define DART_TCR_TRANSLATE_ENABLE BIT(7) #define DART_TCR_BYPASS0_ENABLE BIT(8) @@ -301,6 +303,9 @@ static int apple_dart_hw_reset(struct apple_dart *dart) apple_dart_hw_disable_dma(&stream_map); apple_dart_hw_clear_all_ttbrs(&stream_map); + /* enable all streams globally since TCR is used to control isolation */ + writel(DART_STREAM_ALL, dart->regs + DART_STREAMS_ENABLE); + /* clear any pending errors before the interrupt is unmasked */ writel(readl(dart->regs + DART_ERROR), dart->regs + DART_ERROR); -- 2.33.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.15 59/82] iommu/vt-d: Do not falsely log intel_iommu is unsupported kernel option
From: Tvrtko Ursulin [ Upstream commit 5240aed2cd2594fb392239f11b9681e5e1591619 ] Handling of intel_iommu kernel command line option should return "true" to indicate option is valid and so avoid logging it as unknown by the core parsing code. Also log unknown sub-options at the notice level to let user know of potential typos or similar. Reported-by: Eero Tamminen Signed-off-by: Tvrtko Ursulin Link: https://lore.kernel.org/r/20210831112947.310080-1-tvrtko.ursu...@linux.intel.com Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20211014053839.727419-2-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/iommu.c | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index d75f59ae28e6e..9a356075d3450 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -412,6 +412,7 @@ static int __init intel_iommu_setup(char *str) { if (!str) return -EINVAL; + while (*str) { if (!strncmp(str, "on", 2)) { dmar_disabled = 0; @@ -441,13 +442,16 @@ static int __init intel_iommu_setup(char *str) } else if (!strncmp(str, "tboot_noforce", 13)) { pr_info("Intel-IOMMU: not forcing on after tboot. This could expose security risk for tboot\n"); intel_iommu_tboot_noforce = 1; + } else { + pr_notice("Unknown option - '%s'\n", str); } str += strcspn(str, ","); while (*str == ',') str++; } - return 0; + + return 1; } __setup("intel_iommu=", intel_iommu_setup); -- 2.33.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 29/40] swiotlb-xen: ensure to issue well-formed XENMEM_exchange requests
From: Jan Beulich [ Upstream commit 9074c79b62b6e0d91d7f716c6e4e9968eaf9e043 ] While the hypervisor hasn't been enforcing this, we would still better avoid issuing requests with GFNs not aligned to the requested order. Instead of altering the value also in the call to panic(), drop it there for being static and hence easy to determine without being part of the panic message. Signed-off-by: Jan Beulich Reviewed-by: Stefano Stabellini Link: https://lore.kernel.org/r/7b3998e3-1233-4e5a-89ec-d740e77eb...@suse.com Signed-off-by: Juergen Gross Signed-off-by: Sasha Levin --- drivers/xen/swiotlb-xen.c | 7 --- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c index dbb18dc956f3..de4f55154d49 100644 --- a/drivers/xen/swiotlb-xen.c +++ b/drivers/xen/swiotlb-xen.c @@ -232,10 +232,11 @@ void __init xen_swiotlb_init_early(void) /* * Get IO TLB memory from any location. */ - start = memblock_alloc(PAGE_ALIGN(bytes), PAGE_SIZE); + start = memblock_alloc(PAGE_ALIGN(bytes), + IO_TLB_SEGSIZE << IO_TLB_SHIFT); if (!start) - panic("%s: Failed to allocate %lu bytes align=0x%lx\n", - __func__, PAGE_ALIGN(bytes), PAGE_SIZE); + panic("%s: Failed to allocate %lu bytes\n", + __func__, PAGE_ALIGN(bytes)); /* * And replace that memory with pages under 4GB. -- 2.33.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 08/34] dma-debug: prevent an error message from causing runtime problems
From: Hamza Mahfooz [ Upstream commit 510e1a724ab1bf38150be2c1acabb303f98d0047 ] For some drivers, that use the DMA API. This error message can be reached several millions of times per second, causing spam to the kernel's printk buffer and bringing the CPU usage up to 100% (so, it should be rate limited). However, since there is at least one driver that is in the mainline and suffers from the error condition, it is more useful to err_printk() here instead of just rate limiting the error message (in hopes that it will make it easier for other drivers that suffer from this issue to be spotted). Link: https://lkml.kernel.org/r/fd67fbac-64bf-f0ea-01e1-5938ccfab...@arm.com Reported-by: Jeremy Linton Signed-off-by: Hamza Mahfooz Signed-off-by: Christoph Hellwig Signed-off-by: Sasha Levin --- kernel/dma/debug.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c index f2faa13534e5..70519f67556f 100644 --- a/kernel/dma/debug.c +++ b/kernel/dma/debug.c @@ -567,7 +567,8 @@ static void add_dma_entry(struct dma_debug_entry *entry) pr_err("cacheline tracking ENOMEM, dma-debug disabled\n"); global_disable = true; } else if (rc == -EEXIST) { - pr_err("cacheline tracking EEXIST, overlapping mappings aren't supported\n"); + err_printk(entry->dev, entry, + "cacheline tracking EEXIST, overlapping mappings aren't supported\n"); } } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 27/53] iommu/vt-d: Add present bit check in pasid entry setup helpers
From: Liu Yi L [ Upstream commit 423d39d8518c1bba12e0889a92beeddbb1502392 ] The helper functions should not modify the pasid entries which are still in use. Add a check against present bit. Signed-off-by: Liu Yi L Link: https://lore.kernel.org/r/20210817042425.1784279-1-yi.l@intel.com Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20210818134852.1847070-10-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/pasid.c | 16 1 file changed, 16 insertions(+) diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c index fb911b6c418f..6db3400df484 100644 --- a/drivers/iommu/intel/pasid.c +++ b/drivers/iommu/intel/pasid.c @@ -523,6 +523,10 @@ void intel_pasid_tear_down_entry(struct intel_iommu *iommu, struct device *dev, devtlb_invalidation_with_pasid(iommu, dev, pasid); } +/* + * This function flushes cache for a newly setup pasid table entry. + * Caller of it should not modify the in-use pasid table entries. + */ static void pasid_flush_caches(struct intel_iommu *iommu, struct pasid_entry *pte, u32 pasid, u16 did) @@ -558,6 +562,10 @@ int intel_pasid_setup_first_level(struct intel_iommu *iommu, if (WARN_ON(!pte)) return -EINVAL; + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); /* Setup the first level page table pointer: */ @@ -654,6 +662,10 @@ int intel_pasid_setup_second_level(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_slptr(pte, pgd_val); @@ -693,6 +705,10 @@ int intel_pasid_setup_pass_through(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_address_width(pte, iommu->agaw); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 21/53] iommu/arm-smmu: Fix race condition during iommu_group creation
From: Krishna Reddy [ Upstream commit b1a1347912a742a4e1fcdc9df6302dd9dd2c3405 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_group sometimes is getting allocated more than once as call to arm_smmu_device_group() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_group and domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_group allocation from concurrency in arm_smmu_device_group(). Signed-off-by: Krishna Reddy Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-3-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index df24bbe3ea4f..7a0fea12d0e5 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1475,6 +1475,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) struct iommu_group *group = NULL; int i, idx; + mutex_lock(&smmu->stream_map_mutex); for_each_cfg_sme(cfg, fwspec, i, idx) { if (group && smmu->s2crs[idx].group && group != smmu->s2crs[idx].group) @@ -1483,8 +1484,10 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) group = smmu->s2crs[idx].group; } - if (group) + if (group) { + mutex_unlock(&smmu->stream_map_mutex); return iommu_group_ref_get(group); + } if (dev_is_pci(dev)) group = pci_device_group(dev); @@ -1498,6 +1501,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) for_each_cfg_sme(cfg, fwspec, i, idx) smmu->s2crs[idx].group = group; + mutex_unlock(&smmu->stream_map_mutex); return group; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 20/53] iommu: Fix race condition during default domain allocation
From: Ashish Mhetre [ Upstream commit 211ff31b3d33b56aa12937e898c9280d07daf0d9 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_domain sometimes is getting allocated more than once as call to iommu_alloc_default_domain() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_alloc_default_domain() call with group->mutex and let all devices with same SID share same iommu_domain. Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-2-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index bcf060b5cf85..a103a503a4db 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -262,7 +262,9 @@ int iommu_probe_device(struct device *dev) * support default domains, so the return value is not yet * checked. */ + mutex_lock(&group->mutex); iommu_alloc_default_domain(group, dev); + mutex_unlock(&group->mutex); if (group->default_domain) { ret = __iommu_attach_device(group->default_domain, dev); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 06/53] iommu/amd: Fix printing of IOMMU events when rate limiting kicks in
From: Lennert Buytenhek [ Upstream commit ee974d9625c405977ef5d9aedc476be1d0362ebf ] For the printing of RMP_HW_ERROR / RMP_PAGE_FAULT / IO_PAGE_FAULT events, the AMD IOMMU code uses such logic: if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); if (dev_data && __ratelimit(&dev_data->rs)) { pci_err(pdev, ... } else { printk_ratelimit() / pr_err{,_ratelimited}(... } This means that if we receive an event for a PCI devid which actually does have a struct pci_dev and an attached struct iommu_dev_data, but rate limiting kicks in, we'll fall back to the non-PCI branch of the test, and print the event in a different format. Fix this by changing the logic to: if (dev_data) { if (__ratelimit(&dev_data->rs)) { pci_err(pdev, ... } } else { pr_err_ratelimited(... } Suggested-by: Suravee Suthikulpanit Signed-off-by: Lennert Buytenhek Reviewed-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/ypgk1dd1gpmhj...@wantstofly.org Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/iommu.c | 28 +--- 1 file changed, 17 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index 5f1195791cb1..4c97aa81b836 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -502,9 +502,11 @@ static void amd_iommu_report_rmp_hw_error(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", - vmg_tag, spa, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", + vmg_tag, spa, flags); + } } else { pr_err_ratelimited("Event logged [RMP_HW_ERROR device=%02x:%02x.%x, vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -533,9 +535,11 @@ static void amd_iommu_report_rmp_fault(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", - vmg_tag, gpa, flags_rmp, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", + vmg_tag, gpa, flags_rmp, flags); + } } else { pr_err_ratelimited("Event logged [RMP_PAGE_FAULT device=%02x:%02x.%x, vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -557,11 +561,13 @@ static void amd_iommu_report_page_fault(u16 devid, u16 domain_id, if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", - domain_id, address, flags); - } else if (printk_ratelimit()) { - pr_err("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", + domain_id, address, flags); + } + } else { + pr_err_ratelimited("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), domain_id, address, flags); } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 44/88] iommu/vt-d: Add present bit check in pasid entry setup helpers
From: Liu Yi L [ Upstream commit 423d39d8518c1bba12e0889a92beeddbb1502392 ] The helper functions should not modify the pasid entries which are still in use. Add a check against present bit. Signed-off-by: Liu Yi L Link: https://lore.kernel.org/r/20210817042425.1784279-1-yi.l@intel.com Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20210818134852.1847070-10-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/pasid.c | 16 1 file changed, 16 insertions(+) diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c index 581c694b7cf4..a4e79d52be27 100644 --- a/drivers/iommu/intel/pasid.c +++ b/drivers/iommu/intel/pasid.c @@ -540,6 +540,10 @@ void intel_pasid_tear_down_entry(struct intel_iommu *iommu, struct device *dev, devtlb_invalidation_with_pasid(iommu, dev, pasid); } +/* + * This function flushes cache for a newly setup pasid table entry. + * Caller of it should not modify the in-use pasid table entries. + */ static void pasid_flush_caches(struct intel_iommu *iommu, struct pasid_entry *pte, u32 pasid, u16 did) @@ -591,6 +595,10 @@ int intel_pasid_setup_first_level(struct intel_iommu *iommu, if (WARN_ON(!pte)) return -EINVAL; + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); /* Setup the first level page table pointer: */ @@ -690,6 +698,10 @@ int intel_pasid_setup_second_level(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_slptr(pte, pgd_val); @@ -729,6 +741,10 @@ int intel_pasid_setup_pass_through(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_address_width(pte, iommu->agaw); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 36/88] iommu/arm-smmu: Fix race condition during iommu_group creation
From: Krishna Reddy [ Upstream commit b1a1347912a742a4e1fcdc9df6302dd9dd2c3405 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_group sometimes is getting allocated more than once as call to arm_smmu_device_group() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_group and domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_group allocation from concurrency in arm_smmu_device_group(). Signed-off-by: Krishna Reddy Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-3-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index 1a647e0ea3eb..5b82a08ef4b4 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1462,6 +1462,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) struct iommu_group *group = NULL; int i, idx; + mutex_lock(&smmu->stream_map_mutex); for_each_cfg_sme(cfg, fwspec, i, idx) { if (group && smmu->s2crs[idx].group && group != smmu->s2crs[idx].group) @@ -1470,8 +1471,10 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) group = smmu->s2crs[idx].group; } - if (group) + if (group) { + mutex_unlock(&smmu->stream_map_mutex); return iommu_group_ref_get(group); + } if (dev_is_pci(dev)) group = pci_device_group(dev); @@ -1485,6 +1488,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) for_each_cfg_sme(cfg, fwspec, i, idx) smmu->s2crs[idx].group = group; + mutex_unlock(&smmu->stream_map_mutex); return group; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 35/88] iommu: Fix race condition during default domain allocation
From: Ashish Mhetre [ Upstream commit 211ff31b3d33b56aa12937e898c9280d07daf0d9 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_domain sometimes is getting allocated more than once as call to iommu_alloc_default_domain() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_alloc_default_domain() call with group->mutex and let all devices with same SID share same iommu_domain. Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-2-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index db966a7841fe..f4d0bab6be10 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -273,7 +273,9 @@ int iommu_probe_device(struct device *dev) * support default domains, so the return value is not yet * checked. */ + mutex_lock(&group->mutex); iommu_alloc_default_domain(group, dev); + mutex_unlock(&group->mutex); if (group->default_domain) { ret = __iommu_attach_device(group->default_domain, dev); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 22/88] iommu/amd: Sync once for scatter-gather operations
From: Nadav Amit [ Upstream commit 3b122a5666cb7c0bb9a439fba0c9a6cf59f999c3 ] On virtual machines, software must flush the IOTLB after each page table entry update. The iommu_map_sg() code iterates through the given scatter-gather list and invokes iommu_map() for each element in the scatter-gather list, which calls into the vendor IOMMU driver through iommu_ops callback. As the result, a single sg mapping may lead to multiple IOTLB flushes. Fix this by adding amd_iotlb_sync_map() callback and flushing at this point after all sg mappings we set. This commit is followed and inspired by commit 933fcd01e97e2 ("iommu/vt-d: Add iotlb_sync_map callback"). Cc: Joerg Roedel Cc: Will Deacon Cc: Jiajun Cao Cc: Robin Murphy Cc: Lu Baolu Cc: iommu@lists.linux-foundation.org Cc: linux-ker...@vger.kernel.org Signed-off-by: Nadav Amit Link: https://lore.kernel.org/r/20210723093209.714328-7-na...@vmware.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/iommu.c | 15 --- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index 69d441fc1e32..d11386177301 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -2028,6 +2028,16 @@ static int amd_iommu_attach_device(struct iommu_domain *dom, return ret; } +static void amd_iommu_iotlb_sync_map(struct iommu_domain *dom, +unsigned long iova, size_t size) +{ + struct protection_domain *domain = to_pdomain(dom); + struct io_pgtable_ops *ops = &domain->iop.iop.ops; + + if (ops->map) + domain_flush_np_cache(domain, iova, size); +} + static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, phys_addr_t paddr, size_t page_size, int iommu_prot, gfp_t gfp) @@ -2046,10 +2056,8 @@ static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, if (iommu_prot & IOMMU_WRITE) prot |= IOMMU_PROT_IW; - if (ops->map) { + if (ops->map) ret = ops->map(ops, iova, paddr, page_size, prot, gfp); - domain_flush_np_cache(domain, iova, page_size); - } return ret; } @@ -2197,6 +2205,7 @@ const struct iommu_ops amd_iommu_ops = { .attach_dev = amd_iommu_attach_device, .detach_dev = amd_iommu_detach_device, .map = amd_iommu_map, + .iotlb_sync_map = amd_iommu_iotlb_sync_map, .unmap = amd_iommu_unmap, .iova_to_phys = amd_iommu_iova_to_phys, .probe_device = amd_iommu_probe_device, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 16/88] iommu/amd: Fix printing of IOMMU events when rate limiting kicks in
From: Lennert Buytenhek [ Upstream commit ee974d9625c405977ef5d9aedc476be1d0362ebf ] For the printing of RMP_HW_ERROR / RMP_PAGE_FAULT / IO_PAGE_FAULT events, the AMD IOMMU code uses such logic: if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); if (dev_data && __ratelimit(&dev_data->rs)) { pci_err(pdev, ... } else { printk_ratelimit() / pr_err{,_ratelimited}(... } This means that if we receive an event for a PCI devid which actually does have a struct pci_dev and an attached struct iommu_dev_data, but rate limiting kicks in, we'll fall back to the non-PCI branch of the test, and print the event in a different format. Fix this by changing the logic to: if (dev_data) { if (__ratelimit(&dev_data->rs)) { pci_err(pdev, ... } } else { pr_err_ratelimited(... } Suggested-by: Suravee Suthikulpanit Signed-off-by: Lennert Buytenhek Reviewed-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/ypgk1dd1gpmhj...@wantstofly.org Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/iommu.c | 28 +--- 1 file changed, 17 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index c46dde88a132..69d441fc1e32 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -425,9 +425,11 @@ static void amd_iommu_report_rmp_hw_error(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", - vmg_tag, spa, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", + vmg_tag, spa, flags); + } } else { pr_err_ratelimited("Event logged [RMP_HW_ERROR device=%02x:%02x.%x, vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -456,9 +458,11 @@ static void amd_iommu_report_rmp_fault(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", - vmg_tag, gpa, flags_rmp, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", + vmg_tag, gpa, flags_rmp, flags); + } } else { pr_err_ratelimited("Event logged [RMP_PAGE_FAULT device=%02x:%02x.%x, vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -480,11 +484,13 @@ static void amd_iommu_report_page_fault(u16 devid, u16 domain_id, if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", - domain_id, address, flags); - } else if (printk_ratelimit()) { - pr_err("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", + domain_id, address, flags); + } + } else { + pr_err_ratelimited("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), domain_id, address, flags); } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 02/88] swiotlb: add overflow checks to swiotlb_bounce
From: Dominique Martinet [ Upstream commit 868c9ddc182bc6728bb380cbfb3170734f72c599 ] This is a follow-up on 5f89468e2f06 ("swiotlb: manipulate orig_addr when tlb_addr has offset") which fixed unaligned dma mappings, making sure the following overflows are caught: - offset of the start of the slot within the device bigger than requested address' offset, in other words if the base address given in swiotlb_tbl_map_single to create the mapping (orig_addr) was after the requested address for the sync (tlb_offset) in the same block: |--| block <> mapped part of the block ^ orig_addr ^ invalid tlb_addr for sync - if the resulting offset was bigger than the allocation size this one could happen if the mapping was not until the end. e.g. |--| block <-> mapped part of the block ^ ^ orig_addr invalid tlb_addr Both should never happen so print a warning and bail out without trying to adjust the sizes/offsets: the first one could try to sync from orig_addr to whatever is left of the requested size, but the later really has nothing to sync there... Signed-off-by: Dominique Martinet Cc: Konrad Rzeszutek Wilk Reviewed-by: Bumyong Lee Cc: Christoph Hellwig Signed-off-by: Konrad Rzeszutek Wilk Signed-off-by: Sasha Levin --- kernel/dma/swiotlb.c | 20 +--- 1 file changed, 17 insertions(+), 3 deletions(-) diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index e50df8d8f87e..23f8d0b168c5 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -354,13 +354,27 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size size_t alloc_size = mem->slots[index].alloc_size; unsigned long pfn = PFN_DOWN(orig_addr); unsigned char *vaddr = phys_to_virt(tlb_addr); - unsigned int tlb_offset; + unsigned int tlb_offset, orig_addr_offset; if (orig_addr == INVALID_PHYS_ADDR) return; - tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) - -swiotlb_align_offset(dev, orig_addr); + tlb_offset = tlb_addr & (IO_TLB_SIZE - 1); + orig_addr_offset = swiotlb_align_offset(dev, orig_addr); + if (tlb_offset < orig_addr_offset) { + dev_WARN_ONCE(dev, 1, + "Access before mapping start detected. orig offset %u, requested offset %u.\n", + orig_addr_offset, tlb_offset); + return; + } + + tlb_offset -= orig_addr_offset; + if (tlb_offset > alloc_size) { + dev_WARN_ONCE(dev, 1, + "Buffer overflow detected. Allocation size: %zu. Mapping size: %zu+%u.\n", + alloc_size, size, tlb_offset); + return; + } orig_addr += tlb_offset; alloc_size -= tlb_offset; -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 50/99] iommu/vt-d: Add present bit check in pasid entry setup helpers
From: Liu Yi L [ Upstream commit 423d39d8518c1bba12e0889a92beeddbb1502392 ] The helper functions should not modify the pasid entries which are still in use. Add a check against present bit. Signed-off-by: Liu Yi L Link: https://lore.kernel.org/r/20210817042425.1784279-1-yi.l@intel.com Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20210818134852.1847070-10-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/pasid.c | 16 1 file changed, 16 insertions(+) diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c index 9ec374e17469..60c07050f6a7 100644 --- a/drivers/iommu/intel/pasid.c +++ b/drivers/iommu/intel/pasid.c @@ -540,6 +540,10 @@ void intel_pasid_tear_down_entry(struct intel_iommu *iommu, struct device *dev, devtlb_invalidation_with_pasid(iommu, dev, pasid); } +/* + * This function flushes cache for a newly setup pasid table entry. + * Caller of it should not modify the in-use pasid table entries. + */ static void pasid_flush_caches(struct intel_iommu *iommu, struct pasid_entry *pte, u32 pasid, u16 did) @@ -591,6 +595,10 @@ int intel_pasid_setup_first_level(struct intel_iommu *iommu, if (WARN_ON(!pte)) return -EINVAL; + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); /* Setup the first level page table pointer: */ @@ -690,6 +698,10 @@ int intel_pasid_setup_second_level(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_slptr(pte, pgd_val); @@ -729,6 +741,10 @@ int intel_pasid_setup_pass_through(struct intel_iommu *iommu, return -ENODEV; } + /* Caller must ensure PASID entry is not in use. */ + if (pasid_pte_is_present(pte)) + return -EBUSY; + pasid_clear_entry(pte); pasid_set_domain_id(pte, did); pasid_set_address_width(pte, iommu->agaw); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 42/99] iommu/arm-smmu: Fix race condition during iommu_group creation
From: Krishna Reddy [ Upstream commit b1a1347912a742a4e1fcdc9df6302dd9dd2c3405 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_group sometimes is getting allocated more than once as call to arm_smmu_device_group() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_group and domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_group allocation from concurrency in arm_smmu_device_group(). Signed-off-by: Krishna Reddy Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-3-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index f22dbeb1e510..3b4743d830b8 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1478,6 +1478,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) struct iommu_group *group = NULL; int i, idx; + mutex_lock(&smmu->stream_map_mutex); for_each_cfg_sme(cfg, fwspec, i, idx) { if (group && smmu->s2crs[idx].group && group != smmu->s2crs[idx].group) @@ -1486,8 +1487,10 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) group = smmu->s2crs[idx].group; } - if (group) + if (group) { + mutex_unlock(&smmu->stream_map_mutex); return iommu_group_ref_get(group); + } if (dev_is_pci(dev)) group = pci_device_group(dev); @@ -1501,6 +1504,7 @@ static struct iommu_group *arm_smmu_device_group(struct device *dev) for_each_cfg_sme(cfg, fwspec, i, idx) smmu->s2crs[idx].group = group; + mutex_unlock(&smmu->stream_map_mutex); return group; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 41/99] iommu: Fix race condition during default domain allocation
From: Ashish Mhetre [ Upstream commit 211ff31b3d33b56aa12937e898c9280d07daf0d9 ] When two devices with same SID are getting probed concurrently through iommu_probe_device(), the iommu_domain sometimes is getting allocated more than once as call to iommu_alloc_default_domain() is not protected for concurrency. Furthermore, it leads to each device holding a different iommu_domain pointer, separate IOVA space and only one of the devices' domain is used for translations from IOMMU. This causes accesses from other device to fault or see incorrect translations. Fix this by protecting iommu_alloc_default_domain() call with group->mutex and let all devices with same SID share same iommu_domain. Signed-off-by: Ashish Mhetre Link: https://lore.kernel.org/r/1628570641-9127-2-git-send-email-amhe...@nvidia.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/iommu.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index 63f0af10c403..3050cc4b02bc 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -273,7 +273,9 @@ int iommu_probe_device(struct device *dev) * support default domains, so the return value is not yet * checked. */ + mutex_lock(&group->mutex); iommu_alloc_default_domain(group, dev); + mutex_unlock(&group->mutex); if (group->default_domain) { ret = __iommu_attach_device(group->default_domain, dev); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 26/99] iommu/amd: Sync once for scatter-gather operations
From: Nadav Amit [ Upstream commit 3b122a5666cb7c0bb9a439fba0c9a6cf59f999c3 ] On virtual machines, software must flush the IOTLB after each page table entry update. The iommu_map_sg() code iterates through the given scatter-gather list and invokes iommu_map() for each element in the scatter-gather list, which calls into the vendor IOMMU driver through iommu_ops callback. As the result, a single sg mapping may lead to multiple IOTLB flushes. Fix this by adding amd_iotlb_sync_map() callback and flushing at this point after all sg mappings we set. This commit is followed and inspired by commit 933fcd01e97e2 ("iommu/vt-d: Add iotlb_sync_map callback"). Cc: Joerg Roedel Cc: Will Deacon Cc: Jiajun Cao Cc: Robin Murphy Cc: Lu Baolu Cc: iommu@lists.linux-foundation.org Cc: linux-ker...@vger.kernel.org Signed-off-by: Nadav Amit Link: https://lore.kernel.org/r/20210723093209.714328-7-na...@vmware.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/iommu.c | 15 --- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index a7d6d78147b7..e0cd6e42d349 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -2028,6 +2028,16 @@ static int amd_iommu_attach_device(struct iommu_domain *dom, return ret; } +static void amd_iommu_iotlb_sync_map(struct iommu_domain *dom, +unsigned long iova, size_t size) +{ + struct protection_domain *domain = to_pdomain(dom); + struct io_pgtable_ops *ops = &domain->iop.iop.ops; + + if (ops->map) + domain_flush_np_cache(domain, iova, size); +} + static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, phys_addr_t paddr, size_t page_size, int iommu_prot, gfp_t gfp) @@ -2046,10 +2056,8 @@ static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, if (iommu_prot & IOMMU_WRITE) prot |= IOMMU_PROT_IW; - if (ops->map) { + if (ops->map) ret = ops->map(ops, iova, paddr, page_size, prot, gfp); - domain_flush_np_cache(domain, iova, page_size); - } return ret; } @@ -2197,6 +2205,7 @@ const struct iommu_ops amd_iommu_ops = { .attach_dev = amd_iommu_attach_device, .detach_dev = amd_iommu_detach_device, .map = amd_iommu_map, + .iotlb_sync_map = amd_iommu_iotlb_sync_map, .unmap = amd_iommu_unmap, .iova_to_phys = amd_iommu_iova_to_phys, .probe_device = amd_iommu_probe_device, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 20/99] iommu/amd: Fix printing of IOMMU events when rate limiting kicks in
From: Lennert Buytenhek [ Upstream commit ee974d9625c405977ef5d9aedc476be1d0362ebf ] For the printing of RMP_HW_ERROR / RMP_PAGE_FAULT / IO_PAGE_FAULT events, the AMD IOMMU code uses such logic: if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); if (dev_data && __ratelimit(&dev_data->rs)) { pci_err(pdev, ... } else { printk_ratelimit() / pr_err{,_ratelimited}(... } This means that if we receive an event for a PCI devid which actually does have a struct pci_dev and an attached struct iommu_dev_data, but rate limiting kicks in, we'll fall back to the non-PCI branch of the test, and print the event in a different format. Fix this by changing the logic to: if (dev_data) { if (__ratelimit(&dev_data->rs)) { pci_err(pdev, ... } } else { pr_err_ratelimited(... } Suggested-by: Suravee Suthikulpanit Signed-off-by: Lennert Buytenhek Reviewed-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/ypgk1dd1gpmhj...@wantstofly.org Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/iommu.c | 28 +--- 1 file changed, 17 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index 811a49a95d04..a7d6d78147b7 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -425,9 +425,11 @@ static void amd_iommu_report_rmp_hw_error(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", - vmg_tag, spa, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_HW_ERROR vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", + vmg_tag, spa, flags); + } } else { pr_err_ratelimited("Event logged [RMP_HW_ERROR device=%02x:%02x.%x, vmg_tag=0x%04x, spa=0x%llx, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -456,9 +458,11 @@ static void amd_iommu_report_rmp_fault(volatile u32 *event) if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", - vmg_tag, gpa, flags_rmp, flags); + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [RMP_PAGE_FAULT vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", + vmg_tag, gpa, flags_rmp, flags); + } } else { pr_err_ratelimited("Event logged [RMP_PAGE_FAULT device=%02x:%02x.%x, vmg_tag=0x%04x, gpa=0x%llx, flags_rmp=0x%04x, flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), @@ -480,11 +484,13 @@ static void amd_iommu_report_page_fault(u16 devid, u16 domain_id, if (pdev) dev_data = dev_iommu_priv_get(&pdev->dev); - if (dev_data && __ratelimit(&dev_data->rs)) { - pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", - domain_id, address, flags); - } else if (printk_ratelimit()) { - pr_err("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", + if (dev_data) { + if (__ratelimit(&dev_data->rs)) { + pci_err(pdev, "Event logged [IO_PAGE_FAULT domain=0x%04x address=0x%llx flags=0x%04x]\n", + domain_id, address, flags); + } + } else { + pr_err_ratelimited("Event logged [IO_PAGE_FAULT device=%02x:%02x.%x domain=0x%04x address=0x%llx flags=0x%04x]\n", PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid), domain_id, address, flags); } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.14 02/99] swiotlb: add overflow checks to swiotlb_bounce
From: Dominique Martinet [ Upstream commit 868c9ddc182bc6728bb380cbfb3170734f72c599 ] This is a follow-up on 5f89468e2f06 ("swiotlb: manipulate orig_addr when tlb_addr has offset") which fixed unaligned dma mappings, making sure the following overflows are caught: - offset of the start of the slot within the device bigger than requested address' offset, in other words if the base address given in swiotlb_tbl_map_single to create the mapping (orig_addr) was after the requested address for the sync (tlb_offset) in the same block: |--| block <> mapped part of the block ^ orig_addr ^ invalid tlb_addr for sync - if the resulting offset was bigger than the allocation size this one could happen if the mapping was not until the end. e.g. |--| block <-> mapped part of the block ^ ^ orig_addr invalid tlb_addr Both should never happen so print a warning and bail out without trying to adjust the sizes/offsets: the first one could try to sync from orig_addr to whatever is left of the requested size, but the later really has nothing to sync there... Signed-off-by: Dominique Martinet Cc: Konrad Rzeszutek Wilk Reviewed-by: Bumyong Lee Cc: Christoph Hellwig Signed-off-by: Konrad Rzeszutek Wilk Signed-off-by: Sasha Levin --- kernel/dma/swiotlb.c | 20 +--- 1 file changed, 17 insertions(+), 3 deletions(-) diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index e50df8d8f87e..23f8d0b168c5 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -354,13 +354,27 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size size_t alloc_size = mem->slots[index].alloc_size; unsigned long pfn = PFN_DOWN(orig_addr); unsigned char *vaddr = phys_to_virt(tlb_addr); - unsigned int tlb_offset; + unsigned int tlb_offset, orig_addr_offset; if (orig_addr == INVALID_PHYS_ADDR) return; - tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) - -swiotlb_align_offset(dev, orig_addr); + tlb_offset = tlb_addr & (IO_TLB_SIZE - 1); + orig_addr_offset = swiotlb_align_offset(dev, orig_addr); + if (tlb_offset < orig_addr_offset) { + dev_WARN_ONCE(dev, 1, + "Access before mapping start detected. orig offset %u, requested offset %u.\n", + orig_addr_offset, tlb_offset); + return; + } + + tlb_offset -= orig_addr_offset; + if (tlb_offset > alloc_size) { + dev_WARN_ONCE(dev, 1, + "Buffer overflow detected. Allocation size: %zu. Mapping size: %zu+%u.\n", + alloc_size, size, tlb_offset); + return; + } orig_addr += tlb_offset; alloc_size -= tlb_offset; -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 37/63] iommu/arm-smmu: Fix arm_smmu_device refcount leak in address translation
From: Xiyu Yang [ Upstream commit 7c8f176d6a3fa18aa0f8875da6f7c672ed2a8554 ] The reference counting issue happens in several exception handling paths of arm_smmu_iova_to_phys_hard(). When those error scenarios occur, the function forgets to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by jumping to "out" label when those error scenarios occur. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Reviewed-by: Rob Clark Link: https://lore.kernel.org/r/1623293391-17261-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu.c | 8 ++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c index abf4cf285548..2185ea5191c1 100644 --- a/drivers/iommu/arm-smmu.c +++ b/drivers/iommu/arm-smmu.c @@ -1231,6 +1231,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, u64 phys; unsigned long va, flags; int ret, idx = cfg->cbndx; + phys_addr_t addr = 0; ret = arm_smmu_rpm_get(smmu); if (ret < 0) @@ -1249,6 +1250,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, dev_err(dev, "iova to phys timed out on %pad. Falling back to software table walk.\n", &iova); + arm_smmu_rpm_put(smmu); return ops->iova_to_phys(ops, iova); } @@ -1257,12 +1259,14 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, if (phys & CB_PAR_F) { dev_err(dev, "translation fault!\n"); dev_err(dev, "PAR = 0x%llx\n", phys); - return 0; + goto out; } + addr = (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); +out: arm_smmu_rpm_put(smmu); - return (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); + return addr; } static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 36/63] iommu/arm-smmu: Fix arm_smmu_device refcount leak when arm_smmu_rpm_get fails
From: Xiyu Yang [ Upstream commit 1adf30f198c26539a62d761e45af72cde570413d ] arm_smmu_rpm_get() invokes pm_runtime_get_sync(), which increases the refcount of the "smmu" even though the return value is less than 0. The reference counting issue happens in some error handling paths of arm_smmu_rpm_get() in its caller functions. When arm_smmu_rpm_get() fails, the caller functions forget to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by calling pm_runtime_resume_and_get() instead of pm_runtime_get_sync() in arm_smmu_rpm_get(), which can keep the refcount balanced in case of failure. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Link: https://lore.kernel.org/r/1623293672-17954-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm-smmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c index 7c503a6bc585..abf4cf285548 100644 --- a/drivers/iommu/arm-smmu.c +++ b/drivers/iommu/arm-smmu.c @@ -114,7 +114,7 @@ static bool using_legacy_binding, using_generic_binding; static inline int arm_smmu_rpm_get(struct arm_smmu_device *smmu) { if (pm_runtime_enabled(smmu->dev)) - return pm_runtime_get_sync(smmu->dev); + return pm_runtime_resume_and_get(smmu->dev); return 0; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 57/93] iommu/arm-smmu: Fix arm_smmu_device refcount leak in address translation
From: Xiyu Yang [ Upstream commit 7c8f176d6a3fa18aa0f8875da6f7c672ed2a8554 ] The reference counting issue happens in several exception handling paths of arm_smmu_iova_to_phys_hard(). When those error scenarios occur, the function forgets to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by jumping to "out" label when those error scenarios occur. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Reviewed-by: Rob Clark Link: https://lore.kernel.org/r/1623293391-17261-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 8 ++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index 052f0a1bf037..df24bbe3ea4f 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1284,6 +1284,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, u64 phys; unsigned long va, flags; int ret, idx = cfg->cbndx; + phys_addr_t addr = 0; ret = arm_smmu_rpm_get(smmu); if (ret < 0) @@ -1303,6 +1304,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, dev_err(dev, "iova to phys timed out on %pad. Falling back to software table walk.\n", &iova); + arm_smmu_rpm_put(smmu); return ops->iova_to_phys(ops, iova); } @@ -1311,12 +1313,14 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, if (phys & ARM_SMMU_CB_PAR_F) { dev_err(dev, "translation fault!\n"); dev_err(dev, "PAR = 0x%llx\n", phys); - return 0; + goto out; } + addr = (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); +out: arm_smmu_rpm_put(smmu); - return (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); + return addr; } static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 56/93] iommu/arm-smmu: Fix arm_smmu_device refcount leak when arm_smmu_rpm_get fails
From: Xiyu Yang [ Upstream commit 1adf30f198c26539a62d761e45af72cde570413d ] arm_smmu_rpm_get() invokes pm_runtime_get_sync(), which increases the refcount of the "smmu" even though the return value is less than 0. The reference counting issue happens in some error handling paths of arm_smmu_rpm_get() in its caller functions. When arm_smmu_rpm_get() fails, the caller functions forget to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by calling pm_runtime_resume_and_get() instead of pm_runtime_get_sync() in arm_smmu_rpm_get(), which can keep the refcount balanced in case of failure. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Link: https://lore.kernel.org/r/1623293672-17954-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index bcbacf22331d..052f0a1bf037 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -74,7 +74,7 @@ static bool using_legacy_binding, using_generic_binding; static inline int arm_smmu_rpm_get(struct arm_smmu_device *smmu) { if (pm_runtime_enabled(smmu->dev)) - return pm_runtime_get_sync(smmu->dev); + return pm_runtime_resume_and_get(smmu->dev); return 0; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 061/104] iommu/arm-smmu: Fix arm_smmu_device refcount leak in address translation
From: Xiyu Yang [ Upstream commit 7c8f176d6a3fa18aa0f8875da6f7c672ed2a8554 ] The reference counting issue happens in several exception handling paths of arm_smmu_iova_to_phys_hard(). When those error scenarios occur, the function forgets to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by jumping to "out" label when those error scenarios occur. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Reviewed-by: Rob Clark Link: https://lore.kernel.org/r/1623293391-17261-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 8 ++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index 128c2c87b4e5..c6ff32797a23 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1268,6 +1268,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, u64 phys; unsigned long va, flags; int ret, idx = cfg->cbndx; + phys_addr_t addr = 0; ret = arm_smmu_rpm_get(smmu); if (ret < 0) @@ -1287,6 +1288,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, dev_err(dev, "iova to phys timed out on %pad. Falling back to software table walk.\n", &iova); + arm_smmu_rpm_put(smmu); return ops->iova_to_phys(ops, iova); } @@ -1295,12 +1297,14 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, if (phys & ARM_SMMU_CB_PAR_F) { dev_err(dev, "translation fault!\n"); dev_err(dev, "PAR = 0x%llx\n", phys); - return 0; + goto out; } + addr = (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); +out: arm_smmu_rpm_put(smmu); - return (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); + return addr; } static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 060/104] iommu/arm-smmu: Fix arm_smmu_device refcount leak when arm_smmu_rpm_get fails
From: Xiyu Yang [ Upstream commit 1adf30f198c26539a62d761e45af72cde570413d ] arm_smmu_rpm_get() invokes pm_runtime_get_sync(), which increases the refcount of the "smmu" even though the return value is less than 0. The reference counting issue happens in some error handling paths of arm_smmu_rpm_get() in its caller functions. When arm_smmu_rpm_get() fails, the caller functions forget to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by calling pm_runtime_resume_and_get() instead of pm_runtime_get_sync() in arm_smmu_rpm_get(), which can keep the refcount balanced in case of failure. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Link: https://lore.kernel.org/r/1623293672-17954-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index d8c6bfde6a61..128c2c87b4e5 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -74,7 +74,7 @@ static bool using_legacy_binding, using_generic_binding; static inline int arm_smmu_rpm_get(struct arm_smmu_device *smmu) { if (pm_runtime_enabled(smmu->dev)) - return pm_runtime_get_sync(smmu->dev); + return pm_runtime_resume_and_get(smmu->dev); return 0; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 056/104] iommu/arm-smmu-qcom: Skip the TTBR1 quirk for db820c.
From: Eric Anholt [ Upstream commit a242f4297cfe3f4589a7620dcd42cc503607fc6b ] db820c wants to use the qcom smmu path to get HUPCF set (which keeps the GPU from wedging and then sometimes wedging the kernel after a page fault), but it doesn't have separate pagetables support yet in drm/msm so we can't go all the way to the TTBR1 path. Signed-off-by: Eric Anholt Reviewed-by: Bjorn Andersson Link: https://lore.kernel.org/r/20210326231303.3071950-1-e...@anholt.net Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 13 - 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c index 98b3a1c2a181..44a427833385 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c @@ -130,6 +130,16 @@ static int qcom_adreno_smmu_alloc_context_bank(struct arm_smmu_domain *smmu_doma return __arm_smmu_alloc_bitmap(smmu->context_map, start, count); } +static bool qcom_adreno_can_do_ttbr1(struct arm_smmu_device *smmu) +{ + const struct device_node *np = smmu->dev->of_node; + + if (of_device_is_compatible(np, "qcom,msm8996-smmu-v2")) + return false; + + return true; +} + static int qcom_adreno_smmu_init_context(struct arm_smmu_domain *smmu_domain, struct io_pgtable_cfg *pgtbl_cfg, struct device *dev) { @@ -144,7 +154,8 @@ static int qcom_adreno_smmu_init_context(struct arm_smmu_domain *smmu_domain, * be AARCH64 stage 1 but double check because the arm-smmu code assumes * that is the case when the TTBR1 quirk is enabled */ - if ((smmu_domain->stage == ARM_SMMU_DOMAIN_S1) && + if (qcom_adreno_can_do_ttbr1(smmu_domain->smmu) && + (smmu_domain->stage == ARM_SMMU_DOMAIN_S1) && (smmu_domain->cfg.fmt == ARM_SMMU_CTX_FMT_AARCH64)) pgtbl_cfg->quirks |= IO_PGTABLE_QUIRK_ARM_TTBR1; -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 065/114] iommu/arm-smmu: Fix arm_smmu_device refcount leak when arm_smmu_rpm_get fails
From: Xiyu Yang [ Upstream commit 1adf30f198c26539a62d761e45af72cde570413d ] arm_smmu_rpm_get() invokes pm_runtime_get_sync(), which increases the refcount of the "smmu" even though the return value is less than 0. The reference counting issue happens in some error handling paths of arm_smmu_rpm_get() in its caller functions. When arm_smmu_rpm_get() fails, the caller functions forget to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by calling pm_runtime_resume_and_get() instead of pm_runtime_get_sync() in arm_smmu_rpm_get(), which can keep the refcount balanced in case of failure. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Link: https://lore.kernel.org/r/1623293672-17954-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index 6f72c4d208ca..ee6cac9e7c02 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -74,7 +74,7 @@ static bool using_legacy_binding, using_generic_binding; static inline int arm_smmu_rpm_get(struct arm_smmu_device *smmu) { if (pm_runtime_enabled(smmu->dev)) - return pm_runtime_get_sync(smmu->dev); + return pm_runtime_resume_and_get(smmu->dev); return 0; } -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 066/114] iommu/arm-smmu: Fix arm_smmu_device refcount leak in address translation
From: Xiyu Yang [ Upstream commit 7c8f176d6a3fa18aa0f8875da6f7c672ed2a8554 ] The reference counting issue happens in several exception handling paths of arm_smmu_iova_to_phys_hard(). When those error scenarios occur, the function forgets to decrease the refcount of "smmu" increased by arm_smmu_rpm_get(), causing a refcount leak. Fix this issue by jumping to "out" label when those error scenarios occur. Signed-off-by: Xiyu Yang Signed-off-by: Xin Tan Reviewed-by: Rob Clark Link: https://lore.kernel.org/r/1623293391-17261-1-git-send-email-xiyuyan...@fudan.edu.cn Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu.c | 8 ++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c index ee6cac9e7c02..1a647e0ea3eb 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c @@ -1271,6 +1271,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, u64 phys; unsigned long va, flags; int ret, idx = cfg->cbndx; + phys_addr_t addr = 0; ret = arm_smmu_rpm_get(smmu); if (ret < 0) @@ -1290,6 +1291,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, dev_err(dev, "iova to phys timed out on %pad. Falling back to software table walk.\n", &iova); + arm_smmu_rpm_put(smmu); return ops->iova_to_phys(ops, iova); } @@ -1298,12 +1300,14 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain, if (phys & ARM_SMMU_CB_PAR_F) { dev_err(dev, "translation fault!\n"); dev_err(dev, "PAR = 0x%llx\n", phys); - return 0; + goto out; } + addr = (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); +out: arm_smmu_rpm_put(smmu); - return (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff); + return addr; } static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.13 061/114] iommu/arm-smmu-qcom: Skip the TTBR1 quirk for db820c.
From: Eric Anholt [ Upstream commit a242f4297cfe3f4589a7620dcd42cc503607fc6b ] db820c wants to use the qcom smmu path to get HUPCF set (which keeps the GPU from wedging and then sometimes wedging the kernel after a page fault), but it doesn't have separate pagetables support yet in drm/msm so we can't go all the way to the TTBR1 path. Signed-off-by: Eric Anholt Reviewed-by: Bjorn Andersson Link: https://lore.kernel.org/r/20210326231303.3071950-1-e...@anholt.net Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 13 - 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c index 98b3a1c2a181..44a427833385 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c @@ -130,6 +130,16 @@ static int qcom_adreno_smmu_alloc_context_bank(struct arm_smmu_domain *smmu_doma return __arm_smmu_alloc_bitmap(smmu->context_map, start, count); } +static bool qcom_adreno_can_do_ttbr1(struct arm_smmu_device *smmu) +{ + const struct device_node *np = smmu->dev->of_node; + + if (of_device_is_compatible(np, "qcom,msm8996-smmu-v2")) + return false; + + return true; +} + static int qcom_adreno_smmu_init_context(struct arm_smmu_domain *smmu_domain, struct io_pgtable_cfg *pgtbl_cfg, struct device *dev) { @@ -144,7 +154,8 @@ static int qcom_adreno_smmu_init_context(struct arm_smmu_domain *smmu_domain, * be AARCH64 stage 1 but double check because the arm-smmu code assumes * that is the case when the TTBR1 quirk is enabled */ - if ((smmu_domain->stage == ARM_SMMU_DOMAIN_S1) && + if (qcom_adreno_can_do_ttbr1(smmu_domain->smmu) && + (smmu_domain->stage == ARM_SMMU_DOMAIN_S1) && (smmu_domain->cfg.fmt == ARM_SMMU_CTX_FMT_AARCH64)) pgtbl_cfg->quirks |= IO_PGTABLE_QUIRK_ARM_TTBR1; -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 25/46] iommu/amd: Remove performance counter pre-initialization test
From: Suravee Suthikulpanit [ Upstream commit 994d6608efe4a4c8834bdc5014c86f4bc6aceea6 ] In early AMD desktop/mobile platforms (during 2013), when the IOMMU Performance Counter (PMC) support was first introduced in commit 30861ddc9cca ("perf/x86/amd: Add IOMMU Performance Counter resource management"), there was a HW bug where the counters could not be accessed. The result was reading of the counter always return zero. At the time, the suggested workaround was to add a test logic prior to initializing the PMC feature to check if the counters can be programmed and read back the same value. This has been working fine until the more recent desktop/mobile platforms start enabling power gating for the PMC, which prevents access to the counters. This results in the PMC support being disabled unnecesarily. Unfortunatly, there is no documentation of since which generation of hardware the original PMC HW bug was fixed. Although, it was fixed soon after the first introduction of the PMC. Base on this, we assume that the buggy platforms are less likely to be in used, and it should be relatively safe to remove this legacy logic. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Cc: Paul Menzel Signed-off-by: Suravee Suthikulpanit Tested-by: Shuah Khan Link: https://lore.kernel.org/r/20210409085848.3908-3-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_init.c | 24 +--- 1 file changed, 1 insertion(+), 23 deletions(-) diff --git a/drivers/iommu/amd_iommu_init.c b/drivers/iommu/amd_iommu_init.c index 31d7e2d4f304..692401e941a7 100644 --- a/drivers/iommu/amd_iommu_init.c +++ b/drivers/iommu/amd_iommu_init.c @@ -1672,33 +1672,16 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - static void init_iommu_perf_ctr(struct amd_iommu *iommu) { + u64 val; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; amd_iommu_pc_present = true; - /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) - goto pc_false; - - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) - goto pc_false; - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) - goto pc_false; - pci_info(pdev, "IOMMU performance counters supported\n"); val = readl(iommu->mmio_base + MMIO_CNTR_CONF_OFFSET); @@ -1706,11 +1689,6 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) iommu->max_counters = (u8) ((val >> 7) & 0xf); return; - -pc_false: - pci_err(pdev, "Unable to read/write to IOMMU perf counter.\n"); - amd_iommu_pc_present = false; - return; } static ssize_t amd_iommu_show_cap(struct device *dev, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 24/46] Revert "iommu/amd: Fix performance counter initialization"
From: Paul Menzel [ Upstream commit 715601e4e36903a653cd4294dfd3ed0019101991 ] This reverts commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b. The original commit tries to address an issue, where PMC power-gating causing the IOMMU PMC pre-init test to fail on certain desktop/mobile platforms where the power-gating is normally enabled. There have been several reports that the workaround still does not guarantee to work, and can add up to 100 ms (on the worst case) to the boot process on certain platforms such as the MSI B350M MORTAR with AMD Ryzen 3 2200G. Therefore, revert this commit as a prelude to removing the pre-init test. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Signed-off-by: Paul Menzel Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20210409085848.3908-2-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_init.c | 45 +- 1 file changed, 11 insertions(+), 34 deletions(-) diff --git a/drivers/iommu/amd_iommu_init.c b/drivers/iommu/amd_iommu_init.c index ad714ff375f8..31d7e2d4f304 100644 --- a/drivers/iommu/amd_iommu_init.c +++ b/drivers/iommu/amd_iommu_init.c @@ -12,7 +12,6 @@ #include #include #include -#include #include #include #include @@ -254,8 +253,6 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1675,11 +1672,13 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); + +static void init_iommu_perf_ctr(struct amd_iommu *iommu) { - int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg, save_src; + u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1687,39 +1686,17 @@ static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) - goto pc_false; - - /* -* Disable power gating by programing the performance counter -* source to 20 (i.e. counts the reads and writes from/to IOMMU -* Reserved Register [MMIO Offset 1FF8h] that are ignored.), -* which never get incremented during this init phase. -* (Note: The event is also deprecated.) -*/ - val = 20; - if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) goto pc_false; /* Check if the performance counters can be written to */ - val = 0xabcd; - for (retry = 5; retry; retry--) { - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || - val2) - break; - - /* Wait about 20 msec for power gating to disable and retry. */ - msleep(20); - } - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || + (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || + (val != val2)) goto pc_false; - if (val != val2) + /* restore */ + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 51/85] iommu/amd: Remove performance counter pre-initialization test
From: Suravee Suthikulpanit [ Upstream commit 994d6608efe4a4c8834bdc5014c86f4bc6aceea6 ] In early AMD desktop/mobile platforms (during 2013), when the IOMMU Performance Counter (PMC) support was first introduced in commit 30861ddc9cca ("perf/x86/amd: Add IOMMU Performance Counter resource management"), there was a HW bug where the counters could not be accessed. The result was reading of the counter always return zero. At the time, the suggested workaround was to add a test logic prior to initializing the PMC feature to check if the counters can be programmed and read back the same value. This has been working fine until the more recent desktop/mobile platforms start enabling power gating for the PMC, which prevents access to the counters. This results in the PMC support being disabled unnecesarily. Unfortunatly, there is no documentation of since which generation of hardware the original PMC HW bug was fixed. Although, it was fixed soon after the first introduction of the PMC. Base on this, we assume that the buggy platforms are less likely to be in used, and it should be relatively safe to remove this legacy logic. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Cc: Paul Menzel Signed-off-by: Suravee Suthikulpanit Tested-by: Shuah Khan Link: https://lore.kernel.org/r/20210409085848.3908-3-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 24 +--- 1 file changed, 1 insertion(+), 23 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index c842545368fd..50307c2d42c0 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -1717,33 +1717,16 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - static void init_iommu_perf_ctr(struct amd_iommu *iommu) { + u64 val; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; amd_iommu_pc_present = true; - /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) - goto pc_false; - - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) - goto pc_false; - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) - goto pc_false; - pci_info(pdev, "IOMMU performance counters supported\n"); val = readl(iommu->mmio_base + MMIO_CNTR_CONF_OFFSET); @@ -1751,11 +1734,6 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) iommu->max_counters = (u8) ((val >> 7) & 0xf); return; - -pc_false: - pci_err(pdev, "Unable to read/write to IOMMU perf counter.\n"); - amd_iommu_pc_present = false; - return; } static ssize_t amd_iommu_show_cap(struct device *dev, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 50/85] Revert "iommu/amd: Fix performance counter initialization"
From: Paul Menzel [ Upstream commit 715601e4e36903a653cd4294dfd3ed0019101991 ] This reverts commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b. The original commit tries to address an issue, where PMC power-gating causing the IOMMU PMC pre-init test to fail on certain desktop/mobile platforms where the power-gating is normally enabled. There have been several reports that the workaround still does not guarantee to work, and can add up to 100 ms (on the worst case) to the boot process on certain platforms such as the MSI B350M MORTAR with AMD Ryzen 3 2200G. Therefore, revert this commit as a prelude to removing the pre-init test. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Signed-off-by: Paul Menzel Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20210409085848.3908-2-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 45 ++-- 1 file changed, 11 insertions(+), 34 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 3c215f0a6052..c842545368fd 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -12,7 +12,6 @@ #include #include #include -#include #include #include #include @@ -255,8 +254,6 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1720,11 +1717,13 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); + +static void init_iommu_perf_ctr(struct amd_iommu *iommu) { - int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg, save_src; + u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1732,39 +1731,17 @@ static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) - goto pc_false; - - /* -* Disable power gating by programing the performance counter -* source to 20 (i.e. counts the reads and writes from/to IOMMU -* Reserved Register [MMIO Offset 1FF8h] that are ignored.), -* which never get incremented during this init phase. -* (Note: The event is also deprecated.) -*/ - val = 20; - if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) goto pc_false; /* Check if the performance counters can be written to */ - val = 0xabcd; - for (retry = 5; retry; retry--) { - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || - val2) - break; - - /* Wait about 20 msec for power gating to disable and retry. */ - msleep(20); - } - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || + (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || + (val != val2)) goto pc_false; - if (val != val2) + /* restore */ + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.11 067/104] iommu/amd: Remove performance counter pre-initialization test
From: Suravee Suthikulpanit [ Upstream commit 994d6608efe4a4c8834bdc5014c86f4bc6aceea6 ] In early AMD desktop/mobile platforms (during 2013), when the IOMMU Performance Counter (PMC) support was first introduced in commit 30861ddc9cca ("perf/x86/amd: Add IOMMU Performance Counter resource management"), there was a HW bug where the counters could not be accessed. The result was reading of the counter always return zero. At the time, the suggested workaround was to add a test logic prior to initializing the PMC feature to check if the counters can be programmed and read back the same value. This has been working fine until the more recent desktop/mobile platforms start enabling power gating for the PMC, which prevents access to the counters. This results in the PMC support being disabled unnecesarily. Unfortunatly, there is no documentation of since which generation of hardware the original PMC HW bug was fixed. Although, it was fixed soon after the first introduction of the PMC. Base on this, we assume that the buggy platforms are less likely to be in used, and it should be relatively safe to remove this legacy logic. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Cc: Paul Menzel Signed-off-by: Suravee Suthikulpanit Tested-by: Shuah Khan Link: https://lore.kernel.org/r/20210409085848.3908-3-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 24 +--- 1 file changed, 1 insertion(+), 23 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 7720dcfa2b29..3c86b6ff978f 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -1712,33 +1712,16 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - static void init_iommu_perf_ctr(struct amd_iommu *iommu) { + u64 val; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; amd_iommu_pc_present = true; - /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) - goto pc_false; - - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) - goto pc_false; - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) - goto pc_false; - pci_info(pdev, "IOMMU performance counters supported\n"); val = readl(iommu->mmio_base + MMIO_CNTR_CONF_OFFSET); @@ -1746,11 +1729,6 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) iommu->max_counters = (u8) ((val >> 7) & 0xf); return; - -pc_false: - pci_err(pdev, "Unable to read/write to IOMMU perf counter.\n"); - amd_iommu_pc_present = false; - return; } static ssize_t amd_iommu_show_cap(struct device *dev, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.11 066/104] Revert "iommu/amd: Fix performance counter initialization"
From: Paul Menzel [ Upstream commit 715601e4e36903a653cd4294dfd3ed0019101991 ] This reverts commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b. The original commit tries to address an issue, where PMC power-gating causing the IOMMU PMC pre-init test to fail on certain desktop/mobile platforms where the power-gating is normally enabled. There have been several reports that the workaround still does not guarantee to work, and can add up to 100 ms (on the worst case) to the boot process on certain platforms such as the MSI B350M MORTAR with AMD Ryzen 3 2200G. Therefore, revert this commit as a prelude to removing the pre-init test. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Signed-off-by: Paul Menzel Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20210409085848.3908-2-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 45 ++-- 1 file changed, 11 insertions(+), 34 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 78339b0bb8e5..7720dcfa2b29 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -12,7 +12,6 @@ #include #include #include -#include #include #include #include @@ -255,8 +254,6 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1715,11 +1712,13 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); + +static void init_iommu_perf_ctr(struct amd_iommu *iommu) { - int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg, save_src; + u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1727,39 +1726,17 @@ static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) - goto pc_false; - - /* -* Disable power gating by programing the performance counter -* source to 20 (i.e. counts the reads and writes from/to IOMMU -* Reserved Register [MMIO Offset 1FF8h] that are ignored.), -* which never get incremented during this init phase. -* (Note: The event is also deprecated.) -*/ - val = 20; - if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) goto pc_false; /* Check if the performance counters can be written to */ - val = 0xabcd; - for (retry = 5; retry; retry--) { - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || - val2) - break; - - /* Wait about 20 msec for power gating to disable and retry. */ - msleep(20); - } - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || + (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || + (val != val2)) goto pc_false; - if (val != val2) + /* restore */ + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 076/116] iommu/amd: Remove performance counter pre-initialization test
From: Suravee Suthikulpanit [ Upstream commit 994d6608efe4a4c8834bdc5014c86f4bc6aceea6 ] In early AMD desktop/mobile platforms (during 2013), when the IOMMU Performance Counter (PMC) support was first introduced in commit 30861ddc9cca ("perf/x86/amd: Add IOMMU Performance Counter resource management"), there was a HW bug where the counters could not be accessed. The result was reading of the counter always return zero. At the time, the suggested workaround was to add a test logic prior to initializing the PMC feature to check if the counters can be programmed and read back the same value. This has been working fine until the more recent desktop/mobile platforms start enabling power gating for the PMC, which prevents access to the counters. This results in the PMC support being disabled unnecesarily. Unfortunatly, there is no documentation of since which generation of hardware the original PMC HW bug was fixed. Although, it was fixed soon after the first introduction of the PMC. Base on this, we assume that the buggy platforms are less likely to be in used, and it should be relatively safe to remove this legacy logic. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Cc: Paul Menzel Signed-off-by: Suravee Suthikulpanit Tested-by: Shuah Khan Link: https://lore.kernel.org/r/20210409085848.3908-3-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 24 +--- 1 file changed, 1 insertion(+), 23 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 648cdfd03074..247cdda5d683 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -1714,33 +1714,16 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - static void init_iommu_perf_ctr(struct amd_iommu *iommu) { + u64 val; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; amd_iommu_pc_present = true; - /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) - goto pc_false; - - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) - goto pc_false; - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) - goto pc_false; - pci_info(pdev, "IOMMU performance counters supported\n"); val = readl(iommu->mmio_base + MMIO_CNTR_CONF_OFFSET); @@ -1748,11 +1731,6 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) iommu->max_counters = (u8) ((val >> 7) & 0xf); return; - -pc_false: - pci_err(pdev, "Unable to read/write to IOMMU perf counter.\n"); - amd_iommu_pc_present = false; - return; } static ssize_t amd_iommu_show_cap(struct device *dev, -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 075/116] Revert "iommu/amd: Fix performance counter initialization"
From: Paul Menzel [ Upstream commit 715601e4e36903a653cd4294dfd3ed0019101991 ] This reverts commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b. The original commit tries to address an issue, where PMC power-gating causing the IOMMU PMC pre-init test to fail on certain desktop/mobile platforms where the power-gating is normally enabled. There have been several reports that the workaround still does not guarantee to work, and can add up to 100 ms (on the worst case) to the boot process on certain platforms such as the MSI B350M MORTAR with AMD Ryzen 3 2200G. Therefore, revert this commit as a prelude to removing the pre-init test. Link: https://lore.kernel.org/linux-iommu/alpine.lnx.3.20.13.2006030935570.3...@monopod.intra.ispras.ru/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Cc: Tj (Elloe Linux) Cc: Shuah Khan Cc: Alexander Monakov Cc: David Coe Signed-off-by: Paul Menzel Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20210409085848.3908-2-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 45 ++-- 1 file changed, 11 insertions(+), 34 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 321f5906e6ed..648cdfd03074 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -12,7 +12,6 @@ #include #include #include -#include #include #include #include @@ -257,8 +256,6 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1717,11 +1714,13 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); + +static void init_iommu_perf_ctr(struct amd_iommu *iommu) { - int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg, save_src; + u64 val = 0xabcd, val2 = 0, save_reg = 0; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1729,39 +1728,17 @@ static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) - goto pc_false; - - /* -* Disable power gating by programing the performance counter -* source to 20 (i.e. counts the reads and writes from/to IOMMU -* Reserved Register [MMIO Offset 1FF8h] that are ignored.), -* which never get incremented during this init phase. -* (Note: The event is also deprecated.) -*/ - val = 20; - if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) goto pc_false; /* Check if the performance counters can be written to */ - val = 0xabcd; - for (retry = 5; retry; retry--) { - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || - val2) - break; - - /* Wait about 20 msec for power gating to disable and retry. */ - msleep(20); - } - - /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || - iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || + (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || + (val != val2)) goto pc_false; - if (val != val2) + /* restore */ + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.12 036/116] iommu/arm-smmu-v3: Add a check to avoid invalid iotlb sync
From: Xiang Chen [ Upstream commit 6cc7e5a9c6b02507b9be5a99b51e970afa91c85f ] It may send a invalid tlb sync for smmuv3 if iotlb_gather is not valid (iotlb_gather->pgsize = 0). So add a check to avoid invalid iotlb sync for it. Signed-off-by: Xiang Chen Link: https://lore.kernel.org/r/1617109106-121844-1-git-send-email-chenxian...@hisilicon.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c index 8594b4a83043..941ba5484731 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c @@ -2305,6 +2305,9 @@ static void arm_smmu_iotlb_sync(struct iommu_domain *domain, { struct arm_smmu_domain *smmu_domain = to_smmu_domain(domain); + if (!gather->pgsize) + return; + arm_smmu_tlb_inv_range_domain(gather->start, gather->end - gather->start + 1, gather->pgsize, true, smmu_domain); -- 2.30.2 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 15/33] iommu/amd: Fix performance counter initialization
From: Suravee Suthikulpanit [ Upstream commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b ] Certain AMD platforms enable power gating feature for IOMMU PMC, which prevents the IOMMU driver from updating the counter while trying to validate the PMC functionality in the init_iommu_perf_ctr(). This results in disabling PMC support and the following error message: "AMD-Vi: Unable to read/write to IOMMU perf counter" To workaround this issue, disable power gating temporarily by programming the counter source to non-zero value while validating the counter, and restore the prior state afterward. Signed-off-by: Suravee Suthikulpanit Tested-by: Tj (Elloe Linux) Link: https://lore.kernel.org/r/20210208122712.5048-1-suravee.suthikulpa...@amd.com Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_init.c | 45 +- 1 file changed, 34 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd_iommu_init.c b/drivers/iommu/amd_iommu_init.c index 31d7e2d4f304..ad714ff375f8 100644 --- a/drivers/iommu/amd_iommu_init.c +++ b/drivers/iommu/amd_iommu_init.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include #include @@ -253,6 +254,8 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1672,13 +1675,11 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - -static void init_iommu_perf_ctr(struct amd_iommu *iommu) +static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) { + int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; + u64 val = 0xabcd, val2 = 0, save_reg, save_src; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1686,17 +1687,39 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) goto pc_false; - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) + /* +* Disable power gating by programing the performance counter +* source to 20 (i.e. counts the reads and writes from/to IOMMU +* Reserved Register [MMIO Offset 1FF8h] that are ignored.), +* which never get incremented during this init phase. +* (Note: The event is also deprecated.) +*/ + val = 20; + if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) goto pc_false; + /* Check if the performance counters can be written to */ + val = 0xabcd; + for (retry = 5; retry; retry--) { + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || + val2) + break; + + /* Wait about 20 msec for power gating to disable and retry. */ + msleep(20); + } + /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + goto pc_false; + + if (val != val2) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 21/47] iommu/amd: Fix performance counter initialization
From: Suravee Suthikulpanit [ Upstream commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b ] Certain AMD platforms enable power gating feature for IOMMU PMC, which prevents the IOMMU driver from updating the counter while trying to validate the PMC functionality in the init_iommu_perf_ctr(). This results in disabling PMC support and the following error message: "AMD-Vi: Unable to read/write to IOMMU perf counter" To workaround this issue, disable power gating temporarily by programming the counter source to non-zero value while validating the counter, and restore the prior state afterward. Signed-off-by: Suravee Suthikulpanit Tested-by: Tj (Elloe Linux) Link: https://lore.kernel.org/r/20210208122712.5048-1-suravee.suthikulpa...@amd.com Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 45 ++-- 1 file changed, 34 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index c842545368fd..3c215f0a6052 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include #include @@ -254,6 +255,8 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1717,13 +1720,11 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - -static void init_iommu_perf_ctr(struct amd_iommu *iommu) +static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) { + int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; + u64 val = 0xabcd, val2 = 0, save_reg, save_src; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1731,17 +1732,39 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) goto pc_false; - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) + /* +* Disable power gating by programing the performance counter +* source to 20 (i.e. counts the reads and writes from/to IOMMU +* Reserved Register [MMIO Offset 1FF8h] that are ignored.), +* which never get incremented during this init phase. +* (Note: The event is also deprecated.) +*/ + val = 20; + if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) goto pc_false; + /* Check if the performance counters can be written to */ + val = 0xabcd; + for (retry = 5; retry; retry--) { + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || + val2) + break; + + /* Wait about 20 msec for power gating to disable and retry. */ + msleep(20); + } + /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + goto pc_false; + + if (val != val2) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 07/47] iommu/vt-d: Clear PRQ overflow only when PRQ is empty
From: Lu Baolu [ Upstream commit 28a77185f1cd0650b664f546141433a7a615 ] It is incorrect to always clear PRO when it's set w/o first checking whether the overflow condition has been cleared. Current code assumes that if an overflow condition occurs it must have been cleared by earlier loop. However since the code runs in a threaded context, the overflow condition could occur even after setting the head to the tail under some extreme condition. To be sane, we should read both head/tail again when seeing a pending PRO and only clear PRO after all pending PRs have been handled. Suggested-by: Kevin Tian Signed-off-by: Lu Baolu Link: https://lore.kernel.org/linux-iommu/mwhpr11mb18862d2ea5bd432bf22d99a48c...@mwhpr11mb1886.namprd11.prod.outlook.com/ Link: https://lore.kernel.org/r/20210126080730.2232859-2-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/svm.c | 13 +++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c index 43f392d27d31..b200a3acc6ed 100644 --- a/drivers/iommu/intel/svm.c +++ b/drivers/iommu/intel/svm.c @@ -1079,8 +1079,17 @@ prq_advance: * Clear the page request overflow bit and wake up all threads that * are waiting for the completion of this handling. */ - if (readl(iommu->reg + DMAR_PRS_REG) & DMA_PRS_PRO) - writel(DMA_PRS_PRO, iommu->reg + DMAR_PRS_REG); + if (readl(iommu->reg + DMAR_PRS_REG) & DMA_PRS_PRO) { + pr_info_ratelimited("IOMMU: %s: PRQ overflow detected\n", + iommu->name); + head = dmar_readq(iommu->reg + DMAR_PQH_REG) & PRQ_RING_MASK; + tail = dmar_readq(iommu->reg + DMAR_PQT_REG) & PRQ_RING_MASK; + if (head == tail) { + writel(DMA_PRS_PRO, iommu->reg + DMAR_PRS_REG); + pr_info_ratelimited("IOMMU: %s: PRQ overflow cleared", + iommu->name); + } + } if (!completion_done(&iommu->prq_complete)) complete(&iommu->prq_complete); -- 2.30.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.11 25/52] iommu/amd: Fix performance counter initialization
From: Suravee Suthikulpanit [ Upstream commit 6778ff5b21bd8e78c8bd547fd66437cf2657fd9b ] Certain AMD platforms enable power gating feature for IOMMU PMC, which prevents the IOMMU driver from updating the counter while trying to validate the PMC functionality in the init_iommu_perf_ctr(). This results in disabling PMC support and the following error message: "AMD-Vi: Unable to read/write to IOMMU perf counter" To workaround this issue, disable power gating temporarily by programming the counter source to non-zero value while validating the counter, and restore the prior state afterward. Signed-off-by: Suravee Suthikulpanit Tested-by: Tj (Elloe Linux) Link: https://lore.kernel.org/r/20210208122712.5048-1-suravee.suthikulpa...@amd.com Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753 Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/init.c | 45 ++-- 1 file changed, 34 insertions(+), 11 deletions(-) diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c index 83d8ab2aed9f..01da76dc1caa 100644 --- a/drivers/iommu/amd/init.c +++ b/drivers/iommu/amd/init.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include #include @@ -254,6 +255,8 @@ static enum iommu_init_state init_state = IOMMU_START_STATE; static int amd_iommu_enable_interrupts(void); static int __init iommu_go_to_state(enum iommu_init_state state); static void init_device_table_dma(void); +static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, + u8 fxn, u64 *value, bool is_write); static bool amd_iommu_pre_enabled = true; @@ -1712,13 +1715,11 @@ static int __init init_iommu_all(struct acpi_table_header *table) return 0; } -static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr, - u8 fxn, u64 *value, bool is_write); - -static void init_iommu_perf_ctr(struct amd_iommu *iommu) +static void __init init_iommu_perf_ctr(struct amd_iommu *iommu) { + int retry; struct pci_dev *pdev = iommu->dev; - u64 val = 0xabcd, val2 = 0, save_reg = 0; + u64 val = 0xabcd, val2 = 0, save_reg, save_src; if (!iommu_feature(iommu, FEATURE_PC)) return; @@ -1726,17 +1727,39 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu) amd_iommu_pc_present = true; /* save the value to restore, if writable */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, false) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, false)) goto pc_false; - /* Check if the performance counters can be written to */ - if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) || - (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) || - (val != val2)) + /* +* Disable power gating by programing the performance counter +* source to 20 (i.e. counts the reads and writes from/to IOMMU +* Reserved Register [MMIO Offset 1FF8h] that are ignored.), +* which never get incremented during this init phase. +* (Note: The event is also deprecated.) +*/ + val = 20; + if (iommu_pc_get_set_reg(iommu, 0, 0, 8, &val, true)) goto pc_false; + /* Check if the performance counters can be written to */ + val = 0xabcd; + for (retry = 5; retry; retry--) { + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false) || + val2) + break; + + /* Wait about 20 msec for power gating to disable and retry. */ + msleep(20); + } + /* restore */ - if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true)) + if (iommu_pc_get_set_reg(iommu, 0, 0, 0, &save_reg, true) || + iommu_pc_get_set_reg(iommu, 0, 0, 8, &save_src, true)) + goto pc_false; + + if (val != val2) goto pc_false; pci_info(pdev, "IOMMU performance counters supported\n"); -- 2.30.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.11 09/52] iommu/vt-d: Clear PRQ overflow only when PRQ is empty
From: Lu Baolu [ Upstream commit 28a77185f1cd0650b664f546141433a7a615 ] It is incorrect to always clear PRO when it's set w/o first checking whether the overflow condition has been cleared. Current code assumes that if an overflow condition occurs it must have been cleared by earlier loop. However since the code runs in a threaded context, the overflow condition could occur even after setting the head to the tail under some extreme condition. To be sane, we should read both head/tail again when seeing a pending PRO and only clear PRO after all pending PRs have been handled. Suggested-by: Kevin Tian Signed-off-by: Lu Baolu Link: https://lore.kernel.org/linux-iommu/mwhpr11mb18862d2ea5bd432bf22d99a48c...@mwhpr11mb1886.namprd11.prod.outlook.com/ Link: https://lore.kernel.org/r/20210126080730.2232859-2-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel/svm.c | 13 +++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c index 18a9f05df407..b3bcd6dec93e 100644 --- a/drivers/iommu/intel/svm.c +++ b/drivers/iommu/intel/svm.c @@ -1079,8 +1079,17 @@ prq_advance: * Clear the page request overflow bit and wake up all threads that * are waiting for the completion of this handling. */ - if (readl(iommu->reg + DMAR_PRS_REG) & DMA_PRS_PRO) - writel(DMA_PRS_PRO, iommu->reg + DMAR_PRS_REG); + if (readl(iommu->reg + DMAR_PRS_REG) & DMA_PRS_PRO) { + pr_info_ratelimited("IOMMU: %s: PRQ overflow detected\n", + iommu->name); + head = dmar_readq(iommu->reg + DMAR_PQH_REG) & PRQ_RING_MASK; + tail = dmar_readq(iommu->reg + DMAR_PQT_REG) & PRQ_RING_MASK; + if (head == tail) { + writel(DMA_PRS_PRO, iommu->reg + DMAR_PRS_REG); + pr_info_ratelimited("IOMMU: %s: PRQ overflow cleared", + iommu->name); + } + } if (!completion_done(&iommu->prq_complete)) complete(&iommu->prq_complete); -- 2.30.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 47/51] iommu/vt-d: Fix lockdep splat in sva bind()/unbind()
From: Lu Baolu [ Upstream commit 420d42f6f9db27d88bc4f83e3e668fcdacbf7e29 ] Lock(&iommu->lock) without disabling irq causes lockdep warnings. WARNING: possible irq lock inversion dependency detected 5.11.0-rc1+ #828 Not tainted kworker/0:1H/120 just changed the state of lock: ad9ea1b8 (device_domain_lock){..-.}-{2:2}, at: iommu_flush_dev_iotlb.part.0+0x32/0x120 but this lock took another, SOFTIRQ-unsafe lock in the past: (&iommu->lock){+.+.}-{2:2} and interrupts could create inverse lock ordering between them. other info that might help us debug this: Possible interrupt unsafe locking scenario: CPU0CPU1 lock(&iommu->lock); local_irq_disable(); lock(device_domain_lock); lock(&iommu->lock); lock(device_domain_lock); *** DEADLOCK *** Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20201231005323.2178523-5-baolu...@linux.intel.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/intel/svm.c | 14 -- 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c index 3242ebd0bca36..267001e8076cd 100644 --- a/drivers/iommu/intel/svm.c +++ b/drivers/iommu/intel/svm.c @@ -281,6 +281,7 @@ int intel_svm_bind_gpasid(struct iommu_domain *domain, struct device *dev, struct dmar_domain *dmar_domain; struct device_domain_info *info; struct intel_svm *svm = NULL; + unsigned long iflags; int ret = 0; if (WARN_ON(!iommu) || !data) @@ -381,12 +382,12 @@ int intel_svm_bind_gpasid(struct iommu_domain *domain, struct device *dev, * each bind of a new device even with an existing PASID, we need to * call the nested mode setup function here. */ - spin_lock(&iommu->lock); + spin_lock_irqsave(&iommu->lock, iflags); ret = intel_pasid_setup_nested(iommu, dev, (pgd_t *)(uintptr_t)data->gpgd, data->hpasid, &data->vendor.vtd, dmar_domain, data->addr_width); - spin_unlock(&iommu->lock); + spin_unlock_irqrestore(&iommu->lock, iflags); if (ret) { dev_err_ratelimited(dev, "Failed to set up PASID %llu in nested mode, Err %d\n", data->hpasid, ret); @@ -486,6 +487,7 @@ intel_svm_bind_mm(struct device *dev, unsigned int flags, struct device_domain_info *info; struct intel_svm_dev *sdev; struct intel_svm *svm = NULL; + unsigned long iflags; int pasid_max; int ret; @@ -605,14 +607,14 @@ intel_svm_bind_mm(struct device *dev, unsigned int flags, } } - spin_lock(&iommu->lock); + spin_lock_irqsave(&iommu->lock, iflags); ret = intel_pasid_setup_first_level(iommu, dev, mm ? mm->pgd : init_mm.pgd, svm->pasid, FLPT_DEFAULT_DID, (mm ? 0 : PASID_FLAG_SUPERVISOR_MODE) | (cpu_feature_enabled(X86_FEATURE_LA57) ? PASID_FLAG_FL5LP : 0)); - spin_unlock(&iommu->lock); + spin_unlock_irqrestore(&iommu->lock, iflags); if (ret) { if (mm) mmu_notifier_unregister(&svm->notifier, mm); @@ -632,14 +634,14 @@ intel_svm_bind_mm(struct device *dev, unsigned int flags, * Binding a new device with existing PASID, need to setup * the PASID entry. */ - spin_lock(&iommu->lock); + spin_lock_irqsave(&iommu->lock, iflags); ret = intel_pasid_setup_first_level(iommu, dev, mm ? mm->pgd : init_mm.pgd, svm->pasid, FLPT_DEFAULT_DID, (mm ? 0 : PASID_FLAG_SUPERVISOR_MODE) | (cpu_feature_enabled(X86_FEATURE_LA57) ? PASID_FLAG_FL5LP : 0)); - spin_unlock(&iommu->lock); + spin_unlock_irqrestore(&iommu->lock, iflags); if (ret) { kfree(sdev); goto out; -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.4 26/38] iommu/tegra-smmu: Expand mutex protection range
From: Nicolin Chen [ Upstream commit d5f583bf8654c231b781096bc1a186065cda72b3 ] This is used to protect potential race condition at use_count. since probes of client drivers, calling attach_dev(), may run concurrently. Signed-off-by: Nicolin Chen Tested-by: Dmitry Osipenko Reviewed-by: Dmitry Osipenko Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20201125101013.14953-3-nicoleots...@gmail.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/tegra-smmu.c | 34 +- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c index 04cec050e42bf..ae796342717a6 100644 --- a/drivers/iommu/tegra-smmu.c +++ b/drivers/iommu/tegra-smmu.c @@ -241,26 +241,19 @@ static int tegra_smmu_alloc_asid(struct tegra_smmu *smmu, unsigned int *idp) { unsigned long id; - mutex_lock(&smmu->lock); - id = find_first_zero_bit(smmu->asids, smmu->soc->num_asids); - if (id >= smmu->soc->num_asids) { - mutex_unlock(&smmu->lock); + if (id >= smmu->soc->num_asids) return -ENOSPC; - } set_bit(id, smmu->asids); *idp = id; - mutex_unlock(&smmu->lock); return 0; } static void tegra_smmu_free_asid(struct tegra_smmu *smmu, unsigned int id) { - mutex_lock(&smmu->lock); clear_bit(id, smmu->asids); - mutex_unlock(&smmu->lock); } static bool tegra_smmu_capable(enum iommu_cap cap) @@ -395,17 +388,21 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { u32 value; - int err; + int err = 0; + + mutex_lock(&smmu->lock); if (as->use_count > 0) { as->use_count++; - return 0; + goto unlock; } as->pd_dma = dma_map_page(smmu->dev, as->pd, 0, SMMU_SIZE_PD, DMA_TO_DEVICE); - if (dma_mapping_error(smmu->dev, as->pd_dma)) - return -ENOMEM; + if (dma_mapping_error(smmu->dev, as->pd_dma)) { + err = -ENOMEM; + goto unlock; + } /* We can't handle 64-bit DMA addresses */ if (!smmu_dma_addr_valid(smmu, as->pd_dma)) { @@ -428,24 +425,35 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, as->smmu = smmu; as->use_count++; + mutex_unlock(&smmu->lock); + return 0; err_unmap: dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); +unlock: + mutex_unlock(&smmu->lock); + return err; } static void tegra_smmu_as_unprepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { - if (--as->use_count > 0) + mutex_lock(&smmu->lock); + + if (--as->use_count > 0) { + mutex_unlock(&smmu->lock); return; + } tegra_smmu_free_asid(smmu, as->id); dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); as->smmu = NULL; + + mutex_unlock(&smmu->lock); } static int tegra_smmu_attach_dev(struct iommu_domain *domain, -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.9 31/48] iommu/tegra-smmu: Expand mutex protection range
From: Nicolin Chen [ Upstream commit d5f583bf8654c231b781096bc1a186065cda72b3 ] This is used to protect potential race condition at use_count. since probes of client drivers, calling attach_dev(), may run concurrently. Signed-off-by: Nicolin Chen Tested-by: Dmitry Osipenko Reviewed-by: Dmitry Osipenko Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20201125101013.14953-3-nicoleots...@gmail.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/tegra-smmu.c | 34 +- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c index 04cec050e42bf..ae796342717a6 100644 --- a/drivers/iommu/tegra-smmu.c +++ b/drivers/iommu/tegra-smmu.c @@ -241,26 +241,19 @@ static int tegra_smmu_alloc_asid(struct tegra_smmu *smmu, unsigned int *idp) { unsigned long id; - mutex_lock(&smmu->lock); - id = find_first_zero_bit(smmu->asids, smmu->soc->num_asids); - if (id >= smmu->soc->num_asids) { - mutex_unlock(&smmu->lock); + if (id >= smmu->soc->num_asids) return -ENOSPC; - } set_bit(id, smmu->asids); *idp = id; - mutex_unlock(&smmu->lock); return 0; } static void tegra_smmu_free_asid(struct tegra_smmu *smmu, unsigned int id) { - mutex_lock(&smmu->lock); clear_bit(id, smmu->asids); - mutex_unlock(&smmu->lock); } static bool tegra_smmu_capable(enum iommu_cap cap) @@ -395,17 +388,21 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { u32 value; - int err; + int err = 0; + + mutex_lock(&smmu->lock); if (as->use_count > 0) { as->use_count++; - return 0; + goto unlock; } as->pd_dma = dma_map_page(smmu->dev, as->pd, 0, SMMU_SIZE_PD, DMA_TO_DEVICE); - if (dma_mapping_error(smmu->dev, as->pd_dma)) - return -ENOMEM; + if (dma_mapping_error(smmu->dev, as->pd_dma)) { + err = -ENOMEM; + goto unlock; + } /* We can't handle 64-bit DMA addresses */ if (!smmu_dma_addr_valid(smmu, as->pd_dma)) { @@ -428,24 +425,35 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, as->smmu = smmu; as->use_count++; + mutex_unlock(&smmu->lock); + return 0; err_unmap: dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); +unlock: + mutex_unlock(&smmu->lock); + return err; } static void tegra_smmu_as_unprepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { - if (--as->use_count > 0) + mutex_lock(&smmu->lock); + + if (--as->use_count > 0) { + mutex_unlock(&smmu->lock); return; + } tegra_smmu_free_asid(smmu, as->id); dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); as->smmu = NULL; + + mutex_unlock(&smmu->lock); } static int tegra_smmu_attach_dev(struct iommu_domain *domain, -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.14 41/66] iommu/tegra-smmu: Expand mutex protection range
From: Nicolin Chen [ Upstream commit d5f583bf8654c231b781096bc1a186065cda72b3 ] This is used to protect potential race condition at use_count. since probes of client drivers, calling attach_dev(), may run concurrently. Signed-off-by: Nicolin Chen Tested-by: Dmitry Osipenko Reviewed-by: Dmitry Osipenko Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20201125101013.14953-3-nicoleots...@gmail.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/tegra-smmu.c | 34 +- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c index 848dac3e4580f..30f0c72d685fe 100644 --- a/drivers/iommu/tegra-smmu.c +++ b/drivers/iommu/tegra-smmu.c @@ -244,26 +244,19 @@ static int tegra_smmu_alloc_asid(struct tegra_smmu *smmu, unsigned int *idp) { unsigned long id; - mutex_lock(&smmu->lock); - id = find_first_zero_bit(smmu->asids, smmu->soc->num_asids); - if (id >= smmu->soc->num_asids) { - mutex_unlock(&smmu->lock); + if (id >= smmu->soc->num_asids) return -ENOSPC; - } set_bit(id, smmu->asids); *idp = id; - mutex_unlock(&smmu->lock); return 0; } static void tegra_smmu_free_asid(struct tegra_smmu *smmu, unsigned int id) { - mutex_lock(&smmu->lock); clear_bit(id, smmu->asids); - mutex_unlock(&smmu->lock); } static bool tegra_smmu_capable(enum iommu_cap cap) @@ -398,17 +391,21 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { u32 value; - int err; + int err = 0; + + mutex_lock(&smmu->lock); if (as->use_count > 0) { as->use_count++; - return 0; + goto unlock; } as->pd_dma = dma_map_page(smmu->dev, as->pd, 0, SMMU_SIZE_PD, DMA_TO_DEVICE); - if (dma_mapping_error(smmu->dev, as->pd_dma)) - return -ENOMEM; + if (dma_mapping_error(smmu->dev, as->pd_dma)) { + err = -ENOMEM; + goto unlock; + } /* We can't handle 64-bit DMA addresses */ if (!smmu_dma_addr_valid(smmu, as->pd_dma)) { @@ -431,24 +428,35 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, as->smmu = smmu; as->use_count++; + mutex_unlock(&smmu->lock); + return 0; err_unmap: dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); +unlock: + mutex_unlock(&smmu->lock); + return err; } static void tegra_smmu_as_unprepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { - if (--as->use_count > 0) + mutex_lock(&smmu->lock); + + if (--as->use_count > 0) { + mutex_unlock(&smmu->lock); return; + } tegra_smmu_free_asid(smmu, as->id); dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); as->smmu = NULL; + + mutex_unlock(&smmu->lock); } static int tegra_smmu_attach_dev(struct iommu_domain *domain, -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.19 49/87] iommu/tegra-smmu: Expand mutex protection range
From: Nicolin Chen [ Upstream commit d5f583bf8654c231b781096bc1a186065cda72b3 ] This is used to protect potential race condition at use_count. since probes of client drivers, calling attach_dev(), may run concurrently. Signed-off-by: Nicolin Chen Tested-by: Dmitry Osipenko Reviewed-by: Dmitry Osipenko Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20201125101013.14953-3-nicoleots...@gmail.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/tegra-smmu.c | 34 +- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c index fa0ecb5e63809..63679cce95054 100644 --- a/drivers/iommu/tegra-smmu.c +++ b/drivers/iommu/tegra-smmu.c @@ -252,26 +252,19 @@ static int tegra_smmu_alloc_asid(struct tegra_smmu *smmu, unsigned int *idp) { unsigned long id; - mutex_lock(&smmu->lock); - id = find_first_zero_bit(smmu->asids, smmu->soc->num_asids); - if (id >= smmu->soc->num_asids) { - mutex_unlock(&smmu->lock); + if (id >= smmu->soc->num_asids) return -ENOSPC; - } set_bit(id, smmu->asids); *idp = id; - mutex_unlock(&smmu->lock); return 0; } static void tegra_smmu_free_asid(struct tegra_smmu *smmu, unsigned int id) { - mutex_lock(&smmu->lock); clear_bit(id, smmu->asids); - mutex_unlock(&smmu->lock); } static bool tegra_smmu_capable(enum iommu_cap cap) @@ -406,17 +399,21 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { u32 value; - int err; + int err = 0; + + mutex_lock(&smmu->lock); if (as->use_count > 0) { as->use_count++; - return 0; + goto unlock; } as->pd_dma = dma_map_page(smmu->dev, as->pd, 0, SMMU_SIZE_PD, DMA_TO_DEVICE); - if (dma_mapping_error(smmu->dev, as->pd_dma)) - return -ENOMEM; + if (dma_mapping_error(smmu->dev, as->pd_dma)) { + err = -ENOMEM; + goto unlock; + } /* We can't handle 64-bit DMA addresses */ if (!smmu_dma_addr_valid(smmu, as->pd_dma)) { @@ -439,24 +436,35 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, as->smmu = smmu; as->use_count++; + mutex_unlock(&smmu->lock); + return 0; err_unmap: dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); +unlock: + mutex_unlock(&smmu->lock); + return err; } static void tegra_smmu_as_unprepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { - if (--as->use_count > 0) + mutex_lock(&smmu->lock); + + if (--as->use_count > 0) { + mutex_unlock(&smmu->lock); return; + } tegra_smmu_free_asid(smmu, as->id); dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); as->smmu = NULL; + + mutex_unlock(&smmu->lock); } static int tegra_smmu_attach_dev(struct iommu_domain *domain, -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 128/130] x86, swiotlb: Adjust SWIOTLB bounce buffer size for SEV guests
From: Ashish Kalra [ Upstream commit e998879d4fb7991856916972168cf27c0d86ed12 ] For SEV, all DMA to and from guest has to use shared (un-encrypted) pages. SEV uses SWIOTLB to make this happen without requiring changes to device drivers. However, depending on the workload being run, the default 64MB of it might not be enough and it may run out of buffers to use for DMA, resulting in I/O errors and/or performance degradation for high I/O workloads. Adjust the default size of SWIOTLB for SEV guests using a percentage of the total memory available to guest for the SWIOTLB buffers. Adds a new sev_setup_arch() function which is invoked from setup_arch() and it calls into a new swiotlb generic code function swiotlb_adjust_size() to do the SWIOTLB buffer adjustment. v5 fixed build errors and warnings as Reported-by: kbuild test robot Signed-off-by: Ashish Kalra Co-developed-by: Borislav Petkov Signed-off-by: Borislav Petkov Signed-off-by: Konrad Rzeszutek Wilk Signed-off-by: Sasha Levin --- arch/x86/include/asm/mem_encrypt.h | 2 ++ arch/x86/kernel/setup.c| 6 ++ arch/x86/mm/mem_encrypt.c | 31 ++ include/linux/swiotlb.h| 8 kernel/dma/swiotlb.c | 20 +-- 5 files changed, 65 insertions(+), 2 deletions(-) diff --git a/arch/x86/include/asm/mem_encrypt.h b/arch/x86/include/asm/mem_encrypt.h index 848ce43b9040d..6c36956452ca6 100644 --- a/arch/x86/include/asm/mem_encrypt.h +++ b/arch/x86/include/asm/mem_encrypt.h @@ -36,6 +36,7 @@ void __init sme_map_bootdata(char *real_mode_data); void __init sme_unmap_bootdata(char *real_mode_data); void __init sme_early_init(void); +void __init sev_setup_arch(void); void __init sme_encrypt_kernel(struct boot_params *bp); void __init sme_enable(struct boot_params *bp); @@ -65,6 +66,7 @@ static inline void __init sme_map_bootdata(char *real_mode_data) { } static inline void __init sme_unmap_bootdata(char *real_mode_data) { } static inline void __init sme_early_init(void) { } +static inline void __init sev_setup_arch(void) { } static inline void __init sme_encrypt_kernel(struct boot_params *bp) { } static inline void __init sme_enable(struct boot_params *bp) { } diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 77ea96b794bd1..9dc0c94e1544f 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1120,6 +1120,12 @@ void __init setup_arch(char **cmdline_p) memblock_set_current_limit(ISA_END_ADDRESS); e820__memblock_setup(); + /* +* Needs to run after memblock setup because it needs the physical +* memory size. +*/ + sev_setup_arch(); + reserve_bios_regions(); if (efi_enabled(EFI_MEMMAP)) { diff --git a/arch/x86/mm/mem_encrypt.c b/arch/x86/mm/mem_encrypt.c index 9268c12458c84..5911cfa31bc6d 100644 --- a/arch/x86/mm/mem_encrypt.c +++ b/arch/x86/mm/mem_encrypt.c @@ -196,6 +196,37 @@ void __init sme_early_init(void) swiotlb_force = SWIOTLB_FORCE; } +void __init sev_setup_arch(void) +{ + phys_addr_t total_mem = memblock_phys_mem_size(); + unsigned long size; + + if (!sev_active()) + return; + + /* +* For SEV, all DMA has to occur via shared/unencrypted pages. +* SEV uses SWIOTLB to make this happen without changing device +* drivers. However, depending on the workload being run, the +* default 64MB of SWIOTLB may not be enough and SWIOTLB may +* run out of buffers for DMA, resulting in I/O errors and/or +* performance degradation especially with high I/O workloads. +* +* Adjust the default size of SWIOTLB for SEV guests using +* a percentage of guest memory for SWIOTLB buffers. +* Also, as the SWIOTLB bounce buffer memory is allocated +* from low memory, ensure that the adjusted size is within +* the limits of low available memory. +* +* The percentage of guest memory used here for SWIOTLB buffers +* is more of an approximation of the static adjustment which +* 64MB for <1G, and ~128M to 256M for 1G-to-4G, i.e., the 6% +*/ + size = total_mem * 6 / 100; + size = clamp_val(size, IO_TLB_DEFAULT_SIZE, SZ_1G); + swiotlb_adjust_size(size); +} + static void __init __set_clr_pte_enc(pte_t *kpte, int level, bool enc) { pgprot_t old_prot, new_prot; diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h index 0a8fced6aaec4..522a0942114ad 100644 --- a/include/linux/swiotlb.h +++ b/include/linux/swiotlb.h @@ -30,6 +30,9 @@ enum swiotlb_force { */ #define IO_TLB_SHIFT 11 +/* default to 64MB */ +#define IO_TLB_DEFAULT_SIZE (64UL<<20) + extern void swiotlb_init(int verbose); int swiotlb_init_with_tbl(char *tlb, unsigned long nslabs, int verbose); extern unsigned long swiotlb_nr_tbl(void); @@ -80,6 +83,7 @@ void __init swi
[PATCH AUTOSEL 5.4 070/130] iommu/tegra-smmu: Expand mutex protection range
From: Nicolin Chen [ Upstream commit d5f583bf8654c231b781096bc1a186065cda72b3 ] This is used to protect potential race condition at use_count. since probes of client drivers, calling attach_dev(), may run concurrently. Signed-off-by: Nicolin Chen Tested-by: Dmitry Osipenko Reviewed-by: Dmitry Osipenko Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20201125101013.14953-3-nicoleots...@gmail.com Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/tegra-smmu.c | 34 +- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c index dd486233e2828..41be3e2202971 100644 --- a/drivers/iommu/tegra-smmu.c +++ b/drivers/iommu/tegra-smmu.c @@ -247,26 +247,19 @@ static int tegra_smmu_alloc_asid(struct tegra_smmu *smmu, unsigned int *idp) { unsigned long id; - mutex_lock(&smmu->lock); - id = find_first_zero_bit(smmu->asids, smmu->soc->num_asids); - if (id >= smmu->soc->num_asids) { - mutex_unlock(&smmu->lock); + if (id >= smmu->soc->num_asids) return -ENOSPC; - } set_bit(id, smmu->asids); *idp = id; - mutex_unlock(&smmu->lock); return 0; } static void tegra_smmu_free_asid(struct tegra_smmu *smmu, unsigned int id) { - mutex_lock(&smmu->lock); clear_bit(id, smmu->asids); - mutex_unlock(&smmu->lock); } static bool tegra_smmu_capable(enum iommu_cap cap) @@ -404,17 +397,21 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { u32 value; - int err; + int err = 0; + + mutex_lock(&smmu->lock); if (as->use_count > 0) { as->use_count++; - return 0; + goto unlock; } as->pd_dma = dma_map_page(smmu->dev, as->pd, 0, SMMU_SIZE_PD, DMA_TO_DEVICE); - if (dma_mapping_error(smmu->dev, as->pd_dma)) - return -ENOMEM; + if (dma_mapping_error(smmu->dev, as->pd_dma)) { + err = -ENOMEM; + goto unlock; + } /* We can't handle 64-bit DMA addresses */ if (!smmu_dma_addr_valid(smmu, as->pd_dma)) { @@ -437,24 +434,35 @@ static int tegra_smmu_as_prepare(struct tegra_smmu *smmu, as->smmu = smmu; as->use_count++; + mutex_unlock(&smmu->lock); + return 0; err_unmap: dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); +unlock: + mutex_unlock(&smmu->lock); + return err; } static void tegra_smmu_as_unprepare(struct tegra_smmu *smmu, struct tegra_smmu_as *as) { - if (--as->use_count > 0) + mutex_lock(&smmu->lock); + + if (--as->use_count > 0) { + mutex_unlock(&smmu->lock); return; + } tegra_smmu_free_asid(smmu, as->id); dma_unmap_page(smmu->dev, as->pd_dma, SMMU_SIZE_PD, DMA_TO_DEVICE); as->smmu = NULL; + + mutex_unlock(&smmu->lock); } static int tegra_smmu_attach_dev(struct iommu_domain *domain, -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.10 065/217] pinctrl: qcom: Kconfig: Rework PINCTRL_MSM to be a depenency rather then a selected config
From: John Stultz [ Upstream commit be117ca32261c3331b614f440c737650791a6998 ] This patch reworks PINCTRL_MSM to be a visible option, and instead of having the various SoC specific drivers select PINCTRL_MSM, this switches those configs to depend on PINCTRL_MSM. This is useful, as it will be needed in order to cleanly support having the qcom-scm driver, which pinctrl-msm calls into, configured as a module. Without this change, we would eventually have to add dependency lines to every config that selects PINCTRL_MSM, and that would becomes a maintenance headache. We also add PINCTRL_MSM to the arm64 defconfig to avoid surprises as otherwise PINCTRL_MSM/IPQ* options previously enabled, will be off. Signed-off-by: John Stultz Reviewed-by: Bjorn Andersson Cc: Catalin Marinas Cc: Will Deacon Cc: Andy Gross Cc: Bjorn Andersson Cc: Joerg Roedel Cc: Thomas Gleixner Cc: Jason Cooper Cc: Marc Zyngier Cc: Linus Walleij Cc: Vinod Koul Cc: Kalle Valo Cc: Maulik Shah Cc: Lina Iyer Cc: Saravana Kannan Cc: Todd Kjos Cc: Greg Kroah-Hartman Cc: linux-arm-...@vger.kernel.org Cc: iommu@lists.linux-foundation.org Cc: linux-g...@vger.kernel.org Link: https://lore.kernel.org/r/20201106042710.55979-1-john.stu...@linaro.org Signed-off-by: Linus Walleij Signed-off-by: Sasha Levin --- arch/arm64/configs/defconfig | 1 + drivers/pinctrl/qcom/Kconfig | 48 ++-- 2 files changed, 25 insertions(+), 24 deletions(-) diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig index 5cfe3cf6f2acb..2ac53d8ae76a3 100644 --- a/arch/arm64/configs/defconfig +++ b/arch/arm64/configs/defconfig @@ -483,6 +483,7 @@ CONFIG_PINCTRL_IMX8MP=y CONFIG_PINCTRL_IMX8MQ=y CONFIG_PINCTRL_IMX8QXP=y CONFIG_PINCTRL_IMX8DXL=y +CONFIG_PINCTRL_MSM=y CONFIG_PINCTRL_IPQ8074=y CONFIG_PINCTRL_IPQ6018=y CONFIG_PINCTRL_MSM8916=y diff --git a/drivers/pinctrl/qcom/Kconfig b/drivers/pinctrl/qcom/Kconfig index 5fe7b8aaf69d8..c9bb2d9e49d47 100644 --- a/drivers/pinctrl/qcom/Kconfig +++ b/drivers/pinctrl/qcom/Kconfig @@ -2,7 +2,7 @@ if (ARCH_QCOM || COMPILE_TEST) config PINCTRL_MSM - bool + bool "Qualcomm core pin controller driver" select PINMUX select PINCONF select GENERIC_PINCONF @@ -13,7 +13,7 @@ config PINCTRL_MSM config PINCTRL_APQ8064 tristate "Qualcomm APQ8064 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm TLMM block found in the Qualcomm APQ8064 platform. @@ -21,7 +21,7 @@ config PINCTRL_APQ8064 config PINCTRL_APQ8084 tristate "Qualcomm APQ8084 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm TLMM block found in the Qualcomm APQ8084 platform. @@ -29,7 +29,7 @@ config PINCTRL_APQ8084 config PINCTRL_IPQ4019 tristate "Qualcomm IPQ4019 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm TLMM block found in the Qualcomm IPQ4019 platform. @@ -37,7 +37,7 @@ config PINCTRL_IPQ4019 config PINCTRL_IPQ8064 tristate "Qualcomm IPQ8064 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm TLMM block found in the Qualcomm IPQ8064 platform. @@ -45,7 +45,7 @@ config PINCTRL_IPQ8064 config PINCTRL_IPQ8074 tristate "Qualcomm Technologies, Inc. IPQ8074 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm Technologies Inc. TLMM block found on the @@ -55,7 +55,7 @@ config PINCTRL_IPQ8074 config PINCTRL_IPQ6018 tristate "Qualcomm Technologies, Inc. IPQ6018 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm Technologies Inc. TLMM block found on the @@ -65,7 +65,7 @@ config PINCTRL_IPQ6018 config PINCTRL_MSM8226 tristate "Qualcomm 8226 pin controller driver" depends on GPIOLIB && OF - select PINCTRL_MSM + depends on PINCTRL_MSM help This is the pinctrl, pinmux, pinconf and gpiolib driver for the Qualcomm Technologies Inc TLMM block foun
[PATCH AUTOSEL 5.9 22/39] arm-smmu-qcom: Ensure the qcom_scm driver has finished probing
From: John Stultz [ Upstream commit 72b55c96f3a5ae6e486c20b5dacf5114060ed042 ] Robin Murphy pointed out that if the arm-smmu driver probes before the qcom_scm driver, we may call qcom_scm_qsmmu500_wait_safe_toggle() before the __scm is initialized. Now, getting this to happen is a bit contrived, as in my efforts it required enabling asynchronous probing for both drivers, moving the firmware dts node to the end of the dtsi file, as well as forcing a long delay in the qcom_scm_probe function. With those tweaks we ran into the following crash: [2.631040] arm-smmu 1500.iommu: Stage-1: 48-bit VA -> 48-bit IPA [2.633372] Unable to handle kernel NULL pointer dereference at virtual address ... [2.633402] [] user address but active_mm is swapper [2.633409] Internal error: Oops: 9605 [#1] PREEMPT SMP [2.633415] Modules linked in: [2.633427] CPU: 5 PID: 117 Comm: kworker/u16:2 Tainted: GW 5.10.0-rc1-mainline-00025-g272a618fc36-dirty #3971 [2.633430] Hardware name: Thundercomm Dragonboard 845c (DT) [2.633448] Workqueue: events_unbound async_run_entry_fn [2.633456] pstate: 80c5 (Nzcv daif +PAN +UAO -TCO BTYPE=--) [2.633465] pc : qcom_scm_qsmmu500_wait_safe_toggle+0x78/0xb0 [2.633473] lr : qcom_smmu500_reset+0x58/0x78 [2.633476] sp : ffc0105a3b60 ... [2.633567] Call trace: [2.633572] qcom_scm_qsmmu500_wait_safe_toggle+0x78/0xb0 [2.633576] qcom_smmu500_reset+0x58/0x78 [2.633581] arm_smmu_device_reset+0x194/0x270 [2.633585] arm_smmu_device_probe+0xc94/0xeb8 [2.633592] platform_drv_probe+0x58/0xa8 [2.633597] really_probe+0xec/0x398 [2.633601] driver_probe_device+0x5c/0xb8 [2.633606] __driver_attach_async_helper+0x64/0x88 [2.633610] async_run_entry_fn+0x4c/0x118 [2.633617] process_one_work+0x20c/0x4b0 [2.633621] worker_thread+0x48/0x460 [2.633628] kthread+0x14c/0x158 [2.633634] ret_from_fork+0x10/0x18 [2.633642] Code: a9034fa0 d0007f73 29107fa0 91342273 (f9400020) To avoid this, this patch adds a check on qcom_scm_is_available() in the qcom_smmu_impl_init() function, returning -EPROBE_DEFER if its not ready. This allows the driver to try to probe again later after qcom_scm has finished probing. Reported-by: Robin Murphy Signed-off-by: John Stultz Reviewed-by: Robin Murphy Cc: Robin Murphy Cc: Will Deacon Cc: Andy Gross Cc: Maulik Shah Cc: Bjorn Andersson Cc: Saravana Kannan Cc: Marc Zyngier Cc: Lina Iyer Cc: iommu@lists.linux-foundation.org Cc: linux-arm-msm Link: https://lore.kernel.org/r/20201112220520.48159-1-john.stu...@linaro.org Signed-off-by: Will Deacon Signed-off-by: Sasha Levin --- drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 4 1 file changed, 4 insertions(+) diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c index be4318044f96c..702fbaa6c9ada 100644 --- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c @@ -69,6 +69,10 @@ struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu) { struct qcom_smmu *qsmmu; + /* Check to make sure qcom_scm has finished probing */ + if (!qcom_scm_is_available()) + return ERR_PTR(-EPROBE_DEFER); + qsmmu = devm_kzalloc(smmu->dev, sizeof(*qsmmu), GFP_KERNEL); if (!qsmmu) return ERR_PTR(-ENOMEM); -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.4 08/10] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h index 695d4e235438c..90832bf00538e 100644 --- a/drivers/iommu/amd_iommu_types.h +++ b/drivers/iommu/amd_iommu_types.h @@ -351,7 +351,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.9 10/12] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h index da3fbf82d1cf4..e19c05d9e84ba 100644 --- a/drivers/iommu/amd_iommu_types.h +++ b/drivers/iommu/amd_iommu_types.h @@ -383,7 +383,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.14 11/14] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h index 74c8638aac2b9..ac3cac052af9d 100644 --- a/drivers/iommu/amd_iommu_types.h +++ b/drivers/iommu/amd_iommu_types.h @@ -404,7 +404,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 4.19 14/21] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h index 859b06424e5c4..df6f3cc958e5e 100644 --- a/drivers/iommu/amd_iommu_types.h +++ b/drivers/iommu/amd_iommu_types.h @@ -410,7 +410,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 29/42] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h index 0679896b9e2e1..3ec090adcdae7 100644 --- a/drivers/iommu/amd_iommu_types.h +++ b/drivers/iommu/amd_iommu_types.h @@ -406,7 +406,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.9 37/55] iommu/amd: Increase interrupt remapping table limit to 512 entries
From: Suravee Suthikulpanit [ Upstream commit 73db2fc595f358460ce32bcaa3be1f0cce4a2db1 ] Certain device drivers allocate IO queues on a per-cpu basis. On AMD EPYC platform, which can support up-to 256 cpu threads, this can exceed the current MAX_IRQ_PER_TABLE limit of 256, and result in the error message: AMD-Vi: Failed to allocate IRTE This has been observed with certain NVME devices. AMD IOMMU hardware can actually support upto 512 interrupt remapping table entries. Therefore, update the driver to match the hardware limit. Please note that this also increases the size of interrupt remapping table to 8KB per device when using the 128-bit IRTE format. Signed-off-by: Suravee Suthikulpanit Link: https://lore.kernel.org/r/20201015025002.87997-1-suravee.suthikulpa...@amd.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/amd/amd_iommu_types.h | 6 +- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/iommu/amd/amd_iommu_types.h b/drivers/iommu/amd/amd_iommu_types.h index 30a5d412255a4..427484c455891 100644 --- a/drivers/iommu/amd/amd_iommu_types.h +++ b/drivers/iommu/amd/amd_iommu_types.h @@ -406,7 +406,11 @@ extern bool amd_iommu_np_cache; /* Only true if all IOMMUs support device IOTLBs */ extern bool amd_iommu_iotlb_sup; -#define MAX_IRQS_PER_TABLE 256 +/* + * AMD IOMMU hardware only support 512 IRTEs despite + * the architectural limitation of 2048 entries. + */ +#define MAX_IRQS_PER_TABLE 512 #define IRQ_TABLE_ALIGNMENT128 struct irq_remap_table { -- 2.27.0 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH AUTOSEL 5.4 4/4] iommu/vt-d: Fix lockdep splat in iommu_flush_dev_iotlb()
From: Lu Baolu [ Upstream commit 1a3f2fd7fc4e8f24510830e265de2ffb8e3300d2 ] Lock(&iommu->lock) without disabling irq causes lockdep warnings. [ 12.703950] [ 12.703962] WARNING: possible irq lock inversion dependency detected [ 12.703975] 5.9.0-rc6+ #659 Not tainted [ 12.703983] [ 12.703995] systemd-udevd/284 just changed the state of lock: [ 12.704007] bd6ff4d8 (device_domain_lock){..-.}-{2:2}, at: iommu_flush_dev_iotlb.part.57+0x2e/0x90 [ 12.704031] but this lock took another, SOFTIRQ-unsafe lock in the past: [ 12.704043] (&iommu->lock){+.+.}-{2:2} [ 12.704045] and interrupts could create inverse lock ordering between them. [ 12.704073] other info that might help us debug this: [ 12.704085] Possible interrupt unsafe locking scenario: [ 12.704097]CPU0CPU1 [ 12.704106] [ 12.704115] lock(&iommu->lock); [ 12.704123]local_irq_disable(); [ 12.704134]lock(device_domain_lock); [ 12.704146]lock(&iommu->lock); [ 12.704158] [ 12.704164] lock(device_domain_lock); [ 12.704174] *** DEADLOCK *** Signed-off-by: Lu Baolu Link: https://lore.kernel.org/r/20200927062428.13713-1-baolu...@linux.intel.com Signed-off-by: Joerg Roedel Signed-off-by: Sasha Levin --- drivers/iommu/intel-iommu.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c index 2ffec65df3889..1147626f0d253 100644 --- a/drivers/iommu/intel-iommu.c +++ b/drivers/iommu/intel-iommu.c @@ -2560,14 +2560,14 @@ static struct dmar_domain *dmar_insert_one_dev_info(struct intel_iommu *iommu, } /* Setup the PASID entry for requests without PASID: */ - spin_lock(&iommu->lock); + spin_lock_irqsave(&iommu->lock, flags); if (hw_pass_through && domain_type_is_si(domain)) ret = intel_pasid_setup_pass_through(iommu, domain, dev, PASID_RID2PASID); else ret = intel_pasid_setup_second_level(iommu, domain, dev, PASID_RID2PASID); - spin_unlock(&iommu->lock); + spin_unlock_irqrestore(&iommu->lock, flags); if (ret) { dev_err(dev, "Setup RID2PASID failed\n"); dmar_remove_one_dev_info(dev); -- 2.25.1 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu