]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/commitdiff
iommu/arm-smmu-v3: Fix size calculation in arm_smmu_mm_invalidate_range()
authorNicolin Chen <nicolinc@nvidia.com>
Tue, 19 Apr 2022 21:01:58 +0000 (14:01 -0700)
committerStefan Bader <stefan.bader@canonical.com>
Wed, 22 Jun 2022 12:23:09 +0000 (14:23 +0200)
BugLink: https://bugs.launchpad.net/bugs/1978240
commit 95d4782c34a60800ccf91d9f0703137d4367a2fc upstream.

The arm_smmu_mm_invalidate_range function is designed to be called
by mm core for Shared Virtual Addressing purpose between IOMMU and
CPU MMU. However, the ways of two subsystems defining their "end"
addresses are slightly different. IOMMU defines its "end" address
using the last address of an address range, while mm core defines
that using the following address of an address range:

include/linux/mm_types.h:
unsigned long vm_end;
/* The first byte after our end address ...

This mismatch resulted in an incorrect calculation for size so it
failed to be page-size aligned. Further, it caused a dead loop at
"while (iova < end)" check in __arm_smmu_tlb_inv_range function.

This patch fixes the issue by doing the calculation correctly.

Fixes: 2f7e8c553e98 ("iommu/arm-smmu-v3: Hook up ATC invalidation to mm ops")
Cc: stable@vger.kernel.org
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Link: https://lore.kernel.org/r/20220419210158.21320-1-nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c

index ee66d1f4cb81e67d5befe77fecffe7b2f0509806..f763c1430d152135b040be1dafa36767caad51e6 100644 (file)
@@ -183,7 +183,14 @@ static void arm_smmu_mm_invalidate_range(struct mmu_notifier *mn,
 {
        struct arm_smmu_mmu_notifier *smmu_mn = mn_to_smmu(mn);
        struct arm_smmu_domain *smmu_domain = smmu_mn->domain;
-       size_t size = end - start + 1;
+       size_t size;
+
+       /*
+        * The mm_types defines vm_end as the first byte after the end address,
+        * different from IOMMU subsystem using the last address of an address
+        * range. So do a simple translation here by calculating size correctly.
+        */
+       size = end - start;
 
        if (!(smmu_domain->smmu->features & ARM_SMMU_FEAT_BTM))
                arm_smmu_tlb_inv_range_asid(start, size, smmu_mn->cd->asid,