Merge tag 'iommu-updates-v6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

Pull iommu updates from Joerg Roedel:

 - Consolidate iommu_map/unmap functions.

   There have been blocking and atomic variants so far, but that was
   problematic as this approach does not scale with required new
   variants which just differ in the GFP flags used. So Jason
   consolidated this back into single functions that take a GFP
   parameter.

 - Retire the detach_dev() call-back in iommu_ops

 - Arm SMMU updates from Will:
     - Device-tree binding updates:
         - Cater for three power domains on SM6375
         - Document existing compatible strings for Qualcomm SoCs
         - Tighten up clocks description for platform-specific
           compatible strings
     - Enable Qualcomm workarounds for some additional platforms that
       need them

 - Intel VT-d updates from Lu Baolu:
     - Add Intel IOMMU performance monitoring support
     - Set No Execute Enable bit in PASID table entry
     - Two performance optimizations
     - Fix PASID directory pointer coherency
     - Fix missed rollbacks in error path
     - Cleanups

 - Apple t8110 DART support

 - Exynos IOMMU:
     - Implement better fault handling
     - Error handling fixes

 - Renesas IPMMU:
     - Add device tree bindings for r8a779g0

 - AMD IOMMU:
     - Various fixes for handling on SNP-enabled systems and
       handling of faults with unknown request-ids
     - Cleanups and other small fixes

 - Various other smaller fixes and cleanups

* tag 'iommu-updates-v6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (71 commits)
  iommu/amd: Skip attach device domain is same as new domain
  iommu: Attach device group to old domain in error path
  iommu/vt-d: Allow to use flush-queue when first level is default
  iommu/vt-d: Fix PASID directory pointer coherency
  iommu/vt-d: Avoid superfluous IOTLB tracking in lazy mode
  iommu/vt-d: Fix error handling in sva enable/disable paths
  iommu/amd: Improve page fault error reporting
  iommu/amd: Do not identity map v2 capable device when snp is enabled
  iommu: Fix error unwind in iommu_group_alloc()
  iommu/of: mark an unused function as __maybe_unused
  iommu: dart: DART_T8110_ERROR range should be 0 to 5
  iommu/vt-d: Enable IOMMU perfmon support
  iommu/vt-d: Add IOMMU perfmon overflow handler support
  iommu/vt-d: Support cpumask for IOMMU perfmon
  iommu/vt-d: Add IOMMU perfmon support
  iommu/vt-d: Support Enhanced Command Interface
  iommu/vt-d: Retrieve IOMMU perfmon capability information
  iommu/vt-d: Support size of the register set in DRHD
  iommu/vt-d: Set No Execute Enable bit in PASID table entry
  iommu/vt-d: Remove sva from intel_svm_dev
  ...

1 files changed, 4 insertions, 2 deletions

diff --git a/drivers/net/ipa/ipa_mem.c b/drivers/net/ipa/ipa_mem.c
index 85096d1efe5b..db6ada2343af 100644
--- a/drivers/net/ipa/ipa_mem.c
+++ b/drivers/net/ipa/ipa_mem.c
@@ -472,7 +472,8 @@ static int ipa_imem_init(struct ipa *ipa, unsigned long addr, size_t size)
 	size = PAGE_ALIGN(size + addr - phys);
 	iova = phys;	/* We just want a direct mapping */
 
-	ret = iommu_map(domain, iova, phys, size, IOMMU_READ | IOMMU_WRITE);
+	ret = iommu_map(domain, iova, phys, size, IOMMU_READ | IOMMU_WRITE,
+			GFP_KERNEL);
 	if (ret)
 		return ret;
 
@@ -580,7 +581,8 @@ static int ipa_smem_init(struct ipa *ipa, u32 item, size_t size)
 	size = PAGE_ALIGN(size + addr - phys);
 	iova = phys;	/* We just want a direct mapping */
 
-	ret = iommu_map(domain, iova, phys, size, IOMMU_READ | IOMMU_WRITE);
+	ret = iommu_map(domain, iova, phys, size, IOMMU_READ | IOMMU_WRITE,
+			GFP_KERNEL);
 	if (ret)
 		return ret;