Skip to content

iommu/arm-smmu-v3: Tegra264 invalidation workaround

Ashish Mhetre linux-arm-kernel linux-doc linux-iommu linux-tegra lkml 9 versions – Applied

Status from this message: Applied to iommu (arm/smmu/updates), thanks!

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v9 Latest4 patches 6 repliesApplied
  2. v83 patches 4 repliesSuperseded
  3. v73 patches 7 repliesSuperseded
Earlier versions · 6
VersionPostedPatchesRepliesStatus
v63 patches2 repliesSuperseded
v53 patches6 repliesSuperseded
v43 patches9 repliesSuperseded
v33 patches20 repliesSuperseded
v22 patches4 repliesSuperseded
v12 patches7 repliesSuperseded

Changes in v9 from the cover letter

- Split the iommufd changes into a separate patch, leaving the final patch to enable the workaround and document the erratum.
- Make arm_smmu_erratum_cmd_needs_repeating() file-local and remove its unused declaration from arm-smmu-v3.h.
- Restyle the SMMUv3 hw_info flag kdoc following enum iommufd_option.

Patches in v9 Applied

#Subject and filesSizeRepliesApplied
1/4iommu/arm-smmu-v3: Factor out CMDQ batch force-sync conditionsdrivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c+17 -60 replies
2/4iommu/arm-smmu-v3: Add CFGI/TLBI-repeat workarounddrivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h+57 -40 replies
3/4iommu/arm-smmu-v3-iommufd: Report CFGI/TLBI-repeat erratumdrivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +1 more+24 -42 replies
4/4iommu/arm-smmu-v3: Enable CFGI/TLBI-repeat workaround on Tegra264Documentation/arch/arm64/silicon-errata.rst drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c+5 -12 replies

Reviews trailers per patch

All 4 patches have a Reviewed-by or Acked-by.

PatchNicolin ChenJason Gunthorpe
1/4R-b
2/4R-b R-b
3/4R-b R-b
4/4R-b R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v9 thread

Cover letter [PATCH v9 0/4] iommu/arm-smmu-v3: Tegra264 invalidation workaround

Nvidia Tegra264 SMMUs are affected by an erratum where a TLB entry can survive an invalidation that races with concurrent traffic targeting the same entry. The hardware-recommended software workaround is to issue every CFGI/TLBI command (each followed by CMD_SYNC) twice. The second issue must execute only after the first issue's CMD_SYNC has completed, giving the sequence:

TLBI/CFGI ... CMD_SYNC TLBI/CFGI ... CMD_SYNC

ATC_INV is not affected and must not be doubled.

The erratum is not flagged by any SMMUv3 IDR/IIDR register, so it cannot be detected from hardware ID. Tegra264 is device-tree-only (no ACPI/IORT support), so detection is purely by compatible string.

This series is structured so each step is reviewable in isolation:

1/4 Refactor the existing CMDQ batch force-sync conditions into
arm_smmu_cmdq_batch_force_sync(). No functional change. Authored by Nicolin Chen.

2/4 Add the static key, command classifier, CMDQ submission wrapper,
and batch force-sync condition without enabling the workaround.

3/4 Report the erratum to user space through
IOMMU_HW_INFO_ARM_SMMUV3_ERRATA_REPEAT_TLBI_CFGI and submit user-provided invalidations through the raw CMDQ helper so the host does not apply the workaround a second time.

4/4 Enable the workaround for "nvidia,tegra264-smmu" and document

     T264-SMMU-3 in silicon-errata.rst.

The series applies cleanly on linux-next/master (base-commit below).

Changes since v8:
- Split the iommufd changes into a separate patch, leaving the final patch to enable the workaround and document the erratum.
- Make arm_smmu_erratum_cmd_needs_repeating() file-local and remove its unused declaration from arm-smmu-v3.h.
- Restyle the SMMUv3 hw_info flag kdoc following enum iommufd_option.

Changes since v7:
- Rework the nesting/iommufd handling to report the erratum to user space and stop repeating user-issued invalidations in the host.
- Drop arm_vsmmu_can_batch_cmd() and its iommufd batching split. Expose __arm_smmu_cmdq_issue_cmdlist() and make arm_vsmmu_cache_invalidate() call it directly.
- Add the UAPI flag and report it from arm_smmu_hw_info() through a new arm_smmu_erratum_repeat_tlbi_cfgi() accessor.
- Carry Jason Gunthorpe's Reviewed-by on the first two patches and Nicolin Chen's Reviewed-by on the workaround infrastructure patch.

Changes since v6:
- Add #include <linux/jump_label.h> now that the static key is defined in arm-smmu-v3.c.
- Drop the unused smmu parameter from arm_vsmmu_can_batch_cmd().
- Expand the arm_smmu_cmdq_batch_force_sync() comment to note that batches never mix CFGI/TLBI with other commands, so checking cmds[0] alone is enough.
- Note that a guest kernel enabling CMDQV on Tegra264 must also apply this workaround, since guest-level VCMDQs issue commands directly to the hardware.

Changes since v5:
- Move arm_smmu_erratum_cmd_needs_repeating() into arm-smmu-v3.c and make arm_smmu_erratum_repeat_tlbi_cfgi_key file-local static.
- Add an inline erratum/workaround description at the static key, referenced from arm_smmu_cmdq_batch_force_sync().
- Fix the misleading !n comment above arm_smmu_cmdq_issue_cmdlist(); keep the defensive !n guard.
- Remove the unused smmu parameter from the predicate.

Changes since v4:
- Drop ARM_SMMU_OPT_REPEAT_TLBI_CFGI: the option bit was set and read on the same "nvidia,tegra264-smmu" compatible as the static key.
- Reorder the series so compatible-string detection lands last, after all workaround infrastructure exists.
- Update the force-sync comment to reference the static key description.

Changes since v3:
- Drop the cmds->num == 0 early-return so the refactor is truly "no functional change".
- Rename ARM_SMMU_OPT_TLBI_TWICE to ARM_SMMU_OPT_REPEAT_TLBI_CFGI.
- Rename arm_smmu_cmd_needs_tlbi_twice() to arm_smmu_erratum_cmd_needs_repeating().
- Replace the explicit opcode switch with a range check.
- Introduce arm_smmu_erratum_repeat_tlbi_cfgi_key.
- Document the erratum in silicon-errata.rst.
- Guard the repeat path against an empty cmdlist.

Changes since v2:
- Split the changes into a reviewable and bisectable series.
- Share the command classifier with the iommufd path.
- Split iommufd batches at "needs repeating" transitions.
- Document why detection uses DT instead of IIDR/ACPI.

Changes since v1:
- Detect the erratum from the existing "nvidia,tegra264-smmu" compatible instead of adding a new property.
- Centralise doubling at the CMDQ submission layer and apply it only to CFGI/TLBI, not ATC_INV.
- Drop the binding/dtsi patches.

Ashish Mhetre (3):
iommu/arm-smmu-v3: Add CFGI/TLBI-repeat workaround iommu/arm-smmu-v3-iommufd: Report CFGI/TLBI-repeat erratum iommu/arm-smmu-v3: Enable CFGI/TLBI-repeat workaround on Tegra264

Nicolin Chen (1):
iommu/arm-smmu-v3: Factor out CMDQ batch force-sync conditions

 Documentation/arch/arm64/silicon-errata.rst   |  2 +
 .../arm/arm-smmu-v3/arm-smmu-v3-iommufd.c     |  7 +-
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   | 89 ++++++++++++++++---
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h   |  5 ++
 include/uapi/linux/iommufd.h                  | 13 ++-
 5 files changed, 102 insertions(+), 14 deletions(-)

base-commit: 9eebf259d5352b87080d67758f483583d9e763d7

Signature · 1 line
2.50.1