CC to Graeme, Peter Pan and lsk-interest list.
On 04/19/2016 06:59 PM, Robin Murphy wrote:
Hi Alex,
On 19/04/16 10:15, Alex Shi wrote:
Hi All,
The backport candidate branch is pushed on git://git.linaro.org/kernel/linux-linaro-stable.git v4.1/topic/IOMMU-DMA
Is there any idea on testing?
If you want to test the DMA ops, you'll probably need to backport one of the users - the Exynos or Mediatek IOMMUs - and get hold of a suitable platform. The SMMU drivers currently only support VFIO.
Thanks a lot Robin!
Seems the Documentation/vfio.txt has a example to show how to write a testcase. Another usage maybe running a qemu with vfio-pci parameters on a ARM64 board?
The bad things is that I have no IOMMU supported board in hands. Is there someone like play with my branch? I merged the last lsk into the branch for irq/pci backport.
Thanks Alex
Robin.
Regards Alex
Picked up commits as following: bottom up order.
33c2f1d iommu/dma: Use correct offset in map_sg ed63aac iommu/dma: Avoid unlikely high-order allocations ff7bd94 iommu/dma: Add some missing #includes f37083c iommu: Move default domain allocation to iommu_group_get_for_dev() 313e923 arm64: Hook up IOMMU dma_ops 6eccff2 arm64 : Introduce support for ACPI _CCA object 4b80e6b arm64: Add IOMMU dma_ops d638ff0 iommu: Implement common IOMMU ops for DMA mapping 4193c6c iommu: iova: Move iova cache management to the iova library 73695fd iommu/arm-smmu: Remove redundant calculation of gr0 base address 922f859 iommu/arm-smmu: Fix error checking for ASID and VMID allocation 336a8e3 iommu/arm-smmu: ThunderX mis-extends 64bit registers f43316d iommu/arm-smmu: Remove unneeded '0x' annotation e5645ff iommu/arm-smmu: Use drvdata instead of maintaining smmu_devices list ab18815 iommu/arm-smmu: Use correct address mask for CMD_TLBI_S2_IPA 4528957 iommu/arm-smmu: Ensure IAS is set correctly for AArch32-capable SMMUs 024a081 iommu/io-pgtable-arm: Don't use dma_to_phys() e60ca7d of: iommu: Silence misleading warning 909227b iommu/io-pgtable-arm: Unmap and free table when overwriting with block 292d37e iommu/io-pgtable-arm: Move init-fn declarations to io-pgtable.h eabfcc5 iommu/io-pgtable: Remove flush_pgtable callback 7042bf7 iommu/arm-smmu: Remove arm_smmu_flush_pgtable() e93fcae iommu/arm-smmu: Remove arm_smmu_flush_pgtable() b43e145 iommu/io-pgtable-arm: Centralise sync points 6a22413 iommu/ipmmu-vmsa: Clean up DMA API usage 5d56615 iommu/arm-smmu: Clean up DMA API usage 906e8647 iommu/arm-smmu: Clean up DMA API usage 81d7c3c iommu/io-pgtable-arm: Allow appropriate DMA API use e13f14c iommu/arm-smmu: Treat unknown OAS as 48-bit 7d5f529 iommu/arm-smmu: Sort out coherency b608ba0 iommu/arm-smmu: Limit 2-level strtab allocation for small SID sizes be6db28 iommu/arm-smmu: Fix MSI memory attributes to match specification 7f897f1 iommu/arm-smmu: Fix enabling of PRIQ interrupt f378b6d iommu/arm-smmu: Skip the execution of CMD_PREFETCH_CONFIG 3279db5 iommu/arm-smmu: Enlarge STRTAB_L1_SZ_SHIFT to support larger sidsize 28a0c40 iommu/arm-smmu: Fix the values of ARM64_TCR_{I,O}RGN0_SHIFT 4af6733 iommu/arm-smmu: Fix LOG2SIZE setting for 2-level stream tables 9ba251c iommu/arm-smmu: Fix the index calculation of strtab dd7b9dd iommu/arm-smmu: Delete an unnecessary check before the function call "free_io_pgtable_ops" bb6156c iommu: Ignore -ENODEV errors from add_device call-back 5c9a9c1 iommu: Fix checkpatch warnings for Missing a blank line after declarations cfee8dc iommu: Checking for NULL instead of IS_ERR 5d9d60a9 iommu: Introduce iommu_request_dm_for_dev() f0e5d1b iommu: Add function to query the default domain of a group cb5a55a iommu: Create direct mappings in default domains 53ad160 iommu: Introduce direct mapped region handling 6b3f934 iommu: Add iommu_get_domain_for_dev function dbaf936 iommu: Make sure a device is always attached to a domain afe92d7 iommu: Limit iommu_attach/detach_device to devices with their own group 14fa803 iommu: Allocate a default domain for iommu groups abd3972 iommu: Call remove_device call-back after driver release f965234 iommu: Clean up after a failed bus initialization 7123d0f iommu: Propagate error in add_iommu_group 5fbf0f7 iommu: Add a few printk messages to group handling code 781acc5 iommu: Remove function name from pr_fmt() 3a5fb11 iommu: Init iommu-groups support earlier, in core_initcall e9ef9c3 drivers/vfio: Allow type-1 IOMMU instantiation on top of an ARM SMMUv3 99417cc iommu/arm-smmu: Add initial driver support for ARM SMMUv3 devices 546f557 Documentation: dt-bindings: Add device-tree binding for ARM SMMUv3 IOMMU 8a3b1d5 iommu/arm-smmu: Make force_stage module param read-only in sysfs 6dd0ea6 iommu/arm-smmu: Fix ATS1* register writes
On 04/18/2016 05:41 PM, Mark Brown wrote:
On 18 April 2016 at 10:39, Alex Shi <alex.shi@linaro.org mailto:alex.shi@linaro.org> wrote:
This commit (166bdbd2316116 iommu/arm-smmu: Add support for MSI on SMMUv3) depends on a huge MSI change on 4.2/4.3 which couldcause PCI and MSI related driver changes.
Consider the following concern: 1, big effort to backport the MSI domain/list changes. 2, msi_desc struct change would cause MSI related downstream driver rewriting which works on 4.1 kernel. 3, I have no suitable arm64 board in hands, my test relay onNaresh who living in India. So the testing would become unbearable for such big changes.
So I want to skip this commit in the IOMMU backporting. and suggest using LSK 4.4 kernel if someone really want the platform MSI. Any comments?Seems reasonable to me, it looked like we didn't need to backport all the user changes to get the core infrastructure.
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.