iommu/amd: Use IVHD EFR for early initialization of IOMMU features
authorSuravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Wed, 20 Jan 2021 13:50:02 +0000 (07:50 -0600)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 3 Feb 2021 22:28:50 +0000 (23:28 +0100)
commit974b6289e8dac1cdd700c73b555a54cade760feb
treeee41744ff90e6835028694ef51c8443b6827bcda
parent57e2393f5de0368b2e89a69d39cc9c3f1e1043cd
iommu/amd: Use IVHD EFR for early initialization of IOMMU features

[ Upstream commit a44092e326d403c7878018ba532369f84d31dbfa ]

IOMMU Extended Feature Register (EFR) is used to communicate
the supported features for each IOMMU to the IOMMU driver.
This is normally read from the PCI MMIO register offset 0x30,
and used by the iommu_feature() helper function.

However, there are certain scenarios where the information is needed
prior to PCI initialization, and the iommu_feature() function is used
prematurely w/o warning. This has caused incorrect initialization of IOMMU.
This is the case for the commit 6d39bdee238f ("iommu/amd: Enforce 4k
mapping for certain IOMMU data structures")

Since, the EFR is also available in the IVHD header, and is available to
the driver prior to PCI initialization. Therefore, default to using
the IVHD EFR instead.

Fixes: 6d39bdee238f ("iommu/amd: Enforce 4k mapping for certain IOMMU data structures")
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Brijesh Singh <brijesh.singh@amd.com>
Reviewed-by: Robert Richter <rrichter@amd.com>
Link: https://lore.kernel.org/r/20210120135002.2682-1-suravee.suthikulpanit@amd.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/iommu/amd/amd_iommu.h
drivers/iommu/amd/amd_iommu_types.h
drivers/iommu/amd/init.c