[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL 19/19] intel_iommu: Fix for IQA reg read dropped DW field
From: |
Michael S. Tsirkin |
Subject: |
[PULL 19/19] intel_iommu: Fix for IQA reg read dropped DW field |
Date: |
Thu, 1 Aug 2024 06:37:18 -0400 |
From: yeeli <seven.yi.lee@gmail.com>
If VT-D hardware supports scalable mode, Linux will set the IQA DW field
(bit11). In qemu, the vtd_mem_write and vtd_update_iq_dw set DW field well.
However, vtd_mem_read the DW field wrong because "& VTD_IQA_QS" dropped the
value of DW.
Replace "&VTD_IQA_QS" with "& (VTD_IQA_QS | VTD_IQA_DW_MASK)" could save
the DW field.
Test patch as below:
config the "x-scalable-mode" option:
"-device intel-iommu,caching-mode=on,x-scalable-mode=on,aw-bits=48"
After Linux OS boot, check the IQA_REG DW Field by usage 1 or 2:
1. IOMMU_DEBUGFS:
Before fix:
cat /sys/kernel/debug/iommu/intel/iommu_regset |grep IQA
IQA 0x90 0x00000001001da001
After fix:
cat /sys/kernel/debug/iommu/intel/iommu_regset |grep IQA
IQA 0x90 0x00000001001da801
Check DW field(bit11) is 1.
2. devmem2 read the IQA_REG (offset 0x90):
Before fix:
devmem2 0xfed90090
/dev/mem opened.
Memory mapped at address 0x7f72c795b000.
Value at address 0xFED90090 (0x7f72c795b090): 0x1DA001
After fix:
devmem2 0xfed90090
/dev/mem opened.
Memory mapped at address 0x7fc95281c000.
Value at address 0xFED90090 (0x7fc95281c090): 0x1DA801
Check DW field(bit11) is 1.
Signed-off-by: yeeli <seven.yi.lee@gmail.com>
Message-Id: <20240725031858.1529902-1-seven.yi.lee@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
hw/i386/intel_iommu.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index 9a768f0b44..16d2885fcc 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -2947,7 +2947,9 @@ static uint64_t vtd_mem_read(void *opaque, hwaddr addr,
unsigned size)
/* Invalidation Queue Address Register, 64-bit */
case DMAR_IQA_REG:
- val = s->iq | (vtd_get_quad(s, DMAR_IQA_REG) & VTD_IQA_QS);
+ val = s->iq |
+ (vtd_get_quad(s, DMAR_IQA_REG) &
+ (VTD_IQA_QS | VTD_IQA_DW_MASK));
if (size == 4) {
val = val & ((1ULL << 32) - 1);
}
--
MST
- [PULL 09/19] Revert "pcie_sriov: Register VFs after migration", (continued)
- [PULL 09/19] Revert "pcie_sriov: Register VFs after migration", Michael S. Tsirkin, 2024/08/01
- [PULL 10/19] Revert "pcie_sriov: Remove num_vfs from PCIESriovPF", Michael S. Tsirkin, 2024/08/01
- [PULL 11/19] Revert "pcie_sriov: Release VFs failed to realize", Michael S. Tsirkin, 2024/08/01
- [PULL 12/19] Revert "pcie_sriov: Reuse SR-IOV VF device instances", Michael S. Tsirkin, 2024/08/01
- [PULL 13/19] Revert "pcie_sriov: Ensure VF function number does not overflow", Michael S. Tsirkin, 2024/08/01
- [PULL 14/19] Revert "pcie_sriov: Do not manually unrealize", Michael S. Tsirkin, 2024/08/01
- [PULL 15/19] Revert "hw/ppc/spapr_pci: Do not reject VFs created after a PF", Michael S. Tsirkin, 2024/08/01
- [PULL 16/19] Revert "hw/ppc/spapr_pci: Do not create DT for disabled PCI device", Michael S. Tsirkin, 2024/08/01
- [PULL 17/19] Revert "hw/pci: Rename has_power to enabled", Michael S. Tsirkin, 2024/08/01
- [PULL 18/19] hw/i386/amd_iommu: Don't leak memory in amdvi_update_iotlb(), Michael S. Tsirkin, 2024/08/01
- [PULL 19/19] intel_iommu: Fix for IQA reg read dropped DW field,
Michael S. Tsirkin <=
- Re: [PULL 00/19] virtio,pci,pc: fixes, Richard Henderson, 2024/08/01