Hi Michael,
On 2/25/2025 2:17 PM, Michael Tokarev wrote: > 07.02.2025 07:53, Sairaj Kodilkar wrote: >> This series provides few bug fixes for emulated AMD IOMMU. The series is >> based >> on top of qemu upstream master commit d922088eb4ba. >> >> Patch 1: The code was using wrong DTE field to determine interrupt >> passthrough. >> Hence replaced it with correct field according to [1]. >> >> Patch 2: Current code sets the PCI capability BAR low and high to the >> lower and upper 16 bits of AMDVI_BASE_ADDR respectively, which is >> wrong. Instead use 32 bit mask to set the PCI capability BAR low >> and >> high. >> The guest IOMMU driver works with current qemu code because it uses >> base address from the IVRS table and not the one provided by >> PCI capability. >> >> Sairaj Kodilkar (2): >> amd_iommu: Use correct DTE field for interrupt passthrough >> amd_iommu: Use correct bitmask to set capability BAR >> >> hw/i386/amd_iommu.c | 10 +++++----- >> hw/i386/amd_iommu.h | 2 +- >> 2 files changed, 6 insertions(+), 6 deletions(-) > > Is this qemu-stable material (current series: 7.2, 8.2, 9.2)? Linux kernel doesn't use these changes. So its fine. But I believe we care for other OS as well? if yes then better to backport. > > 3684717b74 "amd_iommu: Use correct bitmask to set capability BAR" does > not apply to 7.2, since v8.0.0-10-g6291a28645 "hw/i386/amd_iommu: Explicit > use of AMDVI_BASE_ADDR in amdvi_init" in not in 7.2, but the change can be > adjusted for 7.2 easily, or 6291a28645 can be picked up too. How is this works? You will pick it up -OR- you want us to backport and send it to stable mailing list? -Vasant
