On 2016-02-22 12:05, David Kiarie wrote: > On Mon, Feb 22, 2016 at 10:29 AM, Jan Kiszka <jan.kis...@web.de> wrote: >> On 2016-02-22 06:57, David Kiarie wrote: >>> On Sun, Feb 21, 2016 at 11:20 PM, Jan Kiszka <jan.kis...@web.de> wrote: >>>> On 2016-02-21 19:10, David Kiarie wrote: >>>>> Hello there, >>>>> >>>>> Repost, AMD IOMMU patches version 6. >>>>> >>>>> Changes since version 5 >>>>> -Fixed macro formating issues >>>>> -changed occurences of IO MMU to IOMMU for consistency >>>>> -Fixed capability registers duplication >>>>> -Rebased to current master >>>> >>>> I suspect this still has some subtle bugs: I'm running the patches over >>>> master with standard Linux distro as guest, full desktop, and I'm >>>> getting sporadic segfaults of arbitrary programs. These disappear once I >>>> disable the IOMMU or switch to the Intel version. >>> >>> Is this L1 guest or L2 guest ? - haven't got any such so far. >> >> It's L1 only. >> >>> >>>> >>>> How did you test so far? >>> >>> I mainly test by logging. I've tested L1 without any iommu-related >>> command line parameters and with L1 with 'iommu=1 iommu=pt'. L2 guest; >>> passed-through a device checked it's working correctly, that all. >>> These guests barely have any load though. >> >> I quickly reproduced the issue by starting some "heavier" applications, >> a browser or an office suite. Something is apparently always corrupted >> then, data or code, thus the crashes. > > Can't reproduce this issue with my ubuntu/debian VMs. Are you using > any iommu-related command line parameters ?
This is my command line:
qemu-system-x86_64
-drive file=/path/to/64-bit-linux.img,discard=unmap,if=none,id=disk
-device ide-hd,drive=disk -snapshot -m 1G -enable-kvm -smp 4
-machine q35,iommu=amd
64-bit guest userland, linus.git head guest kernel.
Jan
signature.asc
Description: OpenPGP digital signature
