[vfio-users] edk2.git-ovmf-x64 from kraxel.org/repos doesn't boot?

[Doug Applegate]

Thanks for the replies. I updated again and the issue seems resolved.
The older version from the fedora repository seems like it's a bit
buggy. Especially when the Windows 10 guest would go into sleep mode,
it seemed to lock up and wouldn't wake up. However the one shipped
from kraxel.org seems happy as ever again.

I'm sure the reason I'm using the kraxel repo was because of following
a gpu passthrough tutorial somewhere and have since really forgotten
what I have set up. I've been using the same configuration for over a
year now so it goes to show how reliable it is I suppose!

Doug

On Wed, Feb 22, 2017 at 6:02 PM, Laszlo Ersek <lersek at redhat.com> wrote:
> On 02/22/17 08:17, Doug Applegate wrote:
>> I've never had problems booting using ovmf firmware from the rpm repo
>> https://www.kraxel.org/repos/ but today I updated (fedora 25) and
>> could no longer boot (black screen, no Tiano Core logo)
>
> Commit dc4c770763d0 ("BaseTools: add error check for Macro usage in the
> INF file", 2017-02-20) broke the build tools in edk2, and they now
> mis-build OVMF. I just ran into the same a few minutes ago (after
> pulling master so I could rebase and send out a new series). I found the
> culprit with bisection.
>
>>
>> In my libvirt configuration I use:
>>  <loader readonly='yes'
>> type='pflash'>/usr/share/edk2.git/ovmf-x64/OVMF_CODE-pure-efi.fd</loader>
>>
>> which I'm assuming is the latest build from the repo,
>
> Yep, looks good.
>
>> however I
>> switched it to this:
>> <loader readonly='yes' type='pflash'>/usr/share/edk2/ovmf/OVMF_CODE.fd</loader>
>>
>> And I can boot again.
>
> You have this other file from... wherever, whenever. :) It is likely
> very old and doesn't suffer from this regression.
>
>>
>> Can someone explain to me what the difference between the two files
>> are? Should I not be running the latest builds from
>> www.kraxel.org/repos/jenkins/edk2/ or is the other one preferred?
>
> The first one (from Gerd) is the right one (even though it broke now).
> The other one -- I have no clue. Sticking with Gerd's builds is a sane
> idea, although, if you update frequently, any fresh upstream regressions
> will hit you immediately, because Gerd's builds track upstream edk2 very
> closely.
>
> Anyway I reported the issue on edk2-devel, so in a few days hopefully
> it'll be fixed.
>
> Laszlo
>
>>
>> Thanks!
>> Doug
>>
>> _______________________________________________
>> vfio-users mailing list
>> vfio-users at redhat.com
>> https://www.redhat.com/mailman/listinfo/vfio-users
>>
>