On 2/2/21 3:56 PM, Daniel P. Berrangé wrote:
> On Tue, Feb 02, 2021 at 03:46:11PM +0100, Philippe Mathieu-Daudé wrote:
>> On 2/2/21 3:28 PM, Kevin Wolf wrote:
>>> If the qemu-system-{arch} binary for the host architecture can't be
>>> found, the old 'check' implementation selected the alphabetically first
>>> system emulator binary that it could find. The new Python implementation
>>> just uses the first result of glob.iglob(), which has an undefined
>>> order.
>>>
>>> This is a problem that breaks CI because the iotests aren't actually
>>> prepared to run on any emulator. They should be, so this is really a bug
>>> in the failing test cases that should be fixed there, but as a quick
>>> fix, let's revert to the old behaviour to let CI runs succeed again.
>>
>> FWIW this is the same problem I had 1 year ago and tried to
>> fix it by sending QMP 'query-version' (introduced in v0.14):
>> https://www.mail-archive.com/qemu-devel@nongnu.org/msg675075.html
>
> In the current failures the issue isn't the version number. Rather some
> of the tests (mistakenly) assume the emulator supports PCI, and we're
> randomly sometimes picking emulator targets that lack PCI.
Then this patch from the same series:
https://www.mail-archive.com/qemu-devel@nongnu.org/msg675088.html
It queries TYPE_DEVICE, but we can query TYPE_PCI_DEVICE too,
which should be selected if a machine have a TYPE_PCI_HOST_BRIDGE
providing a TYPE_PCI_BUS.
Anyway the idea is to query the binary with QMP to see if it makes
sens to run the test with it.
Regards,
Phil.
