On 19.03.21 11:51, Max Reitz wrote:
On 19.03.21 11:50, Laurent Vivier wrote:
Le 19/03/2021 à 10:20, Max Reitz a écrit :
On 19.03.21 07:32, Thomas Huth wrote:
On 18/03/2021 18.28, Max Reitz wrote:
[...]
From that it follows that I don’t see much use in testing
specific devices either. Say there’s
a platform that provides both virtio-pci and virtio-mmio, the
default (say virtio-pci) is fine
for the iotests. I see little value in testing virtio-mmio as
well. (Perhaps I’m short-sighted,
though.)
That's a fair point. But still, if someone compiled QEMU only with a
target that only provided
virtio-mmio, the iotests should not fail when running "make check".
To avoid that we continue playing whack-a-mole here in the future,
maybe it would be better to
restrict the iotests to the "main" targets only, e.g. modify
check-block.sh so that the tests only
run with x86, aarch64, s390x and ppc64 ?
Right, that would certainly be the simplest solution.
The problem with that is we can't run the tests if target-list doesn't
contain one of these targets.
Yes, but is that really a problem?
I should add: The thing is, I wouldn’t really call it a problem. But
still, as I said before somewhere in this thread, in theory we want to
allow running the tests with every configuration. It’s just that it’s a
tradeoff between how much it helps and how much work it is to make them
work. (I gave s390 as an example, where effort was undertaken to make
the iotests work.)
You’ve sent patches, so it seems you’re willing to invest the work.
Sounds good to me, as long as we know it won’t rot.
Max
