Re: [Qemu-devel] [PATCHv2] tests: re-enable vhost-user-test

[Michael S. Tsirkin]

On Tue, Oct 27, 2015 at 10:33:26AM -0400, Marc-André Lureau wrote:
> Hi
> 
> ----- Original Message -----
> > On 26 October 2015 at 14:32,  <address@hidden> wrote:
> > > From: Marc-André Lureau <address@hidden>
> > >
> > > Commit 7fe34ca9c2e actually disabled vhost-user-test altogether,
> > > since CONFIG_VHOST_NET is a per-target config variable.
> > >
> > > tests/vhost-user-test is already x86/x64 softmmu specific test, in order
> > > to enable it correctly, kvm & vhost-net are also conditions. To check
> > > that, set CONFIG_VHOST_NET_TEST_$target when kvm is also enabled.
> > >
> > > Since "check-qtest-x86_64-y = $(check-qtest-i386-y)", avoid duplication
> > > when both x86 & x64 are enabled.
> > >
> > > Other targets than x86 aren't enabled yet, and is intentionally left as
> > > a future improvement, since I can't easily test those.
> > >
> > > Signed-off-by: Marc-André Lureau <address@hidden>
> > 
> > I ran this through my build-tests, which pass, but:
> > 
> > (a) there's still the clang warning about the negative shifts
> > in target-i386/. This is an ancient bug and we can fix it later.
> 
> I sent a patch to fix this.
> 
> > (b) there are new warning messages:
> > Warning: path not on HugeTLBFS: /tmp/vhost-test-xaGJRK
> > Warning: path not on HugeTLBFS: /tmp/vhost-test-xaGJRK
> > Warning: path not on HugeTLBFS: /tmp/vhost-test-xaGJRK
> 
> These are part of qemu file_ram_alloc()
> 
> > I would like (b) fixed -- tests should either:
> >  (1) complete without printing "warning" about anything
> 
> We would need to change the qemu warning.

Rather, disable it if running under test.

> >  (2) fail the test if the warning is actually important
> >  (3) skip the test if the test requires something that the host
> >      machine doesn't have (like a hugetlbfs)
> 
> It is not important for the test to succeed.
> 
> I imagine a few options to get rid of the warning:
> 1. only run the test on hugetlbfs
> 2. remove the warning from qemu
> 3. silence qemu errors in the test
> 4. add an option to memory-backend-file to require hugetlbfs: something like 
> ...,require-hugetlbfs=true,false,warn
> 
> I guess 4. is the most interesting, although I would need some advice on how 
> to express this best.
> 
> (tbh, I think this could be addressed later)

Well not annoying people is important, otherwise they stop
paying attention to warnings.

-- 
MST