Re: [Qemu-devel] [PATCH 33/34] qemu-iotests: Test cache mode option inheritance

[Max Reitz]

On 18.05.2015 16:39, Kevin Wolf wrote:
> Am 15.05.2015 um 21:16 hat Max Reitz geschrieben:
> > On 08.05.2015 19:22, Kevin Wolf wrote:
> > > This is doing a more complete test on setting cache modes both while
> > > opening an image (i.e. in a -drive command line) and in reopen
> > > situations. It checks that reopen can specify options for child nodes
> > > and that cache modes are correctly inherited from parent nodes where
> > > they are not specified.
> > >
> > > Signed-off-by: Kevin Wolf <address@hidden>
> > > ---
> > >   tests/qemu-iotests/132     | 336 ++++++++++++++++++++
> > >   tests/qemu-iotests/132.out | 767 +++++++++++++++++++++++++++++++++++++++++++++
> > >   tests/qemu-iotests/group   |   1 +
> > >   3 files changed, 1104 insertions(+)
> > >   create mode 100755 tests/qemu-iotests/132
> > >   create mode 100644 tests/qemu-iotests/132.out
> >
> > So starting from now I once again need to specify -c writethrough
> > for running tests in tmpfs... Welp.
> >
> > ("Welp" because there is no actual strict requirement to have
> > O_DIRECT, because we don't actually use it but only check the
> > configuration; so using null-co would be fine, too, but we can only
> > use that with raw as the format driver, and raw doesn't support
> > backing files...)
> Perhaps we should add a switch to tell qemu-iotests to skip test cases
> that would require cache.direct=on? Or create a group 'direct' so you
> can use '-x direct'?

Well, it would be 'non-direct', and in this case it's not better than -c 
writethrough. ;-)

I was lucky I got away without specifying the cache mode until now, and 
now I'll just have to live with it. If we (I) would want to make it 
easier, we'd have to probe (in ./check) whether $TEST_DIR supports 
O_DIRECT or not.

Max