Re: [Qemu-devel] [PATCH v2 13/13] qemu-iotests: More qcow2 reopen tests

[Kevin Wolf]

Am 04.09.2015 um 21:52 hat Max Reitz geschrieben:
> On 04.09.2015 19:18, Kevin Wolf wrote:
> > Signed-off-by: Kevin Wolf <address@hidden>
> > ---
> >  tests/qemu-iotests/137     | 143 
> > +++++++++++++++++++++++++++++++++++++++++++++
> >  tests/qemu-iotests/137.out |  43 ++++++++++++++
> >  tests/qemu-iotests/group   |   1 +
> >  3 files changed, 187 insertions(+)
> >  create mode 100755 tests/qemu-iotests/137
> >  create mode 100644 tests/qemu-iotests/137.out
> > 
> > diff --git a/tests/qemu-iotests/137 b/tests/qemu-iotests/137
> > new file mode 100755
> > index 0000000..ad52f71
> > --- /dev/null
> > +++ b/tests/qemu-iotests/137
> 
> 
> [snip]
> 
> > +echo
> > +echo === Try setting some invalid values ===
> > +echo
> > +
> > +$QEMU_IO \
> > +    -c "reopen -o lazy-refcounts=42" \
> > +    -c "reopen -o cache-size=1M,l2-cache-size=64k,refcount-cache-size=64k" 
> > \
> > +    -c "reopen -o cache-size=1M,l2-cache-size=2M" \
> > +    -c "reopen -o cache-size=1M,refcount-cache-size=2M" \
> > +    -c "reopen -o l2-cache-size=256T" \
> > +    -c "reopen -o l2-cache-size=64G" \
> 
> Testing this seems risky. The rest looks fine, but I'm not sure whether
> this is really needed so much that we should risk the test failing on
> some machines, especially in the future.

I think we should be testing the "too large, but not much too large"
case. A risk of false positives is better than a risk to break things
without noticing. I could make it something like 64T, though, that
should last a bit longer. The "much too large" case starts at 128T
(INT_MAX * s->cluster_size == 2G * 64k).

> Or actually even on my machine:
> 
> $ touch ~/tmp/swap
> $ truncate ~/tmp/swap --size 128G
> $ fallocate -l 128G ~/tmp/swap
> # mkswap ~maxx/tmp/swap
> # swapon ~max/tmp/swap

This looks faked. :-P

> $ cd /to/qemu/build
> $ ./qemu-io -c 'reopen -o l2-cache-size=64G' test.qcow2
> $
> 
> (Running the test itself is apparently a bad idea, because then for some
> reason Linux thinks I actually want to use the swap and everything slows
> down to a crawl)

Well, if you do that intentionally, I guess you deserve it...

Kevin

pgpPUC9ODvm49.pgp
Description: PGP signature