Re: [PATCH RFC v2 4/4] 9p: fix race issue in fid contention.

[Dominique Martinet <asmadeus@codewreck.org>]

Christian Schoenebeck wrote on Wed, Nov 04, 2020:
> > Greg, Christian - from what I understood (in private, hopefully I'm
> > allowed to repeat!), he won't be able to contribute to qemu because of
> > company policies and I'm unlikely to take the time either right now.
> > I don't think it's a problem to continue as is though, we can land linux
> > kernel support (it's still useful for non-qemu servers) and if someone
> > is interested later on they'll just need to finish that bit.
> 
> Hmm, no idea what kind of policy that is; there is no GPL3 in qemu at least 
> that some companies are concerned about, but OK not my business.
> 
> I actually thought this would still take a while on kernel side,

To be honest, so did I -- the original patches are so old I had more or
less given up on it :P

But I don't see any more problem now and we'll want to get there
eventually so now's a good time as any... I just want to get fault
injection to work to test various refcounting cornercases but shouldn't
be much longer.

> so in the 
> meantime we layed the ground in qemu for resolving this issue independent of 
> clients and independent of any guest OS installation by introducing test cases 
> using the 9pfs 'local' filesystem driver:
> 
> https://github.com/qemu/qemu/blob/master/tests/qtest/virtio-9p-test.c
> 
> So the idea was to resolve that chicken egg problem of this issue that way and 
> also handle it a bit more systematically. If you now run qemu's 9p tests with 
> latest git version (or at least with yesterday's QEMU 5.2 rc1 tarball):
> 
> cd qemu/build
> make
> export QTEST_QEMU_BINARY=x86_64-softmmu/qemu-system-x86_64
> tests/qtest/qos-test
> 
> these tests will now create a test directory qtest-9p-local-XXXXXX under the 
> current directory (i.e. the build directory) where they are creating real 
> directories and files like on a production system would do, just without a 
> guest OS.
> 
> As you can see, there are already 9p tests for creating and deleting 
> directories, files, symlinks and hard links, etc.
> 
> Maybe somebody interested to see this issue resolved in qemu might help by 
> rebasing Greg's old patches and testing it with some test cases this way. 
> Personally I need to work on some other things in the next couple weeks, but 
> if somebody needs help, questions, review, etc., I'll be there.

Great news, nice work there.
I see the new tests it doesn't look hard to add new ones reproducing
open-unlink-fstat for example; I think it's good to have regardless of
kernel progress.

We'll get there!
-- 
Dominique