Mattias Engdegård (CC'd) reports on IRC that apr_atomic_xchgptr() in
recent APR's is broken, and the brokenness can affect Subversion under
certain circumstances:
<@danielsh> wayita: t
< wayita> gcc 4.0.x or older
< wayita> apr 1.3.x or newer
< wayita> x86-64 only
< wayita> WORKAROUND: --disable-nonportable-atomics to APR's configure
< wayita> OS: any unix
< wayita> nature of the bug: apr_atomic_xchgptr() does not initialize the return value
< wayita> https://issues.apache.org/bugzilla/show_bug.cgi?id=51851
< wayita> manifests as a segfault in svn.
I suppose we could recommend --disable-nonportable-atomics in the
release notes. It is undetermined at this time whether Windows 64-bit
is affected.
Daniel
Hyrum K Wright wrote on Mon, Sep 19, 2011 at 10:54:19 -0500:
> All,
> I'm currently planning on rolling the final 1.7.0 release candidate
> this Thursday evening (UTC +0100). Assuming the tests pass and
> signatures are valid, I'll cut the final release from the same magic
> revision early next week, with the hope of releasing by weeks' end
> (folks won't need to retest if they already tested rc4).
>
> However, these dates are fluid, and depend on our feeling of
> releasability as a community. I'm uncomfortable making that decision
> on my own; I know there may be outstanding issues, but don't know
> their severity or impact. I therefore request an up-or-down vote of
> the above plan.
>
> Please, no +/-0 votes on this one. I'm asking you to take a stand,
> either in favor or against proceeding with the release. I *don't*
> want a couple of +1s and a bunch of +/-0's; we need to be sure we have
> broad community support behind this much-anticipated release.
>
> In other words, if you've got an objection, please voice it *NOW*.
> (Even if you don't have any objections, it's important to know that,
> too.)
>
> Thanks,
> -Hyrum
>
>
> --
>
> uberSVN: Apache Subversion Made Easy
> http://www.uberSVN.com/
Received on 2011-09-20 20:54:07 CEST