Re: lock_tests-9 is segfaulting over ra_dav

On 2005-04-18 10:21-0500, Brian W. Fitzpatrick wrote:
>
> On Apr 17, 2005, at 6:11 PM, Philip Martin wrote:
> >
> >Something like the following perhaps?
> >
> >Fix a neon hook lifetime problem.
> >
> >* subversion/libsvn_ra_dav/session.c
> > (setup_neon_request_hook): Use the session pool instead of passing a
> > pool argument, don't setup the hook more than once.
> > (shim_svn_ra_dav__lock, shim_svn_ra_dav__unlock): memset the
> > log_request_baton struct not the pointer.
> > (svn_ra_dav__get_lock): Use setup_neon_request_hook to setup the
> > hook rather than coding it directly.
>
> +1 Please commit and drop an item into 1.2.x/STATUS. I'll run it
> through the test suite and vote for it as well.

Test suite won't catch it alone, you have to have a system where
segfaults are reported on system logs. The lock_tests-9 is using as a
sentinel command

  svntest.actions.run_and_verify_svn(None, None, None, 'unlock',

which won't detect segfaults and is passing even when svn has fallen
flat on its face. This and the fact that I have been traveling was the
reason why this failure has been going for so long time. This was
second time in short time range when svntest has slipped through
segfaulting tests.

BR, Jani

-- 
Jani Averbach 
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org