[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [Issue 3981] serf infinite loop over https

From: Philip Martin <philip.martin_at_wandisco.com>
Date: Wed, 13 Jun 2012 23:19:20 +0100

Justin Erenkrantz <justin_at_erenkrantz.com> writes:

> (I do not have an easy way to login to the issue tracker from my iPad...so,
> moving back to the list...)
>
> I really do not think the assert is due to memory corruption - nothing I
> saw indicated that was happening. I think it is far more likely that there
> is another bug somewhere. (It is the simplest explanation, IMHO.)
>
> If the checkout continues beyond the xxx lines in that patch (the xxx lines
> indicate when we would have otherwise gone into a bad state and hung later
> on) and does not end up in an infinite loop or hang, then I think this
> issue can be resolved. And, someone who knows more about libsvn_wc can
> investigate the assert. In the meantime, I will clean up the patch
> tomorrow and commit it to serf... -- justin

>> svn: E235000: In file '../src/subversion/libsvn_wc/wc_db.c' line 1707:
>> assertion
>> failed (SVN_IS_VALID_REVNUM(changed_rev))

When this happens all of changed_rev changed_date and changed_author are
invalid. It appears the update editor's close_file callback has been
invoked before receiving the properties that set those fields. If I
patch the client to set changed_rev to a valid revision then the client
no longer asserts at this point and the close_file callback adds the
file to wc.db. Then the client SEGVs or asserts in serf a short while
later when attempting to process the properties that should have been
processed before calling close_file.

-- 
uberSVN: Apache Subversion Made Easy
http://www.uberSVN.com
Received on 2012-06-14 00:19:58 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.