Re: git diff-tree commit detail bug in 2.0.2 and 2.0.3
- From
Jeff King <peff@peff.net>
- Date
- Jul 28, 2014, 10:35 UTC
- Message-ID
- <20140728103504.GB10737@peff.net>
- In-Reply-To
- <CAGyf7-HKpfyi5OqXS9BhtfXUEZXbisawpTPK9UFOQObz1qhRUw@mail.gmail.com>
On Mon, Jul 28, 2014 at 07:42:16PM +1000, Bryan Turner wrote:
Show 9 quoted lines
> Running a git bisect between v2.0.1, which does not manifest this > issue, and v2.0.2 fingers the following commit: > bturner@ubuntu:~/Development/oss/git/git$ git bisect bad > c1b3c71f4b4571abb2b2a457122fd100dc9f7eb0 is the first bad commit > commit c1b3c71f4b4571abb2b2a457122fd100dc9f7eb0 > Author: Jeff King <peff@peff.net> > Date: Tue Jun 10 17:43:02 2014 -0400 > > commit: convert commit->buffer to a slab
I haven't reproduced here yet, but this is almost certainly the bug where lookup_unknown_object causes a bogus commit->index field (and prior to the commit you found, diff-tree did not use commit->index).
The series that Junio has in jk/alloc-commit-id should fix the problem (it's in master already, and slated for v2.1.0).
Junio, we should consider a v2.0.4 with that series, I think. This is a pretty serious regression in diff-tree (I didn't even realize that the buffer-slab work went into the maint series; that may have been a little ambitious).
-Peff