Re: git hang with corrupted .pack

[Nicolas Pitre <nico@fluxnic.net>]

On Wed, 14 Oct 2009, Shawn O. Pearce wrote:

> Nicolas Pitre <nico@fluxnic.net> wrote:
> > > Some types of corruption to a pack may confuse the deflate stream
> > > which stores an object.  In Andy's reported case a 36 byte region
> > > of the pack was overwritten, leading to what appeared to be a valid
> > > deflate stream that was trying to produce a result larger than our
> > > allocated output buffer could accept.
> ...
> > This is unfortunate that making a test case for this isn't exactly 
> > trivial.
> 
> Hmmm.  We could do something like manually create a pack file of
> one non-delta blob whose pack header length is 16, but use a zlib
> stream whose result body is 64.  Prior to this fix, we'd be stuck
> in the infinite loop.  :-)

Ah, of course.

> Its a PITA to create though, you have to hand-craft the test vector
> and save it in the repository, we can't produce such a pack with
> any real code we ship.

Can be done easily with dd though, see do_corrupt_object() in t5303 for 
example.


Nicolas