Bert Huijben wrote on Tue, Aug 30, 2016 at 12:42:14 +0200: > > > > -----Original Message----- > > From: Daniel Shahaf [mailto:d...@daniel.shahaf.name] > > Sent: dinsdag 30 augustus 2016 03:06 > > To: Stefan Fuhrmann <stefanfuhrm...@alice-dsl.de> > > Cc: dev@subversion.apache.org > > Subject: Re: fsfs: Segfault when rep line lists the all-zeroes checksum > > > > Stefan Fuhrmann wrote on Mon, Aug 29, 2016 at 22:10:07 +0200: > > > On 29.08.2016 18:57, Daniel Shahaf wrote: > > > >Line 801 sets CHECKSUM to NULL (as promised by > > svn_checksum_parse_hex()'s > > > >docstring), line 803 dereferences it unconditionally. > > > I vaguely remember that we use(d) all-0 checksums > > > as a 'no checksum' indicator. There may have been > > > some mix-up when rep structure got flattened. > > > > The all-zeroes checksum compares equal to any other checksum. > > I would guess that the all 0 checksum is *also* the checksum of some > very unlikely data, so users might be able to trigger this under some > very unusual circumstances.
md5 has no known preimage attacks, so the probability that somebody would accidentally commit a preimage of any particular checksum is for all practical purposes zero.