[PATCH #14090] md5_process_block() produces incorrect result with large block sizes

[Paul Eggert]

On 05/24/2012 08:11 AM, Andreas Jaeger wrote:
> Ping, what should be done with these two patches?

Your email had just one patch, which unfortunately
was incorrect because it has no effect even on hypothetical
hosts with 128-bit size_t, as its expression
((uint64_t) len>>  63>>  1) is always zero.

I just now attached a corrected sha512.c patch to
the report for bug#14090.  The two non-obsolete patches that
are now attached to that bug report should both
be applied, though I hope somebody can test them
first -- I haven't even compiled them, much less
tested them.