Re: sha512_process_bytes doesn't like odd sized buffers

[J. Scott Edwards]

Yes, my name and e-mail address look correct.

I assume that since it doesn't affect the coreutils, you won't be
releasing an updated version just for that?  It is okay with me
because I have already checked my fix into the CVS for my project.
But I would like to update to the official release when it comes out.

Thanks
  -Scott

On Sat, Mar 15, 2008 at 8:59 AM, Jim Meyering <address@hidden> wrote:
> Andreas Schwab <address@hidden> wrote:
>  > Jim Meyering <address@hidden> writes:
>  >
>  >> The bug strikes whenever ctx->buflen|64 != 0.
>  >
>  > Btw, ctx->buflen|64 != 0 is always true because it is the same as
>  > ctx->buflen|(64 != 0).
>
>  Shameful ;-)  I meant "ctx->buflen&64", of course.
>
>
>  >> -      sha256_process_block (ctx->buffer, ctx->buflen & ~63, ctx);
>  >> -
>  >>        ctx->buflen &= 63;
>  >> +      sha256_process_block (ctx->buffer, ctx->buflen, ctx);
>  >
>  > I hope you are aware of that a & 63 is different from a & ~63?
>
>  Of course.  My mistake was in not *seeing* that "63 != ~63"
>
>  Thanks.  So I won't be making that latter change.
>