On Sun, 13 Jan 2008, David Shaw wrote:
> > As can be seen at [0] paperkey fails to build from source on sparc[1].
> > The testsuite fails due to unaligned memory access in sha1_read_ctx.
>
> Interesting. I actually tested on Solaris running sparc, too. Your
> patch seems fine, but the sha1 code actually comes from gnulib, so the
> alignment fix should be sent there. With your permission, I'll
> forward it.
By all means.
> Until gnulib fixes the alignment issue, how about this patch for
> paperkey? It mallocs the buffer, which should ensure that it is
> correctly aligned.
> else if(packet->buf[0]==4)
> {
> - struct sha1_ctx sha;
> + struct sha1_ctx *sha;
> unsigned char head[3];
>
> - sha1_init_ctx(&sha);
> + sha=xmalloc(sizeof(*sha));
>
> + sha1_init_ctx(sha);
> +
> head[0]=0x99;
> head[1]=public_len>>8;
> head[2]=public_len&0xFF;
>
> - sha1_process_bytes(head,3,&sha);
> - sha1_process_bytes(packet->buf,public_len,&sha);
> - sha1_finish_ctx(&sha,fingerprint);
> + sha1_process_bytes(head,3,sha);
> + sha1_process_bytes(packet->buf,public_len,sha);
> + sha1_finish_ctx(sha,fingerprint);
> +
> + free(sha);
> }
I don't think that'd fix it - the sha struct was aligned properly, the
fingerprint wasn't.
Peter
--
| .''`. ** Debian GNU/Linux **
Peter Palfrader | : :' : The universal
http://www.palfrader.org/ | `. `' Operating System
| `- http://www.debian.org/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
