Hi,
Syn wrote:
> The value returned by csum_block_sub() may have the 16 MSB bits set.
> th->check = tcp_v4_check(th, len, inet->saddr, inet->daddr,
> csum_partial((char *)th,
> th->doff << 2,
> skb->csum));
There indeed seems to be an issue in csum_partial: quoting beginning and
end of csum_partial in checksum.S:
.align 32
.globl csum_partial
csum_partial: /* %o0=buff, %o1=len, %o2=sum */
/* skb->csum is in %o2 */
... a lot of computing that doesn't involve %o2 ...
1: add %o2, %o4, %o2
csum_partial_finish:
retl
mov %o2, %o0
The last add doesn't take care of the potential carry into the 33th
bit. I guess something like
1: add %o2, %o4, %o2
srlx %o2, 32, %o4
srl %o2, 0, %o2
add %o2, %o4, %o2
csum_partial_finish:
retl
mov %o2, %o0
should be correct.
Samuel
-
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
