From: David Miller <davem@xxxxxxxxxxxxx> Date: Sun, 04 Jun 2006 21:05:14 -0700 (PDT) > Let me know if it works for you. It turns out the checksum+copy code had the same exact problem, so let's kill that bug too. Here is the current patch: diff-tree ae5de0ff0bc24664a053109c6caa782ba2ad7c53 (from 672c6108a51bf559d19595d9f8193dfd81f0f752) Author: David S. Miller <davem@xxxxxxxxxxxxxxxxxxxx> Date: Sun Jun 4 21:32:01 2006 -0700 [SPARC64]: Fix missing fold at end of checksums. Both csum_partial() and the csum_partial_copy*() family of routines forget to do a final fold on the computed checksum value on sparc64. So do the standard Sparc "add + set condition codes, add carry" sequence, then make sure the high 32-bits of the return value are clear. Based upon some excellent detective work and debugging done by Richard Braun and Samuel Thibault. Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx> diff --git a/arch/sparc64/lib/checksum.S b/arch/sparc64/lib/checksum.S index ba9cd3c..1d230f6 100644 --- a/arch/sparc64/lib/checksum.S +++ b/arch/sparc64/lib/checksum.S @@ -165,8 +165,9 @@ csum_partial_end_cruft: sll %g1, 8, %g1 or %o5, %g1, %o4 -1: add %o2, %o4, %o2 +1: addcc %o2, %o4, %o2 + addc %g0, %o2, %o2 csum_partial_finish: retl - mov %o2, %o0 + srl %o2, 0, %o0 diff --git a/arch/sparc64/lib/csum_copy.S b/arch/sparc64/lib/csum_copy.S index 71af488..e566c77 100644 --- a/arch/sparc64/lib/csum_copy.S +++ b/arch/sparc64/lib/csum_copy.S @@ -221,11 +221,12 @@ FUNC_NAME: /* %o0=src, %o1=dst, %o2=len sll %g1, 8, %g1 or %o5, %g1, %o4 -1: add %o3, %o4, %o3 +1: addcc %o3, %o4, %o3 + addc %g0, %o3, %o3 70: retl - mov %o3, %o0 + srl %o3, 0, %o0 95: mov 0, GLOBAL_SPARE brlez,pn %o2, 4f - 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