On Wed, Nov 24, 2021 at 9:09 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
>
>
> Although I see slightly worse performance with aligned `buff`  in
> the branch-free approach. Imagine if non-aligned `buff` is that
> uncommon might be better to speculate past the work of `ror`.

Yes, no clear win here removing the conditional (same cost really),
although using a ror32() is removing the from32to16() helper and get
rid of one folding.

I will formally submit this change, thanks !

diff --git a/arch/x86/lib/csum-partial_64.c b/arch/x86/lib/csum-partial_64.c
index 1eb8f2d11f7c785be624eba315fe9ca7989fd56d..cf4bd3ef66e56c681b3435d43011ece78438376d
100644
--- a/arch/x86/lib/csum-partial_64.c
+++ b/arch/x86/lib/csum-partial_64.c
@@ -11,16 +11,6 @@
 #include <asm/checksum.h>
 #include <asm/word-at-a-time.h>

-static inline unsigned short from32to16(unsigned a)
-{
-       unsigned short b = a >> 16;
-       asm("addw %w2,%w0\n\t"
-           "adcw $0,%w0\n"
-           : "=r" (b)
-           : "0" (b), "r" (a));
-       return b;
-}
-
 /*
  * Do a checksum on an arbitrary memory area.
  * Returns a 32bit checksum.
@@ -41,6 +31,7 @@ __wsum csum_partial(const void *buff, int len, __wsum sum)
        if (unlikely(odd)) {
                if (unlikely(len == 0))
                        return sum;
+               temp64 = ror32((__force u32)sum, 8);
                temp64 += (*(unsigned char *)buff << 8);
                len--;
                buff++;
@@ -129,10 +120,8 @@ __wsum csum_partial(const void *buff, int len, __wsum sum)
 #endif
        }
        result = add32_with_carry(temp64 >> 32, temp64 & 0xffffffff);
-       if (unlikely(odd)) {
-               result = from32to16(result);
-               result = ((result >> 8) & 0xff) | ((result & 0xff) << 8);
-       }
+       if (unlikely(odd))
+               result = ror32(result, 8);
        return (__force __wsum)result;
 }
 EXPORT_SYMBOL(csum_partial);