[avr-libc-dev] [ Bug #1720 ] memcmp bug ?

[Theodore A. Roth]

Regarding:

  http://savannah.nongnu.org/bugs/?func=detailbug&bug_id=1720&group_id=2140

Is this really a bug?

I did a google search and found these:

http://tigcc.ticalc.org/doc/mem.html#memcmp
http://216.239.33.100/search?q=cache:MOUn1YX5j6sC:www.htsoft.com/htmlman/picc18/library.htm+memcmp+signed+bytes&hl=en&ie=UTF-8

which seem to imply that the compare is signed. But these imply the
compare is unsigned:

http://www.cs.vu.nl/pub/minix/2.0.2/wwwman/man3/string.3.html
http://soma.bkbits.net:8080/linux-arm-2.4-somaport/anno/lib/address@hidden@address@hidden

I'm leaning toward leaving the behaviour as is, but adding a note in the
dox explaining the signed nature of the compare operation.

If the change should be made, I think it would be the following patch
(don't sign extend the result). Is this patch correct?

Index: libc/string/memcmp.S
===================================================================
RCS file: /cvsroot/avr-libc/avr-libc/libc/string/memcmp.S,v
retrieving revision 1.3
diff -u -r1.3 memcmp.S
--- libc/string/memcmp.S        5 Aug 2002 19:21:36 -0000       1.3
+++ libc/string/memcmp.S        18 Nov 2002 19:28:15 -0000
@@ -68,8 +68,6 @@
 ; strings are equal, so clear both ret_lo and carry
        sub     ret_lo, ret_lo
 .memcmp_done:
-; ret_hi = SREG.C ? 0xFF : 0
-       sbc     ret_hi, ret_hi
        ret
 .memcmp_end:
        .size   _U(memcmp), .memcmp_end - _U(memcmp)



Ted Roth