Re: bugs in strlen, strstr setlocale -- i386

Hi,

at least 32-Bit Intel-Arch should not segfault due to jecxz:

> cat dietlibc-0.31/i386/strlen.S
[...]
strlen:
        movl    4(%esp),%ecx
        xorl    %eax,%eax
        jecxz   .Lnull
        decl    %eax
.Llbl:  incl    %eax
        cmpb    $0,(%ecx, %eax)
        jne     .Llbl
.Lnull: ret
[...]

Test:

> cat test_strlen.c
#include <string.h>
int main() {
  return strlen((char *)0);
}
> diet gcc -O2 -Wall -o test_strlen test_strlen.c
test_strlen.c: In function 'main':
test_strlen.c:3: warning: null argument where non-null required (argument 1)
> ./test_strlen;echo $?
0
>

It's not easy to keep optimizations for different archs always in a 
unified state.
But with some old K&R docs in mind I guess that the result of strlen(0) is 
not defined (see also gcc warning).

Frank

In 3 Apr 2009, Nikola Vladov cum veritate scripsit :

> #include <string.h>
> int main() {
>   return strlen((char *)0);
> }
> 
> This gives segmentation fault!  
> As a result strstr(0, "UTF-8") works also wrong.
[...]

-- 
#!/bin/sh
$0 &; $0 &

Sending unsolicited commercial email to this address may be a violation
of the Washington State Consumer Protection Act, chapter 19.86 RCW.
Das Verschicken unverlangter kommerzieller email an diese Adresse ist
verboten (LG Traunstein, 2 HK O 3755/97 vom 14.10.1997, CR 1998, 171f).