broken strndup on AIX 5.1

[Doug VanLeuven]

Sorry about the error in cut & paste.  I disagree strndup is OK.  It 
passes simple tests, but not complicated ones in conjunction with malloc.
Refer bug 1097
https://bugzilla.samba.org/show_bug.cgi?id=1097
and the recent contribution by Heinrich.Mislik to prove this.

I can test 4.3.3
Using your code
rs1scwa /root/src # uname -a
AIX rs6 3 4 000216FF4C00 unknown
rs1scwa /root/src # more strnlen.c
#include <stdio.h>
#include <string.h>

int main(void)
{
  char *test = "1234567890";
  int i;
  for (i=0;i<9;i++) {
    printf("%d\n", strnlen(test,i));
  }
  return 0;
}
rs1scwa /root/src # gcc strnlen.c -o strnlen                   
rs1scwa /root/src # ./strnlen
0
3
2
1
0
7
6
5
4


Simo Sorce wrote:

>On Mon, 2004-08-23 at 23:20, William Jojo wrote:
>  
>
>>If I can help here. I do not have the gcc used, but I ran these test
>>programs using the IBM compiler on 5.2 and the listed filesets described
>>below. It appears that strndup is fine, but strnlen returns questionable
>>results.
>>    
>>
>
>Yeah I just reproduced the same result on a 5.1 AIX box.
>
>strndup works fine, but strnlen give this result:
>
>strnlen.c
>-----------------------------------------------
>#include <stdio.h>
>#include <string.h>
>
>int main(void)
>{
>  char *test = "1234567890";
>  int i;
>  for (i=0;i<9;i++) {
>    printf("%d\n", strnlen(test,i));
>  }
>  return 0;
>}
>-----------------------------------------------
>./strnlen
>0
>3
>2
>1
>0
>7
>6
>5
>4
>
>Do anyone here have access to an AIX 4.3.3 box and check strndup is ok
>and strnlen broken here too?
>
>Simo.
>
>  
>