RE: malloc and threads

["Mathew, Tisson K" <tisson dot k dot mathew at intel dot com>]

Thanks Wolfram,

Here is the code; I'm running glibc 2.2.2 on XScale iq80320. 

-------------------------- 
#define MYSIZE                          (1*1024)
#define SLEEP_TIME                      (100*1000) // usec

void *AllocThread(void *pvArg);

int main(int argc, char **argv)
{
  pthread_t hThread;
  int       ii;
  int       iLimit=12;

  for(ii=0;ii<iLimit;ii++)
    pthread_create(&hThread,NULL,AllocThread,NULL);

  AllocThread(NULL);

  return 0;
}

void *AllocThread(void *pvArg)
{
  struct timeval tv;
  char          *pBuff=NULL;

  for(;;)
    {
      if((pBuff=(char *)malloc(MYSIZE))==NULL)
        printf("Could not alloc in parent!\n");
      else
        memset(pBuff,0x05,MYSIZE);

      tv.tv_sec=0;
      tv.tv_usec=SLEEP_TIME;
      select(0,NULL,NULL,NULL,&tv);
      if(pBuff)
        free(pBuff);
    }

  return NULL;
}
------------------------------------------------------------

thanks
_tisson


-----Original Message-----
From: Wolfram Gloger [mailto:Wolfram.Gloger@dent.med.uni-muenchen.de] 
Sent: Tuesday, November 26, 2002 9:12 AM
To: tisson.k.mathew@intel.com
Cc: libc-alpha@sources.redhat.com
Subject: Re: malloc and threads

Hi,

> A program that simply launches 10 threads that just malloc, wait 100
> milliseconds, and free the data and then loop.  

Please post the code!

> This simple program segmentation faults. I'm using glibc version 2.2.2. I
> was wondering if malloc **still** has reentrancy problems or not ? 

malloc never was reentrant and probably never will be.

If you mean "thread-safe", I'm not aware of any thread-related
problems in malloc, in particular not anything that would cause the
program described above to fail.  That also holds for glicbc-2.2.2,
even though that's _really_ old now.

Regards,
Wolfram.