RE: 2.4.31 TLB corruption

From: Smarduch Mario-CMS063 <>
Date: 2005-06-23 06:37:48
Attached is a patch produced by HP LOSL its not quite against
2.4.31  but I it probably should apply cleanly, the related code
is indentical to latest 2.4 kernel. If needed I could inquire
for them to develop a 2.6 patch (I myself don't have 2.6 kernel up)
please ask if needed. 

To reproduce the RID range was shrunk to lim/maxctx=1023/2048.
on an 8-way about 180 copies with an argument of 5 was ran.
We'd also appreciate feedback on the patch itself. There is stil
another scenario on NUMA systems I'm concerned about I'll ask
on the maillist when time permits.

- mario

#include <stdio.h>
#include <errno.h>
#include <pthread.h>
#include <string.h>
#include <unistd.h>
#include <stdlib.h>
#include <sys/wait.h>

volatile unsigned long udummy1, udummy2, udummy3;
volatile unsigned long idummy1=100, idummy2=200, idummy3=300;

void * worker(void *self) {
  int x = -1;
  FILE *pp = 0;
  long junk[8192];
  char buf[1024];

    x = -1;
    errno = 0;
    //printf("Thread: %u\n", getpid());
    pp = popen("ls", "r");
    x = errno;
    if(pp) {
      memset(buf, 0, sizeof(buf));
      while(fgets(buf, sizeof(buf)-1, pp) != 0) {

    else printf("popen failed: %s\n", strerror(errno));
    //printf("Thread Exit: %ud\n", getpid());

int main(int argc, char **argv) {
  pthread_t pt[512];
  pthread_t pthr;
  int x = 0;
  int thrcnt=5;

  if(argc > 1)

  while(1) {
        //printf("Creating Thread...");
        for (x = 0; x < thrcnt; x++)
                pthread_create(&(pt[x]), 0, worker, (void *)&x);
        for (x = 0; x < thrcnt; x++)
                pthread_join( pt[x], NULL);
  return 0;

-----Original Message-----
From: David Mosberger [] 
Sent: Wednesday, June 22, 2005 12:56 PM
To: Smarduch Mario-CMS063
Subject: RE: 2.4.31 TLB corruption

>>>>> On Tue, 21 Jun 2005 16:22:50 -0500, Smarduch Mario-CMS063 
>>>>> <> said:

  Mario> FYI, shrinking the RID range does reproduce the problem
  Mario> easily. Its also reproducable on 2.6.7.

That's progress.  Could you provide the kernel patch and description of minimal test-case to reproduce the problem?



To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to
More majordomo info at

Received on Wed Jun 22 16:45:56 2005

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:39 EST