Problem with atomicity of TIF for parisc ?

Do you have a testcase? I'm still seeing the occasional thread test fail in the GCC testsuite fail. These are usually timeouts. I'm not seeing hung processes leading to a kernel crash anymore ;)

I also see output from a test truncated occasionally when a test generates a lot of output.

The bb test clearly isn't atomic...

Is disabling interrupts sufficient on a SMP system?

Dave

J. David Anglin                                  [email protected]
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

No. :/

Hi Dave,

Randolph pointed me once to the thread_test2.c testprogram in parisc's 'userspace' CVS repository. Source is attached, but I haven't tested it since a few weeks...

Me neither.

Helge

/* gcc -Wall -O2 -o thread_test2 thread_test2.c -lpthread */ #include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <pthread.h>
#include <sys/types.h>
#include <sys/wait.h>

pthread_cond_t started, terminated;
pthread_mutex_t started_lock, terminated_lock;
pthread_t thread;

static void *threadfunc(void *data)

{
	int *count = (int *)data;
	pid_t pid;
	int status;

	printf("Thread %d started\n", *count);

	pthread_cond_signal(&started);
	pid = fork();
	switch (pid) {
		case 0: /* Child */
			printf("Process %d started\n", *count);
			sleep(1);
			exit(10);
			break;

		case -1: /* Error */
			perror("fork");
			return NULL;
	}

	printf("Waiting on pid %d\n", pid);
	pid = waitpid(-1, &status, 0);
	if (pid < 0) {
		perror("waitpid");
		return NULL;
	}

	printf("Thread %d terminated (%d)\n", *count, WEXITSTATUS(status));
	pthread_cond_signal(&terminated);

	pthread_exit(NULL);

	return 0;
}

int main(int argc, char **argv)

{
	int count, ret;

	for (count = 1; ; count++) {
		ret = pthread_cond_init(&started, NULL);
		if (ret) {
			perror("pthread_cond_init(started)");
			break;
		}

		ret = pthread_mutex_init(&started_lock, NULL);
		if (ret) {
			perror("pthread_mutex_init(started_lock)");
			break;
		}

		ret = pthread_cond_init(&terminated, NULL);
		if (ret) {
			perror("pthread_cond_init(terminated)");
			break;
		}

		ret = pthread_mutex_init(&terminated_lock, NULL);
		if (ret) {
			perror("pthread_mutex_init(terminated_lock)");
			break;
		}

		ret = pthread_create(&thread, NULL, threadfunc, &count);
		if (ret) {
			perror("pthread_create");
			break;
		}

		pthread_mutex_lock(&started_lock);
		do {
			ret = pthread_cond_wait(&started, &started_lock);
		} while (ret != 0);
		pthread_mutex_unlock(&started_lock);
		
		pthread_mutex_destroy(&started_lock);
		pthread_cond_destroy(&started);

		pthread_mutex_lock(&terminated_lock);
		do {
			ret = pthread_cond_wait(&terminated, &terminated_lock);
		} while (ret != 0);
		pthread_mutex_unlock(&terminated_lock);

		pthread_mutex_destroy(&terminated_lock);
		pthread_cond_destroy(&terminated);
	}

	return 0;
}

_______________________________________________
parisc-linux mailing list
[email protected]
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux

Hi Kyle,

Independend of the fact if irq-disabling is enough on SMP or not. The question still remains: May this be one of the reasons we see the hangs ?

Helge
_______________________________________________
parisc-linux mailing list
[email protected]
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux