This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

segfault with pthread and detached attribute


Hi,

I had strange hangs and segfaults with a program, and it happens that
detached thread termination may segfault:

#include <windows.h>
#include <pthread.h>
#include <stdio.h>

void *f(void *foo) {
	return(NULL);
}

pthread_t t;
pthread_attr_t attr;

int main(void) {

	if (pthread_attr_init(&attr))
		perror("pthread_attr_init");
	if (pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED))
		perror("pthread_attr_setdetachstate");
	if (pthread_create(&t,&attr,f,NULL))
		perror("pthread_create");
	Sleep(10000);

	return 0;
}

segfaults with latest net cygwin:

Stack trace:
Frame     Function  Args
0022EAC8  7C802542  (00000708, 0000EA60, 000000A4, 0022EB04) WaitForSingleObject+18 (which is pop %ebp after calling WaitForSingleObjectEx)
0022EBD8  61088618  (00000000, 7C859BB0, 40010006, 00000000) sigemptyset+4616
0022ECC8  610865FC  (00000000, 00000000, 00000001, 0022ED78) writev+1821
0022ED48  61086B1B  (00000006, 00000001, 00000000, 00000000) writev+3132
0022EDF8  61086CD3  (00000B38, 00000006, 0022EE28, 61086EC7) writev+3572
0022EE08  61086CFC  (00000006, 00000000, 004B2000, 004B2000) writev+3613
0022EE28  61086EC7  (7C809F90, FFFFFFFF, 7C809F8A, 004B0000) sigfillset+343
0022EE58  610AA8B2  (00403020, 00403030, 0022EE88, 6109B63C) setstate+2562
0022EE88  6104EC66  (004B2620, FFFFFFFF, 004B2600, 00000000) lsearch+118
0022EEB8  610A11CE  (00403030, 00403020, 00401050, 00000000) pthread_getspecific+13806
0022EEE8  610844FF  (00000001, 004B25A8, 004B0090, 610A5E22) cygwin1!aclcheck+62431
0022EFD8  61004DD2  (0022EFF0, 0022F03C, 00342320, 002310A0) exit+1474
0022FF88  6100594F  (00000000, 00000000, 00000000, 00000000) dll_crt0+207
End of stack trace

I don't know why registers are not dumped. Running it through gdb or
strace just makes it work...

Adding a Sleep(1000); in the f() function before returning prevents from
the crash, so I guess f() returning fast triggers a race condition in
the pthread code.

Regards,
Samuel

Attachment: cygcheck.out
Description: Text document

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]