https://bugs.kde.org/show_bug.cgi?id=484480
Bug ID: 484480
Summary: False positives when using sem_trywait
Classification: Developer tools
Product: valgrind
Version: 3.22.0
Platform: Ubuntu
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: helgrind
Assignee: jsew...@acm.org
Reporter: jmcgee01...@gmail.com
Target Milestone: ---
SUMMARY
Using sem_trywait leads to false positives, even when it is guaranteed to have
succeeded by the time that the access is made.
STEPS TO REPRODUCE
Compile the following code with `gcc -o test test.c -pthread`, then run with
`valgrind --tool=helgrind ./test`
```
#include <stdio.h>
#include <semaphore.h>
#include <pthread.h>
static char* result;
static sem_t sem;
static void* func(void* data)
{
result = "Finished!";
sem_post(&sem);
return NULL;
}
int main()
{
sem_init(&sem, 0, 0);
pthread_t tid;
pthread_create(&tid, NULL, func, NULL);
while (sem_trywait(&sem))
; // do nothing but keep retrying
/* The above loop could be replaced this instead:
* if (sem_trywait(&sem))
* sem_wait(&sem);
*/
printf("%s\n", result);
}
```
OBSERVED RESULT
Helgrind reports the following:
==787273== Helgrind, a thread error detector
==787273== Copyright (C) 2007-2017, and GNU GPL'd, by OpenWorks LLP et al.
==787273== Using Valgrind-3.22.0 and LibVEX; rerun with -h for copyright info
==787273== Command: ./test
==787273==
==787273== ---Thread-Announcement------------------------------------------
==787273==
==787273== Thread #1 is the program's root thread
==787273==
==787273== ---Thread-Announcement------------------------------------------
==787273==
==787273== Thread #2 was created
==787273== at 0x4980F23: clone (in /usr/lib64/libc.so.6)
==787273== by 0x4981CDE: __clone_internal (in /usr/lib64/libc.so.6)
==787273== by 0x48FC834: create_thread (in /usr/lib64/libc.so.6)
==787273== by 0x48FD339: pthread_create@@GLIBC_2.34 (in
/usr/lib64/libc.so.6)
==787273== by 0x4852D1C: pthread_create_WRK (hg_intercepts.c:445)
==787273== by 0x4854239: pthread_create@* (hg_intercepts.c:478)
==787273== by 0x4011C4: main (in /home/ugrads/majors/colinm25/test)
==787273==
==787273== ----------------------------------------------------------------
==787273==
==787273== Possible data race during read of size 8 at 0x404080 by thread #1
==787273== Locks held: none
==787273== at 0x4011D4: main (in /home/ugrads/majors/colinm25/test)
==787273==
==787273== This conflicts with a previous write of size 8 by thread #2
==787273== Locks held: none
==787273== at 0x401172: func (in /home/ugrads/majors/colinm25/test)
==787273== by 0x4852F16: mythread_wrapper (hg_intercepts.c:406)
==787273== by 0x48FCC01: start_thread (in /usr/lib64/libc.so.6)
==787273== by 0x4980F33: clone (in /usr/lib64/libc.so.6)
==787273== Address 0x404080 is 0 bytes inside data symbol "result"
==787273==
Finished!
==787273==
==787273== Use --history-level=approx or =none to gain increased speed, at
==787273== the cost of reduced accuracy of conflicting-access information
==787273== For lists of detected and suppressed errors, rerun with: -s
==787273== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 0 from 0)
EXPECTED RESULT
Helgrind should not report any warnings.
SOFTWARE/OS VERSIONS
Tested on CentOS Stream 9 with Valgrind 3.22.0 and Ubuntu Server 20.04 with
Valgrind 3.15.0.
--
You are receiving this mail because:
You are watching all bug changes.
