https://bugs.kde.org/show_bug.cgi?id=451327
Bug ID: 451327
Summary: Potential false positive when posting to stack
allocated semaphore
Product: valgrind
Version: 3.18.1
Platform: Compiled Sources
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: helgrind
Assignee: jsew...@acm.org
Reporter: god...@gmail.com
Target Milestone: ---
Created attachment 147408
--> https://bugs.kde.org/attachment.cgi?id=147408&action=edit
semtest.c that shows the potential false positive
SUMMARY
Helgrind 3.18.1 appears to flag what I believe to be a false positive when a
stack-allocated semaphore is signaled to and the stack space on which it is
allocated is then reused.
STEPS TO REPRODUCE
1. Compile the attached program with gcc -g -pthread semtest.c -o semtest (gcc
9.4.0 on Ubuntu 20.04 with GNU libc 2.2.5 I believe)
2. Run it with valgrind --tool=helgrind ./semtest
OBSERVED RESULT
==2965601== Helgrind, a thread error detector
==2965601== Copyright (C) 2007-2017, and GNU GPL'd, by OpenWorks LLP et al.
==2965601== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
==2965601== Command: ./semtest
==2965601==
==2965601== ---Thread-Announcement------------------------------------------
==2965601==
==2965601== Thread #1 is the program's root thread
==2965601==
==2965601== ---Thread-Announcement------------------------------------------
==2965601==
==2965601== Thread #2 was created
==2965601== at 0x49BC152: clone (clone.S:71)
==2965601== by 0x48812EB: create_thread (createthread.c:101)
==2965601== by 0x4882E0F: pthread_create@@GLIBC_2.2.5 (pthread_create.c:817)
==2965601== by 0x4846FBA: pthread_create_WRK (hg_intercepts.c:445)
==2965601== by 0x48480BD: pthread_create@* (hg_intercepts.c:478)
==2965601== by 0x1092C9: use_stack_allocated_semaphore (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601== by 0x10931C: main (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601==
==2965601== ----------------------------------------------------------------
==2965601==
==2965601== Possible data race during write of size 8 at 0x1FFEFFFDE0 by thread
#1
==2965601== Locks held: none
==2965601== at 0x484BCD9: memset (vg_replace_strmem.c:1358)
==2965601== by 0x10923F: touch_the_stack (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601== by 0x109317: main (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601==
==2965601== This conflicts with a previous read of size 8 by thread #2
==2965601== Locks held: none
==2965601== at 0x488CA37: sem_post@@GLIBC_2.2.5 (sem_post.c:44)
==2965601== by 0x4847ED1: sem_post_WRK (hg_intercepts.c:2993)
==2965601== by 0x4848AB6: sem_post@* (hg_intercepts.c:3013)
==2965601== by 0x109272: post_thread (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601== by 0x48471B2: mythread_wrapper (hg_intercepts.c:406)
==2965601== by 0x4882608: start_thread (pthread_create.c:477)
==2965601== by 0x49BC162: clone (clone.S:95)
==2965601== Address 0x1ffefffde0 is on thread #1's stack
==2965601== in frame #1, created by touch_the_stack (???:)
==2965601==
==2965601== ----------------------------------------------------------------
==2965601==
==2965601== Possible data race during write of size 4 at 0x1FFEFFFDE8 by thread
#1
==2965601== Locks held: none
==2965601== at 0x484BCDC: memset (vg_replace_strmem.c:1358)
==2965601== by 0x10923F: touch_the_stack (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601== by 0x109317: main (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601==
==2965601== This conflicts with a previous read of size 4 by thread #2
==2965601== Locks held: none
==2965601== at 0x488CA34: sem_post@@GLIBC_2.2.5 (sem_post.c:36)
==2965601== by 0x4847ED1: sem_post_WRK (hg_intercepts.c:2993)
==2965601== by 0x4848AB6: sem_post@* (hg_intercepts.c:3013)
==2965601== by 0x109272: post_thread (in
/home/gback/cs3214/dutchblitz.sem/tests/semtest)
==2965601== by 0x48471B2: mythread_wrapper (hg_intercepts.c:406)
==2965601== by 0x4882608: start_thread (pthread_create.c:477)
==2965601== by 0x49BC162: clone (clone.S:95)
==2965601== Address 0x1ffefffde8 is on thread #1's stack
==2965601== in frame #1, created by touch_the_stack (???:)
==2965601==
==2965601==
==2965601== Use --history-level=approx or =none to gain increased speed, at
==2965601== the cost of reduced accuracy of conflicting-access information
==2965601== For lists of detected and suppressed errors, rerun with: -s
==2965601== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 0 from 0)
EXPECTED RESULT
There should be no errors shown.
SOFTWARE/OS VERSIONS
- Valgrind 3.18.1 (built from source)
- gcc 9.4.0
- Ubuntu 20.04.4
ADDITIONAL INFORMATION
The error is not shown if the semaphore is dynamically allocated.
--
You are receiving this mail because:
You are watching all bug changes.
