https://bugs.kde.org/show_bug.cgi?id=397670
Bug ID: 397670
Summary: [Helgrind] False positive race detected with
sig_atomic_t
Product: valgrind
Version: 3.13.0
Platform: Compiled Sources
OS: Linux
Status: UNCONFIRMED
Severity: normal
Priority: NOR
Component: helgrind
Assignee: jsew...@acm.org
Reporter: nikita.leont...@gmail.com
Target Milestone: ---
Helgrind 3.13.0 detects false positive data race with sig_atomic_t in libev
4.19.
In ev_vars.h we have pipe_write_skipped defined as:
#define VARx(type,name) VAR(name, type name)
VARx(EV_ATOMIC_T, pipe_write_skipped)
ev.h:
#ifndef EV_ATOMIC_T
# include <signal.h>
# define EV_ATOMIC_T sig_atomic_t volatile
#endif
Helgrind result:
==983== Possible data race during write of size 4 at 0x5BFD084 by thread #1
==983== Locks held: none
==983== at 0x50001C0: evpipe_write (ev.c:2237)
==983== by 0x5005515: ev_async_send (ev.c:4706)
...
==983== This conflicts with a previous read of size 4 by thread #3
==983== Locks held: none
==983== at 0x5002F5C: ev_run (ev.c:3407)
==983== by 0x5261607: ev_loop (ev.h:828)
...
==983== Address 0x5bfd084 is 228 bytes inside a block of size 768 alloc'd
==983== at 0x483D4FC: malloc (vg_replace_malloc.c:298)
==983== by 0x483FDBB: realloc (vg_replace_malloc.c:785)
==983== by 0x4FFF4FD: ev_realloc_emul (ev.c:1518)
==983== by 0x4FFF54D: ev_realloc (ev.c:1535)
==983== by 0x5001DBD: ev_loop_new (ev.c:2878)
...
evpipe_write (ev.c:2237) is:
inline_speed void
evpipe_write (EV_P_ EV_ATOMIC_T *flag)
{
...
pipe_write_skipped = 1;
...
}
and ev_run (ev.c:3407) is:
int
ev_run (EV_P_ int flags)
{
...
if (expect_true (!(flags & EVRUN_NOWAIT || idleall || !activecnt ||
pipe_write_skipped)))
...
}
so both access performed on sig_atomic_t that's is thread safe.
--
You are receiving this mail because:
You are watching all bug changes.
