我正在检查我的程序是否存在线程错误，并使用这个最简单的代码重现了我得到的helgrind错误：
int main() {
int num = 24;
#pragma omp parallel for firstprivate(num)
for (int i= 0; i <num; ++i) {
int test = 0;
}
}
使用g++-4.7 -fopenmp -g test.cpp -o test
这会产生300行错误＆＃34;可能的数据竞争＆＃34;。我在这里缺乏理解是什么？我做错了什么？
以下是错误日志的示例：
1 ==11990== Helgrind, a thread error detector
2 ==11990== Copyright (C) 2007-2013, and GNU GPL'd, by OpenWorks LLP et al.
3 ==11990== Using Valgrind-3.9.0 and LibVEX; rerun with -h for copyright info
4 ==11990== Command: /home/windenergy/test
5 ==11990== Parent PID: 10832
6 ==11990==
7
8 parse_type_DIE: confused by:
9 <2><197>: DW_TAG_typedef
10 DW_AT_type : <1f4>
11
12 --11990-- WARNING: Serious error when reading debug info
13 --11990-- When reading debug info from /home/windenergy/test:
14 --11990-- parse_type_DIE: confused by the above DIE
15 ==11990== ---Thread-Announcement------------------------------------------
16 ==11990==
17 ==11990== Thread #1 is the program's root thread
18 ==11990==
19 ==11990== ---Thread-Announcement------------------------------------------
20 ==11990==
21 ==11990== Thread #24 was created
22 ==11990== at 0x5ACE05E: clone (in /lib64/libc-2.11.3.so)
23 ==11990== by 0x57DC990: do_clone (in /lib64/libpthread-2.11.3.so)
24 ==11990== by 0x57DCF77: pthread_create@@GLIBC_2.2.5 (in /lib64/libpthread-2.11.3.so)
25 ==11990== by 0x4C2DB78: pthread_create_WRK (hg_intercepts.c:269)
26 ==11990== by 0x4C2DC89: pthread_create@* (hg_intercepts.c:300)
27 ==11990== by 0x53BA56D: ??? (in /usr/lib64/libgomp.so.1.0.0)
28 ==11990== by 0x400816: main (test.cpp:4)
29 ==11990==
30 ==11990== ----------------------------------------------------------------
31 ==11990==
32 ==11990== Possible data race during write of size 4 at 0x5F750C0 by thread #1
33 ==11990== Locks held: none
34 ==11990== at 0x53BB85B: ??? (in /usr/lib64/libgomp.so.1.0.0)
35 ==11990== by 0x53BA5B0: ??? (in /usr/lib64/libgomp.so.1.0.0)
36 ==11990== by 0x400816: main (test.cpp:4)
37 ==11990==
38 ==11990== This conflicts with a previous read of size 4 by thread #24
39 ==11990== Locks held: none
40 ==11990== at 0x53BB8C6: ??? (in /usr/lib64/libgomp.so.1.0.0)
41 ==11990== by 0x53BA0B8: ??? (in /usr/lib64/libgomp.so.1.0.0)
42 ==11990== by 0x4C2DD12: mythread_wrapper (hg_intercepts.c:233)
43 ==11990== by 0x57DD7F5: start_thread (in /lib64/libpthread-2.11.3.so)
44 ==11990== by 0x5ACE09C: clone (in /lib64/libc-2.11.3.so)
45 ==11990==
46 ==11990== Address 0x5F750C0 is 128 bytes inside a block of size 192 alloc'd
47 ==11990== at 0x4C2819E: malloc (vg_replace_malloc.c:291)
48 ==11990== by 0x53B6038: ??? (in /usr/lib64/libgomp.so.1.0.0)
49 ==11990== by 0x53BA7DF: ??? (in /usr/lib64/libgomp.so.1.0.0)
50 ==11990== by 0x400816: main (test.cpp:4)
51 ==11990==
52 ==11990== ----------------------------------------------------------------
53 ==11990==
54 ==11990== Possible data race during write of size 4 at 0x5F75084 by thread #1
55 ==11990== Locks held: none
56 ==11990== at 0x53BB869: ??? (in /usr/lib64/libgomp.so.1.0.0)
57 ==11990== by 0x53BA5B0: ??? (in /usr/lib64/libgomp.so.1.0.0)
58 ==11990== by 0x400816: main (test.cpp:4)
59 ==11990==
60 ==11990== This conflicts with a previous read of size 4 by thread #24
61 ==11990== Locks held: none
62 ==11990== at 0x53BB8C0: ??? (in /usr/lib64/libgomp.so.1.0.0)
63 ==11990== by 0x53BA0B8: ??? (in /usr/lib64/libgomp.so.1.0.0)
64 ==11990== by 0x4C2DD12: mythread_wrapper (hg_intercepts.c:233)
65 ==11990== by 0x57DD7F5: start_thread (in /lib64/libpthread-2.11.3.so)
66 ==11990== by 0x5ACE09C: clone (in /lib64/libc-2.11.3.so)
67 ==11990==
68 ==11990== Address 0x5F75084 is 68 bytes inside a block of size 192 alloc'd
69 ==11990== at 0x4C2819E: malloc (vg_replace_malloc.c:291)
70 ==11990== by 0x53B6038: ??? (in /usr/lib64/libgomp.so.1.0.0)
71 ==11990== by 0x53BA7DF: ??? (in /usr/lib64/libgomp.so.1.0.0)
72 ==11990== by 0x400816: main (test.cpp:4)
73 ==11990==