From: dje@google.com (Doug Evans)
To: gdb-patches@sourceware.org
Subject: [RFA] Make watchthreads.exp more robust
Date: Sun, 26 Apr 2009 19:32:00 -0000 [thread overview]
Message-ID: <20090426193156.9F5B384890@localhost> (raw)
Hi.
watchthreads.exp can errantly fail if the o/s schedules, for example,
the thread for args[0] enough times before the thread for args[1] gets
a chance to run.
Ok to check in?
2009-04-26 Doug Evans <dje@google.com>
* gdb.threads/watchthreads.c (main): Initialize args before starting
the threads. Plus formatting cleanup.
* gdb.threads/watchthreads.exp: Avoid errant failures due to
biased scheduling of one thread.
Index: watchthreads.c
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.threads/watchthreads.c,v
retrieving revision 1.6
diff -u -p -u -p -r1.6 watchthreads.c
--- watchthreads.c 3 Jan 2009 05:58:07 -0000 1.6
+++ watchthreads.c 26 Apr 2009 18:54:22 -0000
@@ -22,34 +22,49 @@
#include <stdlib.h>
#include <pthread.h>
-void *thread_function(void *arg); /* Pointer to function executed by each thread */
+void *thread_function (void *arg); /* Function executed by each thread. */
#define NUM 5
unsigned int args[NUM+1];
-int main() {
+int
+main ()
+{
int res;
pthread_t threads[NUM];
void *thread_result;
long i;
+ /* To keep the test determinative, initialize args first,
+ then start all the threads. Otherwise, the way watchthreads.exp
+ is written, we have to worry about things like threads[0] getting
+ to 29 hits of args[0] before args[1] gets changed. */
+
+ for (i = 0; i < NUM; i++)
+ {
+ /* The call to usleep is so that when the watchpoint triggers,
+ the pc is still on the same line. */
+ args[i] = 1; usleep (1); /* Init value. */
+ }
+
for (i = 0; i < NUM; i++)
{
- args[i] = 1; /* Init value. */
- res = pthread_create(&threads[i],
- NULL,
- thread_function,
- (void *) i);
+ res = pthread_create (&threads[i],
+ NULL,
+ thread_function,
+ (void *) i);
}
args[i] = 1;
thread_function ((void *) i);
- exit(EXIT_SUCCESS);
+ exit (EXIT_SUCCESS);
}
-void *thread_function(void *arg) {
+void *
+thread_function (void *arg)
+{
int my_number = (long) arg;
int *myp = (int *) &args[my_number];
@@ -59,6 +74,6 @@ void *thread_function(void *arg) {
(*myp) ++; usleep (1); /* Loop increment. */
}
- pthread_exit(NULL);
+ pthread_exit (NULL);
}
Index: watchthreads.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.threads/watchthreads.exp,v
retrieving revision 1.7
diff -u -p -u -p -r1.7 watchthreads.exp
--- watchthreads.exp 3 Jan 2009 05:58:07 -0000 1.7
+++ watchthreads.exp 26 Apr 2009 18:54:22 -0000
@@ -62,8 +62,10 @@ set args_1 0
# Watch values that will be modified by distinct threads.
gdb_test "watch args\[0\]" "Hardware watchpoint 2: args\\\[0\\\]"
gdb_test "watch args\[1\]" "Hardware watchpoint 3: args\\\[1\\\]"
+set hwwp_2_enabled 1
+set hwwp_3_enabled 1
-set init_line [expr [gdb_get_line_number "Init value"]+1]
+set init_line [gdb_get_line_number "Init value"]
set inc_line [gdb_get_line_number "Loop increment"]
set main_loc "main \\\(\\\) at .*watchthreads.c:$init_line"
set thread0_loc "thread_function \\\(arg=0x0\\\) at .*watchthreads.c:$inc_line"
@@ -114,6 +116,27 @@ for {set i 0} {$i < 30} {incr i} {
} else {
fail "threaded watch loop"
}
+
+ # If one of the watchpoints is disabled, we'd better not stop there.
+ if { !$hwwp_2_enabled && $test_flag_0 } {
+ fail "disabled hw watchpoint 2 triggered"
+ }
+ if { !$hwwp_3_enabled && $test_flag_1 } {
+ fail "disabled hw watchpoint 3 triggered"
+ }
+
+ # If we get to 10 in one of the watched locations disable it so we
+ # see some of the other watched location.
+ # 10 is chosen so we're guaranteed to come through here.
+ if { $hwwp_2_enabled && $hwwp_3_enabled } {
+ if { $args_0 >= 10 && $hwwp_2_enabled } {
+ gdb_test "disable 2" ""
+ set hwwp_2_enabled 0
+ } elseif { $args_1 >= 10 && $hwwp_3_enabled } {
+ gdb_test "disable 3" ""
+ set hwwp_3_enabled 0
+ }
+ }
}
}
next reply other threads:[~2009-04-26 19:32 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-04-26 19:32 Doug Evans [this message]
2009-04-27 12:52 ` Stan Shebs
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090426193156.9F5B384890@localhost \
--to=dje@google.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox