Hi,
Please see attached a patch from [EMAIL PROTECTED] for fixing times03
testcase.
Regards--
Subrata
The times03 testcase assumes that the time spent in userspace would be more than 1/USER_HZ seconds.
But there is no such guarantee with the testcase design. Especially for CPU's having higher frequency,
this is not going to give the expcted result.
The fix makes sure that the testcase spends some time in userspace using an alarm() to determine, the
to be spent in the userspace.
Signed-off-by : Suzuki K P <[EMAIL PROTECTED]>
Index: times/times03.c
===================================================================
--- times.orig/times03.c 2005-07-12 03:59:10.000000000 +0530
+++ times/times03.c 2007-08-01 19:40:44.000000000 +0530
@@ -52,11 +52,16 @@
#include <time.h>
#include <test.h>
#include <usctest.h>
+#include <signal.h>
char *TCID = "times03";
int TST_TOTAL = 1;
extern int Tst_count;
int exp_enos[]={0};
+int timeout; /* Did we timeout in alarm() ? */
+
+void work(void);
+void sighandler(int signal, siginfo_t *info, void *uc);
void setup(void);
void cleanup(void);
@@ -67,21 +72,40 @@
struct tms buf1, buf2;
time_t start_time, end_time;
- int i, pid2, status, fail=0;
+ int pid2, status, fail=0;
+ struct sigaction sa;
/* parse standard options */
if ((msg = parse_opts(argc, argv, (option_t *)NULL, NULL)) !=
(char *)NULL) {
tst_brkm(TBROK, cleanup, "OPTION PARSING ERROR - %s", msg);
- /*NOTREACHED*/
+ /*NOT REACHED*/
}
setup();
+ /*
+ * We spend time in userspace using the following mechanism :
+ * Setup an alarm() for 3 secs and do some simple loop operations
+ * until we get the signal. This makes the test independent of
+ * processor speed.
+ */
+ sa.sa_sigaction = sighandler;
+ sigemptyset(&sa.sa_mask);
+ sa.sa_flags = SA_SIGINFO;
+
+ if (sigaction(SIGALRM, &sa, NULL) < 0) {
+ tst_brkm(TBROK, cleanup, "Sigaction failed !\n");
+ /* NOT REACHED */
+ }
+
+ timeout = 0;
+ alarm(3);
+
+ work();
- for (i = 0; i < 1000000; i++);
/*
- * At least some CPU time must be used. This is
+ * At least some CPU time must be used in system space. This is
* achieved by executing the times(2) call for
* atleast 5 secs. This logic makes it independant
* of the processor speed.
@@ -135,9 +159,17 @@
tst_brkm(TFAIL, cleanup, "Fork failed");
/*NOTREACHED*/
} else if (pid2 == 0) {
- for (i = 0; i < 2000000; i++);
+
+ /* Spend some cycles in userspace */
+
+ timeout = 0;
+ alarm(3);
+
+ work();
+
+
/*
- * Atleast some CPU time must be used
+ * Atleast some CPU system ime must be used
* even in the child process (thereby
* making it independent of the
* processor speed). In fact the child
@@ -161,6 +193,7 @@
}
exit(0);
}
+
waitpid(pid2, &status, 0);
if (WEXITSTATUS(status) != 0) {
tst_resm(TFAIL, "Call to times(2) "
@@ -212,6 +245,35 @@
}
/*
+ * sighandler
+ * Set the timeout to indicate we timed out in the alarm().
+ */
+
+void sighandler (int signal, siginfo_t *info, void *uc)
+{
+ if (signal == SIGALRM)
+ timeout = 1;
+ else
+ tst_brkm (TBROK, cleanup, "Unexpected signal %d\n", signal);
+}
+
+/*
+ * work
+ * Do some work in user space, until we get a timeout.
+ */
+
+void work(void)
+{
+ int i, j, k;
+
+ while (!timeout)
+ for (i = 0; i < 10000; i ++)
+ for (j = 0; j < 100; j ++)
+ k = i * j;
+ timeout = 0;
+}
+
+/*
* setup()
* performs all ONE TIME setup for this test
*/
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
