It is very hard to predict runtime for fsstress. In many cases it
is useful to give test to run a reasonable time, and then kill it.
But currently there is no reliable way to kill test without leaving
running children.
This patch add sanity cleanup logic which looks follow:
- On sigterm received by parent, it resend signal to it's children
- Wait for each child to terminates
- EXTRA_SANITY: Even if parent was killed by other signal, children
will be terminated with SIGKILL to preven staled children.
So now one can simply run fsstress like this:
./fsstress -p 1000 -n999999999 -d $TEST_DIR &
PID=$!
sleep 300
kill $PID
wait $PID
Signed-off-by: Dmitry Monakhov <dmonakhov@xxxxxxxxxx>
---
aclocal.m4 | 5 +++++
configure.in | 1 +
ltp/fsstress.c | 37 ++++++++++++++++++++++++++++++++++++-
3 files changed, 42 insertions(+), 1 deletions(-)
diff --git a/aclocal.m4 b/aclocal.m4
index 168eb59..5532606 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -16,6 +16,11 @@ AC_DEFUN([AC_PACKAGE_WANT_LINUX_FIEMAP_H],
AC_SUBST(have_fiemap)
])
+AC_DEFUN([AC_PACKAGE_WANT_LINUX_PRCTL_H],
+ [ AC_CHECK_HEADERS([sys/prctl.h], [ have_prctl=true ], [ have_prctl=false ])
+ AC_SUBST(have_prctl)
+ ])
+
AC_DEFUN([AC_PACKAGE_WANT_FALLOCATE],
[ AC_MSG_CHECKING([for fallocate])
AC_TRY_LINK([
diff --git a/configure.in b/configure.in
index c697b4f..76d23e4 100644
--- a/configure.in
+++ b/configure.in
@@ -67,6 +67,7 @@ in
AC_PACKAGE_WANT_DMAPI
AC_PACKAGE_WANT_LINUX_FIEMAP_H
AC_PACKAGE_WANT_FALLOCATE
+ AC_PACKAGE_WANT_LINUX_PRCTL_H
;;
esac
diff --git a/ltp/fsstress.c b/ltp/fsstress.c
index c7001f3..133a247 100644
--- a/ltp/fsstress.c
+++ b/ltp/fsstress.c
@@ -28,7 +28,9 @@
#ifndef HAVE_ATTR_LIST
#define attr_list(path, buf, size, flags, cursor) (errno = -ENOSYS, -1)
#endif
-
+#ifdef HAVE_SYS_PRCTL_H
+#include <sys/prctl.h>
+#endif
#include <math.h>
#define XFS_ERRTAG_MAX 17
#define XFS_IDMODULO_MAX 31 /* user/group IDs (1 << x) */
@@ -209,6 +211,7 @@ int rtpct;
unsigned long seed = 0;
ino_t top_ino;
int verbose = 0;
+sig_atomic_t should_stop = 0;
void add_to_flist(int, int, int);
void append_pathname(pathname_t *, char *);
@@ -253,6 +256,11 @@ void usage(void);
void write_freq(void);
void zero_freq(void);
+void sg_handler(int signum)
+{
+ should_stop = 1;
+}
+
int main(int argc, char **argv)
{
char buf[10];
@@ -269,6 +277,7 @@ int main(int argc, char **argv)
ptrdiff_t srval;
int nousage = 0;
xfs_error_injection_t err_inj;
+ struct sigaction action;
errrange = errtag = 0;
umask(0);
@@ -429,8 +438,27 @@ int main(int argc, char **argv)
}
} else
close(fd);
+
+ setpgid(0, 0);
+ action.sa_handler = sg_handler;
+ sigemptyset(&action.sa_mask);
+ action.sa_flags = 0;
+ if (sigaction(SIGTERM, &action, 0)) {
+ perror("sigaction failed");
+ exit(1);
+ }
+
for (i = 0; i < nproc; i++) {
if (fork() == 0) {
+ action.sa_handler = SIG_DFL;
+ sigemptyset(&action.sa_mask);
+ if (sigaction(SIGTERM, &action, 0))
+ return 1;
+#ifdef HAVE_SYS_PRCTL_H
+ prctl(PR_SET_PDEATHSIG, SIGKILL);
+ if (getppid() == 1) /* parent died already? */
+ return 0;
+#endif
if (logname) {
char path[PATH_MAX];
snprintf(path, sizeof(path), "%s/%s.%d",
@@ -445,8 +473,15 @@ int main(int argc, char **argv)
return 0;
}
}
+ while (wait(&stat) > 0 && !should_stop) {
+ continue;
+ }
+ action.sa_flags = SA_RESTART;
+ sigaction(SIGTERM, &action, 0);
+ kill(-getpid(), SIGTERM);
while (wait(&stat) > 0)
continue;
+
if (errtag != 0) {
err_inj.errtag = 0;
err_inj.fd = fd;
--
1.7.1
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
