Skip to contentSkip to conversation

Add a multi-process IPC test

PATCH In review Dan Smith 4 participants 7 messages Apr 9, 2009 – Apr 12, 2009 3 messages not in the archive Kbuild

lore
Share & tools
lore

Dan Smith

author
#0 lore raw

PATCH 1/1 Add a multi-process IPC test

This test forks off several children in various states to ensure that the IPC namespace information is properly restored after restart.

---
 Makefile           |    2 +-
 tst_ipcshm_multi.c |  266 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 267 insertions(+), 1 deletions(-)
 create mode 100644 tst_ipcshm_multi.c
Makefile +1 -1
index c847a0c..2da03bb 100644
@@ -19,7 +19,7 @@ CFLAGS += -g $(WARNS) $(PATHS) $(DEBUG)
 
 PROGS =	self ckpt rstr mktree
 TESTS=	tst_onetask tst_multitask tst_bigmem tst_pipes tst_pipes2 tst_shmem \
-	tst_ipcshm tst_ipcmsq
+	tst_ipcshm tst_ipcmsq tst_ipcshm_multi
 
 LDLIBS = -lm
 
tst_ipcshm_multi.c new file +266 -0

266 lines not on this page · raw message

Signature · 1 line
1.6.0.3

A message that is not in the archive (sent off-list or not yet received): 1239299170-14827-1-git-send-email-danms-r/[email protected]

author
#2 lore raw

DS> This test forks off several children in various states to ensure DS> that the IPC namespace information is properly restored after DS> restart.

I should have mentioned that this test behaves as expected with the latest IPC patches from Oren. If I run it and wait about ten seconds to checkpoint it, on restart everything gets hooked back up properly.

Signature · 3 lines
Dan Smith
IBM Linux Technology Center
email: danms-r/[email protected]
#3 lore raw
In reply to a message not in the archive
Quoted 2 lines · Dan Smith

This test forks off several children in various states to ensure that the IPC namespace information is properly restored after restart.

Thanks, Dan. I've got a few other simple c/r tests collected which I want to automate in the next few days so one can just run 'check_cr' and get a list of pass/fails. I'll also add some tests which Nathan had written for IPC. Hmm, I suppose the easiest thing for now will be to put a tarball on lxc.sf.net.

-serge

A message that is not in the archive (sent off-list or not yet received): 20090409185159.GA32610-r/[email protected]

author
#5 lore raw

SH> Thanks, Dan. I've got a few other simple c/r tests collected SH> which I want to automate in the next few days so one can just run SH> 'check_cr' and get a list of pass/fails. I'll also add some tests SH> which Nathan had written for IPC. Hmm, I suppose the easiest SH> thing for now will be to put a tarball on lxc.sf.net.

Okay. I should probably add a little more logic to make this test validate the result (instead of requiring an eye-parse of the log) so it can be automated.

Signature · 3 lines
Dan Smith
IBM Linux Technology Center
email: danms-r/[email protected]
#6 lore raw
In reply to a message not in the archive
Quoted 9 lines · Serge E. Hallyn
Quoted 2 lines · Dan Smith

This test forks off several children in various states to ensure that the IPC namespace information is properly restored after restart.

Thanks, Dan. I've got a few other simple c/r tests collected which I want to automate in the next few days so one can just run 'check_cr' and get a list of pass/fails. I'll also add some tests which Nathan had written for IPC. Hmm, I suppose the easiest thing for now will be to put a tarball on lxc.sf.net.

I include the c/r tests with the c/r userland tools (ckpt, mktree etc). That's a natural repository to collect/hold them.

Oren.

Quoted 7 lines

-serge
_______________________________________________
Containers mailing list Containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
https://lists.linux-foundation.org/mailman/listinfo/containers

A message that is not in the archive (sent off-list or not yet received): [email protected]

#8 lore raw
Quoted 7 lines · Oren Laadan
Quoted 4 lines · Serge E. Hallyn
Quoted 2 lines · Dan Smith

This test forks off several children in various states to ensure that the IPC namespace information is properly restored after restart.

Quoted 8 lines
Quoted 5 lines

Thanks, Dan. I've got a few other simple c/r tests collected which I want to automate in the next few days so one can just run 'check_cr' and get a list of pass/fails. I'll also add some tests which Nathan had written for IPC. Hmm, I suppose the easiest thing for now will be to put a tarball on lxc.sf.net.

I include the c/r tests with the c/r userland tools (ckpt, mktree etc). That's a natural repository to collect/hold them.

The natural place is ltp :) Meanwhile I wanted to automate a few tests, but if you wanted to add those to your user-cr tree that'd be great. I don't actually *want* them on lxc.sf.net, just had nowhere else to put them.

thanks,
-serge

#9 lore raw
In reply to a message not in the archive

Thanks, added to user-cr.git for next version.

Oren.

Quoted 296 lines · Dan Smith

This test forks off several children in various states to ensure that the IPC namespace information is properly restored after restart.

---
 Makefile           |    2 +-
 tst_ipcshm_multi.c |  266 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 267 insertions(+), 1 deletions(-)
 create mode 100644 tst_ipcshm_multi.c
Makefile +1 -1
index c847a0c..2da03bb 100644
@@ -19,7 +19,7 @@ CFLAGS += -g $(WARNS) $(PATHS) $(DEBUG)
 
 PROGS =	self ckpt rstr mktree
 TESTS=	tst_onetask tst_multitask tst_bigmem tst_pipes tst_pipes2 tst_shmem \
-	tst_ipcshm tst_ipcmsq
+	tst_ipcshm tst_ipcmsq tst_ipcshm_multi
 
 LDLIBS = -lm
 
tst_ipcshm_multi.c new file +266 -0
new file mode 100644
index 0000000..1ef31f7
@@ -0,0 +1,266 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <errno.h>
+#include <sys/ipc.h>
+#include <sys/shm.h>
+
+#include <linux/sched.h>
+#include <sched.h>
+
+#define OUTFILE  "/tmp/cr-test.out"
+#define SEG_SIZE (20 * 4096)
+#define DELAY 20
+
+int attach(unsigned char **seg, int num)
+{
+	int id;
+
+	id = shmget(123, SEG_SIZE, 0700|IPC_EXCL);
+	if (id < 0) {
+		printf("[CHILD %i] shmget: %m\n", num);
+		return -1;
+	}
+
+	*seg = shmat(id, NULL, 0);
+	if (!*seg) {
+		printf("[CHILD %i] shmat: %m\n", num);
+		return -1;
+	}
+
+	printf("[CHILD %i] Attached\n", num);
+
+	return 0;
+}
+
+int validate(unsigned char *seg, int num)
+{
+	int i;
+
+	for (i = 1; i < SEG_SIZE; i++) {
+		if (seg[i] != (unsigned char)(i % 256)) {
+			printf("[CHILD %i] Mismatch at %hhu: %hhu\n",
+			       num, i, seg[i]);
+			return -1;
+		}
+	}
+
+	printf("[CHILD %i] Validated\n", num);
+
+	return 0;
+}
+
+void track(unsigned char *seg, int num)
+{
+	int i;
+
+	for (i = 0; i < 20; i++) {
+		printf("[CHILD %i] Seg[0]: %i\n", num, seg[0]);
+		sleep(1);
+	}
+}
+
+/*
+ * Attach to the shared segment *before* the checkpoint and then track
+ * the changing seg[0] position afterwards.
+ */
+int child1(void)
+{
+	unsigned char *seg;
+	int num = 1;
+
+	printf("[CHILD %i] Running\n", num);
+
+	sleep(1); /* Allow master to finish initializing the array */
+
+	if (attach(&seg, num))
+		return -1;
+
+	if (validate(seg, num))
+		return -1;
+
+	sleep(DELAY - 1); /* Wait until after the checkpoint */
+
+	track(seg, num);
+
+	return 0;
+}
+
+/*
+ * Attempt to attach to the shared segment *after* the checkpoint to
+ * verify that things are hooked up properly after restart by tracking
+ * seg[0].
+ */
+int child2(void)
+{
+	unsigned char *seg;
+	int num = 2;
+  
+	printf("[CHILD %i] Running\n", num);
+
+	sleep(DELAY); /* Wait until after the checkpoint */
+
+	if (attach(&seg, num))
+		return -1;
+
+	if (validate(seg, num))
+		return -1;
+
+	track(seg, num);
+
+	return 0;
+}
+
+int child4(void);
+
+/*
+ * Detach from the parent's IPC namespace and verify that:
+ * 1. We can't attach to the parent's segment
+ * 2. We can create our own of the same ID without conflict
+ * 3. We do not see the parent's changing seg[0] after restart
+ */
+int child3(void)
+{
+	unsigned char *seg;
+	int num = 3;
+
+	if (unshare(CLONE_NEWIPC) != 0) {
+		printf("[CHILD %i] unshare(CLONE_NEWIPC): %m", num);
+		return -1;
+	}
+
+	if (fork() == 0)
+		return child4();
+
+	printf("[CHILD %i] Running (new IPC NS)\n", num);
+
+	printf("[CHILD %i] Attempting parent attach (should fail)\n", num);
+	if (!attach(&seg, num)) {
+		printf("[CHILD %i] Attached to parent's IPC!\n", num);
+		return -1;
+	}
+
+	if (shmget(123, SEG_SIZE, 0700|IPC_CREAT|IPC_EXCL) < 0) {
+		printf("[CHILD %i] Failed to create shm in namespace: %m\n",
+		       num);
+		return -1;
+	}
+
+	if (attach(&seg, num))
+		return -1;
+
+	seg[0] = 123;
+
+	sleep(DELAY); /* Wait until after checkpoint, then attach */
+
+	track(seg, num);
+  
+	return 0;
+}
+
+/*
+ * This child is forked from child3 under the new IPC namespace.
+ * Verify that post-restart, we do not see the changing seg[0]
+ */
+int child4(void)
+{
+	unsigned char *seg;
+	int num = 4;
+
+	printf("[CHILD %i] Running (new IPC NS)\n", num);
+
+	sleep(DELAY); /* Wait past my parent creating the IPC */
+
+	if (attach(&seg, num))
+		return -1;
+
+	track(seg, num);
+
+	return 0;
+}
+
+void shm_destroy(void)
+{
+	int id;
+
+	id = shmget(123, SEG_SIZE, 0700);
+	if (id < 0)
+		return;
+
+	if (shmctl(id, IPC_RMID, NULL) == 0)
+		printf("Deleted SHM %i\n", id);
+}
+
+int main(int argc, char *argv[])
+{
+	int id;
+	unsigned char *seg;
+	int i;
+	int pid1, pid2, pid3;
+#ifndef TEST
+	FILE *file;
+
+	close(0);
+	close(1);
+	close(2);
+
+	unlink(OUTFILE);
+	file = fopen(OUTFILE, "w+");
+	if (!file) {
+		perror("open");
+		exit(1);
+	}
+	if (dup2(0,2) < 0) {
+		perror("dup2");
+		exit(2);
+	}
+	stdout = file;
+	setlinebuf(file);
+#endif
+	shm_destroy();
+
+	pid1 = fork();
+	if (pid1 == 0)
+		return child1();
+
+	pid2 = fork();
+	if (pid2 == 0)
+		return child2();
+
+	pid3 = fork();
+	if (pid3 == 0)
+		return child3();
+
+	id = shmget(123, SEG_SIZE, 0700|IPC_CREAT|IPC_EXCL);
+	if (id < 0) {
+		perror("shmget");
+		exit(1);
+	}
+
+	seg = (unsigned char *) shmat(id, NULL, 0);
+	if(seg == ((void *) -1)) {
+		perror("shmat");
+		exit(1);
+	}
+	
+	for (i = 0; i < SEG_SIZE; i++)
+		seg[i] = i;
+
+	printf("[MSTER] Waiting for checkpoint\n");
+	sleep(DELAY);
+	printf("[MSTER] Woke\n");
+
+	for (i = 0; i < 15; i++) {
+		seg[0] = i;
+		sleep(1);
+	}
+
+	if (shmdt(seg) < 0)
+		perror("shmdt");
+
+	shm_destroy();
+
+	printf("[MSTER] Completed\n");
+
+	return 0;
+}