All of lore.kernel.org
 help / color / mirror / Atom feed
From: teigland@sourceware.org <teigland@sourceware.org>
To: cluster-devel.redhat.com
Subject: [Cluster-devel] cluster/dlm/tests/usertest dlmtest2.c
Date: 12 Dec 2006 21:17:56 -0000	[thread overview]
Message-ID: <20061212211756.1251.qmail@sourceware.org> (raw)

CVSROOT:	/cvs/cluster
Module name:	cluster
Changes by:	teigland at sourceware.org	2006-12-12 21:17:56

Modified files:
	dlm/tests/usertest: dlmtest2.c 

Log message:
	add lock_flood/unlock_flood/unlock_flood-exit commands to test doing
	large volumes of locks/unlocks

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/dlm/tests/usertest/dlmtest2.c.diff?cvsroot=cluster&r1=1.2&r2=1.3

--- cluster/dlm/tests/usertest/dlmtest2.c	2006/12/08 19:25:27	1.2
+++ cluster/dlm/tests/usertest/dlmtest2.c	2006/12/12 21:17:56	1.3
@@ -35,7 +35,7 @@
 static int noqueue = 1;
 
 #define MAX_CLIENTS 4
-#define LOCKS 4
+#define LOCKS 16
 
 struct client {
 	int fd;
@@ -55,6 +55,9 @@
 };
 
 struct lk locks[LOCKS];
+struct lk *locks_flood;
+int locks_flood_n;
+int locks_flood_ast_done;
 
 void unlock(int i);
 
@@ -106,6 +109,26 @@
 	}
 }
 
+void dump_flood(void)
+{
+	struct lk *lk;
+	int i;
+
+	if (!locks_flood_n) {
+		printf("no current locks_flood\n");
+		return;
+	}
+
+	for (i = 0; i < locks_flood_n; i++) {
+		lk = &locks_flood[i];
+		printf("x %d lkid %x grmode %d rqmode %d wait_ast %d\n", i,
+			lk->lksb.sb_lkid,
+			lk->grmode,
+			lk->rqmode,
+			lk->wait_ast);
+	}
+}
+
 void bastfn(void *arg)
 {
 	struct lk *lk = arg;
@@ -148,6 +171,54 @@
 	locks[i].rqmode = -1;
 }
 
+void bastfn_flood(void *arg)
+{
+	struct lk *lk = arg;
+	printf("bastfn_flood %d\n", lk->id);
+}
+
+void astfn_flood(void *arg)
+{
+	struct lk *lk = arg;
+	int i = lk->id, unlock = 0;
+
+	if (!lk->wait_ast) {
+		printf("lk %d not waiting for ast\n", lk->id);
+		exit(-1);
+	}
+
+	lk->wait_ast = 0;
+
+	if (lk->lksb.sb_status == EUNLOCK) {
+		memset(&lk->lksb, 0, sizeof(struct dlm_lksb));
+		lk->grmode = -1;
+		unlock = 1;
+	} else if (lk->lksb.sb_status == EAGAIN) {
+		if (lk->grmode == -1)
+			memset(&lk->lksb, 0, sizeof(struct dlm_lksb));
+	} else {
+		if (lk->lksb.sb_status != 0) {
+			printf("lk %d unknown sb_status %d\n", lk->id,
+				lk->lksb.sb_status);
+			exit(-1);
+		}
+
+		lk->grmode = lk->rqmode;
+	}
+
+	lk->rqmode = -1;
+	locks_flood_ast_done++;
+
+	if (locks_flood_ast_done == locks_flood_n) {
+		printf("astfn_flood all %d done\n", locks_flood_n);
+		if (unlock) {
+			free(locks_flood);
+			locks_flood = NULL;
+			locks_flood_n = 0;
+		}
+	}
+}
+
 void process_libdlm(void)
 {
 	dlm_dispatch(libdlm_fd);
@@ -234,6 +305,44 @@
 		lock(i, mode);
 }
 
+void lock_flood(int n, int mode)
+{
+	struct lk *lk;
+	char name[DLM_RESNAME_MAXLEN];
+	int flags = 0, rv, i;
+
+	if (noqueue)
+		flags |= LKF_NOQUEUE;
+
+	if (locks_flood) {
+		printf("unlock_flood required before another lock_flood\n");
+		return;
+	}
+
+	locks_flood = malloc(n * sizeof(struct lk));
+	if (!locks_flood) {
+		printf("no mem for %d locks\n", n);
+		return;
+	}
+	locks_flood_n = n;
+	locks_flood_ast_done = 0;
+	memset(locks_flood, 0, sizeof(*locks_flood));
+
+	for (i = 0; i < n; i++) {
+		memset(name, 0, sizeof(name));
+		snprintf(name, sizeof(name), "testflood%d", i);
+		lk = &locks_flood[i];
+
+		rv = dlm_ls_lock(dh, mode, &lk->lksb, flags,
+			 name, strlen(name), 0, astfn_flood, (void *) lk,
+			 bastfn_flood, NULL);
+		if (!rv) {
+			lk->wait_ast = 1;
+			lk->rqmode = mode;
+		}
+	}
+}
+
 void unlock(int i)
 {
 	uint32_t lkid;
@@ -302,19 +411,61 @@
 		unlock(i);
 }
 
+void unlock_flood(void)
+{
+	struct lk *lk;
+	uint32_t lkid;
+	int rv, i;
+
+	if (!locks_flood)
+		return;
+
+	if (locks_flood_ast_done != locks_flood_n)
+		printf("warning: locks_flood_ast_done %d locks_flood_n %d\n",
+			locks_flood_ast_done, locks_flood_n);
+
+	locks_flood_ast_done = 0;
+
+	for (i = 0; i < locks_flood_n; i++) {
+		lk = &locks_flood[i];
+		lkid = lk->lksb.sb_lkid;
+		if (!lkid) {
+			printf("unlock_flood %d skip zero lkid\n", i);
+			continue;
+		}
+
+		rv = dlm_ls_unlock(dh, lkid, 0, &lk->lksb, (void *)lk);
+		if (!rv) {
+			lk->wait_ast = 1;
+			lk->rqmode = -1;
+		} else {
+			char input[32];
+			printf("flood: dlm_ls_unlock: %d rv %d errno %d\n",
+				i, rv, errno);
+			printf("press X to exit, D to dispatch, "
+			       "other to continue\n");
+			fgets(input, 32, stdin);
+			if (input[0] == 'X')
+				exit(-1);
+			else if (input[0] == 'D')
+				dlm_dispatch(libdlm_fd);
+		}
+	}
+}
+
 void loop(int i, int num)
 {
 	int n;
 
 	for (n = 0; n < num; n++) {
-		/*
 		lock(i, LKM_PRMODE);
 		dlm_dispatch(libdlm_fd);
 		unlock(i);
 		dlm_dispatch(libdlm_fd);
-		*/
+		/*
 		lock_sync(i, LKM_PRMODE);
 		unlock_sync(i);
+		*/
 	}
 }
 
@@ -380,6 +531,7 @@
 	if (!strncmp(cmd, "EXIT", 4)) {
 		*quit = 1;
 		unlock_all();
+		unlock_flood();
 		return;
 	}
 
@@ -420,6 +572,22 @@
 		exit(0);
 	}
 
+	if (!strncmp(cmd, "lock_flood", 10) && strlen(cmd) == 10) {
+		lock_flood(x, y);
+		return;
+	}
+
+	if (!strncmp(cmd, "unlock_flood", 12) && strlen(cmd) == 12) {
+		unlock_flood();
+		return;
+	}
+
+	if (!strncmp(cmd, "unlock_flood-kill", 17) && strlen(cmd) == 17) {
+		unlock_flood();
+		printf("process exiting\n");
+		exit(0);
+	}
+
 	if (!strncmp(cmd, "ex", 2)) {
 		lock(x, LKM_EXMODE);
 		return;
@@ -440,11 +608,16 @@
 		return;
 	}
 
-	if (!strncmp(cmd, "dump", 4)) {
+	if (!strncmp(cmd, "dump", 4) && strlen(cmd) == 4) {
 		dump();
 		return;
 	}
 
+	if (!strncmp(cmd, "dump_flood", 10) && strlen(cmd) == 10) {
+		dump_flood();
+		return;
+	}
+
 	if (!strncmp(cmd, "loop", 4)) {
 		loop(x, y);
 		return;
@@ -472,11 +645,15 @@
 		printf("unlock_sync x 	 - synchronous version of unlock\n");
 		printf("lock-kill x mode - request/convert lock on resource x, then exit\n");
 		printf("unlock-kill x	 - unlock lock on resource x, then exit\n");
+		printf("lock_flood n mode - request n locks (in flood namespace)\n");
+		printf("unlock_flood     - unlock all from lock_flood\n");
+		printf("unlock_flood-kill - unlock all from lock_flood and exit\n");
 		printf("ex x		 - equivalent to: lock x 5\n");
 		printf("pr x		 - equivalent to: lock x 3\n");
 		printf("hold		 - for x in 0 to MAX, lock x 3\n");
 		printf("release		 - for x in 0 to MAX, unlock x\n");
 		printf("dump		 - show info for all resources\n");
+		printf("dump_flood	 - show info for all flood resources\n");
 		printf("loop x n	 - lock_sync x PR / unlock_sync x, n times\n");
 		printf("hammer n	 - loop doing random lock/unlock on all locks, n times\n");
 		printf("noqueue		 - toggle NOQUEUE flag for all requests\n"); 



             reply	other threads:[~2006-12-12 21:17 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-12-12 21:17 teigland [this message]
  -- strict thread matches above, loose matches on Subject: below --
2008-02-21 17:49 [Cluster-devel] cluster/dlm/tests/usertest dlmtest2.c teigland
2008-01-17 21:49 teigland
2007-08-23 19:13 teigland
2007-08-22 14:01 teigland
2007-05-30 19:31 teigland
2007-04-27 19:09 teigland
2007-03-27 19:04 teigland
2007-02-27 18:55 teigland
2007-01-04 20:34 teigland
2006-12-08 19:25 teigland

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=20061212211756.1251.qmail@sourceware.org \
    --to=teigland@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.