mdmon: allow pid to be stored in different directory.

/var/run probably doesn't persist from early boot.
So if necessary, store in in /lib/init/rw or somewhere else
that does persist.

Signed-off-by: NeilBrown <neilb@suse.de>

9 files changed, 121 insertions, 34 deletions

diff --git a/Makefile b/Makefile
index 5636392..1836b4b 100644
--- a/Makefile
+++ b/Makefile
@@ -58,7 +58,13 @@ CONFFILE = $(SYSCONFDIR)/mdadm.conf
 CONFFILE2 = $(SYSCONFDIR)/mdadm/mdadm.conf
 MAILCMD =/usr/sbin/sendmail -t
 CONFFILEFLAGS = -DCONFFILE=\"$(CONFFILE)\" -DCONFFILE2=\"$(CONFFILE2)\"
-CFLAGS = $(CWFLAGS) $(CXFLAGS) -DSendmail=\""$(MAILCMD)"\" $(CONFFILEFLAGS)
+# ALT_RUN should be somewhere that persists across the pivotroot
+# from early boot to late boot.
+# If you don't have /lib/init/rw you might want to use /dev/.something
+#  e.g. make ALT_RUN=/dev/.mdadm
+ALT_RUN = /lib/init/rw
+ALTFLAGS = -DALT_RUN=\"$(ALT_RUN)\"
+CFLAGS = $(CWFLAGS) $(CXFLAGS) -DSendmail=\""$(MAILCMD)"\" $(CONFFILEFLAGS) $(ALTFLAGS)
 
 # If you want a static binary, you might uncomment these
 # LDFLAGS = -static
diff --git a/managemon.c b/managemon.c
index a4e9a8f..2a73d4b 100644
--- a/managemon.c
+++ b/managemon.c
@@ -702,14 +702,29 @@ void do_manager(struct supertype *container)
 
 			read_sock(container);
 
-			if (container->sock < 0 || socket_hup_requested) {
-				/* If this fails, we hope it already exists
-				 * pid file lives in /var/run/mdadm/mdXX.pid
+			if (socket_hup_requested) {
+				/* Try to create pid file and socket in
+				 * main or alternate RUN directory.
 				 */
-				mkdir("/var/run/mdadm", 0600);
-				close(container->sock);
-				container->sock = make_control_sock(container->devname);
-				make_pidfile(container->devname, 0);
+				char *dir = VAR_RUN;
+				if (mkdir(dir, 0600) < 0 && errno != EEXIST) {
+					char *dir = ALT_RUN;
+					if (mkdir(dir, 0600) < 0 && errno != EEXIST)
+						dir = NULL;
+				} else {
+					if (proc_fd >= 0)
+						close(proc_fd);
+					proc_fd = -1;
+				}
+				if (dir && !sigterm &&
+				    (container->sock < 0 ||
+				     strcmp(dir, pid_dir) != 0)) {
+					close(container->sock);
+					remove_pidfile(container->devname);
+					pid_dir = dir;
+					container->sock = make_control_sock(container->devname);
+					make_pidfile(container->devname);
+				}
 				socket_hup_requested = 0;
 			}
 			if (container->sock < 0)
@@ -726,12 +741,9 @@ void do_manager(struct supertype *container)
 		if (sigterm)
 			wakeup_monitor();
 
-		if (update_queue == NULL) {
-			if (container->sock < 0)
-				mdstat_wait_fd(proc_fd, &set);
-			else
-				mdstat_wait_fd(container->sock, &set);
-		} else
+		if (update_queue == NULL)
+			mdstat_wait_fd(container->sock, proc_fd, &set);
+		else
 			/* If an update is happening, just wait for signal */
 			pselect(0, NULL, NULL, NULL, NULL, &set);
 	} while(1);
diff --git a/mdadm.h b/mdadm.h
index eec5c0c..f65a462 100644
--- a/mdadm.h
+++ b/mdadm.h
@@ -68,6 +68,15 @@ extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
 #define DEFAULT_BITMAP_DELAY 5
 #define DEFAULT_MAX_WRITE_BEHIND 256
 
+#define VAR_RUN "/var/run/mdadm"
+/* ALT_RUN should be somewhere that persists across the pivotroot
+ * from early boot to late boot.
+ * If you don't have /lib/init/rw you might want to use /dev/.something
+ */
+#ifndef ALT_RUN
+#define ALT_RUN "/lib/init/rw/mdadm"
+#endif /* ALT_RUN */
+
 #include	"md_u.h"
 #include	"md_p.h"
 #include	"bitmap.h"
@@ -332,7 +341,7 @@ struct mdstat_ent {
 extern struct mdstat_ent *mdstat_read(int hold, int start);
 extern void free_mdstat(struct mdstat_ent *ms);
 extern void mdstat_wait(int seconds);
-extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
+extern void mdstat_wait_fd(int fd, int fd2, const sigset_t *sigmask);
 extern int mddev_busy(int devnum);
 
 struct map_ent {
@@ -882,6 +891,7 @@ extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
 extern int open_mddev(char *dev, int report_errors);
 extern int open_container(int fd);
 
+extern char *pid_dir;
 extern int mdmon_running(int devnum);
 extern int mdmon_pid(int devnum);
 extern int check_env(char *name);
diff --git a/mdmon.8 b/mdmon.8
index 05aaa50..1dc7844 100644
--- a/mdmon.8
+++ b/mdmon.8
@@ -5,7 +5,7 @@ mdmon \- monitor MD external metadata arrays
 
 .SH SYNOPSIS
 
-.BI mdmon " CONTAINER [NEWROOT]"
+.BI mdmon " CONTAINER [NEWROOT | PID]"
 
 .SH OVERVIEW
 The 2.6.27 kernel brings the ability to support external metadata arrays.
@@ -148,6 +148,45 @@ will terminate any
 instances that are running in the current namespace,
 .IR chroot (2)
 to NEWROOT, and continue monitoring the container.
+
+Alternately the new
+.I mdmon
+can be started after the new root has been installed with
+.I pivotroot
+by passing
+.B /
+as the NEWROOT.  For
+.I mdmon
+to be able to find and kill the old
+.I mdmon
+there must be some part of the filesystem that persists from before
+the
+.I pivotroot
+to afterwards.  This can be
+.B /var/run/mdadm
+or some other directory.  The default other directory is
+.B /lib/init/rw/mdadm
+but this can easily be changed when
+.I mdmon
+is compiled.
+If
+.I mdmon
+cannot store the pid file in
+.B /var/run/mdadm
+it will store it in the other directory until
+.B /var/run/mdadm
+becomes available, or until it is killed.
+.TP
+[PID]
+If the second argument to
+.I mdmon
+does not start with a slash it is assumed to be the pid of a previous
+instance of
+.I mdmon
+to kill.  This is used internally by
+.I mdmon
+to pass the pid across a chroot/exec and should not be used otherwise.
+
 .PP
 Note that
 .I mdmon
@@ -159,6 +198,17 @@ RAID arrays.  The only times it is run other that by
 is when the boot scripts need to restart it after mounting the new
 root filesystem.
 
+.SH EXMAPLES
+
+.B "  mdmon /proc/mdstat /"
+.br
+Any
+.I mdmon
+which is currently running is killed and a new instance is started.
+This should be run early in the boot sequence after a
+.I pivotroot
+to the final root filesystem, but before that filesystem is remounted
+read-write.
 .SH SEE ALSO
 .IR mdadm (8),
 .IR md (4).
diff --git a/mdmon.c b/mdmon.c
index d20bb3e..fa49706 100644
--- a/mdmon.c
+++ b/mdmon.c
@@ -113,19 +113,16 @@ static struct superswitch *find_metadata_methods(char *vers)
 	return NULL;
 }
 
-int make_pidfile(char *devname, int o_excl)
+int make_pidfile(char *devname)
 {
 	char path[100];
 	char pid[10];
 	int fd;
 	int n;
 
-	if (sigterm)
-		return -1;
-
-	sprintf(path, "/var/run/mdadm/%s.pid", devname);
+	sprintf(path, "%s/%s.pid", pid_dir, devname);
 
-	fd = open(path, O_RDWR|O_CREAT|o_excl, 0600);
+	fd = open(path, O_RDWR|O_CREAT|O_EXCL, 0600);
 	if (fd < 0)
 		return -errno;
 	sprintf(pid, "%d\n", getpid());
@@ -186,13 +183,13 @@ void remove_pidfile(char *devname)
 {
 	char buf[100];
 
-	if (sigterm)
-		return;
-
-	sprintf(buf, "/var/run/mdadm/%s.pid", devname);
+	sprintf(buf, "%s/%s.pid", pid_dir, devname);
 	unlink(buf);
-	sprintf(buf, "/var/run/mdadm/%s.sock", devname);
+	sprintf(buf, "%s/%s.sock", pid_dir, devname);
 	unlink(buf);
+	if (strcmp(pid_dir, ALT_RUN) == 0)
+		/* try to clean up when we are finished with this dir */
+		rmdir(pid_dir);
 }
 
 int make_control_sock(char *devname)
@@ -205,7 +202,7 @@ int make_control_sock(char *devname)
 	if (sigterm)
 		return -1;
 
-	sprintf(path, "/var/run/mdadm/%s.sock", devname);
+	sprintf(path, "%s/%s.sock", pid_dir, devname);
 	unlink(path);
 	sfd = socket(PF_LOCAL, SOCK_STREAM, 0);
 	if (sfd < 0)
diff --git a/mdmon.h b/mdmon.h
index 4494085..a03686f 100644
--- a/mdmon.h
+++ b/mdmon.h
@@ -66,7 +66,7 @@ void remove_pidfile(char *devname);
 void do_monitor(struct supertype *container);
 void do_manager(struct supertype *container);
 int make_control_sock(char *devname);
-int make_pidfile(char *devname, int o_excl);
+int make_pidfile(char *devname);
 extern int socket_hup_requested;
 extern int sigterm;
 
diff --git a/mdstat.c b/mdstat.c
index 4d2f473..4b3f6fe 100644
--- a/mdstat.c
+++ b/mdstat.c
@@ -266,16 +266,20 @@ void mdstat_wait(int seconds)
 	select(maxfd + 1, NULL, NULL, &fds, &tm);
 }
 
-void mdstat_wait_fd(int fd, const sigset_t *sigmask)
+void mdstat_wait_fd(int fd, int fd2, const sigset_t *sigmask)
 {
 	fd_set fds, rfds;
-	int maxfd = fd;
+	int maxfd = 0;
 
 	FD_ZERO(&fds);
 	FD_ZERO(&rfds);
 	if (mdstat_fd >= 0)
 		FD_SET(mdstat_fd, &fds);
-	if (fd >= 0) {
+
+	if (fd < 0)
+		fd = fd2, fd2 = -1;
+
+	while (fd >= 0) {
 		struct stat stb;
 		fstat(fd, &stb);
 		if ((stb.st_mode & S_IFMT) == S_IFREG)
@@ -286,6 +290,12 @@ void mdstat_wait_fd(int fd, const sigset_t *sigmask)
 			FD_SET(fd, &fds);
 		else
 			FD_SET(fd, &rfds);
+
+		if (fd > maxfd)
+			maxfd = fd;
+
+		fd = fd2;
+		fd2 = -1;
 	}
 	if (mdstat_fd > maxfd)
 		maxfd = mdstat_fd;
diff --git a/msg.c b/msg.c
index 8d52b94..cc07b96 100644
--- a/msg.c
+++ b/msg.c
@@ -147,7 +147,7 @@ int connect_monitor(char *devname)
 	int pos;
 	char *c;
 
-	pos = sprintf(path, "/var/run/mdadm/");
+	pos = sprintf(path, "%s/", VAR_RUN);
 	if (is_subarray(devname)) {
 		devname++;
 		c = strchr(devname, '/');
diff --git a/util.c b/util.c
index e5b0c1f..68f048d 100644
--- a/util.c
+++ b/util.c
@@ -1463,13 +1463,15 @@ int fd2devnum(int fd)
 	return NoMdDev;
 }
 
+char *pid_dir = VAR_RUN;
+
 int mdmon_pid(int devnum)
 {
 	char path[100];
 	char pid[10];
 	int fd;
 	int n;
-	sprintf(path, "/var/run/mdadm/%s.pid", devnum2devname(devnum));
+	sprintf(path, "%s/%s.pid", pid_dir, devnum2devname(devnum));
 	fd = open(path, O_RDONLY | O_NOATIME, 0);
 
 	if (fd < 0)