FIX: imsm: Rebuild does not start on second failed disk

[thirdparty/mdadm.git] / monitor.c

diff --git a/monitor.c b/monitor.c
index 1107d4778b3f623f60e6c0ae0800924ecf2c75d1..7ac59072144ca5c309625e52a74bfb433a408667 100644 (file)
--- a/monitor.c
+++ b/monitor.c
@@ -219,6 +219,7 @@ static int read_and_act(struct active_array *a)
        int deactivate = 0;
        struct mdinfo *mdi;
        int dirty = 0;
+       int count = 0;
 
        a->next_state = bad_word;
        a->next_action = bad_action;
@@ -229,6 +230,7 @@ static int read_and_act(struct active_array *a)
        sync_completed = read_sync_completed(a->sync_completed_fd);
        for (mdi = a->info.devs; mdi ; mdi = mdi->next) {
                mdi->next_state = 0;
+               mdi->curr_state = 0;
                if (mdi->state_fd >= 0) {
                        mdi->recovery_start = read_resync_start(mdi->recovery_fd);
                        mdi->curr_state = read_dev_state(mdi->state_fd);
@@ -310,7 +312,10 @@ static int read_and_act(struct active_array *a)
                                                   mdi->curr_state);
                        if (! (mdi->curr_state & DS_INSYNC))
                                check_degraded = 1;
+                       count++;
                }
+               if (count != a->info.array.raid_disks)
+                       check_degraded = 1;
        }
 
        if (!deactivate &&
@@ -363,8 +368,28 @@ static int read_and_act(struct active_array *a)
                /* Reshape has progressed or completed so we need to
                 * update the array state - and possibly the array size
                 */
-               a->last_checkpoint = sync_completed;
+               if (sync_completed != 0)
+                       a->last_checkpoint = sync_completed;
+               /* We might need to update last_checkpoint depending on
+                * the reason that reshape finished.
+                * if array reshape is really finished:
+                *        set check point to the end, this allows
+                *        set_array_state() to finalize reshape in metadata
+                * if reshape if broken: do not set checkpoint to the end
+                *        this allows for reshape restart from checkpoint
+                */
+               if ((a->curr_action != reshape) &&
+                   (a->prev_action == reshape)) {
+                       char buf[40];
+                       if ((sysfs_get_str(&a->info, NULL,
+                                         "reshape_position",
+                                         buf,
+                                         sizeof(buf)) >= 0) &&
+                            strncmp(buf, "none", 4) == 0)
+                               a->last_checkpoint = a->info.component_size;
+               }
                a->container->ss->set_array_state(a, a->curr_state <= clean);
+               a->last_checkpoint = sync_completed;
        }
 
        if (sync_completed > a->last_checkpoint)
@@ -400,6 +425,7 @@ static int read_and_act(struct active_array *a)
                        if (remove_result > 0) {
                                dprintf(" %d:removed", mdi->disk.raid_disk);
                                close(mdi->state_fd);
+                               close(mdi->recovery_fd);
                                mdi->state_fd = -1;
                        }
                }
@@ -513,7 +539,7 @@ static int wait_and_act(struct supertype *container, int nowait)
                /* once an array has been deactivated we want to
                 * ask the manager to discard it.
                 */
-               if (!a->container || (a->info.array.level == 0)) {
+               if (!a->container) {
                        if (discard_this) {
                                ap = &(*ap)->next;
                                continue;
@@ -541,7 +567,11 @@ static int wait_and_act(struct supertype *container, int nowait)
                 * problem as there are no active arrays, there is
                 * nothing that we need to be ready to do.
                 */
-               int fd = open_dev_excl(container->devnum);
+               int fd;
+               if (sigterm)
+                       fd = open_dev_excl(container->devnum);
+               else
+                       fd = open_dev_flags(container->devnum, O_RDONLY|O_EXCL);
                if (fd >= 0 || errno != EBUSY) {
                        /* OK, we are safe to leave */
                        if (sigterm && !dirty_arrays)
@@ -562,10 +592,18 @@ static int wait_and_act(struct supertype *container, int nowait)
 
        if (!nowait) {
                sigset_t set;
+               struct timespec ts;
+               ts.tv_sec = 24*3600;
+               ts.tv_nsec = 0;
+               if (*aap == NULL) {
+                       /* just waiting to get O_EXCL access */
+                       ts.tv_sec = 0;
+                       ts.tv_nsec = 20000000ULL;
+               }
                sigprocmask(SIG_UNBLOCK, NULL, &set);
                sigdelset(&set, SIGUSR1);
                monitor_loop_cnt |= 1;
-               rv = pselect(maxfd+1, NULL, NULL, &rfds, NULL, &set);
+               rv = pselect(maxfd+1, NULL, NULL, &rfds, &ts, &set);
                monitor_loop_cnt += 1;
                if (rv == -1 && errno == EINTR)
                        rv = 0;