]> err.no Git - linux-2.6/commitdiff
[SCSI] zfcp: fix cleanup of dismissed error recovery actions
authorMartin Peschke <mp3@de.ibm.com>
Thu, 15 Nov 2007 12:57:17 +0000 (13:57 +0100)
committerJames Bottomley <James.Bottomley@HansenPartnership.com>
Fri, 16 Nov 2007 19:03:21 +0000 (13:03 -0600)
Calling zfcp_erp_strategy_check_action() after zfcp_erp_action_to_running()
in zfcp_erp_strategy() might cause an unbalanced up() for erp_ready_sem,
which makes the zfcp recovery fail somewhere along the way:

erp thread processing erp_action:
|
| someone waking up erp thread for erp_action
| |
| | someone else dismissing erp_action:
| | |
V V V

write_lock_irqsave(&adapter->erp_lock, flags);
...
if (zfcp_erp_action_exists(erp_action) == ZFCP_ERP_ACTION_RUNNING) {
zfcp_erp_action_to_ready(erp_action);
up(&adapter->erp_ready_sem); /* first up() for erp_action */
}
write_unlock_irqrestore(&adapter->erp_lock, flags);

write_lock_irqsave(&adapter->erp_lock, flags);
...
zfcp_erp_action_to_running(erp_action);
write_unlock_restore(&adapter->erp_lock, flags);
/* processing erp_action */

write_lock_irqsave(&adapter->erp_lock, flags);
...
erp_action->status |= ZFCP_STATUS_ERP_DISMISSED;
if (zfcp_erp_action_exists(erp_action) ==
ZFCP_ERP_ACTION_RUNNING) {
zfcp_erp_action_to_ready(erp_action);
up(&adapter->erp_ready_sem);
/* second, unbalanced up() for erp_action */
}
...
write_unlock_restore(&adapter->erp_lock, flags);

write_lock_irqsave(&adapter->erp_lock, flags);
if (erp_action->status & ZFCP_STATUS_ERP_DISMISSED) {
zfcp_erp_action_dequeue(erp_action);
retval = ZFCP_ERP_DISMISSED;
}
...
write_unlock_restore(&adapter->erp_lock, flags);
down(&adapter->erp_ready_sem);
/* this down() is meant to balance the first up() */

The erp thread must not dismiss an erp_action after moving that action to
erp_running_head. Instead it should just go through the down() operation,
which balances the first up(), and run through zfcp_erp_strategy one more
time for the second up(), which eventually cleans up erp_action. Which
is similar to the normal processing of an event for erp_action doing
something asynchronously (e.g. waiting for the completion of an fsf_req).

This only works if we make sure that a dismissed erp_action is passed to
zfcp_erp_strategy() prior to the other action, which caused actions to be
dismissed. Therefore the patch implements this rule: running actions go to
the head of the ready list; new actions go to the tail of the ready list;
the erp thread picks actions to be processed from the ready list's head.

Signed-off-by: Martin Peschke <mp3@de.ibm.com>
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
drivers/s390/scsi/zfcp_erp.c

index ad5b481ddafacff6662b7bd67872d04f3a5f96c6..07fa824d179f203e3f9bbbf6b099e94206140e5a 100644 (file)
@@ -1065,7 +1065,7 @@ zfcp_erp_thread(void *data)
                                 &adapter->status)) {
 
                write_lock_irqsave(&adapter->erp_lock, flags);
-               next = adapter->erp_ready_head.prev;
+               next = adapter->erp_ready_head.next;
                write_unlock_irqrestore(&adapter->erp_lock, flags);
 
                if (next != &adapter->erp_ready_head) {
@@ -1155,15 +1155,13 @@ zfcp_erp_strategy(struct zfcp_erp_action *erp_action)
 
        /*
         * check for dismissed status again to avoid follow-up actions,
-        * failing of targets and so on for dismissed actions
+        * failing of targets and so on for dismissed actions,
+        * we go through down() here because there has been an up()
         */
-       retval = zfcp_erp_strategy_check_action(erp_action, retval);
+       if (erp_action->status & ZFCP_STATUS_ERP_DISMISSED)
+               retval = ZFCP_ERP_CONTINUES;
 
        switch (retval) {
-       case ZFCP_ERP_DISMISSED:
-               /* leave since this action has ridden to its ancestors */
-               debug_text_event(adapter->erp_dbf, 6, "a_st_dis2");
-               goto unlock;
        case ZFCP_ERP_NOMEM:
                /* no memory to continue immediately, let it sleep */
                if (!(erp_action->status & ZFCP_STATUS_ERP_LOWMEM)) {
@@ -3091,7 +3089,7 @@ zfcp_erp_action_enqueue(int action,
        ++adapter->erp_total_count;
 
        /* finally put it into 'ready' queue and kick erp thread */
-       list_add(&erp_action->list, &adapter->erp_ready_head);
+       list_add_tail(&erp_action->list, &adapter->erp_ready_head);
        up(&adapter->erp_ready_sem);
        retval = 0;
  out: