Sign Up
Log In
Log In
or
Sign Up
Places
All Projects
Status Monitor
Collapse sidebar
home:steve-beattie
kernel
linux-2.6-gfs2-dlm-fix-mount-issues.patch
Overview
Repositories
Revisions
Requests
Users
Attributes
Meta
File linux-2.6-gfs2-dlm-fix-mount-issues.patch of Package kernel
Date: Tue, 31 Oct 2006 11:55:56 -0600 From: David Teigland <teigland@redhat.com> Subject: [RHEL5 PATCH 1/3] dlm: fix requestqueue race BZ 211914 (beta blocker) There's a race between dlm_recoverd (1) enabling locking and (2) clearing out the requestqueue, and dlm_recvd (1) checking if locking is enabled and (2) adding a message to the requestqueue. An order of recoverd(1), recvd(1), recvd(2), recoverd(2) will result in a message being left on the requestqueue. The fix is to have dlm_recvd check if dlm_recoverd has enabled locking after taking the mutex for the requestqueue and if it has processing the message instead of queueing it. With this fix, QA's mount_stress test passes. Index: linux-2.6.19-rc1-rh/fs/dlm/lock.c =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/lock.c 2006-10-30 12:45:38.833044329 -0600 +++ linux-2.6.19-rc1-rh/fs/dlm/lock.c 2006-10-30 12:46:39.696744002 -0600 @@ -3029,10 +3029,17 @@ while (1) { if (dlm_locking_stopped(ls)) { - if (!recovery) - dlm_add_requestqueue(ls, nodeid, hd); - error = -EINTR; - goto out; + if (recovery) { + error = -EINTR; + goto out; + } + error = dlm_add_requestqueue(ls, nodeid, hd); + if (error == -EAGAIN) + continue; + else { + error = -EINTR; + goto out; + } } if (lock_recovery_try(ls)) Index: linux-2.6.19-rc1-rh/fs/dlm/requestqueue.c =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/requestqueue.c 2006-10-05 11:45:35.000000000 -0500 +++ linux-2.6.19-rc1-rh/fs/dlm/requestqueue.c 2006-10-30 12:51:13.067971288 -0600 @@ -30,26 +30,39 @@ * lockspace is enabled on some while still suspended on others. */ -void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd) +int dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd) { struct rq_entry *e; int length = hd->h_length; + int rv = 0; if (dlm_is_removed(ls, nodeid)) - return; + return 0; e = kmalloc(sizeof(struct rq_entry) + length, GFP_KERNEL); if (!e) { log_print("dlm_add_requestqueue: out of memory\n"); - return; + return 0; } e->nodeid = nodeid; memcpy(e->request, hd, length); + /* We need to check dlm_locking_stopped() after taking the mutex to + avoid a race where dlm_recoverd enables locking and runs + process_requestqueue between our earlier dlm_locking_stopped check + and this addition to the requestqueue. */ + mutex_lock(&ls->ls_requestqueue_mutex); - list_add_tail(&e->list, &ls->ls_requestqueue); + if (dlm_locking_stopped(ls)) + list_add_tail(&e->list, &ls->ls_requestqueue); + else { + log_debug(ls, "dlm_add_requestqueue skip from %d", nodeid); + kfree(e); + rv = -EAGAIN; + } mutex_unlock(&ls->ls_requestqueue_mutex); + return rv; } int dlm_process_requestqueue(struct dlm_ls *ls) Index: linux-2.6.19-rc1-rh/fs/dlm/requestqueue.h =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/requestqueue.h 2006-10-05 11:45:35.000000000 -0500 +++ linux-2.6.19-rc1-rh/fs/dlm/requestqueue.h 2006-10-30 12:46:39.698743696 -0600 @@ -13,7 +13,7 @@ #ifndef __REQUESTQUEUE_DOT_H__ #define __REQUESTQUEUE_DOT_H__ -void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd); +int dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd); int dlm_process_requestqueue(struct dlm_ls *ls); void dlm_wait_requestqueue(struct dlm_ls *ls); void dlm_purge_requestqueue(struct dlm_ls *ls); Date: Tue, 31 Oct 2006 11:56:01 -0600 From: David Teigland <teigland@redhat.com> Subject: [RHEL5 PATCH 2/3] dlm: fix aborted recovery during node removal BZ 211914 (beta blocker) With the new cluster infrastructure, dlm recovery for a node removal can be aborted and restarted for a node addition. When this happens, the restarted recovery isn't aware that it's doing recovery for the earlier removal as well as the addition. So, it then skips the recovery steps only required when nodes are removed. This can result in locks not being purged for failed/removed nodes. The fix is to check for removed nodes for which recovery has not been completed at the start of a new recovery sequence. With this fix, QA's mount_stress test passes. Index: linux-2.6.19-rc1-rh/fs/dlm/member.c =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/member.c 2006-10-30 16:32:50.218089404 -0600 +++ linux-2.6.19-rc1-rh/fs/dlm/member.c 2006-10-30 16:33:04.718873600 -0600 @@ -186,6 +186,14 @@ struct dlm_member *memb, *safe; int i, error, found, pos = 0, neg = 0, low = -1; + /* previously removed members that we've not finished removing need to + count as a negative change so the "neg" recovery steps will happen */ + + list_for_each_entry(memb, &ls->ls_nodes_gone, list) { + log_debug(ls, "prev removed member %d", memb->nodeid); + neg++; + } + /* move departed members from ls_nodes to ls_nodes_gone */ list_for_each_entry_safe(memb, safe, &ls->ls_nodes, list) { Index: linux-2.6.19-rc1-rh/fs/dlm/recoverd.c =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/recoverd.c 2006-10-30 16:32:50.219089251 -0600 +++ linux-2.6.19-rc1-rh/fs/dlm/recoverd.c 2006-10-30 16:33:43.321974821 -0600 @@ -164,6 +164,13 @@ */ dlm_recover_rsbs(ls); + } else { + /* + * Other lockspace members may be going through the "neg" steps + * while also adding us to the lockspace, in which case they'll + * be looking for this status bit during dlm_recover_locks(). + */ + dlm_set_recover_status(ls, DLM_RS_LOCKS); } dlm_release_root_list(ls); Date: Wed, 1 Nov 2006 09:31:48 -0600 From: David Teigland <teigland@redhat.com> Subject: [RHEL5 PATCH] dlm: do full recover_locks barrier BZ 211914 (beta blocker) The previous patch "[RHEL5 PATCH 2/3] dlm: fix aborted recovery during node removal" was incomplete as discovered with further testing. It set the bit for the RS_LOCKS barrier but did not then wait for the barrier. This is often ok, but sometimes it will cause yet another recovery hang. If it's a new node that also has the lowest nodeid that skips the barrier wait, then it misses the important step of collecting and reporting the barrier status from the other nodes (which is the job of the low nodeid in the barrier wait routine). Index: linux-2.6.19-rc1-quilt/fs/dlm/recoverd.c =================================================================== --- linux-2.6.19-rc1-quilt.orig/fs/dlm/recoverd.c 2006-10-31 13:07:26.146856607 -0600 +++ linux-2.6.19-rc1-quilt/fs/dlm/recoverd.c 2006-10-31 13:37:44.099062935 -0600 @@ -168,9 +168,15 @@ /* * Other lockspace members may be going through the "neg" steps * while also adding us to the lockspace, in which case they'll - * be looking for this status bit during dlm_recover_locks(). + * be doing the recover_locks (RS_LOCKS) barrier. */ dlm_set_recover_status(ls, DLM_RS_LOCKS); + + error = dlm_recover_locks_wait(ls); + if (error) { + log_error(ls, "recover_locks_wait failed %d", error); + goto fail; + } } dlm_release_root_list(ls); Date: Tue, 31 Oct 2006 11:56:08 -0600 From: David Teigland <teigland@redhat.com> Subject: [RHEL5 PATCH 3/3] dlm: fix stopping unstarted recovery BZ 211914 (beta blocker) When many nodes are joining a lockspace simultaneously, the dlm gets a quick sequence of stop/start events, a pair for adding each node. dlm_controld in user space sends dlm_recoverd in the kernel each stop and start event. dlm_controld will sometimes send the stop before dlm_recoverd has had a chance to take up the previously queued start. The stop aborts the processing of the previous start by setting the RECOVERY_STOP flag. dlm_recoverd is erroneously clearing this flag and ignoring the stop/abort if it happens to take up the start after the stop meant to abort it. The fix is to check the sequence number that's incremented for each stop/start before clearing the flag. With this fix, QA's mount_stress test passes. Index: linux-2.6.19-rc1-rh/fs/dlm/recoverd.c =================================================================== --- linux-2.6.19-rc1-rh.orig/fs/dlm/recoverd.c 2006-10-30 16:33:43.321974821 -0600 +++ linux-2.6.19-rc1-rh/fs/dlm/recoverd.c 2006-10-30 16:34:15.811010308 -0600 @@ -219,6 +219,10 @@ return error; } +/* The dlm_ls_start() that created the rv we take here may already have been + stopped via dlm_ls_stop(); in that case we need to leave the RECOVERY_STOP + flag set. */ + static void do_ls_recovery(struct dlm_ls *ls) { struct dlm_recover *rv = NULL; @@ -226,7 +230,8 @@ spin_lock(&ls->ls_recover_lock); rv = ls->ls_recover_args; ls->ls_recover_args = NULL; - clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags); + if (rv && ls->ls_recover_seq == rv->seq) + clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags); spin_unlock(&ls->ls_recover_lock); if (rv) {
Locations
Projects
Search
Status Monitor
Help
OpenBuildService.org
Documentation
API Documentation
Code of Conduct
Contact
Support
@OBShq
Terms
openSUSE Build Service is sponsored by
The Open Build Service is an
openSUSE project
.
Sign Up
Log In
Places
Places
All Projects
Status Monitor
