ocfs2: dlmlock_master should return DLM_NORMAL after adding lock to blocked list

Function dlmlock_master() returns DLM_RECOVERING/DLM_MIGRATING/ DLM_FORWAR after adding lock to blocked list if lockres has the state DLM_LOCK_RES_RECOVERING/DLM_LOCK_RES_MIGRATING/ DLM_LOCK_RES_IN_PROGRESS. so it will retry in dlmlock(). And this may cause dlm_thread fall into an infinite loop

Thread1 dlm_thread
calls dlm_lock->dlmlock_master,
if lockresA is in state
DLM_LOCK_RES_RECOVERING, calls
__dlm_wait_on_lockres() and waits
until others threads clear this
state;

If cannot grant this lock,
adding lock to blocked list,
and return DLM_RECOVERING;

Grant this lock and move it to grant list;

After a while, retry and
calls list_add_tail(), adding lock
to blocked list again.

Granted and blocked list of this lockres will become the following conditions:

    lock_res->granted.next = dlm_lock->list_head;
    lock_res->blocked.next = dlm_lock->list_head;
    dlm_lock->list_head.next = dlm_lock_resource->blocked;

When dlm_thread traverses the granted list, it will fall into an endless loop, checking dlm_lock.list_head, dlm_lock->list_head.next (i.e.lock_res->blocked), lock_res->blocked.next(i.e.dlm_lock.list_head again) .....

Signed-off-by joyce <[email protected]>

---
 fs/ocfs2/dlm/dlmlock.c |    1 +
 1 file changed, 1 insertion(+)

fs/ocfs2/dlm/dlmlock.c +1 -0

index 975810b..47e67c2 100644
@@ -178,6 +178,7 @@ static enum dlm_status dlmlock_master(struct dlm_ctxt *dlm,
 				     lock->ml.node);
 			}
 		} else {
+			status = DLM_NORMAL;
 			dlm_lock_get(lock);
 			list_add_tail(&lock->list, &res->blocked);
 			kick_thread = 1;

Function dlmlock_master() returns DLM_RECOVERING/DLM_MIGRATING/ DLM_FORWAR after adding lock to blocked list if lockres has the state DLM_LOCK_RES_RECOVERING/DLM_LOCK_RES_MIGRATING/ DLM_LOCK_RES_IN_PROGRESS. so it will retry in dlmlock(). And this may cause dlm_thread fall into an infinite loop

Thread1 dlm_thread
calls dlm_lock->dlmlock_master,
if lockresA is in state
DLM_LOCK_RES_RECOVERING, calls
__dlm_wait_on_lockres() and waits
until others threads clear this
state;

If cannot grant this lock,
adding lock to blocked list,
and return DLM_RECOVERING;

Grant this lock and move it to grant list;

After a while, retry and
calls list_add_tail(), adding lock
to blocked list again.

Granted and blocked list of this lockres will become the following conditions:

    lock_res->granted.next = dlm_lock->list_head;
    lock_res->blocked.next = dlm_lock->list_head;
    dlm_lock->list_head.next = dlm_lock_resource->blocked;

When dlm_thread traverses the granted list, it will fall into an endless loop, checking dlm_lock.list_head, dlm_lock->list_head.next (i.e.lock_res->blocked), lock_res->blocked.next(i.e.dlm_lock.list_head again) .....

Signed-off-by joyce <[email protected]>

---
 fs/ocfs2/dlm/dlmlock.c |    1 +
 1 file changed, 1 insertion(+)

fs/ocfs2/dlm/dlmlock.c +1 -0

index 975810b..47e67c2 100644
@@ -178,6 +178,7 @@ static enum dlm_status dlmlock_master(struct dlm_ctxt *dlm,
 				     lock->ml.node);
 			}
 		} else {
+			status = DLM_NORMAL;
 			dlm_lock_get(lock);
 			list_add_tail(&lock->list, &res->blocked);
 			kick_thread = 1;

Hi Jiufei,

Quoted 27 lines · Xue jiufei

Function dlmlock_master() returns DLM_RECOVERING/DLM_MIGRATING/ DLM_FORWAR after adding lock to blocked list if lockres has the state DLM_LOCK_RES_RECOVERING/DLM_LOCK_RES_MIGRATING/ DLM_LOCK_RES_IN_PROGRESS. so it will retry in dlmlock(). And this may cause dlm_thread fall into an infinite loop

Thread1 dlm_thread
calls dlm_lock->dlmlock_master,
if lockresA is in state
DLM_LOCK_RES_RECOVERING, calls
__dlm_wait_on_lockres() and waits
until others threads clear this
state;

If cannot grant this lock,
adding lock to blocked list,
and return DLM_RECOVERING;

Grant this lock and move it to grant list;

After a while, retry and
calls list_add_tail(), adding lock
to blocked list again.

Granted and blocked list of this lockres will become the following conditions:

Quoted 7 lines

    lock_res->granted.next = dlm_lock->list_head;
    lock_res->blocked.next = dlm_lock->list_head;
    dlm_lock->list_head.next = dlm_lock_resource->blocked;

When dlm_thread traverses the granted list, it will fall into an endless loop, checking dlm_lock.list_head, dlm_lock->list_head.next (i.e.lock_res->blocked), lock_res->blocked.next(i.e.dlm_lock.list_head again) .....

Thanks for your nice description of this problem and this fix looks good. Let's waiting for an ACK from either Sunil, Mark or Joel.

Quoted 39 lines · On Sun, 23 Jun 2013 18:39:16 +0800 Jeff Liu <[email protected]> wrote:

Hi Jiufei,

Quoted 27 lines · Xue jiufei

Function dlmlock_master() returns DLM_RECOVERING/DLM_MIGRATING/ DLM_FORWAR after adding lock to blocked list if lockres has the state DLM_LOCK_RES_RECOVERING/DLM_LOCK_RES_MIGRATING/ DLM_LOCK_RES_IN_PROGRESS. so it will retry in dlmlock(). And this may cause dlm_thread fall into an infinite loop

Thread1 dlm_thread
calls dlm_lock->dlmlock_master,
if lockresA is in state
DLM_LOCK_RES_RECOVERING, calls
__dlm_wait_on_lockres() and waits
until others threads clear this
state;

If cannot grant this lock,
adding lock to blocked list,
and return DLM_RECOVERING;

Grant this lock and move it to grant list;

After a while, retry and
calls list_add_tail(), adding lock
to blocked list again.

Granted and blocked list of this lockres will become the following conditions:

Quoted 7 lines

    lock_res->granted.next = dlm_lock->list_head;
    lock_res->blocked.next = dlm_lock->list_head;
    dlm_lock->list_head.next = dlm_lock_resource->blocked;

When dlm_thread traverses the granted list, it will fall into an endless loop, checking dlm_lock.list_head, dlm_lock->list_head.next (i.e.lock_res->blocked), lock_res->blocked.next(i.e.dlm_lock.list_head again) .....

Quoted 2 lines

Thanks for your nice description of this problem and this fix looks good. Let's waiting for an ACK from either Sunil, Mark or Joel.

Still waiting ;)

From: Xue jiufei <[email protected]>
Subject: ocfs2: dlmlock_master() should return DLM_NORMAL after adding lock to blocked list

dlmlock_master() returns DLM_RECOVERING/DLM_MIGRATING/ DLM_FORWAR after adding lock to blocked list if lockres has the state DLM_LOCK_RES_RECOVERING/DLM_LOCK_RES_MIGRATING/ DLM_LOCK_RES_IN_PROGRESS. so it will retry in dlmlock(). And this may cause dlm_thread fall into an infinite loop

Thread1 dlm_thread
calls dlm_lock->dlmlock_master,
if lockresA is in state
DLM_LOCK_RES_RECOVERING, calls
__dlm_wait_on_lockres() and waits
until others threads clear this
state;

If cannot grant this lock,
adding lock to blocked list,
and return DLM_RECOVERING;

Grant this lock and move it to grant list;

After a while, retry and
calls list_add_tail(), adding lock
to blocked list again.

Granted and blocked list of this lockres will become the following conditions:

    lock_res->granted.next = dlm_lock->list_head;
    lock_res->blocked.next = dlm_lock->list_head;
    dlm_lock->list_head.next = dlm_lock_resource->blocked;

When dlm_thread traverses the granted list, it will fall into an endless loop, checking dlm_lock.list_head, dlm_lock->list_head.next (i.e.lock_res->blocked), lock_res->blocked.next(i.e.dlm_lock.list_head again) .....

Signed-off-by joyce <[email protected]>

Reviewed-by jensen <[email protected]>

Cc Jeff Liu <[email protected]>

Cc Sunil Mushran <[email protected]>

Cc Mark Fasheh <[email protected]>

Cc Joel Becker <[email protected]>

Signed-off-by Andrew Morton <[email protected]>

---

 fs/ocfs2/dlm/dlmlock.c |    1 +
 1 file changed, 1 insertion(+)

diff -puN fs/ocfs2/dlm/dlmlock.c~ocfs2-dlmlock_master-should-return-dlm_normal-after-adding-lock-to-blocked-list fs/ocfs2/dlm/dlmlock.c
---
a/fs/ocfs2/dlm/dlmlock.c~ocfs2-dlmlock_master-should-return-dlm_normal-after-adding-lock-to-blocked-list +++ a/fs/ocfs2/dlm/dlmlock.c

fs/ocfs2/dlm/dlmlock.c renamed from fs/ocfs2/dlm/dlmlock.c~ocfs2-dlmlock_master-should-return-dlm_normal-after-adding-lock-to-blocked-list +1 -0

@@ -178,6 +178,7 @@ static enum dlm_status dlmlock_master(st
                                     lock->ml.node);
                        }
                } else {
+                       status = DLM_NORMAL;
                        dlm_lock_get(lock);
                        list_add_tail(&lock->list, &res->blocked);
                        kick_thread = 1;

_

_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel at oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel