Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 4670c46d authored by Joel Becker's avatar Joel Becker Committed by Mark Fasheh
Browse files

ocfs2: Introduce the new ocfs2_cluster_connect/disconnect() API. 



This step introduces a cluster stack agnostic API for initializing and
exiting.  fs/ocfs2/dlmglue.c no longer uses o2cb/o2dlm knowledge to
connect to the stack.  It is all handled in stackglue.c.

heartbeat.c no longer needs to know how it gets called.
ocfs2_do_node_down() is now a clean recovery trigger.

The big gotcha is the ordering of initializations and de-initializations done
underneath ocfs2_cluster_connect().  ocfs2_dlm_init() used to do all
o2dlm initialization in one block.  Thus, the o2dlm functionality of
ocfs2_cluster_connect() is very straightforward.  ocfs2_dlm_shutdown(),
however, did a few things between de-registration of the eviction
callback and actually shutting down the domain.  Now de-registration and
shutdown of the domain are wrapped within the single
ocfs2_cluster_disconnect() call.  I've checked the code paths to make
sure we can safely tear down things in ocfs2_dlm_shutdown() before
calling ocfs2_cluster_disconnect().  The filesystem has already set
itself to ignore the callback.

Signed-off-by: default avatarJoel Becker <joel.becker@oracle.com>
Signed-off-by: default avatarMark Fasheh <mfasheh@suse.com>
parent 8f2c9c1b

fs/ocfs2/dlmglue.c

+48 −49
Original line number Diff line number Diff line
@@ -27,7 +27,6 @@ 
#include <linux/slab.h>

#include <linux/highmem.h>

#include <linux/mm.h>

#include <linux/crc32.h>

#include <linux/kthread.h>

#include <linux/pagemap.h>

#include <linux/debugfs.h>
@@ -259,31 +258,6 @@ static struct ocfs2_lock_res_ops ocfs2_flock_lops = { 
	.flags		= 0,

};



/*

 * This is the filesystem locking protocol version.

 *

 * Whenever the filesystem does new things with locks (adds or removes a

 * lock, orders them differently, does different things underneath a lock),

 * the version must be changed.  The protocol is negotiated when joining

 * the dlm domain.  A node may join the domain if its major version is

 * identical to all other nodes and its minor version is greater than

 * or equal to all other nodes.  When its minor version is greater than

 * the other nodes, it will run at the minor version specified by the

 * other nodes.

 *

 * If a locking change is made that will not be compatible with older

 * versions, the major number must be increased and the minor version set

 * to zero.  If a change merely adds a behavior that can be disabled when

 * speaking to older versions, the minor version must be increased.  If a

 * change adds a fully backwards compatible change (eg, LVB changes that

 * are just ignored by older versions), the version does not need to be

 * updated.

 */

const struct dlm_protocol_version ocfs2_locking_protocol = {

	.pv_major = OCFS2_LOCKING_PROTOCOL_MAJOR,

	.pv_minor = OCFS2_LOCKING_PROTOCOL_MINOR,

};



static inline int ocfs2_is_inode_lock(struct ocfs2_lock_res *lockres)

{

	return lockres->l_type == OCFS2_LOCK_TYPE_META ||
@@ -886,7 +860,7 @@ static int ocfs2_lock_create(struct ocfs2_super *osb, 
	lockres_or_flags(lockres, OCFS2_LOCK_BUSY);

	spin_unlock_irqrestore(&lockres->l_lock, flags);



	ret = ocfs2_dlm_lock(osb->dlm,

	ret = ocfs2_dlm_lock(osb->cconn,

			     level,

			     &lockres->l_lksb,

			     dlm_flags,
@@ -1085,7 +1059,7 @@ again: 
		     lockres->l_name, lockres->l_level, level);



		/* call dlm_lock to upgrade lock now */

		ret = ocfs2_dlm_lock(osb->dlm,

		ret = ocfs2_dlm_lock(osb->cconn,

				     level,

				     &lockres->l_lksb,

				     lkm_flags,
@@ -1492,7 +1466,7 @@ int ocfs2_file_lock(struct file *file, int ex, int trylock) 
	lockres_add_mask_waiter(lockres, &mw, OCFS2_LOCK_BUSY, 0);

	spin_unlock_irqrestore(&lockres->l_lock, flags);



	ret = ocfs2_dlm_lock(osb->dlm, level, &lockres->l_lksb, lkm_flags,

	ret = ocfs2_dlm_lock(osb->cconn, level, &lockres->l_lksb, lkm_flags,

			     lockres->l_name, OCFS2_LOCK_ID_MAX_LEN - 1,

			     lockres);

	if (ret) {
@@ -2485,8 +2459,7 @@ static void ocfs2_dlm_shutdown_debug(struct ocfs2_super *osb) 
int ocfs2_dlm_init(struct ocfs2_super *osb)

{

	int status = 0;

	u32 dlm_key;

	struct dlm_ctxt *dlm = NULL;

	struct ocfs2_cluster_connection *conn = NULL;



	mlog_entry_void();
@@ -2508,26 +2481,21 @@ int ocfs2_dlm_init(struct ocfs2_super *osb) 
		goto bail;

	}



	/* used by the dlm code to make message headers unique, each

	 * node in this domain must agree on this. */

	dlm_key = crc32_le(0, osb->uuid_str, strlen(osb->uuid_str));



	/* for now, uuid == domain */

	dlm = dlm_register_domain(osb->uuid_str, dlm_key,

				  &osb->osb_locking_proto);

	if (IS_ERR(dlm)) {

		status = PTR_ERR(dlm);

	status = ocfs2_cluster_connect(osb->uuid_str,

				       strlen(osb->uuid_str),

				       ocfs2_do_node_down, osb,

				       &conn);

	if (status) {

		mlog_errno(status);

		goto bail;

	}



	dlm_register_eviction_cb(dlm, &osb->osb_eviction_cb);



local:

	ocfs2_super_lock_res_init(&osb->osb_super_lockres, osb);

	ocfs2_rename_lock_res_init(&osb->osb_rename_lockres, osb);



	osb->dlm = dlm;

	osb->cconn = conn;



	status = 0;

bail:
@@ -2545,10 +2513,14 @@ void ocfs2_dlm_shutdown(struct ocfs2_super *osb) 
{

	mlog_entry_void();



	dlm_unregister_eviction_cb(&osb->osb_eviction_cb);



	ocfs2_drop_osb_locks(osb);



	/*

	 * Now that we have dropped all locks and ocfs2_dismount_volume()

	 * has disabled recovery, the DLM won't be talking to us.  It's

	 * safe to tear things down before disconnecting the cluster.

	 */



	if (osb->dc_task) {

		kthread_stop(osb->dc_task);

		osb->dc_task = NULL;
@@ -2557,8 +2529,8 @@ void ocfs2_dlm_shutdown(struct ocfs2_super *osb) 
	ocfs2_lock_res_free(&osb->osb_super_lockres);

	ocfs2_lock_res_free(&osb->osb_rename_lockres);



	dlm_unregister_domain(osb->dlm);

	osb->dlm = NULL;

	ocfs2_cluster_disconnect(osb->cconn);

	osb->cconn = NULL;



	ocfs2_dlm_shutdown_debug(osb);
@@ -2689,7 +2661,7 @@ static int ocfs2_drop_lock(struct ocfs2_super *osb, 


	mlog(0, "lock %s\n", lockres->l_name);



	ret = ocfs2_dlm_unlock(osb->dlm, &lockres->l_lksb, lkm_flags,

	ret = ocfs2_dlm_unlock(osb->cconn, &lockres->l_lksb, lkm_flags,

			       lockres);

	if (ret) {

		ocfs2_log_dlm_error("ocfs2_dlm_unlock", ret, lockres);
@@ -2823,7 +2795,7 @@ static int ocfs2_downconvert_lock(struct ocfs2_super *osb, 
	if (lvb)

		dlm_flags |= DLM_LKF_VALBLK;



	ret = ocfs2_dlm_lock(osb->dlm,

	ret = ocfs2_dlm_lock(osb->cconn,

			     new_level,

			     &lockres->l_lksb,

			     dlm_flags,
@@ -2882,7 +2854,7 @@ static int ocfs2_cancel_convert(struct ocfs2_super *osb, 
	mlog_entry_void();

	mlog(0, "lock %s\n", lockres->l_name);



	ret = ocfs2_dlm_unlock(osb->dlm, &lockres->l_lksb,

	ret = ocfs2_dlm_unlock(osb->cconn, &lockres->l_lksb,

			       DLM_LKF_CANCEL, lockres);

	if (ret) {

		ocfs2_log_dlm_error("ocfs2_dlm_unlock", ret, lockres);
@@ -3193,7 +3165,34 @@ static int ocfs2_dentry_convert_worker(struct ocfs2_lock_res *lockres, 
	return UNBLOCK_CONTINUE_POST;

}



/*

 * This is the filesystem locking protocol.  It provides the lock handling

 * hooks for the underlying DLM.  It has a maximum version number.

 * The version number allows interoperability with systems running at

 * the same major number and an equal or smaller minor number.

 *

 * Whenever the filesystem does new things with locks (adds or removes a

 * lock, orders them differently, does different things underneath a lock),

 * the version must be changed.  The protocol is negotiated when joining

 * the dlm domain.  A node may join the domain if its major version is

 * identical to all other nodes and its minor version is greater than

 * or equal to all other nodes.  When its minor version is greater than

 * the other nodes, it will run at the minor version specified by the

 * other nodes.

 *

 * If a locking change is made that will not be compatible with older

 * versions, the major number must be increased and the minor version set

 * to zero.  If a change merely adds a behavior that can be disabled when

 * speaking to older versions, the minor version must be increased.  If a

 * change adds a fully backwards compatible change (eg, LVB changes that

 * are just ignored by older versions), the version does not need to be

 * updated.

 */

static struct ocfs2_locking_protocol lproto = {

	.lp_max_version = {

		.pv_major = OCFS2_LOCKING_PROTOCOL_MAJOR,

		.pv_minor = OCFS2_LOCKING_PROTOCOL_MINOR,

	},

	.lp_lock_ast		= ocfs2_locking_ast,

	.lp_blocking_ast	= ocfs2_blocking_ast,

	.lp_unlock_ast		= ocfs2_unlock_ast,

fs/ocfs2/dlmglue.h

+0 −1
Original line number Diff line number Diff line
@@ -117,5 +117,4 @@ void ocfs2_put_dlm_debug(struct ocfs2_dlm_debug *dlm_debug); 
void dlmglue_init_stack(void);

void dlmglue_exit_stack(void);



extern const struct dlm_protocol_version ocfs2_locking_protocol;

#endif	/* DLMGLUE_H */

fs/ocfs2/heartbeat.c

+8 −32
Original line number Diff line number Diff line
@@ -30,8 +30,6 @@ 
#include <linux/highmem.h>

#include <linux/kmod.h>



#include <dlm/dlmapi.h>



#define MLOG_MASK_PREFIX ML_SUPER

#include <cluster/masklog.h>
@@ -64,19 +62,20 @@ void ocfs2_init_node_maps(struct ocfs2_super *osb) 
	ocfs2_node_map_init(&osb->osb_recovering_orphan_dirs);

}



static void ocfs2_do_node_down(int node_num,

			       struct ocfs2_super *osb)

void ocfs2_do_node_down(int node_num, void *data)

{

	struct ocfs2_super *osb = data;



	BUG_ON(osb->node_num == node_num);



	mlog(0, "ocfs2: node down event for %d\n", node_num);



	if (!osb->dlm) {

	if (!osb->cconn) {

		/*

		 * No DLM means we're not even ready to participate yet.

		 * We check the slots after the DLM comes up, so we will

		 * notice the node death then.  We can safely ignore it

		 * here.

		 * No cluster connection means we're not even ready to

		 * participate yet.  We check the slots after the cluster

		 * comes up, so we will notice the node death then.  We

		 * can safely ignore it here.

		 */

		return;

	}
@@ -84,29 +83,6 @@ static void ocfs2_do_node_down(int node_num, 
	ocfs2_recovery_thread(osb, node_num);

}



/* Called from the dlm when it's about to evict a node. We may also

 * get a heartbeat callback later. */

static void ocfs2_dlm_eviction_cb(int node_num,

				  void *data)

{

	struct ocfs2_super *osb = (struct ocfs2_super *) data;

	struct super_block *sb = osb->sb;



	mlog(ML_NOTICE, "device (%u,%u): dlm has evicted node %d\n",

	     MAJOR(sb->s_dev), MINOR(sb->s_dev), node_num);



	ocfs2_do_node_down(node_num, osb);

}



void ocfs2_setup_hb_callbacks(struct ocfs2_super *osb)

{

	/* Not exactly a heartbeat callback, but leads to essentially

	 * the same path so we set it up here. */

	dlm_setup_eviction_cb(&osb->osb_eviction_cb,

			      ocfs2_dlm_eviction_cb,

			      osb);

}



void ocfs2_stop_heartbeat(struct ocfs2_super *osb)

{

	int ret;

fs/ocfs2/heartbeat.h

+1 −1
Original line number Diff line number Diff line
@@ -28,7 +28,7 @@ 


void ocfs2_init_node_maps(struct ocfs2_super *osb);



void ocfs2_setup_hb_callbacks(struct ocfs2_super *osb);

void ocfs2_do_node_down(int node_num, void *data);

void ocfs2_stop_heartbeat(struct ocfs2_super *osb);



/* node map functions - used to keep track of mounted and in-recovery

fs/ocfs2/ocfs2.h

+1 −3
Original line number Diff line number Diff line
@@ -248,12 +248,10 @@ struct ocfs2_super 
	struct ocfs2_alloc_stats alloc_stats;

	char dev_str[20];		/* "major,minor" of the device */



	struct dlm_ctxt *dlm;

	struct ocfs2_cluster_connection *cconn;

	struct ocfs2_lock_res osb_super_lockres;

	struct ocfs2_lock_res osb_rename_lockres;

	struct dlm_eviction_cb osb_eviction_cb;

	struct ocfs2_dlm_debug *osb_dlm_debug;

	struct dlm_protocol_version osb_locking_proto;



	struct dentry *osb_debug_root;