]> err.no Git - linux-2.6/commitdiff
[SCSI] scsi_dh: Use SCSI device handler in dm-multipath
authorChandra Seetharaman <sekharan@us.ibm.com>
Thu, 1 May 2008 21:50:11 +0000 (14:50 -0700)
committerJames Bottomley <James.Bottomley@HansenPartnership.com>
Thu, 5 Jun 2008 14:23:41 +0000 (09:23 -0500)
This patch converts dm-mpath to use scsi device handlers instead of
dm's hardware handlers.

This patch does not add any new functionality. Old behaviors remain and
userspace tools work as is except that arguments supplied with hardware
handler are ignored.

One behavioral exception is: Activation of a path is synchronous in this
patch, opposed to the older behavior of being asynchronous (changed in
patch 07: scsi_dh: Add a single threaded workqueue for initializing a path)

Note: There is no need to get a reference for the device handler module
(as it was done in the dm hardware handler case) here as the reference
is held when the device was first found. Instead we check and make sure
that support for the specified device is present at table load time.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
drivers/md/Kconfig
drivers/md/dm-mpath.c

index 610af916891ec2328ef14e20bacf99d9475fefb9..5303af55d2c7502f9c44c5934860cc66798ba8c6 100644 (file)
@@ -252,6 +252,7 @@ config DM_ZERO
 config DM_MULTIPATH
        tristate "Multipath target"
        depends on BLK_DEV_DM
+       select SCSI_DH
        ---help---
          Allow volume managers to support multipath hardware.
 
index e7ee59e655d58a66e53658794e5e0dc2b2042ed3..e54ff372d711b53e652220e8616ddf3d07995f5c 100644 (file)
@@ -20,6 +20,7 @@
 #include <linux/slab.h>
 #include <linux/time.h>
 #include <linux/workqueue.h>
+#include <scsi/scsi_dh.h>
 #include <asm/atomic.h>
 
 #define DM_MSG_PREFIX "multipath"
@@ -61,7 +62,7 @@ struct multipath {
 
        spinlock_t lock;
 
-       struct hw_handler hw_handler;
+       const char *hw_handler_name;
        unsigned nr_priority_groups;
        struct list_head priority_groups;
        unsigned pg_init_required;      /* pg_init needs calling? */
@@ -109,6 +110,7 @@ static struct kmem_cache *_mpio_cache;
 static struct workqueue_struct *kmultipathd;
 static void process_queued_ios(struct work_struct *work);
 static void trigger_event(struct work_struct *work);
+static void pg_init_done(struct dm_path *, int);
 
 
 /*-----------------------------------------------
@@ -193,18 +195,13 @@ static struct multipath *alloc_multipath(struct dm_target *ti)
 static void free_multipath(struct multipath *m)
 {
        struct priority_group *pg, *tmp;
-       struct hw_handler *hwh = &m->hw_handler;
 
        list_for_each_entry_safe(pg, tmp, &m->priority_groups, list) {
                list_del(&pg->list);
                free_priority_group(pg, m->ti);
        }
 
-       if (hwh->type) {
-               hwh->type->destroy(hwh);
-               dm_put_hw_handler(hwh->type);
-       }
-
+       kfree(m->hw_handler_name);
        mempool_destroy(m->mpio_pool);
        kfree(m);
 }
@@ -216,12 +213,10 @@ static void free_multipath(struct multipath *m)
 
 static void __switch_pg(struct multipath *m, struct pgpath *pgpath)
 {
-       struct hw_handler *hwh = &m->hw_handler;
-
        m->current_pg = pgpath->pg;
 
        /* Must we initialise the PG first, and queue I/O till it's ready? */
-       if (hwh->type && hwh->type->pg_init) {
+       if (m->hw_handler_name) {
                m->pg_init_required = 1;
                m->queue_io = 1;
        } else {
@@ -409,7 +404,6 @@ static void process_queued_ios(struct work_struct *work)
 {
        struct multipath *m =
                container_of(work, struct multipath, process_queued_ios);
-       struct hw_handler *hwh = &m->hw_handler;
        struct pgpath *pgpath = NULL;
        unsigned init_required = 0, must_queue = 1;
        unsigned long flags;
@@ -438,8 +432,11 @@ static void process_queued_ios(struct work_struct *work)
 out:
        spin_unlock_irqrestore(&m->lock, flags);
 
-       if (init_required)
-               hwh->type->pg_init(hwh, pgpath->pg->bypassed, &pgpath->path);
+       if (init_required) {
+               struct dm_path *path = &pgpath->path;
+               int ret = scsi_dh_activate(bdev_get_queue(path->dev->bdev));
+               pg_init_done(path, ret);
+       }
 
        if (!must_queue)
                dispatch_queued_ios(m);
@@ -652,8 +649,6 @@ static struct priority_group *parse_priority_group(struct arg_set *as,
 
 static int parse_hw_handler(struct arg_set *as, struct multipath *m)
 {
-       int r;
-       struct hw_handler_type *hwht;
        unsigned hw_argc;
        struct dm_target *ti = m->ti;
 
@@ -661,30 +656,18 @@ static int parse_hw_handler(struct arg_set *as, struct multipath *m)
                {0, 1024, "invalid number of hardware handler args"},
        };
 
-       r = read_param(_params, shift(as), &hw_argc, &ti->error);
-       if (r)
+       if (read_param(_params, shift(as), &hw_argc, &ti->error))
                return -EINVAL;
 
        if (!hw_argc)
                return 0;
 
-       hwht = dm_get_hw_handler(shift(as));
-       if (!hwht) {
+       m->hw_handler_name = kstrdup(shift(as), GFP_KERNEL);
+       request_module("scsi_dh_%s", m->hw_handler_name);
+       if (scsi_dh_handler_exist(m->hw_handler_name) == 0) {
                ti->error = "unknown hardware handler type";
                return -EINVAL;
        }
-
-       m->hw_handler.md = dm_table_get_md(ti->table);
-       dm_put(m->hw_handler.md);
-
-       r = hwht->create(&m->hw_handler, hw_argc - 1, as->argv);
-       if (r) {
-               dm_put_hw_handler(hwht);
-               ti->error = "hardware handler constructor failed";
-               return r;
-       }
-
-       m->hw_handler.type = hwht;
        consume(as, hw_argc - 1);
 
        return 0;
@@ -1063,14 +1046,74 @@ void dm_pg_init_complete(struct dm_path *path, unsigned err_flags)
        spin_unlock_irqrestore(&m->lock, flags);
 }
 
+static void pg_init_done(struct dm_path *path, int errors)
+{
+       struct pgpath *pgpath = path_to_pgpath(path);
+       struct priority_group *pg = pgpath->pg;
+       struct multipath *m = pg->m;
+       unsigned long flags;
+
+       /* device or driver problems */
+       switch (errors) {
+       case SCSI_DH_OK:
+               break;
+       case SCSI_DH_NOSYS:
+               if (!m->hw_handler_name) {
+                       errors = 0;
+                       break;
+               }
+               DMERR("Cannot failover device because scsi_dh_%s was not "
+                     "loaded.", m->hw_handler_name);
+               /*
+                * Fail path for now, so we do not ping pong
+                */
+               fail_path(pgpath);
+               break;
+       case SCSI_DH_DEV_TEMP_BUSY:
+               /*
+                * Probably doing something like FW upgrade on the
+                * controller so try the other pg.
+                */
+               bypass_pg(m, pg, 1);
+               break;
+       /* TODO: For SCSI_DH_RETRY we should wait a couple seconds */
+       case SCSI_DH_RETRY:
+       case SCSI_DH_IMM_RETRY:
+       case SCSI_DH_RES_TEMP_UNAVAIL:
+               if (pg_init_limit_reached(m, pgpath))
+                       fail_path(pgpath);
+               errors = 0;
+               break;
+       default:
+               /*
+                * We probably do not want to fail the path for a device
+                * error, but this is what the old dm did. In future
+                * patches we can do more advanced handling.
+                */
+               fail_path(pgpath);
+       }
+
+       spin_lock_irqsave(&m->lock, flags);
+       if (errors) {
+               DMERR("Could not failover device. Error %d.", errors);
+               m->current_pgpath = NULL;
+               m->current_pg = NULL;
+       } else if (!m->pg_init_required) {
+               m->queue_io = 0;
+               pg->bypassed = 0;
+       }
+
+       m->pg_init_in_progress = 0;
+       queue_work(kmultipathd, &m->process_queued_ios);
+       spin_unlock_irqrestore(&m->lock, flags);
+}
+
 /*
  * end_io handling
  */
 static int do_end_io(struct multipath *m, struct bio *bio,
                     int error, struct dm_mpath_io *mpio)
 {
-       struct hw_handler *hwh = &m->hw_handler;
-       unsigned err_flags = MP_FAIL_PATH;      /* Default behavior */
        unsigned long flags;
 
        if (!error)
@@ -1097,19 +1140,8 @@ static int do_end_io(struct multipath *m, struct bio *bio,
        }
        spin_unlock_irqrestore(&m->lock, flags);
 
-       if (hwh->type && hwh->type->error)
-               err_flags = hwh->type->error(hwh, bio);
-
-       if (mpio->pgpath) {
-               if (err_flags & MP_FAIL_PATH)
-                       fail_path(mpio->pgpath);
-
-               if (err_flags & MP_BYPASS_PG)
-                       bypass_pg(m, mpio->pgpath->pg, 1);
-       }
-
-       if (err_flags & MP_ERROR_IO)
-               return -EIO;
+       if (mpio->pgpath)
+               fail_path(mpio->pgpath);
 
       requeue:
        dm_bio_restore(&mpio->details, bio);
@@ -1194,7 +1226,6 @@ static int multipath_status(struct dm_target *ti, status_type_t type,
        int sz = 0;
        unsigned long flags;
        struct multipath *m = (struct multipath *) ti->private;
-       struct hw_handler *hwh = &m->hw_handler;
        struct priority_group *pg;
        struct pgpath *p;
        unsigned pg_num;
@@ -1214,12 +1245,10 @@ static int multipath_status(struct dm_target *ti, status_type_t type,
                        DMEMIT("pg_init_retries %u ", m->pg_init_retries);
        }
 
-       if (hwh->type && hwh->type->status)
-               sz += hwh->type->status(hwh, type, result + sz, maxlen - sz);
-       else if (!hwh->type || type == STATUSTYPE_INFO)
+       if (!m->hw_handler_name || type == STATUSTYPE_INFO)
                DMEMIT("0 ");
        else
-               DMEMIT("1 %s ", hwh->type->name);
+               DMEMIT("1 %s ", m->hw_handler_name);
 
        DMEMIT("%u ", m->nr_priority_groups);