Merge branch 'for-linus' of git://git.open-osd.org/linux-open-osd
[pandora-kernel.git] / fs / exofs / super.c
index 2748940..057b237 100644 (file)
@@ -266,7 +266,7 @@ static int __sbi_read_stats(struct exofs_sb_info *sbi)
        struct ore_io_state *ios;
        int ret;
 
-       ret = ore_get_io_state(&sbi->layout, &sbi->comps, &ios);
+       ret = ore_get_io_state(&sbi->layout, &sbi->oc, &ios);
        if (unlikely(ret)) {
                EXOFS_ERR("%s: ore_get_io_state failed.\n", __func__);
                return ret;
@@ -321,7 +321,7 @@ int exofs_sbi_write_stats(struct exofs_sb_info *sbi)
        struct ore_io_state *ios;
        int ret;
 
-       ret = ore_get_io_state(&sbi->layout, &sbi->comps, &ios);
+       ret = ore_get_io_state(&sbi->layout, &sbi->oc, &ios);
        if (unlikely(ret)) {
                EXOFS_ERR("%s: ore_get_io_state failed.\n", __func__);
                return ret;
@@ -355,12 +355,12 @@ static const struct export_operations exofs_export_ops;
 /*
  * Write the superblock to the OSD
  */
-int exofs_sync_fs(struct super_block *sb, int wait)
+static int exofs_sync_fs(struct super_block *sb, int wait)
 {
        struct exofs_sb_info *sbi;
        struct exofs_fscb *fscb;
        struct ore_comp one_comp;
-       struct ore_components comps;
+       struct ore_components oc;
        struct ore_io_state *ios;
        int ret = -ENOMEM;
 
@@ -378,9 +378,9 @@ int exofs_sync_fs(struct super_block *sb, int wait)
         * the writeable info is set in exofs_sbi_write_stats() above.
         */
 
-       exofs_init_comps(&comps, &one_comp, sbi, EXOFS_SUPER_ID);
+       exofs_init_comps(&oc, &one_comp, sbi, EXOFS_SUPER_ID);
 
-       ret = ore_get_io_state(&sbi->layout, &comps, &ios);
+       ret = ore_get_io_state(&sbi->layout, &oc, &ios);
        if (unlikely(ret))
                goto out;
 
@@ -429,19 +429,20 @@ static void _exofs_print_device(const char *msg, const char *dev_path,
                msg, dev_path ?: "", odi->osdname, _LLU(pid));
 }
 
-void exofs_free_sbi(struct exofs_sb_info *sbi)
+static void exofs_free_sbi(struct exofs_sb_info *sbi)
 {
-       while (sbi->comps.numdevs) {
-               int i = --sbi->comps.numdevs;
-               struct osd_dev *od = sbi->comps.ods[i];
+       unsigned numdevs = sbi->oc.numdevs;
+
+       while (numdevs) {
+               unsigned i = --numdevs;
+               struct osd_dev *od = ore_comp_dev(&sbi->oc, i);
 
                if (od) {
-                       sbi->comps.ods[i] = NULL;
+                       ore_comp_set_dev(&sbi->oc, i, NULL);
                        osduld_put_device(od);
                }
        }
-       if (sbi->comps.ods != sbi->_min_one_dev)
-               kfree(sbi->comps.ods);
+       kfree(sbi->oc.ods);
        kfree(sbi);
 }
 
@@ -468,7 +469,7 @@ static void exofs_put_super(struct super_block *sb)
                                  msecs_to_jiffies(100));
        }
 
-       _exofs_print_device("Unmounting", NULL, sbi->comps.ods[0],
+       _exofs_print_device("Unmounting", NULL, ore_comp_dev(&sbi->oc, 0),
                            sbi->one_comp.obj.partition);
 
        bdi_destroy(&sbi->bdi);
@@ -479,76 +480,20 @@ static void exofs_put_super(struct super_block *sb)
 static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs,
                                    struct exofs_device_table *dt)
 {
-       u64 stripe_length;
+       int ret;
 
-       sbi->data_map.odm_num_comps   =
-                               le32_to_cpu(dt->dt_data_map.cb_num_comps);
-       sbi->data_map.odm_stripe_unit =
+       sbi->layout.stripe_unit =
                                le64_to_cpu(dt->dt_data_map.cb_stripe_unit);
-       sbi->data_map.odm_group_width =
+       sbi->layout.group_width =
                                le32_to_cpu(dt->dt_data_map.cb_group_width);
-       sbi->data_map.odm_group_depth =
+       sbi->layout.group_depth =
                                le32_to_cpu(dt->dt_data_map.cb_group_depth);
-       sbi->data_map.odm_mirror_cnt  =
-                               le32_to_cpu(dt->dt_data_map.cb_mirror_cnt);
-       sbi->data_map.odm_raid_algorithm  =
+       sbi->layout.mirrors_p1  =
+                               le32_to_cpu(dt->dt_data_map.cb_mirror_cnt) + 1;
+       sbi->layout.raid_algorithm  =
                                le32_to_cpu(dt->dt_data_map.cb_raid_algorithm);
 
-/* FIXME: Only raid0 for now. if not so, do not mount */
-       if (sbi->data_map.odm_num_comps != numdevs) {
-               EXOFS_ERR("odm_num_comps(%u) != numdevs(%u)\n",
-                         sbi->data_map.odm_num_comps, numdevs);
-               return -EINVAL;
-       }
-       if (sbi->data_map.odm_raid_algorithm != PNFS_OSD_RAID_0) {
-               EXOFS_ERR("Only RAID_0 for now\n");
-               return -EINVAL;
-       }
-       if (0 != (numdevs % (sbi->data_map.odm_mirror_cnt + 1))) {
-               EXOFS_ERR("Data Map wrong, numdevs=%d mirrors=%d\n",
-                         numdevs, sbi->data_map.odm_mirror_cnt);
-               return -EINVAL;
-       }
-
-       if (0 != (sbi->data_map.odm_stripe_unit & ~PAGE_MASK)) {
-               EXOFS_ERR("Stripe Unit(0x%llx)"
-                         " must be Multples of PAGE_SIZE(0x%lx)\n",
-                         _LLU(sbi->data_map.odm_stripe_unit), PAGE_SIZE);
-               return -EINVAL;
-       }
-
-       sbi->layout.stripe_unit = sbi->data_map.odm_stripe_unit;
-       sbi->layout.mirrors_p1 = sbi->data_map.odm_mirror_cnt + 1;
-
-       if (sbi->data_map.odm_group_width) {
-               sbi->layout.group_width = sbi->data_map.odm_group_width;
-               sbi->layout.group_depth = sbi->data_map.odm_group_depth;
-               if (!sbi->layout.group_depth) {
-                       EXOFS_ERR("group_depth == 0 && group_width != 0\n");
-                       return -EINVAL;
-               }
-               sbi->layout.group_count = sbi->data_map.odm_num_comps /
-                                               sbi->layout.mirrors_p1 /
-                                               sbi->data_map.odm_group_width;
-       } else {
-               if (sbi->data_map.odm_group_depth) {
-                       printk(KERN_NOTICE "Warning: group_depth ignored "
-                               "group_width == 0 && group_depth == %d\n",
-                               sbi->data_map.odm_group_depth);
-                       sbi->data_map.odm_group_depth = 0;
-               }
-               sbi->layout.group_width = sbi->data_map.odm_num_comps /
-                                                       sbi->layout.mirrors_p1;
-               sbi->layout.group_depth = -1;
-               sbi->layout.group_count = 1;
-       }
-
-       stripe_length = (u64)sbi->layout.group_width * sbi->layout.stripe_unit;
-       if (stripe_length >= (1ULL << 32)) {
-               EXOFS_ERR("Total Stripe length(0x%llx)"
-                         " >= 32bit is not supported\n", _LLU(stripe_length));
-               return -EINVAL;
-       }
+       ret = ore_verify_layout(numdevs, &sbi->layout);
 
        EXOFS_DBGMSG("exofs: layout: "
                "num_comps=%u stripe_unit=0x%x group_width=%u "
@@ -558,8 +503,8 @@ static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs,
                sbi->layout.group_width,
                _LLU(sbi->layout.group_depth),
                sbi->layout.mirrors_p1,
-               sbi->data_map.odm_raid_algorithm);
-       return 0;
+               sbi->layout.raid_algorithm);
+       return ret;
 }
 
 static unsigned __ra_pages(struct ore_layout *layout)
@@ -605,12 +550,40 @@ static int exofs_devs_2_odi(struct exofs_dt_device_info *dt_dev,
        return !(odi->systemid_len || odi->osdname_len);
 }
 
+int __alloc_dev_table(struct exofs_sb_info *sbi, unsigned numdevs,
+                     struct exofs_dev **peds)
+{
+       struct __alloc_ore_devs_and_exofs_devs {
+               /* Twice bigger table: See exofs_init_comps() and comment at
+                * exofs_read_lookup_dev_table()
+                */
+               struct ore_dev *oreds[numdevs * 2 - 1];
+               struct exofs_dev eds[numdevs];
+       } *aoded;
+       struct exofs_dev *eds;
+       unsigned i;
+
+       aoded = kzalloc(sizeof(*aoded), GFP_KERNEL);
+       if (unlikely(!aoded)) {
+               EXOFS_ERR("ERROR: faild allocating Device array[%d]\n",
+                         numdevs);
+               return -ENOMEM;
+       }
+
+       sbi->oc.ods = aoded->oreds;
+       *peds = eds = aoded->eds;
+       for (i = 0; i < numdevs; ++i)
+               aoded->oreds[i] = &eds[i].ored;
+       return 0;
+}
+
 static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
                                       struct osd_dev *fscb_od,
                                       unsigned table_count)
 {
        struct ore_comp comp;
        struct exofs_device_table *dt;
+       struct exofs_dev *eds;
        unsigned table_bytes = table_count * sizeof(dt->dt_dev_table[0]) +
                                             sizeof(*dt);
        unsigned numdevs, i;
@@ -623,7 +596,7 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
                return -ENOMEM;
        }
 
-       sbi->comps.numdevs = 0;
+       sbi->oc.numdevs = 0;
 
        comp.obj.partition = sbi->one_comp.obj.partition;
        comp.obj.id = EXOFS_DEVTABLE_ID;
@@ -647,20 +620,16 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
        if (unlikely(ret))
                goto out;
 
-       if (likely(numdevs > 1)) {
-               unsigned size = numdevs * sizeof(sbi->comps.ods[0]);
-
-               /* Twice bigger table: See exofs_init_comps() and below
-                * comment
-                */
-               sbi->comps.ods = kzalloc(size + size - 1, GFP_KERNEL);
-               if (unlikely(!sbi->comps.ods)) {
-                       EXOFS_ERR("ERROR: faild allocating Device array[%d]\n",
-                                 numdevs);
-                       ret = -ENOMEM;
-                       goto out;
-               }
-       }
+       ret = __alloc_dev_table(sbi, numdevs, &eds);
+       if (unlikely(ret))
+               goto out;
+       /* exofs round-robins the device table view according to inode
+        * number. We hold a: twice bigger table hence inodes can point
+        * to any device and have a sequential view of the table
+        * starting at this device. See exofs_init_comps()
+        */
+       memcpy(&sbi->oc.ods[numdevs], &sbi->oc.ods[0],
+               (numdevs - 1) * sizeof(sbi->oc.ods[0]));
 
        for (i = 0; i < numdevs; i++) {
                struct exofs_fscb fscb;
@@ -676,13 +645,16 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
                printk(KERN_NOTICE "Add device[%d]: osd_name-%s\n",
                       i, odi.osdname);
 
+               /* the exofs id is currently the table index */
+               eds[i].did = i;
+
                /* On all devices the device table is identical. The user can
                 * specify any one of the participating devices on the command
                 * line. We always keep them in device-table order.
                 */
                if (fscb_od && osduld_device_same(fscb_od, &odi)) {
-                       sbi->comps.ods[i] = fscb_od;
-                       ++sbi->comps.numdevs;
+                       eds[i].ored.od = fscb_od;
+                       ++sbi->oc.numdevs;
                        fscb_od = NULL;
                        continue;
                }
@@ -695,8 +667,8 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
                        goto out;
                }
 
-               sbi->comps.ods[i] = od;
-               ++sbi->comps.numdevs;
+               eds[i].ored.od = od;
+               ++sbi->oc.numdevs;
 
                /* Read the fscb of the other devices to make sure the FS
                 * partition is there.
@@ -718,21 +690,10 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
 
 out:
        kfree(dt);
-       if (likely(!ret)) {
-               unsigned numdevs = sbi->comps.numdevs;
-
-               if (unlikely(fscb_od)) {
+       if (unlikely(fscb_od && !ret)) {
                        EXOFS_ERR("ERROR: Bad device-table container device not present\n");
                        osduld_put_device(fscb_od);
                        return -EINVAL;
-               }
-               /* exofs round-robins the device table view according to inode
-                * number. We hold a: twice bigger table hence inodes can point
-                * to any device and have a sequential view of the table
-                * starting at this device. See exofs_init_comps()
-                */
-               for (i = 0; i < numdevs - 1; ++i)
-                       sbi->comps.ods[i + numdevs] = sbi->comps.ods[i];
        }
        return ret;
 }
@@ -783,10 +744,9 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
        sbi->one_comp.obj.partition = opts->pid;
        sbi->one_comp.obj.id = 0;
        exofs_make_credential(sbi->one_comp.cred, &sbi->one_comp.obj);
-       sbi->comps.numdevs = 1;
-       sbi->comps.single_comp = EC_SINGLE_COMP;
-       sbi->comps.comps = &sbi->one_comp;
-       sbi->comps.ods = sbi->_min_one_dev;
+       sbi->oc.numdevs = 1;
+       sbi->oc.single_comp = EC_SINGLE_COMP;
+       sbi->oc.comps = &sbi->one_comp;
 
        /* fill in some other data by hand */
        memset(sb->s_id, 0, sizeof(sb->s_id));
@@ -835,7 +795,13 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
                if (unlikely(ret))
                        goto free_sbi;
        } else {
-               sbi->comps.ods[0] = od;
+               struct exofs_dev *eds;
+
+               ret = __alloc_dev_table(sbi, 1, &eds);
+               if (unlikely(ret))
+                       goto free_sbi;
+
+               ore_comp_set_dev(&sbi->oc, 0, od);
        }
 
        __sbi_read_stats(sbi);
@@ -875,7 +841,8 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
                goto free_sbi;
        }
 
-       _exofs_print_device("Mounting", opts->dev_name, sbi->comps.ods[0],
+       _exofs_print_device("Mounting", opts->dev_name,
+                           ore_comp_dev(&sbi->oc, 0),
                            sbi->one_comp.obj.partition);
        return 0;
 
@@ -924,7 +891,7 @@ static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf)
        uint64_t used = ULLONG_MAX;
        int ret;
 
-       ret = ore_get_io_state(&sbi->layout, &sbi->comps, &ios);
+       ret = ore_get_io_state(&sbi->layout, &sbi->oc, &ios);
        if (ret) {
                EXOFS_DBGMSG("ore_get_io_state failed.\n");
                return ret;
@@ -981,7 +948,7 @@ static const struct super_operations exofs_sops = {
  * EXPORT OPERATIONS
  *****************************************************************************/
 
-struct dentry *exofs_get_parent(struct dentry *child)
+static struct dentry *exofs_get_parent(struct dentry *child)
 {
        unsigned long ino = exofs_parent_ino(child);