summaryrefslogtreecommitdiffstats
path: root/fs/exofs/super.c
diff options
context:
space:
mode:
authorBoaz Harrosh <bharrosh@panasas.com>2011-08-06 00:06:04 +0200
committerBoaz Harrosh <bharrosh@panasas.com>2011-08-07 04:35:32 +0200
commit9e9db45649eb5d3ee5622fdad741914ecf1016a0 (patch)
tree19ab9e1431e3d6535cef3f2cba6fcff12bb6ba6c /fs/exofs/super.c
parentexofs: Move exofs specific osd operations out of ios.c (diff)
downloadlinux-9e9db45649eb5d3ee5622fdad741914ecf1016a0.tar.xz
linux-9e9db45649eb5d3ee5622fdad741914ecf1016a0.zip
exofs: ios: Move to a per inode components & device-table
Exofs raid engine was saving on memory space by having a single layout-info, single pid, and a single device-table, global to the filesystem. Then passing a credential and object_id info at the io_state level, private for each inode. It would also devise this contraption of rotating the device table view for each inode->ino to spread out the device usage. This is not compatible with the pnfs-objects standard, demanding that each inode can have it's own layout-info, device-table, and each object component it's own pid, oid and creds. So: Bring exofs raid engine to be usable for generic pnfs-objects use by: * Define an exofs_comp structure that holds obj_id and credential info. * Break up exofs_layout struct to an exofs_components structure that holds a possible array of exofs_comp and the array of devices + the size of the arrays. * Add a "comps" parameter to get_io_state() that specifies the ids creds and device array to use for each IO. This enables to keep the layout global, but the device-table view, creds and IDs at the inode level. It only adds two 64bit to each inode, since some of these members already existed in another form. * ios raid engine now access layout-info and comps-info through the passed pointers. Everything is pre-prepared by caller for generic access of these structures and arrays. At the exofs Level: * Super block holds an exofs_components struct that holds the device array, previously in layout. The devices there are in device-table order. The device-array is twice bigger and repeats the device-table twice so now each inode's device array can point to a random device and have a round-robin view of the table, making it compatible to previous exofs versions. * Each inode has an exofs_components struct that is initialized at load time, with it's own view of the device table IDs and creds. When doing IO this gets passed to the io_state together with the layout. While preforming this change. Bugs where found where credentials with the wrong IDs where used to access the different SB objects (super.c). As well as some dead code. It was never noticed because the target we use does not check the credentials. Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Diffstat (limited to '')
-rw-r--r--fs/exofs/super.c136
1 files changed, 70 insertions, 66 deletions
diff --git a/fs/exofs/super.c b/fs/exofs/super.c
index 8783f3d33c4a..4d6877967c87 100644
--- a/fs/exofs/super.c
+++ b/fs/exofs/super.c
@@ -253,22 +253,6 @@ out:
return ret;
}
-unsigned exofs_layout_od_id(struct exofs_layout *layout,
- osd_id obj_no, unsigned layout_index)
-{
-/* switch (layout->lay_func) {
- case LAYOUT_MOVING_WINDOW:
- {*/
- unsigned dev_mod = obj_no;
-
- return (layout_index + dev_mod * layout->mirrors_p1) %
- layout->s_numdevs;
-/* }
- case LAYOUT_FUNC_IMPLICT:
- return layout->devs[layout_index];
- }*/
-}
-
static const struct osd_attr g_attr_sb_stats = ATTR_DEF(
EXOFS_APAGE_SB_DATA,
EXOFS_ATTR_SB_STATS,
@@ -282,14 +266,12 @@ static int __sbi_read_stats(struct exofs_sb_info *sbi)
struct exofs_io_state *ios;
int ret;
- ret = exofs_get_io_state(&sbi->layout, &ios);
+ ret = exofs_get_io_state(&sbi->layout, &sbi->comps, &ios);
if (unlikely(ret)) {
EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__);
return ret;
}
- ios->cred = sbi->s_cred;
-
ios->in_attr = attrs;
ios->in_attr_len = ARRAY_SIZE(attrs);
@@ -339,7 +321,7 @@ int exofs_sbi_write_stats(struct exofs_sb_info *sbi)
struct exofs_io_state *ios;
int ret;
- ret = exofs_get_io_state(&sbi->layout, &ios);
+ ret = exofs_get_io_state(&sbi->layout, &sbi->comps, &ios);
if (unlikely(ret)) {
EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__);
return ret;
@@ -349,7 +331,7 @@ int exofs_sbi_write_stats(struct exofs_sb_info *sbi)
sbi->s_ess.s_numfiles = cpu_to_le64(sbi->s_numfiles);
attrs[0].val_ptr = &sbi->s_ess;
- ios->cred = sbi->s_cred;
+
ios->done = stats_done;
ios->private = sbi;
ios->out_attr = attrs;
@@ -377,6 +359,8 @@ int exofs_sync_fs(struct super_block *sb, int wait)
{
struct exofs_sb_info *sbi;
struct exofs_fscb *fscb;
+ struct exofs_comp one_comp;
+ struct exofs_components comps;
struct exofs_io_state *ios;
int ret = -ENOMEM;
@@ -393,7 +377,10 @@ int exofs_sync_fs(struct super_block *sb, int wait)
* version). Otherwise the exofs_fscb is read-only from mkfs time. All
* the writeable info is set in exofs_sbi_write_stats() above.
*/
- ret = exofs_get_io_state(&sbi->layout, &ios);
+
+ exofs_init_comps(&comps, &one_comp, sbi, EXOFS_SUPER_ID);
+
+ ret = exofs_get_io_state(&sbi->layout, &comps, &ios);
if (unlikely(ret))
goto out;
@@ -407,10 +394,8 @@ int exofs_sync_fs(struct super_block *sb, int wait)
fscb->s_newfs = 0;
fscb->s_version = EXOFS_FSCB_VER;
- ios->obj.id = EXOFS_SUPER_ID;
ios->offset = 0;
ios->kern_buff = fscb;
- ios->cred = sbi->s_cred;
ret = exofs_sbi_write(ios);
if (unlikely(ret))
@@ -446,17 +431,17 @@ static void _exofs_print_device(const char *msg, const char *dev_path,
void exofs_free_sbi(struct exofs_sb_info *sbi)
{
- while (sbi->layout.s_numdevs) {
- int i = --sbi->layout.s_numdevs;
- struct osd_dev *od = sbi->layout.s_ods[i];
+ while (sbi->comps.numdevs) {
+ int i = --sbi->comps.numdevs;
+ struct osd_dev *od = sbi->comps.ods[i];
if (od) {
- sbi->layout.s_ods[i] = NULL;
+ sbi->comps.ods[i] = NULL;
osduld_put_device(od);
}
}
- if (sbi->layout.s_ods != sbi->_min_one_dev)
- kfree(sbi->layout.s_ods);
+ if (sbi->comps.ods != sbi->_min_one_dev)
+ kfree(sbi->comps.ods);
kfree(sbi);
}
@@ -483,8 +468,8 @@ static void exofs_put_super(struct super_block *sb)
msecs_to_jiffies(100));
}
- _exofs_print_device("Unmounting", NULL, sbi->layout.s_ods[0],
- sbi->layout.s_pid);
+ _exofs_print_device("Unmounting", NULL, sbi->comps.ods[0],
+ sbi->one_comp.obj.partition);
bdi_destroy(&sbi->bdi);
exofs_free_sbi(sbi);
@@ -624,8 +609,7 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
struct osd_dev *fscb_od,
unsigned table_count)
{
- struct osd_obj_id obj = {.partition = sbi->layout.s_pid,
- .id = EXOFS_DEVTABLE_ID};
+ struct exofs_comp comp;
struct exofs_device_table *dt;
unsigned table_bytes = table_count * sizeof(dt->dt_dev_table[0]) +
sizeof(*dt);
@@ -639,8 +623,14 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
return -ENOMEM;
}
- sbi->layout.s_numdevs = 0;
- ret = exofs_read_kern(fscb_od, sbi->s_cred, &obj, 0, dt, table_bytes);
+ sbi->comps.numdevs = 0;
+
+ comp.obj.partition = sbi->one_comp.obj.partition;
+ comp.obj.id = EXOFS_DEVTABLE_ID;
+ exofs_make_credential(comp.cred, &comp.obj);
+
+ ret = exofs_read_kern(fscb_od, comp.cred, &comp.obj, 0, dt,
+ table_bytes);
if (unlikely(ret)) {
EXOFS_ERR("ERROR: reading device table\n");
goto out;
@@ -658,10 +648,13 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
goto out;
if (likely(numdevs > 1)) {
- unsigned size = numdevs * sizeof(sbi->layout.s_ods[0]);
+ unsigned size = numdevs * sizeof(sbi->comps.ods[0]);
- sbi->layout.s_ods = kzalloc(size, GFP_KERNEL);
- if (unlikely(!sbi->layout.s_ods)) {
+ /* 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;
@@ -688,8 +681,8 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
* line. We always keep them in device-table order.
*/
if (fscb_od && osduld_device_same(fscb_od, &odi)) {
- sbi->layout.s_ods[i] = fscb_od;
- ++sbi->layout.s_numdevs;
+ sbi->comps.ods[i] = fscb_od;
+ ++sbi->comps.numdevs;
fscb_od = NULL;
continue;
}
@@ -702,13 +695,13 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
goto out;
}
- sbi->layout.s_ods[i] = od;
- ++sbi->layout.s_numdevs;
+ sbi->comps.ods[i] = od;
+ ++sbi->comps.numdevs;
/* Read the fscb of the other devices to make sure the FS
* partition is there.
*/
- ret = exofs_read_kern(od, sbi->s_cred, &obj, 0, &fscb,
+ ret = exofs_read_kern(od, comp.cred, &comp.obj, 0, &fscb,
sizeof(fscb));
if (unlikely(ret)) {
EXOFS_ERR("ERROR: Malformed participating device "
@@ -725,13 +718,22 @@ static int exofs_read_lookup_dev_table(struct exofs_sb_info *sbi,
out:
kfree(dt);
- if (unlikely(!ret && fscb_od)) {
- EXOFS_ERR(
- "ERROR: Bad device-table container device not present\n");
- osduld_put_device(fscb_od);
- ret = -EINVAL;
- }
+ if (likely(!ret)) {
+ unsigned numdevs = sbi->comps.numdevs;
+ if (unlikely(fscb_od)) {
+ 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;
}
@@ -745,7 +747,7 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
struct exofs_sb_info *sbi; /*extended info */
struct osd_dev *od; /* Master device */
struct exofs_fscb fscb; /*on-disk superblock info */
- struct osd_obj_id obj;
+ struct exofs_comp comp;
unsigned table_count;
int ret;
@@ -776,11 +778,16 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
sbi->layout.group_width = 1;
sbi->layout.group_depth = -1;
sbi->layout.group_count = 1;
- sbi->layout.s_ods = sbi->_min_one_dev;
- sbi->layout.s_numdevs = 1;
- sbi->layout.s_pid = opts->pid;
sbi->s_timeout = opts->timeout;
+ 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;
+
/* fill in some other data by hand */
memset(sb->s_id, 0, sizeof(sb->s_id));
strcpy(sb->s_id, "exofs");
@@ -791,11 +798,11 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
sb->s_bdev = NULL;
sb->s_dev = 0;
- obj.partition = sbi->layout.s_pid;
- obj.id = EXOFS_SUPER_ID;
- exofs_make_credential(sbi->s_cred, &obj);
+ comp.obj.partition = sbi->one_comp.obj.partition;
+ comp.obj.id = EXOFS_SUPER_ID;
+ exofs_make_credential(comp.cred, &comp.obj);
- ret = exofs_read_kern(od, sbi->s_cred, &obj, 0, &fscb, sizeof(fscb));
+ ret = exofs_read_kern(od, comp.cred, &comp.obj, 0, &fscb, sizeof(fscb));
if (unlikely(ret))
goto free_sbi;
@@ -828,7 +835,7 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
if (unlikely(ret))
goto free_sbi;
} else {
- sbi->layout.s_ods[0] = od;
+ sbi->comps.ods[0] = od;
}
__sbi_read_stats(sbi);
@@ -868,13 +875,13 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
goto free_sbi;
}
- _exofs_print_device("Mounting", opts->dev_name, sbi->layout.s_ods[0],
- sbi->layout.s_pid);
+ _exofs_print_device("Mounting", opts->dev_name, sbi->comps.ods[0],
+ sbi->one_comp.obj.partition);
return 0;
free_sbi:
EXOFS_ERR("Unable to mount exofs on %s pid=0x%llx err=%d\n",
- opts->dev_name, sbi->layout.s_pid, ret);
+ opts->dev_name, sbi->one_comp.obj.partition, ret);
exofs_free_sbi(sbi);
return ret;
}
@@ -915,17 +922,14 @@ static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf)
};
uint64_t capacity = ULLONG_MAX;
uint64_t used = ULLONG_MAX;
- uint8_t cred_a[OSD_CAP_LEN];
int ret;
- ret = exofs_get_io_state(&sbi->layout, &ios);
+ ret = exofs_get_io_state(&sbi->layout, &sbi->comps, &ios);
if (ret) {
EXOFS_DBGMSG("exofs_get_io_state failed.\n");
return ret;
}
- exofs_make_credential(cred_a, &ios->obj);
- ios->cred = sbi->s_cred;
ios->in_attr = attrs;
ios->in_attr_len = ARRAY_SIZE(attrs);