int gfs2_lm_withdraw(struct gfs2_sbd *sdp, char *fmt, ...)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
const struct lm_lockops *lm = ls->ls_ops;
va_list args;
if (sdp->sd_args.ar_errors == GFS2_ERRORS_WITHDRAW &&
test_and_set_bit(SDF_SHUTDOWN, &sdp->sd_flags))
return 0;
va_start(args, fmt);
vprintk(fmt, args);
va_end(args);
if (sdp->sd_args.ar_errors == GFS2_ERRORS_WITHDRAW) {
fs_err(sdp, "about to withdraw this file system\n");
BUG_ON(sdp->sd_args.ar_debug);
kobject_uevent(&sdp->sd_kobj, KOBJ_OFFLINE);
if (lm->lm_unmount) {
fs_err(sdp, "telling LM to unmount\n");
lm->lm_unmount(sdp);
}
fs_err(sdp, "withdrawn\n");
dump_stack();
}
if (sdp->sd_args.ar_errors == GFS2_ERRORS_PANIC)
panic("GFS2: fsid=%s: panic requested.\n", sdp->sd_fsname);
return -1;
}
static int init_threads(struct gfs2_sbd *sdp)
{
struct task_struct *p;
int error = 0;
p = kthread_run(gfs2_logd, sdp, "gfs2_logd");
if (IS_ERR(p)) {
error = PTR_ERR(p);
fs_err(sdp, "can't start logd thread: %d\n", error);
return error;
}
sdp->sd_logd_process = p;
p = kthread_run(gfs2_quotad, sdp, "gfs2_quotad");
if (IS_ERR(p)) {
error = PTR_ERR(p);
fs_err(sdp, "can't start quotad thread: %d\n", error);
goto fail;
}
sdp->sd_quotad_process = p;
return 0;
fail:
kthread_stop(sdp->sd_logd_process);
return error;
}
static void gfs2_write_super_lockfs(struct super_block *sb)
{
struct gfs2_sbd *sdp = sb->s_fs_info;
int error;
if (test_bit(SDF_SHUTDOWN, &sdp->sd_flags))
return;
for (;;) {
error = gfs2_freeze_fs(sdp);
if (!error)
break;
switch (error) {
case -EBUSY:
fs_err(sdp, "waiting for recovery before freeze\n");
break;
default:
fs_err(sdp, "error freezing FS: %d\n", error);
break;
}
fs_err(sdp, "retrying...\n");
msleep(1000);
}
}
static int sync_lock(struct gfs2_sbd *sdp, int mode, uint32_t flags,
unsigned int num, struct dlm_lksb *lksb, char *name)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char strname[GDLM_STRNAME_BYTES];
int error, status;
memset(strname, 0, GDLM_STRNAME_BYTES);
snprintf(strname, GDLM_STRNAME_BYTES, "%8x%16x", LM_TYPE_NONDISK, num);
error = dlm_lock(ls->ls_dlm, mode, lksb, flags,
strname, GDLM_STRNAME_BYTES - 1,
0, sync_wait_cb, ls, NULL);
if (error) {
fs_err(sdp, "%s lkid %x flags %x mode %d error %d\n",
name, lksb->sb_lkid, flags, mode, error);
return error;
}
wait_for_completion(&ls->ls_sync_wait);
status = lksb->sb_status;
if (status && status != -EAGAIN) {
fs_err(sdp, "%s lkid %x flags %x mode %d status %d\n",
name, lksb->sb_lkid, flags, mode, status);
}
return status;
}
static int control_first_done(struct gfs2_sbd *sdp)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char lvb_bits[GDLM_LVB_SIZE];
uint32_t start_gen, block_gen;
int error;
restart:
spin_lock(&ls->ls_recover_spin);
start_gen = ls->ls_recover_start;
block_gen = ls->ls_recover_block;
if (test_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags) ||
!test_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags) ||
!test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags)) {
/* */
fs_err(sdp, "control_first_done start %u block %u flags %lx\n",
start_gen, block_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
control_unlock(sdp);
return -1;
}
if (start_gen == block_gen) {
/*
*/
spin_unlock(&ls->ls_recover_spin);
fs_info(sdp, "control_first_done wait gen %u\n", start_gen);
wait_on_bit(&ls->ls_recover_flags, DFL_DLM_RECOVERY,
dlm_recovery_wait, TASK_UNINTERRUPTIBLE);
goto restart;
}
clear_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
set_bit(DFL_FIRST_MOUNT_DONE, &ls->ls_recover_flags);
memset(ls->ls_recover_submit, 0, ls->ls_recover_size*sizeof(uint32_t));
memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t));
spin_unlock(&ls->ls_recover_spin);
memset(lvb_bits, 0, sizeof(lvb_bits));
control_lvb_write(ls, start_gen, lvb_bits);
error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT);
if (error)
fs_err(sdp, "control_first_done mounted PR error %d\n", error);
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT|DLM_LKF_VALBLK);
if (error)
fs_err(sdp, "control_first_done control NL error %d\n", error);
return error;
}
static int control_first_done(struct gfs2_sbd *sdp)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char lvb_bits[GDLM_LVB_SIZE];
uint32_t start_gen, block_gen;
int error;
restart:
spin_lock(&ls->ls_recover_spin);
start_gen = ls->ls_recover_start;
block_gen = ls->ls_recover_block;
if (test_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags) ||
!test_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags) ||
!test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags)) {
/* sanity check, should not happen */
fs_err(sdp, "control_first_done start %u block %u flags %lx\n",
start_gen, block_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
control_unlock(sdp);
return -1;
}
if (start_gen == block_gen) {
/*
* Wait for the end of a dlm recovery cycle to switch from
* first mounter recovery. We can ignore any recover_slot
* callbacks between the recover_prep and next recover_done
* because we are still the first mounter and any failed nodes
* have not fully mounted, so they don't need recovery.
*/
spin_unlock(&ls->ls_recover_spin);
fs_info(sdp, "control_first_done wait gen %u\n", start_gen);
wait_on_bit(&ls->ls_recover_flags, DFL_DLM_RECOVERY,
dlm_recovery_wait, TASK_UNINTERRUPTIBLE);
goto restart;
}
clear_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
set_bit(DFL_FIRST_MOUNT_DONE, &ls->ls_recover_flags);
memset(ls->ls_recover_submit, 0, ls->ls_recover_size*sizeof(uint32_t));
memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t));
spin_unlock(&ls->ls_recover_spin);
memset(lvb_bits, 0, sizeof(lvb_bits));
control_lvb_write(ls, start_gen, lvb_bits);
error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT);
if (error)
fs_err(sdp, "control_first_done mounted PR error %d\n", error);
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT|DLM_LKF_VALBLK);
if (error)
fs_err(sdp, "control_first_done control NL error %d\n", error);
return error;
}
static void gfs2_ail_error(struct gfs2_glock *gl, const struct buffer_head *bh)
{
fs_err(gl->gl_sbd, "AIL buffer %p: blocknr %llu state 0x%08lx mapping %p page state 0x%lx\n",
bh, (unsigned long long)bh->b_blocknr, bh->b_state,
bh->b_page->mapping, bh->b_page->flags);
fs_err(gl->gl_sbd, "AIL glock %u:%llu mapping %p\n",
gl->gl_name.ln_type, gl->gl_name.ln_number,
gfs2_glock2aspace(gl));
gfs2_lm_withdraw(gl->gl_sbd, "AIL error\n");
}
static void gdlm_recovery_result(struct gfs2_sbd *sdp, unsigned int jid,
unsigned int result)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
if (test_bit(DFL_NO_DLM_OPS, &ls->ls_recover_flags))
return;
if (jid == ls->ls_jid)
return;
spin_lock(&ls->ls_recover_spin);
if (test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags)) {
spin_unlock(&ls->ls_recover_spin);
return;
}
if (ls->ls_recover_size < jid + 1) {
fs_err(sdp, "recovery_result jid %d short size %d",
jid, ls->ls_recover_size);
spin_unlock(&ls->ls_recover_spin);
return;
}
fs_info(sdp, "recover jid %d result %s\n", jid,
result == LM_RD_GAVEUP ? "busy" : "success");
ls->ls_recover_result[jid] = result;
if (!test_bit(DFL_UNMOUNT, &ls->ls_recover_flags))
queue_delayed_work(gfs2_control_wq, &sdp->sd_control_work,
result == LM_RD_GAVEUP ? HZ : 0);
spin_unlock(&ls->ls_recover_spin);
}
int gfs2_lm_withdraw(struct gfs2_sbd *sdp, const char *fmt, ...)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
const struct lm_lockops *lm = ls->ls_ops;
va_list args;
struct va_format vaf;
if (sdp->sd_args.ar_errors == GFS2_ERRORS_WITHDRAW &&
test_and_set_bit(SDF_SHUTDOWN, &sdp->sd_flags))
return 0;
if (fmt) {
va_start(args, fmt);
vaf.fmt = fmt;
vaf.va = &args;
fs_err(sdp, "%pV", &vaf);
va_end(args);
}
if (sdp->sd_args.ar_errors == GFS2_ERRORS_WITHDRAW) {
fs_err(sdp, "about to withdraw this file system\n");
BUG_ON(sdp->sd_args.ar_debug);
kobject_uevent(&sdp->sd_kobj, KOBJ_OFFLINE);
if (!strcmp(sdp->sd_lockstruct.ls_ops->lm_proto_name, "lock_dlm"))
wait_for_completion(&sdp->sd_wdack);
if (lm->lm_unmount) {
fs_err(sdp, "telling LM to unmount\n");
lm->lm_unmount(sdp);
}
set_bit(SDF_SKIP_DLM_UNLOCK, &sdp->sd_flags);
fs_err(sdp, "withdrawn\n");
dump_stack();
}
if (sdp->sd_args.ar_errors == GFS2_ERRORS_PANIC)
panic("GFS2: fsid=%s: panic requested\n", sdp->sd_fsname);
return -1;
}
int gfs2_quotad(void *data)
{
struct gfs2_sbd *sdp = data;
unsigned long t;
int error;
while (!kthread_should_stop()) {
/* Update the master statfs file */
t = sdp->sd_statfs_sync_time +
gfs2_tune_get(sdp, gt_statfs_quantum) * HZ;
if (time_after_eq(jiffies, t)) {
error = gfs2_statfs_sync(sdp);
if (error &&
error != -EROFS &&
!test_bit(SDF_SHUTDOWN, &sdp->sd_flags))
fs_err(sdp, "quotad: (1) error=%d\n", error);
sdp->sd_statfs_sync_time = jiffies;
}
/* Update quota file */
t = sdp->sd_quota_sync_time +
gfs2_tune_get(sdp, gt_quota_quantum) * HZ;
if (time_after_eq(jiffies, t)) {
error = gfs2_quota_sync(sdp);
if (error &&
error != -EROFS &&
!test_bit(SDF_SHUTDOWN, &sdp->sd_flags))
fs_err(sdp, "quotad: (2) error=%d\n", error);
sdp->sd_quota_sync_time = jiffies;
}
gfs2_quota_scan(sdp);
t = gfs2_tune_get(sdp, gt_quotad_secs) * HZ;
if (freezing(current))
refrigerator();
schedule_timeout_interruptible(t);
}
return 0;
}
int gfs2_lm_withdraw(struct gfs2_sbd *sdp, char *fmt, ...)
{
va_list args;
if (test_and_set_bit(SDF_SHUTDOWN, &sdp->sd_flags))
return 0;
va_start(args, fmt);
vprintk(fmt, args);
va_end(args);
fs_err(sdp, "about to withdraw this file system\n");
BUG_ON(sdp->sd_args.ar_debug);
fs_err(sdp, "telling LM to withdraw\n");
gfs2_withdraw_lockproto(&sdp->sd_lockstruct);
fs_err(sdp, "withdrawn\n");
dump_stack();
return -1;
}
static void gdlm_first_done(struct gfs2_sbd *sdp)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
int error;
if (test_bit(DFL_NO_DLM_OPS, &ls->ls_recover_flags))
return;
error = control_first_done(sdp);
if (error)
fs_err(sdp, "mount first_done error %d\n", error);
}
static int sync_unlock(struct gfs2_sbd *sdp, struct dlm_lksb *lksb, char *name)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
int error;
error = dlm_unlock(ls->ls_dlm, lksb->sb_lkid, 0, lksb, ls);
if (error) {
fs_err(sdp, "%s lkid %x error %d\n",
name, lksb->sb_lkid, error);
return error;
}
wait_for_completion(&ls->ls_sync_wait);
if (lksb->sb_status != -DLM_EUNLOCK) {
fs_err(sdp, "%s lkid %x status %d\n",
name, lksb->sb_lkid, lksb->sb_status);
return -1;
}
return 0;
}
void gfs2_io_error_bh_i(struct gfs2_sbd *sdp, struct buffer_head *bh,
const char *function, char *file, unsigned int line,
bool withdraw)
{
fs_err(sdp,
"fatal: I/O error\n"
" block = %llu\n"
" function = %s, file = %s, line = %u\n",
(unsigned long long)bh->b_blocknr,
function, file, line);
if (withdraw)
gfs2_lm_withdraw(sdp, NULL);
}
void gfs2_trans_end(struct gfs2_sbd *sdp)
{
struct gfs2_trans *tr = current->journal_info;
BUG_ON(!tr);
current->journal_info = NULL;
if (!tr->tr_touched) {
gfs2_log_release(sdp, tr->tr_reserved);
if (tr->tr_t_gh.gh_gl) {
gfs2_glock_dq(&tr->tr_t_gh);
gfs2_holder_uninit(&tr->tr_t_gh);
kfree(tr);
}
return;
}
if (gfs2_assert_withdraw(sdp, tr->tr_num_buf <= tr->tr_blocks)) {
fs_err(sdp, "tr_num_buf = %u, tr_blocks = %u ",
tr->tr_num_buf, tr->tr_blocks);
print_symbol(KERN_WARNING "GFS2: Transaction created at: %s\n", tr->tr_ip);
}
if (gfs2_assert_withdraw(sdp, tr->tr_num_revoke <= tr->tr_revokes)) {
fs_err(sdp, "tr_num_revoke = %u, tr_revokes = %u ",
tr->tr_num_revoke, tr->tr_revokes);
print_symbol(KERN_WARNING "GFS2: Transaction created at: %s\n", tr->tr_ip);
}
gfs2_log_commit(sdp, tr);
if (tr->tr_t_gh.gh_gl) {
gfs2_glock_dq(&tr->tr_t_gh);
gfs2_holder_uninit(&tr->tr_t_gh);
kfree(tr);
}
if (sdp->sd_vfs->s_flags & MS_SYNCHRONOUS)
gfs2_log_flush(sdp, NULL);
}
static int make_mode(struct gfs2_sbd *sdp, const unsigned int lmstate)
{
switch (lmstate) {
case LM_ST_UNLOCKED:
return DLM_LOCK_NL;
case LM_ST_EXCLUSIVE:
return DLM_LOCK_EX;
case LM_ST_DEFERRED:
return DLM_LOCK_CW;
case LM_ST_SHARED:
return DLM_LOCK_PR;
}
fs_err(sdp, "unknown LM state %d\n", lmstate);
BUG();
return -1;
}
static void gdlm_bast(void *arg, int mode)
{
struct gfs2_glock *gl = arg;
switch (mode) {
case DLM_LOCK_EX:
gfs2_glock_cb(gl, LM_ST_UNLOCKED);
break;
case DLM_LOCK_CW:
gfs2_glock_cb(gl, LM_ST_DEFERRED);
break;
case DLM_LOCK_PR:
gfs2_glock_cb(gl, LM_ST_SHARED);
break;
default:
fs_err(gl->gl_name.ln_sbd, "unknown bast mode %d\n", mode);
BUG();
}
}
int gfs2_sys_fs_add(struct gfs2_sbd *sdp)
{
int error;
sdp->sd_kobj.kset = gfs2_kset;
error = kobject_init_and_add(&sdp->sd_kobj, &gfs2_ktype, NULL,
"%s", sdp->sd_table_name);
if (error)
goto fail;
error = sysfs_create_group(&sdp->sd_kobj, &lockstruct_group);
if (error)
goto fail_reg;
error = sysfs_create_group(&sdp->sd_kobj, &counters_group);
if (error)
goto fail_lockstruct;
error = sysfs_create_group(&sdp->sd_kobj, &args_group);
if (error)
goto fail_counters;
error = sysfs_create_group(&sdp->sd_kobj, &tune_group);
if (error)
goto fail_args;
kobject_uevent(&sdp->sd_kobj, KOBJ_ADD);
return 0;
fail_args:
sysfs_remove_group(&sdp->sd_kobj, &args_group);
fail_counters:
sysfs_remove_group(&sdp->sd_kobj, &counters_group);
fail_lockstruct:
sysfs_remove_group(&sdp->sd_kobj, &lockstruct_group);
fail_reg:
kobject_put(&sdp->sd_kobj);
fail:
fs_err(sdp, "error %d adding sysfs files", error);
return error;
}
static void gdlm_recover_slot(void *arg, struct dlm_slot *slot)
{
struct gfs2_sbd *sdp = arg;
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
int jid = slot->slot - 1;
spin_lock(&ls->ls_recover_spin);
if (ls->ls_recover_size < jid + 1) {
fs_err(sdp, "recover_slot jid %d gen %u short size %d",
jid, ls->ls_recover_block, ls->ls_recover_size);
spin_unlock(&ls->ls_recover_spin);
return;
}
if (ls->ls_recover_submit[jid]) {
fs_info(sdp, "recover_slot jid %d gen %u prev %u",
jid, ls->ls_recover_block, ls->ls_recover_submit[jid]);
}
ls->ls_recover_submit[jid] = ls->ls_recover_block;
spin_unlock(&ls->ls_recover_spin);
}
int gfs2_sys_fs_add(struct gfs2_sbd *sdp)
{
int error;
sdp->sd_kobj.kset = &gfs2_kset;
sdp->sd_kobj.ktype = &gfs2_ktype;
error = kobject_set_name(&sdp->sd_kobj, "%s", sdp->sd_table_name);
if (error)
goto fail;
error = kobject_register(&sdp->sd_kobj);
if (error)
goto fail;
error = sysfs_create_group(&sdp->sd_kobj, &lockstruct_group);
if (error)
goto fail_reg;
error = sysfs_create_group(&sdp->sd_kobj, &args_group);
if (error)
goto fail_lockstruct;
error = sysfs_create_group(&sdp->sd_kobj, &tune_group);
if (error)
goto fail_args;
return 0;
fail_args:
sysfs_remove_group(&sdp->sd_kobj, &args_group);
fail_lockstruct:
sysfs_remove_group(&sdp->sd_kobj, &lockstruct_group);
fail_reg:
kobject_unregister(&sdp->sd_kobj);
fail:
fs_err(sdp, "error %d adding sysfs files", error);
return error;
}
static void gdlm_recovery_result(struct gfs2_sbd *sdp, unsigned int jid,
unsigned int result)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
if (test_bit(DFL_NO_DLM_OPS, &ls->ls_recover_flags))
return;
/* don't care about the recovery of own journal during mount */
if (jid == ls->ls_jid)
return;
spin_lock(&ls->ls_recover_spin);
if (test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags)) {
spin_unlock(&ls->ls_recover_spin);
return;
}
if (ls->ls_recover_size < jid + 1) {
fs_err(sdp, "recovery_result jid %d short size %d",
jid, ls->ls_recover_size);
spin_unlock(&ls->ls_recover_spin);
return;
}
fs_info(sdp, "recover jid %d result %s\n", jid,
result == LM_RD_GAVEUP ? "busy" : "success");
ls->ls_recover_result[jid] = result;
/* GAVEUP means another node is recovering the journal; delay our
next attempt to recover it, to give the other node a chance to
finish before trying again */
if (!test_bit(DFL_UNMOUNT, &ls->ls_recover_flags))
queue_delayed_work(gfs2_control_wq, &sdp->sd_control_work,
result == LM_RD_GAVEUP ? HZ : 0);
spin_unlock(&ls->ls_recover_spin);
}
static void gdlm_put_lock(struct gfs2_glock *gl)
{
struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
int lvb_needs_unlock = 0;
int error;
if (gl->gl_lksb.sb_lkid == 0) {
gfs2_glock_free(gl);
return;
}
clear_bit(GLF_BLOCKING, &gl->gl_flags);
gfs2_glstats_inc(gl, GFS2_LKS_DCOUNT);
gfs2_sbstats_inc(gl, GFS2_LKS_DCOUNT);
gfs2_update_request_times(gl);
/* don't want to skip dlm_unlock writing the lvb when lock is ex */
if (gl->gl_lksb.sb_lvbptr && (gl->gl_state == LM_ST_EXCLUSIVE))
lvb_needs_unlock = 1;
if (test_bit(SDF_SKIP_DLM_UNLOCK, &sdp->sd_flags) &&
!lvb_needs_unlock) {
gfs2_glock_free(gl);
return;
}
error = dlm_unlock(ls->ls_dlm, gl->gl_lksb.sb_lkid, DLM_LKF_VALBLK,
NULL, gl);
if (error) {
fs_err(sdp, "gdlm_unlock %x,%llx err=%d\n",
gl->gl_name.ln_type,
(unsigned long long)gl->gl_name.ln_number, error);
return;
}
}
static int gfs2_read_sb(struct gfs2_sbd *sdp, int silent)
{
u32 hash_blocks, ind_blocks, leaf_blocks;
u32 tmp_blocks;
unsigned int x;
int error;
error = gfs2_read_super(sdp, GFS2_SB_ADDR >> sdp->sd_fsb2bb_shift, silent);
if (error) {
if (!silent)
fs_err(sdp, "can't read superblock\n");
return error;
}
sdp->sd_fsb2bb_shift = sdp->sd_sb.sb_bsize_shift -
GFS2_BASIC_BLOCK_SHIFT;
sdp->sd_fsb2bb = 1 << sdp->sd_fsb2bb_shift;
sdp->sd_diptrs = (sdp->sd_sb.sb_bsize -
sizeof(struct gfs2_dinode)) / sizeof(u64);
sdp->sd_inptrs = (sdp->sd_sb.sb_bsize -
sizeof(struct gfs2_meta_header)) / sizeof(u64);
sdp->sd_jbsize = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_meta_header);
sdp->sd_hash_bsize = sdp->sd_sb.sb_bsize / 2;
sdp->sd_hash_bsize_shift = sdp->sd_sb.sb_bsize_shift - 1;
sdp->sd_hash_ptrs = sdp->sd_hash_bsize / sizeof(u64);
sdp->sd_qc_per_block = (sdp->sd_sb.sb_bsize -
sizeof(struct gfs2_meta_header)) /
sizeof(struct gfs2_quota_change);
/* Compute maximum reservation required to add a entry to a directory */
hash_blocks = DIV_ROUND_UP(sizeof(u64) * (1 << GFS2_DIR_MAX_DEPTH),
sdp->sd_jbsize);
ind_blocks = 0;
for (tmp_blocks = hash_blocks; tmp_blocks > sdp->sd_diptrs;) {
tmp_blocks = DIV_ROUND_UP(tmp_blocks, sdp->sd_inptrs);
ind_blocks += tmp_blocks;
}
leaf_blocks = 2 + GFS2_DIR_MAX_DEPTH;
sdp->sd_max_dirres = hash_blocks + ind_blocks + leaf_blocks;
sdp->sd_heightsize[0] = sdp->sd_sb.sb_bsize -
sizeof(struct gfs2_dinode);
sdp->sd_heightsize[1] = sdp->sd_sb.sb_bsize * sdp->sd_diptrs;
for (x = 2;; x++) {
u64 space, d;
u32 m;
space = sdp->sd_heightsize[x - 1] * sdp->sd_inptrs;
d = space;
m = do_div(d, sdp->sd_inptrs);
if (d != sdp->sd_heightsize[x - 1] || m)
break;
sdp->sd_heightsize[x] = space;
}
sdp->sd_max_height = x;
sdp->sd_heightsize[x] = ~0;
gfs2_assert(sdp, sdp->sd_max_height <= GFS2_MAX_META_HEIGHT);
sdp->sd_jheightsize[0] = sdp->sd_sb.sb_bsize -
sizeof(struct gfs2_dinode);
sdp->sd_jheightsize[1] = sdp->sd_jbsize * sdp->sd_diptrs;
for (x = 2;; x++) {
u64 space, d;
u32 m;
space = sdp->sd_jheightsize[x - 1] * sdp->sd_inptrs;
d = space;
m = do_div(d, sdp->sd_inptrs);
if (d != sdp->sd_jheightsize[x - 1] || m)
break;
sdp->sd_jheightsize[x] = space;
}
sdp->sd_max_jheight = x;
sdp->sd_jheightsize[x] = ~0;
gfs2_assert(sdp, sdp->sd_max_jheight <= GFS2_MAX_META_HEIGHT);
return 0;
}
static int control_mount(struct gfs2_sbd *sdp)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char lvb_bits[GDLM_LVB_SIZE];
uint32_t start_gen, block_gen, mount_gen, lvb_gen;
int mounted_mode;
int retries = 0;
int error;
memset(&ls->ls_mounted_lksb, 0, sizeof(struct dlm_lksb));
memset(&ls->ls_control_lksb, 0, sizeof(struct dlm_lksb));
memset(&ls->ls_control_lvb, 0, GDLM_LVB_SIZE);
ls->ls_control_lksb.sb_lvbptr = ls->ls_control_lvb;
init_completion(&ls->ls_sync_wait);
set_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_VALBLK);
if (error) {
fs_err(sdp, "control_mount control_lock NL error %d\n", error);
return error;
}
error = mounted_lock(sdp, DLM_LOCK_NL, 0);
if (error) {
fs_err(sdp, "control_mount mounted_lock NL error %d\n", error);
control_unlock(sdp);
return error;
}
mounted_mode = DLM_LOCK_NL;
restart:
if (retries++ && signal_pending(current)) {
error = -EINTR;
goto fail;
}
if (mounted_mode != DLM_LOCK_NL) {
error = mounted_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT);
if (error)
goto fail;
mounted_mode = DLM_LOCK_NL;
}
msleep_interruptible(500);
error = control_lock(sdp, DLM_LOCK_EX, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE|DLM_LKF_VALBLK);
if (error == -EAGAIN) {
goto restart;
} else if (error) {
fs_err(sdp, "control_mount control_lock EX error %d\n", error);
goto fail;
}
error = mounted_lock(sdp, DLM_LOCK_EX, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE);
if (!error) {
mounted_mode = DLM_LOCK_EX;
goto locks_done;
} else if (error != -EAGAIN) {
fs_err(sdp, "control_mount mounted_lock EX error %d\n", error);
goto fail;
}
error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE);
if (!error) {
mounted_mode = DLM_LOCK_PR;
goto locks_done;
} else {
fs_err(sdp, "control_mount mounted_lock PR error %d\n", error);
goto fail;
}
locks_done:
control_lvb_read(ls, &lvb_gen, lvb_bits);
if (lvb_gen == 0xFFFFFFFF) {
fs_err(sdp, "control_mount control_lock disabled\n");
error = -EINVAL;
goto fail;
}
if (mounted_mode == DLM_LOCK_EX) {
spin_lock(&ls->ls_recover_spin);
clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
set_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags);
set_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
fs_info(sdp, "first mounter control generation %u\n", lvb_gen);
return 0;
}
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT);
if (error)
goto fail;
if (!all_jid_bits_clear(lvb_bits)) {
fs_info(sdp, "control_mount wait for journal recovery\n");
goto restart;
}
spin_lock(&ls->ls_recover_spin);
block_gen = ls->ls_recover_block;
start_gen = ls->ls_recover_start;
mount_gen = ls->ls_recover_mount;
if (lvb_gen < mount_gen) {
fs_info(sdp, "control_mount wait1 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
if (lvb_gen != start_gen) {
fs_info(sdp, "control_mount wait2 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
if (block_gen == start_gen) {
fs_info(sdp, "control_mount wait3 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
set_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags);
memset(ls->ls_recover_submit, 0, ls->ls_recover_size*sizeof(uint32_t));
memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t));
spin_unlock(&ls->ls_recover_spin);
return 0;
fail:
mounted_unlock(sdp);
control_unlock(sdp);
return error;
}
static int gdlm_mount(struct gfs2_sbd *sdp, const char *table)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char cluster[GFS2_LOCKNAME_LEN];
const char *fsname;
uint32_t flags;
int error, ops_result;
INIT_DELAYED_WORK(&sdp->sd_control_work, gfs2_control_func);
spin_lock_init(&ls->ls_recover_spin);
ls->ls_recover_flags = 0;
ls->ls_recover_mount = 0;
ls->ls_recover_start = 0;
ls->ls_recover_block = 0;
ls->ls_recover_size = 0;
ls->ls_recover_submit = NULL;
ls->ls_recover_result = NULL;
error = set_recover_size(sdp, NULL, 0);
if (error)
goto fail;
fsname = strchr(table, ':');
if (!fsname) {
fs_info(sdp, "no fsname found\n");
error = -EINVAL;
goto fail_free;
}
memset(cluster, 0, sizeof(cluster));
memcpy(cluster, table, strlen(table) - strlen(fsname));
fsname++;
flags = DLM_LSFL_FS | DLM_LSFL_NEWEXCL;
if (ls->ls_nodir)
flags |= DLM_LSFL_NODIR;
error = dlm_new_lockspace(fsname, cluster, flags, GDLM_LVB_SIZE,
&gdlm_lockspace_ops, sdp, &ops_result,
&ls->ls_dlm);
if (error) {
fs_err(sdp, "dlm_new_lockspace error %d\n", error);
goto fail_free;
}
if (ops_result < 0) {
fs_info(sdp, "dlm lockspace ops not used\n");
free_recover_size(ls);
set_bit(DFL_NO_DLM_OPS, &ls->ls_recover_flags);
return 0;
}
if (!test_bit(SDF_NOJOURNALID, &sdp->sd_flags)) {
fs_err(sdp, "dlm lockspace ops disallow jid preset\n");
error = -EINVAL;
goto fail_release;
}
error = control_mount(sdp);
if (error) {
fs_err(sdp, "mount control error %d\n", error);
goto fail_release;
}
ls->ls_first = !!test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
clear_bit(SDF_NOJOURNALID, &sdp->sd_flags);
smp_mb__after_clear_bit();
wake_up_bit(&sdp->sd_flags, SDF_NOJOURNALID);
return 0;
fail_release:
dlm_release_lockspace(ls->ls_dlm, 2);
fail_free:
free_recover_size(ls);
fail:
return error;
}
static int gfs2_bmap_alloc(struct inode *inode, const sector_t lblock,
struct buffer_head *bh_map, struct metapath *mp,
const unsigned int sheight,
const unsigned int height,
const unsigned int maxlen)
{
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_sbd *sdp = GFS2_SB(inode);
struct super_block *sb = sdp->sd_vfs;
struct buffer_head *dibh = mp->mp_bh[0];
u64 bn, dblock = 0;
unsigned n, i, blks, alloced = 0, iblks = 0, branch_start = 0;
unsigned dblks = 0;
unsigned ptrs_per_blk;
const unsigned end_of_metadata = height - 1;
int ret;
int eob = 0;
enum alloc_state state;
__be64 *ptr;
__be64 zero_bn = 0;
BUG_ON(sheight < 1);
BUG_ON(dibh == NULL);
gfs2_trans_add_bh(ip->i_gl, dibh, 1);
if (height == sheight) {
struct buffer_head *bh;
ptr = metapointer(end_of_metadata, mp);
bh = mp->mp_bh[end_of_metadata];
dblks = gfs2_extent_length(bh->b_data, bh->b_size, ptr, maxlen,
&eob);
BUG_ON(dblks < 1);
state = ALLOC_DATA;
} else {
ptrs_per_blk = height > 1 ? sdp->sd_inptrs : sdp->sd_diptrs;
dblks = min(maxlen, ptrs_per_blk - mp->mp_list[end_of_metadata]);
if (height == ip->i_height) {
iblks = height - sheight;
state = ALLOC_GROW_DEPTH;
} else {
state = ALLOC_GROW_HEIGHT;
iblks = height - ip->i_height;
branch_start = metapath_branch_start(mp);
iblks += (height - branch_start);
}
}
blks = dblks + iblks;
i = sheight;
do {
int error;
n = blks - alloced;
error = gfs2_alloc_blocks(ip, &bn, &n, 0, NULL);
if (error)
return error;
alloced += n;
if (state != ALLOC_DATA || gfs2_is_jdata(ip))
gfs2_trans_add_unrevoke(sdp, bn, n);
switch (state) {
case ALLOC_GROW_HEIGHT:
if (i == 1) {
ptr = (__be64 *)(dibh->b_data +
sizeof(struct gfs2_dinode));
zero_bn = *ptr;
}
for (; i - 1 < height - ip->i_height && n > 0; i++, n--)
gfs2_indirect_init(mp, ip->i_gl, i, 0, bn++);
if (i - 1 == height - ip->i_height) {
i--;
gfs2_buffer_copy_tail(mp->mp_bh[i],
sizeof(struct gfs2_meta_header),
dibh, sizeof(struct gfs2_dinode));
gfs2_buffer_clear_tail(dibh,
sizeof(struct gfs2_dinode) +
sizeof(__be64));
ptr = (__be64 *)(mp->mp_bh[i]->b_data +
sizeof(struct gfs2_meta_header));
*ptr = zero_bn;
state = ALLOC_GROW_DEPTH;
for(i = branch_start; i < height; i++) {
if (mp->mp_bh[i] == NULL)
break;
brelse(mp->mp_bh[i]);
mp->mp_bh[i] = NULL;
}
i = branch_start;
}
if (n == 0)
break;
case ALLOC_GROW_DEPTH:
if (i > 1 && i < height)
gfs2_trans_add_bh(ip->i_gl, mp->mp_bh[i-1], 1);
for (; i < height && n > 0; i++, n--)
gfs2_indirect_init(mp, ip->i_gl, i,
mp->mp_list[i-1], bn++);
if (i == height)
state = ALLOC_DATA;
if (n == 0)
break;
case ALLOC_DATA:
BUG_ON(n > dblks);
BUG_ON(mp->mp_bh[end_of_metadata] == NULL);
gfs2_trans_add_bh(ip->i_gl, mp->mp_bh[end_of_metadata], 1);
dblks = n;
ptr = metapointer(end_of_metadata, mp);
dblock = bn;
while (n-- > 0)
*ptr++ = cpu_to_be64(bn++);
if (buffer_zeronew(bh_map)) {
ret = sb_issue_zeroout(sb, dblock, dblks,
GFP_NOFS);
if (ret) {
fs_err(sdp,
"Failed to zero data buffers\n");
clear_buffer_zeronew(bh_map);
}
}
break;
}
} while ((state != ALLOC_DATA) || !dblock);
ip->i_height = height;
gfs2_add_inode_blocks(&ip->i_inode, alloced);
gfs2_dinode_out(ip, mp->mp_bh[0]->b_data);
map_bh(bh_map, inode->i_sb, dblock);
bh_map->b_size = dblks << inode->i_blkbits;
set_buffer_new(bh_map);
return 0;
}
/*
* struct fs32_chdir_parms {
* unsigned short iCurDirEnd;
* PTR16 pDir;
* PTR16 pcdfsd;
* PTR16 pcdfsi;
* unsigned short flag;
* };
*/
int FS32ENTRY fs32_chdir(struct fs32_chdir_parms *parms)
{
char *pDir;
struct cdfsi32 *pcdfsi;
union cdfsd32 *pcdfsd;
int rc;
struct file *p_file;
struct super_block *sb;
int valid;
char *tmp;
struct inode *base;
switch (parms->flag)
{
case CD_EXPLICIT:
pcdfsi = VDHQueryLin(parms->pcdfsi);
pcdfsd = VDHQueryLin(parms->pcdfsd);
pDir = VDHQueryLin(parms->pDir);
if (trace_FS_CHDIR)
{
kernel_printf("FS_CHDIR( CD_EXPLICIT, %s )", pDir);
}
rc = ERROR_INVALID_PARAMETER;
if (parms->iCurDirEnd != CURDIREND_INVALID)
{
tmp = pDir + parms->iCurDirEnd;
if ((pcdfsd->u.p_file) && (pcdfsd->u.p_file->f_magic == FILE_MAGIC))
{
base = pcdfsd->u.p_file->f_inode;
if (base)
{
rc = NO_ERROR;
}
}
}
else
{
sb = getvolume(pcdfsi->cdi_hVPB);
if ((sb) && (sb->s_magic_internal == SUPER_MAGIC))
{
tmp = skip_drive(pDir);
base = sb->s_mounted;
if (base)
{
rc = NO_ERROR;
}
}
}
if (rc == NO_ERROR)
{
if ((p_file = open_by_name(base, tmp, OPENMODE_READONLY)) == 0)
{
rc = ERROR_PATH_NOT_FOUND;
}
else
{
if (!S_ISDIR(p_file->f_inode->i_mode))
{
kernel_printf("FS_CHDIR( %s ) Not a directory", pDir);
vfs_close(p_file);
rc = ERROR_ACCESS_DENIED;
}
else
{
pcdfsd->u.is_valid = 1;
pcdfsd->u.p_file = p_file;
rc = NO_ERROR;
}
}
}
break;
case CD_VERIFY:
pcdfsi = VDHQueryLin(parms->pcdfsi);
pcdfsd = VDHQueryLin(parms->pcdfsd);
if (trace_FS_CHDIR)
{
kernel_printf("FS_CHDIR : flag = CD_VERIFY hVPB=0x%04X", pcdfsi->cdi_hVPB);
}
//
// Gets the superblock from pcdfsi
//
sb = getvolume(pcdfsi->cdi_hVPB);
valid = 1;
if ((p_file = open_by_name(sb->s_mounted, skip_drive(pcdfsi->cdi_curdir), OPENMODE_READONLY)) == 0)
{
valid = 0;
}
else
{
if (!S_ISDIR(p_file->f_inode->i_mode))
{
vfs_close(p_file);
valid = 0;
}
if (pcdfsi->cdi_flags & CDI_ISVALID)
{
vfs_close(pcdfsd->u.p_file);
}
if (valid)
{
pcdfsd->u.is_valid = 1;
pcdfsd->u.p_file = p_file;
return NO_ERROR;
}
else
{
vfs_close(p_file);
return ERROR_PATH_NOT_FOUND;
}
}
break;
case CD_FREE:
pcdfsd = VDHQueryLin(parms->pcdfsd);
if (trace_FS_CHDIR)
{
kernel_printf("FS_CHDIR( CD_FREE )");
}
if (pcdfsd->u.is_valid == 1)
{
pcdfsd->u.is_valid = 0;
if ((rc = vfs_close(pcdfsd->u.p_file)) != NO_ERROR)
{
fs_err(FUNC_FS_CHDIR, FUNC_CLOSE, rc, THISFILE, __LINE__);
}
// return rc;
}
else
{
rc = NO_ERROR;
}
break;
default:
fs_log("FS_CHDIR : invalid flag");
rc = ERROR_INVALID_PARAMETER;
}
return rc;
}
static int control_mount(struct gfs2_sbd *sdp)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
uint32_t start_gen, block_gen, mount_gen, lvb_gen;
int mounted_mode;
int retries = 0;
int error;
memset(&ls->ls_mounted_lksb, 0, sizeof(struct dlm_lksb));
memset(&ls->ls_control_lksb, 0, sizeof(struct dlm_lksb));
memset(&ls->ls_control_lvb, 0, GDLM_LVB_SIZE);
ls->ls_control_lksb.sb_lvbptr = ls->ls_control_lvb;
init_completion(&ls->ls_sync_wait);
set_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_VALBLK);
if (error) {
fs_err(sdp, "control_mount control_lock NL error %d\n", error);
return error;
}
error = mounted_lock(sdp, DLM_LOCK_NL, 0);
if (error) {
fs_err(sdp, "control_mount mounted_lock NL error %d\n", error);
control_unlock(sdp);
return error;
}
mounted_mode = DLM_LOCK_NL;
restart:
if (retries++ && signal_pending(current)) {
error = -EINTR;
goto fail;
}
/*
* We always start with both locks in NL. control_lock is
* demoted to NL below so we don't need to do it here.
*/
if (mounted_mode != DLM_LOCK_NL) {
error = mounted_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT);
if (error)
goto fail;
mounted_mode = DLM_LOCK_NL;
}
/*
* Other nodes need to do some work in dlm recovery and gfs2_control
* before the recover_done and control_lock will be ready for us below.
* A delay here is not required but often avoids having to retry.
*/
msleep_interruptible(500);
/*
* Acquire control_lock in EX and mounted_lock in either EX or PR.
* control_lock lvb keeps track of any pending journal recoveries.
* mounted_lock indicates if any other nodes have the fs mounted.
*/
error = control_lock(sdp, DLM_LOCK_EX, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE|DLM_LKF_VALBLK);
if (error == -EAGAIN) {
goto restart;
} else if (error) {
fs_err(sdp, "control_mount control_lock EX error %d\n", error);
goto fail;
}
error = mounted_lock(sdp, DLM_LOCK_EX, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE);
if (!error) {
mounted_mode = DLM_LOCK_EX;
goto locks_done;
} else if (error != -EAGAIN) {
fs_err(sdp, "control_mount mounted_lock EX error %d\n", error);
goto fail;
}
error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT|DLM_LKF_NOQUEUE);
if (!error) {
mounted_mode = DLM_LOCK_PR;
goto locks_done;
} else {
/* not even -EAGAIN should happen here */
fs_err(sdp, "control_mount mounted_lock PR error %d\n", error);
goto fail;
}
locks_done:
/*
* If we got both locks above in EX, then we're the first mounter.
* If not, then we need to wait for the control_lock lvb to be
* updated by other mounted nodes to reflect our mount generation.
*
* In simple first mounter cases, first mounter will see zero lvb_gen,
* but in cases where all existing nodes leave/fail before mounting
* nodes finish control_mount, then all nodes will be mounting and
* lvb_gen will be non-zero.
*/
control_lvb_read(ls, &lvb_gen, ls->ls_lvb_bits);
if (lvb_gen == 0xFFFFFFFF) {
/* special value to force mount attempts to fail */
fs_err(sdp, "control_mount control_lock disabled\n");
error = -EINVAL;
goto fail;
}
if (mounted_mode == DLM_LOCK_EX) {
/* first mounter, keep both EX while doing first recovery */
spin_lock(&ls->ls_recover_spin);
clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
set_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags);
set_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
fs_info(sdp, "first mounter control generation %u\n", lvb_gen);
return 0;
}
error = control_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT);
if (error)
goto fail;
/*
* We are not first mounter, now we need to wait for the control_lock
* lvb generation to be >= the generation from our first recover_done
* and all lvb bits to be clear (no pending journal recoveries.)
*/
if (!all_jid_bits_clear(ls->ls_lvb_bits)) {
/* journals need recovery, wait until all are clear */
fs_info(sdp, "control_mount wait for journal recovery\n");
goto restart;
}
spin_lock(&ls->ls_recover_spin);
block_gen = ls->ls_recover_block;
start_gen = ls->ls_recover_start;
mount_gen = ls->ls_recover_mount;
if (lvb_gen < mount_gen) {
/* wait for mounted nodes to update control_lock lvb to our
generation, which might include new recovery bits set */
fs_info(sdp, "control_mount wait1 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
if (lvb_gen != start_gen) {
/* wait for mounted nodes to update control_lock lvb to the
latest recovery generation */
fs_info(sdp, "control_mount wait2 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
if (block_gen == start_gen) {
/* dlm recovery in progress, wait for it to finish */
fs_info(sdp, "control_mount wait3 block %u start %u mount %u "
"lvb %u flags %lx\n", block_gen, start_gen, mount_gen,
lvb_gen, ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
goto restart;
}
clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
set_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags);
memset(ls->ls_recover_submit, 0, ls->ls_recover_size*sizeof(uint32_t));
memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t));
spin_unlock(&ls->ls_recover_spin);
return 0;
fail:
mounted_unlock(sdp);
control_unlock(sdp);
return error;
}
static void gfs2_control_func(struct work_struct *work)
{
struct gfs2_sbd *sdp = container_of(work, struct gfs2_sbd, sd_control_work.work);
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
uint32_t block_gen, start_gen, lvb_gen, flags;
int recover_set = 0;
int write_lvb = 0;
int recover_size;
int i, error;
spin_lock(&ls->ls_recover_spin);
/*
* No MOUNT_DONE means we're still mounting; control_mount()
* will set this flag, after which this thread will take over
* all further clearing of BLOCK_LOCKS.
*
* FIRST_MOUNT means this node is doing first mounter recovery,
* for which recovery control is handled by
* control_mount()/control_first_done(), not this thread.
*/
if (!test_bit(DFL_MOUNT_DONE, &ls->ls_recover_flags) ||
test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags)) {
spin_unlock(&ls->ls_recover_spin);
return;
}
block_gen = ls->ls_recover_block;
start_gen = ls->ls_recover_start;
spin_unlock(&ls->ls_recover_spin);
/*
* Equal block_gen and start_gen implies we are between
* recover_prep and recover_done callbacks, which means
* dlm recovery is in progress and dlm locking is blocked.
* There's no point trying to do any work until recover_done.
*/
if (block_gen == start_gen)
return;
/*
* Propagate recover_submit[] and recover_result[] to lvb:
* dlm_recoverd adds to recover_submit[] jids needing recovery
* gfs2_recover adds to recover_result[] journal recovery results
*
* set lvb bit for jids in recover_submit[] if the lvb has not
* yet been updated for the generation of the failure
*
* clear lvb bit for jids in recover_result[] if the result of
* the journal recovery is SUCCESS
*/
error = control_lock(sdp, DLM_LOCK_EX, DLM_LKF_CONVERT|DLM_LKF_VALBLK);
if (error) {
fs_err(sdp, "control lock EX error %d\n", error);
return;
}
control_lvb_read(ls, &lvb_gen, ls->ls_lvb_bits);
spin_lock(&ls->ls_recover_spin);
if (block_gen != ls->ls_recover_block ||
start_gen != ls->ls_recover_start) {
fs_info(sdp, "recover generation %u block1 %u %u\n",
start_gen, block_gen, ls->ls_recover_block);
spin_unlock(&ls->ls_recover_spin);
control_lock(sdp, DLM_LOCK_NL, DLM_LKF_CONVERT);
return;
}
recover_size = ls->ls_recover_size;
if (lvb_gen <= start_gen) {
/*
* Clear lvb bits for jids we've successfully recovered.
* Because all nodes attempt to recover failed journals,
* a journal can be recovered multiple times successfully
* in succession. Only the first will really do recovery,
* the others find it clean, but still report a successful
* recovery. So, another node may have already recovered
* the jid and cleared the lvb bit for it.
*/
for (i = 0; i < recover_size; i++) {
if (ls->ls_recover_result[i] != LM_RD_SUCCESS)
continue;
ls->ls_recover_result[i] = 0;
if (!test_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET))
continue;
__clear_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET);
write_lvb = 1;
}
}
if (lvb_gen == start_gen) {
/*
* Failed slots before start_gen are already set in lvb.
*/
for (i = 0; i < recover_size; i++) {
if (!ls->ls_recover_submit[i])
continue;
if (ls->ls_recover_submit[i] < lvb_gen)
ls->ls_recover_submit[i] = 0;
}
} else if (lvb_gen < start_gen) {
/*
* Failed slots before start_gen are not yet set in lvb.
*/
for (i = 0; i < recover_size; i++) {
if (!ls->ls_recover_submit[i])
continue;
if (ls->ls_recover_submit[i] < start_gen) {
ls->ls_recover_submit[i] = 0;
__set_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET);
}
}
/* even if there are no bits to set, we need to write the
latest generation to the lvb */
write_lvb = 1;
} else {
/*
* we should be getting a recover_done() for lvb_gen soon
*/
}
spin_unlock(&ls->ls_recover_spin);
if (write_lvb) {
control_lvb_write(ls, start_gen, ls->ls_lvb_bits);
flags = DLM_LKF_CONVERT | DLM_LKF_VALBLK;
} else {
flags = DLM_LKF_CONVERT;
}
error = control_lock(sdp, DLM_LOCK_NL, flags);
if (error) {
fs_err(sdp, "control lock NL error %d\n", error);
return;
}
/*
* Everyone will see jid bits set in the lvb, run gfs2_recover_set(),
* and clear a jid bit in the lvb if the recovery is a success.
* Eventually all journals will be recovered, all jid bits will
* be cleared in the lvb, and everyone will clear BLOCK_LOCKS.
*/
for (i = 0; i < recover_size; i++) {
if (test_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET)) {
fs_info(sdp, "recover generation %u jid %d\n",
start_gen, i);
gfs2_recover_set(sdp, i);
recover_set++;
}
}
if (recover_set)
return;
/*
* No more jid bits set in lvb, all recovery is done, unblock locks
* (unless a new recover_prep callback has occured blocking locks
* again while working above)
*/
spin_lock(&ls->ls_recover_spin);
if (ls->ls_recover_block == block_gen &&
ls->ls_recover_start == start_gen) {
clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
spin_unlock(&ls->ls_recover_spin);
fs_info(sdp, "recover generation %u done\n", start_gen);
gfs2_glock_thaw(sdp);
} else {
fs_info(sdp, "recover generation %u block2 %u %u\n",
start_gen, block_gen, ls->ls_recover_block);
spin_unlock(&ls->ls_recover_spin);
}
}
static int gdlm_mount(struct gfs2_sbd *sdp, const char *table)
{
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
char cluster[GFS2_LOCKNAME_LEN];
const char *fsname;
uint32_t flags;
int error, ops_result;
/*
* initialize everything
*/
INIT_DELAYED_WORK(&sdp->sd_control_work, gfs2_control_func);
spin_lock_init(&ls->ls_recover_spin);
ls->ls_recover_flags = 0;
ls->ls_recover_mount = 0;
ls->ls_recover_start = 0;
ls->ls_recover_block = 0;
ls->ls_recover_size = 0;
ls->ls_recover_submit = NULL;
ls->ls_recover_result = NULL;
ls->ls_lvb_bits = NULL;
error = set_recover_size(sdp, NULL, 0);
if (error)
goto fail;
/*
* prepare dlm_new_lockspace args
*/
fsname = strchr(table, ':');
if (!fsname) {
fs_info(sdp, "no fsname found\n");
error = -EINVAL;
goto fail_free;
}
memset(cluster, 0, sizeof(cluster));
memcpy(cluster, table, strlen(table) - strlen(fsname));
fsname++;
flags = DLM_LSFL_FS | DLM_LSFL_NEWEXCL;
/*
* create/join lockspace
*/
error = dlm_new_lockspace(fsname, cluster, flags, GDLM_LVB_SIZE,
&gdlm_lockspace_ops, sdp, &ops_result,
&ls->ls_dlm);
if (error) {
fs_err(sdp, "dlm_new_lockspace error %d\n", error);
goto fail_free;
}
if (ops_result < 0) {
/*
* dlm does not support ops callbacks,
* old dlm_controld/gfs_controld are used, try without ops.
*/
fs_info(sdp, "dlm lockspace ops not used\n");
free_recover_size(ls);
set_bit(DFL_NO_DLM_OPS, &ls->ls_recover_flags);
return 0;
}
if (!test_bit(SDF_NOJOURNALID, &sdp->sd_flags)) {
fs_err(sdp, "dlm lockspace ops disallow jid preset\n");
error = -EINVAL;
goto fail_release;
}
/*
* control_mount() uses control_lock to determine first mounter,
* and for later mounts, waits for any recoveries to be cleared.
*/
error = control_mount(sdp);
if (error) {
fs_err(sdp, "mount control error %d\n", error);
goto fail_release;
}
ls->ls_first = !!test_bit(DFL_FIRST_MOUNT, &ls->ls_recover_flags);
clear_bit(SDF_NOJOURNALID, &sdp->sd_flags);
smp_mb__after_atomic();
wake_up_bit(&sdp->sd_flags, SDF_NOJOURNALID);
return 0;
fail_release:
dlm_release_lockspace(ls->ls_dlm, 2);
fail_free:
free_recover_size(ls);
fail:
return error;
}
