Commit 71e330b5 authored by Dave Chinner's avatar Dave Chinner Committed by Alex Elder

xfs: Introduce delayed logging core code

The delayed logging code only changes in-memory structures and as
such can be enabled and disabled with a mount option. Add the mount
option and emit a warning that this is an experimental feature that
should not be used in production yet.

We also need infrastructure to track committed items that have not
yet been written to the log. This is what the Committed Item List
(CIL) is for.

The log item also needs to be extended to track the current log
vector, the associated memory buffer and it's location in the Commit
Item List. Extend the log item and log vector structures to enable
this tracking.

To maintain the current log format for transactions with delayed
logging, we need to introduce a checkpoint transaction and a context
for tracking each checkpoint from initiation to transaction
completion.  This includes adding a log ticket for tracking space
log required/used by the context checkpoint.

To track all the changes we need an io vector array per log item,
rather than a single array for the entire transaction. Using the new
log vector structure for this requires two passes - the first to
allocate the log vector structures and chain them together, and the
second to fill them out.  This log vector chain can then be passed
to the CIL for formatting, pinning and insertion into the CIL.

Formatting of the log vector chain is relatively simple - it's just
a loop over the iovecs on each log vector, but it is made slightly
more complex because we re-write the iovec after the copy to point
back at the memory buffer we just copied into.

This code also needs to pin log items. If the log item is not
already tracked in this checkpoint context, then it needs to be
pinned. Otherwise it is already pinned and we don't need to pin it
again.

The only other complexity is calculating the amount of new log space
the formatting has consumed. This needs to be accounted to the
transaction in progress, and the accounting is made more complex
becase we need also to steal space from it for log metadata in the
checkpoint transaction. Calculate all this at insert time and update
all the tickets, counters, etc correctly.

Once we've formatted all the log items in the transaction, attach
the busy extents to the checkpoint context so the busy extents live
until checkpoint completion and can be processed at that point in
time. Transactions can then be freed at this point in time.

Now we need to issue checkpoints - we are tracking the amount of log space
used by the items in the CIL, so we can trigger background checkpoints when the
space usage gets to a certain threshold. Otherwise, checkpoints need ot be
triggered when a log synchronisation point is reached - a log force event.

Because the log write code already handles chained log vectors, writing the
transaction is trivial, too. Construct a transaction header, add it
to the head of the chain and write it into the log, then issue a
commit record write. Then we can release the checkpoint log ticket
and attach the context to the log buffer so it can be called during
Io completion to complete the checkpoint.

We also need to allow for synchronising multiple in-flight
checkpoints. This is needed for two things - the first is to ensure
that checkpoint commit records appear in the log in the correct
sequence order (so they are replayed in the correct order). The
second is so that xfs_log_force_lsn() operates correctly and only
flushes and/or waits for the specific sequence it was provided with.

To do this we need a wait variable and a list tracking the
checkpoint commits in progress. We can walk this list and wait for
the checkpoints to change state or complete easily, an this provides
the necessary synchronisation for correct operation in both cases.
Signed-off-by: default avatarDave Chinner <dchinner@redhat.com>
Reviewed-by: default avatarChristoph Hellwig <hch@lst.de>
Signed-off-by: default avatarAlex Elder <aelder@sgi.com>
parent a9a745da
...@@ -77,6 +77,7 @@ xfs-y += xfs_alloc.o \ ...@@ -77,6 +77,7 @@ xfs-y += xfs_alloc.o \
xfs_itable.o \ xfs_itable.o \
xfs_dfrag.o \ xfs_dfrag.o \
xfs_log.o \ xfs_log.o \
xfs_log_cil.o \
xfs_log_recover.o \ xfs_log_recover.o \
xfs_mount.o \ xfs_mount.o \
xfs_mru_cache.o \ xfs_mru_cache.o \
......
...@@ -119,6 +119,8 @@ mempool_t *xfs_ioend_pool; ...@@ -119,6 +119,8 @@ mempool_t *xfs_ioend_pool;
#define MNTOPT_DMAPI "dmapi" /* DMI enabled (DMAPI / XDSM) */ #define MNTOPT_DMAPI "dmapi" /* DMI enabled (DMAPI / XDSM) */
#define MNTOPT_XDSM "xdsm" /* DMI enabled (DMAPI / XDSM) */ #define MNTOPT_XDSM "xdsm" /* DMI enabled (DMAPI / XDSM) */
#define MNTOPT_DMI "dmi" /* DMI enabled (DMAPI / XDSM) */ #define MNTOPT_DMI "dmi" /* DMI enabled (DMAPI / XDSM) */
#define MNTOPT_DELAYLOG "delaylog" /* Delayed loging enabled */
#define MNTOPT_NODELAYLOG "nodelaylog" /* Delayed loging disabled */
/* /*
* Table driven mount option parser. * Table driven mount option parser.
...@@ -374,6 +376,13 @@ xfs_parseargs( ...@@ -374,6 +376,13 @@ xfs_parseargs(
mp->m_flags |= XFS_MOUNT_DMAPI; mp->m_flags |= XFS_MOUNT_DMAPI;
} else if (!strcmp(this_char, MNTOPT_DMI)) { } else if (!strcmp(this_char, MNTOPT_DMI)) {
mp->m_flags |= XFS_MOUNT_DMAPI; mp->m_flags |= XFS_MOUNT_DMAPI;
} else if (!strcmp(this_char, MNTOPT_DELAYLOG)) {
mp->m_flags |= XFS_MOUNT_DELAYLOG;
cmn_err(CE_WARN,
"Enabling EXPERIMENTAL delayed logging feature "
"- use at your own risk.\n");
} else if (!strcmp(this_char, MNTOPT_NODELAYLOG)) {
mp->m_flags &= ~XFS_MOUNT_DELAYLOG;
} else if (!strcmp(this_char, "ihashsize")) { } else if (!strcmp(this_char, "ihashsize")) {
cmn_err(CE_WARN, cmn_err(CE_WARN,
"XFS: ihashsize no longer used, option is deprecated."); "XFS: ihashsize no longer used, option is deprecated.");
...@@ -535,6 +544,7 @@ xfs_showargs( ...@@ -535,6 +544,7 @@ xfs_showargs(
{ XFS_MOUNT_FILESTREAMS, "," MNTOPT_FILESTREAM }, { XFS_MOUNT_FILESTREAMS, "," MNTOPT_FILESTREAM },
{ XFS_MOUNT_DMAPI, "," MNTOPT_DMAPI }, { XFS_MOUNT_DMAPI, "," MNTOPT_DMAPI },
{ XFS_MOUNT_GRPID, "," MNTOPT_GRPID }, { XFS_MOUNT_GRPID, "," MNTOPT_GRPID },
{ XFS_MOUNT_DELAYLOG, "," MNTOPT_DELAYLOG },
{ 0, NULL } { 0, NULL }
}; };
static struct proc_xfs_info xfs_info_unset[] = { static struct proc_xfs_info xfs_info_unset[] = {
......
...@@ -54,9 +54,6 @@ STATIC xlog_t * xlog_alloc_log(xfs_mount_t *mp, ...@@ -54,9 +54,6 @@ STATIC xlog_t * xlog_alloc_log(xfs_mount_t *mp,
STATIC int xlog_space_left(xlog_t *log, int cycle, int bytes); STATIC int xlog_space_left(xlog_t *log, int cycle, int bytes);
STATIC int xlog_sync(xlog_t *log, xlog_in_core_t *iclog); STATIC int xlog_sync(xlog_t *log, xlog_in_core_t *iclog);
STATIC void xlog_dealloc_log(xlog_t *log); STATIC void xlog_dealloc_log(xlog_t *log);
STATIC int xlog_write(struct log *log, struct xfs_log_vec *log_vector,
struct xlog_ticket *tic, xfs_lsn_t *start_lsn,
xlog_in_core_t **commit_iclog, uint flags);
/* local state machine functions */ /* local state machine functions */
STATIC void xlog_state_done_syncing(xlog_in_core_t *iclog, int); STATIC void xlog_state_done_syncing(xlog_in_core_t *iclog, int);
...@@ -86,12 +83,6 @@ STATIC int xlog_regrant_write_log_space(xlog_t *log, ...@@ -86,12 +83,6 @@ STATIC int xlog_regrant_write_log_space(xlog_t *log,
STATIC void xlog_ungrant_log_space(xlog_t *log, STATIC void xlog_ungrant_log_space(xlog_t *log,
xlog_ticket_t *ticket); xlog_ticket_t *ticket);
/* local ticket functions */
STATIC xlog_ticket_t *xlog_ticket_alloc(xlog_t *log, int unit_bytes, int count,
char clientid, uint flags,
int alloc_flags);
#if defined(DEBUG) #if defined(DEBUG)
STATIC void xlog_verify_dest_ptr(xlog_t *log, char *ptr); STATIC void xlog_verify_dest_ptr(xlog_t *log, char *ptr);
STATIC void xlog_verify_grant_head(xlog_t *log, int equals); STATIC void xlog_verify_grant_head(xlog_t *log, int equals);
...@@ -460,6 +451,13 @@ xfs_log_mount( ...@@ -460,6 +451,13 @@ xfs_log_mount(
/* Normal transactions can now occur */ /* Normal transactions can now occur */
mp->m_log->l_flags &= ~XLOG_ACTIVE_RECOVERY; mp->m_log->l_flags &= ~XLOG_ACTIVE_RECOVERY;
/*
* Now the log has been fully initialised and we know were our
* space grant counters are, we can initialise the permanent ticket
* needed for delayed logging to work.
*/
xlog_cil_init_post_recovery(mp->m_log);
return 0; return 0;
out_destroy_ail: out_destroy_ail:
...@@ -666,6 +664,10 @@ xfs_log_item_init( ...@@ -666,6 +664,10 @@ xfs_log_item_init(
item->li_ailp = mp->m_ail; item->li_ailp = mp->m_ail;
item->li_type = type; item->li_type = type;
item->li_ops = ops; item->li_ops = ops;
item->li_lv = NULL;
INIT_LIST_HEAD(&item->li_ail);
INIT_LIST_HEAD(&item->li_cil);
} }
/* /*
...@@ -1176,6 +1178,9 @@ xlog_alloc_log(xfs_mount_t *mp, ...@@ -1176,6 +1178,9 @@ xlog_alloc_log(xfs_mount_t *mp,
*iclogp = log->l_iclog; /* complete ring */ *iclogp = log->l_iclog; /* complete ring */
log->l_iclog->ic_prev = prev_iclog; /* re-write 1st prev ptr */ log->l_iclog->ic_prev = prev_iclog; /* re-write 1st prev ptr */
error = xlog_cil_init(log);
if (error)
goto out_free_iclog;
return log; return log;
out_free_iclog: out_free_iclog:
...@@ -1502,6 +1507,8 @@ xlog_dealloc_log(xlog_t *log) ...@@ -1502,6 +1507,8 @@ xlog_dealloc_log(xlog_t *log)
xlog_in_core_t *iclog, *next_iclog; xlog_in_core_t *iclog, *next_iclog;
int i; int i;
xlog_cil_destroy(log);
iclog = log->l_iclog; iclog = log->l_iclog;
for (i=0; i<log->l_iclog_bufs; i++) { for (i=0; i<log->l_iclog_bufs; i++) {
sv_destroy(&iclog->ic_force_wait); sv_destroy(&iclog->ic_force_wait);
...@@ -1544,8 +1551,10 @@ xlog_state_finish_copy(xlog_t *log, ...@@ -1544,8 +1551,10 @@ xlog_state_finish_copy(xlog_t *log,
* print out info relating to regions written which consume * print out info relating to regions written which consume
* the reservation * the reservation
*/ */
STATIC void void
xlog_print_tic_res(xfs_mount_t *mp, xlog_ticket_t *ticket) xlog_print_tic_res(
struct xfs_mount *mp,
struct xlog_ticket *ticket)
{ {
uint i; uint i;
uint ophdr_spc = ticket->t_res_num_ophdrs * (uint)sizeof(xlog_op_header_t); uint ophdr_spc = ticket->t_res_num_ophdrs * (uint)sizeof(xlog_op_header_t);
...@@ -1877,7 +1886,7 @@ xlog_write_copy_finish( ...@@ -1877,7 +1886,7 @@ xlog_write_copy_finish(
* we don't update ic_offset until the end when we know exactly how many * we don't update ic_offset until the end when we know exactly how many
* bytes have been written out. * bytes have been written out.
*/ */
STATIC int int
xlog_write( xlog_write(
struct log *log, struct log *log,
struct xfs_log_vec *log_vector, struct xfs_log_vec *log_vector,
...@@ -1901,9 +1910,26 @@ xlog_write( ...@@ -1901,9 +1910,26 @@ xlog_write(
*start_lsn = 0; *start_lsn = 0;
len = xlog_write_calc_vec_length(ticket, log_vector); len = xlog_write_calc_vec_length(ticket, log_vector);
if (ticket->t_curr_res < len) if (log->l_cilp) {
/*
* Region headers and bytes are already accounted for.
* We only need to take into account start records and
* split regions in this function.
*/
if (ticket->t_flags & XLOG_TIC_INITED)
ticket->t_curr_res -= sizeof(xlog_op_header_t);
/*
* Commit record headers need to be accounted for. These
* come in as separate writes so are easy to detect.
*/
if (flags & (XLOG_COMMIT_TRANS | XLOG_UNMOUNT_TRANS))
ticket->t_curr_res -= sizeof(xlog_op_header_t);
} else
ticket->t_curr_res -= len;
if (ticket->t_curr_res < 0)
xlog_print_tic_res(log->l_mp, ticket); xlog_print_tic_res(log->l_mp, ticket);
ticket->t_curr_res -= len;
index = 0; index = 0;
lv = log_vector; lv = log_vector;
...@@ -2999,6 +3025,8 @@ _xfs_log_force( ...@@ -2999,6 +3025,8 @@ _xfs_log_force(
XFS_STATS_INC(xs_log_force); XFS_STATS_INC(xs_log_force);
xlog_cil_push(log, 1);
spin_lock(&log->l_icloglock); spin_lock(&log->l_icloglock);
iclog = log->l_iclog; iclog = log->l_iclog;
...@@ -3148,6 +3176,12 @@ _xfs_log_force_lsn( ...@@ -3148,6 +3176,12 @@ _xfs_log_force_lsn(
XFS_STATS_INC(xs_log_force); XFS_STATS_INC(xs_log_force);
if (log->l_cilp) {
lsn = xlog_cil_push_lsn(log, lsn);
if (lsn == NULLCOMMITLSN)
return 0;
}
try_again: try_again:
spin_lock(&log->l_icloglock); spin_lock(&log->l_icloglock);
iclog = log->l_iclog; iclog = log->l_iclog;
...@@ -3322,7 +3356,7 @@ xfs_log_get_trans_ident( ...@@ -3322,7 +3356,7 @@ xfs_log_get_trans_ident(
/* /*
* Allocate and initialise a new log ticket. * Allocate and initialise a new log ticket.
*/ */
STATIC xlog_ticket_t * xlog_ticket_t *
xlog_ticket_alloc( xlog_ticket_alloc(
struct log *log, struct log *log,
int unit_bytes, int unit_bytes,
......
...@@ -113,6 +113,9 @@ struct xfs_log_vec { ...@@ -113,6 +113,9 @@ struct xfs_log_vec {
struct xfs_log_vec *lv_next; /* next lv in build list */ struct xfs_log_vec *lv_next; /* next lv in build list */
int lv_niovecs; /* number of iovecs in lv */ int lv_niovecs; /* number of iovecs in lv */
struct xfs_log_iovec *lv_iovecp; /* iovec array */ struct xfs_log_iovec *lv_iovecp; /* iovec array */
struct xfs_log_item *lv_item; /* owner */
char *lv_buf; /* formatted buffer */
int lv_buf_len; /* size of formatted buffer */
}; };
/* /*
...@@ -187,11 +190,15 @@ int xfs_log_need_covered(struct xfs_mount *mp); ...@@ -187,11 +190,15 @@ int xfs_log_need_covered(struct xfs_mount *mp);
void xlog_iodone(struct xfs_buf *); void xlog_iodone(struct xfs_buf *);
struct xlog_ticket * xfs_log_ticket_get(struct xlog_ticket *ticket); struct xlog_ticket *xfs_log_ticket_get(struct xlog_ticket *ticket);
void xfs_log_ticket_put(struct xlog_ticket *ticket); void xfs_log_ticket_put(struct xlog_ticket *ticket);
xlog_tid_t xfs_log_get_trans_ident(struct xfs_trans *tp); xlog_tid_t xfs_log_get_trans_ident(struct xfs_trans *tp);
int xfs_log_commit_cil(struct xfs_mount *mp, struct xfs_trans *tp,
struct xfs_log_vec *log_vector,
xfs_lsn_t *commit_lsn, int flags);
#endif #endif
......
This diff is collapsed.
...@@ -376,6 +376,54 @@ typedef struct xlog_in_core { ...@@ -376,6 +376,54 @@ typedef struct xlog_in_core {
#define ic_header ic_data->hic_header #define ic_header ic_data->hic_header
} xlog_in_core_t; } xlog_in_core_t;
/*
* The CIL context is used to aggregate per-transaction details as well be
* passed to the iclog for checkpoint post-commit processing. After being
* passed to the iclog, another context needs to be allocated for tracking the
* next set of transactions to be aggregated into a checkpoint.
*/
struct xfs_cil;
struct xfs_cil_ctx {
struct xfs_cil *cil;
xfs_lsn_t sequence; /* chkpt sequence # */
xfs_lsn_t start_lsn; /* first LSN of chkpt commit */
xfs_lsn_t commit_lsn; /* chkpt commit record lsn */
struct xlog_ticket *ticket; /* chkpt ticket */
int nvecs; /* number of regions */
int space_used; /* aggregate size of regions */
struct list_head busy_extents; /* busy extents in chkpt */
struct xfs_log_vec *lv_chain; /* logvecs being pushed */
xfs_log_callback_t log_cb; /* completion callback hook. */
struct list_head committing; /* ctx committing list */
};
/*
* Committed Item List structure
*
* This structure is used to track log items that have been committed but not
* yet written into the log. It is used only when the delayed logging mount
* option is enabled.
*
* This structure tracks the list of committing checkpoint contexts so
* we can avoid the problem of having to hold out new transactions during a
* flush until we have a the commit record LSN of the checkpoint. We can
* traverse the list of committing contexts in xlog_cil_push_lsn() to find a
* sequence match and extract the commit LSN directly from there. If the
* checkpoint is still in the process of committing, we can block waiting for
* the commit LSN to be determined as well. This should make synchronous
* operations almost as efficient as the old logging methods.
*/
struct xfs_cil {
struct log *xc_log;
struct list_head xc_cil;
spinlock_t xc_cil_lock;
struct xfs_cil_ctx *xc_ctx;
struct rw_semaphore xc_ctx_lock;
struct list_head xc_committing;
sv_t xc_commit_wait;
};
/* /*
* The reservation head lsn is not made up of a cycle number and block number. * The reservation head lsn is not made up of a cycle number and block number.
* Instead, it uses a cycle number and byte number. Logs don't expect to * Instead, it uses a cycle number and byte number. Logs don't expect to
...@@ -386,6 +434,7 @@ typedef struct log { ...@@ -386,6 +434,7 @@ typedef struct log {
/* The following fields don't need locking */ /* The following fields don't need locking */
struct xfs_mount *l_mp; /* mount point */ struct xfs_mount *l_mp; /* mount point */
struct xfs_ail *l_ailp; /* AIL log is working with */ struct xfs_ail *l_ailp; /* AIL log is working with */
struct xfs_cil *l_cilp; /* CIL log is working with */
struct xfs_buf *l_xbuf; /* extra buffer for log struct xfs_buf *l_xbuf; /* extra buffer for log
* wrapping */ * wrapping */
struct xfs_buftarg *l_targ; /* buftarg of log */ struct xfs_buftarg *l_targ; /* buftarg of log */
...@@ -436,14 +485,17 @@ typedef struct log { ...@@ -436,14 +485,17 @@ typedef struct log {
#define XLOG_FORCED_SHUTDOWN(log) ((log)->l_flags & XLOG_IO_ERROR) #define XLOG_FORCED_SHUTDOWN(log) ((log)->l_flags & XLOG_IO_ERROR)
/* common routines */ /* common routines */
extern xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp); extern xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
extern int xlog_recover(xlog_t *log); extern int xlog_recover(xlog_t *log);
extern int xlog_recover_finish(xlog_t *log); extern int xlog_recover_finish(xlog_t *log);
extern void xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int); extern void xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
extern kmem_zone_t *xfs_log_ticket_zone; extern kmem_zone_t *xfs_log_ticket_zone;
struct xlog_ticket *xlog_ticket_alloc(struct log *log, int unit_bytes,
int count, char client, uint xflags,
int alloc_flags);
static inline void static inline void
xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes) xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes)
...@@ -453,6 +505,21 @@ xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes) ...@@ -453,6 +505,21 @@ xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes)
*off += bytes; *off += bytes;
} }
void xlog_print_tic_res(struct xfs_mount *mp, struct xlog_ticket *ticket);
int xlog_write(struct log *log, struct xfs_log_vec *log_vector,
struct xlog_ticket *tic, xfs_lsn_t *start_lsn,
xlog_in_core_t **commit_iclog, uint flags);
/*
* Committed Item List interfaces
*/
int xlog_cil_init(struct log *log);
void xlog_cil_init_post_recovery(struct log *log);
void xlog_cil_destroy(struct log *log);
int xlog_cil_push(struct log *log, int push_now);
xfs_lsn_t xlog_cil_push_lsn(struct log *log, xfs_lsn_t push_sequence);
/* /*
* Unmount record type is used as a pseudo transaction type for the ticket. * Unmount record type is used as a pseudo transaction type for the ticket.
* It's value must be outside the range of XFS_TRANS_* values. * It's value must be outside the range of XFS_TRANS_* values.
......
...@@ -268,6 +268,7 @@ typedef struct xfs_mount { ...@@ -268,6 +268,7 @@ typedef struct xfs_mount {
#define XFS_MOUNT_WSYNC (1ULL << 0) /* for nfs - all metadata ops #define XFS_MOUNT_WSYNC (1ULL << 0) /* for nfs - all metadata ops
must be synchronous except must be synchronous except
for space allocations */ for space allocations */
#define XFS_MOUNT_DELAYLOG (1ULL << 1) /* delayed logging is enabled */
#define XFS_MOUNT_DMAPI (1ULL << 2) /* dmapi is enabled */ #define XFS_MOUNT_DMAPI (1ULL << 2) /* dmapi is enabled */
#define XFS_MOUNT_WAS_CLEAN (1ULL << 3) #define XFS_MOUNT_WAS_CLEAN (1ULL << 3)
#define XFS_MOUNT_FS_SHUTDOWN (1ULL << 4) /* atomic stop of all filesystem #define XFS_MOUNT_FS_SHUTDOWN (1ULL << 4) /* atomic stop of all filesystem
......
...@@ -655,7 +655,7 @@ xfs_trans_apply_sb_deltas( ...@@ -655,7 +655,7 @@ xfs_trans_apply_sb_deltas(
* XFS_TRANS_SB_DIRTY will not be set when the transaction is updated but we * XFS_TRANS_SB_DIRTY will not be set when the transaction is updated but we
* still need to update the incore superblock with the changes. * still need to update the incore superblock with the changes.
*/ */
STATIC void void
xfs_trans_unreserve_and_mod_sb( xfs_trans_unreserve_and_mod_sb(
xfs_trans_t *tp) xfs_trans_t *tp)
{ {
...@@ -883,7 +883,7 @@ xfs_trans_fill_vecs( ...@@ -883,7 +883,7 @@ xfs_trans_fill_vecs(
* they could be immediately flushed and we'd have to race with the flusher * they could be immediately flushed and we'd have to race with the flusher
* trying to pull the item from the AIL as we add it. * trying to pull the item from the AIL as we add it.
*/ */
static void void
xfs_trans_item_committed( xfs_trans_item_committed(
struct xfs_log_item *lip, struct xfs_log_item *lip,
xfs_lsn_t commit_lsn, xfs_lsn_t commit_lsn,
...@@ -994,7 +994,7 @@ xfs_trans_uncommit( ...@@ -994,7 +994,7 @@ xfs_trans_uncommit(
xfs_trans_unreserve_and_mod_sb(tp); xfs_trans_unreserve_and_mod_sb(tp);
xfs_trans_unreserve_and_mod_dquots(tp); xfs_trans_unreserve_and_mod_dquots(tp);
xfs_trans_free_items(tp, flags); xfs_trans_free_items(tp, NULLCOMMITLSN, flags);
xfs_trans_free(tp); xfs_trans_free(tp);
} }
...@@ -1144,6 +1144,93 @@ xfs_trans_commit_iclog( ...@@ -1144,6 +1144,93 @@ xfs_trans_commit_iclog(
return xfs_log_release_iclog(mp, commit_iclog); return xfs_log_release_iclog(mp, commit_iclog);
} }
/*
* Walk the log items and allocate log vector structures for
* each item large enough to fit all the vectors they require.
* Note that this format differs from the old log vector format in
* that there is no transaction header in these log vectors.
*/
STATIC struct xfs_log_vec *
xfs_trans_alloc_log_vecs(
xfs_trans_t *tp)
{
xfs_log_item_desc_t *lidp;
struct xfs_log_vec *lv = NULL;
struct xfs_log_vec *ret_lv = NULL;
lidp = xfs_trans_first_item(tp);
/* Bail out if we didn't find a log item. */
if (!lidp) {
ASSERT(0);
return NULL;
}
while (lidp != NULL) {
struct xfs_log_vec *new_lv;
/* Skip items which aren't dirty in this transaction. */
if (!(lidp->lid_flags & XFS_LID_DIRTY)) {
lidp = xfs_trans_next_item(tp, lidp);
continue;
}
/* Skip items that do not have any vectors for writing */
lidp->lid_size = IOP_SIZE(lidp->lid_item);
if (!lidp->lid_size) {
lidp = xfs_trans_next_item(tp, lidp);
continue;
}
new_lv = kmem_zalloc(sizeof(*new_lv) +
lidp->lid_size * sizeof(struct xfs_log_iovec),
KM_SLEEP);
/* The allocated iovec region lies beyond the log vector. */
new_lv->lv_iovecp = (struct xfs_log_iovec *)&new_lv[1];
new_lv->lv_niovecs = lidp->lid_size;
new_lv->lv_item = lidp->lid_item;
if (!ret_lv)
ret_lv = new_lv;
else
lv->lv_next = new_lv;
lv = new_lv;
lidp = xfs_trans_next_item(tp, lidp);
}
return ret_lv;
}
static int
xfs_trans_commit_cil(
struct xfs_mount *mp,
struct xfs_trans *tp,
xfs_lsn_t *commit_lsn,
int flags)
{
struct xfs_log_vec *log_vector;
int error;
/*
* Get each log item to allocate a vector structure for
* the log item to to pass to the log write code. The
* CIL commit code will format the vector and save it away.
*/
log_vector = xfs_trans_alloc_log_vecs(tp);
if (!log_vector)
return ENOMEM;
error = xfs_log_commit_cil(mp, tp, log_vector, commit_lsn, flags);
if (error)
return error;
current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
/* xfs_trans_free_items() unlocks them first */
xfs_trans_free_items(tp, *commit_lsn, 0);
xfs_trans_free(tp);
return 0;
}
/* /*
* xfs_trans_commit * xfs_trans_commit
...@@ -1204,7 +1291,11 @@ _xfs_trans_commit( ...@@ -1204,7 +1291,11 @@ _xfs_trans_commit(
xfs_trans_apply_sb_deltas(tp); xfs_trans_apply_sb_deltas(tp);
xfs_trans_apply_dquot_deltas(tp); xfs_trans_apply_dquot_deltas(tp);
error = xfs_trans_commit_iclog(mp, tp, &commit_lsn, flags); if (mp->m_flags & XFS_MOUNT_DELAYLOG)
error = xfs_trans_commit_cil(mp, tp, &commit_lsn, flags);
else
error = xfs_trans_commit_iclog(mp, tp, &commit_lsn, flags);
if (error == ENOMEM) { if (error == ENOMEM) {
xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR); xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
error = XFS_ERROR(EIO); error = XFS_ERROR(EIO);
...@@ -1242,7 +1333,7 @@ _xfs_trans_commit( ...@@ -1242,7 +1333,7 @@ _xfs_trans_commit(
error = XFS_ERROR(EIO); error = XFS_ERROR(EIO);
} }
current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS); current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
xfs_trans_free_items(tp, error ? XFS_TRANS_ABORT : 0); xfs_trans_free_items(tp, NULLCOMMITLSN, error ? XFS_TRANS_ABORT : 0);
xfs_trans_free(tp); xfs_trans_free(tp);
XFS_STATS_INC(xs_trans_empty); XFS_STATS_INC(xs_trans_empty);
...@@ -1320,7 +1411,7 @@ xfs_trans_cancel( ...@@ -1320,7 +1411,7 @@ xfs_trans_cancel(
/* mark this thread as no longer being in a transaction */ /* mark this thread as no longer being in a transaction */
current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS); current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
xfs_trans_free_items(tp, flags); xfs_trans_free_items(tp, NULLCOMMITLSN, flags);
xfs_trans_free(tp); xfs_trans_free(tp);
} }
......
...@@ -106,7 +106,8 @@ typedef struct xfs_trans_header { ...@@ -106,7 +106,8 @@ typedef struct xfs_trans_header {
#define XFS_TRANS_GROWFSRT_FREE 39 #define XFS_TRANS_GROWFSRT_FREE 39
#define XFS_TRANS_SWAPEXT 40 #define XFS_TRANS_SWAPEXT 40
#define XFS_TRANS_SB_COUNT 41 #define XFS_TRANS_SB_COUNT 41
#define XFS_TRANS_TYPE_MAX 41 #define XFS_TRANS_CHECKPOINT 42
#define XFS_TRANS_TYPE_MAX 42
/* new transaction types need to be reflected in xfs_logprint(8) */ /* new transaction types need to be reflected in xfs_logprint(8) */
#define XFS_TRANS_TYPES \ #define XFS_TRANS_TYPES \
...@@ -148,6 +149,7 @@ typedef struct xfs_trans_header { ...@@ -148,6 +149,7 @@ typedef struct xfs_trans_header {
{ XFS_TRANS_GROWFSRT_FREE, "GROWFSRT_FREE" }, \ { XFS_TRANS_GROWFSRT_FREE, "GROWFSRT_FREE" }, \
{ XFS_TRANS_SWAPEXT, "SWAPEXT" }, \ { XFS_TRANS_SWAPEXT, "SWAPEXT" }, \
{ XFS_TRANS_SB_COUNT, "SB_COUNT" }, \ { XFS_TRANS_SB_COUNT, "SB_COUNT" }, \
{ XFS_TRANS_CHECKPOINT, "CHECKPOINT" }, \
{ XFS_TRANS_DUMMY1, "DUMMY1" }, \ { XFS_TRANS_DUMMY1, "DUMMY1" }, \
{ XFS_TRANS_DUMMY2, "DUMMY2" }, \ { XFS_TRANS_DUMMY2, "DUMMY2" }, \
{ XLOG_UNMOUNT_REC_TYPE, "UNMOUNT" } { XLOG_UNMOUNT_REC_TYPE, "UNMOUNT" }
...@@ -829,6 +831,10 @@ typedef struct xfs_log_item { ...@@ -829,6 +831,10 @@ typedef struct xfs_log_item {
/* buffer item iodone */ /* buffer item iodone */
/* callback func */ /* callback func */
struct xfs_item_ops *li_ops; /* function list */ struct xfs_item_ops *li_ops; /* function list */
/* delayed logging */
struct list_head li_cil; /* CIL pointers */
struct xfs_log_vec *li_lv; /* active log vector */
} xfs_log_item_t; } xfs_log_item_t;
#define XFS_LI_IN_AIL 0x1 #define XFS_LI_IN_AIL 0x1
......
...@@ -299,6 +299,7 @@ xfs_trans_next_item(xfs_trans_t *tp, xfs_log_item_desc_t *lidp) ...@@ -299,6 +299,7 @@ xfs_trans_next_item(xfs_trans_t *tp, xfs_log_item_desc_t *lidp)
void void
xfs_trans_free_items( xfs_trans_free_items(
xfs_trans_t *tp, xfs_trans_t *tp,
xfs_lsn_t commit_lsn,
int flags) int flags)
{ {
xfs_log_item_chunk_t *licp; xfs_log_item_chunk_t *licp;
...@@ -311,7 +312,7 @@ xfs_trans_free_items( ...@@ -311,7 +312,7 @@ xfs_trans_free_items(
* Special case the embedded chunk so we don't free it below. * Special case the embedded chunk so we don't free it below.
*/ */
if (!xfs_lic_are_all_free(licp)) { if (!xfs_lic_are_all_free(licp)) {
(void) xfs_trans_unlock_chunk(licp, 1, abort, NULLCOMMITLSN); (void) xfs_trans_unlock_chunk(licp, 1, abort, commit_lsn);
xfs_lic_all_free(licp); xfs_lic_all_free(licp);
licp->lic_unused = 0; licp->lic_unused = 0;
} }
...@@ -322,7 +323,7 @@ xfs_trans_free_items( ...@@ -322,7 +323,7 @@ xfs_trans_free_items(
*/ */
while (licp != NULL) { while (licp != NULL) {
ASSERT(!xfs_lic_are_all_free(licp)); ASSERT(!xfs_lic_are_all_free(licp));
(void) xfs_trans_unlock_chunk(licp, 1, abort, NULLCOMMITLSN); (void) xfs_trans_unlock_chunk(licp, 1, abort, commit_lsn);
next_licp = licp->lic_next; next_licp = licp->lic_next;
kmem_free(licp); kmem_free(licp);
licp = next_licp; licp = next_licp;
......
...@@ -35,9 +35,14 @@ struct xfs_log_item_desc *xfs_trans_find_item(struct xfs_trans *, ...@@ -35,9 +35,14 @@ struct xfs_log_item_desc *xfs_trans_find_item(struct xfs_trans *,
struct xfs_log_item_desc *xfs_trans_first_item(struct xfs_trans *); struct xfs_log_item_desc *xfs_trans_first_item(struct xfs_trans *);
struct xfs_log_item_desc *xfs_trans_next_item(struct xfs_trans *, struct xfs_log_item_desc *xfs_trans_next_item(struct xfs_trans *,
struct xfs_log_item_desc *); struct xfs_log_item_desc *);
void xfs_trans_free_items(struct xfs_trans *, int);
void xfs_trans_unlock_items(struct xfs_trans *, void xfs_trans_unlock_items(struct xfs_trans *tp, xfs_lsn_t commit_lsn);
xfs_lsn_t); void xfs_trans_free_items(struct xfs_trans *tp, xfs_lsn_t commit_lsn,
int flags);
void xfs_trans_item_committed(struct xfs_log_item *lip,
xfs_lsn_t commit_lsn, int aborted);
void xfs_trans_unreserve_and_mod_sb(struct xfs_trans *tp);
/* /*
* AIL traversal cursor. * AIL traversal cursor.
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment