*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
+ * the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
- *
+ *
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
* the GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
+ * along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#ifndef _H_JFS_LOGMGR
/*
* log logical volume
*
- * a log is used to make the commit operation on journalled
+ * a log is used to make the commit operation on journalled
* files within the same logical volume group atomic.
* a log is implemented with a logical volume.
- * there is one log per logical volume group.
+ * there is one log per logical volume group.
*
* block 0 of the log logical volume is not used (ipl etc).
* block 1 contains a log "superblock" and is used by logFormat(),
- * lmLogInit(), lmLogShutdown(), and logRedo() to record status
- * of the log but is not otherwise used during normal processing.
+ * lmLogInit(), lmLogShutdown(), and logRedo() to record status
+ * of the log but is not otherwise used during normal processing.
* blocks 2 - (N-1) are used to contain log records.
*
- * when a volume group is varied-on-line, logRedo() must have
- * been executed before the file systems (logical volumes) in
+ * when a volume group is varied-on-line, logRedo() must have
+ * been executed before the file systems (logical volumes) in
* the volume group can be mounted.
*/
/*
#define MAX_ACTIVE 128 /* Max active file systems sharing log */
struct logsuper {
- u32 magic; /* 4: log lv identifier */
- s32 version; /* 4: version number */
- s32 serial; /* 4: log open/mount counter */
- s32 size; /* 4: size in number of LOGPSIZE blocks */
- s32 bsize; /* 4: logical block size in byte */
- s32 l2bsize; /* 4: log2 of bsize */
+ __le32 magic; /* 4: log lv identifier */
+ __le32 version; /* 4: version number */
+ __le32 serial; /* 4: log open/mount counter */
+ __le32 size; /* 4: size in number of LOGPSIZE blocks */
+ __le32 bsize; /* 4: logical block size in byte */
+ __le32 l2bsize; /* 4: log2 of bsize */
- u32 flag; /* 4: option */
- u32 state; /* 4: state - see below */
+ __le32 flag; /* 4: option */
+ __le32 state; /* 4: state - see below */
- s32 end; /* 4: addr of last log record set by logredo */
+ __le32 end; /* 4: addr of last log record set by logredo */
char uuid[16]; /* 16: 128-bit journal uuid */
char label[16]; /* 16: journal label */
struct {
* log logical page
*
* (this comment should be rewritten !)
- * the header and trailer structures (h,t) will normally have
+ * the header and trailer structures (h,t) will normally have
* the same page and eor value.
- * An exception to this occurs when a complete page write is not
+ * An exception to this occurs when a complete page write is not
* accomplished on a power failure. Since the hardware may "split write"
- * sectors in the page, any out of order sequence may occur during powerfail
+ * sectors in the page, any out of order sequence may occur during powerfail
* and needs to be recognized during log replay. The xor value is
* an "exclusive or" of all log words in the page up to eor. This
* 32 bit eor is stored with the top 16 bits in the header and the
* bottom 16 bits in the trailer. logredo can easily recognize pages
- * that were not completed by reconstructing this eor and checking
+ * that were not completed by reconstructing this eor and checking
* the log page.
*
- * Previous versions of the operating system did not allow split
- * writes and detected partially written records in logredo by
- * ordering the updates to the header, trailer, and the move of data
- * into the logdata area. The order: (1) data is moved (2) header
- * is updated (3) trailer is updated. In logredo, when the header
- * differed from the trailer, the header and trailer were reconciled
- * as follows: if h.page != t.page they were set to the smaller of
- * the two and h.eor and t.eor set to 8 (i.e. empty page). if (only)
+ * Previous versions of the operating system did not allow split
+ * writes and detected partially written records in logredo by
+ * ordering the updates to the header, trailer, and the move of data
+ * into the logdata area. The order: (1) data is moved (2) header
+ * is updated (3) trailer is updated. In logredo, when the header
+ * differed from the trailer, the header and trailer were reconciled
+ * as follows: if h.page != t.page they were set to the smaller of
+ * the two and h.eor and t.eor set to 8 (i.e. empty page). if (only)
* h.eor != t.eor they were set to the smaller of their two values.
*/
struct logpage {
struct { /* header */
- s32 page; /* 4: log sequence page number */
- s16 rsrvd; /* 2: */
- s16 eor; /* 2: end-of-log offset of lasrt record write */
+ __le32 page; /* 4: log sequence page number */
+ __le16 rsrvd; /* 2: */
+ __le16 eor; /* 2: end-of-log offset of lasrt record write */
} h;
- s32 data[LOGPSIZE / 4 - 4]; /* log record area */
+ __le32 data[LOGPSIZE / 4 - 4]; /* log record area */
struct { /* trailer */
- s32 page; /* 4: normally the same as h.page */
- s16 rsrvd; /* 2: */
- s16 eor; /* 2: normally the same as h.eor */
+ __le32 page; /* 4: normally the same as h.page */
+ __le16 rsrvd; /* 2: */
+ __le16 eor; /* 2: normally the same as h.eor */
} t;
};
* in a page, pages are written to temporary paging space if
* if they must be written to disk before commit, and i/o is
* scheduled for modified pages to their home location after
- * the log records containing the after values and the commit
+ * the log records containing the after values and the commit
* record is written to the log on disk, undo discards the copy
* in main-memory.)
*
- * a log record consists of a data area of variable length followed by
+ * a log record consists of a data area of variable length followed by
* a descriptor of fixed size LOGRDSIZE bytes.
- * the data area is rounded up to an integral number of 4-bytes and
+ * the data area is rounded up to an integral number of 4-bytes and
* must be no longer than LOGPSIZE.
- * the descriptor is of size of multiple of 4-bytes and aligned on a
- * 4-byte boundary.
+ * the descriptor is of size of multiple of 4-bytes and aligned on a
+ * 4-byte boundary.
* records are packed one after the other in the data area of log pages.
- * (sometimes a DUMMY record is inserted so that at least one record ends
+ * (sometimes a DUMMY record is inserted so that at least one record ends
* on every page or the longest record is placed on at most two pages).
- * the field eor in page header/trailer points to the byte following
+ * the field eor in page header/trailer points to the byte following
* the last record on a page.
*/
/*
* type independent area
*/
- s32 logtid; /* 4: log transaction identifier */
- s32 backchain; /* 4: ptr to prev record of same transaction */
- u16 type; /* 2: record type */
- s16 length; /* 2: length of data in record (in byte) */
- u32 aggregate; /* 4: file system lv/aggregate */
+ __le32 logtid; /* 4: log transaction identifier */
+ __le32 backchain; /* 4: ptr to prev record of same transaction */
+ __le16 type; /* 2: record type */
+ __le16 length; /* 2: length of data in record (in byte) */
+ __le32 aggregate; /* 4: file system lv/aggregate */
/* (16) */
/*
* N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
*/
struct {
- u32 fileset; /* 4: fileset number */
- u32 inode; /* 4: inode number */
- u16 type; /* 2: REDOPAGE record type */
- s16 l2linesize; /* 2: log2 of line size */
+ __le32 fileset; /* 4: fileset number */
+ __le32 inode; /* 4: inode number */
+ __le16 type; /* 2: REDOPAGE record type */
+ __le16 l2linesize; /* 2: log2 of line size */
pxd_t pxd; /* 8: on-disk page pxd */
} redopage; /* (20) */
* N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
*/
struct {
- s32 fileset; /* 4: fileset number */
- u32 inode; /* 4: inode number */
- u16 type; /* 2: NOREDOPAGE record type */
- s16 rsrvd; /* 2: reserved */
+ __le32 fileset; /* 4: fileset number */
+ __le32 inode; /* 4: inode number */
+ __le16 type; /* 2: NOREDOPAGE record type */
+ __le16 rsrvd; /* 2: reserved */
pxd_t pxd; /* 8: on-disk page pxd */
} noredopage; /* (20) */
* N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
*/
struct {
- u32 fileset; /* 4: fileset number */
- u32 inode; /* 4: inode number */
- u16 type; /* 2: UPDATEMAP record type */
- s16 nxd; /* 2: number of extents */
+ __le32 fileset; /* 4: fileset number */
+ __le32 inode; /* 4: inode number */
+ __le16 type; /* 2: UPDATEMAP record type */
+ __le16 nxd; /* 2: number of extents */
pxd_t pxd; /* 8: pxd */
} updatemap; /* (20) */
/*
* NOREDOINOEXT: the inode extent is freed
*
- * do not apply after-image records which precede this
- * record in the log with the any of the 4 page block
- * numbers in this inode extent.
- *
- * NOTE: The fileset and pxd fields MUST remain in
+ * do not apply after-image records which precede this
+ * record in the log with the any of the 4 page block
+ * numbers in this inode extent.
+ *
+ * NOTE: The fileset and pxd fields MUST remain in
* the same fields in the REDOPAGE record format.
*
*/
struct {
- s32 fileset; /* 4: fileset number */
- s32 iagnum; /* 4: IAG number */
- s32 inoext_idx; /* 4: inode extent index */
+ __le32 fileset; /* 4: fileset number */
+ __le32 iagnum; /* 4: IAG number */
+ __le32 inoext_idx; /* 4: inode extent index */
pxd_t pxd; /* 8: on-disk page pxd */
} noredoinoext; /* (20) */
* replay log upto syncpt address specified;
*/
struct {
- s32 sync; /* 4: syncpt address (0 = here) */
+ __le32 sync; /* 4: syncpt address (0 = here) */
} syncpt;
/*
* N.B.: nextents should be length of data/sizeof(xad_t)
*/
struct {
- s32 type; /* 4: FREEXTENT record type */
- s32 nextent; /* 4: number of extents */
+ __le32 type; /* 4: FREEXTENT record type */
+ __le32 nextent; /* 4: number of extents */
/* data: PXD or XAD list */
} freextent;
* do not apply records which precede this record in the log
* with the same inode number.
*
- * NOREDILE must be the first to be written at commit
+ * NOREDOFILE must be the first to be written at commit
* (last to be read in logredo()) - it prevents
* replay of preceding updates of all preceding generations
- * of the inumber esp. the on-disk inode itself,
- * but does NOT prevent
- * replay of the
+ * of the inumber esp. the on-disk inode itself.
*/
struct {
- s32 fileset; /* 4: fileset number */
- u32 inode; /* 4: inode number */
+ __le32 fileset; /* 4: fileset number */
+ __le32 inode; /* 4: inode number */
} noredofile;
/*
- * ? NEWPAGE:
+ * ? NEWPAGE:
*
* metadata type dependent
*/
struct {
- s32 fileset; /* 4: fileset number */
- u32 inode; /* 4: inode number */
- s32 type; /* 4: NEWPAGE record type */
+ __le32 fileset; /* 4: fileset number */
+ __le32 inode; /* 4: inode number */
+ __le32 type; /* 4: NEWPAGE record type */
pxd_t pxd; /* 8: on-disk page pxd */
} newpage;
* line vector descriptor
*/
struct lvd {
- s16 offset;
- s16 length;
+ __le16 offset;
+ __le16 length;
};
*/
struct list_head journal_list; /* Global list */
struct block_device *bdev; /* 4: log lv pointer */
- s32 serial; /* 4: log mount serial number */
+ int serial; /* 4: log mount serial number */
s64 base; /* @8: log extent address (inline log ) */
int size; /* 4: log size in log page (in page) */
int eor; /* 4: eor of last record in eol page */
struct lbuf *bp; /* 4: current log page buffer */
- struct semaphore loglock; /* 4: log write serialization lock */
+ struct mutex loglock; /* 4: log write serialization lock */
/* syncpt */
int nextsync; /* 4: bytes to write before next syncpt */
/* commit */
uint cflag; /* 4: */
- struct list_head cqueue; /* FIFO commit queue */
+ struct list_head cqueue; /* FIFO commit queue */
struct tblock *flush_tblk; /* tblk we're waiting on for flush */
int gcrtc; /* 4: GC_READY transaction count */
struct tblock *gclrt; /* 4: latest GC_READY transaction */
s64 l_blkno; /* 8: log page block number */
caddr_t l_ldata; /* 4: data page */
+ struct page *l_page; /* The page itself */
+ uint l_offset; /* Offset of l_ldata within the page */
wait_queue_head_t l_ioevent; /* 4: i/o done event */
- struct page *l_page; /* The page itself */
};
/* Reuse l_freelist for redrive list */
*/
#define LOGSYNC_LOCK_INIT(log) spin_lock_init(&(log)->synclock)
-#define LOGSYNC_LOCK(log) spin_lock(&(log)->synclock)
-#define LOGSYNC_UNLOCK(log) spin_unlock(&(log)->synclock)
+#define LOGSYNC_LOCK(log, flags) spin_lock_irqsave(&(log)->synclock, flags)
+#define LOGSYNC_UNLOCK(log, flags) \
+ spin_unlock_irqrestore(&(log)->synclock, flags)
/* compute the difference in bytes of lsn from sync point */
#define logdiff(diff, lsn, log)\
extern int lmLogShutdown(struct jfs_log * log);
extern int lmLogInit(struct jfs_log * log);
extern int lmLogFormat(struct jfs_log *log, s64 logAddress, int logSize);
+extern int lmGroupCommit(struct jfs_log *, struct tblock *);
+extern int jfsIOWait(void *);
extern void jfs_flush_journal(struct jfs_log * log, int wait);
+extern void jfs_syncpt(struct jfs_log *log, int hard_sync);
#endif /* _H_JFS_LOGMGR */
git://git.onelab.eu / linux-2.6.git / blobdiff