summaryrefslogtreecommitdiffstats
path: root/fs/xfs/linux-2.6/xfs_buf.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/xfs/linux-2.6/xfs_buf.c')
-rw-r--r--fs/xfs/linux-2.6/xfs_buf.c227
1 files changed, 63 insertions, 164 deletions
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index e82cf72ac599..6fe21d2b8847 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -1,46 +1,20 @@
/*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc. All Rights Reserved.
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
*
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of version 2 of the GNU General Public License as
+ * 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.
*
- * This program is distributed in the hope that it would be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
- *
- * Further, this software is distributed without any warranty that it is
- * free of the rightful claim of any third person regarding infringement
- * or the like. Any license provided herein, whether implied or
- * otherwise, applies only to this software file. Patent licenses, if
- * any, provided herein do not apply to combinations of this program with
- * other software, or any other product whatsoever.
- *
- * You should have received a copy of the GNU General Public License along
- * with this program; if not, write the Free Software Foundation, Inc., 59
- * Temple Place - Suite 330, Boston MA 02111-1307, USA.
- *
- * Contact information: Silicon Graphics, Inc., 1600 Amphitheatre Pkwy,
- * Mountain View, CA 94043, or:
- *
- * http://www.sgi.com
- *
- * For further information regarding this notice, see:
- *
- * http://oss.sgi.com/projects/GenInfo/SGIGPLNoticeExplan/
- */
-
-/*
- * The xfs_buf.c code provides an abstract buffer cache model on top
- * of the Linux page cache. Cached metadata blocks for a file system
- * are hashed to the inode for the block device. xfs_buf.c assembles
- * buffers (xfs_buf_t) on demand to aggregate such cached pages for I/O.
- *
- * Written by Steve Lord, Jim Mostek, Russell Cattelan
- * and Rajagopal Ananthanarayanan ("ananth") at SGI.
+ * This program is distributed in the hope that it would 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 the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
-
#include <linux/stddef.h>
#include <linux/errno.h>
#include <linux/slab.h>
@@ -55,25 +29,16 @@
#include <linux/blkdev.h>
#include <linux/hash.h>
#include <linux/kthread.h>
-
#include "xfs_linux.h"
-/*
- * File wide globals
- */
-
STATIC kmem_cache_t *pagebuf_zone;
STATIC kmem_shaker_t pagebuf_shake;
-STATIC int xfsbufd_wakeup(int, unsigned int);
+STATIC int xfsbufd_wakeup(int, gfp_t);
STATIC void pagebuf_delwri_queue(xfs_buf_t *, int);
STATIC struct workqueue_struct *xfslogd_workqueue;
struct workqueue_struct *xfsdatad_workqueue;
-/*
- * Pagebuf debugging
- */
-
#ifdef PAGEBUF_TRACE
void
pagebuf_trace(
@@ -112,10 +77,6 @@ ktrace_t *pagebuf_trace_buf;
# define PB_GET_OWNER(pb) do { } while (0)
#endif
-/*
- * Pagebuf allocation / freeing.
- */
-
#define pb_to_gfp(flags) \
((((flags) & PBF_READ_AHEAD) ? __GFP_NORETRY : \
((flags) & PBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
@@ -123,7 +84,6 @@ ktrace_t *pagebuf_trace_buf;
#define pb_to_km(flags) \
(((flags) & PBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
-
#define pagebuf_allocate(flags) \
kmem_zone_alloc(pagebuf_zone, pb_to_km(flags))
#define pagebuf_deallocate(pb) \
@@ -181,8 +141,9 @@ set_page_region(
size_t offset,
size_t length)
{
- page->private |= page_region_mask(offset, length);
- if (page->private == ~0UL)
+ set_page_private(page,
+ page_private(page) | page_region_mask(offset, length));
+ if (page_private(page) == ~0UL)
SetPageUptodate(page);
}
@@ -194,7 +155,7 @@ test_page_region(
{
unsigned long mask = page_region_mask(offset, length);
- return (mask && (page->private & mask) == mask);
+ return (mask && (page_private(page) & mask) == mask);
}
/*
@@ -285,7 +246,7 @@ _pagebuf_initialize(
* most cases but may be reset (e.g. XFS recovery).
*/
pb->pb_buffer_length = pb->pb_count_desired = range_length;
- pb->pb_flags = flags | PBF_NONE;
+ pb->pb_flags = flags;
pb->pb_bn = XFS_BUF_DADDR_NULL;
atomic_set(&pb->pb_pin_count, 0);
init_waitqueue_head(&pb->pb_waiters);
@@ -383,7 +344,7 @@ _pagebuf_lookup_pages(
size_t blocksize = bp->pb_target->pbr_bsize;
size_t size = bp->pb_count_desired;
size_t nbytes, offset;
- int gfp_mask = pb_to_gfp(flags);
+ gfp_t gfp_mask = pb_to_gfp(flags);
unsigned short page_count, i;
pgoff_t first;
loff_t end;
@@ -457,14 +418,8 @@ _pagebuf_lookup_pages(
unlock_page(bp->pb_pages[i]);
}
- if (page_count) {
- /* if we have any uptodate pages, mark that in the buffer */
- bp->pb_flags &= ~PBF_NONE;
-
- /* if some pages aren't uptodate, mark that in the buffer */
- if (page_count != bp->pb_page_count)
- bp->pb_flags |= PBF_PARTIAL;
- }
+ if (page_count == bp->pb_page_count)
+ bp->pb_flags |= PBF_DONE;
PB_TRACE(bp, "lookup_pages", (long)page_count);
return error;
@@ -675,7 +630,7 @@ xfs_buf_read_flags(
pb = xfs_buf_get_flags(target, ioff, isize, flags);
if (pb) {
- if (PBF_NOT_DONE(pb)) {
+ if (!XFS_BUF_ISDONE(pb)) {
PB_TRACE(pb, "read", (unsigned long)flags);
XFS_STATS_INC(pb_get_read);
pagebuf_iostart(pb, flags);
@@ -812,7 +767,7 @@ pagebuf_get_no_daddr(
bp = pagebuf_allocate(0);
if (unlikely(bp == NULL))
goto fail;
- _pagebuf_initialize(bp, target, 0, len, PBF_FORCEIO);
+ _pagebuf_initialize(bp, target, 0, len, 0);
try_again:
data = kmem_alloc(malloc_len, KM_SLEEP | KM_MAYFAIL);
@@ -875,39 +830,18 @@ pagebuf_rele(
PB_TRACE(pb, "rele", pb->pb_relse);
- /*
- * pagebuf_lookup buffers are not hashed, not delayed write,
- * and don't have their own release routines. Special case.
- */
- if (unlikely(!hash)) {
- ASSERT(!pb->pb_relse);
- if (atomic_dec_and_test(&pb->pb_hold))
- xfs_buf_free(pb);
- return;
- }
-
if (atomic_dec_and_lock(&pb->pb_hold, &hash->bh_lock)) {
- int do_free = 1;
-
if (pb->pb_relse) {
atomic_inc(&pb->pb_hold);
spin_unlock(&hash->bh_lock);
(*(pb->pb_relse)) (pb);
- spin_lock(&hash->bh_lock);
- do_free = 0;
- }
-
- if (pb->pb_flags & PBF_FS_MANAGED) {
- do_free = 0;
- }
-
- if (do_free) {
- ASSERT((pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)) == 0);
- list_del_init(&pb->pb_hash_list);
+ } else if (pb->pb_flags & PBF_FS_MANAGED) {
spin_unlock(&hash->bh_lock);
- pagebuf_free(pb);
} else {
+ ASSERT(!(pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)));
+ list_del_init(&pb->pb_hash_list);
spin_unlock(&hash->bh_lock);
+ pagebuf_free(pb);
}
} else {
/*
@@ -1120,21 +1054,18 @@ pagebuf_iodone_work(
void
pagebuf_iodone(
xfs_buf_t *pb,
- int dataio,
int schedule)
{
pb->pb_flags &= ~(PBF_READ | PBF_WRITE);
- if (pb->pb_error == 0) {
- pb->pb_flags &= ~(PBF_PARTIAL | PBF_NONE);
- }
+ if (pb->pb_error == 0)
+ pb->pb_flags |= PBF_DONE;
PB_TRACE(pb, "iodone", pb->pb_iodone);
if ((pb->pb_iodone) || (pb->pb_flags & PBF_ASYNC)) {
if (schedule) {
INIT_WORK(&pb->pb_iodone_work, pagebuf_iodone_work, pb);
- queue_work(dataio ? xfsdatad_workqueue :
- xfslogd_workqueue, &pb->pb_iodone_work);
+ queue_work(xfslogd_workqueue, &pb->pb_iodone_work);
} else {
pagebuf_iodone_work(pb);
}
@@ -1234,7 +1165,7 @@ _pagebuf_iodone(
{
if (atomic_dec_and_test(&pb->pb_io_remaining) == 1) {
pb->pb_locked = 0;
- pagebuf_iodone(pb, (pb->pb_flags & PBF_FS_DATAIOD), schedule);
+ pagebuf_iodone(pb, schedule);
}
}
@@ -1303,6 +1234,11 @@ _pagebuf_ioapply(
rw = (pb->pb_flags & PBF_READ) ? READ : WRITE;
}
+ if (pb->pb_flags & PBF_ORDERED) {
+ ASSERT(!(pb->pb_flags & PBF_READ));
+ rw = WRITE_BARRIER;
+ }
+
/* Special code path for reading a sub page size pagebuf in --
* we populate up the whole page, and hence the other metadata
* in the same page. This optimization is only valid when the
@@ -1750,7 +1686,7 @@ STATIC int xfsbufd_force_sleep;
STATIC int
xfsbufd_wakeup(
int priority,
- unsigned int mask)
+ gfp_t mask)
{
if (xfsbufd_force_sleep)
return 0;
@@ -1780,8 +1716,8 @@ xfsbufd(
xfsbufd_force_sleep = 0;
}
- schedule_timeout_interruptible
- (xfs_buf_timer_centisecs * msecs_to_jiffies(10));
+ schedule_timeout_interruptible(
+ xfs_buf_timer_centisecs * msecs_to_jiffies(10));
age = xfs_buf_age_centisecs * msecs_to_jiffies(10);
spin_lock(&pbd_delwrite_lock);
@@ -1890,14 +1826,22 @@ xfs_flush_buftarg(
return pincount;
}
-STATIC int
-xfs_buf_daemons_start(void)
+int __init
+pagebuf_init(void)
{
int error = -ENOMEM;
+#ifdef PAGEBUF_TRACE
+ pagebuf_trace_buf = ktrace_alloc(PAGEBUF_TRACE_SIZE, KM_SLEEP);
+#endif
+
+ pagebuf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
+ if (!pagebuf_zone)
+ goto out_free_trace_buf;
+
xfslogd_workqueue = create_workqueue("xfslogd");
if (!xfslogd_workqueue)
- goto out;
+ goto out_free_buf_zone;
xfsdatad_workqueue = create_workqueue("xfsdatad");
if (!xfsdatad_workqueue)
@@ -1908,82 +1852,37 @@ xfs_buf_daemons_start(void)
error = PTR_ERR(xfsbufd_task);
goto out_destroy_xfsdatad_workqueue;
}
+
+ pagebuf_shake = kmem_shake_register(xfsbufd_wakeup);
+ if (!pagebuf_shake)
+ goto out_stop_xfsbufd;
+
return 0;
+ out_stop_xfsbufd:
+ kthread_stop(xfsbufd_task);
out_destroy_xfsdatad_workqueue:
destroy_workqueue(xfsdatad_workqueue);
out_destroy_xfslogd_workqueue:
destroy_workqueue(xfslogd_workqueue);
- out:
- return error;
-}
-
-/*
- * Note: do not mark as __exit, it is called from pagebuf_terminate.
- */
-STATIC void
-xfs_buf_daemons_stop(void)
-{
- kthread_stop(xfsbufd_task);
- destroy_workqueue(xfslogd_workqueue);
- destroy_workqueue(xfsdatad_workqueue);
-}
-
-/*
- * Initialization and Termination
- */
-
-int __init
-pagebuf_init(void)
-{
- int error = -ENOMEM;
-
- pagebuf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
- if (!pagebuf_zone)
- goto out;
-
-#ifdef PAGEBUF_TRACE
- pagebuf_trace_buf = ktrace_alloc(PAGEBUF_TRACE_SIZE, KM_SLEEP);
-#endif
-
- error = xfs_buf_daemons_start();
- if (error)
- goto out_free_buf_zone;
-
- pagebuf_shake = kmem_shake_register(xfsbufd_wakeup);
- if (!pagebuf_shake) {
- error = -ENOMEM;
- goto out_stop_daemons;
- }
-
- return 0;
-
- out_stop_daemons:
- xfs_buf_daemons_stop();
out_free_buf_zone:
+ kmem_zone_destroy(pagebuf_zone);
+ out_free_trace_buf:
#ifdef PAGEBUF_TRACE
ktrace_free(pagebuf_trace_buf);
#endif
- kmem_zone_destroy(pagebuf_zone);
- out:
return error;
}
-
-/*
- * pagebuf_terminate.
- *
- * Note: do not mark as __exit, this is also called from the __init code.
- */
void
pagebuf_terminate(void)
{
- xfs_buf_daemons_stop();
-
+ kmem_shake_deregister(pagebuf_shake);
+ kthread_stop(xfsbufd_task);
+ destroy_workqueue(xfsdatad_workqueue);
+ destroy_workqueue(xfslogd_workqueue);
+ kmem_zone_destroy(pagebuf_zone);
#ifdef PAGEBUF_TRACE
ktrace_free(pagebuf_trace_buf);
#endif
-
- kmem_zone_destroy(pagebuf_zone);
- kmem_shake_deregister(pagebuf_shake);
}
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-29 14:33:31 +0100