summaryrefslogtreecommitdiffstats
path: root/fs/xfs/linux-2.6
diff options
context:
space:
mode:
Diffstat (limited to 'fs/xfs/linux-2.6')
-rw-r--r--fs/xfs/linux-2.6/mutex.h18
-rw-r--r--fs/xfs/linux-2.6/xfs_aops.c1124
-rw-r--r--fs/xfs/linux-2.6/xfs_aops.h10
-rw-r--r--fs/xfs/linux-2.6/xfs_buf.c1373
-rw-r--r--fs/xfs/linux-2.6/xfs_buf.h696
-rw-r--r--fs/xfs/linux-2.6/xfs_cred.h4
-rw-r--r--fs/xfs/linux-2.6/xfs_file.c6
-rw-r--r--fs/xfs/linux-2.6/xfs_fs_subr.c3
-rw-r--r--fs/xfs/linux-2.6/xfs_ioctl.c16
-rw-r--r--fs/xfs/linux-2.6/xfs_ioctl32.c1
-rw-r--r--fs/xfs/linux-2.6/xfs_iops.c137
-rw-r--r--fs/xfs/linux-2.6/xfs_iops.h5
-rw-r--r--fs/xfs/linux-2.6/xfs_linux.h6
-rw-r--r--fs/xfs/linux-2.6/xfs_lrw.c76
-rw-r--r--fs/xfs/linux-2.6/xfs_stats.c2
-rw-r--r--fs/xfs/linux-2.6/xfs_stats.h18
-rw-r--r--fs/xfs/linux-2.6/xfs_super.c19
-rw-r--r--fs/xfs/linux-2.6/xfs_vnode.c1
-rw-r--r--fs/xfs/linux-2.6/xfs_vnode.h19
19 files changed, 1759 insertions, 1775 deletions
diff --git a/fs/xfs/linux-2.6/mutex.h b/fs/xfs/linux-2.6/mutex.h
index ce773d89a92..2a88d56c4dc 100644
--- a/fs/xfs/linux-2.6/mutex.h
+++ b/fs/xfs/linux-2.6/mutex.h
@@ -18,22 +18,8 @@
#ifndef __XFS_SUPPORT_MUTEX_H__
#define __XFS_SUPPORT_MUTEX_H__
-#include <linux/spinlock.h>
-#include <asm/semaphore.h>
+#include <linux/mutex.h>
-/*
- * Map the mutex'es from IRIX to Linux semaphores.
- *
- * Destroy just simply initializes to -99 which should block all other
- * callers.
- */
-#define MUTEX_DEFAULT 0x0
-typedef struct semaphore mutex_t;
-
-#define mutex_init(lock, type, name) sema_init(lock, 1)
-#define mutex_destroy(lock) sema_init(lock, -99)
-#define mutex_lock(lock, num) down(lock)
-#define mutex_trylock(lock) (down_trylock(lock) ? 0 : 1)
-#define mutex_unlock(lock) up(lock)
+typedef struct mutex mutex_t;
#endif /* __XFS_SUPPORT_MUTEX_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index c6108971b4e..12062678940 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -40,11 +40,10 @@
#include "xfs_rw.h"
#include "xfs_iomap.h"
#include <linux/mpage.h>
+#include <linux/pagevec.h>
#include <linux/writeback.h>
STATIC void xfs_count_page_state(struct page *, int *, int *, int *);
-STATIC void xfs_convert_page(struct inode *, struct page *, xfs_iomap_t *,
- struct writeback_control *wbc, void *, int, int);
#if defined(XFS_RW_TRACE)
void
@@ -55,17 +54,15 @@ xfs_page_trace(
int mask)
{
xfs_inode_t *ip;
- bhv_desc_t *bdp;
vnode_t *vp = LINVFS_GET_VP(inode);
loff_t isize = i_size_read(inode);
- loff_t offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
+ loff_t offset = page_offset(page);
int delalloc = -1, unmapped = -1, unwritten = -1;
if (page_has_buffers(page))
xfs_count_page_state(page, &delalloc, &unmapped, &unwritten);
- bdp = vn_bhv_lookup(VN_BHV_HEAD(vp), &xfs_vnodeops);
- ip = XFS_BHVTOI(bdp);
+ ip = xfs_vtoi(vp);
if (!ip->i_rwtrace)
return;
@@ -103,15 +100,56 @@ xfs_finish_ioend(
queue_work(xfsdatad_workqueue, &ioend->io_work);
}
+/*
+ * We're now finished for good with this ioend structure.
+ * Update the page state via the associated buffer_heads,
+ * release holds on the inode and bio, and finally free
+ * up memory. Do not use the ioend after this.
+ */
STATIC void
xfs_destroy_ioend(
xfs_ioend_t *ioend)
{
+ struct buffer_head *bh, *next;
+
+ for (bh = ioend->io_buffer_head; bh; bh = next) {
+ next = bh->b_private;
+ bh->b_end_io(bh, ioend->io_uptodate);
+ }
+
vn_iowake(ioend->io_vnode);
mempool_free(ioend, xfs_ioend_pool);
}
/*
+ * Buffered IO write completion for delayed allocate extents.
+ * TODO: Update ondisk isize now that we know the file data
+ * has been flushed (i.e. the notorious "NULL file" problem).
+ */
+STATIC void
+xfs_end_bio_delalloc(
+ void *data)
+{
+ xfs_ioend_t *ioend = data;
+
+ xfs_destroy_ioend(ioend);
+}
+
+/*
+ * Buffered IO write completion for regular, written extents.
+ */
+STATIC void
+xfs_end_bio_written(
+ void *data)
+{
+ xfs_ioend_t *ioend = data;
+
+ xfs_destroy_ioend(ioend);
+}
+
+/*
+ * IO write completion for unwritten extents.
+ *
* Issue transactions to convert a buffer range from unwritten
* to written extents.
*/
@@ -123,21 +161,10 @@ xfs_end_bio_unwritten(
vnode_t *vp = ioend->io_vnode;
xfs_off_t offset = ioend->io_offset;
size_t size = ioend->io_size;
- struct buffer_head *bh, *next;
int error;
if (ioend->io_uptodate)
VOP_BMAP(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL, error);
-
- /* ioend->io_buffer_head is only non-NULL for buffered I/O */
- for (bh = ioend->io_buffer_head; bh; bh = next) {
- next = bh->b_private;
-
- bh->b_end_io = NULL;
- clear_buffer_unwritten(bh);
- end_buffer_async_write(bh, ioend->io_uptodate);
- }
-
xfs_destroy_ioend(ioend);
}
@@ -149,7 +176,8 @@ xfs_end_bio_unwritten(
*/
STATIC xfs_ioend_t *
xfs_alloc_ioend(
- struct inode *inode)
+ struct inode *inode,
+ unsigned int type)
{
xfs_ioend_t *ioend;
@@ -162,45 +190,25 @@ xfs_alloc_ioend(
*/
atomic_set(&ioend->io_remaining, 1);
ioend->io_uptodate = 1; /* cleared if any I/O fails */
+ ioend->io_list = NULL;
+ ioend->io_type = type;
ioend->io_vnode = LINVFS_GET_VP(inode);
ioend->io_buffer_head = NULL;
+ ioend->io_buffer_tail = NULL;
atomic_inc(&ioend->io_vnode->v_iocount);
ioend->io_offset = 0;
ioend->io_size = 0;
- INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
+ if (type == IOMAP_UNWRITTEN)
+ INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
+ else if (type == IOMAP_DELAY)
+ INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc, ioend);
+ else
+ INIT_WORK(&ioend->io_work, xfs_end_bio_written, ioend);
return ioend;
}
-void
-linvfs_unwritten_done(
- struct buffer_head *bh,
- int uptodate)
-{
- xfs_ioend_t *ioend = bh->b_private;
- static spinlock_t unwritten_done_lock = SPIN_LOCK_UNLOCKED;
- unsigned long flags;
-
- ASSERT(buffer_unwritten(bh));
- bh->b_end_io = NULL;
-
- if (!uptodate)
- ioend->io_uptodate = 0;
-
- /*
- * Deep magic here. We reuse b_private in the buffer_heads to build
- * a chain for completing the I/O from user context after we've issued
- * a transaction to convert the unwritten extent.
- */
- spin_lock_irqsave(&unwritten_done_lock, flags);
- bh->b_private = ioend->io_buffer_head;
- ioend->io_buffer_head = bh;
- spin_unlock_irqrestore(&unwritten_done_lock, flags);
-
- xfs_finish_ioend(ioend);
-}
-
STATIC int
xfs_map_blocks(
struct inode *inode,
@@ -218,138 +226,283 @@ xfs_map_blocks(
return -error;
}
+STATIC inline int
+xfs_iomap_valid(
+ xfs_iomap_t *iomapp,
+ loff_t offset)
+{
+ return offset >= iomapp->iomap_offset &&
+ offset < iomapp->iomap_offset + iomapp->iomap_bsize;
+}
+
/*
- * Finds the corresponding mapping in block @map array of the
- * given @offset within a @page.
+ * BIO completion handler for buffered IO.
*/
-STATIC xfs_iomap_t *
-xfs_offset_to_map(
+STATIC int
+xfs_end_bio(
+ struct bio *bio,
+ unsigned int bytes_done,
+ int error)
+{
+ xfs_ioend_t *ioend = bio->bi_private;
+
+ if (bio->bi_size)
+ return 1;
+
+ ASSERT(ioend);
+ ASSERT(atomic_read(&bio->bi_cnt) >= 1);
+
+ /* Toss bio and pass work off to an xfsdatad thread */
+ if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
+ ioend->io_uptodate = 0;
+ bio->bi_private = NULL;
+ bio->bi_end_io = NULL;
+
+ bio_put(bio);
+ xfs_finish_ioend(ioend);
+ return 0;
+}
+
+STATIC void
+xfs_submit_ioend_bio(
+ xfs_ioend_t *ioend,
+ struct bio *bio)
+{
+ atomic_inc(&ioend->io_remaining);
+
+ bio->bi_private = ioend;
+ bio->bi_end_io = xfs_end_bio;
+
+ submit_bio(WRITE, bio);
+ ASSERT(!bio_flagged(bio, BIO_EOPNOTSUPP));
+ bio_put(bio);
+}
+
+STATIC struct bio *
+xfs_alloc_ioend_bio(
+ struct buffer_head *bh)
+{
+ struct bio *bio;
+ int nvecs = bio_get_nr_vecs(bh->b_bdev);
+
+ do {
+ bio = bio_alloc(GFP_NOIO, nvecs);
+ nvecs >>= 1;
+ } while (!bio);
+
+ ASSERT(bio->bi_private == NULL);
+ bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
+ bio->bi_bdev = bh->b_bdev;
+ bio_get(bio);
+ return bio;
+}
+
+STATIC void
+xfs_start_buffer_writeback(
+ struct buffer_head *bh)
+{
+ ASSERT(buffer_mapped(bh));
+ ASSERT(buffer_locked(bh));
+ ASSERT(!buffer_delay(bh));
+ ASSERT(!buffer_unwritten(bh));
+
+ mark_buffer_async_write(bh);
+ set_buffer_uptodate(bh);
+ clear_buffer_dirty(bh);
+}
+
+STATIC void
+xfs_start_page_writeback(
struct page *page,
- xfs_iomap_t *iomapp,
- unsigned long offset)
+ struct writeback_control *wbc,
+ int clear_dirty,
+ int buffers)
+{
+ ASSERT(PageLocked(page));
+ ASSERT(!PageWriteback(page));
+ set_page_writeback(page);
+ if (clear_dirty)
+ clear_page_dirty(page);
+ unlock_page(page);
+ if (!buffers) {
+ end_page_writeback(page);
+ wbc->pages_skipped++; /* We didn't write this page */
+ }
+}
+
+static inline int bio_add_buffer(struct bio *bio, struct buffer_head *bh)
+{
+ return bio_add_page(bio, bh->b_page, bh->b_size, bh_offset(bh));
+}
+
+/*
+ * Submit all of the bios for all of the ioends we have saved up, covering the
+ * initial writepage page and also any probed pages.
+ *
+ * Because we may have multiple ioends spanning a page, we need to start
+ * writeback on all the buffers before we submit them for I/O. If we mark the
+ * buffers as we got, then we can end up with a page that only has buffers
+ * marked async write and I/O complete on can occur before we mark the other
+ * buffers async write.
+ *
+ * The end result of this is that we trip a bug in end_page_writeback() because
+ * we call it twice for the one page as the code in end_buffer_async_write()
+ * assumes that all buffers on the page are started at the same time.
+ *
+ * The fix is two passes across the ioend list - one to start writeback on the
+ * bufferheads, and then the second one submit them for I/O.
+ */
+STATIC void
+xfs_submit_ioend(
+ xfs_ioend_t *ioend)
{
- loff_t full_offset; /* offset from start of file */
+ xfs_ioend_t *head = ioend;
+ xfs_ioend_t *next;
+ struct buffer_head *bh;
+ struct bio *bio;
+ sector_t lastblock = 0;
+
+ /* Pass 1 - start writeback */
+ do {
+ next = ioend->io_list;
+ for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
+ xfs_start_buffer_writeback(bh);
+ }
+ } while ((ioend = next) != NULL);
+
+ /* Pass 2 - submit I/O */
+ ioend = head;
+ do {
+ next = ioend->io_list;
+ bio = NULL;
+
+ for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
+
+ if (!bio) {
+ retry:
+ bio = xfs_alloc_ioend_bio(bh);
+ } else if (bh->b_blocknr != lastblock + 1) {
+ xfs_submit_ioend_bio(ioend, bio);
+ goto retry;
+ }
+
+ if (bio_add_buffer(bio, bh) != bh->b_size) {
+ xfs_submit_ioend_bio(ioend, bio);
+ goto retry;
+ }
- ASSERT(offset < PAGE_CACHE_SIZE);
+ lastblock = bh->b_blocknr;
+ }
+ if (bio)
+ xfs_submit_ioend_bio(ioend, bio);
+ xfs_finish_ioend(ioend);
+ } while ((ioend = next) != NULL);
+}
+
+/*
+ * Cancel submission of all buffer_heads so far in this endio.
+ * Toss the endio too. Only ever called for the initial page
+ * in a writepage request, so only ever one page.
+ */
+STATIC void
+xfs_cancel_ioend(
+ xfs_ioend_t *ioend)
+{
+ xfs_ioend_t *next;
+ struct buffer_head *bh, *next_bh;
- full_offset = page->index; /* NB: using 64bit number */
- full_offset <<= PAGE_CACHE_SHIFT; /* offset from file start */
- full_offset += offset; /* offset from page start */
+ do {
+ next = ioend->io_list;
+ bh = ioend->io_buffer_head;
+ do {
+ next_bh = bh->b_private;
+ clear_buffer_async_write(bh);
+ unlock_buffer(bh);
+ } while ((bh = next_bh) != NULL);
- if (full_offset < iomapp->iomap_offset)
- return NULL;
- if (iomapp->iomap_offset + (iomapp->iomap_bsize -1) >= full_offset)
- return iomapp;
- return NULL;
+ vn_iowake(ioend->io_vnode);
+ mempool_free(ioend, xfs_ioend_pool);
+ } while ((ioend = next) != NULL);
+}
+
+/*
+ * Test to see if we've been building up a completion structure for
+ * earlier buffers -- if so, we try to append to this ioend if we
+ * can, otherwise we finish off any current ioend and start another.
+ * Return true if we've finished the given ioend.
+ */
+STATIC void
+xfs_add_to_ioend(
+ struct inode *inode,
+ struct buffer_head *bh,
+ xfs_off_t offset,
+ unsigned int type,
+ xfs_ioend_t **result,
+ int need_ioend)
+{
+ xfs_ioend_t *ioend = *result;
+
+ if (!ioend || need_ioend || type != ioend->io_type) {
+ xfs_ioend_t *previous = *result;
+
+ ioend = xfs_alloc_ioend(inode, type);
+ ioend->io_offset = offset;
+ ioend->io_buffer_head = bh;
+ ioend->io_buffer_tail = bh;
+ if (previous)
+ previous->io_list = ioend;
+ *result = ioend;
+ } else {
+ ioend->io_buffer_tail->b_private = bh;
+ ioend->io_buffer_tail = bh;
+ }
+
+ bh->b_private = NULL;
+ ioend->io_size += bh->b_size;
}
STATIC void
xfs_map_at_offset(
- struct page *page,
struct buffer_head *bh,
- unsigned long offset,
+ loff_t offset,
int block_bits,
xfs_iomap_t *iomapp)
{
xfs_daddr_t bn;
- loff_t delta;
int sector_shift;
ASSERT(!(iomapp->iomap_flags & IOMAP_HOLE));
ASSERT(!(iomapp->iomap_flags & IOMAP_DELAY));
ASSERT(iomapp->iomap_bn != IOMAP_DADDR_NULL);
- delta = page->index;
- delta <<= PAGE_CACHE_SHIFT;
- delta += offset;
- delta -= iomapp->iomap_offset;
- delta >>= block_bits;
-
sector_shift = block_bits - BBSHIFT;
- bn = iomapp->iomap_bn >> sector_shift;
- bn += delta;
- BUG_ON(!bn && !(iomapp->iomap_flags & IOMAP_REALTIME));
+ bn = (iomapp->iomap_bn >> sector_shift) +
+ ((offset - iomapp->iomap_offset) >> block_bits);
+
+ ASSERT(bn || (iomapp->iomap_flags & IOMAP_REALTIME));
ASSERT((bn << sector_shift) >= iomapp->iomap_bn);
lock_buffer(bh);
bh->b_blocknr = bn;
- bh->b_bdev = iomapp->iomap_target->pbr_bdev;
+ bh->b_bdev = iomapp->iomap_target->bt_bdev;
set_buffer_mapped(bh);
clear_buffer_delay(bh);
+ clear_buffer_unwritten(bh);
}
/*
- * Look for a page at index which is unlocked and contains our
- * unwritten extent flagged buffers at its head. Returns page
- * locked and with an extra reference count, and length of the
- * unwritten extent component on this page that we can write,
- * in units of filesystem blocks.
- */
-STATIC struct page *
-xfs_probe_unwritten_page(
- struct address_space *mapping,
- pgoff_t index,
- xfs_iomap_t *iomapp,
- xfs_ioend_t *ioend,
- unsigned long max_offset,
- unsigned long *fsbs,
- unsigned int bbits)
-{
- struct page *page;
-
- page = find_trylock_page(mapping, index);
- if (!page)
- return NULL;
- if (PageWriteback(page))
- goto out;
-
- if (page->mapping && page_has_buffers(page)) {
- struct buffer_head *bh, *head;
- unsigned long p_offset = 0;
-
- *fsbs = 0;
- bh = head = page_buffers(page);
- do {
- if (!buffer_unwritten(bh) || !buffer_uptodate(bh))
- break;
- if (!xfs_offset_to_map(page, iomapp, p_offset))
- break;
- if (p_offset >= max_offset)
- break;
- xfs_map_at_offset(page, bh, p_offset, bbits, iomapp);
- set_buffer_unwritten_io(bh);
- bh->b_private = ioend;
- p_offset += bh->b_size;
- (*fsbs)++;
- } while ((bh = bh->b_this_page) != head);
-
- if (p_offset)
- return page;
- }
-
-out:
- unlock_page(page);
- return NULL;
-}
-
-/*
- * Look for a page at index which is unlocked and not mapped
- * yet - clustering for mmap write case.
+ * Look for a page at index that is suitable for clustering.
*/
STATIC unsigned int
-xfs_probe_unmapped_page(
- struct address_space *mapping,
- pgoff_t index,
- unsigned int pg_offset)
+xfs_probe_page(
+ struct page *page,
+ unsigned int pg_offset,
+ int mapped)
{
- struct page *page;
int ret = 0;
- page = find_trylock_page(mapping, index);
- if (!page)
- return 0;
if (PageWriteback(page))
- goto out;
+ return 0;
if (page->mapping && PageDirty(page)) {
if (page_has_buffers(page)) {
@@ -357,79 +510,101 @@ xfs_probe_unmapped_page(
bh = head = page_buffers(page);
do {
- if (buffer_mapped(bh) || !buffer_uptodate(bh))
+ if (!buffer_uptodate(bh))
+ break;
+ if (mapped != buffer_mapped(bh))
break;
ret += bh->b_size;
if (ret >= pg_offset)
break;
} while ((bh = bh->b_this_page) != head);
} else
- ret = PAGE_CACHE_SIZE;
+ ret = mapped ? 0 : PAGE_CACHE_SIZE;
}
-out:
- unlock_page(page);
return ret;
}
-STATIC unsigned int
-xfs_probe_unmapped_cluster(
+STATIC size_t
+xfs_probe_cluster(
struct inode *inode,
struct page *startpage,
struct buffer_head *bh,
- struct buffer_head *head)
+ struct buffer_head *head,
+ int mapped)
{
+ struct pagevec pvec;
pgoff_t tindex, tlast, tloff;
- unsigned int pg_offset, len, total = 0;
- struct address_space *mapping = inode->i_mapping;
+ size_t total = 0;
+ int done = 0, i;
/* First sum forwards in this page */
do {
- if (buffer_mapped(bh))
- break;
+ if (mapped != buffer_mapped(bh))
+ return total;
total += bh->b_size;
} while ((bh = bh->b_this_page) != head);
- /* If we reached the end of the page, sum forwards in
- * following pages.
- */
- if (bh == head) {
- tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
- /* Prune this back to avoid pathological behavior */
- tloff = min(tlast, startpage->index + 64);
- for (tindex = startpage->index + 1; tindex < tloff; tindex++) {
- len = xfs_probe_unmapped_page(mapping, tindex,
- PAGE_CACHE_SIZE);
- if (!len)
- return total;
+ /* if we reached the end of the page, sum forwards in following pages */
+ tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
+ tindex = startpage->index + 1;
+
+ /* Prune this back to avoid pathological behavior */
+ tloff = min(tlast, startpage->index + 64);
+
+ pagevec_init(&pvec, 0);
+ while (!done && tindex <= tloff) {
+ unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
+
+ if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
+ break;
+
+ for (i = 0; i < pagevec_count(&pvec); i++) {
+ struct page *page = pvec.pages[i];
+ size_t pg_offset, len = 0;
+
+ if (tindex == tlast) {
+ pg_offset =
+ i_size_read(inode) & (PAGE_CACHE_SIZE - 1);
+ if (!pg_offset) {
+ done = 1;
+ break;
+ }
+ } else
+ pg_offset = PAGE_CACHE_SIZE;
+
+ if (page->index == tindex && !TestSetPageLocked(page)) {
+ len = xfs_probe_page(page, pg_offset, mapped);
+ unlock_page(page);
+ }
+
+ if (!len) {
+ done = 1;
+ break;
+ }
+
total += len;
+ tindex++;
}
- if (tindex == tlast &&
- (pg_offset = i_size_read(inode) & (PAGE_CACHE_SIZE - 1))) {
- total += xfs_probe_unmapped_page(mapping,
- tindex, pg_offset);
- }
+
+ pagevec_release(&pvec);
+ cond_resched();
}
+
return total;
}
/*
- * Probe for a given page (index) in the inode and test if it is delayed
- * and without unwritten buffers. Returns page locked and with an extra
- * reference count.
+ * Test if a given page is suitable for writing as part of an unwritten
+ * or delayed allocate extent.
*/
-STATIC struct page *
-xfs_probe_delalloc_page(
- struct inode *inode,
- pgoff_t index)
+STATIC int
+xfs_is_delayed_page(
+ struct page *page,
+ unsigned int type)
{
- struct page *page;
-
- page = find_trylock_page(inode->i_mapping, index);
- if (!page)
- return NULL;
if (PageWriteback(page))
- goto out;
+ return 0;
if (page->mapping && page_has_buffers(page)) {
struct buffer_head *bh, *head;
@@ -437,243 +612,156 @@ xfs_probe_delalloc_page(
bh = head = page_buffers(page);
do {
- if (buffer_unwritten(bh)) {
- acceptable = 0;
+ if (buffer_unwritten(bh))
+ acceptable = (type == IOMAP_UNWRITTEN);
+ else if (buffer_delay(bh))
+ acceptable = (type == IOMAP_DELAY);
+ else if (buffer_mapped(bh))
+ acceptable = (type == 0);
+ else
break;
- } else if (buffer_delay(bh)) {
- acceptable = 1;
- }
} while ((bh = bh->b_this_page) != head);
if (acceptable)
- return page;
+ return 1;
}
-out:
- unlock_page(page);
- return NULL;
-}
-
-STATIC int
-xfs_map_unwritten(
- struct inode *inode,
- struct page *start_page,
- struct buffer_head *head,
- struct buffer_head *curr,
- unsigned long p_offset,
- int block_bits,
- xfs_iomap_t *iomapp,
- struct writeback_control *wbc,
- int startio,
- int all_bh)
-{
- struct buffer_head *bh = curr;
- xfs_iomap_t *tmp;
- xfs_ioend_t *ioend;
- loff_t offset;
- unsigned long nblocks = 0;
-
- offset = start_page->index;
- offset <<= PAGE_CACHE_SHIFT;
- offset += p_offset;
-
- ioend = xfs_alloc_ioend(inode);
-
- /* First map forwards in the page consecutive buffers
- * covering this unwritten extent
- */
- do {
- if (!buffer_unwritten(bh))
- break;
- tmp = xfs_offset_to_map(start_page, iomapp, p_offset);
- if (!tmp)
- break;
- xfs_map_at_offset(start_page, bh, p_offset, block_bits, iomapp);
- set_buffer_unwritten_io(bh);
- bh->b_private = ioend;
- p_offset += bh->b_size;
- nblocks++;
- } while ((bh = bh->b_this_page) != head);
-
- atomic_add(nblocks, &ioend->io_remaining);
-
- /* If we reached the end of the page, map forwards in any
- * following pages which are also covered by this extent.
- */
- if (bh == head) {
- struct address_space *mapping = inode->i_mapping;
- pgoff_t tindex, tloff, tlast;
- unsigned long bs;
- unsigned int pg_offset, bbits = inode->i_blkbits;
- struct page *page;
-
- tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
- tloff = (iomapp->iomap_offset + iomapp->iomap_bsize) >> PAGE_CACHE_SHIFT;
- tloff = min(tlast, tloff);
- for (tindex = start_page->index + 1; tindex < tloff; tindex++) {
- page = xfs_probe_unwritten_page(mapping,
- tindex, iomapp, ioend,
- PAGE_CACHE_SIZE, &bs, bbits);
- if (!page)
- break;
- nblocks += bs;
- atomic_add(bs, &ioend->io_remaining);
- xfs_convert_page(inode, page, iomapp, wbc, ioend,
- startio, all_bh);
- /* stop if converting the next page might add
- * enough blocks that the corresponding byte
- * count won't fit in our ulong page buf length */
- if (nblocks >= ((ULONG_MAX - PAGE_SIZE) >> block_bits))
- goto enough;
- }
-
- if (tindex == tlast &&
- (pg_offset = (i_size_read(inode) & (PAGE_CACHE_SIZE - 1)))) {
- page = xfs_probe_unwritten_page(mapping,
- tindex, iomapp, ioend,
- pg_offset, &bs, bbits);
- if (page) {
- nblocks += bs;
- atomic_add(bs, &ioend->io_remaining);
- xfs_convert_page(inode, page, iomapp, wbc, ioend,
- startio, all_bh);
- if (nblocks >= ((ULONG_MAX - PAGE_SIZE) >> block_bits))
- goto enough;
- }
- }
- }
-
-enough:
- ioend->io_size = (xfs_off_t)nblocks << block_bits;
- ioend->io_offset = offset;
- xfs_finish_ioend(ioend);
return 0;
}
-STATIC void
-xfs_submit_page(
- struct page *page,
- struct writeback_control *wbc,
- struct buffer_head *bh_arr[],
- int bh_count,
- int probed_page,
- int clear_dirty)
-{
- struct buffer_head *bh;
- int i;
-
- BUG_ON(PageWriteback(page));
- if (bh_count)
- set_page_writeback(page);
- if (clear_dirty)
- clear_page_dirty(page);
- unlock_page(page);
-
- if (bh_count) {
- for (i = 0; i < bh_count; i++) {
- bh = bh_arr[i];
- mark_buffer_async_write(bh);
- if (buffer_unwritten(bh))
- set_buffer_unwritten_io(bh);
- set_buffer_uptodate(bh);
- clear_buffer_dirty(bh);
- }
-
- for (i = 0; i < bh_count; i++)
- submit_bh(WRITE, bh_arr[i]);
-
- if (probed_page && clear_dirty)
- wbc->nr_to_write--; /* Wrote an "extra" page */
- }
-}
-
/*
* Allocate & map buffers for page given the extent map. Write it out.
* except for the original page of a writepage, this is called on
* delalloc/unwritten pages only, for the original page it is possible
* that the page has no mapping at all.
*/
-STATIC void
+STATIC int
xfs_convert_page(
struct inode *inode,
struct page *page,
- xfs_iomap_t *iomapp,
+ loff_t tindex,
+ xfs_iomap_t *mp,
+ xfs_ioend_t **ioendp,
struct writeback_control *wbc,
- void *private,
int startio,
int all_bh)
{
- struct buffer_head *bh_arr[MAX_BUF_PER_PAGE], *bh, *head;
- xfs_iomap_t *mp = iomapp, *tmp;
- unsigned long offset, end_offset;
- int index = 0;
+ struct buffer_head *bh, *head;
+ xfs_off_t end_offset;
+ unsigned long p_offset;
+ unsigned int type;
int bbits = inode->i_blkbits;
int len, page_dirty;
+ int count = 0, done = 0, uptodate = 1;
+ xfs_off_t offset = page_offset(page);
- end_offset = (i_size_read(inode) & (PAGE_CACHE_SIZE - 1));
+ if (page->index != tindex)
+ goto fail;
+ if (TestSetPageLocked(page))
+ goto fail;
+ if (PageWriteback(page))
+ goto fail_unlock_page;
+ if (page->mapping != inode->i_mapping)
+ goto fail_unlock_page;
+ if (!xfs_is_delayed_page(page, (*ioendp)->io_type))
+ goto fail_unlock_page;
/*
* page_dirty is initially a count of buffers on the page before
* EOF and is decrememted as we move each into a cleanable state.
+ *
+ * Derivation:
+ *
+ * End offset is the highest offset that this page should represent.
+ * If we are on the last page, (end_offset & (PAGE_CACHE_SIZE - 1))
+ * will evaluate non-zero and be less than PAGE_CACHE_SIZE and
+ * hence give us the correct page_dirty count. On any other page,
+ * it will be zero and in that case we need page_dirty to be the
+ * count of buffers on the page.
*/
+ end_offset = min_t(unsigned long long,
+ (xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT,
+ i_size_read(inode));
+
len = 1 << inode->i_blkbits;
- end_offset = max(end_offset, PAGE_CACHE_SIZE);
- end_offset = roundup(end_offset, len);
- page_dirty = end_offset / len;
+ p_offset = min_t(unsigned long, end_offset & (PAGE_CACHE_SIZE - 1),
+ PAGE_CACHE_SIZE);
+ p_offset = p_offset ? roundup(p_offset, len) : PAGE_CACHE_SIZE;
+ page_dirty = p_offset / len;
- offset = 0;
bh = head = page_buffers(page);
do {
if (offset >= end_offset)
break;
- if (!(PageUptodate(page) || buffer_uptodate(bh)))
+ if (!buffer_uptodate(bh))
+ uptodate = 0;
+ if (!(PageUptodate(page) || buffer_uptodate(bh))) {
+ done = 1;
continue;
- if (buffer_mapped(bh) && all_bh &&
- !(buffer_unwritten(bh) || buffer_delay(bh))) {
+ }
+
+ if (buffer_unwritten(bh) || buffer_delay(bh)) {
+ if (buffer_unwritten(bh))
+ type = IOMAP_UNWRITTEN;
+ else
+ type = IOMAP_DELAY;
+
+ if (!xfs_iomap_valid(mp, offset)) {
+ done = 1;
+ continue;
+ }
+
+ ASSERT(!(mp->iomap_flags & IOMAP_HOLE));
+ ASSERT(!(mp->iomap_flags & IOMAP_DELAY));
+
+ xfs_map_at_offset(bh, offset, bbits, mp);
if (startio) {
+ xfs_add_to_ioend(inode, bh, offset,
+ type, ioendp, done);
+ } else {
+ set_buffer_dirty(bh);
+ unlock_buffer(bh);
+ mark_buffer_dirty(bh);
+ }
+ page_dirty--;
+ count++;
+ } else {
+ type = 0;
+ if (buffer_mapped(bh) && all_bh && startio) {
lock_buffer(bh);
- bh_arr[index++] = bh;
+ xfs_add_to_ioend(inode, bh, offset,
+ type, ioendp, done);
+ count++;
page_dirty--;
+ } else {
+ done = 1;
}
- continue;
}
- tmp = xfs_offset_to_map(page, mp, offset);
- if (!tmp)
- continue;
- ASSERT(!(tmp->iomap_flags & IOMAP_HOLE));
- ASSERT(!(tmp->iomap_flags & IOMAP_DELAY));
+ } while (offset += len, (bh = bh->b_this_page) != head);
- /* If this is a new unwritten extent buffer (i.e. one
- * that we haven't passed in private data for, we must
- * now map this buffer too.
- */
- if (buffer_unwritten(bh) && !bh->b_end_io) {
- ASSERT(tmp->iomap_flags & IOMAP_UNWRITTEN);
- xfs_map_unwritten(inode, page, head, bh, offset,
- bbits, tmp, wbc, startio, all_bh);
- } else if (! (buffer_unwritten(bh) && buffer_locked(bh))) {
- xfs_map_at_offset(page, bh, offset, bbits, tmp);
- if (buffer_unwritten(bh)) {
- set_buffer_unwritten_io(bh);
- bh->b_private = private;
- ASSERT(private);
+ if (uptodate && bh == head)
+ SetPageUptodate(page);
+
+ if (startio) {
+ if (count) {
+ struct backing_dev_info *bdi;
+
+ bdi = inode->i_mapping->backing_dev_info;
+ if (bdi_write_congested(bdi)) {
+ wbc->encountered_congestion = 1;
+ done = 1;
+ } else if (--wbc->nr_to_write <= 0) {
+ done = 1;
}
}
- if (startio) {
- bh_arr[index++] = bh;
- } else {
- set_buffer_dirty(bh);
- unlock_buffer(bh);
- mark_buffer_dirty(bh);
- }
- page_dirty--;
- } while (offset += len, (bh = bh->b_this_page) != head);
-
- if (startio && index) {
- xfs_submit_page(page, wbc, bh_arr, index, 1, !page_dirty);
- } else {
- unlock_page(page);
+ xfs_start_page_writeback(page, wbc, !page_dirty, count);
}
+
+ return done;
+ fail_unlock_page:
+ unlock_page(page);
+ fail:
+ return 1;
}
/*
@@ -685,19 +773,31 @@ xfs_cluster_write(
struct inode *inode,
pgoff_t tindex,
xfs_iomap_t *iomapp,
+ xfs_ioend_t **ioendp,
struct writeback_control *wbc,
int startio,
int all_bh,
pgoff_t tlast)
{
- struct page *page;
+ struct pagevec pvec;
+ int done = 0, i;
+
+ pagevec_init(&pvec, 0);
+ while (!done && tindex <= tlast) {
+ unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
- for (; tindex <= tlast; tindex++) {
- page = xfs_probe_delalloc_page(inode, tindex);
- if (!page)
+ if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
break;
- xfs_convert_page(inode, page, iomapp, wbc, NULL,
- startio, all_bh);
+
+ for (i = 0; i < pagevec_count(&pvec); i++) {
+ done = xfs_convert_page(inode, pvec.pages[i], tindex++,
+ iomapp, ioendp, wbc, startio, all_bh);
+ if (done)
+ break;
+ }
+
+ pagevec_release(&pvec);
+ cond_resched();
}
}
@@ -728,18 +828,22 @@ xfs_page_state_convert(
int startio,
int unmapped) /* also implies page uptodate */
{
- struct buffer_head *bh_arr[MAX_BUF_PER_PAGE], *bh, *head;
- xfs_iomap_t *iomp, iomap;
+ struct buffer_head *bh, *head;
+ xfs_iomap_t iomap;
+ xfs_ioend_t *ioend = NULL, *iohead = NULL;
loff_t offset;
unsigned long p_offset = 0;
+ unsigned int type;
__uint64_t end_offset;
pgoff_t end_index, last_index, tlast;
- int len, err, i, cnt = 0, uptodate = 1;
- int flags;
- int page_dirty;
+ ssize_t size, len;
+ int flags, err, iomap_valid = 0, uptodate = 1;
+ int page_dirty, count = 0, trylock_flag = 0;
+ int all_bh = unmapped;
/* wait for other IO threads? */
- flags = (startio && wbc->sync_mode != WB_SYNC_NONE) ? 0 : BMAPI_TRYLOCK;
+ if (startio && (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking))
+ trylock_flag |= BMAPI_TRYLOCK;
/* Is this page beyond the end of the file? */
offset = i_size_read(inode);
@@ -754,161 +858,173 @@ xfs_page_state_convert(
}
}
- end_offset = min_t(unsigned long long,
- (loff_t)(page->index + 1) << PAGE_CACHE_SHIFT, offset);
- offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
-
/*
* page_dirty is initially a count of buffers on the page before
* EOF and is decrememted as we move each into a cleanable state.
- */
+ *
+ * Derivation:
+ *
+ * End offset is the highest offset that this page should represent.
+ * If we are on the last page, (end_offset & (PAGE_CACHE_SIZE - 1))
+ * will evaluate non-zero and be less than PAGE_CACHE_SIZE and
+ * hence give us the correct page_dirty count. On any other page,
+ * it will be zero and in that case we need page_dirty to be the
+ * count of buffers on the page.
+ */
+ end_offset = min_t(unsigned long long,
+ (xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT, offset);
len = 1 << inode->i_blkbits;
- p_offset = max(p_offset, PAGE_CACHE_SIZE);
- p_offset = roundup(p_offset, len);
+ p_offset = min_t(unsigned long, end_offset & (PAGE_CACHE_SIZE - 1),
+ PAGE_CACHE_SIZE);
+ p_offset = p_offset ? roundup(p_offset, len) : PAGE_CACHE_SIZE;
page_dirty = p_offset / len;
- iomp = NULL;
- p_offset = 0;
bh = head = page_buffers(page);
+ offset = page_offset(page);
+ flags = -1;
+ type = 0;
+
+ /* TODO: cleanup count and page_dirty */
do {
if (offset >= end_offset)
break;
if (!buffer_uptodate(bh))
uptodate = 0;
- if (!(PageUptodate(page) || buffer_uptodate(bh)) && !startio)
+ if (!(PageUptodate(page) || buffer_uptodate(bh)) && !startio) {
+ /*
+ * the iomap is actually still valid, but the ioend
+ * isn't. shouldn't happen too often.
+ */
+ iomap_valid = 0;
continue;
-
- if (iomp) {
- iomp = xfs_offset_to_map(page, &iomap, p_offset);
}
+ if (iomap_valid)
+ iomap_valid = xfs_iomap_valid(&iomap, offset);
+
/*
* First case, map an unwritten extent and prepare for
* extent state conversion transaction on completion.
- */
- if (buffer_unwritten(bh)) {
- if (!startio)
- continue;
- if (!iomp) {
- err = xfs_map_blocks(inode, offset, len, &iomap,
- BMAPI_WRITE|BMAPI_IGNSTATE);
- if (err) {
- goto error;
- }
- iomp = xfs_offset_to_map(page, &iomap,
- p_offset);
+ *
+ * Second case, allocate space for a delalloc buffer.
+ * We can return EAGAIN here in the release page case.
+ *
+ * Third case, an unmapped buffer was found, and we are
+ * in a path where we need to write the whole page out.
+ */
+ if (buffer_unwritten(bh) || buffer_delay(bh) ||
+ ((buffer_uptodate(bh) || PageUptodate(page)) &&
+ !buffer_mapped(bh) && (unmapped || startio))) {
+ /*
+ * Make sure we don't use a read-only iomap
+ */
+ if (flags == BMAPI_READ)
+ iomap_valid = 0;
+
+ if (buffer_unwritten(bh)) {
+ type = IOMAP_UNWRITTEN;
+ flags = BMAPI_WRITE|BMAPI_IGNSTATE;
+ } else if (buffer_delay(bh)) {
+ type = IOMAP_DELAY;
+ flags = BMAPI_ALLOCATE;
+ if (!startio)
+ flags |= trylock_flag;
+ } else {
+ type = IOMAP_NEW;
+ flags = BMAPI_WRITE|BMAPI_MMAP;
}
- if (iomp) {
- if (!bh->b_end_io) {
- err = xfs_map_unwritten(inode, page,
- head, bh, p_offset,
- inode->i_blkbits, iomp,
- wbc, startio, unmapped);
- if (err) {
- goto error;
- }
+
+ if (!iomap_valid) {
+ if (type == IOMAP_NEW) {
+ size = xfs_probe_cluster(inode,
+ page, bh, head, 0);
} else {
- set_bit(BH_Lock, &bh->b_state);
+ size = len;
}
- BUG_ON(!buffer_locked(bh));
- bh_arr[cnt++] = bh;
- page_dirty--;
- }
- /*
- * Second case, allocate space for a delalloc buffer.
- * We can return EAGAIN here in the release page case.
- */
- } else if (buffer_delay(bh)) {
- if (!iomp) {
- err = xfs_map_blocks(inode, offset, len, &iomap,
- BMAPI_ALLOCATE | flags);
- if (err) {
+
+ err = xfs_map_blocks(inode, offset, size,
+ &iomap, flags);
+ if (err)
goto error;
- }
- iomp = xfs_offset_to_map(page, &iomap,
- p_offset);
+ iomap_valid = xfs_iomap_valid(&iomap, offset);
}
- if (iomp) {
- xfs_map_at_offset(page, bh, p_offset,
- inode->i_blkbits, iomp);
+ if (iomap_valid) {
+ xfs_map_at_offset(bh, offset,
+ inode->i_blkbits, &iomap);
if (startio) {
- bh_arr[cnt++] = bh;
+ xfs_add_to_ioend(inode, bh, offset,
+ type, &ioend,
+ !iomap_valid);
} else {
set_buffer_dirty(bh);
unlock_buffer(bh);
mark_buffer_dirty(bh);
}
page_dirty--;
+ count++;
+ }
+ } else if (buffer_uptodate(bh) && startio) {
+ /*
+ * we got here because the buffer is already mapped.
+ * That means it must already have extents allocated
+ * underneath it. Map the extent by reading it.
+ */
+ if (!iomap_valid || type != 0) {
+ flags = BMAPI_READ;
+ size = xfs_probe_cluster(inode, page, bh,
+ head, 1);
+ err = xfs_map_blocks(inode, offset, size,
+ &iomap, flags);
+ if (err)
+ goto error;
+ iomap_valid = xfs_iomap_valid(&iomap, offset);
}
- } else if ((buffer_uptodate(bh) || PageUptodate(page)) &&
- (unmapped || startio)) {
- if (!buffer_mapped(bh)) {
- int size;
-
- /*
- * Getting here implies an unmapped buffer
- * was found, and we are in a path where we
- * need to write the whole page out.
- */
- if (!iomp) {
- size = xfs_probe_unmapped_cluster(
- inode, page, bh, head);
- err = xfs_map_blocks(inode, offset,
- size, &iomap,
- BMAPI_WRITE|BMAPI_MMAP);
- if (err) {
- goto error;
- }
- iomp = xfs_offset_to_map(page, &iomap,
- p_offset);
- }
- if (iomp) {
- xfs_map_at_offset(page,
- bh, p_offset,
- inode->i_blkbits, iomp);
- if (startio) {
- bh_arr[cnt++] = bh;
- } else {
- set_buffer_dirty(bh);
- unlock_buffer(bh);
- mark_buffer_dirty(bh);
- }
- page_dirty--;
- }
- } else if (startio) {
- if (buffer_uptodate(bh) &&
- !test_and_set_bit(BH_Lock, &bh->b_state)) {
- bh_arr[cnt++] = bh;
- page_dirty--;
- }
+ type = 0;
+ if (!test_and_set_bit(BH_Lock, &bh->b_state)) {
+ ASSERT(buffer_mapped(bh));
+ if (iomap_valid)
+ all_bh = 1;
+ xfs_add_to_ioend(inode, bh, offset, type,
+ &ioend, !iomap_valid);
+ page_dirty--;
+ count++;
+ } else {
+ iomap_valid = 0;
}
+ } else if ((buffer_uptodate(bh) || PageUptodate(page)) &&
+ (unmapped || startio)) {
+ iomap_valid = 0;
}
- } while (offset += len, p_offset += len,
- ((bh = bh->b_this_page) != head));
+
+ if (!iohead)
+ iohead = ioend;
+
+ } while (offset += len, ((bh = bh->b_this_page) != head));
if (uptodate && bh == head)
SetPageUptodate(page);
- if (startio) {
- xfs_submit_page(page, wbc, bh_arr, cnt, 0, !page_dirty);
- }
+ if (startio)
+ xfs_start_page_writeback(page, wbc, 1, count);
- if (iomp) {
- offset = (iomp->iomap_offset + iomp->iomap_bsize - 1) >>
+ if (ioend && iomap_valid) {
+ offset = (iomap.iomap_offset + iomap.iomap_bsize - 1) >>
PAGE_CACHE_SHIFT;
tlast = min_t(pgoff_t, offset, last_index);
- xfs_cluster_write(inode, page->index + 1, iomp, wbc,
- startio, unmapped, tlast);
+ xfs_cluster_write(inode, page->index + 1, &iomap, &ioend,
+ wbc, startio, all_bh, tlast);
}
+ if (iohead)
+ xfs_submit_ioend(iohead);
+
return page_dirty;
error:
- for (i = 0; i < cnt; i++) {
- unlock_buffer(bh_arr[i]);
- }
+ if (iohead)
+ xfs_cancel_ioend(iohead);
/*
* If it's delalloc and we have nowhere to put it,
@@ -916,9 +1032,8 @@ error:
* us to try again.
*/
if (err != -EAGAIN) {
- if (!unmapped) {
+ if (!unmapped)
block_invalidatepage(page, 0);
- }
ClearPageUptodate(page);
}
return err;
@@ -941,13 +1056,12 @@ __linvfs_get_block(
int retpbbm = 1;
int error;
- if (blocks) {
- offset = blocks << inode->i_blkbits; /* 64 bit goodness */
- size = (ssize_t) min_t(xfs_off_t, offset, LONG_MAX);
- } else {
- size = 1 << inode->i_blkbits;
- }
offset = (xfs_off_t)iblock << inode->i_blkbits;
+ if (blocks)
+ size = (ssize_t) min_t(xfs_off_t, LONG_MAX,
+ (xfs_off_t)blocks << inode->i_blkbits);
+ else
+ size = 1 << inode->i_blkbits;
VOP_BMAP(vp, offset, size,
create ? flags : BMAPI_READ, &iomap, &retpbbm, error);
@@ -983,7 +1097,7 @@ __linvfs_get_block(
}
/* If this is a realtime file, data might be on a new device */
- bh_result->b_bdev = iomap.iomap_target->pbr_bdev;
+ bh_result->b_bdev = iomap.iomap_target->bt_bdev;
/* If we previously allocated a block out beyond eof and
* we are now coming back to use it then we will need to
@@ -1007,7 +1121,7 @@ __linvfs_get_block(
ASSERT(iomap.iomap_bsize - iomap.iomap_delta > 0);
offset = min_t(xfs_off_t,
iomap.iomap_bsize - iomap.iomap_delta,
- blocks << inode->i_blkbits);
+ (xfs_off_t)blocks << inode->i_blkbits);
bh_result->b_size = (u32) min_t(xfs_off_t, UINT_MAX, offset);
}
@@ -1095,10 +1209,10 @@ linvfs_direct_IO(
if (error)
return -error;
- iocb->private = xfs_alloc_ioend(inode);
+ iocb->private = xfs_alloc_ioend(inode, IOMAP_UNWRITTEN);
ret = blockdev_direct_IO_own_locking(rw, iocb, inode,
- iomap.iomap_target->pbr_bdev,
+ iomap.iomap_target->bt_bdev,
iov, offset, nr_segs,
linvfs_get_blocks_direct,
linvfs_end_io_direct);
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
index 4720758a9ad..55339dd5a30 100644
--- a/fs/xfs/linux-2.6/xfs_aops.h
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -23,14 +23,24 @@ extern mempool_t *xfs_ioend_pool;
typedef void (*xfs_ioend_func_t)(void *);
+/*
+ * xfs_ioend struct manages large extent writes for XFS.
+ * It can manage several multi-page bio's at once.
+ */
typedef struct xfs_ioend {
+ struct xfs_ioend *io_list; /* next ioend in chain */
+ unsigned int io_type; /* delalloc / unwritten */
unsigned int io_uptodate; /* I/O status register */
atomic_t io_remaining; /* hold count */
struct vnode *io_vnode; /* file being written to */
struct buffer_head *io_buffer_head;/* buffer linked list head */
+ struct buffer_head *io_buffer_tail;/* buffer linked list tail */
size_t io_size; /* size of the extent */
xfs_off_t io_offset; /* offset in the file */
struct work_struct io_work; /* xfsdatad work queue */
} xfs_ioend_t;
+extern struct address_space_operations linvfs_aops;
+extern int linvfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
+
#endif /* __XFS_IOPS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 6fe21d2b884..e44b7c1a3a3 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -31,76 +31,77 @@
#include <linux/kthread.h>
#include "xfs_linux.h"
-STATIC kmem_cache_t *pagebuf_zone;
-STATIC kmem_shaker_t pagebuf_shake;
+STATIC kmem_zone_t *xfs_buf_zone;
+STATIC kmem_shaker_t xfs_buf_shake;
+STATIC int xfsbufd(void *);
STATIC int xfsbufd_wakeup(int, gfp_t);
-STATIC void pagebuf_delwri_queue(xfs_buf_t *, int);
+STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
STATIC struct workqueue_struct *xfslogd_workqueue;
struct workqueue_struct *xfsdatad_workqueue;
-#ifdef PAGEBUF_TRACE
+#ifdef XFS_BUF_TRACE
void
-pagebuf_trace(
- xfs_buf_t *pb,
+xfs_buf_trace(
+ xfs_buf_t *bp,
char *id,
void *data,
void *ra)
{
- ktrace_enter(pagebuf_trace_buf,
- pb, id,
- (void *)(unsigned long)pb->pb_flags,
- (void *)(unsigned long)pb->pb_hold.counter,
- (void *)(unsigned long)pb->pb_sema.count.counter,
+ ktrace_enter(xfs_buf_trace_buf,
+ bp, id,
+ (void *)(unsigned long)bp->b_flags,
+ (void *)(unsigned long)bp->b_hold.counter,
+ (void *)(unsigned long)bp->b_sema.count.counter,
(void *)current,
data, ra,
- (void *)(unsigned long)((pb->pb_file_offset>>32) & 0xffffffff),
- (void *)(unsigned long)(pb->pb_file_offset & 0xffffffff),
- (void *)(unsigned long)pb->pb_buffer_length,
+ (void *)(unsigned long)((bp->b_file_offset>>32) & 0xffffffff),
+ (void *)(unsigned long)(bp->b_file_offset & 0xffffffff),
+ (void *)(unsigned long)bp->b_buffer_length,
NULL, NULL, NULL, NULL, NULL);
}
-ktrace_t *pagebuf_trace_buf;
-#define PAGEBUF_TRACE_SIZE 4096
-#define PB_TRACE(pb, id, data) \
- pagebuf_trace(pb, id, (void *)data, (void *)__builtin_return_address(0))
+ktrace_t *xfs_buf_trace_buf;
+#define XFS_BUF_TRACE_SIZE 4096
+#define XB_TRACE(bp, id, data) \
+ xfs_buf_trace(bp, id, (void *)data, (void *)__builtin_return_address(0))
#else
-#define PB_TRACE(pb, id, data) do { } while (0)
+#define XB_TRACE(bp, id, data) do { } while (0)
#endif
-#ifdef PAGEBUF_LOCK_TRACKING
-# define PB_SET_OWNER(pb) ((pb)->pb_last_holder = current->pid)
-# define PB_CLEAR_OWNER(pb) ((pb)->pb_last_holder = -1)
-# define PB_GET_OWNER(pb) ((pb)->pb_last_holder)
+#ifdef XFS_BUF_LOCK_TRACKING
+# define XB_SET_OWNER(bp) ((bp)->b_last_holder = current->pid)
+# define XB_CLEAR_OWNER(bp) ((bp)->b_last_holder = -1)
+# define XB_GET_OWNER(bp) ((bp)->b_last_holder)
#else
-# define PB_SET_OWNER(pb) do { } while (0)
-# define PB_CLEAR_OWNER(pb) do { } while (0)
-# define PB_GET_OWNER(pb) do { } while (0)
+# define XB_SET_OWNER(bp) do { } while (0)
+# define XB_CLEAR_OWNER(bp) do { } while (0)
+# define XB_GET_OWNER(bp) do { } while (0)
#endif
-#define pb_to_gfp(flags) \
- ((((flags) & PBF_READ_AHEAD) ? __GFP_NORETRY : \
- ((flags) & PBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
+#define xb_to_gfp(flags) \
+ ((((flags) & XBF_READ_AHEAD) ? __GFP_NORETRY : \
+ ((flags) & XBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
-#define pb_to_km(flags) \
- (((flags) & PBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
+#define xb_to_km(flags) \
+ (((flags) & XBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
-#define pagebuf_allocate(flags) \
- kmem_zone_alloc(pagebuf_zone, pb_to_km(flags))
-#define pagebuf_deallocate(pb) \
- kmem_zone_free(pagebuf_zone, (pb));
+#define xfs_buf_allocate(flags) \
+ kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags))
+#define xfs_buf_deallocate(bp) \
+ kmem_zone_free(xfs_buf_zone, (bp));
/*
- * Page Region interfaces.
+ * Page Region interfaces.
*
- * For pages in filesystems where the blocksize is smaller than the
- * pagesize, we use the page->private field (long) to hold a bitmap
- * of uptodate regions within the page.
+ * For pages in filesystems where the blocksize is smaller than the
+ * pagesize, we use the page->private field (long) to hold a bitmap
+ * of uptodate regions within the page.
*
- * Each such region is "bytes per page / bits per long" bytes long.
+ * Each such region is "bytes per page / bits per long" bytes long.
*
- * NBPPR == number-of-bytes-per-page-region
- * BTOPR == bytes-to-page-region (rounded up)
- * BTOPRT == bytes-to-page-region-truncated (rounded down)
+ * NBPPR == number-of-bytes-per-page-region
+ * BTOPR == bytes-to-page-region (rounded up)
+ * BTOPRT == bytes-to-page-region-truncated (rounded down)
*/
#if (BITS_PER_LONG == 32)
#define PRSHIFT (PAGE_CACHE_SHIFT - 5) /* (32 == 1<<5) */
@@ -159,7 +160,7 @@ test_page_region(
}
/*
- * Mapping of multi-page buffers into contiguous virtual space
+ * Mapping of multi-page buffers into contiguous virtual space
*/
typedef struct a_list {
@@ -172,7 +173,7 @@ STATIC int as_list_len;
STATIC DEFINE_SPINLOCK(as_lock);
/*
- * Try to batch vunmaps because they are costly.
+ * Try to batch vunmaps because they are costly.
*/
STATIC void
free_address(
@@ -215,83 +216,83 @@ purge_addresses(void)
}
/*
- * Internal pagebuf object manipulation
+ * Internal xfs_buf_t object manipulation
*/
STATIC void
-_pagebuf_initialize(
- xfs_buf_t *pb,
+_xfs_buf_initialize(
+ xfs_buf_t *bp,
xfs_buftarg_t *target,
- loff_t range_base,
+ xfs_off_t range_base,
size_t range_length,
- page_buf_flags_t flags)
+ xfs_buf_flags_t flags)
{
/*
- * We don't want certain flags to appear in pb->pb_flags.
+ * We don't want certain flags to appear in b_flags.
*/
- flags &= ~(PBF_LOCK|PBF_MAPPED|PBF_DONT_BLOCK|PBF_READ_AHEAD);
-
- memset(pb, 0, sizeof(xfs_buf_t));
- atomic_set(&pb->pb_hold, 1);
- init_MUTEX_LOCKED(&pb->pb_iodonesema);
- INIT_LIST_HEAD(&pb->pb_list);
- INIT_LIST_HEAD(&pb->pb_hash_list);
- init_MUTEX_LOCKED(&pb->pb_sema); /* held, no waiters */
- PB_SET_OWNER(pb);
- pb->pb_target = target;
- pb->pb_file_offset = range_base;
+ flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);
+
+ memset(bp, 0, sizeof(xfs_buf_t));
+ atomic_set(&bp->b_hold, 1);
+ init_MUTEX_LOCKED(&bp->b_iodonesema);
+ INIT_LIST_HEAD(&bp->b_list);
+ INIT_LIST_HEAD(&bp->b_hash_list);
+ init_MUTEX_LOCKED(&bp->b_sema); /* held, no waiters */
+ XB_SET_OWNER(bp);
+ bp->b_target = target;
+ bp->b_file_offset = range_base;
/*
* Set buffer_length and count_desired to the same value initially.
* I/O routines should use count_desired, which will be the same in
* most cases but may be reset (e.g. XFS recovery).
*/
- pb->pb_buffer_length = pb->pb_count_desired = range_length;
- pb->pb_flags = flags;
- pb->pb_bn = XFS_BUF_DADDR_NULL;
- atomic_set(&pb->pb_pin_count, 0);
- init_waitqueue_head(&pb->pb_waiters);
-
- XFS_STATS_INC(pb_create);
- PB_TRACE(pb, "initialize", target);
+ bp->b_buffer_length = bp->b_count_desired = range_length;
+ bp->b_flags = flags;
+ bp->b_bn = XFS_BUF_DADDR_NULL;
+ atomic_set(&bp->b_pin_count, 0);
+ init_waitqueue_head(&bp->b_waiters);
+
+ XFS_STATS_INC(xb_create);
+ XB_TRACE(bp, "initialize", target);
}
/*
- * Allocate a page array capable of holding a specified number
- * of pages, and point the page buf at it.
+ * Allocate a page array capable of holding a specified number
+ * of pages, and point the page buf at it.
*/
STATIC int
-_pagebuf_get_pages(
- xfs_buf_t *pb,
+_xfs_buf_get_pages(
+ xfs_buf_t *bp,
int page_count,
- page_buf_flags_t flags)
+ xfs_buf_flags_t flags)
{
/* Make sure that we have a page list */
- if (pb->pb_pages == NULL) {
- pb->pb_offset = page_buf_poff(pb->pb_file_offset);
- pb->pb_page_count = page_count;
- if (page_count <= PB_PAGES) {
- pb->pb_pages = pb->pb_page_array;
+ if (bp->b_pages == NULL) {
+ bp->b_offset = xfs_buf_poff(bp->b_file_offset);
+ bp->b_page_count = page_count;
+ if (page_count <= XB_PAGES) {
+ bp->b_pages = bp->b_page_array;
} else {
- pb->pb_pages = kmem_alloc(sizeof(struct page *) *
- page_count, pb_to_km(flags));
- if (pb->pb_pages == NULL)
+ bp->b_pages = kmem_alloc(sizeof(struct page *) *
+ page_count, xb_to_km(flags));
+ if (bp->b_pages == NULL)
return -ENOMEM;
}
- memset(pb->pb_pages, 0, sizeof(struct page *) * page_count);
+ memset(bp->b_pages, 0, sizeof(struct page *) * page_count);
}
return 0;
}
/*
- * Frees pb_pages if it was malloced.
+ * Frees b_pages if it was allocated.
*/
STATIC void
-_pagebuf_free_pages(
+_xfs_buf_free_pages(
xfs_buf_t *bp)
{
- if (bp->pb_pages != bp->pb_page_array) {
- kmem_free(bp->pb_pages,
- bp->pb_page_count * sizeof(struct page *));
+ if (bp->b_pages != bp->b_page_array) {
+ kmem_free(bp->b_pages,
+ bp->b_page_count * sizeof(struct page *));
}
}
@@ -299,79 +300,79 @@ _pagebuf_free_pages(
* Releases the specified buffer.
*
* The modification state of any associated pages is left unchanged.
- * The buffer most not be on any hash - use pagebuf_rele instead for
+ * The buffer most not be on any hash - use xfs_buf_rele instead for
* hashed and refcounted buffers
*/
void
-pagebuf_free(
+xfs_buf_free(
xfs_buf_t *bp)
{
- PB_TRACE(bp, "free", 0);
+ XB_TRACE(bp, "free", 0);
- ASSERT(list_empty(&bp->pb_hash_list));
+ ASSERT(list_empty(&bp->b_hash_list));
- if (bp->pb_flags & _PBF_PAGE_CACHE) {
+ if (bp->b_flags & _XBF_PAGE_CACHE) {
uint i;
- if ((bp->pb_flags & PBF_MAPPED) && (bp->pb_page_count > 1))
- free_address(bp->pb_addr - bp->pb_offset);
+ if ((bp->b_flags & XBF_MAPPED) && (bp->b_page_count > 1))
+ free_address(bp->b_addr - bp->b_offset);
- for (i = 0; i < bp->pb_page_count; i++)
- page_cache_release(bp->pb_pages[i]);
- _pagebuf_free_pages(bp);
- } else if (bp->pb_flags & _PBF_KMEM_ALLOC) {
+ for (i = 0; i < bp->b_page_count; i++)
+ page_cache_release(bp->b_pages[i]);
+ _xfs_buf_free_pages(bp);
+ } else if (bp->b_flags & _XBF_KMEM_ALLOC) {
/*
- * XXX(hch): bp->pb_count_desired might be incorrect (see
- * pagebuf_associate_memory for details), but fortunately
+ * XXX(hch): bp->b_count_desired might be incorrect (see
+ * xfs_buf_associate_memory for details), but fortunately
* the Linux version of kmem_free ignores the len argument..
*/
- kmem_free(bp->pb_addr, bp->pb_count_desired);
- _pagebuf_free_pages(bp);
+ kmem_free(bp->b_addr, bp->b_count_desired);
+ _xfs_buf_free_pages(bp);
}
- pagebuf_deallocate(bp);
+ xfs_buf_deallocate(bp);
}
/*
* Finds all pages for buffer in question and builds it's page list.
*/
STATIC int
-_pagebuf_lookup_pages(
+_xfs_buf_lookup_pages(
xfs_buf_t *bp,
uint flags)
{
- struct address_space *mapping = bp->pb_target->pbr_mapping;
- size_t blocksize = bp->pb_target->pbr_bsize;
- size_t size = bp->pb_count_desired;
+ struct address_space *mapping = bp->b_target->bt_mapping;
+ size_t blocksize = bp->b_target->bt_bsize;
+ size_t size = bp->b_count_desired;
size_t nbytes, offset;
- gfp_t gfp_mask = pb_to_gfp(flags);
+ gfp_t gfp_mask = xb_to_gfp(flags);
unsigned short page_count, i;
pgoff_t first;
- loff_t end;
+ xfs_off_t end;
int error;
- end = bp->pb_file_offset + bp->pb_buffer_length;
- page_count = page_buf_btoc(end) - page_buf_btoct(bp->pb_file_offset);
+ end = bp->b_file_offset + bp->b_buffer_length;
+ page_count = xfs_buf_btoc(end) - xfs_buf_btoct(bp->b_file_offset);
- error = _pagebuf_get_pages(bp, page_count, flags);
+ error = _xfs_buf_get_pages(bp, page_count, flags);
if (unlikely(error))
return error;
- bp->pb_flags |= _PBF_PAGE_CACHE;
+ bp->b_flags |= _XBF_PAGE_CACHE;
- offset = bp->pb_offset;
- first = bp->pb_file_offset >> PAGE_CACHE_SHIFT;
+ offset = bp->b_offset;
+ first = bp->b_file_offset >> PAGE_CACHE_SHIFT;
- for (i = 0; i < bp->pb_page_count; i++) {
+ for (i = 0; i < bp->b_page_count; i++) {
struct page *page;
uint retries = 0;
retry:
page = find_or_create_page(mapping, first + i, gfp_mask);
if (unlikely(page == NULL)) {
- if (flags & PBF_READ_AHEAD) {
- bp->pb_page_count = i;
- for (i = 0; i < bp->pb_page_count; i++)
- unlock_page(bp->pb_pages[i]);
+ if (flags & XBF_READ_AHEAD) {
+ bp->b_page_count = i;
+ for (i = 0; i < bp->b_page_count; i++)
+ unlock_page(bp->b_pages[i]);
return -ENOMEM;
}
@@ -387,13 +388,13 @@ _pagebuf_lookup_pages(
"deadlock in %s (mode:0x%x)\n",
__FUNCTION__, gfp_mask);
- XFS_STATS_INC(pb_page_retries);
+ XFS_STATS_INC(xb_page_retries);
xfsbufd_wakeup(0, gfp_mask);
blk_congestion_wait(WRITE, HZ/50);
goto retry;
}
- XFS_STATS_INC(pb_page_found);
+ XFS_STATS_INC(xb_page_found);
nbytes = min_t(size_t, size, PAGE_CACHE_SIZE - offset);
size -= nbytes;
@@ -401,27 +402,27 @@ _pagebuf_lookup_pages(
if (!PageUptodate(page)) {
page_count--;
if (blocksize >= PAGE_CACHE_SIZE) {
- if (flags & PBF_READ)
- bp->pb_locked = 1;
+ if (flags & XBF_READ)
+ bp->b_locked = 1;
} else if (!PagePrivate(page)) {
if (test_page_region(page, offset, nbytes))
page_count++;
}
}
- bp->pb_pages[i] = page;
+ bp->b_pages[i] = page;
offset = 0;
}
- if (!bp->pb_locked) {
- for (i = 0; i < bp->pb_page_count; i++)
- unlock_page(bp->pb_pages[i]);
+ if (!bp->b_locked) {
+ for (i = 0; i < bp->b_page_count; i++)
+ unlock_page(bp->b_pages[i]);
}
- if (page_count == bp->pb_page_count)
- bp->pb_flags |= PBF_DONE;
+ if (page_count == bp->b_page_count)
+ bp->b_flags |= XBF_DONE;
- PB_TRACE(bp, "lookup_pages", (long)page_count);
+ XB_TRACE(bp, "lookup_pages", (long)page_count);
return error;
}
@@ -429,23 +430,23 @@ _pagebuf_lookup_pages(
* Map buffer into kernel address-space if nessecary.
*/
STATIC int
-_pagebuf_map_pages(
+_xfs_buf_map_pages(
xfs_buf_t *bp,
uint flags)
{
/* A single page buffer is always mappable */
- if (bp->pb_page_count == 1) {
- bp->pb_addr = page_address(bp->pb_pages[0]) + bp->pb_offset;
- bp->pb_flags |= PBF_MAPPED;
- } else if (flags & PBF_MAPPED) {
+ if (bp->b_page_count == 1) {
+ bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset;
+ bp->b_flags |= XBF_MAPPED;
+ } else if (flags & XBF_MAPPED) {
if (as_list_len > 64)
purge_addresses();
- bp->pb_addr = vmap(bp->pb_pages, bp->pb_page_count,
- VM_MAP, PAGE_KERNEL);
- if (unlikely(bp->pb_addr == NULL))
+ bp->b_addr = vmap(bp->b_pages, bp->b_page_count,
+ VM_MAP, PAGE_KERNEL);
+ if (unlikely(bp->b_addr == NULL))
return -ENOMEM;
- bp->pb_addr += bp->pb_offset;
- bp->pb_flags |= PBF_MAPPED;
+ bp->b_addr += bp->b_offset;
+ bp->b_flags |= XBF_MAPPED;
}
return 0;
@@ -456,9 +457,7 @@ _pagebuf_map_pages(
*/
/*
- * _pagebuf_find
- *
- * Looks up, and creates if absent, a lockable buffer for
+ * Look up, and creates if absent, a lockable buffer for
* a given range of an inode. The buffer is returned
* locked. If other overlapping buffers exist, they are
* released before the new buffer is created and locked,
@@ -466,55 +465,55 @@ _pagebuf_map_pages(
* are unlocked. No I/O is implied by this call.
*/
xfs_buf_t *
-_pagebuf_find(
+_xfs_buf_find(
xfs_buftarg_t *btp, /* block device target */
- loff_t ioff, /* starting offset of range */
+ xfs_off_t ioff, /* starting offset of range */
size_t isize, /* length of range */
- page_buf_flags_t flags, /* PBF_TRYLOCK */
- xfs_buf_t *new_pb)/* newly allocated buffer */
+ xfs_buf_flags_t flags,
+ xfs_buf_t *new_bp)
{
- loff_t range_base;
+ xfs_off_t range_base;
size_t range_length;
xfs_bufhash_t *hash;
- xfs_buf_t *pb, *n;
+ xfs_buf_t *bp, *n;
range_base = (ioff << BBSHIFT);
range_length = (isize << BBSHIFT);
/* Check for IOs smaller than the sector size / not sector aligned */
- ASSERT(!(range_length < (1 << btp->pbr_sshift)));
- ASSERT(!(range_base & (loff_t)btp->pbr_smask));
+ ASSERT(!(range_length < (1 << btp->bt_sshift)));
+ ASSERT(!(range_base & (xfs_off_t)btp->bt_smask));
hash = &btp->bt_hash[hash_long((unsigned long)ioff, btp->bt_hashshift)];
spin_lock(&hash->bh_lock);
- list_for_each_entry_safe(pb, n, &hash->bh_list, pb_hash_list) {
- ASSERT(btp == pb->pb_target);
- if (pb->pb_file_offset == range_base &&
- pb->pb_buffer_length == range_length) {
+ list_for_each_entry_safe(bp, n, &hash->bh_list, b_hash_list) {
+ ASSERT(btp == bp->b_target);
+ if (bp->b_file_offset == range_base &&
+ bp->b_buffer_length == range_length) {
/*
- * If we look at something bring it to the
+ * If we look at something, bring it to the
* front of the list for next time.
*/
- atomic_inc(&pb->pb_hold);
- list_move(&pb->pb_hash_list, &hash->bh_list);
+ atomic_inc(&bp->b_hold);
+ list_move(&bp->b_hash_list, &hash->bh_list);
goto found;
}
}
/* No match found */
- if (new_pb) {
- _pagebuf_initialize(new_pb, btp, range_base,
+ if (new_bp) {
+ _xfs_buf_initialize(new_bp, btp, range_base,
range_length, flags);
- new_pb->pb_hash = hash;
- list_add(&new_pb->pb_hash_list, &hash->bh_list);
+ new_bp->b_hash = hash;
+ list_add(&new_bp->b_hash_list, &hash->bh_list);
} else {
- XFS_STATS_INC(pb_miss_locked);
+ XFS_STATS_INC(xb_miss_locked);
}
spin_unlock(&hash->bh_lock);
- return new_pb;
+ return new_bp;
found:
spin_unlock(&hash->bh_lock);
@@ -523,74 +522,72 @@ found:
* if this does not work then we need to drop the
* spinlock and do a hard attempt on the semaphore.
*/
- if (down_trylock(&pb->pb_sema)) {
- if (!(flags & PBF_TRYLOCK)) {
+ if (down_trylock(&bp->b_sema)) {
+ if (!(flags & XBF_TRYLOCK)) {
/* wait for buffer ownership */
- PB_TRACE(pb, "get_lock", 0);
- pagebuf_lock(pb);
- XFS_STATS_INC(pb_get_locked_waited);
+ XB_TRACE(bp, "get_lock", 0);
+ xfs_buf_lock(bp);
+ XFS_STATS_INC(xb_get_locked_waited);
} else {
/* We asked for a trylock and failed, no need
* to look at file offset and length here, we
- * know that this pagebuf at least overlaps our
- * pagebuf and is locked, therefore our buffer
- * either does not exist, or is this buffer
+ * know that this buffer at least overlaps our
+ * buffer and is locked, therefore our buffer
+ * either does not exist, or is this buffer.
*/
-
- pagebuf_rele(pb);
- XFS_STATS_INC(pb_busy_locked);
- return (NULL);
+ xfs_buf_rele(bp);
+ XFS_STATS_INC(xb_busy_locked);
+ return NULL;
}
} else {
/* trylock worked */
- PB_SET_OWNER(pb);
+ XB_SET_OWNER(bp);
}
- if (pb->pb_flags & PBF_STALE) {
- ASSERT((pb->pb_flags & _PBF_DELWRI_Q) == 0);
- pb->pb_flags &= PBF_MAPPED;
+ if (bp->b_flags & XBF_STALE) {
+ ASSERT((bp->b_flags & _XBF_DELWRI_Q) == 0);
+ bp->b_flags &= XBF_MAPPED;
}
- PB_TRACE(pb, "got_lock", 0);
- XFS_STATS_INC(pb_get_locked);
- return (pb);
+ XB_TRACE(bp, "got_lock", 0);
+ XFS_STATS_INC(xb_get_locked);
+ return bp;
}
/*
- * xfs_buf_get_flags assembles a buffer covering the specified range.
- *
+ * Assembles a buffer covering the specified range.
* Storage in memory for all portions of the buffer will be allocated,
* although backing storage may not be.
*/
xfs_buf_t *
-xfs_buf_get_flags( /* allocate a buffer */
+xfs_buf_get_flags(
xfs_buftarg_t *target,/* target for buffer */
- loff_t ioff, /* starting offset of range */
+ xfs_off_t ioff, /* starting offset of range */
size_t isize, /* length of range */
- page_buf_flags_t flags) /* PBF_TRYLOCK */
+ xfs_buf_flags_t flags)
{
- xfs_buf_t *pb, *new_pb;
+ xfs_buf_t *bp, *new_bp;
int error = 0, i;
- new_pb = pagebuf_allocate(flags);
- if (unlikely(!new_pb))
+ new_bp = xfs_buf_allocate(flags);
+ if (unlikely(!new_bp))
return NULL;
- pb = _pagebuf_find(target, ioff, isize, flags, new_pb);
- if (pb == new_pb) {
- error = _pagebuf_lookup_pages(pb, flags);
+ bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
+ if (bp == new_bp) {
+ error = _xfs_buf_lookup_pages(bp, flags);
if (error)
goto no_buffer;
} else {
- pagebuf_deallocate(new_pb);
- if (unlikely(pb == NULL))
+ xfs_buf_deallocate(new_bp);
+ if (unlikely(bp == NULL))
return NULL;
}
- for (i = 0; i < pb->pb_page_count; i++)
- mark_page_accessed(pb->pb_pages[i]);
+ for (i = 0; i < bp->b_page_count; i++)
+ mark_page_accessed(bp->b_pages[i]);
- if (!(pb->pb_flags & PBF_MAPPED)) {
- error = _pagebuf_map_pages(pb, flags);
+ if (!(bp->b_flags & XBF_MAPPED)) {
+ error = _xfs_buf_map_pages(bp, flags);
if (unlikely(error)) {
printk(KERN_WARNING "%s: failed to map pages\n",
__FUNCTION__);
@@ -598,97 +595,97 @@ xfs_buf_get_flags( /* allocate a buffer */
}
}
- XFS_STATS_INC(pb_get);
+ XFS_STATS_INC(xb_get);
/*
* Always fill in the block number now, the mapped cases can do
* their own overlay of this later.
*/
- pb->pb_bn = ioff;
- pb->pb_count_desired = pb->pb_buffer_length;
+ bp->b_bn = ioff;
+ bp->b_count_desired = bp->b_buffer_length;
- PB_TRACE(pb, "get", (unsigned long)flags);
- return pb;
+ XB_TRACE(bp, "get", (unsigned long)flags);
+ return bp;
no_buffer:
- if (flags & (PBF_LOCK | PBF_TRYLOCK))
- pagebuf_unlock(pb);
- pagebuf_rele(pb);
+ if (flags & (XBF_LOCK | XBF_TRYLOCK))
+ xfs_buf_unlock(bp);
+ xfs_buf_rele(bp);
return NULL;
}
xfs_buf_t *
xfs_buf_read_flags(
xfs_buftarg_t *target,
- loff_t ioff,
+ xfs_off_t ioff,
size_t isize,
- page_buf_flags_t flags)
+ xfs_buf_flags_t flags)
{
- xfs_buf_t *pb;
-
- flags |= PBF_READ;
-
- pb = xfs_buf_get_flags(target, ioff, isize, flags);
- if (pb) {
- if (!XFS_BUF_ISDONE(pb)) {
- PB_TRACE(pb, "read", (unsigned long)flags);
- XFS_STATS_INC(pb_get_read);
- pagebuf_iostart(pb, flags);
- } else if (flags & PBF_ASYNC) {
- PB_TRACE(pb, "read_async", (unsigned long)flags);
+ xfs_buf_t *bp;
+
+ flags |= XBF_READ;
+
+ bp = xfs_buf_get_flags(target, ioff, isize, flags);
+ if (bp) {
+ if (!XFS_BUF_ISDONE(bp)) {
+ XB_TRACE(bp, "read", (unsigned long)flags);
+ XFS_STATS_INC(xb_get_read);
+ xfs_buf_iostart(bp, flags);
+ } else if (flags & XBF_ASYNC) {
+ XB_TRACE(bp, "read_async", (unsigned long)flags);
/*
* Read ahead call which is already satisfied,
* drop the buffer
*/
goto no_buffer;
} else {
- PB_TRACE(pb, "read_done", (unsigned long)flags);
+ XB_TRACE(bp, "read_done", (unsigned long)flags);
/* We do not want read in the flags */
- pb->pb_flags &= ~PBF_READ;
+ bp->b_flags &= ~XBF_READ;
}
}
- return pb;
+ return bp;
no_buffer:
- if (flags & (PBF_LOCK | PBF_TRYLOCK))
- pagebuf_unlock(pb);
- pagebuf_rele(pb);
+ if (flags & (XBF_LOCK | XBF_TRYLOCK))
+ xfs_buf_unlock(bp);
+ xfs_buf_rele(bp);
return NULL;
}
/*
- * If we are not low on memory then do the readahead in a deadlock
- * safe manner.
+ * If we are not low on memory then do the readahead in a deadlock
+ * safe manner.
*/
void
-pagebuf_readahead(
+xfs_buf_readahead(
xfs_buftarg_t *target,
- loff_t ioff,
+ xfs_off_t ioff,
size_t isize,
- page_buf_flags_t flags)
+ xfs_buf_flags_t flags)
{
struct backing_dev_info *bdi;
- bdi = target->pbr_mapping->backing_dev_info;
+ bdi = target->bt_mapping->backing_dev_info;
if (bdi_read_congested(bdi))
return;
- flags |= (PBF_TRYLOCK|PBF_ASYNC|PBF_READ_AHEAD);
+ flags |= (XBF_TRYLOCK|XBF_ASYNC|XBF_READ_AHEAD);
xfs_buf_read_flags(target, ioff, isize, flags);
}
xfs_buf_t *
-pagebuf_get_empty(
+xfs_buf_get_empty(
size_t len,
xfs_buftarg_t *target)
{
- xfs_buf_t *pb;
+ xfs_buf_t *bp;
- pb = pagebuf_allocate(0);
- if (pb)
- _pagebuf_initialize(pb, target, 0, len, 0);
- return pb;
+ bp = xfs_buf_allocate(0);
+ if (bp)
+ _xfs_buf_initialize(bp, target, 0, len, 0);
+ return bp;
}
static inline struct page *
@@ -704,8 +701,8 @@ mem_to_page(
}
int
-pagebuf_associate_memory(
- xfs_buf_t *pb,
+xfs_buf_associate_memory(
+ xfs_buf_t *bp,
void *mem,
size_t len)
{
@@ -722,40 +719,40 @@ pagebuf_associate_memory(
page_count++;
/* Free any previous set of page pointers */
- if (pb->pb_pages)
- _pagebuf_free_pages(pb);
+ if (bp->b_pages)
+ _xfs_buf_free_pages(bp);
- pb->pb_pages = NULL;
- pb->pb_addr = mem;
+ bp->b_pages = NULL;
+ bp->b_addr = mem;
- rval = _pagebuf_get_pages(pb, page_count, 0);
+ rval = _xfs_buf_get_pages(bp, page_count, 0);
if (rval)
return rval;
- pb->pb_offset = offset;
+ bp->b_offset = offset;
ptr = (size_t) mem & PAGE_CACHE_MASK;
end = PAGE_CACHE_ALIGN((size_t) mem + len);
end_cur = end;
/* set up first page */
- pb->pb_pages[0] = mem_to_page(mem);
+ bp->b_pages[0] = mem_to_page(mem);
ptr += PAGE_CACHE_SIZE;
- pb->pb_page_count = ++i;
+ bp->b_page_count = ++i;
while (ptr < end) {
- pb->pb_pages[i] = mem_to_page((void *)ptr);
- pb->pb_page_count = ++i;
+ bp->b_pages[i] = mem_to_page((void *)ptr);
+ bp->b_page_count = ++i;
ptr += PAGE_CACHE_SIZE;
}
- pb->pb_locked = 0;
+ bp->b_locked = 0;
- pb->pb_count_desired = pb->pb_buffer_length = len;
- pb->pb_flags |= PBF_MAPPED;
+ bp->b_count_desired = bp->b_buffer_length = len;
+ bp->b_flags |= XBF_MAPPED;
return 0;
}
xfs_buf_t *
-pagebuf_get_no_daddr(
+xfs_buf_get_noaddr(
size_t len,
xfs_buftarg_t *target)
{
@@ -764,10 +761,10 @@ pagebuf_get_no_daddr(
void *data;
int error;
- bp = pagebuf_allocate(0);
+ bp = xfs_buf_allocate(0);
if (unlikely(bp == NULL))
goto fail;
- _pagebuf_initialize(bp, target, 0, len, 0);
+ _xfs_buf_initialize(bp, target, 0, len, 0);
try_again:
data = kmem_alloc(malloc_len, KM_SLEEP | KM_MAYFAIL);
@@ -776,78 +773,73 @@ pagebuf_get_no_daddr(
/* check whether alignment matches.. */
if ((__psunsigned_t)data !=
- ((__psunsigned_t)data & ~target->pbr_smask)) {
+ ((__psunsigned_t)data & ~target->bt_smask)) {
/* .. else double the size and try again */
kmem_free(data, malloc_len);
malloc_len <<= 1;
goto try_again;
}
- error = pagebuf_associate_memory(bp, data, len);
+ error = xfs_buf_associate_memory(bp, data, len);
if (error)
goto fail_free_mem;
- bp->pb_flags |= _PBF_KMEM_ALLOC;
+ bp->b_flags |= _XBF_KMEM_ALLOC;
- pagebuf_unlock(bp);
+ xfs_buf_unlock(bp);
- PB_TRACE(bp, "no_daddr", data);
+ XB_TRACE(bp, "no_daddr", data);
return bp;
fail_free_mem:
kmem_free(data, malloc_len);
fail_free_buf:
- pagebuf_free(bp);
+ xfs_buf_free(bp);
fail:
return NULL;
}
/*
- * pagebuf_hold
- *
* Increment reference count on buffer, to hold the buffer concurrently
* with another thread which may release (free) the buffer asynchronously.
- *
* Must hold the buffer already to call this function.
*/
void
-pagebuf_hold(
- xfs_buf_t *pb)
+xfs_buf_hold(
+ xfs_buf_t *bp)
{
- atomic_inc(&pb->pb_hold);
- PB_TRACE(pb, "hold", 0);
+ atomic_inc(&bp->b_hold);
+ XB_TRACE(bp, "hold", 0);
}
/*
- * pagebuf_rele
- *
- * pagebuf_rele releases a hold on the specified buffer. If the
- * the hold count is 1, pagebuf_rele calls pagebuf_free.
+ * Releases a hold on the specified buffer. If the
+ * the hold count is 1, calls xfs_buf_free.
*/
void
-pagebuf_rele(
- xfs_buf_t *pb)
+xfs_buf_rele(
+ xfs_buf_t *bp)
{
- xfs_bufhash_t *hash = pb->pb_hash;
+ xfs_bufhash_t *hash = bp->b_hash;
- PB_TRACE(pb, "rele", pb->pb_relse);
+ XB_TRACE(bp, "rele", bp->b_relse);
- if (atomic_dec_and_lock(&pb->pb_hold, &hash->bh_lock)) {
- if (pb->pb_relse) {
- atomic_inc(&pb->pb_hold);
+ if (atomic_dec_and_lock(&bp->b_hold, &hash->bh_lock)) {
+ if (bp->b_relse) {
+ atomic_inc(&bp->b_hold);
spin_unlock(&hash->bh_lock);
- (*(pb->pb_relse)) (pb);
- } else if (pb->pb_flags & PBF_FS_MANAGED) {
+ (*(bp->b_relse)) (bp);
+ } else if (bp->b_flags & XBF_FS_MANAGED) {
spin_unlock(&hash->bh_lock);
} else {
- ASSERT(!(pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)));
- list_del_init(&pb->pb_hash_list);
+ ASSERT(!(bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)));
+ list_del_init(&bp->b_hash_list);
spin_unlock(&hash->bh_lock);
- pagebuf_free(pb);
+ xfs_buf_free(bp);
}
} else {
/*
* Catch reference count leaks
*/
- ASSERT(atomic_read(&pb->pb_hold) >= 0);
+ ASSERT(atomic_read(&bp->b_hold) >= 0);
}
}
@@ -863,168 +855,122 @@ pagebuf_rele(
*/
/*
- * pagebuf_cond_lock
- *
- * pagebuf_cond_lock locks a buffer object, if it is not already locked.
- * Note that this in no way
- * locks the underlying pages, so it is only useful for synchronizing
- * concurrent use of page buffer objects, not for synchronizing independent
- * access to the underlying pages.
+ * Locks a buffer object, if it is not already locked.
+ * Note that this in no way locks the underlying pages, so it is only
+ * useful for synchronizing concurrent use of buffer objects, not for
+ * synchronizing independent access to the underlying pages.
*/
int
-pagebuf_cond_lock( /* lock buffer, if not locked */
- /* returns -EBUSY if locked) */
- xfs_buf_t *pb)
+xfs_buf_cond_lock(
+ xfs_buf_t *bp)
{
int locked;
- locked = down_trylock(&pb->pb_sema) == 0;
+ locked = down_trylock(&bp->b_sema) == 0;
if (locked) {
- PB_SET_OWNER(pb);
+ XB_SET_OWNER(bp);
}
- PB_TRACE(pb, "cond_lock", (long)locked);
- return(locked ? 0 : -EBUSY);
+ XB_TRACE(bp, "cond_lock", (long)locked);
+ return locked ? 0 : -EBUSY;
}
#if defined(DEBUG) || defined(XFS_BLI_TRACE)
-/*
- * pagebuf_lock_value
- *
- * Return lock value for a pagebuf
- */
int
-pagebuf_lock_value(
- xfs_buf_t *pb)
+xfs_buf_lock_value(
+ xfs_buf_t *bp)
{
- return(atomic_read(&pb->pb_sema.count));
+ return atomic_read(&bp->b_sema.count);
}
#endif
/*
- * pagebuf_lock
- *
- * pagebuf_lock locks a buffer object. Note that this in no way
- * locks the underlying pages, so it is only useful for synchronizing
- * concurrent use of page buffer objects, not for synchronizing independent
- * access to the underlying pages.
+ * Locks a buffer object.
+ * Note that this in no way locks the underlying pages, so it is only
+ * useful for synchronizing concurrent use of buffer objects, not for
+ * synchronizing independent access to the underlying pages.
*/
-int
-pagebuf_lock(
- xfs_buf_t *pb)
+void
+xfs_buf_lock(
+ xfs_buf_t *bp)
{
- PB_TRACE(pb, "lock", 0);
- if (atomic_read(&pb->pb_io_remaining))
- blk_run_address_space(pb->pb_target->pbr_mapping);
- down(&pb->pb_sema);
- PB_SET_OWNER(pb);
- PB_TRACE(pb, "locked", 0);
- return 0;
+ XB_TRACE(bp, "lock", 0);
+ if (atomic_read(&bp->b_io_remaining))
+ blk_run_address_space(bp->b_target->bt_mapping);
+ down(&bp->b_sema);
+ XB_SET_OWNER(bp);
+ XB_TRACE(bp, "locked", 0);
}
/*
- * pagebuf_unlock
- *
- * pagebuf_unlock releases the lock on the buffer object created by
- * pagebuf_lock or pagebuf_cond_lock (not any pinning of underlying pages
- * created by pagebuf_pin).
- *
+ * Releases the lock on the buffer object.
* If the buffer is marked delwri but is not queued, do so before we
- * unlock the buffer as we need to set flags correctly. We also need to
+ * unlock the buffer as we need to set flags correctly. We also need to
* take a reference for the delwri queue because the unlocker is going to
* drop their's and they don't know we just queued it.
*/
void
-pagebuf_unlock( /* unlock buffer */
- xfs_buf_t *pb) /* buffer to unlock */
+xfs_buf_unlock(
+ xfs_buf_t *bp)
{
- if ((pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)) == PBF_DELWRI) {
- atomic_inc(&pb->pb_hold);
- pb->pb_flags |= PBF_ASYNC;
- pagebuf_delwri_queue(pb, 0);
+ if ((bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)) == XBF_DELWRI) {
+ atomic_inc(&bp->b_hold);
+ bp->b_flags |= XBF_ASYNC;
+ xfs_buf_delwri_queue(bp, 0);
}
- PB_CLEAR_OWNER(pb);
- up(&pb->pb_sema);
- PB_TRACE(pb, "unlock", 0);
+ XB_CLEAR_OWNER(bp);
+ up(&bp->b_sema);
+ XB_TRACE(bp, "unlock", 0);
}
/*
* Pinning Buffer Storage in Memory
- */
-
-/*
- * pagebuf_pin
- *
- * pagebuf_pin locks all of the memory represented by a buffer in
- * memory. Multiple calls to pagebuf_pin and pagebuf_unpin, for
- * the same or different buffers affecting a given page, will
- * properly count the number of outstanding "pin" requests. The
- * buffer may be released after the pagebuf_pin and a different
- * buffer used when calling pagebuf_unpin, if desired.
- * pagebuf_pin should be used by the file system when it wants be
- * assured that no attempt will be made to force the affected
- * memory to disk. It does not assure that a given logical page
- * will not be moved to a different physical page.
+ * Ensure that no attempt to force a buffer to disk will succeed.
*/
void
-pagebuf_pin(
- xfs_buf_t *pb)
+xfs_buf_pin(
+ xfs_buf_t *bp)
{
- atomic_inc(&pb->pb_pin_count);
- PB_TRACE(pb, "pin", (long)pb->pb_pin_count.counter);
+ atomic_inc(&bp->b_pin_count);
+ XB_TRACE(bp, "pin", (long)bp->b_pin_count.counter);
}
-/*
- * pagebuf_unpin
- *
- * pagebuf_unpin reverses the locking of memory performed by
- * pagebuf_pin. Note that both functions affected the logical
- * pages associated with the buffer, not the buffer itself.
- */
void
-pagebuf_unpin(
- xfs_buf_t *pb)
+xfs_buf_unpin(
+ xfs_buf_t *bp)
{
- if (atomic_dec_and_test(&pb->pb_pin_count)) {
- wake_up_all(&pb->pb_waiters);
- }
- PB_TRACE(pb, "unpin", (long)pb->pb_pin_count.counter);
+ if (atomic_dec_and_test(&bp->b_pin_count))
+ wake_up_all(&bp->b_waiters);
+ XB_TRACE(bp, "unpin", (long)bp->b_pin_count.counter);
}
int
-pagebuf_ispin(
- xfs_buf_t *pb)
+xfs_buf_ispin(
+ xfs_buf_t *bp)
{
- return atomic_read(&pb->pb_pin_count);
+ return atomic_read(&bp->b_pin_count);
}
-/*
- * pagebuf_wait_unpin
- *
- * pagebuf_wait_unpin waits until all of the memory associated
- * with the buffer is not longer locked in memory. It returns
- * immediately if none of the affected pages are locked.
- */
-static inline void
-_pagebuf_wait_unpin(
- xfs_buf_t *pb)
+STATIC void
+xfs_buf_wait_unpin(
+ xfs_buf_t *bp)
{
DECLARE_WAITQUEUE (wait, current);
- if (atomic_read(&pb->pb_pin_count) == 0)
+ if (atomic_read(&bp->b_pin_count) == 0)
return;
- add_wait_queue(&pb->pb_waiters, &wait);
+ add_wait_queue(&bp->b_waiters, &wait);
for (;;) {
set_current_state(TASK_UNINTERRUPTIBLE);
- if (atomic_read(&pb->pb_pin_count) == 0)
+ if (atomic_read(&bp->b_pin_count) == 0)
break;
- if (atomic_read(&pb->pb_io_remaining))
- blk_run_address_space(pb->pb_target->pbr_mapping);
+ if (atomic_read(&bp->b_io_remaining))
+ blk_run_address_space(bp->b_target->bt_mapping);
schedule();
}
- remove_wait_queue(&pb->pb_waiters, &wait);
+ remove_wait_queue(&bp->b_waiters, &wait);
set_current_state(TASK_RUNNING);
}
@@ -1032,241 +978,216 @@ _pagebuf_wait_unpin(
* Buffer Utility Routines
*/
-/*
- * pagebuf_iodone
- *
- * pagebuf_iodone marks a buffer for which I/O is in progress
- * done with respect to that I/O. The pb_iodone routine, if
- * present, will be called as a side-effect.
- */
STATIC void
-pagebuf_iodone_work(
+xfs_buf_iodone_work(
void *v)
{
xfs_buf_t *bp = (xfs_buf_t *)v;
- if (bp->pb_iodone)
- (*(bp->pb_iodone))(bp);
- else if (bp->pb_flags & PBF_ASYNC)
+ if (bp->b_iodone)
+ (*(bp->b_iodone))(bp);
+ else if (bp->b_flags & XBF_ASYNC)
xfs_buf_relse(bp);
}
void
-pagebuf_iodone(
- xfs_buf_t *pb,
+xfs_buf_ioend(
+ xfs_buf_t *bp,
int schedule)
{
- pb->pb_flags &= ~(PBF_READ | PBF_WRITE);
- if (pb->pb_error == 0)
- pb->pb_flags |= PBF_DONE;
+ bp->b_flags &= ~(XBF_READ | XBF_WRITE);
+ if (bp->b_error == 0)
+ bp->b_flags |= XBF_DONE;
- PB_TRACE(pb, "iodone", pb->pb_iodone);
+ XB_TRACE(bp, "iodone", bp->b_iodone);
- if ((pb->pb_iodone) || (pb->pb_flags & PBF_ASYNC)) {
+ if ((bp->b_iodone) || (bp->b_flags & XBF_ASYNC)) {
if (schedule) {
- INIT_WORK(&pb->pb_iodone_work, pagebuf_iodone_work, pb);
- queue_work(xfslogd_workqueue, &pb->pb_iodone_work);
+ INIT_WORK(&bp->b_iodone_work, xfs_buf_iodone_work, bp);
+ queue_work(xfslogd_workqueue, &bp->b_iodone_work);
} else {
- pagebuf_iodone_work(pb);
+ xfs_buf_iodone_work(bp);
}
} else {
- up(&pb->pb_iodonesema);
+ up(&bp->b_iodonesema);
}
}
-/*
- * pagebuf_ioerror
- *
- * pagebuf_ioerror sets the error code for a buffer.
- */
void
-pagebuf_ioerror( /* mark/clear buffer error flag */
- xfs_buf_t *pb, /* buffer to mark */
- int error) /* error to store (0 if none) */
+xfs_buf_ioerror(
+ xfs_buf_t *bp,
+ int error)
{
ASSERT(error >= 0 && error <= 0xffff);
- pb->pb_error = (unsigned short)error;
- PB_TRACE(pb, "ioerror", (unsigned long)error);
+ bp->b_error = (unsigned short)error;
+ XB_TRACE(bp, "ioerror", (unsigned long)error);
}
/*
- * pagebuf_iostart
- *
- * pagebuf_iostart initiates I/O on a buffer, based on the flags supplied.
- * If necessary, it will arrange for any disk space allocation required,
- * and it will break up the request if the block mappings require it.
- * The pb_iodone routine in the buffer supplied will only be called
+ * Initiate I/O on a buffer, based on the flags supplied.
+ * The b_iodone routine in the buffer supplied will only be called
* when all of the subsidiary I/O requests, if any, have been completed.
- * pagebuf_iostart calls the pagebuf_ioinitiate routine or
- * pagebuf_iorequest, if the former routine is not defined, to start
- * the I/O on a given low-level request.
*/
int
-pagebuf_iostart( /* start I/O on a buffer */
- xfs_buf_t *pb, /* buffer to start */
- page_buf_flags_t flags) /* PBF_LOCK, PBF_ASYNC, PBF_READ, */
- /* PBF_WRITE, PBF_DELWRI, */
- /* PBF_DONT_BLOCK */
+xfs_buf_iostart(
+ xfs_buf_t *bp,
+ xfs_buf_flags_t flags)
{
int status = 0;
- PB_TRACE(pb, "iostart", (unsigned long)flags);
+ XB_TRACE(bp, "iostart", (unsigned long)flags);
- if (flags & PBF_DELWRI) {
- pb->pb_flags &= ~(PBF_READ | PBF_WRITE | PBF_ASYNC);
- pb->pb_flags |= flags & (PBF_DELWRI | PBF_ASYNC);
- pagebuf_delwri_queue(pb, 1);
+ if (flags & XBF_DELWRI) {
+ bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC);
+ bp->b_flags |= flags & (XBF_DELWRI | XBF_ASYNC);
+ xfs_buf_delwri_queue(bp, 1);
return status;
}
- pb->pb_flags &= ~(PBF_READ | PBF_WRITE | PBF_ASYNC | PBF_DELWRI | \
- PBF_READ_AHEAD | _PBF_RUN_QUEUES);
- pb->pb_flags |= flags & (PBF_READ | PBF_WRITE | PBF_ASYNC | \
- PBF_READ_AHEAD | _PBF_RUN_QUEUES);
+ bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC | XBF_DELWRI | \
+ XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+ bp->b_flags |= flags & (XBF_READ | XBF_WRITE | XBF_ASYNC | \
+ XBF_READ_AHEAD | _XBF_RUN_QUEUES);
- BUG_ON(pb->pb_bn == XFS_BUF_DADDR_NULL);
+ BUG_ON(bp->b_bn == XFS_BUF_DADDR_NULL);
/* For writes allow an alternate strategy routine to precede
* the actual I/O request (which may not be issued at all in
* a shutdown situation, for example).
*/
- status = (flags & PBF_WRITE) ?
- pagebuf_iostrategy(pb) : pagebuf_iorequest(pb);
+ status = (flags & XBF_WRITE) ?
+ xfs_buf_iostrategy(bp) : xfs_buf_iorequest(bp);
/* Wait for I/O if we are not an async request.
* Note: async I/O request completion will release the buffer,
* and that can already be done by this point. So using the
* buffer pointer from here on, after async I/O, is invalid.
*/
- if (!status && !(flags & PBF_ASYNC))
- status = pagebuf_iowait(pb);
+ if (!status && !(flags & XBF_ASYNC))
+ status = xfs_buf_iowait(bp);
return status;
}
-/*
- * Helper routine for pagebuf_iorequest
- */
-
STATIC __inline__ int
-_pagebuf_iolocked(
- xfs_buf_t *pb)
+_xfs_buf_iolocked(
+ xfs_buf_t *bp)
{
- ASSERT(pb->pb_flags & (PBF_READ|PBF_WRITE));
- if (pb->pb_flags & PBF_READ)
- return pb->pb_locked;
+ ASSERT(bp->b_flags & (XBF_READ | XBF_WRITE));
+ if (bp->b_flags & XBF_READ)
+ return bp->b_locked;
return 0;
}
STATIC __inline__ void
-_pagebuf_iodone(
- xfs_buf_t *pb,
+_xfs_buf_ioend(
+ xfs_buf_t *bp,
int schedule)
{
- if (atomic_dec_and_test(&pb->pb_io_remaining) == 1) {
- pb->pb_locked = 0;
- pagebuf_iodone(pb, schedule);
+ if (atomic_dec_and_test(&bp->b_io_remaining) == 1) {
+ bp->b_locked = 0;
+ xfs_buf_ioend(bp, schedule);
}
}
STATIC int
-bio_end_io_pagebuf(
+xfs_buf_bio_end_io(
struct bio *bio,
unsigned int bytes_done,
int error)
{
- xfs_buf_t *pb = (xfs_buf_t *)bio->bi_private;
- unsigned int blocksize = pb->pb_target->pbr_bsize;
+ xfs_buf_t *bp = (xfs_buf_t *)bio->bi_private;
+ unsigned int blocksize = bp->b_target->bt_bsize;
struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
if (bio->bi_size)
return 1;
if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
- pb->pb_error = EIO;
+ bp->b_error = EIO;
do {
struct page *page = bvec->bv_page;
- if (unlikely(pb->pb_error)) {
- if (pb->pb_flags & PBF_READ)
+ if (unlikely(bp->b_error)) {
+ if (bp->b_flags & XBF_READ)
ClearPageUptodate(page);
SetPageError(page);
- } else if (blocksize == PAGE_CACHE_SIZE) {
+ } else if (blocksize >= PAGE_CACHE_SIZE) {
SetPageUptodate(page);
} else if (!PagePrivate(page) &&
- (pb->pb_flags & _PBF_PAGE_CACHE)) {
+ (bp->b_flags & _XBF_PAGE_CACHE)) {
set_page_region(page, bvec->bv_offset, bvec->bv_len);
}
if (--bvec >= bio->bi_io_vec)
prefetchw(&bvec->bv_page->flags);
- if (_pagebuf_iolocked(pb)) {
+ if (_xfs_buf_iolocked(bp)) {
unlock_page(page);
}
} while (bvec >= bio->bi_io_vec);
- _pagebuf_iodone(pb, 1);
+ _xfs_buf_ioend(bp, 1);
bio_put(bio);
return 0;
}
STATIC void
-_pagebuf_ioapply(
- xfs_buf_t *pb)
+_xfs_buf_ioapply(
+ xfs_buf_t *bp)
{
int i, rw, map_i, total_nr_pages, nr_pages;
struct bio *bio;
- int offset = pb->pb_offset;
- int size = pb->pb_count_desired;
- sector_t sector = pb->pb_bn;
- unsigned int blocksize = pb->pb_target->pbr_bsize;
- int locking = _pagebuf_iolocked(pb);
+ int offset = bp->b_offset;
+ int size = bp->b_count_desired;
+ sector_t sector = bp->b_bn;
+ unsigned int blocksize = bp->b_target->bt_bsize;
+ int locking = _xfs_buf_iolocked(bp);
- total_nr_pages = pb->pb_page_count;
+ total_nr_pages = bp->b_page_count;
map_i = 0;
- if (pb->pb_flags & _PBF_RUN_QUEUES) {
- pb->pb_flags &= ~_PBF_RUN_QUEUES;
- rw = (pb->pb_flags & PBF_READ) ? READ_SYNC : WRITE_SYNC;
+ if (bp->b_flags & _XBF_RUN_QUEUES) {
+ bp->b_flags &= ~_XBF_RUN_QUEUES;
+ rw = (bp->b_flags & XBF_READ) ? READ_SYNC : WRITE_SYNC;
} else {
- rw = (pb->pb_flags & PBF_READ) ? READ : WRITE;
+ rw = (bp->b_flags & XBF_READ) ? READ : WRITE;
}
- if (pb->pb_flags & PBF_ORDERED) {
- ASSERT(!(pb->pb_flags & PBF_READ));
+ if (bp->b_flags & XBF_ORDERED) {
+ ASSERT(!(bp->b_flags & XBF_READ));
rw = WRITE_BARRIER;
}
- /* Special code path for reading a sub page size pagebuf in --
+ /* Special code path for reading a sub page size buffer in --
* we populate up the whole page, and hence the other metadata
* in the same page. This optimization is only valid when the
- * filesystem block size and the page size are equal.
+ * filesystem block size is not smaller than the page size.
*/
- if ((pb->pb_buffer_length < PAGE_CACHE_SIZE) &&
- (pb->pb_flags & PBF_READ) && locking &&
- (blocksize == PAGE_CACHE_SIZE)) {
+ if ((bp->b_buffer_length < PAGE_CACHE_SIZE) &&
+ (bp->b_flags & XBF_READ) && locking &&
+ (blocksize >= PAGE_CACHE_SIZE)) {
bio = bio_alloc(GFP_NOIO, 1);
- bio->bi_bdev = pb->pb_target->pbr_bdev;
+ bio->bi_bdev = bp->b_target->bt_bdev;
bio->bi_sector = sector - (offset >> BBSHIFT);
- bio->bi_end_io = bio_end_io_pagebuf;
- bio->bi_private = pb;
+ bio->bi_end_io = xfs_buf_bio_end_io;
+ bio->bi_private = bp;
- bio_add_page(bio, pb->pb_pages[0], PAGE_CACHE_SIZE, 0);
+ bio_add_page(bio, bp->b_pages[0], PAGE_CACHE_SIZE, 0);
size = 0;
- atomic_inc(&pb->pb_io_remaining);
+ atomic_inc(&bp->b_io_remaining);
goto submit_io;
}
/* Lock down the pages which we need to for the request */
- if (locking && (pb->pb_flags & PBF_WRITE) && (pb->pb_locked == 0)) {
+ if (locking && (bp->b_flags & XBF_WRITE) && (bp->b_locked == 0)) {
for (i = 0; size; i++) {
int nbytes = PAGE_CACHE_SIZE - offset;
- struct page *page = pb->pb_pages[i];
+ struct page *page = bp->b_pages[i];
if (nbytes > size)
nbytes = size;
@@ -1276,30 +1197,30 @@ _pagebuf_ioapply(
size -= nbytes;
offset = 0;
}
- offset = pb->pb_offset;
- size = pb->pb_count_desired;
+ offset = bp->b_offset;
+ size = bp->b_count_desired;
}
next_chunk:
- atomic_inc(&pb->pb_io_remaining);
+ atomic_inc(&bp->b_io_remaining);
nr_pages = BIO_MAX_SECTORS >> (PAGE_SHIFT - BBSHIFT);
if (nr_pages > total_nr_pages)
nr_pages = total_nr_pages;
bio = bio_alloc(GFP_NOIO, nr_pages);
- bio->bi_bdev = pb->pb_target->pbr_bdev;
+ bio->bi_bdev = bp->b_target->bt_bdev;
bio->bi_sector = sector;
- bio->bi_end_io = bio_end_io_pagebuf;
- bio->bi_private = pb;
+ bio->bi_end_io = xfs_buf_bio_end_io;
+ bio->bi_private = bp;
for (; size && nr_pages; nr_pages--, map_i++) {
- int nbytes = PAGE_CACHE_SIZE - offset;
+ int rbytes, nbytes = PAGE_CACHE_SIZE - offset;
if (nbytes > size)
nbytes = size;
- if (bio_add_page(bio, pb->pb_pages[map_i],
- nbytes, offset) < nbytes)
+ rbytes = bio_add_page(bio, bp->b_pages[map_i], nbytes, offset);
+ if (rbytes < nbytes)
break;
offset = 0;
@@ -1315,107 +1236,102 @@ submit_io:
goto next_chunk;
} else {
bio_put(bio);
- pagebuf_ioerror(pb, EIO);
+ xfs_buf_ioerror(bp, EIO);
}
}
-/*
- * pagebuf_iorequest -- the core I/O request routine.
- */
int
-pagebuf_iorequest( /* start real I/O */
- xfs_buf_t *pb) /* buffer to convey to device */
+xfs_buf_iorequest(
+ xfs_buf_t *bp)
{
- PB_TRACE(pb, "iorequest", 0);
+ XB_TRACE(bp, "iorequest", 0);
- if (pb->pb_flags & PBF_DELWRI) {
- pagebuf_delwri_queue(pb, 1);
+ if (bp->b_flags & XBF_DELWRI) {
+ xfs_buf_delwri_queue(bp, 1);
return 0;
}
- if (pb->pb_flags & PBF_WRITE) {
- _pagebuf_wait_unpin(pb);
+ if (bp->b_flags & XBF_WRITE) {
+ xfs_buf_wait_unpin(bp);
}
- pagebuf_hold(pb);
+ xfs_buf_hold(bp);
/* Set the count to 1 initially, this will stop an I/O
* completion callout which happens before we have started
- * all the I/O from calling pagebuf_iodone too early.
+ * all the I/O from calling xfs_buf_ioend too early.
*/
- atomic_set(&pb->pb_io_remaining, 1);
- _pagebuf_ioapply(pb);
- _pagebuf_iodone(pb, 0);
+ atomic_set(&bp->b_io_remaining, 1);
+ _xfs_buf_ioapply(bp);
+ _xfs_buf_ioend(bp, 0);
- pagebuf_rele(pb);
+ xfs_buf_rele(bp);
return 0;
}
/*
- * pagebuf_iowait
- *
- * pagebuf_iowait waits for I/O to complete on the buffer supplied.
- * It returns immediately if no I/O is pending. In any case, it returns
- * the error code, if any, or 0 if there is no error.
+ * Waits for I/O to complete on the buffer supplied.
+ * It returns immediately if no I/O is pending.
+ * It returns the I/O error code, if any, or 0 if there was no error.
*/
int
-pagebuf_iowait(
- xfs_buf_t *pb)
+xfs_buf_iowait(
+ xfs_buf_t *bp)
{
- PB_TRACE(pb, "iowait", 0);
- if (atomic_read(&pb->pb_io_remaining))
- blk_run_address_space(pb->pb_target->pbr_mapping);
- down(&pb->pb_iodonesema);
- PB_TRACE(pb, "iowaited", (long)pb->pb_error);
- return pb->pb_error;
+ XB_TRACE(bp, "iowait", 0);
+ if (atomic_read(&bp->b_io_remaining))
+ blk_run_address_space(bp->b_target->bt_mapping);
+ down(&bp->b_iodonesema);
+ XB_TRACE(bp, "iowaited", (long)bp->b_error);
+ return bp->b_error;
}
-caddr_t
-pagebuf_offset(
- xfs_buf_t *pb,
+xfs_caddr_t
+xfs_buf_offset(
+ xfs_buf_t *bp,
size_t offset)
{
struct page *page;
- offset += pb->pb_offset;
+ if (bp->b_flags & XBF_MAPPED)
+ return XFS_BUF_PTR(bp) + offset;
- page = pb->pb_pages[offset >> PAGE_CACHE_SHIFT];
- return (caddr_t) page_address(page) + (offset & (PAGE_CACHE_SIZE - 1));
+ offset += bp->b_offset;
+ page = bp->b_pages[offset >> PAGE_CACHE_SHIFT];
+ return (xfs_caddr_t)page_address(page) + (offset & (PAGE_CACHE_SIZE-1));
}
/*
- * pagebuf_iomove
- *
* Move data into or out of a buffer.
*/
void
-pagebuf_iomove(
- xfs_buf_t *pb, /* buffer to process */
+xfs_buf_iomove(
+ xfs_buf_t *bp, /* buffer to process */
size_t boff, /* starting buffer offset */
size_t bsize, /* length to copy */
caddr_t data, /* data address */
- page_buf_rw_t mode) /* read/write flag */
+ xfs_buf_rw_t mode) /* read/write/zero flag */
{
size_t bend, cpoff, csize;
struct page *page;
bend = boff + bsize;
while (boff < bend) {
- page = pb->pb_pages[page_buf_btoct(boff + pb->pb_offset)];
- cpoff = page_buf_poff(boff + pb->pb_offset);
+ page = bp->b_pages[xfs_buf_btoct(boff + bp->b_offset)];
+ cpoff = xfs_buf_poff(boff + bp->b_offset);
csize = min_t(size_t,
- PAGE_CACHE_SIZE-cpoff, pb->pb_count_desired-boff);
+ PAGE_CACHE_SIZE-cpoff, bp->b_count_desired-boff);
ASSERT(((csize + cpoff) <= PAGE_CACHE_SIZE));
switch (mode) {
- case PBRW_ZERO:
+ case XBRW_ZERO:
memset(page_address(page) + cpoff, 0, csize);
break;
- case PBRW_READ:
+ case XBRW_READ:
memcpy(data, page_address(page) + cpoff, csize);
break;
- case PBRW_WRITE:
+ case XBRW_WRITE:
memcpy(page_address(page) + cpoff, data, csize);
}
@@ -1425,12 +1341,12 @@ pagebuf_iomove(
}
/*
- * Handling of buftargs.
+ * Handling of buffer targets (buftargs).
*/
/*
- * Wait for any bufs with callbacks that have been submitted but
- * have not yet returned... walk the hash list for the target.
+ * Wait for any bufs with callbacks that have been submitted but
+ * have not yet returned... walk the hash list for the target.
*/
void
xfs_wait_buftarg(
@@ -1444,15 +1360,15 @@ xfs_wait_buftarg(
hash = &btp->bt_hash[i];
again:
spin_lock(&hash->bh_lock);
- list_for_each_entry_safe(bp, n, &hash->bh_list, pb_hash_list) {
- ASSERT(btp == bp->pb_target);
- if (!(bp->pb_flags & PBF_FS_MANAGED)) {
+ list_for_each_entry_safe(bp, n, &hash->bh_list, b_hash_list) {
+ ASSERT(btp == bp->b_target);
+ if (!(bp->b_flags & XBF_FS_MANAGED)) {
spin_unlock(&hash->bh_lock);
/*
* Catch superblock reference count leaks
* immediately
*/
- BUG_ON(bp->pb_bn == 0);
+ BUG_ON(bp->b_bn == 0);
delay(100);
goto again;
}
@@ -1462,9 +1378,9 @@ again:
}
/*
- * Allocate buffer hash table for a given target.
- * For devices containing metadata (i.e. not the log/realtime devices)
- * we need to allocate a much larger hash table.
+ * Allocate buffer hash table for a given target.
+ * For devices containing metadata (i.e. not the log/realtime devices)
+ * we need to allocate a much larger hash table.
*/
STATIC void
xfs_alloc_bufhash(
@@ -1487,11 +1403,34 @@ STATIC void
xfs_free_bufhash(
xfs_buftarg_t *btp)
{
- kmem_free(btp->bt_hash,
- (1 << btp->bt_hashshift) * sizeof(xfs_bufhash_t));
+ kmem_free(btp->bt_hash, (1<<btp->bt_hashshift) * sizeof(xfs_bufhash_t));
btp->bt_hash = NULL;
}
+/*
+ * buftarg list for delwrite queue processing
+ */
+STATIC LIST_HEAD(xfs_buftarg_list);
+STATIC DEFINE_SPINLOCK(xfs_buftarg_lock);
+
+STATIC void
+xfs_register_buftarg(
+ xfs_buftarg_t *btp)
+{
+ spin_lock(&xfs_buftarg_lock);
+ list_add(&btp->bt_list, &xfs_buftarg_list);
+ spin_unlock(&xfs_buftarg_lock);
+}
+
+STATIC void
+xfs_unregister_buftarg(
+ xfs_buftarg_t *btp)
+{
+ spin_lock(&xfs_buftarg_lock);
+ list_del(&btp->bt_list);
+ spin_unlock(&xfs_buftarg_lock);
+}
+
void
xfs_free_buftarg(
xfs_buftarg_t *btp,
@@ -1499,9 +1438,16 @@ xfs_free_buftarg(
{
xfs_flush_buftarg(btp, 1);
if (external)
- xfs_blkdev_put(btp->pbr_bdev);
+ xfs_blkdev_put(btp->bt_bdev);
xfs_free_bufhash(btp);
- iput(btp->pbr_mapping->host);
+ iput(btp->bt_mapping->host);
+
+ /* Unregister the buftarg first so that we don't get a
+ * wakeup finding a non-existent task
+ */
+ xfs_unregister_buftarg(btp);
+ kthread_stop(btp->bt_task);
+
kmem_free(btp, sizeof(*btp));
}
@@ -1512,11 +1458,11 @@ xfs_setsize_buftarg_flags(
unsigned int sectorsize,
int verbose)
{
- btp->pbr_bsize = blocksize;
- btp->pbr_sshift = ffs(sectorsize) - 1;
- btp->pbr_smask = sectorsize - 1;
+ btp->bt_bsize = blocksize;
+ btp->bt_sshift = ffs(sectorsize) - 1;
+ btp->bt_smask = sectorsize - 1;
- if (set_blocksize(btp->pbr_bdev, sectorsize)) {
+ if (set_blocksize(btp->bt_bdev, sectorsize)) {
printk(KERN_WARNING
"XFS: Cannot set_blocksize to %u on device %s\n",
sectorsize, XFS_BUFTARG_NAME(btp));
@@ -1536,10 +1482,10 @@ xfs_setsize_buftarg_flags(
}
/*
-* When allocating the initial buffer target we have not yet
-* read in the superblock, so don't know what sized sectors
-* are being used is at this early stage. Play safe.
-*/
+ * When allocating the initial buffer target we have not yet
+ * read in the superblock, so don't know what sized sectors
+ * are being used is at this early stage. Play safe.
+ */
STATIC int
xfs_setsize_buftarg_early(
xfs_buftarg_t *btp,
@@ -1587,10 +1533,30 @@ xfs_mapping_buftarg(
mapping->a_ops = &mapping_aops;
mapping->backing_dev_info = bdi;
mapping_set_gfp_mask(mapping, GFP_NOFS);
- btp->pbr_mapping = mapping;
+ btp->bt_mapping = mapping;
return 0;
}
+STATIC int
+xfs_alloc_delwrite_queue(
+ xfs_buftarg_t *btp)
+{
+ int error = 0;
+
+ INIT_LIST_HEAD(&btp->bt_list);
+ INIT_LIST_HEAD(&btp->bt_delwrite_queue);
+ spinlock_init(&btp->bt_delwrite_lock, "delwri_lock");
+ btp->bt_flags = 0;
+ btp->bt_task = kthread_run(xfsbufd, btp, "xfsbufd");
+ if (IS_ERR(btp->bt_task)) {
+ error = PTR_ERR(btp->bt_task);
+ goto out_error;
+ }
+ xfs_register_buftarg(btp);
+out_error:
+ return error;
+}
+
xfs_buftarg_t *
xfs_alloc_buftarg(
struct block_device *bdev,
@@ -1600,12 +1566,14 @@ xfs_alloc_buftarg(
btp = kmem_zalloc(sizeof(*btp), KM_SLEEP);
- btp->pbr_dev = bdev->bd_dev;
- btp->pbr_bdev = bdev;
+ btp->bt_dev = bdev->bd_dev;
+ btp->bt_bdev = bdev;
if (xfs_setsize_buftarg_early(btp, bdev))
goto error;
if (xfs_mapping_buftarg(btp, bdev))
goto error;
+ if (xfs_alloc_delwrite_queue(btp))
+ goto error;
xfs_alloc_bufhash(btp, external);
return btp;
@@ -1616,83 +1584,81 @@ error:
/*
- * Pagebuf delayed write buffer handling
+ * Delayed write buffer handling
*/
-
-STATIC LIST_HEAD(pbd_delwrite_queue);
-STATIC DEFINE_SPINLOCK(pbd_delwrite_lock);
-
STATIC void
-pagebuf_delwri_queue(
- xfs_buf_t *pb,
+xfs_buf_delwri_queue(
+ xfs_buf_t *bp,
int unlock)
{
- PB_TRACE(pb, "delwri_q", (long)unlock);
- ASSERT((pb->pb_flags & (PBF_DELWRI|PBF_ASYNC)) ==
- (PBF_DELWRI|PBF_ASYNC));
+ struct list_head *dwq = &bp->b_target->bt_delwrite_queue;
+ spinlock_t *dwlk = &bp->b_target->bt_delwrite_lock;
+
+ XB_TRACE(bp, "delwri_q", (long)unlock);
+ ASSERT((bp->b_flags&(XBF_DELWRI|XBF_ASYNC)) == (XBF_DELWRI|XBF_ASYNC));
- spin_lock(&pbd_delwrite_lock);
+ spin_lock(dwlk);
/* If already in the queue, dequeue and place at tail */
- if (!list_empty(&pb->pb_list)) {
- ASSERT(pb->pb_flags & _PBF_DELWRI_Q);
- if (unlock) {
- atomic_dec(&pb->pb_hold);
- }
- list_del(&pb->pb_list);
+ if (!list_empty(&bp->b_list)) {
+ ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+ if (unlock)
+ atomic_dec(&bp->b_hold);
+ list_del(&bp->b_list);
}
- pb->pb_flags |= _PBF_DELWRI_Q;
- list_add_tail(&pb->pb_list, &pbd_delwrite_queue);
- pb->pb_queuetime = jiffies;
- spin_unlock(&pbd_delwrite_lock);
+ bp->b_flags |= _XBF_DELWRI_Q;
+ list_add_tail(&bp->b_list, dwq);
+ bp->b_queuetime = jiffies;
+ spin_unlock(dwlk);
if (unlock)
- pagebuf_unlock(pb);
+ xfs_buf_unlock(bp);
}
void
-pagebuf_delwri_dequeue(
- xfs_buf_t *pb)
+xfs_buf_delwri_dequeue(
+ xfs_buf_t *bp)
{
+ spinlock_t *dwlk = &bp->b_target->bt_delwrite_lock;
int dequeued = 0;
- spin_lock(&pbd_delwrite_lock);
- if ((pb->pb_flags & PBF_DELWRI) && !list_empty(&pb->pb_list)) {
- ASSERT(pb->pb_flags & _PBF_DELWRI_Q);
- list_del_init(&pb->pb_list);
+ spin_lock(dwlk);
+ if ((bp->b_flags & XBF_DELWRI) && !list_empty(&bp->b_list)) {
+ ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+ list_del_init(&bp->b_list);
dequeued = 1;
}
- pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
- spin_unlock(&pbd_delwrite_lock);
+ bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+ spin_unlock(dwlk);
if (dequeued)
- pagebuf_rele(pb);
+ xfs_buf_rele(bp);
- PB_TRACE(pb, "delwri_dq", (long)dequeued);
+ XB_TRACE(bp, "delwri_dq", (long)dequeued);
}
STATIC void
-pagebuf_runall_queues(
+xfs_buf_runall_queues(
struct workqueue_struct *queue)
{
flush_workqueue(queue);
}
-/* Defines for pagebuf daemon */
-STATIC struct task_struct *xfsbufd_task;
-STATIC int xfsbufd_force_flush;
-STATIC int xfsbufd_force_sleep;
-
STATIC int
xfsbufd_wakeup(
int priority,
gfp_t mask)
{
- if (xfsbufd_force_sleep)
- return 0;
- xfsbufd_force_flush = 1;
- barrier();
- wake_up_process(xfsbufd_task);
+ xfs_buftarg_t *btp;
+
+ spin_lock(&xfs_buftarg_lock);
+ list_for_each_entry(btp, &xfs_buftarg_list, bt_list) {
+ if (test_bit(XBT_FORCE_SLEEP, &btp->bt_flags))
+ continue;
+ set_bit(XBT_FORCE_FLUSH, &btp->bt_flags);
+ wake_up_process(btp->bt_task);
+ }
+ spin_unlock(&xfs_buftarg_lock);
return 0;
}
@@ -1702,67 +1668,70 @@ xfsbufd(
{
struct list_head tmp;
unsigned long age;
- xfs_buftarg_t *target;
- xfs_buf_t *pb, *n;
+ xfs_buftarg_t *target = (xfs_buftarg_t *)data;
+ xfs_buf_t *bp, *n;
+ struct list_head *dwq = &target->bt_delwrite_queue;
+ spinlock_t *dwlk = &target->bt_delwrite_lock;
current->flags |= PF_MEMALLOC;
INIT_LIST_HEAD(&tmp);
do {
if (unlikely(freezing(current))) {
- xfsbufd_force_sleep = 1;
+ set_bit(XBT_FORCE_SLEEP, &target->bt_flags);
refrigerator();
} else {
- xfsbufd_force_sleep = 0;
+ clear_bit(XBT_FORCE_SLEEP, &target->bt_flags);
}
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);
- list_for_each_entry_safe(pb, n, &pbd_delwrite_queue, pb_list) {
- PB_TRACE(pb, "walkq1", (long)pagebuf_ispin(pb));
- ASSERT(pb->pb_flags & PBF_DELWRI);
-
- if (!pagebuf_ispin(pb) && !pagebuf_cond_lock(pb)) {
- if (!xfsbufd_force_flush &&
+ spin_lock(dwlk);
+ list_for_each_entry_safe(bp, n, dwq, b_list) {
+ XB_TRACE(bp, "walkq1", (long)xfs_buf_ispin(bp));
+ ASSERT(bp->b_flags & XBF_DELWRI);
+
+ if (!xfs_buf_ispin(bp) && !xfs_buf_cond_lock(bp)) {
+ if (!test_bit(XBT_FORCE_FLUSH,
+ &target->bt_flags) &&
time_before(jiffies,
- pb->pb_queuetime + age)) {
- pagebuf_unlock(pb);
+ bp->b_queuetime + age)) {
+ xfs_buf_unlock(bp);
break;
}
- pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
- pb->pb_flags |= PBF_WRITE;
- list_move(&pb->pb_list, &tmp);
+ bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+ bp->b_flags |= XBF_WRITE;
+ list_move(&bp->b_list, &tmp);
}
}
- spin_unlock(&pbd_delwrite_lock);
+ spin_unlock(dwlk);
while (!list_empty(&tmp)) {
- pb = list_entry(tmp.next, xfs_buf_t, pb_list);
- target = pb->pb_target;
+ bp = list_entry(tmp.next, xfs_buf_t, b_list);
+ ASSERT(target == bp->b_target);
- list_del_init(&pb->pb_list);
- pagebuf_iostrategy(pb);
+ list_del_init(&bp->b_list);
+ xfs_buf_iostrategy(bp);
- blk_run_address_space(target->pbr_mapping);
+ blk_run_address_space(target->bt_mapping);
}
if (as_list_len > 0)
purge_addresses();
- xfsbufd_force_flush = 0;
+ clear_bit(XBT_FORCE_FLUSH, &target->bt_flags);
} while (!kthread_should_stop());
return 0;
}
/*
- * Go through all incore buffers, and release buffers if they belong to
- * the given device. This is used in filesystem error handling to
- * preserve the consistency of its metadata.
+ * Go through all incore buffers, and release buffers if they belong to
+ * the given device. This is used in filesystem error handling to
+ * preserve the consistency of its metadata.
*/
int
xfs_flush_buftarg(
@@ -1770,73 +1739,72 @@ xfs_flush_buftarg(
int wait)
{
struct list_head tmp;
- xfs_buf_t *pb, *n;
+ xfs_buf_t *bp, *n;
int pincount = 0;
+ struct list_head *dwq = &target->bt_delwrite_queue;
+ spinlock_t *dwlk = &target->bt_delwrite_lock;
- pagebuf_runall_queues(xfsdatad_workqueue);
- pagebuf_runall_queues(xfslogd_workqueue);
+ xfs_buf_runall_queues(xfsdatad_workqueue);
+ xfs_buf_runall_queues(xfslogd_workqueue);
INIT_LIST_HEAD(&tmp);
- spin_lock(&pbd_delwrite_lock);
- list_for_each_entry_safe(pb, n, &pbd_delwrite_queue, pb_list) {
-
- if (pb->pb_target != target)
- continue;
-
- ASSERT(pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q));
- PB_TRACE(pb, "walkq2", (long)pagebuf_ispin(pb));
- if (pagebuf_ispin(pb)) {
+ spin_lock(dwlk);
+ list_for_each_entry_safe(bp, n, dwq, b_list) {
+ ASSERT(bp->b_target == target);
+ ASSERT(bp->b_flags & (XBF_DELWRI | _XBF_DELWRI_Q));
+ XB_TRACE(bp, "walkq2", (long)xfs_buf_ispin(bp));
+ if (xfs_buf_ispin(bp)) {
pincount++;
continue;
}
- list_move(&pb->pb_list, &tmp);
+ list_move(&bp->b_list, &tmp);
}
- spin_unlock(&pbd_delwrite_lock);
+ spin_unlock(dwlk);
/*
* Dropped the delayed write list lock, now walk the temporary list
*/
- list_for_each_entry_safe(pb, n, &tmp, pb_list) {
- pagebuf_lock(pb);
- pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
- pb->pb_flags |= PBF_WRITE;
+ list_for_each_entry_safe(bp, n, &tmp, b_list) {
+ xfs_buf_lock(bp);
+ bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+ bp->b_flags |= XBF_WRITE;
if (wait)
- pb->pb_flags &= ~PBF_ASYNC;
+ bp->b_flags &= ~XBF_ASYNC;
else
- list_del_init(&pb->pb_list);
+ list_del_init(&bp->b_list);
- pagebuf_iostrategy(pb);
+ xfs_buf_iostrategy(bp);
}
/*
* Remaining list items must be flushed before returning
*/
while (!list_empty(&tmp)) {
- pb = list_entry(tmp.next, xfs_buf_t, pb_list);
+ bp = list_entry(tmp.next, xfs_buf_t, b_list);
- list_del_init(&pb->pb_list);
- xfs_iowait(pb);
- xfs_buf_relse(pb);
+ list_del_init(&bp->b_list);
+ xfs_iowait(bp);
+ xfs_buf_relse(bp);
}
if (wait)
- blk_run_address_space(target->pbr_mapping);
+ blk_run_address_space(target->bt_mapping);
return pincount;
}
int __init
-pagebuf_init(void)
+xfs_buf_init(void)
{
int error = -ENOMEM;
-#ifdef PAGEBUF_TRACE
- pagebuf_trace_buf = ktrace_alloc(PAGEBUF_TRACE_SIZE, KM_SLEEP);
+#ifdef XFS_BUF_TRACE
+ xfs_buf_trace_buf = ktrace_alloc(XFS_BUF_TRACE_SIZE, KM_SLEEP);
#endif
- pagebuf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
- if (!pagebuf_zone)
+ xfs_buf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
+ if (!xfs_buf_zone)
goto out_free_trace_buf;
xfslogd_workqueue = create_workqueue("xfslogd");
@@ -1847,42 +1815,33 @@ pagebuf_init(void)
if (!xfsdatad_workqueue)
goto out_destroy_xfslogd_workqueue;
- xfsbufd_task = kthread_run(xfsbufd, NULL, "xfsbufd");
- if (IS_ERR(xfsbufd_task)) {
- error = PTR_ERR(xfsbufd_task);
+ xfs_buf_shake = kmem_shake_register(xfsbufd_wakeup);
+ if (!xfs_buf_shake)
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_free_buf_zone:
- kmem_zone_destroy(pagebuf_zone);
+ kmem_zone_destroy(xfs_buf_zone);
out_free_trace_buf:
-#ifdef PAGEBUF_TRACE
- ktrace_free(pagebuf_trace_buf);
+#ifdef XFS_BUF_TRACE
+ ktrace_free(xfs_buf_trace_buf);
#endif
return error;
}
void
-pagebuf_terminate(void)
+xfs_buf_terminate(void)
{
- kmem_shake_deregister(pagebuf_shake);
- kthread_stop(xfsbufd_task);
+ kmem_shake_deregister(xfs_buf_shake);
destroy_workqueue(xfsdatad_workqueue);
destroy_workqueue(xfslogd_workqueue);
- kmem_zone_destroy(pagebuf_zone);
-#ifdef PAGEBUF_TRACE
- ktrace_free(pagebuf_trace_buf);
+ kmem_zone_destroy(xfs_buf_zone);
+#ifdef XFS_BUF_TRACE
+ ktrace_free(xfs_buf_trace_buf);
#endif
}
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index 237a35b915d..4dd6592d5a4 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -32,44 +32,47 @@
* Base types
*/
-#define XFS_BUF_DADDR_NULL ((xfs_daddr_t) (-1LL))
-
-#define page_buf_ctob(pp) ((pp) * PAGE_CACHE_SIZE)
-#define page_buf_btoc(dd) (((dd) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT)
-#define page_buf_btoct(dd) ((dd) >> PAGE_CACHE_SHIFT)
-#define page_buf_poff(aa) ((aa) & ~PAGE_CACHE_MASK)
-
-typedef enum page_buf_rw_e {
- PBRW_READ = 1, /* transfer into target memory */
- PBRW_WRITE = 2, /* transfer from target memory */
- PBRW_ZERO = 3 /* Zero target memory */
-} page_buf_rw_t;
-
-
-typedef enum page_buf_flags_e { /* pb_flags values */
- PBF_READ = (1 << 0), /* buffer intended for reading from device */
- PBF_WRITE = (1 << 1), /* buffer intended for writing to device */
- PBF_MAPPED = (1 << 2), /* buffer mapped (pb_addr valid) */
- PBF_ASYNC = (1 << 4), /* initiator will not wait for completion */
- PBF_DONE = (1 << 5), /* all pages in the buffer uptodate */
- PBF_DELWRI = (1 << 6), /* buffer has dirty pages */
- PBF_STALE = (1 << 7), /* buffer has been staled, do not find it */
- PBF_FS_MANAGED = (1 << 8), /* filesystem controls freeing memory */
- PBF_ORDERED = (1 << 11), /* use ordered writes */
- PBF_READ_AHEAD = (1 << 12), /* asynchronous read-ahead */
+#define XFS_BUF_DADDR_NULL ((xfs_daddr_t) (-1LL))
+
+#define xfs_buf_ctob(pp) ((pp) * PAGE_CACHE_SIZE)
+#define xfs_buf_btoc(dd) (((dd) + PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_btoct(dd) ((dd) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_poff(aa) ((aa) & ~PAGE_CACHE_MASK)
+
+typedef enum {
+ XBRW_READ = 1, /* transfer into target memory */
+ XBRW_WRITE = 2, /* transfer from target memory */
+ XBRW_ZERO = 3, /* Zero target memory */
+} xfs_buf_rw_t;
+
+typedef enum {
+ XBF_READ = (1 << 0), /* buffer intended for reading from device */
+ XBF_WRITE = (1 << 1), /* buffer intended for writing to device */
+ XBF_MAPPED = (1 << 2), /* buffer mapped (b_addr valid) */
+ XBF_ASYNC = (1 << 4), /* initiator will not wait for completion */
+ XBF_DONE = (1 << 5), /* all pages in the buffer uptodate */
+ XBF_DELWRI = (1 << 6), /* buffer has dirty pages */
+ XBF_STALE = (1 << 7), /* buffer has been staled, do not find it */
+ XBF_FS_MANAGED = (1 << 8), /* filesystem controls freeing memory */
+ XBF_ORDERED = (1 << 11), /* use ordered writes */
+ XBF_READ_AHEAD = (1 << 12), /* asynchronous read-ahead */
/* flags used only as arguments to access routines */
- PBF_LOCK = (1 << 14), /* lock requested */
- PBF_TRYLOCK = (1 << 15), /* lock requested, but do not wait */
- PBF_DONT_BLOCK = (1 << 16), /* do not block in current thread */
+ XBF_LOCK = (1 << 14), /* lock requested */
+ XBF_TRYLOCK = (1 << 15), /* lock requested, but do not wait */
+ XBF_DONT_BLOCK = (1 << 16), /* do not block in current thread */
/* flags used only internally */
- _PBF_PAGE_CACHE = (1 << 17),/* backed by pagecache */
- _PBF_KMEM_ALLOC = (1 << 18),/* backed by kmem_alloc() */
- _PBF_RUN_QUEUES = (1 << 19),/* run block device task queue */
- _PBF_DELWRI_Q = (1 << 21), /* buffer on delwri queue */
-} page_buf_flags_t;
+ _XBF_PAGE_CACHE = (1 << 17),/* backed by pagecache */
+ _XBF_KMEM_ALLOC = (1 << 18),/* backed by kmem_alloc() */
+ _XBF_RUN_QUEUES = (1 << 19),/* run block device task queue */
+ _XBF_DELWRI_Q = (1 << 21), /* buffer on delwri queue */
+} xfs_buf_flags_t;
+typedef enum {
+ XBT_FORCE_SLEEP = (0 << 1),
+ XBT_FORCE_FLUSH = (1 << 1),
+} xfs_buftarg_flags_t;
typedef struct xfs_bufhash {
struct list_head bh_list;
@@ -77,477 +80,350 @@ typedef struct xfs_bufhash {
} xfs_bufhash_t;
typedef struct xfs_buftarg {
- dev_t pbr_dev;
- struct block_device *pbr_bdev;
- struct address_space *pbr_mapping;
- unsigned int pbr_bsize;
- unsigned int pbr_sshift;
- size_t pbr_smask;
-
- /* per-device buffer hash table */
+ dev_t bt_dev;
+ struct block_device *bt_bdev;
+ struct address_space *bt_mapping;
+ unsigned int bt_bsize;
+ unsigned int bt_sshift;
+ size_t bt_smask;
+
+ /* per device buffer hash table */
uint bt_hashmask;
uint bt_hashshift;
xfs_bufhash_t *bt_hash;
+
+ /* per device delwri queue */
+ struct task_struct *bt_task;
+ struct list_head bt_list;
+ struct list_head bt_delwrite_queue;
+ spinlock_t bt_delwrite_lock;
+ unsigned long bt_flags;
} xfs_buftarg_t;
/*
- * xfs_buf_t: Buffer structure for page cache-based buffers
+ * xfs_buf_t: Buffer structure for pagecache-based buffers
+ *
+ * This buffer structure is used by the pagecache buffer management routines
+ * to refer to an assembly of pages forming a logical buffer.
*
- * This buffer structure is used by the page cache buffer management routines
- * to refer to an assembly of pages forming a logical buffer. The actual I/O
- * is performed with buffer_head structures, as required by drivers.
- *
- * The buffer structure is used on temporary basis only, and discarded when
- * released. The real data storage is recorded in the page cache. Metadata is
+ * The buffer structure is used on a temporary basis only, and discarded when
+ * released. The real data storage is recorded in the pagecache. Buffers are
* hashed to the block device on which the file system resides.
*/
struct xfs_buf;
+typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);
+typedef void (*xfs_buf_relse_t)(struct xfs_buf *);
+typedef int (*xfs_buf_bdstrat_t)(struct xfs_buf *);
-/* call-back function on I/O completion */
-typedef void (*page_buf_iodone_t)(struct xfs_buf *);
-/* call-back function on I/O completion */
-typedef void (*page_buf_relse_t)(struct xfs_buf *);
-/* pre-write function */
-typedef int (*page_buf_bdstrat_t)(struct xfs_buf *);
-
-#define PB_PAGES 2
+#define XB_PAGES 2
typedef struct xfs_buf {
- struct semaphore pb_sema; /* semaphore for lockables */
- unsigned long pb_queuetime; /* time buffer was queued */
- atomic_t pb_pin_count; /* pin count */
- wait_queue_head_t pb_waiters; /* unpin waiters */
- struct list_head pb_list;
- page_buf_flags_t pb_flags; /* status flags */
- struct list_head pb_hash_list; /* hash table list */
- xfs_bufhash_t *pb_hash; /* hash table list start */
- xfs_buftarg_t *pb_target; /* buffer target (device) */
- atomic_t pb_hold; /* reference count */
- xfs_daddr_t pb_bn; /* block number for I/O */
- loff_t pb_file_offset; /* offset in file */
- size_t pb_buffer_length; /* size of buffer in bytes */
- size_t pb_count_desired; /* desired transfer size */
- void *pb_addr; /* virtual address of buffer */
- struct work_struct pb_iodone_work;
- atomic_t pb_io_remaining;/* #outstanding I/O requests */
- page_buf_iodone_t pb_iodone; /* I/O completion function */
- page_buf_relse_t pb_relse; /* releasing function */
- page_buf_bdstrat_t pb_strat; /* pre-write function */
- struct semaphore pb_iodonesema; /* Semaphore for I/O waiters */
- void *pb_fspriv;
- void *pb_fspriv2;
- void *pb_fspriv3;
- unsigned short pb_error; /* error code on I/O */
- unsigned short pb_locked; /* page array is locked */
- unsigned int pb_page_count; /* size of page array */
- unsigned int pb_offset; /* page offset in first page */
- struct page **pb_pages; /* array of page pointers */
- struct page *pb_page_array[PB_PAGES]; /* inline pages */
-#ifdef PAGEBUF_LOCK_TRACKING
- int pb_last_holder;
+ struct semaphore b_sema; /* semaphore for lockables */
+ unsigned long b_queuetime; /* time buffer was queued */
+ atomic_t b_pin_count; /* pin count */
+ wait_queue_head_t b_waiters; /* unpin waiters */
+ struct list_head b_list;
+ xfs_buf_flags_t b_flags; /* status flags */
+ struct list_head b_hash_list; /* hash table list */
+ xfs_bufhash_t *b_hash; /* hash table list start */
+ xfs_buftarg_t *b_target; /* buffer target (device) */
+ atomic_t b_hold; /* reference count */
+ xfs_daddr_t b_bn; /* block number for I/O */
+ xfs_off_t b_file_offset; /* offset in file */
+ size_t b_buffer_length;/* size of buffer in bytes */
+ size_t b_count_desired;/* desired transfer size */
+ void *b_addr; /* virtual address of buffer */
+ struct work_struct b_iodone_work;
+ atomic_t b_io_remaining; /* #outstanding I/O requests */
+ xfs_buf_iodone_t b_iodone; /* I/O completion function */
+ xfs_buf_relse_t b_relse; /* releasing function */
+ xfs_buf_bdstrat_t b_strat; /* pre-write function */
+ struct semaphore b_iodonesema; /* Semaphore for I/O waiters */
+ void *b_fspriv;
+ void *b_fspriv2;
+ void *b_fspriv3;
+ unsigned short b_error; /* error code on I/O */
+ unsigned short b_locked; /* page array is locked */
+ unsigned int b_page_count; /* size of page array */
+ unsigned int b_offset; /* page offset in first page */
+ struct page **b_pages; /* array of page pointers */
+ struct page *b_page_array[XB_PAGES]; /* inline pages */
+#ifdef XFS_BUF_LOCK_TRACKING
+ int b_last_holder;
#endif
} xfs_buf_t;
/* Finding and Reading Buffers */
-
-extern xfs_buf_t *_pagebuf_find( /* find buffer for block if */
- /* the block is in memory */
- xfs_buftarg_t *, /* inode for block */
- loff_t, /* starting offset of range */
- size_t, /* length of range */
- page_buf_flags_t, /* PBF_LOCK */
- xfs_buf_t *); /* newly allocated buffer */
-
+extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t,
+ xfs_buf_flags_t, xfs_buf_t *);
#define xfs_incore(buftarg,blkno,len,lockit) \
- _pagebuf_find(buftarg, blkno ,len, lockit, NULL)
-
-extern xfs_buf_t *xfs_buf_get_flags( /* allocate a buffer */
- xfs_buftarg_t *, /* inode for buffer */
- loff_t, /* starting offset of range */
- size_t, /* length of range */
- page_buf_flags_t); /* PBF_LOCK, PBF_READ, */
- /* PBF_ASYNC */
+ _xfs_buf_find(buftarg, blkno ,len, lockit, NULL)
+extern xfs_buf_t *xfs_buf_get_flags(xfs_buftarg_t *, xfs_off_t, size_t,
+ xfs_buf_flags_t);
#define xfs_buf_get(target, blkno, len, flags) \
- xfs_buf_get_flags((target), (blkno), (len), PBF_LOCK | PBF_MAPPED)
-
-extern xfs_buf_t *xfs_buf_read_flags( /* allocate and read a buffer */
- xfs_buftarg_t *, /* inode for buffer */
- loff_t, /* starting offset of range */
- size_t, /* length of range */
- page_buf_flags_t); /* PBF_LOCK, PBF_ASYNC */
+ xfs_buf_get_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
+extern xfs_buf_t *xfs_buf_read_flags(xfs_buftarg_t *, xfs_off_t, size_t,
+ xfs_buf_flags_t);
#define xfs_buf_read(target, blkno, len, flags) \
- xfs_buf_read_flags((target), (blkno), (len), PBF_LOCK | PBF_MAPPED)
-
-extern xfs_buf_t *pagebuf_get_empty( /* allocate pagebuf struct with */
- /* no memory or disk address */
- size_t len,
- xfs_buftarg_t *); /* mount point "fake" inode */
-
-extern xfs_buf_t *pagebuf_get_no_daddr(/* allocate pagebuf struct */
- /* without disk address */
- size_t len,
- xfs_buftarg_t *); /* mount point "fake" inode */
-
-extern int pagebuf_associate_memory(
- xfs_buf_t *,
- void *,
- size_t);
-
-extern void pagebuf_hold( /* increment reference count */
- xfs_buf_t *); /* buffer to hold */
+ xfs_buf_read_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
-extern void pagebuf_readahead( /* read ahead into cache */
- xfs_buftarg_t *, /* target for buffer (or NULL) */
- loff_t, /* starting offset of range */
- size_t, /* length of range */
- page_buf_flags_t); /* additional read flags */
+extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
+extern xfs_buf_t *xfs_buf_get_noaddr(size_t, xfs_buftarg_t *);
+extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
+extern void xfs_buf_hold(xfs_buf_t *);
+extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t,
+ xfs_buf_flags_t);
/* Releasing Buffers */
-
-extern void pagebuf_free( /* deallocate a buffer */
- xfs_buf_t *); /* buffer to deallocate */
-
-extern void pagebuf_rele( /* release hold on a buffer */
- xfs_buf_t *); /* buffer to release */
+extern void xfs_buf_free(xfs_buf_t *);
+extern void xfs_buf_rele(xfs_buf_t *);
/* Locking and Unlocking Buffers */
-
-extern int pagebuf_cond_lock( /* lock buffer, if not locked */
- /* (returns -EBUSY if locked) */
- xfs_buf_t *); /* buffer to lock */
-
-extern int pagebuf_lock_value( /* return count on lock */
- xfs_buf_t *); /* buffer to check */
-
-extern int pagebuf_lock( /* lock buffer */
- xfs_buf_t *); /* buffer to lock */
-
-extern void pagebuf_unlock( /* unlock buffer */
- xfs_buf_t *); /* buffer to unlock */
+extern int xfs_buf_cond_lock(xfs_buf_t *);
+extern int xfs_buf_lock_value(xfs_buf_t *);
+extern void xfs_buf_lock(xfs_buf_t *);
+extern void xfs_buf_unlock(xfs_buf_t *);
/* Buffer Read and Write Routines */
-
-extern void pagebuf_iodone( /* mark buffer I/O complete */
- xfs_buf_t *, /* buffer to mark */
- int); /* run completion locally, or in
- * a helper thread. */
-
-extern void pagebuf_ioerror( /* mark buffer in error (or not) */
- xfs_buf_t *, /* buffer to mark */
- int); /* error to store (0 if none) */
-
-extern int pagebuf_iostart( /* start I/O on a buffer */
- xfs_buf_t *, /* buffer to start */
- page_buf_flags_t); /* PBF_LOCK, PBF_ASYNC, */
- /* PBF_READ, PBF_WRITE, */
- /* PBF_DELWRI */
-
-extern int pagebuf_iorequest( /* start real I/O */
- xfs_buf_t *); /* buffer to convey to device */
-
-extern int pagebuf_iowait( /* wait for buffer I/O done */
- xfs_buf_t *); /* buffer to wait on */
-
-extern void pagebuf_iomove( /* move data in/out of pagebuf */
- xfs_buf_t *, /* buffer to manipulate */
- size_t, /* starting buffer offset */
- size_t, /* length in buffer */
- caddr_t, /* data pointer */
- page_buf_rw_t); /* direction */
-
-static inline int pagebuf_iostrategy(xfs_buf_t *pb)
+extern void xfs_buf_ioend(xfs_buf_t *, int);
+extern void xfs_buf_ioerror(xfs_buf_t *, int);
+extern int xfs_buf_iostart(xfs_buf_t *, xfs_buf_flags_t);
+extern int xfs_buf_iorequest(xfs_buf_t *);
+extern int xfs_buf_iowait(xfs_buf_t *);
+extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, xfs_caddr_t,
+ xfs_buf_rw_t);
+
+static inline int xfs_buf_iostrategy(xfs_buf_t *bp)
{
- return pb->pb_strat ? pb->pb_strat(pb) : pagebuf_iorequest(pb);
+ return bp->b_strat ? bp->b_strat(bp) : xfs_buf_iorequest(bp);
}
-static inline int pagebuf_geterror(xfs_buf_t *pb)
+static inline int xfs_buf_geterror(xfs_buf_t *bp)
{
- return pb ? pb->pb_error : ENOMEM;
+ return bp ? bp->b_error : ENOMEM;
}
/* Buffer Utility Routines */
-
-extern caddr_t pagebuf_offset( /* pointer at offset in buffer */
- xfs_buf_t *, /* buffer to offset into */
- size_t); /* offset */
+extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t);
/* Pinning Buffer Storage in Memory */
-
-extern void pagebuf_pin( /* pin buffer in memory */
- xfs_buf_t *); /* buffer to pin */
-
-extern void pagebuf_unpin( /* unpin buffered data */
- xfs_buf_t *); /* buffer to unpin */
-
-extern int pagebuf_ispin( /* check if buffer is pinned */
- xfs_buf_t *); /* buffer to check */
+extern void xfs_buf_pin(xfs_buf_t *);
+extern void xfs_buf_unpin(xfs_buf_t *);
+extern int xfs_buf_ispin(xfs_buf_t *);
/* Delayed Write Buffer Routines */
-
-extern void pagebuf_delwri_dequeue(xfs_buf_t *);
+extern void xfs_buf_delwri_dequeue(xfs_buf_t *);
/* Buffer Daemon Setup Routines */
+extern int xfs_buf_init(void);
+extern void xfs_buf_terminate(void);
-extern int pagebuf_init(void);
-extern void pagebuf_terminate(void);
-
-
-#ifdef PAGEBUF_TRACE
-extern ktrace_t *pagebuf_trace_buf;
-extern void pagebuf_trace(
- xfs_buf_t *, /* buffer being traced */
- char *, /* description of operation */
- void *, /* arbitrary diagnostic value */
- void *); /* return address */
+#ifdef XFS_BUF_TRACE
+extern ktrace_t *xfs_buf_trace_buf;
+extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
#else
-# define pagebuf_trace(pb, id, ptr, ra) do { } while (0)
+#define xfs_buf_trace(bp,id,ptr,ra) do { } while (0)
#endif
-#define pagebuf_target_name(target) \
- ({ char __b[BDEVNAME_SIZE]; bdevname((target)->pbr_bdev, __b); __b; })
+#define xfs_buf_target_name(target) \
+ ({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; })
+#define XFS_B_ASYNC XBF_ASYNC
+#define XFS_B_DELWRI XBF_DELWRI
+#define XFS_B_READ XBF_READ
+#define XFS_B_WRITE XBF_WRITE
+#define XFS_B_STALE XBF_STALE
-/* These are just for xfs_syncsub... it sets an internal variable
- * then passes it to VOP_FLUSH_PAGES or adds the flags to a newly gotten buf_t
- */
-#define XFS_B_ASYNC PBF_ASYNC
-#define XFS_B_DELWRI PBF_DELWRI
-#define XFS_B_READ PBF_READ
-#define XFS_B_WRITE PBF_WRITE
-#define XFS_B_STALE PBF_STALE
-
-#define XFS_BUF_TRYLOCK PBF_TRYLOCK
-#define XFS_INCORE_TRYLOCK PBF_TRYLOCK
-#define XFS_BUF_LOCK PBF_LOCK
-#define XFS_BUF_MAPPED PBF_MAPPED
-
-#define BUF_BUSY PBF_DONT_BLOCK
-
-#define XFS_BUF_BFLAGS(x) ((x)->pb_flags)
-#define XFS_BUF_ZEROFLAGS(x) \
- ((x)->pb_flags &= ~(PBF_READ|PBF_WRITE|PBF_ASYNC|PBF_DELWRI))
-
-#define XFS_BUF_STALE(x) ((x)->pb_flags |= XFS_B_STALE)
-#define XFS_BUF_UNSTALE(x) ((x)->pb_flags &= ~XFS_B_STALE)
-#define XFS_BUF_ISSTALE(x) ((x)->pb_flags & XFS_B_STALE)
-#define XFS_BUF_SUPER_STALE(x) do { \
- XFS_BUF_STALE(x); \
- pagebuf_delwri_dequeue(x); \
- XFS_BUF_DONE(x); \
- } while (0)
+#define XFS_BUF_TRYLOCK XBF_TRYLOCK
+#define XFS_INCORE_TRYLOCK XBF_TRYLOCK
+#define XFS_BUF_LOCK XBF_LOCK
+#define XFS_BUF_MAPPED XBF_MAPPED
-#define XFS_BUF_MANAGE PBF_FS_MANAGED
-#define XFS_BUF_UNMANAGE(x) ((x)->pb_flags &= ~PBF_FS_MANAGED)
-
-#define XFS_BUF_DELAYWRITE(x) ((x)->pb_flags |= PBF_DELWRI)
-#define XFS_BUF_UNDELAYWRITE(x) pagebuf_delwri_dequeue(x)
-#define XFS_BUF_ISDELAYWRITE(x) ((x)->pb_flags & PBF_DELWRI)
-
-#define XFS_BUF_ERROR(x,no) pagebuf_ioerror(x,no)
-#define XFS_BUF_GETERROR(x) pagebuf_geterror(x)
-#define XFS_BUF_ISERROR(x) (pagebuf_geterror(x)?1:0)
-
-#define XFS_BUF_DONE(x) ((x)->pb_flags |= PBF_DONE)
-#define XFS_BUF_UNDONE(x) ((x)->pb_flags &= ~PBF_DONE)
-#define XFS_BUF_ISDONE(x) ((x)->pb_flags & PBF_DONE)
-
-#define XFS_BUF_BUSY(x) do { } while (0)
-#define XFS_BUF_UNBUSY(x) do { } while (0)
-#define XFS_BUF_ISBUSY(x) (1)
-
-#define XFS_BUF_ASYNC(x) ((x)->pb_flags |= PBF_ASYNC)
-#define XFS_BUF_UNASYNC(x) ((x)->pb_flags &= ~PBF_ASYNC)
-#define XFS_BUF_ISASYNC(x) ((x)->pb_flags & PBF_ASYNC)
-
-#define XFS_BUF_ORDERED(x) ((x)->pb_flags |= PBF_ORDERED)
-#define XFS_BUF_UNORDERED(x) ((x)->pb_flags &= ~PBF_ORDERED)
-#define XFS_BUF_ISORDERED(x) ((x)->pb_flags & PBF_ORDERED)
-
-#define XFS_BUF_SHUT(x) printk("XFS_BUF_SHUT not implemented yet\n")
-#define XFS_BUF_UNSHUT(x) printk("XFS_BUF_UNSHUT not implemented yet\n")
-#define XFS_BUF_ISSHUT(x) (0)
-
-#define XFS_BUF_HOLD(x) pagebuf_hold(x)
-#define XFS_BUF_READ(x) ((x)->pb_flags |= PBF_READ)
-#define XFS_BUF_UNREAD(x) ((x)->pb_flags &= ~PBF_READ)
-#define XFS_BUF_ISREAD(x) ((x)->pb_flags & PBF_READ)
-
-#define XFS_BUF_WRITE(x) ((x)->pb_flags |= PBF_WRITE)
-#define XFS_BUF_UNWRITE(x) ((x)->pb_flags &= ~PBF_WRITE)
-#define XFS_BUF_ISWRITE(x) ((x)->pb_flags & PBF_WRITE)
-
-#define XFS_BUF_ISUNINITIAL(x) (0)
-#define XFS_BUF_UNUNINITIAL(x) (0)
-
-#define XFS_BUF_BP_ISMAPPED(bp) 1
-
-#define XFS_BUF_IODONE_FUNC(buf) (buf)->pb_iodone
-#define XFS_BUF_SET_IODONE_FUNC(buf, func) \
- (buf)->pb_iodone = (func)
-#define XFS_BUF_CLR_IODONE_FUNC(buf) \
- (buf)->pb_iodone = NULL
-#define XFS_BUF_SET_BDSTRAT_FUNC(buf, func) \
- (buf)->pb_strat = (func)
-#define XFS_BUF_CLR_BDSTRAT_FUNC(buf) \
- (buf)->pb_strat = NULL
-
-#define XFS_BUF_FSPRIVATE(buf, type) \
- ((type)(buf)->pb_fspriv)
-#define XFS_BUF_SET_FSPRIVATE(buf, value) \
- (buf)->pb_fspriv = (void *)(value)
-#define XFS_BUF_FSPRIVATE2(buf, type) \
- ((type)(buf)->pb_fspriv2)
-#define XFS_BUF_SET_FSPRIVATE2(buf, value) \
- (buf)->pb_fspriv2 = (void *)(value)
-#define XFS_BUF_FSPRIVATE3(buf, type) \
- ((type)(buf)->pb_fspriv3)
-#define XFS_BUF_SET_FSPRIVATE3(buf, value) \
- (buf)->pb_fspriv3 = (void *)(value)
-#define XFS_BUF_SET_START(buf)
-
-#define XFS_BUF_SET_BRELSE_FUNC(buf, value) \
- (buf)->pb_relse = (value)
-
-#define XFS_BUF_PTR(bp) (xfs_caddr_t)((bp)->pb_addr)
-
-static inline xfs_caddr_t xfs_buf_offset(xfs_buf_t *bp, size_t offset)
-{
- if (bp->pb_flags & PBF_MAPPED)
- return XFS_BUF_PTR(bp) + offset;
- return (xfs_caddr_t) pagebuf_offset(bp, offset);
-}
+#define BUF_BUSY XBF_DONT_BLOCK
+
+#define XFS_BUF_BFLAGS(bp) ((bp)->b_flags)
+#define XFS_BUF_ZEROFLAGS(bp) \
+ ((bp)->b_flags &= ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI))
+
+#define XFS_BUF_STALE(bp) ((bp)->b_flags |= XFS_B_STALE)
+#define XFS_BUF_UNSTALE(bp) ((bp)->b_flags &= ~XFS_B_STALE)
+#define XFS_BUF_ISSTALE(bp) ((bp)->b_flags & XFS_B_STALE)
+#define XFS_BUF_SUPER_STALE(bp) do { \
+ XFS_BUF_STALE(bp); \
+ xfs_buf_delwri_dequeue(bp); \
+ XFS_BUF_DONE(bp); \
+ } while (0)
-#define XFS_BUF_SET_PTR(bp, val, count) \
- pagebuf_associate_memory(bp, val, count)
-#define XFS_BUF_ADDR(bp) ((bp)->pb_bn)
-#define XFS_BUF_SET_ADDR(bp, blk) \
- ((bp)->pb_bn = (xfs_daddr_t)(blk))
-#define XFS_BUF_OFFSET(bp) ((bp)->pb_file_offset)
-#define XFS_BUF_SET_OFFSET(bp, off) \
- ((bp)->pb_file_offset = (off))
-#define XFS_BUF_COUNT(bp) ((bp)->pb_count_desired)
-#define XFS_BUF_SET_COUNT(bp, cnt) \
- ((bp)->pb_count_desired = (cnt))
-#define XFS_BUF_SIZE(bp) ((bp)->pb_buffer_length)
-#define XFS_BUF_SET_SIZE(bp, cnt) \
- ((bp)->pb_buffer_length = (cnt))
-#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)
-#define XFS_BUF_SET_VTYPE(bp, type)
-#define XFS_BUF_SET_REF(bp, ref)
-
-#define XFS_BUF_ISPINNED(bp) pagebuf_ispin(bp)
-
-#define XFS_BUF_VALUSEMA(bp) pagebuf_lock_value(bp)
-#define XFS_BUF_CPSEMA(bp) (pagebuf_cond_lock(bp) == 0)
-#define XFS_BUF_VSEMA(bp) pagebuf_unlock(bp)
-#define XFS_BUF_PSEMA(bp,x) pagebuf_lock(bp)
-#define XFS_BUF_V_IODONESEMA(bp) up(&bp->pb_iodonesema);
-
-/* setup the buffer target from a buftarg structure */
-#define XFS_BUF_SET_TARGET(bp, target) \
- (bp)->pb_target = (target)
-#define XFS_BUF_TARGET(bp) ((bp)->pb_target)
-#define XFS_BUFTARG_NAME(target) \
- pagebuf_target_name(target)
-
-#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)
-#define XFS_BUF_SET_VTYPE(bp, type)
-#define XFS_BUF_SET_REF(bp, ref)
-
-static inline int xfs_bawrite(void *mp, xfs_buf_t *bp)
+#define XFS_BUF_MANAGE XBF_FS_MANAGED
+#define XFS_BUF_UNMANAGE(bp) ((bp)->b_flags &= ~XBF_FS_MANAGED)
+
+#define XFS_BUF_DELAYWRITE(bp) ((bp)->b_flags |= XBF_DELWRI)
+#define XFS_BUF_UNDELAYWRITE(bp) xfs_buf_delwri_dequeue(bp)
+#define XFS_BUF_ISDELAYWRITE(bp) ((bp)->b_flags & XBF_DELWRI)
+
+#define XFS_BUF_ERROR(bp,no) xfs_buf_ioerror(bp,no)
+#define XFS_BUF_GETERROR(bp) xfs_buf_geterror(bp)
+#define XFS_BUF_ISERROR(bp) (xfs_buf_geterror(bp) ? 1 : 0)
+
+#define XFS_BUF_DONE(bp) ((bp)->b_flags |= XBF_DONE)
+#define XFS_BUF_UNDONE(bp) ((bp)->b_flags &= ~XBF_DONE)
+#define XFS_BUF_ISDONE(bp) ((bp)->b_flags & XBF_DONE)
+
+#define XFS_BUF_BUSY(bp) do { } while (0)
+#define XFS_BUF_UNBUSY(bp) do { } while (0)
+#define XFS_BUF_ISBUSY(bp) (1)
+
+#define XFS_BUF_ASYNC(bp) ((bp)->b_flags |= XBF_ASYNC)
+#define XFS_BUF_UNASYNC(bp) ((bp)->b_flags &= ~XBF_ASYNC)
+#define XFS_BUF_ISASYNC(bp) ((bp)->b_flags & XBF_ASYNC)
+
+#define XFS_BUF_ORDERED(bp) ((bp)->b_flags |= XBF_ORDERED)
+#define XFS_BUF_UNORDERED(bp) ((bp)->b_flags &= ~XBF_ORDERED)
+#define XFS_BUF_ISORDERED(bp) ((bp)->b_flags & XBF_ORDERED)
+
+#define XFS_BUF_SHUT(bp) do { } while (0)
+#define XFS_BUF_UNSHUT(bp) do { } while (0)
+#define XFS_BUF_ISSHUT(bp) (0)
+
+#define XFS_BUF_HOLD(bp) xfs_buf_hold(bp)
+#define XFS_BUF_READ(bp) ((bp)->b_flags |= XBF_READ)
+#define XFS_BUF_UNREAD(bp) ((bp)->b_flags &= ~XBF_READ)
+#define XFS_BUF_ISREAD(bp) ((bp)->b_flags & XBF_READ)
+
+#define XFS_BUF_WRITE(bp) ((bp)->b_flags |= XBF_WRITE)
+#define XFS_BUF_UNWRITE(bp) ((bp)->b_flags &= ~XBF_WRITE)
+#define XFS_BUF_ISWRITE(bp) ((bp)->b_flags & XBF_WRITE)
+
+#define XFS_BUF_ISUNINITIAL(bp) (0)
+#define XFS_BUF_UNUNINITIAL(bp) (0)
+
+#define XFS_BUF_BP_ISMAPPED(bp) (1)
+
+#define XFS_BUF_IODONE_FUNC(bp) ((bp)->b_iodone)
+#define XFS_BUF_SET_IODONE_FUNC(bp, func) ((bp)->b_iodone = (func))
+#define XFS_BUF_CLR_IODONE_FUNC(bp) ((bp)->b_iodone = NULL)
+#define XFS_BUF_SET_BDSTRAT_FUNC(bp, func) ((bp)->b_strat = (func))
+#define XFS_BUF_CLR_BDSTRAT_FUNC(bp) ((bp)->b_strat = NULL)
+
+#define XFS_BUF_FSPRIVATE(bp, type) ((type)(bp)->b_fspriv)
+#define XFS_BUF_SET_FSPRIVATE(bp, val) ((bp)->b_fspriv = (void*)(val))
+#define XFS_BUF_FSPRIVATE2(bp, type) ((type)(bp)->b_fspriv2)
+#define XFS_BUF_SET_FSPRIVATE2(bp, val) ((bp)->b_fspriv2 = (void*)(val))
+#define XFS_BUF_FSPRIVATE3(bp, type) ((type)(bp)->b_fspriv3)
+#define XFS_BUF_SET_FSPRIVATE3(bp, val) ((bp)->b_fspriv3 = (void*)(val))
+#define XFS_BUF_SET_START(bp) do { } while (0)
+#define XFS_BUF_SET_BRELSE_FUNC(bp, func) ((bp)->b_relse = (func))
+
+#define XFS_BUF_PTR(bp) (xfs_caddr_t)((bp)->b_addr)
+#define XFS_BUF_SET_PTR(bp, val, cnt) xfs_buf_associate_memory(bp, val, cnt)
+#define XFS_BUF_ADDR(bp) ((bp)->b_bn)
+#define XFS_BUF_SET_ADDR(bp, bno) ((bp)->b_bn = (xfs_daddr_t)(bno))
+#define XFS_BUF_OFFSET(bp) ((bp)->b_file_offset)
+#define XFS_BUF_SET_OFFSET(bp, off) ((bp)->b_file_offset = (off))
+#define XFS_BUF_COUNT(bp) ((bp)->b_count_desired)
+#define XFS_BUF_SET_COUNT(bp, cnt) ((bp)->b_count_desired = (cnt))
+#define XFS_BUF_SIZE(bp) ((bp)->b_buffer_length)
+#define XFS_BUF_SET_SIZE(bp, cnt) ((bp)->b_buffer_length = (cnt))
+
+#define XFS_BUF_SET_VTYPE_REF(bp, type, ref) do { } while (0)
+#define XFS_BUF_SET_VTYPE(bp, type) do { } while (0)
+#define XFS_BUF_SET_REF(bp, ref) do { } while (0)
+
+#define XFS_BUF_ISPINNED(bp) xfs_buf_ispin(bp)
+
+#define XFS_BUF_VALUSEMA(bp) xfs_buf_lock_value(bp)
+#define XFS_BUF_CPSEMA(bp) (xfs_buf_cond_lock(bp) == 0)
+#define XFS_BUF_VSEMA(bp) xfs_buf_unlock(bp)
+#define XFS_BUF_PSEMA(bp,x) xfs_buf_lock(bp)
+#define XFS_BUF_V_IODONESEMA(bp) up(&bp->b_iodonesema);
+
+#define XFS_BUF_SET_TARGET(bp, target) ((bp)->b_target = (target))
+#define XFS_BUF_TARGET(bp) ((bp)->b_target)
+#define XFS_BUFTARG_NAME(target) xfs_buf_target_name(target)
+
+static inline int xfs_bawrite(void *mp, xfs_buf_t *bp)
{
- bp->pb_fspriv3 = mp;
- bp->pb_strat = xfs_bdstrat_cb;
- pagebuf_delwri_dequeue(bp);
- return pagebuf_iostart(bp, PBF_WRITE | PBF_ASYNC | _PBF_RUN_QUEUES);
+ bp->b_fspriv3 = mp;
+ bp->b_strat = xfs_bdstrat_cb;
+ xfs_buf_delwri_dequeue(bp);
+ return xfs_buf_iostart(bp, XBF_WRITE | XBF_ASYNC | _XBF_RUN_QUEUES);
}
-static inline void xfs_buf_relse(xfs_buf_t *bp)
+static inline void xfs_buf_relse(xfs_buf_t *bp)
{
- if (!bp->pb_relse)
- pagebuf_unlock(bp);
- pagebuf_rele(bp);
+ if (!bp->b_relse)
+ xfs_buf_unlock(bp);
+ xfs_buf_rele(bp);
}
-#define xfs_bpin(bp) pagebuf_pin(bp)
-#define xfs_bunpin(bp) pagebuf_unpin(bp)
+#define xfs_bpin(bp) xfs_buf_pin(bp)
+#define xfs_bunpin(bp) xfs_buf_unpin(bp)
#define xfs_buftrace(id, bp) \
- pagebuf_trace(bp, id, NULL, (void *)__builtin_return_address(0))
+ xfs_buf_trace(bp, id, NULL, (void *)__builtin_return_address(0))
-#define xfs_biodone(pb) \
- pagebuf_iodone(pb, 0)
+#define xfs_biodone(bp) xfs_buf_ioend(bp, 0)
-#define xfs_biomove(pb, off, len, data, rw) \
- pagebuf_iomove((pb), (off), (len), (data), \
- ((rw) == XFS_B_WRITE) ? PBRW_WRITE : PBRW_READ)
+#define xfs_biomove(bp, off, len, data, rw) \
+ xfs_buf_iomove((bp), (off), (len), (data), \
+ ((rw) == XFS_B_WRITE) ? XBRW_WRITE : XBRW_READ)
-#define xfs_biozero(pb, off, len) \
- pagebuf_iomove((pb), (off), (len), NULL, PBRW_ZERO)
+#define xfs_biozero(bp, off, len) \
+ xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO)
-static inline int XFS_bwrite(xfs_buf_t *pb)
+static inline int XFS_bwrite(xfs_buf_t *bp)
{
- int iowait = (pb->pb_flags & PBF_ASYNC) == 0;
+ int iowait = (bp->b_flags & XBF_ASYNC) == 0;
int error = 0;
if (!iowait)
- pb->pb_flags |= _PBF_RUN_QUEUES;
+ bp->b_flags |= _XBF_RUN_QUEUES;
- pagebuf_delwri_dequeue(pb);
- pagebuf_iostrategy(pb);
+ xfs_buf_delwri_dequeue(bp);
+ xfs_buf_iostrategy(bp);
if (iowait) {
- error = pagebuf_iowait(pb);
- xfs_buf_relse(pb);
+ error = xfs_buf_iowait(bp);
+ xfs_buf_relse(bp);
}
return error;
}
-#define XFS_bdwrite(pb) \
- pagebuf_iostart(pb, PBF_DELWRI | PBF_ASYNC)
+#define XFS_bdwrite(bp) xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC)
static inline int xfs_bdwrite(void *mp, xfs_buf_t *bp)
{
- bp->pb_strat = xfs_bdstrat_cb;
- bp->pb_fspriv3 = mp;
-
- return pagebuf_iostart(bp, PBF_DELWRI | PBF_ASYNC);
+ bp->b_strat = xfs_bdstrat_cb;
+ bp->b_fspriv3 = mp;
+ return xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC);
}
-#define XFS_bdstrat(bp) pagebuf_iorequest(bp)
+#define XFS_bdstrat(bp) xfs_buf_iorequest(bp)
-#define xfs_iowait(pb) pagebuf_iowait(pb)
+#define xfs_iowait(bp) xfs_buf_iowait(bp)
#define xfs_baread(target, rablkno, ralen) \
- pagebuf_readahead((target), (rablkno), (ralen), PBF_DONT_BLOCK)
-
-#define xfs_buf_get_empty(len, target) pagebuf_get_empty((len), (target))
-#define xfs_buf_get_noaddr(len, target) pagebuf_get_no_daddr((len), (target))
-#define xfs_buf_free(bp) pagebuf_free(bp)
+ xfs_buf_readahead((target), (rablkno), (ralen), XBF_DONT_BLOCK)
/*
* Handling of buftargs.
*/
-
extern xfs_buftarg_t *xfs_alloc_buftarg(struct block_device *, int);
extern void xfs_free_buftarg(xfs_buftarg_t *, int);
extern void xfs_wait_buftarg(xfs_buftarg_t *);
extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int);
extern int xfs_flush_buftarg(xfs_buftarg_t *, int);
-#define xfs_getsize_buftarg(buftarg) \
- block_size((buftarg)->pbr_bdev)
-#define xfs_readonly_buftarg(buftarg) \
- bdev_read_only((buftarg)->pbr_bdev)
-#define xfs_binval(buftarg) \
- xfs_flush_buftarg(buftarg, 1)
-#define XFS_bflush(buftarg) \
- xfs_flush_buftarg(buftarg, 1)
+#define xfs_getsize_buftarg(buftarg) block_size((buftarg)->bt_bdev)
+#define xfs_readonly_buftarg(buftarg) bdev_read_only((buftarg)->bt_bdev)
+
+#define xfs_binval(buftarg) xfs_flush_buftarg(buftarg, 1)
+#define XFS_bflush(buftarg) xfs_flush_buftarg(buftarg, 1)
#endif /* __XFS_BUF_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_cred.h b/fs/xfs/linux-2.6/xfs_cred.h
index 4af49102472..e7f3da61c6c 100644
--- a/fs/xfs/linux-2.6/xfs_cred.h
+++ b/fs/xfs/linux-2.6/xfs_cred.h
@@ -18,6 +18,8 @@
#ifndef __XFS_CRED_H__
#define __XFS_CRED_H__
+#include <linux/capability.h>
+
/*
* Credentials
*/
@@ -27,7 +29,7 @@ typedef struct cred {
extern struct cred *sys_cred;
-/* this is a hack.. (assums sys_cred is the only cred_t in the system) */
+/* this is a hack.. (assumes sys_cred is the only cred_t in the system) */
static __inline int capable_cred(cred_t *cr, int cid)
{
return (cr == sys_cred) ? 1 : capable(cid);
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 06111d0bbae..ced4404339c 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -509,16 +509,14 @@ linvfs_open_exec(
vnode_t *vp = LINVFS_GET_VP(inode);
xfs_mount_t *mp = XFS_VFSTOM(vp->v_vfsp);
int error = 0;
- bhv_desc_t *bdp;
xfs_inode_t *ip;
if (vp->v_vfsp->vfs_flag & VFS_DMI) {
- bdp = vn_bhv_lookup(VN_BHV_HEAD(vp), &xfs_vnodeops);
- if (!bdp) {
+ ip = xfs_vtoi(vp);
+ if (!ip) {
error = -EINVAL;
goto open_exec_out;
}
- ip = XFS_BHVTOI(bdp);
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ)) {
error = -XFS_SEND_DATA(mp, DM_EVENT_READ, vp,
0, 0, 0, NULL);
diff --git a/fs/xfs/linux-2.6/xfs_fs_subr.c b/fs/xfs/linux-2.6/xfs_fs_subr.c
index f89340c61bf..4fa4b1a5187 100644
--- a/fs/xfs/linux-2.6/xfs_fs_subr.c
+++ b/fs/xfs/linux-2.6/xfs_fs_subr.c
@@ -79,8 +79,7 @@ fs_flushinval_pages(
struct inode *ip = LINVFS_GET_IP(vp);
if (VN_CACHED(vp)) {
- filemap_fdatawrite(ip->i_mapping);
- filemap_fdatawait(ip->i_mapping);
+ filemap_write_and_wait(ip->i_mapping);
truncate_inode_pages(ip->i_mapping, first);
}
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index b78b5eb9e96..4db47790415 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -52,6 +52,7 @@
#include "xfs_dfrag.h"
#include "xfs_fsops.h"
+#include <linux/capability.h>
#include <linux/dcache.h>
#include <linux/mount.h>
#include <linux/namei.h>
@@ -145,13 +146,10 @@ xfs_find_handle(
if (cmd != XFS_IOC_PATH_TO_FSHANDLE) {
xfs_inode_t *ip;
- bhv_desc_t *bhv;
int lock_mode;
/* need to get access to the xfs_inode to read the generation */
- bhv = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
- ASSERT(bhv);
- ip = XFS_BHVTOI(bhv);
+ ip = xfs_vtoi(vp);
ASSERT(ip);
lock_mode = xfs_ilock_map_shared(ip);
@@ -530,6 +528,8 @@ xfs_attrmulti_attr_set(
char *kbuf;
int error = EFAULT;
+ if (IS_RDONLY(&vp->v_inode))
+ return -EROFS;
if (IS_IMMUTABLE(&vp->v_inode) || IS_APPEND(&vp->v_inode))
return EPERM;
if (len > XATTR_SIZE_MAX)
@@ -557,6 +557,9 @@ xfs_attrmulti_attr_remove(
{
int error;
+
+ if (IS_RDONLY(&vp->v_inode))
+ return -EROFS;
if (IS_IMMUTABLE(&vp->v_inode) || IS_APPEND(&vp->v_inode))
return EPERM;
@@ -745,9 +748,8 @@ xfs_ioctl(
(ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
mp->m_rtdev_targp : mp->m_ddev_targp;
- da.d_mem = da.d_miniosz = 1 << target->pbr_sshift;
- /* The size dio will do in one go */
- da.d_maxiosz = 64 * PAGE_CACHE_SIZE;
+ da.d_mem = da.d_miniosz = 1 << target->bt_sshift;
+ da.d_maxiosz = INT_MAX & ~(da.d_miniosz - 1);
if (copy_to_user(arg, &da, sizeof(da)))
return -XFS_ERROR(EFAULT);
diff --git a/fs/xfs/linux-2.6/xfs_ioctl32.c b/fs/xfs/linux-2.6/xfs_ioctl32.c
index c83ae15bb0e..a7c9ba1a9f7 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl32.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl32.c
@@ -19,7 +19,6 @@
#include <linux/compat.h>
#include <linux/init.h>
#include <linux/ioctl.h>
-#include <linux/ioctl32.h>
#include <linux/syscalls.h>
#include <linux/types.h>
#include <linux/fs.h>
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index 14215a7db59..76c6df34d0d 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -51,8 +51,44 @@
#include "xfs_buf_item.h"
#include "xfs_utils.h"
+#include <linux/capability.h>
#include <linux/xattr.h>
#include <linux/namei.h>
+#include <linux/security.h>
+
+/*
+ * Get a XFS inode from a given vnode.
+ */
+xfs_inode_t *
+xfs_vtoi(
+ struct vnode *vp)
+{
+ bhv_desc_t *bdp;
+
+ bdp = bhv_lookup_range(VN_BHV_HEAD(vp),
+ VNODE_POSITION_XFS, VNODE_POSITION_XFS);
+ if (unlikely(bdp == NULL))
+ return NULL;
+ return XFS_BHVTOI(bdp);
+}
+
+/*
+ * Bring the atime in the XFS inode uptodate.
+ * Used before logging the inode to disk or when the Linux inode goes away.
+ */
+void
+xfs_synchronize_atime(
+ xfs_inode_t *ip)
+{
+ vnode_t *vp;
+
+ vp = XFS_ITOV_NULL(ip);
+ if (vp) {
+ struct inode *inode = &vp->v_inode;
+ ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
+ ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
+ }
+}
/*
* Change the requested timestamp in the given inode.
@@ -73,23 +109,6 @@ xfs_ichgtime(
struct inode *inode = LINVFS_GET_IP(XFS_ITOV(ip));
timespec_t tv;
- /*
- * We're not supposed to change timestamps in readonly-mounted
- * filesystems. Throw it away if anyone asks us.
- */
- if (unlikely(IS_RDONLY(inode)))
- return;
-
- /*
- * Don't update access timestamps on reads if mounted "noatime".
- * Throw it away if anyone asks us.
- */
- if (unlikely(
- (ip->i_mount->m_flags & XFS_MOUNT_NOATIME || IS_NOATIME(inode)) &&
- (flags & (XFS_ICHGTIME_ACC|XFS_ICHGTIME_MOD|XFS_ICHGTIME_CHG)) ==
- XFS_ICHGTIME_ACC))
- return;
-
nanotime(&tv);
if (flags & XFS_ICHGTIME_MOD) {
inode->i_mtime = tv;
@@ -126,8 +145,6 @@ xfs_ichgtime(
* Variant on the above which avoids querying the system clock
* in situations where we know the Linux inode timestamps have
* just been updated (and so we can update our inode cheaply).
- * We also skip the readonly and noatime checks here, they are
- * also catered for already.
*/
void
xfs_ichgtime_fast(
@@ -138,20 +155,16 @@ xfs_ichgtime_fast(
timespec_t *tvp;
/*
- * We're not supposed to change timestamps in readonly-mounted
- * filesystems. Throw it away if anyone asks us.
+ * Atime updates for read() & friends are handled lazily now, and
+ * explicit updates must go through xfs_ichgtime()
*/
- if (unlikely(IS_RDONLY(inode)))
- return;
+ ASSERT((flags & XFS_ICHGTIME_ACC) == 0);
/*
- * Don't update access timestamps on reads if mounted "noatime".
- * Throw it away if anyone asks us.
+ * We're not supposed to change timestamps in readonly-mounted
+ * filesystems. Throw it away if anyone asks us.
*/
- if (unlikely(
- (ip->i_mount->m_flags & XFS_MOUNT_NOATIME || IS_NOATIME(inode)) &&
- ((flags & (XFS_ICHGTIME_ACC|XFS_ICHGTIME_MOD|XFS_ICHGTIME_CHG)) ==
- XFS_ICHGTIME_ACC)))
+ if (unlikely(IS_RDONLY(inode)))
return;
if (flags & XFS_ICHGTIME_MOD) {
@@ -159,11 +172,6 @@ xfs_ichgtime_fast(
ip->i_d.di_mtime.t_sec = (__int32_t)tvp->tv_sec;
ip->i_d.di_mtime.t_nsec = (__int32_t)tvp->tv_nsec;
}
- if (flags & XFS_ICHGTIME_ACC) {
- tvp = &inode->i_atime;
- ip->i_d.di_atime.t_sec = (__int32_t)tvp->tv_sec;
- ip->i_d.di_atime.t_nsec = (__int32_t)tvp->tv_nsec;
- }
if (flags & XFS_ICHGTIME_CHG) {
tvp = &inode->i_ctime;
ip->i_d.di_ctime.t_sec = (__int32_t)tvp->tv_sec;
@@ -203,13 +211,46 @@ validate_fields(
ip->i_nlink = va.va_nlink;
ip->i_blocks = va.va_nblocks;
- /* we're under i_sem so i_size can't change under us */
+ /* we're under i_mutex so i_size can't change under us */
if (i_size_read(ip) != va.va_size)
i_size_write(ip, va.va_size);
}
}
/*
+ * Hook in SELinux. This is not quite correct yet, what we really need
+ * here (as we do for default ACLs) is a mechanism by which creation of
+ * these attrs can be journalled at inode creation time (along with the
+ * inode, of course, such that log replay can't cause these to be lost).
+ */
+STATIC int
+linvfs_init_security(
+ struct vnode *vp,
+ struct inode *dir)
+{
+ struct inode *ip = LINVFS_GET_IP(vp);
+ size_t length;
+ void *value;
+ char *name;
+ int error;
+
+ error = security_inode_init_security(ip, dir, &name, &value, &length);
+ if (error) {
+ if (error == -EOPNOTSUPP)
+ return 0;
+ return -error;
+ }
+
+ VOP_ATTR_SET(vp, name, value, length, ATTR_SECURE, NULL, error);
+ if (!error)
+ VMODIFY(vp);
+
+ kfree(name);
+ kfree(value);
+ return error;
+}
+
+/*
* Determine whether a process has a valid fs_struct (kernel daemons
* like knfsd don't have an fs_struct).
*
@@ -274,6 +315,9 @@ linvfs_mknod(
break;
}
+ if (!error)
+ error = linvfs_init_security(vp, dir);
+
if (default_acl) {
if (!error) {
error = _ACL_INHERIT(vp, &va, default_acl);
@@ -290,8 +334,6 @@ linvfs_mknod(
teardown.d_inode = ip = LINVFS_GET_IP(vp);
teardown.d_name = dentry->d_name;
- vn_mark_bad(vp);
-
if (S_ISDIR(mode))
VOP_RMDIR(dvp, &teardown, NULL, err2);
else
@@ -429,11 +471,14 @@ linvfs_symlink(
error = 0;
VOP_SYMLINK(dvp, dentry, &va, (char *)symname, &cvp, NULL, error);
- if (!error && cvp) {
- ip = LINVFS_GET_IP(cvp);
- d_instantiate(dentry, ip);
- validate_fields(dir);
- validate_fields(ip); /* size needs update */
+ if (likely(!error && cvp)) {
+ error = linvfs_init_security(cvp, dir);
+ if (likely(!error)) {
+ ip = LINVFS_GET_IP(cvp);
+ d_instantiate(dentry, ip);
+ validate_fields(dir);
+ validate_fields(ip);
+ }
}
return -error;
}
@@ -502,7 +547,7 @@ linvfs_follow_link(
ASSERT(dentry);
ASSERT(nd);
- link = (char *)kmalloc(MAXNAMELEN+1, GFP_KERNEL);
+ link = (char *)kmalloc(MAXPATHLEN+1, GFP_KERNEL);
if (!link) {
nd_set_link(nd, ERR_PTR(-ENOMEM));
return NULL;
@@ -518,12 +563,12 @@ linvfs_follow_link(
vp = LINVFS_GET_VP(dentry->d_inode);
iov.iov_base = link;
- iov.iov_len = MAXNAMELEN;
+ iov.iov_len = MAXPATHLEN;
uio->uio_iov = &iov;
uio->uio_offset = 0;
uio->uio_segflg = UIO_SYSSPACE;
- uio->uio_resid = MAXNAMELEN;
+ uio->uio_resid = MAXPATHLEN;
uio->uio_iovcnt = 1;
VOP_READLINK(vp, uio, 0, NULL, error);
@@ -531,7 +576,7 @@ linvfs_follow_link(
kfree(link);
link = ERR_PTR(-error);
} else {
- link[MAXNAMELEN - uio->uio_resid] = '\0';
+ link[MAXPATHLEN - uio->uio_resid] = '\0';
}
kfree(uio);
diff --git a/fs/xfs/linux-2.6/xfs_iops.h b/fs/xfs/linux-2.6/xfs_iops.h
index ee784b63acb..6899a6b4a50 100644
--- a/fs/xfs/linux-2.6/xfs_iops.h
+++ b/fs/xfs/linux-2.6/xfs_iops.h
@@ -26,11 +26,6 @@ extern struct file_operations linvfs_file_operations;
extern struct file_operations linvfs_invis_file_operations;
extern struct file_operations linvfs_dir_operations;
-extern struct address_space_operations linvfs_aops;
-
-extern int linvfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
-extern void linvfs_unwritten_done(struct buffer_head *, int);
-
extern int xfs_ioctl(struct bhv_desc *, struct inode *, struct file *,
int, unsigned int, void __user *);
diff --git a/fs/xfs/linux-2.6/xfs_linux.h b/fs/xfs/linux-2.6/xfs_linux.h
index d8e21ba0ccc..67389b74552 100644
--- a/fs/xfs/linux-2.6/xfs_linux.h
+++ b/fs/xfs/linux-2.6/xfs_linux.h
@@ -110,10 +110,6 @@
* delalloc and these ondisk-uninitialised buffers.
*/
BUFFER_FNS(PrivateStart, unwritten);
-static inline void set_buffer_unwritten_io(struct buffer_head *bh)
-{
- bh->b_end_io = linvfs_unwritten_done;
-}
#define restricted_chown xfs_params.restrict_chown.val
#define irix_sgid_inherit xfs_params.sgid_inherit.val
@@ -232,7 +228,7 @@ static inline void set_buffer_unwritten_io(struct buffer_head *bh)
#define xfs_itruncate_data(ip, off) \
(-vmtruncate(LINVFS_GET_IP(XFS_ITOV(ip)), (off)))
#define xfs_statvfs_fsid(statp, mp) \
- ({ u64 id = huge_encode_dev((mp)->m_dev); \
+ ({ u64 id = huge_encode_dev((mp)->m_ddev_targp->bt_dev); \
__kernel_fsid_t *fsid = &(statp)->f_fsid; \
(fsid->val[0] = (u32)id, fsid->val[1] = (u32)(id >> 32)); })
diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c
index 279e9bc92ab..e0ab45fbfeb 100644
--- a/fs/xfs/linux-2.6/xfs_lrw.c
+++ b/fs/xfs/linux-2.6/xfs_lrw.c
@@ -233,8 +233,8 @@ xfs_read(
xfs_buftarg_t *target =
(ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
mp->m_rtdev_targp : mp->m_ddev_targp;
- if ((*offset & target->pbr_smask) ||
- (size & target->pbr_smask)) {
+ if ((*offset & target->bt_smask) ||
+ (size & target->bt_smask)) {
if (*offset == ip->i_d.di_size) {
return (0);
}
@@ -254,7 +254,7 @@ xfs_read(
}
if (unlikely(ioflags & IO_ISDIRECT))
- down(&inode->i_sem);
+ mutex_lock(&inode->i_mutex);
xfs_ilock(ip, XFS_IOLOCK_SHARED);
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) &&
@@ -281,12 +281,9 @@ xfs_read(
xfs_iunlock(ip, XFS_IOLOCK_SHARED);
- if (likely(!(ioflags & IO_INVIS)))
- xfs_ichgtime_fast(ip, inode, XFS_ICHGTIME_ACC);
-
unlock_isem:
if (unlikely(ioflags & IO_ISDIRECT))
- up(&inode->i_sem);
+ mutex_unlock(&inode->i_mutex);
return ret;
}
@@ -346,9 +343,6 @@ xfs_sendfile(
if (ret > 0)
XFS_STATS_ADD(xs_read_bytes, ret);
- if (likely(!(ioflags & IO_INVIS)))
- xfs_ichgtime_fast(ip, LINVFS_GET_IP(vp), XFS_ICHGTIME_ACC);
-
return ret;
}
@@ -362,7 +356,6 @@ STATIC int /* error (positive) */
xfs_zero_last_block(
struct inode *ip,
xfs_iocore_t *io,
- xfs_off_t offset,
xfs_fsize_t isize,
xfs_fsize_t end_size)
{
@@ -371,19 +364,16 @@ xfs_zero_last_block(
int nimaps;
int zero_offset;
int zero_len;
- int isize_fsb_offset;
int error = 0;
xfs_bmbt_irec_t imap;
loff_t loff;
- size_t lsize;
ASSERT(ismrlocked(io->io_lock, MR_UPDATE) != 0);
- ASSERT(offset > isize);
mp = io->io_mount;
- isize_fsb_offset = XFS_B_FSB_OFFSET(mp, isize);
- if (isize_fsb_offset == 0) {
+ zero_offset = XFS_B_FSB_OFFSET(mp, isize);
+ if (zero_offset == 0) {
/*
* There are no extra bytes in the last block on disk to
* zero, so return.
@@ -413,10 +403,8 @@ xfs_zero_last_block(
*/
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL| XFS_EXTSIZE_RD);
loff = XFS_FSB_TO_B(mp, last_fsb);
- lsize = XFS_FSB_TO_B(mp, 1);
- zero_offset = isize_fsb_offset;
- zero_len = mp->m_sb.sb_blocksize - isize_fsb_offset;
+ zero_len = mp->m_sb.sb_blocksize - zero_offset;
error = xfs_iozero(ip, loff + zero_offset, zero_len, end_size);
@@ -447,20 +435,17 @@ xfs_zero_eof(
struct inode *ip = LINVFS_GET_IP(vp);
xfs_fileoff_t start_zero_fsb;
xfs_fileoff_t end_zero_fsb;
- xfs_fileoff_t prev_zero_fsb;
xfs_fileoff_t zero_count_fsb;
xfs_fileoff_t last_fsb;
xfs_extlen_t buf_len_fsb;
- xfs_extlen_t prev_zero_count;
xfs_mount_t *mp;
int nimaps;
int error = 0;
xfs_bmbt_irec_t imap;
- loff_t loff;
- size_t lsize;
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
+ ASSERT(offset > isize);
mp = io->io_mount;
@@ -468,7 +453,7 @@ xfs_zero_eof(
* First handle zeroing the block on which isize resides.
* We only zero a part of that block so it is handled specially.
*/
- error = xfs_zero_last_block(ip, io, offset, isize, end_size);
+ error = xfs_zero_last_block(ip, io, isize, end_size);
if (error) {
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
@@ -496,8 +481,6 @@ xfs_zero_eof(
}
ASSERT(start_zero_fsb <= end_zero_fsb);
- prev_zero_fsb = NULLFILEOFF;
- prev_zero_count = 0;
while (start_zero_fsb <= end_zero_fsb) {
nimaps = 1;
zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
@@ -519,10 +502,7 @@ xfs_zero_eof(
* that sits on a hole and sets the page as P_HOLE
* and calls remapf if it is a mapped file.
*/
- prev_zero_fsb = NULLFILEOFF;
- prev_zero_count = 0;
- start_zero_fsb = imap.br_startoff +
- imap.br_blockcount;
+ start_zero_fsb = imap.br_startoff + imap.br_blockcount;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
continue;
}
@@ -543,17 +523,15 @@ xfs_zero_eof(
*/
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
- loff = XFS_FSB_TO_B(mp, start_zero_fsb);
- lsize = XFS_FSB_TO_B(mp, buf_len_fsb);
-
- error = xfs_iozero(ip, loff, lsize, end_size);
+ error = xfs_iozero(ip,
+ XFS_FSB_TO_B(mp, start_zero_fsb),
+ XFS_FSB_TO_B(mp, buf_len_fsb),
+ end_size);
if (error) {
goto out_lock;
}
- prev_zero_fsb = start_zero_fsb;
- prev_zero_count = buf_len_fsb;
start_zero_fsb = imap.br_startoff + buf_len_fsb;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
@@ -640,7 +618,7 @@ xfs_write(
(xip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
mp->m_rtdev_targp : mp->m_ddev_targp;
- if ((pos & target->pbr_smask) || (count & target->pbr_smask))
+ if ((pos & target->bt_smask) || (count & target->bt_smask))
return XFS_ERROR(-EINVAL);
if (!VN_CACHED(vp) && pos < i_size_read(inode))
@@ -655,7 +633,7 @@ relock:
iolock = XFS_IOLOCK_EXCL;
locktype = VRWLOCK_WRITE;
- down(&inode->i_sem);
+ mutex_lock(&inode->i_mutex);
} else {
iolock = XFS_IOLOCK_SHARED;
locktype = VRWLOCK_WRITE_DIRECT;
@@ -686,7 +664,7 @@ start:
int dmflags = FILP_DELAY_FLAG(file);
if (need_isem)
- dmflags |= DM_FLAGS_ISEM;
+ dmflags |= DM_FLAGS_IMUX;
xfs_iunlock(xip, XFS_ILOCK_EXCL);
error = XFS_SEND_DATA(xip->i_mount, DM_EVENT_WRITE, vp,
@@ -713,7 +691,7 @@ start:
}
if (likely(!(ioflags & IO_INVIS))) {
- inode_update_time(inode, 1);
+ file_update_time(file);
xfs_ichgtime_fast(xip, inode,
XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
}
@@ -772,7 +750,7 @@ retry:
if (need_isem) {
/* demote the lock now the cached pages are gone */
XFS_ILOCK_DEMOTE(mp, io, XFS_IOLOCK_EXCL);
- up(&inode->i_sem);
+ mutex_unlock(&inode->i_mutex);
iolock = XFS_IOLOCK_SHARED;
locktype = VRWLOCK_WRITE_DIRECT;
@@ -817,20 +795,24 @@ retry:
xfs_rwunlock(bdp, locktype);
if (need_isem)
- up(&inode->i_sem);
+ mutex_unlock(&inode->i_mutex);
error = XFS_SEND_NAMESP(xip->i_mount, DM_EVENT_NOSPACE, vp,
DM_RIGHT_NULL, vp, DM_RIGHT_NULL, NULL, NULL,
0, 0, 0); /* Delay flag intentionally unused */
if (error)
goto out_nounlocks;
if (need_isem)
- down(&inode->i_sem);
+ mutex_lock(&inode->i_mutex);
xfs_rwlock(bdp, locktype);
pos = xip->i_d.di_size;
ret = 0;
goto retry;
}
+ isize = i_size_read(inode);
+ if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize))
+ *offset = isize;
+
if (*offset > xip->i_d.di_size) {
xfs_ilock(xip, XFS_ILOCK_EXCL);
if (*offset > xip->i_d.di_size) {
@@ -926,7 +908,7 @@ retry:
xfs_rwunlock(bdp, locktype);
if (need_isem)
- up(&inode->i_sem);
+ mutex_unlock(&inode->i_mutex);
error = sync_page_range(inode, mapping, pos, ret);
if (!error)
@@ -938,7 +920,7 @@ retry:
xfs_rwunlock(bdp, locktype);
out_unlock_isem:
if (need_isem)
- up(&inode->i_sem);
+ mutex_unlock(&inode->i_mutex);
out_nounlocks:
return -error;
}
@@ -956,7 +938,7 @@ xfs_bdstrat_cb(struct xfs_buf *bp)
mp = XFS_BUF_FSPRIVATE3(bp, xfs_mount_t *);
if (!XFS_FORCED_SHUTDOWN(mp)) {
- pagebuf_iorequest(bp);
+ xfs_buf_iorequest(bp);
return 0;
} else {
xfs_buftrace("XFS__BDSTRAT IOERROR", bp);
@@ -1009,7 +991,7 @@ xfsbdstrat(
* if (XFS_BUF_IS_GRIO(bp)) {
*/
- pagebuf_iorequest(bp);
+ xfs_buf_iorequest(bp);
return 0;
}
diff --git a/fs/xfs/linux-2.6/xfs_stats.c b/fs/xfs/linux-2.6/xfs_stats.c
index 6c40a74be7c..8955720a2c6 100644
--- a/fs/xfs/linux-2.6/xfs_stats.c
+++ b/fs/xfs/linux-2.6/xfs_stats.c
@@ -34,7 +34,7 @@ xfs_read_xfsstats(
__uint64_t xs_write_bytes = 0;
__uint64_t xs_read_bytes = 0;
- static struct xstats_entry {
+ static const struct xstats_entry {
char *desc;
int endpoint;
} xstats[] = {
diff --git a/fs/xfs/linux-2.6/xfs_stats.h b/fs/xfs/linux-2.6/xfs_stats.h
index 50027c4a561..8ba7a2fa6c1 100644
--- a/fs/xfs/linux-2.6/xfs_stats.h
+++ b/fs/xfs/linux-2.6/xfs_stats.h
@@ -109,15 +109,15 @@ struct xfsstats {
__uint32_t vn_remove; /* # times vn_remove called */
__uint32_t vn_free; /* # times vn_free called */
#define XFSSTAT_END_BUF (XFSSTAT_END_VNODE_OPS+9)
- __uint32_t pb_get;
- __uint32_t pb_create;
- __uint32_t pb_get_locked;
- __uint32_t pb_get_locked_waited;
- __uint32_t pb_busy_locked;
- __uint32_t pb_miss_locked;
- __uint32_t pb_page_retries;
- __uint32_t pb_page_found;
- __uint32_t pb_get_read;
+ __uint32_t xb_get;
+ __uint32_t xb_create;
+ __uint32_t xb_get_locked;
+ __uint32_t xb_get_locked_waited;
+ __uint32_t xb_busy_locked;
+ __uint32_t xb_miss_locked;
+ __uint32_t xb_page_retries;
+ __uint32_t xb_page_found;
+ __uint32_t xb_get_read;
/* Extra precision counters */
__uint64_t xs_xstrat_bytes;
__uint64_t xs_write_bytes;
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 6116b5bf433..f22e426d9e4 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -306,13 +306,15 @@ xfs_mountfs_check_barriers(xfs_mount_t *mp)
xfs_fs_cmn_err(CE_NOTE, mp,
"Disabling barriers, not supported with external log device");
mp->m_flags &= ~XFS_MOUNT_BARRIER;
+ return;
}
- if (mp->m_ddev_targp->pbr_bdev->bd_disk->queue->ordered ==
+ if (mp->m_ddev_targp->bt_bdev->bd_disk->queue->ordered ==
QUEUE_ORDERED_NONE) {
xfs_fs_cmn_err(CE_NOTE, mp,
"Disabling barriers, not supported by the underlying device");
mp->m_flags &= ~XFS_MOUNT_BARRIER;
+ return;
}
error = xfs_barrier_test(mp);
@@ -320,6 +322,7 @@ xfs_mountfs_check_barriers(xfs_mount_t *mp)
xfs_fs_cmn_err(CE_NOTE, mp,
"Disabling barriers, trial barrier write failed");
mp->m_flags &= ~XFS_MOUNT_BARRIER;
+ return;
}
}
@@ -327,7 +330,7 @@ void
xfs_blkdev_issue_flush(
xfs_buftarg_t *buftarg)
{
- blkdev_issue_flush(buftarg->pbr_bdev, NULL);
+ blkdev_issue_flush(buftarg->bt_bdev, NULL);
}
STATIC struct inode *
@@ -576,7 +579,7 @@ xfssyncd(
timeleft = schedule_timeout_interruptible(timeleft);
/* swsusp */
try_to_freeze();
- if (kthread_should_stop())
+ if (kthread_should_stop() && list_empty(&vfsp->vfs_sync_list))
break;
spin_lock(&vfsp->vfs_sync_lock);
@@ -966,9 +969,9 @@ init_xfs_fs( void )
if (error < 0)
goto undo_zones;
- error = pagebuf_init();
+ error = xfs_buf_init();
if (error < 0)
- goto undo_pagebuf;
+ goto undo_buffers;
vn_init();
xfs_init();
@@ -982,9 +985,9 @@ init_xfs_fs( void )
return 0;
undo_register:
- pagebuf_terminate();
+ xfs_buf_terminate();
-undo_pagebuf:
+undo_buffers:
linvfs_destroy_zones();
undo_zones:
@@ -998,7 +1001,7 @@ exit_xfs_fs( void )
XFS_DM_EXIT(&xfs_fs_type);
unregister_filesystem(&xfs_fs_type);
xfs_cleanup();
- pagebuf_terminate();
+ xfs_buf_terminate();
linvfs_destroy_zones();
ktrace_uninit();
}
diff --git a/fs/xfs/linux-2.6/xfs_vnode.c b/fs/xfs/linux-2.6/xfs_vnode.c
index e9bbcb4d624..260dd8415dd 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.c
+++ b/fs/xfs/linux-2.6/xfs_vnode.c
@@ -106,7 +106,6 @@ vn_revalidate_core(
inode->i_blocks = vap->va_nblocks;
inode->i_mtime = vap->va_mtime;
inode->i_ctime = vap->va_ctime;
- inode->i_atime = vap->va_atime;
inode->i_blksize = vap->va_blocksize;
if (vap->va_xflags & XFS_XFLAG_IMMUTABLE)
inode->i_flags |= S_IMMUTABLE;
diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h
index f2bbb327c08..0fe2419461d 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.h
+++ b/fs/xfs/linux-2.6/xfs_vnode.h
@@ -566,6 +566,25 @@ static inline int VN_BAD(struct vnode *vp)
}
/*
+ * Extracting atime values in various formats
+ */
+static inline void vn_atime_to_bstime(struct vnode *vp, xfs_bstime_t *bs_atime)
+{
+ bs_atime->tv_sec = vp->v_inode.i_atime.tv_sec;
+ bs_atime->tv_nsec = vp->v_inode.i_atime.tv_nsec;
+}
+
+static inline void vn_atime_to_timespec(struct vnode *vp, struct timespec *ts)
+{
+ *ts = vp->v_inode.i_atime;
+}
+
+static inline void vn_atime_to_time_t(struct vnode *vp, time_t *tt)
+{
+ *tt = vp->v_inode.i_atime.tv_sec;
+}
+
+/*
* Some useful predicates.
*/
#define VN_MAPPED(vp) mapping_mapped(LINVFS_GET_IP(vp)->i_mapping)
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2024-11-24 16:57:50 +0000