[RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Dave Chinner]

This series is a backport of all the major bug fixes in the current
TOT kernel to the current 3.0.x stable tree.

I won't make any secret of this - the fixes and supporting patches
have been selected as a result of the issues reported in the current
RHEL XFS codebase which currently is 2 commits short of the 3.0 code
base.  With that said, it doesn't take a brain surgeon to work out
the motivations behind this work and eventual destination of the
patch set. ;)

There's no guarantee I have caught every single bug fix that has
been made since 3.0, but I've tried to grab all the bug fix commits
as indicated by the commit headers (hence the importance of good one
line bug summaries).

Over the past couple of weeks of testing and refining, I've had only
three significant problems arise from QA and load testing:

	1) An unreproducable log space hang
	2) An unmount panic due to buffers not being cleaned up
	before tearing down the perag tree
	3) A forced shutdown panic in block_invalidatepage()
	via xfs_aops_discard_page()

It's entirely possible that #1 was due to the CIL space hang we
still haven't got to the bottom of, so i'm not not greatly concerned
by that. #2 implies I haven't quite backported the shutdown ordering
fixes correctly (or I missed one), so I have a bit more work to do
there. And for #3 - I've never seen that before and I haven't been
able to reproduce it, so I really don't know what potential cause or
impact it has.

I've been beating on the series with xfstests, dbench, fsmark,
postmark, compilebench and a few other load scripts that I've got,
and it seems fairly resilient.  Hence, it's time to give the series
wider testing and review to flush out any remaining issues.

For all the folks that run 3.0.x stable kernels, I'd appreciate it if
you could give this a whirl on your test systems to see if there are
any obvious, glaring problems that show up under your particular
workloads. This woul dbe of great benefit to me before I submit the
series to the stable kernel gurus - I'd prefer it there's more
substantial testing than "i've done what I can" when sending them
the series.

For all the XFS developers that have copious amounts of free time
available, I'd appreciate an eye run over the patch list to see if
there's any potential bug fixes that I missed or have made glaring
errors in backporting. Some of the fixes are dependent on cleanups I
haven't included, so some of the patches are a bit different to what
is in mainline (e.g. anything that touches setattr). Most important
to look at is probably the inode i_size changes and the logging of
all metadata changes.

Enjoy!

Cheers,

Dave.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 001/102] xfs: don't serialise adjacent concurrent direct IO appending writes

[Dave Chinner]

Upstream commit: 7271d243f9d1b4106289e4cf876c8b1203de59ab

For append write workloads, extending the file requires a certain
amount of exclusive locking to be done up front to ensure sanity in
things like ensuring that we've zeroed any allocated regions
between the old EOF and the start of the new IO.

For single threads, this typically isn't a problem, and for large
IOs we don't serialise enough for it to be a problem for two
threads on really fast block devices. However for smaller IO and
larger thread counts we have a problem.

Take 4 concurrent sequential, single block sized and aligned IOs.
After the first IO is submitted but before it completes, we end up
with this state:

        IO 1    IO 2    IO 3    IO 4
      +-------+-------+-------+-------+
      ^       ^
      |       |
      |       |
      |       |
      |       \- ip->i_new_size
      \- ip->i_size

And the IO is done without exclusive locking because offset <=
ip->i_size. When we submit IO 2, we see offset > ip->i_size, and
grab the IO lock exclusive, because there is a chance we need to do
EOF zeroing. However, there is already an IO in progress that avoids
the need for IO zeroing because offset <= ip->i_new_size. hence we
could avoid holding the IO lock exlcusive for this. Hence after
submission of the second IO, we'd end up this state:

        IO 1    IO 2    IO 3    IO 4
      +-------+-------+-------+-------+
      ^               ^
      |               |
      |               |
      |               |
      |               \- ip->i_new_size
      \- ip->i_size

There is no need to grab the i_mutex of the IO lock in exclusive
mode if we don't need to invalidate the page cache. Taking these
locks on every direct IO effective serialises them as taking the IO
lock in exclusive mode has to wait for all shared holders to drop
the lock. That only happens when IO is complete, so effective it
prevents dispatch of concurrent direct IO writes to the same inode.

And so you can see that for the third concurrent IO, we'd avoid
exclusive locking for the same reason we avoided the exclusive lock
for the second IO.

Fixing this is a bit more complex than that, because we need to hold
a write-submission local value of ip->i_new_size to that clearing
the value is only done if no other thread has updated it before our
IO completes.....

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |   68 +++++++++++++++++++++++++++++++++----------
 1 file changed, 52 insertions(+), 16 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index b679198..67d392a 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -410,11 +410,13 @@ xfs_aio_write_isize_update(
  */
 STATIC void
 xfs_aio_write_newsize_update(
-	struct xfs_inode	*ip)
+	struct xfs_inode	*ip,
+	xfs_fsize_t		new_size)
 {
-	if (ip->i_new_size) {
+	if (new_size == ip->i_new_size) {
 		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
-		ip->i_new_size = 0;
+		if (new_size == ip->i_new_size)
+			ip->i_new_size = 0;
 		if (ip->i_d.di_size > ip->i_size)
 			ip->i_d.di_size = ip->i_size;
 		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
@@ -465,7 +467,7 @@ xfs_file_splice_write(
 	ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags);
 
 	xfs_aio_write_isize_update(inode, ppos, ret);
-	xfs_aio_write_newsize_update(ip);
+	xfs_aio_write_newsize_update(ip, new_size);
 	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 	return ret;
 }
@@ -662,6 +664,7 @@ xfs_file_aio_write_checks(
 	struct file		*file,
 	loff_t			*pos,
 	size_t			*count,
+	xfs_fsize_t		*new_sizep,
 	int			*iolock)
 {
 	struct inode		*inode = file->f_mapping->host;
@@ -670,6 +673,8 @@ xfs_file_aio_write_checks(
 	int			error = 0;
 
 	xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
+	*new_sizep = 0;
+restart:
 	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
 	if (error) {
 		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
@@ -677,20 +682,41 @@ xfs_file_aio_write_checks(
 		return error;
 	}
 
-	new_size = *pos + *count;
-	if (new_size > ip->i_size)
-		ip->i_new_size = new_size;
-
 	if (likely(!(file->f_mode & FMODE_NOCMTIME)))
 		file_update_time(file);
 
 	/*
 	 * If the offset is beyond the size of the file, we need to zero any
 	 * blocks that fall between the existing EOF and the start of this
-	 * write.
+	 * write. There is no need to issue zeroing if another in-flght IO ends
+	 * at or before this one If zeronig is needed and we are currently
+	 * holding the iolock shared, we need to update it to exclusive which
+	 * involves dropping all locks and relocking to maintain correct locking
+	 * order. If we do this, restart the function to ensure all checks and
+	 * values are still valid.
 	 */
-	if (*pos > ip->i_size)
+	if ((ip->i_new_size && *pos > ip->i_new_size) ||
+	    (!ip->i_new_size && *pos > ip->i_size)) {
+		if (*iolock == XFS_IOLOCK_SHARED) {
+			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
+			*iolock = XFS_IOLOCK_EXCL;
+			xfs_rw_ilock(ip, XFS_ILOCK_EXCL | *iolock);
+			goto restart;
+		}
 		error = -xfs_zero_eof(ip, *pos, ip->i_size);
+	}
+
+	/*
+	 * If this IO extends beyond EOF, we may need to update ip->i_new_size.
+	 * We have already zeroed space beyond EOF (if necessary).  Only update
+	 * ip->i_new_size if this IO ends beyond any other in-flight writes.
+	 */
+	new_size = *pos + *count;
+	if (new_size > ip->i_size) {
+		if (new_size > ip->i_new_size)
+			ip->i_new_size = new_size;
+		*new_sizep = new_size;
+	}
 
 	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
 	if (error)
@@ -737,6 +763,7 @@ xfs_file_dio_aio_write(
 	unsigned long		nr_segs,
 	loff_t			pos,
 	size_t			ocount,
+	xfs_fsize_t		*new_size,
 	int			*iolock)
 {
 	struct file		*file = iocb->ki_filp;
@@ -757,13 +784,20 @@ xfs_file_dio_aio_write(
 	if ((pos & mp->m_blockmask) || ((pos + count) & mp->m_blockmask))
 		unaligned_io = 1;
 
-	if (unaligned_io || mapping->nrpages || pos > ip->i_size)
+	/*
+	 * We don't need to take an exclusive lock unless there page cache needs
+	 * to be invalidated or unaligned IO is being executed. We don't need to
+	 * consider the EOF extension case here because
+	 * xfs_file_aio_write_checks() will relock the inode as necessary for
+	 * EOF zeroing cases and fill out the new inode size as appropriate.
+	 */
+	if (unaligned_io || mapping->nrpages)
 		*iolock = XFS_IOLOCK_EXCL;
 	else
 		*iolock = XFS_IOLOCK_SHARED;
 	xfs_rw_ilock(ip, *iolock);
 
-	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
+	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
 	if (ret)
 		return ret;
 
@@ -812,6 +846,7 @@ xfs_file_buffered_aio_write(
 	unsigned long		nr_segs,
 	loff_t			pos,
 	size_t			ocount,
+	xfs_fsize_t		*new_size,
 	int			*iolock)
 {
 	struct file		*file = iocb->ki_filp;
@@ -825,7 +860,7 @@ xfs_file_buffered_aio_write(
 	*iolock = XFS_IOLOCK_EXCL;
 	xfs_rw_ilock(ip, *iolock);
 
-	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
+	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
 	if (ret)
 		return ret;
 
@@ -865,6 +900,7 @@ xfs_file_aio_write(
 	ssize_t			ret;
 	int			iolock;
 	size_t			ocount = 0;
+	xfs_fsize_t		new_size = 0;
 
 	XFS_STATS_INC(xs_write_calls);
 
@@ -884,10 +920,10 @@ xfs_file_aio_write(
 
 	if (unlikely(file->f_flags & O_DIRECT))
 		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &iolock);
+						ocount, &new_size, &iolock);
 	else
 		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &iolock);
+						ocount, &new_size, &iolock);
 
 	xfs_aio_write_isize_update(inode, &iocb->ki_pos, ret);
 
@@ -912,7 +948,7 @@ xfs_file_aio_write(
 	}
 
 out_unlock:
-	xfs_aio_write_newsize_update(ip);
+	xfs_aio_write_newsize_update(ip, new_size);
 	xfs_rw_iunlock(ip, iolock);
 	return ret;
 }
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 002/102] xfs: remove dead ENODEV handling in xfs_destroy_ioend

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 398d25ef23b10ce75424e0336a8d059dda1dbc8d

No driver returns ENODEV from it bio completion handler, not has this
ever been documented.  Remove the dead code dealing with it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   11 -----------
 1 file changed, 11 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 79ce38b..05df3a8 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -122,17 +122,6 @@ xfs_destroy_ioend(
 		bh->b_end_io(bh, !ioend->io_error);
 	}
 
-	/*
-	 * Volume managers supporting multiple paths can send back ENODEV
-	 * when the final path disappears.  In this case continuing to fill
-	 * the page cache with dirty data which cannot be written out is
-	 * evil, so prevent that.
-	 */
-	if (unlikely(ioend->io_error == -ENODEV)) {
-		xfs_do_force_shutdown(ip->i_mount, SHUTDOWN_DEVICE_REQ,
-				      __FILE__, __LINE__);
-	}
-
 	xfs_ioend_wake(ip);
 	mempool_free(ioend, xfs_ioend_pool);
 }
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 003/102] xfs: defer AIO/DIO completions

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: c859cdd1da008b3825555be3242908088a3de366

We really shouldn't complete AIO or DIO requests until we have finished
the unwritten extent conversion and size update.  This means fsync never
has to pick up any ioends as all work has been completed when signalling
I/O completion.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   21 ++++++++-------------
 fs/xfs/linux-2.6/xfs_aops.h |    1 +
 2 files changed, 9 insertions(+), 13 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 05df3a8..68ac1ce 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -122,6 +122,10 @@ xfs_destroy_ioend(
 		bh->b_end_io(bh, !ioend->io_error);
 	}
 
+	if (ioend->io_iocb) {
+		if (ioend->io_isasync)
+			aio_complete(ioend->io_iocb, ioend->io_result, 0);
+	}
 	xfs_ioend_wake(ip);
 	mempool_free(ioend, xfs_ioend_pool);
 }
@@ -235,8 +239,6 @@ xfs_end_io(
 		/* ensure we don't spin on blocked ioends */
 		delay(1);
 	} else {
-		if (ioend->io_iocb)
-			aio_complete(ioend->io_iocb, ioend->io_result, 0);
 		xfs_destroy_ioend(ioend);
 	}
 }
@@ -273,6 +275,7 @@ xfs_alloc_ioend(
 	 * all the I/O from calling the completion routine too early.
 	 */
 	atomic_set(&ioend->io_remaining, 1);
+	ioend->io_isasync = 0;
 	ioend->io_error = 0;
 	ioend->io_list = NULL;
 	ioend->io_type = type;
@@ -1309,21 +1312,13 @@ xfs_end_io_direct_write(
 
 	ioend->io_offset = offset;
 	ioend->io_size = size;
+	ioend->io_iocb = iocb;
+	ioend->io_result = ret;
 	if (private && size > 0)
 		ioend->io_type = IO_UNWRITTEN;
 
 	if (is_async) {
-		/*
-		 * If we are converting an unwritten extent we need to delay
-		 * the AIO completion until after the unwrittent extent
-		 * conversion has completed, otherwise do it ASAP.
-		 */
-		if (ioend->io_type == IO_UNWRITTEN) {
-			ioend->io_iocb = iocb;
-			ioend->io_result = ret;
-		} else {
-			aio_complete(iocb, ret, 0);
-		}
+		ioend->io_isasync = 1;
 		xfs_finish_ioend(ioend);
 	} else {
 		xfs_finish_ioend_sync(ioend);
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
index 71f721e..ce3dcb5 100644
--- a/fs/xfs/linux-2.6/xfs_aops.h
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -47,6 +47,7 @@ typedef struct xfs_ioend {
 	unsigned int		io_type;	/* delalloc / unwritten */
 	int			io_error;	/* I/O error code */
 	atomic_t		io_remaining;	/* hold count */
+	unsigned int		io_isasync : 1;	/* needs aio_complete */
 	struct inode		*io_inode;	/* file being written to */
 	struct buffer_head	*io_buffer_head;/* buffer linked list head */
 	struct buffer_head	*io_buffer_tail;/* buffer linked list tail */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 004/102] xfs: reduce ioend latency

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: fc0063c4474599b7a066ba76b90902abe21bc675

There is no reason to queue up ioends for processing in user context
unless we actually need it.  Just complete ioends that do not convert
unwritten extents or need a size update from the end_io context.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 68ac1ce..20f356d 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -149,6 +149,15 @@ xfs_ioend_new_eof(
 }
 
 /*
+ * Fast and loose check if this write could update the on-disk inode size.
+ */
+static inline bool xfs_ioend_is_append(struct xfs_ioend *ioend)
+{
+	return ioend->io_offset + ioend->io_size >
+		XFS_I(ioend->io_inode)->i_d.di_size;
+}
+
+/*
  * Update on-disk file size now that data has been written to disk.  The
  * current in-memory file size is i_size.  If a write is beyond eof i_new_size
  * will be the intended file size until i_size is updated.  If this write does
@@ -184,6 +193,9 @@ xfs_setfilesize(
 
 /*
  * Schedule IO completion handling on the final put of an ioend.
+ *
+ * If there is no work to do we might as well call it a day and free the
+ * ioend right now.
  */
 STATIC void
 xfs_finish_ioend(
@@ -192,8 +204,10 @@ xfs_finish_ioend(
 	if (atomic_dec_and_test(&ioend->io_remaining)) {
 		if (ioend->io_type == IO_UNWRITTEN)
 			queue_work(xfsconvertd_workqueue, &ioend->io_work);
-		else
+		else if (xfs_ioend_is_append(ioend))
 			queue_work(xfsdatad_workqueue, &ioend->io_work);
+		else
+			xfs_destroy_ioend(ioend);
 	}
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 005/102] xfs: wait for I/O completion when writing out pages in xfs_setattr_size

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 2b3ffd7eb7b4392e3657c5046b055ca9f1f7cf5e

The current code relies on the xfs_ioend_wait call later on to make sure
all I/O actually has completed.  The xfs_ioend_wait call will go away soon,
so prepare for that by using the waiting filemap function.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_vnodeops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 59509ae..fbdb33b 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -246,7 +246,7 @@ xfs_setattr(
 		    iattr->ia_size > ip->i_d.di_size) {
 			code = xfs_flush_pages(ip,
 					ip->i_d.di_size, iattr->ia_size,
-					XBF_ASYNC, FI_NONE);
+					0, FI_NONE);
 			if (code)
 				goto error_return;
 		}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 006/102] xfs: improve ioend error handling

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 04f658ee229f60dbb9a0dc2f3d6871b12b758051

Return unwritten extent conversion errors to aio_complete.

Skip both unwritten extent conversion and size updates if we had an
I/O error or the filesystem has been shut down.

Return -EIO to the aio/buffer completion handlers in case of a
forced shutdown.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 20f356d..f16207b 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -123,8 +123,10 @@ xfs_destroy_ioend(
 	}
 
 	if (ioend->io_iocb) {
-		if (ioend->io_isasync)
-			aio_complete(ioend->io_iocb, ioend->io_result, 0);
+		if (ioend->io_isasync) {
+			aio_complete(ioend->io_iocb, ioend->io_error ?
+					ioend->io_error : ioend->io_result, 0);
+		}
 	}
 	xfs_ioend_wake(ip);
 	mempool_free(ioend, xfs_ioend_pool);
@@ -175,9 +177,6 @@ xfs_setfilesize(
 	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
 	xfs_fsize_t		isize;
 
-	if (unlikely(ioend->io_error))
-		return 0;
-
 	if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL))
 		return EAGAIN;
 
@@ -222,17 +221,24 @@ xfs_end_io(
 	struct xfs_inode *ip = XFS_I(ioend->io_inode);
 	int		error = 0;
 
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+		error = -EIO;
+		goto done;
+	}
+	if (ioend->io_error)
+		goto done;
+
 	/*
 	 * For unwritten extents we need to issue transactions to convert a
 	 * range to normal written extens after the data I/O has finished.
 	 */
-	if (ioend->io_type == IO_UNWRITTEN &&
-	    likely(!ioend->io_error && !XFS_FORCED_SHUTDOWN(ip->i_mount))) {
-
+	if (ioend->io_type == IO_UNWRITTEN) {
 		error = xfs_iomap_write_unwritten(ip, ioend->io_offset,
 						 ioend->io_size);
-		if (error)
-			ioend->io_error = error;
+		if (error) {
+			ioend->io_error = -error;
+			goto done;
+		}
 	}
 
 	/*
@@ -242,6 +248,7 @@ xfs_end_io(
 	error = xfs_setfilesize(ioend);
 	ASSERT(!error || error == EAGAIN);
 
+done:
 	/*
 	 * If we didn't complete processing of the ioend, requeue it to the
 	 * tail of the workqueue for another attempt later. Otherwise destroy
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 007/102] xfs: Check the return value of xfs_buf_get()

[Dave Chinner]

From: Chandra Seetharaman <sekharan@us.ibm.com>

Upstream commit: b522950f0ab8551f2ef56c210ebd50e6c6396601

Check the return value of xfs_buf_get() and fail appropriately.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_attr.c  |    4 ++--
 fs/xfs/xfs_fsops.c |   20 ++++++++++++++++++++
 2 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_attr.c b/fs/xfs/xfs_attr.c
index 99d4011..fe01931 100644
--- a/fs/xfs/xfs_attr.c
+++ b/fs/xfs/xfs_attr.c
@@ -2110,8 +2110,8 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 
 		bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt,
 				 XBF_LOCK | XBF_DONT_BLOCK);
-		ASSERT(bp);
-		ASSERT(!XFS_BUF_GETERROR(bp));
+		if (!bp)
+			return ENOMEM;
 
 		tmp = (valuelen < XFS_BUF_SIZE(bp)) ? valuelen :
 							XFS_BUF_SIZE(bp);
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
index 9153d2c..ac49ad1 100644
--- a/fs/xfs/xfs_fsops.c
+++ b/fs/xfs/xfs_fsops.c
@@ -194,6 +194,10 @@ xfs_growfs_data_private(
 		bp = xfs_buf_get(mp->m_ddev_targp,
 				 XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
 				 XFS_FSS_TO_BB(mp, 1), XBF_LOCK | XBF_MAPPED);
+		if (!bp) {
+			error = ENOMEM;
+			goto error0;
+		}
 		agf = XFS_BUF_TO_AGF(bp);
 		memset(agf, 0, mp->m_sb.sb_sectsize);
 		agf->agf_magicnum = cpu_to_be32(XFS_AGF_MAGIC);
@@ -226,6 +230,10 @@ xfs_growfs_data_private(
 		bp = xfs_buf_get(mp->m_ddev_targp,
 				 XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
 				 XFS_FSS_TO_BB(mp, 1), XBF_LOCK | XBF_MAPPED);
+		if (!bp) {
+			error = ENOMEM;
+			goto error0;
+		}
 		agi = XFS_BUF_TO_AGI(bp);
 		memset(agi, 0, mp->m_sb.sb_sectsize);
 		agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
@@ -251,6 +259,10 @@ xfs_growfs_data_private(
 				 XFS_AGB_TO_DADDR(mp, agno, XFS_BNO_BLOCK(mp)),
 				 BTOBB(mp->m_sb.sb_blocksize),
 				 XBF_LOCK | XBF_MAPPED);
+		if (!bp) {
+			error = ENOMEM;
+			goto error0;
+		}
 		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_ABTB_MAGIC);
@@ -273,6 +285,10 @@ xfs_growfs_data_private(
 				 XFS_AGB_TO_DADDR(mp, agno, XFS_CNT_BLOCK(mp)),
 				 BTOBB(mp->m_sb.sb_blocksize),
 				 XBF_LOCK | XBF_MAPPED);
+		if (!bp) {
+			error = ENOMEM;
+			goto error0;
+		}
 		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_ABTC_MAGIC);
@@ -296,6 +312,10 @@ xfs_growfs_data_private(
 				 XFS_AGB_TO_DADDR(mp, agno, XFS_IBT_BLOCK(mp)),
 				 BTOBB(mp->m_sb.sb_blocksize),
 				 XBF_LOCK | XBF_MAPPED);
+		if (!bp) {
+			error = ENOMEM;
+			goto error0;
+		}
 		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 008/102] xfs: Check the return value of xfs_trans_get_buf()

[Dave Chinner]

From: Chandra Seetharaman <sekharan@us.ibm.com>

Upstream commit: 2a30f36d9069b0646dcdd73def5fd7ab674bffd6

Check the return value of xfs_trans_get_buf() and fail
appropriately.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/quota/xfs_dquot.c |    3 ++-
 fs/xfs/xfs_attr_leaf.c   |    2 ++
 fs/xfs/xfs_btree.c       |    4 ++--
 fs/xfs/xfs_ialloc.c      |   13 ++++++++-----
 fs/xfs/xfs_inode.c       |    9 ++++++---
 fs/xfs/xfs_vnodeops.c    |    9 ++++++++-
 6 files changed, 28 insertions(+), 12 deletions(-)

diff --git a/fs/xfs/quota/xfs_dquot.c b/fs/xfs/quota/xfs_dquot.c
index 6fa2146..478a11b 100644
--- a/fs/xfs/quota/xfs_dquot.c
+++ b/fs/xfs/quota/xfs_dquot.c
@@ -402,7 +402,8 @@ xfs_qm_dqalloc(
 			       dqp->q_blkno,
 			       mp->m_quotainfo->qi_dqchunklen,
 			       0);
-	if (!bp || (error = XFS_BUF_GETERROR(bp)))
+	error = xfs_buf_geterror(bp);
+	if (error)
 		goto error1;
 	/*
 	 * Make a chunk of dquots out of this buffer and log
diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c
index f49ecf2..ec4f133 100644
--- a/fs/xfs/xfs_attr_leaf.c
+++ b/fs/xfs/xfs_attr_leaf.c
@@ -2962,6 +2962,8 @@ xfs_attr_leaf_freextent(xfs_trans_t **trans, xfs_inode_t *dp,
 			bp = xfs_trans_get_buf(*trans,
 					dp->i_mount->m_ddev_targp,
 					dblkno, dblkcnt, XBF_LOCK);
+			if (!bp)
+				return ENOMEM;
 			xfs_trans_binval(*trans, bp);
 			/*
 			 * Roll to next transaction.
diff --git a/fs/xfs/xfs_btree.c b/fs/xfs/xfs_btree.c
index 2f9e97c..9896456 100644
--- a/fs/xfs/xfs_btree.c
+++ b/fs/xfs/xfs_btree.c
@@ -974,8 +974,8 @@ xfs_btree_get_buf_block(
 	*bpp = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d,
 				 mp->m_bsize, flags);
 
-	ASSERT(*bpp);
-	ASSERT(!XFS_BUF_GETERROR(*bpp));
+	if (!*bpp)
+		return ENOMEM;
 
 	*block = XFS_BUF_TO_BLOCK(*bpp);
 	return 0;
diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
index 84ebeec..53f1e8a 100644
--- a/fs/xfs/xfs_ialloc.c
+++ b/fs/xfs/xfs_ialloc.c
@@ -150,7 +150,7 @@ xfs_check_agi_freecount(
 /*
  * Initialise a new set of inodes.
  */
-STATIC void
+STATIC int
 xfs_ialloc_inode_init(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
@@ -202,8 +202,8 @@ xfs_ialloc_inode_init(
 		fbuf = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
 					 mp->m_bsize * blks_per_cluster,
 					 XBF_LOCK);
-		ASSERT(fbuf);
-		ASSERT(!XFS_BUF_GETERROR(fbuf));
+		if (!fbuf)
+			return ENOMEM;
 
 		/*
 		 * Initialize all inodes in this buffer and then log them.
@@ -226,6 +226,7 @@ xfs_ialloc_inode_init(
 		}
 		xfs_trans_inode_alloc_buf(tp, fbuf);
 	}
+	return 0;
 }
 
 /*
@@ -370,9 +371,11 @@ xfs_ialloc_ag_alloc(
 	 * rather than a linear progression to prevent the next generation
 	 * number from being easily guessable.
 	 */
-	xfs_ialloc_inode_init(args.mp, tp, agno, args.agbno, args.len,
-			      random32());
+	error = xfs_ialloc_inode_init(args.mp, tp, agno, args.agbno,
+			args.len, random32());
 
+	if (error)
+		return error;
 	/*
 	 * Convert the results.
 	 */
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 5715279..2966b4d 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1898,7 +1898,7 @@ xfs_iunlink_remove(
  * inodes that are in memory - they all must be marked stale and attached to
  * the cluster buffer.
  */
-STATIC void
+STATIC int
 xfs_ifree_cluster(
 	xfs_inode_t	*free_ip,
 	xfs_trans_t	*tp,
@@ -1944,6 +1944,8 @@ xfs_ifree_cluster(
 					mp->m_bsize * blks_per_cluster,
 					XBF_LOCK);
 
+		if (!bp)
+			return ENOMEM;
 		/*
 		 * Walk the inodes already attached to the buffer and mark them
 		 * stale. These will all have the flush locks held, so an
@@ -2053,6 +2055,7 @@ retry:
 	}
 
 	xfs_perag_put(pag);
+	return 0;
 }
 
 /*
@@ -2133,10 +2136,10 @@ xfs_ifree(
 	dip->di_mode = 0;
 
 	if (delete) {
-		xfs_ifree_cluster(ip, tp, first_ino);
+		error = xfs_ifree_cluster(ip, tp, first_ino);
 	}
 
-	return 0;
+	return error;
 }
 
 /*
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index fbdb33b..154223c 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -754,6 +754,10 @@ xfs_inactive_symlink_rmt(
 		bp = xfs_trans_get_buf(tp, mp->m_ddev_targp,
 			XFS_FSB_TO_DADDR(mp, mval[i].br_startblock),
 			XFS_FSB_TO_BB(mp, mval[i].br_blockcount), 0);
+		if (!bp) {
+			error = ENOMEM;
+			goto error1;
+		}
 		xfs_trans_binval(tp, bp);
 	}
 	/*
@@ -2116,7 +2120,10 @@ xfs_symlink(
 			byte_cnt = XFS_FSB_TO_B(mp, mval[n].br_blockcount);
 			bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
 					       BTOBB(byte_cnt), 0);
-			ASSERT(bp && !XFS_BUF_GETERROR(bp));
+			if (!bp) {
+				error = ENOMEM;
+				goto error2;
+			}
 			if (pathlen < byte_cnt) {
 				byte_cnt = pathlen;
 			}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 009/102] xfs: dont ignore error code from xfs_bmbt_update

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: b0eab14e74d2d7b22d065e18a1cdebcf7716debf

Fix a case in xfs_bmap_add_extent_unwritten_real where we aren't
passing the returned error on.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_bmap.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index a175933..8193ee5 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -1407,10 +1407,11 @@ xfs_bmap_add_extent_unwritten_real(
 				goto done;
 			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
-			if (xfs_bmbt_update(cur, LEFT.br_startoff,
+			error = xfs_bmbt_update(cur, LEFT.br_startoff,
 				LEFT.br_startblock,
 				LEFT.br_blockcount + new->br_blockcount,
-				LEFT.br_state))
+				LEFT.br_state);
+			if (error)
 				goto done;
 		}
 		break;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 010/102] xfs: fix possible overflow in xfs_ioc_trim()

[Dave Chinner]

From: Lukas Czerner <lczerner@redhat.com>

Upstream commit: c029a50d51b8a9520105ec903639de03389915d0

In xfs_ioc_trim it is possible that computing the last allocation group
to discard might overflow for big start & len values, because the result
might be bigger then xfs_agnumber_t which is 32 bit long. Fix this by not
allowing the start and end block of the range to be beyond the end of the
file system.

Note that if the start is beyond the end of the file system we have to
return -EINVAL, but in the "end" case we have to truncate it to the fs
size.

Also introduce "end" variable, rather than using start+len which which
might be more confusing to get right as this bug shows.

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_discard.c |   20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_discard.c b/fs/xfs/linux-2.6/xfs_discard.c
index 572494f..286a051 100644
--- a/fs/xfs/linux-2.6/xfs_discard.c
+++ b/fs/xfs/linux-2.6/xfs_discard.c
@@ -38,7 +38,7 @@ xfs_trim_extents(
 	struct xfs_mount	*mp,
 	xfs_agnumber_t		agno,
 	xfs_fsblock_t		start,
-	xfs_fsblock_t		len,
+	xfs_fsblock_t		end,
 	xfs_fsblock_t		minlen,
 	__uint64_t		*blocks_trimmed)
 {
@@ -100,7 +100,7 @@ xfs_trim_extents(
 		 * down partially overlapping ranges for now.
 		 */
 		if (XFS_AGB_TO_FSB(mp, agno, fbno) + flen < start ||
-		    XFS_AGB_TO_FSB(mp, agno, fbno) >= start + len) {
+		    XFS_AGB_TO_FSB(mp, agno, fbno) > end) {
 			trace_xfs_discard_exclude(mp, agno, fbno, flen);
 			goto next_extent;
 		}
@@ -145,7 +145,7 @@ xfs_ioc_trim(
 	struct request_queue	*q = mp->m_ddev_targp->bt_bdev->bd_disk->queue;
 	unsigned int		granularity = q->limits.discard_granularity;
 	struct fstrim_range	range;
-	xfs_fsblock_t		start, len, minlen;
+	xfs_fsblock_t		start, end, minlen;
 	xfs_agnumber_t		start_agno, end_agno, agno;
 	__uint64_t		blocks_trimmed = 0;
 	int			error, last_error = 0;
@@ -165,19 +165,19 @@ xfs_ioc_trim(
 	 * matter as trimming blocks is an advisory interface.
 	 */
 	start = XFS_B_TO_FSBT(mp, range.start);
-	len = XFS_B_TO_FSBT(mp, range.len);
+	end = start + XFS_B_TO_FSBT(mp, range.len) - 1;
 	minlen = XFS_B_TO_FSB(mp, max_t(u64, granularity, range.minlen));
 
-	start_agno = XFS_FSB_TO_AGNO(mp, start);
-	if (start_agno >= mp->m_sb.sb_agcount)
+	if (start >= mp->m_sb.sb_dblocks)
 		return -XFS_ERROR(EINVAL);
+	if (end > mp->m_sb.sb_dblocks - 1)
+		end = mp->m_sb.sb_dblocks - 1;
 
-	end_agno = XFS_FSB_TO_AGNO(mp, start + len);
-	if (end_agno >= mp->m_sb.sb_agcount)
-		end_agno = mp->m_sb.sb_agcount - 1;
+	start_agno = XFS_FSB_TO_AGNO(mp, start);
+	end_agno = XFS_FSB_TO_AGNO(mp, end);
 
 	for (agno = start_agno; agno <= end_agno; agno++) {
-		error = -xfs_trim_extents(mp, agno, start, len, minlen,
+		error = -xfs_trim_extents(mp, agno, start, end, minlen,
 					  &blocks_trimmed);
 		if (error)
 			last_error = error;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 011/102] xfs: XFS_TRANS_SWAPEXT is not a valid flag for

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 815cb21662b914e1e14c256a3d662b1352c8509e
 xfs_trans_commit

XFS_TRANS_SWAPEXT is a transaction type, not a flag for xfs_trans_commit, so
don't pass it in xfs_swap_extents.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_dfrag.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 9a84a85..645387b 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -438,7 +438,7 @@ xfs_swap_extents(
 	if (mp->m_flags & XFS_MOUNT_WSYNC)
 		xfs_trans_set_sync(tp);
 
-	error = xfs_trans_commit(tp, XFS_TRANS_SWAPEXT);
+	error = xfs_trans_commit(tp, 0);
 
 	trace_xfs_swap_extent_after(ip, 0);
 	trace_xfs_swap_extent_after(tip, 1);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 012/102] xfs: Don't allocate new buffers on every call to _xfs_buf_find

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 3815832a2aa4df9815d15dac05227e0c8551833f

Stats show that for an 8-way unlink @ ~80,000 unlinks/s we are doing
~1 million cache hit lookups to ~3000 buffer creates. That's almost
3 orders of magnitude more cahce hits than misses, so optimising for
cache hits is quite important. In the cache hit case, we do not need
to allocate a new buffer in case of a cache miss, so we are
effectively hitting the allocator for no good reason for vast the
majority of calls to _xfs_buf_find. 8-way create workloads are
showing similar cache hit/miss ratios.

The result is profiles that look like this:

     samples  pcnt function                        DSO
     _______ _____ _______________________________ _________________

     1036.00 10.0% _xfs_buf_find                   [kernel.kallsyms]
      582.00  5.6% kmem_cache_alloc                [kernel.kallsyms]
      519.00  5.0% __memcpy                        [kernel.kallsyms]
      468.00  4.5% __ticket_spin_lock              [kernel.kallsyms]
      388.00  3.7% kmem_cache_free                 [kernel.kallsyms]
      331.00  3.2% xfs_log_commit_cil              [kernel.kallsyms]


Further, there is a fair bit of work involved in initialising a new
buffer once a cache miss has occurred and we currently do that under
the rbtree spinlock. That increases spinlock hold time on what are
heavily used trees.

To fix this, remove the initialisation of the buffer from
_xfs_buf_find() and only allocate the new buffer once we've had a
cache miss. Initialise the buffer immediately after allocating it in
xfs_buf_get, too, so that is it ready for insert if we get another
cache miss after allocation. This minimises lock hold time and
avoids unnecessary allocator churn. The resulting profiles look
like:

     samples  pcnt function                    DSO
     _______ _____ ___________________________ _________________

     8111.00  9.1% _xfs_buf_find               [kernel.kallsyms]
     4380.00  4.9% __memcpy                    [kernel.kallsyms]
     4341.00  4.8% __ticket_spin_lock          [kernel.kallsyms]
     3401.00  3.8% kmem_cache_alloc            [kernel.kallsyms]
     2856.00  3.2% xfs_log_commit_cil          [kernel.kallsyms]
     2625.00  2.9% __kmalloc                   [kernel.kallsyms]
     2380.00  2.7% kfree                       [kernel.kallsyms]
     2016.00  2.3% kmem_cache_free             [kernel.kallsyms]

Showing a significant reduction in time spent doing allocation and
freeing from slabs (kmem_cache_alloc and kmem_cache_free).

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   48 ++++++++++++++++++++++++++------------------
 1 file changed, 28 insertions(+), 20 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 5e68099..6a2ab95 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -481,8 +481,6 @@ _xfs_buf_find(
 
 	/* No match found */
 	if (new_bp) {
-		_xfs_buf_initialize(new_bp, btp, range_base,
-				range_length, flags);
 		rb_link_node(&new_bp->b_rbnode, parent, rbp);
 		rb_insert_color(&new_bp->b_rbnode, &pag->pag_buf_tree);
 		/* the buffer keeps the perag reference until it is freed */
@@ -527,35 +525,53 @@ found:
 }
 
 /*
- *	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.
+ * Assembles a buffer covering the specified range. The code is optimised for
+ * cache hits, as metadata intensive workloads will see 3 orders of magnitude
+ * more hits than misses.
  */
-xfs_buf_t *
+struct xfs_buf *
 xfs_buf_get(
 	xfs_buftarg_t		*target,/* target for buffer		*/
 	xfs_off_t		ioff,	/* starting offset of range	*/
 	size_t			isize,	/* length of range		*/
 	xfs_buf_flags_t		flags)
 {
-	xfs_buf_t		*bp, *new_bp;
+	struct xfs_buf		*bp;
+	struct xfs_buf		*new_bp;
 	int			error = 0;
 
+	bp = _xfs_buf_find(target, ioff, isize, flags, NULL);
+	if (likely(bp))
+		goto found;
+
 	new_bp = xfs_buf_allocate(flags);
 	if (unlikely(!new_bp))
 		return NULL;
 
+	_xfs_buf_initialize(new_bp, target,
+			    ioff << BBSHIFT, isize << BBSHIFT, flags);
+
 	bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
+	if (!bp) {
+		xfs_buf_deallocate(new_bp);
+		return NULL;
+	}
+
 	if (bp == new_bp) {
 		error = xfs_buf_allocate_memory(bp, flags);
 		if (error)
 			goto no_buffer;
-	} else {
+	} else
 		xfs_buf_deallocate(new_bp);
-		if (unlikely(bp == NULL))
-			return NULL;
-	}
 
+	/*
+	 * Now we have a workable buffer, fill in the block number so
+	 * that we can do IO on it.
+	 */
+	bp->b_bn = ioff;
+	bp->b_count_desired = bp->b_buffer_length;
+
+found:
 	if (!(bp->b_flags & XBF_MAPPED)) {
 		error = _xfs_buf_map_pages(bp, flags);
 		if (unlikely(error)) {
@@ -566,18 +582,10 @@ xfs_buf_get(
 	}
 
 	XFS_STATS_INC(xb_get);
-
-	/*
-	 * Always fill in the block number now, the mapped cases can do
-	 * their own overlay of this later.
-	 */
-	bp->b_bn = ioff;
-	bp->b_count_desired = bp->b_buffer_length;
-
 	trace_xfs_buf_get(bp, flags, _RET_IP_);
 	return bp;
 
- no_buffer:
+no_buffer:
 	if (flags & (XBF_LOCK | XBF_TRYLOCK))
 		xfs_buf_unlock(bp);
 	xfs_buf_rele(bp);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 013/102] xfs: reduce the number of log forces from tail pushing

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 670ce93fef93bba8c8a422a79747385bec8e846a

The AIL push code will issue a log force on ever single push loop
that it exits and has encountered pinned items. It doesn't rescan
these pinned items until it revisits the AIL from the start. Hence
we only need to force the log once per walk from the start of the
AIL to the target LSN.

This results in numbers like this:

	xs_push_ail_flush.....         1456
	xs_log_force.........          1485

For an 8-way 50M inode create workload - almost all the log forces
are coming from the AIL pushing code.

Reduce the number of log forces by only forcing the log if the
previous walk found pinned buffers. This reduces the numbers to:

	xs_push_ail_flush.....          665
	xs_log_force.........           682

For the same test.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_trans_ail.c  |   35 +++++++++++++++++++++--------------
 fs/xfs/xfs_trans_priv.h |    1 +
 2 files changed, 22 insertions(+), 14 deletions(-)

diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index a4c281b..3949a5e 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -414,12 +414,24 @@ xfsaild_push(
 	xfs_lsn_t		lsn;
 	xfs_lsn_t		target;
 	long			tout = 10;
-	int			flush_log = 0;
 	int			stuck = 0;
 	int			count = 0;
 	int			push_xfsbufd = 0;
 
+	/*
+	 * If last time we ran we encountered pinned items, force the log first
+	 * and wait for it before pushing again.
+	 */
 	spin_lock(&ailp->xa_lock);
+	if (ailp->xa_last_pushed_lsn == 0 && ailp->xa_log_flush &&
+	    !list_empty(&ailp->xa_ail)) {
+		ailp->xa_log_flush = 0;
+		spin_unlock(&ailp->xa_lock);
+		XFS_STATS_INC(xs_push_ail_flush);
+		xfs_log_force(mp, XFS_LOG_SYNC);
+		spin_lock(&ailp->xa_lock);
+	}
+
 	target = ailp->xa_target;
 	xfs_trans_ail_cursor_init(ailp, cur);
 	lip = xfs_trans_ail_cursor_first(ailp, cur, ailp->xa_last_pushed_lsn);
@@ -473,7 +485,7 @@ xfsaild_push(
 
 			if (!IOP_PUSHBUF(lip)) {
 				stuck++;
-				flush_log = 1;
+				ailp->xa_log_flush++;
 			} else {
 				ailp->xa_last_pushed_lsn = lsn;
 			}
@@ -483,7 +495,7 @@ xfsaild_push(
 		case XFS_ITEM_PINNED:
 			XFS_STATS_INC(xs_push_ail_pinned);
 			stuck++;
-			flush_log = 1;
+			ailp->xa_log_flush++;
 			break;
 
 		case XFS_ITEM_LOCKED:
@@ -527,16 +539,6 @@ xfsaild_push(
 	xfs_trans_ail_cursor_done(ailp, cur);
 	spin_unlock(&ailp->xa_lock);
 
-	if (flush_log) {
-		/*
-		 * If something we need to push out was pinned, then
-		 * push out the log so it will become unpinned and
-		 * move forward in the AIL.
-		 */
-		XFS_STATS_INC(xs_push_ail_flush);
-		xfs_log_force(mp, 0);
-	}
-
 	if (push_xfsbufd) {
 		/* we've got delayed write buffers to flush */
 		wake_up_process(mp->m_ddev_targp->bt_task);
@@ -547,6 +549,7 @@ out_done:
 	if (!count) {
 		/* We're past our target or empty, so idle */
 		ailp->xa_last_pushed_lsn = 0;
+		ailp->xa_log_flush = 0;
 
 		tout = 50;
 	} else if (XFS_LSN_CMP(lsn, target) >= 0) {
@@ -565,9 +568,13 @@ out_done:
 		 * were stuck.
 		 *
 		 * Backoff a bit more to allow some I/O to complete before
-		 * continuing from where we were.
+		 * restarting from the start of the AIL. This prevents us
+		 * from spinning on the same items, and if they are pinned will
+		 * all the restart to issue a log force to unpin the stuck
+		 * items.
 		 */
 		tout = 20;
+		ailp->xa_last_pushed_lsn = 0;
 	}
 
 	return tout;
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
index fe2e3cb..8c0c465 100644
--- a/fs/xfs/xfs_trans_priv.h
+++ b/fs/xfs/xfs_trans_priv.h
@@ -70,6 +70,7 @@ struct xfs_ail {
 	struct xfs_ail_cursor	xa_cursors;
 	spinlock_t		xa_lock;
 	xfs_lsn_t		xa_last_pushed_lsn;
+	int			xa_log_flush;
 };
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 014/102] xfs: optimize fsync on directories

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 1da2f2dbf2d2aaa1b0f6ca2f61fcf07e24eb659b

Directories are only updated transactionally, which means fsync only
needs to flush the log the inode is currently dirty, but not bother
with checking for dirty data, non-transactional updates, and most
importanly doesn't have to flush disk caches except as part of a
transaction commit.

While the first two optimizations can't easily be measured, the
latter actually makes a difference when doing lots of fsync that do
not actually have to commit the inode, e.g. because an earlier fsync
already pushed the log far enough.

The new xfs_dir_fsync is identical to xfs_nfs_commit_metadata except
for the prototype, but I'm not sure creating a common helper for the
two is worth it given how simple the functions are.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c  |   29 ++++++++++++++++++++++++++++-
 fs/xfs/linux-2.6/xfs_trace.h |    1 +
 2 files changed, 29 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 67d392a..185a337 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -124,6 +124,33 @@ xfs_iozero(
 	return (-status);
 }
 
+/*
+ * Fsync operations on directories are much simpler than on regular files,
+ * as there is no file data to flush, and thus also no need for explicit
+ * cache flush operations, and there are no non-transaction metadata updates
+ * on directories either.
+ */
+STATIC int
+xfs_dir_fsync(
+	struct file		*file,
+	int			datasync)
+{
+	struct xfs_inode	*ip = XFS_I(file->f_mapping->host);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_lsn_t		lsn = 0;
+
+	trace_xfs_dir_fsync(ip);
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	if (xfs_ipincount(ip))
+		lsn = ip->i_itemp->ili_last_lsn;
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	if (!lsn)
+		return 0;
+	return _xfs_log_force_lsn(mp, lsn, XFS_LOG_SYNC, NULL);
+}
+
 STATIC int
 xfs_file_fsync(
 	struct file		*file,
@@ -1141,7 +1168,7 @@ const struct file_operations xfs_dir_file_operations = {
 #ifdef CONFIG_COMPAT
 	.compat_ioctl	= xfs_file_compat_ioctl,
 #endif
-	.fsync		= xfs_file_fsync,
+	.fsync		= xfs_dir_fsync,
 };
 
 static const struct vm_operations_struct xfs_file_vm_ops = {
diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index d48b7a5..fae13a0 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -577,6 +577,7 @@ DEFINE_INODE_EVENT(xfs_vm_bmap);
 DEFINE_INODE_EVENT(xfs_file_ioctl);
 DEFINE_INODE_EVENT(xfs_file_compat_ioctl);
 DEFINE_INODE_EVENT(xfs_ioctl_setattr);
+DEFINE_INODE_EVENT(xfs_dir_fsync);
 DEFINE_INODE_EVENT(xfs_file_fsync);
 DEFINE_INODE_EVENT(xfs_destroy_inode);
 DEFINE_INODE_EVENT(xfs_write_inode);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 015/102] xfs: clean up buffer allocation

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 4347b9d7ad4223474d315c3ab6bc1ce7cce7fa2d

Change _xfs_buf_initialize to allocate the buffer directly and rename it to
xfs_buf_alloc now that is the only buffer allocation routine.  Also remove
the xfs_buf_deallocate wrapper around the kmem_zone_free calls for buffers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   50 ++++++++++++++++----------------------------
 fs/xfs/linux-2.6/xfs_buf.h |    3 ++-
 fs/xfs/xfs_log.c           |    2 +-
 3 files changed, 21 insertions(+), 34 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 6a2ab95..449ba57 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -66,10 +66,6 @@ struct workqueue_struct *xfsconvertd_workqueue;
 #define xb_to_km(flags) \
 	 (((flags) & XBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
 
-#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));
 
 static inline int
 xfs_buf_is_vmapped(
@@ -167,14 +163,19 @@ xfs_buf_stale(
 	ASSERT(atomic_read(&bp->b_hold) >= 1);
 }
 
-STATIC void
-_xfs_buf_initialize(
-	xfs_buf_t		*bp,
-	xfs_buftarg_t		*target,
+struct xfs_buf *
+xfs_buf_alloc(
+	struct xfs_buftarg	*target,
 	xfs_off_t		range_base,
 	size_t			range_length,
 	xfs_buf_flags_t		flags)
 {
+	struct xfs_buf		*bp;
+
+	bp = kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags));
+	if (unlikely(!bp))
+		return NULL;
+
 	/*
 	 * We don't want certain flags to appear in b_flags.
 	 */
@@ -203,8 +204,9 @@ _xfs_buf_initialize(
 	init_waitqueue_head(&bp->b_waiters);
 
 	XFS_STATS_INC(xb_create);
-
 	trace_xfs_buf_init(bp, _RET_IP_);
+
+	return bp;
 }
 
 /*
@@ -277,7 +279,7 @@ xfs_buf_free(
 	} else if (bp->b_flags & _XBF_KMEM)
 		kmem_free(bp->b_addr);
 	_xfs_buf_free_pages(bp);
-	xfs_buf_deallocate(bp);
+	kmem_zone_free(xfs_buf_zone, bp);
 }
 
 /*
@@ -544,16 +546,14 @@ xfs_buf_get(
 	if (likely(bp))
 		goto found;
 
-	new_bp = xfs_buf_allocate(flags);
+	new_bp = xfs_buf_alloc(target, ioff << BBSHIFT, isize << BBSHIFT,
+			       flags);
 	if (unlikely(!new_bp))
 		return NULL;
 
-	_xfs_buf_initialize(new_bp, target,
-			    ioff << BBSHIFT, isize << BBSHIFT, flags);
-
 	bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
 	if (!bp) {
-		xfs_buf_deallocate(new_bp);
+		kmem_zone_free(xfs_buf_zone, new_bp);
 		return NULL;
 	}
 
@@ -562,7 +562,7 @@ xfs_buf_get(
 		if (error)
 			goto no_buffer;
 	} else
-		xfs_buf_deallocate(new_bp);
+		kmem_zone_free(xfs_buf_zone, new_bp);
 
 	/*
 	 * Now we have a workable buffer, fill in the block number so
@@ -703,19 +703,6 @@ xfs_buf_read_uncached(
 	return bp;
 }
 
-xfs_buf_t *
-xfs_buf_get_empty(
-	size_t			len,
-	xfs_buftarg_t		*target)
-{
-	xfs_buf_t		*bp;
-
-	bp = xfs_buf_allocate(0);
-	if (bp)
-		_xfs_buf_initialize(bp, target, 0, len, 0);
-	return bp;
-}
-
 /*
  * Return a buffer allocated as an empty buffer and associated to external
  * memory via xfs_buf_associate_memory() back to it's empty state.
@@ -801,10 +788,9 @@ xfs_buf_get_uncached(
 	int			error, i;
 	xfs_buf_t		*bp;
 
-	bp = xfs_buf_allocate(0);
+	bp = xfs_buf_alloc(target, 0, len, 0);
 	if (unlikely(bp == NULL))
 		goto fail;
-	_xfs_buf_initialize(bp, target, 0, len, 0);
 
 	error = _xfs_buf_get_pages(bp, page_count, 0);
 	if (error)
@@ -834,7 +820,7 @@ xfs_buf_get_uncached(
 		__free_page(bp->b_pages[i]);
 	_xfs_buf_free_pages(bp);
  fail_free_buf:
-	xfs_buf_deallocate(bp);
+	kmem_zone_free(xfs_buf_zone, bp);
  fail:
 	return NULL;
 }
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index 36d6ee4..ee3714b 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -177,7 +177,8 @@ extern xfs_buf_t *xfs_buf_get(xfs_buftarg_t *, xfs_off_t, size_t,
 extern xfs_buf_t *xfs_buf_read(xfs_buftarg_t *, xfs_off_t, size_t,
 				xfs_buf_flags_t);
 
-extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
+struct xfs_buf *xfs_buf_alloc(struct xfs_buftarg *, xfs_off_t, size_t,
+			      xfs_buf_flags_t);
 extern void xfs_buf_set_empty(struct xfs_buf *bp, size_t len);
 extern xfs_buf_t *xfs_buf_get_uncached(struct xfs_buftarg *, size_t, int);
 extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 41d5b8f..8700411 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -1053,7 +1053,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	xlog_get_iclog_buffer_size(mp, log);
 
 	error = ENOMEM;
-	bp = xfs_buf_get_empty(log->l_iclog_size, mp->m_logdev_targp);
+	bp = xfs_buf_alloc(mp->m_logdev_targp, 0, log->l_iclog_size, 0);
 	if (!bp)
 		goto out_free_log;
 	XFS_BUF_SET_IODONE_FUNC(bp, xlog_iodone);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 016/102] xfs: clean up xfs_ioerror_alert

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 901796afca0d31d97bf6d1bf2ab251a93a4b8c83

Instead of passing the block number and mount structure explicitly
get them off the bp and fix make the argument order more natural.

Also move it to xfs_buf.c and stop printing the device name given
that we already get the fs name as part of xfs_alert, and we know
what device is operates on because of the caller that gets printed,
finally rename it to xfs_buf_ioerror_alert and pass __func__ as
argument where it makes sense.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   11 +++++++++++
 fs/xfs/linux-2.6/xfs_buf.h |    1 +
 fs/xfs/xfs_log.c           |   14 +++++++-------
 fs/xfs/xfs_log_recover.c   |   25 +++++++++----------------
 fs/xfs/xfs_mount.c         |    3 +--
 fs/xfs/xfs_rw.c            |   20 +-------------------
 fs/xfs/xfs_rw.h            |    2 --
 fs/xfs/xfs_trans_buf.c     |    9 +++------
 fs/xfs/xfs_vnodeops.c      |   11 +++++------
 9 files changed, 38 insertions(+), 58 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 449ba57..82fe480 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -1028,6 +1028,17 @@ xfs_buf_ioerror(
 	trace_xfs_buf_ioerror(bp, error, _RET_IP_);
 }
 
+void
+xfs_buf_ioerror_alert(
+	struct xfs_buf		*bp,
+	const char		*func)
+{
+	xfs_alert(bp->b_target->bt_mount,
+"metadata I/O error: block 0x%llx (\"%s\") error %d buf count %zd",
+		(__uint64_t)XFS_BUF_ADDR(bp), func,
+		bp->b_error, XFS_BUF_COUNT(bp));
+}
+
 int
 xfs_bwrite(
 	struct xfs_mount	*mp,
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index ee3714b..71d6698 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -207,6 +207,7 @@ extern int xfs_bdstrat_cb(struct xfs_buf *);
 
 extern void xfs_buf_ioend(xfs_buf_t *,	int);
 extern void xfs_buf_ioerror(xfs_buf_t *, int);
+extern void xfs_buf_ioerror_alert(struct xfs_buf *, const char *func);
 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, void *,
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 8700411..b180b55 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -886,7 +886,7 @@ xlog_iodone(xfs_buf_t *bp)
 	 */
 	if (XFS_TEST_ERROR((XFS_BUF_GETERROR(bp)), l->l_mp,
 			XFS_ERRTAG_IODONE_IOERR, XFS_RANDOM_IODONE_IOERR)) {
-		xfs_ioerror_alert("xlog_iodone", l->l_mp, bp, XFS_BUF_ADDR(bp));
+		xfs_buf_ioerror_alert(bp, __func__);
 		XFS_BUF_STALE(bp);
 		xfs_force_shutdown(l->l_mp, SHUTDOWN_LOG_IO_ERROR);
 		/*
@@ -1397,9 +1397,9 @@ xlog_sync(xlog_t		*log,
 	 */
 	XFS_BUF_WRITE(bp);
 
-	if ((error = xlog_bdstrat(bp))) {
-		xfs_ioerror_alert("xlog_sync", log->l_mp, bp,
-				  XFS_BUF_ADDR(bp));
+	error = xlog_bdstrat(bp);
+	if (error) {
+		xfs_buf_ioerror_alert(bp, "xlog_sync");
 		return error;
 	}
 	if (split) {
@@ -1437,9 +1437,9 @@ xlog_sync(xlog_t		*log,
 		/* account for internal log which doesn't start at block #0 */
 		XFS_BUF_SET_ADDR(bp, XFS_BUF_ADDR(bp) + log->l_logBBstart);
 		XFS_BUF_WRITE(bp);
-		if ((error = xlog_bdstrat(bp))) {
-			xfs_ioerror_alert("xlog_sync (split)", log->l_mp,
-					  bp, XFS_BUF_ADDR(bp));
+		error = xlog_bdstrat(bp);
+		if (error) {
+			xfs_buf_ioerror_alert(bp, "xlog_sync (split)");
 			return error;
 		}
 	}
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index b75fd67..dfd872a 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -181,8 +181,7 @@ xlog_bread_noalign(
 	xfsbdstrat(log->l_mp, bp);
 	error = xfs_buf_iowait(bp);
 	if (error)
-		xfs_ioerror_alert("xlog_bread", log->l_mp,
-				  bp, XFS_BUF_ADDR(bp));
+		xfs_buf_ioerror_alert(bp, __func__);
 	return error;
 }
 
@@ -268,9 +267,9 @@ xlog_bwrite(
 	XFS_BUF_SET_COUNT(bp, BBTOB(nbblks));
 	XFS_BUF_SET_TARGET(bp, log->l_mp->m_logdev_targp);
 
-	if ((error = xfs_bwrite(log->l_mp, bp)))
-		xfs_ioerror_alert("xlog_bwrite", log->l_mp,
-				  bp, XFS_BUF_ADDR(bp));
+	error = xfs_bwrite(log->l_mp, bp);
+	if (error)
+		xfs_buf_ioerror_alert(bp, __func__);
 	return error;
 }
 
@@ -361,9 +360,7 @@ xlog_recover_iodone(
 		 * We're not going to bother about retrying
 		 * this during recovery. One strike!
 		 */
-		xfs_ioerror_alert("xlog_recover_iodone",
-					bp->b_target->bt_mount, bp,
-					XFS_BUF_ADDR(bp));
+		xfs_buf_ioerror_alert(bp, __func__);
 		xfs_force_shutdown(bp->b_target->bt_mount,
 					SHUTDOWN_META_IO_ERROR);
 	}
@@ -2132,8 +2129,7 @@ xlog_recover_buffer_pass2(
 	bp = xfs_buf_read(mp->m_ddev_targp, buf_f->blf_blkno, buf_f->blf_len,
 			  buf_flags);
 	if (XFS_BUF_ISERROR(bp)) {
-		xfs_ioerror_alert("xlog_recover_do..(read#1)", mp,
-				  bp, buf_f->blf_blkno);
+		xfs_buf_ioerror_alert(bp, __func__);
 		error = XFS_BUF_GETERROR(bp);
 		xfs_buf_relse(bp);
 		return error;
@@ -2224,8 +2220,7 @@ xlog_recover_inode_pass2(
 	bp = xfs_buf_read(mp->m_ddev_targp, in_f->ilf_blkno, in_f->ilf_len,
 			  XBF_LOCK);
 	if (XFS_BUF_ISERROR(bp)) {
-		xfs_ioerror_alert("xlog_recover_do..(read#2)", mp,
-				  bp, in_f->ilf_blkno);
+		xfs_buf_ioerror_alert(bp, __func__);
 		error = XFS_BUF_GETERROR(bp);
 		xfs_buf_relse(bp);
 		goto error;
@@ -2533,8 +2528,7 @@ xlog_recover_dquot_pass2(
 			     XFS_FSB_TO_BB(mp, dq_f->qlf_len),
 			     0, &bp);
 	if (error) {
-		xfs_ioerror_alert("xlog_recover_do..(read#3)", mp,
-				  bp, dq_f->qlf_blkno);
+		xfs_buf_ioerror_alert(bp, "xlog_recover_do..(read#3)");
 		return error;
 	}
 	ASSERT(bp);
@@ -3674,8 +3668,7 @@ xlog_do_recover(
 	xfsbdstrat(log->l_mp, bp);
 	error = xfs_buf_iowait(bp);
 	if (error) {
-		xfs_ioerror_alert("xlog_do_recover",
-				  log->l_mp, bp, XFS_BUF_ADDR(bp));
+		xfs_buf_ioerror_alert(bp, __func__);
 		ASSERT(0);
 		xfs_buf_relse(bp);
 		return error;
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 9afdd49..3effa7f 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -1626,8 +1626,7 @@ xfs_unmountfs_writesb(xfs_mount_t *mp)
 		xfsbdstrat(mp, sbp);
 		error = xfs_buf_iowait(sbp);
 		if (error)
-			xfs_ioerror_alert("xfs_unmountfs_writesb",
-					  mp, sbp, XFS_BUF_ADDR(sbp));
+			xfs_buf_ioerror_alert(sbp, __func__);
 		xfs_buf_relse(sbp);
 	}
 	return error;
diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
index d6d6fdf..47188fa 100644
--- a/fs/xfs/xfs_rw.c
+++ b/fs/xfs/xfs_rw.c
@@ -92,24 +92,6 @@ xfs_do_force_shutdown(
 }
 
 /*
- * Prints out an ALERT message about I/O error.
- */
-void
-xfs_ioerror_alert(
-	char			*func,
-	struct xfs_mount	*mp,
-	xfs_buf_t		*bp,
-	xfs_daddr_t		blkno)
-{
-	xfs_alert(mp,
-		 "I/O error occurred: meta-data dev %s block 0x%llx"
-		 "       (\"%s\") error %d buf count %zd",
-		XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
-		(__uint64_t)blkno, func,
-		XFS_BUF_GETERROR(bp), XFS_BUF_COUNT(bp));
-}
-
-/*
  * This isn't an absolute requirement, but it is
  * just a good idea to call xfs_read_buf instead of
  * directly doing a read_buf call. For one, we shouldn't
@@ -143,7 +125,7 @@ xfs_read_buf(
 	} else {
 		*bpp = NULL;
 		if (error) {
-			xfs_ioerror_alert("xfs_read_buf", mp, bp, XFS_BUF_ADDR(bp));
+			xfs_buf_ioerror_alert(bp, __func__);
 		} else {
 			error = XFS_ERROR(EIO);
 		}
diff --git a/fs/xfs/xfs_rw.h b/fs/xfs/xfs_rw.h
index 11c41ec..bbdb9ad 100644
--- a/fs/xfs/xfs_rw.h
+++ b/fs/xfs/xfs_rw.h
@@ -42,8 +42,6 @@ xfs_fsb_to_db(struct xfs_inode *ip, xfs_fsblock_t fsb)
 extern int xfs_read_buf(struct xfs_mount *mp, xfs_buftarg_t *btp,
 			xfs_daddr_t blkno, int len, uint flags,
 			struct xfs_buf **bpp);
-extern void xfs_ioerror_alert(char *func, struct xfs_mount *mp,
-				xfs_buf_t *bp, xfs_daddr_t blkno);
 extern xfs_extlen_t xfs_get_extsz_hint(struct xfs_inode *ip);
 
 #endif /* __XFS_RW_H__ */
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c
index 03b3b7f..26abce5 100644
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -294,8 +294,7 @@ xfs_trans_read_buf(
 					EAGAIN : XFS_ERROR(ENOMEM);
 
 		if (XFS_BUF_GETERROR(bp) != 0) {
-			xfs_ioerror_alert("xfs_trans_read_buf", mp,
-					  bp, blkno);
+			xfs_buf_ioerror_alert(bp, __func__);
 			error = XFS_BUF_GETERROR(bp);
 			xfs_buf_relse(bp);
 			return error;
@@ -338,8 +337,7 @@ xfs_trans_read_buf(
 			xfsbdstrat(tp->t_mountp, bp);
 			error = xfs_buf_iowait(bp);
 			if (error) {
-				xfs_ioerror_alert("xfs_trans_read_buf", mp,
-						  bp, blkno);
+				xfs_buf_ioerror_alert(bp, __func__);
 				xfs_buf_relse(bp);
 				/*
 				 * We can gracefully recover from most read
@@ -390,8 +388,7 @@ xfs_trans_read_buf(
 	    XFS_BUF_SUPER_STALE(bp);
 		error = XFS_BUF_GETERROR(bp);
 
-		xfs_ioerror_alert("xfs_trans_read_buf", mp,
-				  bp, blkno);
+		xfs_buf_ioerror_alert(bp, __func__);
 		if (tp->t_flags & XFS_TRANS_DIRTY)
 			xfs_force_shutdown(tp->t_mountp, SHUTDOWN_META_IO_ERROR);
 		xfs_buf_relse(bp);
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 154223c..44f780c 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -509,8 +509,7 @@ xfs_readlink_bmap(
 				  XBF_LOCK | XBF_MAPPED | XBF_DONT_BLOCK);
 		error = XFS_BUF_GETERROR(bp);
 		if (error) {
-			xfs_ioerror_alert("xfs_readlink",
-				  ip->i_mount, bp, XFS_BUF_ADDR(bp));
+			xfs_buf_ioerror_alert(bp, __func__);
 			xfs_buf_relse(bp);
 			goto out;
 		}
@@ -2468,8 +2467,8 @@ xfs_zero_remaining_bytes(
 		xfsbdstrat(mp, bp);
 		error = xfs_buf_iowait(bp);
 		if (error) {
-			xfs_ioerror_alert("xfs_zero_remaining_bytes(read)",
-					  mp, bp, XFS_BUF_ADDR(bp));
+			xfs_buf_ioerror_alert(bp,
+					"xfs_zero_remaining_bytes(read)");
 			break;
 		}
 		memset(XFS_BUF_PTR(bp) +
@@ -2481,8 +2480,8 @@ xfs_zero_remaining_bytes(
 		xfsbdstrat(mp, bp);
 		error = xfs_buf_iowait(bp);
 		if (error) {
-			xfs_ioerror_alert("xfs_zero_remaining_bytes(write)",
-					  mp, bp, XFS_BUF_ADDR(bp));
+			xfs_buf_ioerror_alert(bp,
+					"xfs_zero_remaining_bytes(write)");
 			break;
 		}
 	}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 017/102] xfs: use xfs_ioerror_alert in xfs_buf_iodone_callbacks

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: b38505b09b7854d446b2f60b4414e3231277aa1a

Use xfs_ioerror_alert instead of opencoding a very similar error
message.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/xfs_buf_item.c |    4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
index 7888a75..aedebfb 100644
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -986,9 +986,7 @@ xfs_buf_iodone_callbacks(
 	if (XFS_BUF_TARGET(bp) != lasttarg ||
 	    time_after(jiffies, (lasttime + 5*HZ))) {
 		lasttime = jiffies;
-		xfs_alert(mp, "Device %s: metadata write error block 0x%llx",
-			XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
-		      (__uint64_t)XFS_BUF_ADDR(bp));
+		xfs_buf_ioerror_alert(bp, __func__);
 	}
 	lasttarg = XFS_BUF_TARGET(bp);
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 018/102] xfs: do not flush data workqueues in xfs_flush_buftarg

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 5a93a064d27b42e4af1772b0599b53e3241191ac

When we call xfs_flush_buftarg (generally from sync or umount) it already
is too late to flush the data workqueues, as I/O completion is signalled
for them and we are thus already done with the data we would flush here.

There are places where flushing them might be useful, but the current
sync interface doesn't give us that opportunity.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 82fe480..4508ec2 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -1670,13 +1670,6 @@ xfs_buf_delwri_promote(
 	spin_unlock(&btp->bt_delwrite_lock);
 }
 
-STATIC void
-xfs_buf_runall_queues(
-	struct workqueue_struct	*queue)
-{
-	flush_workqueue(queue);
-}
-
 /*
  * Move as many buffers as specified to the supplied list
  * idicating if we skipped any buffers to prevent deadlocks.
@@ -1811,9 +1804,7 @@ xfs_flush_buftarg(
 	LIST_HEAD(wait_list);
 	struct blk_plug plug;
 
-	xfs_buf_runall_queues(xfsconvertd_workqueue);
-	xfs_buf_runall_queues(xfsdatad_workqueue);
-	xfs_buf_runall_queues(xfslogd_workqueue);
+	flush_workqueue(xfslogd_workqueue);
 
 	set_bit(XBT_FORCE_FLUSH, &target->bt_flags);
 	pincount = xfs_buf_delwri_split(target, &tmp_list, 0);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 019/102] xfs: add AIL pushing tracepoints

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 9e4c109ac822395e0aae650e4e3c9e4903f6602f

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   37 +++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_trans_ail.c       |    8 ++++++++
 2 files changed, 45 insertions(+)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index fae13a0..b5f892a 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -30,6 +30,7 @@ struct xfs_buf_log_item;
 struct xfs_da_args;
 struct xfs_da_node_entry;
 struct xfs_dquot;
+struct xfs_log_item;
 struct xlog_ticket;
 struct log;
 struct xlog_recover;
@@ -854,6 +855,42 @@ DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_sub);
 
+DECLARE_EVENT_CLASS(xfs_log_item_class,
+	TP_PROTO(struct xfs_log_item *lip),
+	TP_ARGS(lip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(void *, lip)
+		__field(uint, type)
+		__field(uint, flags)
+		__field(xfs_lsn_t, lsn)
+	),
+	TP_fast_assign(
+		__entry->dev = lip->li_mountp->m_super->s_dev;
+		__entry->lip = lip;
+		__entry->type = lip->li_type;
+		__entry->flags = lip->li_flags;
+		__entry->lsn = lip->li_lsn;
+	),
+	TP_printk("dev %d:%d lip 0x%p lsn %d/%d type %s flags %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->lip,
+		  CYCLE_LSN(__entry->lsn), BLOCK_LSN(__entry->lsn),
+		  __print_symbolic(__entry->type, XFS_LI_TYPE_DESC),
+		  __print_flags(__entry->flags, "|", XFS_LI_FLAGS))
+)
+
+#define DEFINE_LOG_ITEM_EVENT(name) \
+DEFINE_EVENT(xfs_log_item_class, name, \
+	TP_PROTO(struct xfs_log_item *lip), \
+	TP_ARGS(lip))
+DEFINE_LOG_ITEM_EVENT(xfs_ail_push);
+DEFINE_LOG_ITEM_EVENT(xfs_ail_pushbuf);
+DEFINE_LOG_ITEM_EVENT(xfs_ail_pushbuf_pinned);
+DEFINE_LOG_ITEM_EVENT(xfs_ail_pinned);
+DEFINE_LOG_ITEM_EVENT(xfs_ail_locked);
+
+
 DECLARE_EVENT_CLASS(xfs_file_class,
 	TP_PROTO(struct xfs_inode *ip, size_t count, loff_t offset, int flags),
 	TP_ARGS(ip, count, offset, flags),
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index 3949a5e..409ac37 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -26,6 +26,7 @@
 #include "xfs_ag.h"
 #include "xfs_mount.h"
 #include "xfs_trans_priv.h"
+#include "xfs_trace.h"
 #include "xfs_error.h"
 
 #ifdef DEBUG
@@ -476,14 +477,18 @@ xfsaild_push(
 		switch (lock_result) {
 		case XFS_ITEM_SUCCESS:
 			XFS_STATS_INC(xs_push_ail_success);
+			trace_xfs_ail_push(lip);
+
 			IOP_PUSH(lip);
 			ailp->xa_last_pushed_lsn = lsn;
 			break;
 
 		case XFS_ITEM_PUSHBUF:
 			XFS_STATS_INC(xs_push_ail_pushbuf);
+			trace_xfs_ail_pushbuf(lip);
 
 			if (!IOP_PUSHBUF(lip)) {
+				trace_xfs_ail_pushbuf_pinned(lip);
 				stuck++;
 				ailp->xa_log_flush++;
 			} else {
@@ -494,12 +499,15 @@ xfsaild_push(
 
 		case XFS_ITEM_PINNED:
 			XFS_STATS_INC(xs_push_ail_pinned);
+			trace_xfs_ail_pinned(lip);
+
 			stuck++;
 			ailp->xa_log_flush++;
 			break;
 
 		case XFS_ITEM_LOCKED:
 			XFS_STATS_INC(xs_push_ail_locked);
+			trace_xfs_ail_locked(lip);
 			stuck++;
 			break;
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Dave Chinner]

From: Mel Gorman <mgorman@suse.de>

Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4

Direct reclaim should never writeback pages.  For now, handle the
situation and warn about it.  Ultimately, this will be a BUG_ON.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Alex Elder <aelder@sgi.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
---
 fs/xfs/linux-2.6/xfs_aops.c |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index f16207b..4ff2404 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -943,11 +943,11 @@ xfs_vm_writepage(
 	 * random callers for direct reclaim or memcg reclaim.  We explicitly
 	 * allow reclaim from kswapd as the stack usage there is relatively low.
 	 *
-	 * This should really be done by the core VM, but until that happens
-	 * filesystems like XFS, btrfs and ext4 have to take care of this
-	 * by themselves.
+	 * This should never happen except in the case of a VM regression so
+	 * warn about it.
 	 */
-	if ((current->flags & (PF_MEMALLOC|PF_KSWAPD)) == PF_MEMALLOC)
+	if (WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD)) ==
+			PF_MEMALLOC))
 		goto redirty;
 
 	/*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 021/102] xfs: fix force shutdown handling in xfs_end_io

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 810627d9a6d0e8820c798001875bc4e1b7754ebf

Ensure ioend->io_error gets propagated back to e.g. AIO completions.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Alex Elder <aelder@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 4ff2404..2426ecd 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -222,7 +222,7 @@ xfs_end_io(
 	int		error = 0;
 
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-		error = -EIO;
+		ioend->io_error = -EIO;
 		goto done;
 	}
 	if (ioend->io_error)
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 022/102] xfs: fix allocation length overflow in xfs_bmapi_write()

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: a99ebf43f49f4499ab0e2a8a9132ad6ed6ba2409

When testing the new xfstests --large-fs option that does very large
file preallocations, this assert was tripped deep in
xfs_alloc_vextent():

XFS: Assertion failed: args->minlen <= args->maxlen, file: fs/xfs/xfs_alloc.c, line: 2239

The allocation was trying to allocate a zero length extent because
the lower 32 bits of the allocation length was zero. The remaining
length of the allocation to be done was an exact multiple of 2^32 -
the first case I saw was at 496TB remaining to be allocated.

This turns out to be an overflow when converting the allocation
length (a 64 bit quantity) into the extent length to allocate (a 32
bit quantity), and it requires the length to be allocated an exact
multiple of 2^32 blocks to trip the assert.

Fix it by limiting the extent lenth to allocate to MAXEXTLEN.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_bmap.c |   14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 8193ee5..05ea68f 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -4454,8 +4454,18 @@ xfs_bmapi(
 					xfs_bmbt_get_all(ep, &prev);
 				}
 			} else {
-				alen = (xfs_extlen_t)
-					XFS_FILBLKS_MIN(len, MAXEXTLEN);
+				/*
+				 * There's a 32/64 bit type mismatch between the
+				 * allocation length request (which can be 64
+				 * bits in length) and the bma length request,
+				 * which is xfs_extlen_t and therefore 32 bits.
+				 * Hence we have to check for 32-bit overflows
+				 * and handle them here.
+				 */
+				if (len > (xfs_filblks_t)MAXEXTLEN)
+					alen = MAXEXTLEN;
+				else
+					alen = len;
 				if (!eof)
 					alen = (xfs_extlen_t)
 						XFS_FILBLKS_MIN(alen,
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 023/102] xfs: fix the logspace waiting algorithm

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 9f9c19ec1a59422c7687b11847ed3408128aa0d6

Apply the scheme used in log_regrant_write_log_space to wake up any other
threads waiting for log space before the newly added one to
log_regrant_write_log_space as well, and factor the code into readable
helpers.  For each of the queues we have add two helpers:

 - one to try to wake up all waiting threads.  This helper will also be
   usable by xfs_log_move_tail once we remove the current opportunistic
   wakeups in it.
 - one to sleep on t_wait until enough log space is available, loosely
   modelled after Linux waitqueues.

And use them to reimplement the guts of log_regrant_write_log_space and
log_regrant_write_log_space.  These two function now use one and the same
algorithm for waiting on log space instead of subtly different ones before,
with an option to completely unify them in the near future.

Also move the filesystem shutdown handling to the common caller given
that we had to touch it anyway.

Based on hard debugging and an earlier patch from
Chandra Seetharaman <sekharan@us.ibm.com>.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chandra Seetharaman <sekharan@us.ibm.com>
Tested-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   12 +-
 fs/xfs/xfs_log.c             |  348 +++++++++++++++++++++---------------------
 2 files changed, 177 insertions(+), 183 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index b5f892a..59e8096 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -835,18 +835,14 @@ DEFINE_LOGGRANT_EVENT(xfs_log_umount_write);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_error);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep1);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake1);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep2);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep1);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake1);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep2);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index b180b55..42d8a5d 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -150,6 +150,117 @@ xlog_grant_add_space(
 	} while (head_val != old);
 }
 
+STATIC bool
+xlog_reserveq_wake(
+	struct log		*log,
+	int			*free_bytes)
+{
+	struct xlog_ticket	*tic;
+	int			need_bytes;
+
+	list_for_each_entry(tic, &log->l_reserveq, t_queue) {
+		if (tic->t_flags & XLOG_TIC_PERM_RESERV)
+			need_bytes = tic->t_unit_res * tic->t_cnt;
+		else
+			need_bytes = tic->t_unit_res;
+
+		if (*free_bytes < need_bytes)
+			return false;
+		*free_bytes -= need_bytes;
+
+		trace_xfs_log_grant_wake_up(log, tic);
+		wake_up(&tic->t_wait);
+	}
+
+	return true;
+}
+
+STATIC bool
+xlog_writeq_wake(
+	struct log		*log,
+	int			*free_bytes)
+{
+	struct xlog_ticket	*tic;
+	int			need_bytes;
+
+	list_for_each_entry(tic, &log->l_writeq, t_queue) {
+		ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
+
+		need_bytes = tic->t_unit_res;
+
+		if (*free_bytes < need_bytes)
+			return false;
+		*free_bytes -= need_bytes;
+
+		trace_xfs_log_regrant_write_wake_up(log, tic);
+		wake_up(&tic->t_wait);
+	}
+
+	return true;
+}
+
+STATIC int
+xlog_reserveq_wait(
+	struct log		*log,
+	struct xlog_ticket	*tic,
+	int			need_bytes)
+{
+	list_add_tail(&tic->t_queue, &log->l_reserveq);
+
+	do {
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto shutdown;
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		trace_xfs_log_grant_sleep(log, tic);
+
+		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
+		trace_xfs_log_grant_wake(log, tic);
+
+		spin_lock(&log->l_grant_reserve_lock);
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto shutdown;
+	} while (xlog_space_left(log, &log->l_grant_reserve_head) < need_bytes);
+
+	list_del_init(&tic->t_queue);
+	return 0;
+shutdown:
+	list_del_init(&tic->t_queue);
+	return XFS_ERROR(EIO);
+}
+
+STATIC int
+xlog_writeq_wait(
+	struct log		*log,
+	struct xlog_ticket	*tic,
+	int			need_bytes)
+{
+	list_add_tail(&tic->t_queue, &log->l_writeq);
+
+	do {
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto shutdown;
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		trace_xfs_log_regrant_write_sleep(log, tic);
+
+		xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
+		trace_xfs_log_regrant_write_wake(log, tic);
+
+		spin_lock(&log->l_grant_write_lock);
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto shutdown;
+	} while (xlog_space_left(log, &log->l_grant_write_head) < need_bytes);
+
+	list_del_init(&tic->t_queue);
+	return 0;
+shutdown:
+	list_del_init(&tic->t_queue);
+	return XFS_ERROR(EIO);
+}
+
 static void
 xlog_tic_reset_res(xlog_ticket_t *tic)
 {
@@ -350,8 +461,19 @@ xfs_log_reserve(
 		retval = xlog_grant_log_space(log, internal_ticket);
 	}
 
+	if (unlikely(retval)) {
+		/*
+		 * If we are failing, make sure the ticket doesn't have any
+		 * current reservations.  We don't want to add this back
+		 * when the ticket/ transaction gets cancelled.
+		 */
+		internal_ticket->t_curr_res = 0;
+		/* ungrant will give back unit_res * t_cnt. */
+		internal_ticket->t_cnt = 0;
+	}
+
 	return retval;
-}	/* xfs_log_reserve */
+}
 
 
 /*
@@ -2495,8 +2617,8 @@ restart:
 /*
  * Atomically get the log space required for a log ticket.
  *
- * Once a ticket gets put onto the reserveq, it will only return after
- * the needed reservation is satisfied.
+ * Once a ticket gets put onto the reserveq, it will only return after the
+ * needed reservation is satisfied.
  *
  * This function is structured so that it has a lock free fast path. This is
  * necessary because every new transaction reservation will come through this
@@ -2504,113 +2626,53 @@ restart:
  * every pass.
  *
  * As tickets are only ever moved on and off the reserveq under the
- * l_grant_reserve_lock, we only need to take that lock if we are going
- * to add the ticket to the queue and sleep. We can avoid taking the lock if the
- * ticket was never added to the reserveq because the t_queue list head will be
- * empty and we hold the only reference to it so it can safely be checked
- * unlocked.
+ * l_grant_reserve_lock, we only need to take that lock if we are going to add
+ * the ticket to the queue and sleep. We can avoid taking the lock if the ticket
+ * was never added to the reserveq because the t_queue list head will be empty
+ * and we hold the only reference to it so it can safely be checked unlocked.
  */
 STATIC int
-xlog_grant_log_space(xlog_t	   *log,
-		     xlog_ticket_t *tic)
+xlog_grant_log_space(
+	struct log		*log,
+	struct xlog_ticket	*tic)
 {
-	int		 free_bytes;
-	int		 need_bytes;
+	int			free_bytes, need_bytes;
+	int			error = 0;
 
-#ifdef DEBUG
-	if (log->l_flags & XLOG_ACTIVE_RECOVERY)
-		panic("grant Recovery problem");
-#endif
+	ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
 
 	trace_xfs_log_grant_enter(log, tic);
 
+	/*
+	 * If there are other waiters on the queue then give them a chance at
+	 * logspace before us.  Wake up the first waiters, if we do not wake
+	 * up all the waiters then go to sleep waiting for more free space,
+	 * otherwise try to get some space for this transaction.
+	 */
 	need_bytes = tic->t_unit_res;
 	if (tic->t_flags & XFS_LOG_PERM_RESERV)
 		need_bytes *= tic->t_ocnt;
-
-	/* something is already sleeping; insert new transaction at end */
-	if (!list_empty_careful(&log->l_reserveq)) {
-		spin_lock(&log->l_grant_reserve_lock);
-		/* recheck the queue now we are locked */
-		if (list_empty(&log->l_reserveq)) {
-			spin_unlock(&log->l_grant_reserve_lock);
-			goto redo;
-		}
-		list_add_tail(&tic->t_queue, &log->l_reserveq);
-
-		trace_xfs_log_grant_sleep1(log, tic);
-
-		/*
-		 * Gotta check this before going to sleep, while we're
-		 * holding the grant lock.
-		 */
-		if (XLOG_FORCED_SHUTDOWN(log))
-			goto error_return;
-
-		XFS_STATS_INC(xs_sleep_logspace);
-		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
-
-		/*
-		 * If we got an error, and the filesystem is shutting down,
-		 * we'll catch it down below. So just continue...
-		 */
-		trace_xfs_log_grant_wake1(log, tic);
-	}
-
-redo:
-	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return_unlocked;
-
 	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
-	if (free_bytes < need_bytes) {
+	if (!list_empty_careful(&log->l_reserveq)) {
 		spin_lock(&log->l_grant_reserve_lock);
-		if (list_empty(&tic->t_queue))
-			list_add_tail(&tic->t_queue, &log->l_reserveq);
-
-		trace_xfs_log_grant_sleep2(log, tic);
-
-		if (XLOG_FORCED_SHUTDOWN(log))
-			goto error_return;
-
-		xlog_grant_push_ail(log, need_bytes);
-
-		XFS_STATS_INC(xs_sleep_logspace);
-		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
-
-		trace_xfs_log_grant_wake2(log, tic);
-		goto redo;
-	}
-
-	if (!list_empty(&tic->t_queue)) {
+		if (!xlog_reserveq_wake(log, &free_bytes) ||
+		    free_bytes < need_bytes)
+			error = xlog_reserveq_wait(log, tic, need_bytes);
+		spin_unlock(&log->l_grant_reserve_lock);
+	} else if (free_bytes < need_bytes) {
 		spin_lock(&log->l_grant_reserve_lock);
-		list_del_init(&tic->t_queue);
+		error = xlog_reserveq_wait(log, tic, need_bytes);
 		spin_unlock(&log->l_grant_reserve_lock);
 	}
+	if (error)
+		return error;
 
-	/* we've got enough space */
 	xlog_grant_add_space(log, &log->l_grant_reserve_head, need_bytes);
 	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
 	trace_xfs_log_grant_exit(log, tic);
 	xlog_verify_grant_tail(log);
 	return 0;
-
-error_return_unlocked:
-	spin_lock(&log->l_grant_reserve_lock);
-error_return:
-	list_del_init(&tic->t_queue);
-	spin_unlock(&log->l_grant_reserve_lock);
-	trace_xfs_log_grant_error(log, tic);
-
-	/*
-	 * If we are failing, make sure the ticket doesn't have any
-	 * current reservations. We don't want to add this back when
-	 * the ticket/transaction gets cancelled.
-	 */
-	tic->t_curr_res = 0;
-	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
-	return XFS_ERROR(EIO);
-}	/* xlog_grant_log_space */
-
+}
 
 /*
  * Replenish the byte reservation required by moving the grant write head.
@@ -2619,10 +2681,12 @@ error_return:
  * free fast path.
  */
 STATIC int
-xlog_regrant_write_log_space(xlog_t	   *log,
-			     xlog_ticket_t *tic)
+xlog_regrant_write_log_space(
+	struct log		*log,
+	struct xlog_ticket	*tic)
 {
-	int		free_bytes, need_bytes;
+	int			free_bytes, need_bytes;
+	int			error = 0;
 
 	tic->t_curr_res = tic->t_unit_res;
 	xlog_tic_reset_res(tic);
@@ -2630,104 +2694,38 @@ xlog_regrant_write_log_space(xlog_t	   *log,
 	if (tic->t_cnt > 0)
 		return 0;
 
-#ifdef DEBUG
-	if (log->l_flags & XLOG_ACTIVE_RECOVERY)
-		panic("regrant Recovery problem");
-#endif
+	ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
 
 	trace_xfs_log_regrant_write_enter(log, tic);
-	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return_unlocked;
 
-	/* If there are other waiters on the queue then give them a
-	 * chance at logspace before us. Wake up the first waiters,
-	 * if we do not wake up all the waiters then go to sleep waiting
-	 * for more free space, otherwise try to get some space for
-	 * this transaction.
+	/*
+	 * If there are other waiters on the queue then give them a chance at
+	 * logspace before us.  Wake up the first waiters, if we do not wake
+	 * up all the waiters then go to sleep waiting for more free space,
+	 * otherwise try to get some space for this transaction.
 	 */
 	need_bytes = tic->t_unit_res;
-	if (!list_empty_careful(&log->l_writeq)) {
-		struct xlog_ticket *ntic;
-
-		spin_lock(&log->l_grant_write_lock);
-		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
-		list_for_each_entry(ntic, &log->l_writeq, t_queue) {
-			ASSERT(ntic->t_flags & XLOG_TIC_PERM_RESERV);
-
-			if (free_bytes < ntic->t_unit_res)
-				break;
-			free_bytes -= ntic->t_unit_res;
-			wake_up(&ntic->t_wait);
-		}
-
-		if (ntic != list_first_entry(&log->l_writeq,
-						struct xlog_ticket, t_queue)) {
-			if (list_empty(&tic->t_queue))
-				list_add_tail(&tic->t_queue, &log->l_writeq);
-			trace_xfs_log_regrant_write_sleep1(log, tic);
-
-			xlog_grant_push_ail(log, need_bytes);
-
-			XFS_STATS_INC(xs_sleep_logspace);
-			xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
-			trace_xfs_log_regrant_write_wake1(log, tic);
-		} else
-			spin_unlock(&log->l_grant_write_lock);
-	}
-
-redo:
-	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return_unlocked;
-
 	free_bytes = xlog_space_left(log, &log->l_grant_write_head);
-	if (free_bytes < need_bytes) {
+	if (!list_empty_careful(&log->l_writeq)) {
 		spin_lock(&log->l_grant_write_lock);
-		if (list_empty(&tic->t_queue))
-			list_add_tail(&tic->t_queue, &log->l_writeq);
-
-		if (XLOG_FORCED_SHUTDOWN(log))
-			goto error_return;
-
-		xlog_grant_push_ail(log, need_bytes);
-
-		XFS_STATS_INC(xs_sleep_logspace);
-		trace_xfs_log_regrant_write_sleep2(log, tic);
-		xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
-
-		trace_xfs_log_regrant_write_wake2(log, tic);
-		goto redo;
-	}
-
-	if (!list_empty(&tic->t_queue)) {
+		if (!xlog_writeq_wake(log, &free_bytes) ||
+		    free_bytes < need_bytes)
+			error = xlog_writeq_wait(log, tic, need_bytes);
+		spin_unlock(&log->l_grant_write_lock);
+	} else if (free_bytes < need_bytes) {
 		spin_lock(&log->l_grant_write_lock);
-		list_del_init(&tic->t_queue);
+		error = xlog_writeq_wait(log, tic, need_bytes);
 		spin_unlock(&log->l_grant_write_lock);
 	}
 
-	/* we've got enough space */
+	if (error)
+		return error;
+
 	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
 	trace_xfs_log_regrant_write_exit(log, tic);
 	xlog_verify_grant_tail(log);
 	return 0;
-
-
- error_return_unlocked:
-	spin_lock(&log->l_grant_write_lock);
- error_return:
-	list_del_init(&tic->t_queue);
-	spin_unlock(&log->l_grant_write_lock);
-	trace_xfs_log_regrant_write_error(log, tic);
-
-	/*
-	 * If we are failing, make sure the ticket doesn't have any
-	 * current reservations. We don't want to add this back when
-	 * the ticket/transaction gets cancelled.
-	 */
-	tic->t_curr_res = 0;
-	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
-	return XFS_ERROR(EIO);
-}	/* xlog_regrant_write_log_space */
-
+}
 
 /* The first cnt-1 times through here we don't need to
  * move the grant write head because the permanent
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 024/102] xfs: untangle SYNC_WAIT and SYNC_TRYLOCK meanings for xfs_qm_dqflush

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: fdedf28b9492d69976110d12cc0d02d33c8ea7ea

Only skip pinned dquots if SYNC_TRYLOCK is specified, and adjust the callers
to keep the behaviour unchanged.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/quota/xfs_dquot.c      |    2 +-
 fs/xfs/quota/xfs_dquot_item.c |    2 +-
 fs/xfs/quota/xfs_qm.c         |    4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/quota/xfs_dquot.c b/fs/xfs/quota/xfs_dquot.c
index 478a11b..b8be8a0 100644
--- a/fs/xfs/quota/xfs_dquot.c
+++ b/fs/xfs/quota/xfs_dquot.c
@@ -1173,7 +1173,7 @@ xfs_qm_dqflush(
 	 * If not dirty, or it's pinned and we are not supposed to block, nada.
 	 */
 	if (!XFS_DQ_IS_DIRTY(dqp) ||
-	    (!(flags & SYNC_WAIT) && atomic_read(&dqp->q_pincount) > 0)) {
+	    ((flags & SYNC_TRYLOCK) && atomic_read(&dqp->q_pincount) > 0)) {
 		xfs_dqfunlock(dqp);
 		return 0;
 	}
diff --git a/fs/xfs/quota/xfs_dquot_item.c b/fs/xfs/quota/xfs_dquot_item.c
index 8126fc2..87f10f3 100644
--- a/fs/xfs/quota/xfs_dquot_item.c
+++ b/fs/xfs/quota/xfs_dquot_item.c
@@ -134,7 +134,7 @@ xfs_qm_dquot_logitem_push(
 	 * lock without sleeping, then there must not have been
 	 * anyone in the process of flushing the dquot.
 	 */
-	error = xfs_qm_dqflush(dqp, 0);
+	error = xfs_qm_dqflush(dqp, SYNC_TRYLOCK);
 	if (error)
 		xfs_warn(dqp->q_mount, "%s: push error %d on dqp %p",
 			__func__, error, dqp);
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index e70c7fc..3d50587 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -1703,7 +1703,7 @@ xfs_qm_quotacheck(
 	 * successfully.
 	 */
 	if (!error)
-		error = xfs_qm_dqflush_all(mp, 0);
+		error = xfs_qm_dqflush_all(mp, SYNC_TRYLOCK);
 
 	/*
 	 * We can get this error if we couldn't do a dquot allocation inside
@@ -1918,7 +1918,7 @@ again:
 			 * We flush it delayed write, so don't bother
 			 * releasing the freelist lock.
 			 */
-			error = xfs_qm_dqflush(dqp, 0);
+			error = xfs_qm_dqflush(dqp, SYNC_TRYLOCK);
 			if (error) {
 				xfs_warn(mp, "%s: dquot %p flush failed",
 					__func__, dqp);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 025/102] xfs: make sure to really flush all dquots in xfs_qm_quotacheck

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: f2fba558d3c80dcd10bbadbb8f05c78dc2860b95

Make sure we do not skip any dquots when flushing them out after a
quotacheck to make sure that we will never have any dirty dquots on a
live filesystem.  At this point no dquot should be pinnable, but lets
be pedantic about it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/quota/xfs_qm.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index 3d50587..0785984 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -1703,7 +1703,7 @@ xfs_qm_quotacheck(
 	 * successfully.
 	 */
 	if (!error)
-		error = xfs_qm_dqflush_all(mp, SYNC_TRYLOCK);
+		error = xfs_qm_dqflush_all(mp, 0);
 
 	/*
 	 * We can get this error if we couldn't do a dquot allocation inside
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 026/102] xfs: simplify xfs_qm_detach_gdquots

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 28fb588c9bd810dec273d96e80591392f6ce1e1c

There is no reason to drop qi_dqlist_lock around calls to xfs_qm_dqrele
because the free list lock now nests inside qi_dqlist_lock and the
dquot lock.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/quota/xfs_qm.c |   23 +++++------------------
 1 file changed, 5 insertions(+), 18 deletions(-)

diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index 0785984..eacf7e0 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -518,31 +518,18 @@ xfs_qm_detach_gdquots(
 {
 	struct xfs_quotainfo	*q = mp->m_quotainfo;
 	struct xfs_dquot	*dqp, *gdqp;
-	int			nrecl;
 
- again:
 	ASSERT(mutex_is_locked(&q->qi_dqlist_lock));
 	list_for_each_entry(dqp, &q->qi_dqlist, q_mplist) {
 		xfs_dqlock(dqp);
-		if ((gdqp = dqp->q_gdquot)) {
-			xfs_dqlock(gdqp);
+
+		gdqp = dqp->q_gdquot;
+		if (gdqp)
 			dqp->q_gdquot = NULL;
-		}
 		xfs_dqunlock(dqp);
+		if (gdqp)
+			xfs_qm_dqrele(gdqp);
 
-		if (gdqp) {
-			/*
-			 * Can't hold the mplist lock across a dqput.
-			 * XXXmust convert to marker based iterations here.
-			 */
-			nrecl = q->qi_dqreclaims;
-			mutex_unlock(&q->qi_dqlist_lock);
-			xfs_qm_dqput(gdqp);
-
-			mutex_lock(&q->qi_dqlist_lock);
-			if (nrecl != q->qi_dqreclaims)
-				goto again;
-		}
 	}
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 027/102] xfs: mark the xfssyncd workqueue as non-reentrant

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 40d344ec5ee440596b1f3ae87556e20c7197757a

On a system with lots of memory pressure that is stuck on synchronous inode
reclaim the workqueue code will run one instance of the inode reclaim work
item on every CPU. which is not what we want.  Make sure to mark the
xfssyncd workqueue as non-reentrant to make sure there only is one instace
of each running globally.  Also stop using special paramater for the
workqueue; now that we guarantee each fs has only running one of each works
at a time there is no need to artificially lower max_active and compensate
for that by setting the WQ_CPU_INTENSIVE flag.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_super.c |   10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index e6ac98c..04dd4d4 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1617,12 +1617,12 @@ STATIC int __init
 xfs_init_workqueues(void)
 {
 	/*
-	 * max_active is set to 8 to give enough concurency to allow
-	 * multiple work operations on each CPU to run. This allows multiple
-	 * filesystems to be running sync work concurrently, and scales with
-	 * the number of CPUs in the system.
+	 * We never want to the same work item to run twice, reclaiming inodes
+	 * or idling the log is not going to get any faster by multiple CPUs
+	 * competing for ressources.  Use the default large max_active value
+	 * so that even lots of filesystems can perform these task in parallel.
 	 */
-	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_CPU_INTENSIVE, 8);
+	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_NON_REENTRANT, 0);
 	if (!xfs_syncd_wq)
 		return -ENOMEM;
 	return 0;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 028/102] xfs: make i_flags an unsigned long

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 49e4c70e52a2bc2090e5a4e003e2888af21d6a2b

To be used for bit wakeup i_flags needs to be an unsigned long or we'll
run into trouble on big endian systems.  Because of the 1-byte i_update
field right after it this actually causes a fairly large size increase
on its own (4 or 8 bytes), but that increase will be more than offset
by the next two patches.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Alex Elder <aelder@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_inode.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 28b3596..8a0979a 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -250,7 +250,7 @@ typedef struct xfs_inode {
 	wait_queue_head_t	i_ipin_wait;	/* inode pinning wait queue */
 	spinlock_t		i_flags_lock;	/* inode i_flags lock */
 	/* Miscellaneous state. */
-	unsigned short		i_flags;	/* see defined flags below */
+	unsigned long		i_flags;	/* see defined flags below */
 	unsigned char		i_update_core;	/* timestamps/size is dirty */
 	unsigned int		i_delayed_blks;	/* count of delay alloc blks */
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 029/102] xfs: remove the i_size field in struct xfs_inode

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: ce7ae151ddada3dbf67301464343c154903166b3

There is no fundamental need to keep an in-memory inode size copy in the XFS
inode.  We already have the on-disk value in the dinode, and the separate
in-memory copy that we need for regular files only in the XFS inode.

Remove the xfs_inode i_size field and change the XFS_ISIZE macro to use the
VFS inode i_size field for regular files.  Switch code that was directly
accessing the i_size field in the xfs_inode to XFS_ISIZE, or in cases where
we are limited to regular files direct access of the VFS inode i_size field.

This also allows dropping some fairly complicated code in the write path
which dealt with keeping the xfs_inode i_size uptodate with the VFS i_size
that is getting updated inside ->write_end.

Note that we do not bother resetting the VFS i_size when truncating a file
that gets freed to zero as there is no point in doing so because the VFS inode
is no longer in use at this point.  Just relax the assert in xfs_ifree to
only check the on-disk size instead.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c    |    2 +-
 fs/xfs/linux-2.6/xfs_file.c    |   45 ++++++++++------------------------------
 fs/xfs/linux-2.6/xfs_fs_subr.c |    2 +-
 fs/xfs/xfs_bmap.c              |   15 ++++++--------
 fs/xfs/xfs_iget.c              |    1 -
 fs/xfs/xfs_inode.c             |   20 +++++++-----------
 fs/xfs/xfs_inode.h             |   16 ++++++++++----
 fs/xfs/xfs_iomap.c             |   14 ++++++-------
 fs/xfs/xfs_vnodeops.c          |   39 +++++++++++++++++-----------------
 9 files changed, 65 insertions(+), 89 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 2426ecd..a284ea7 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -145,7 +145,7 @@ xfs_ioend_new_eof(
 	xfs_fsize_t		bsize;
 
 	bsize = ioend->io_offset + ioend->io_size;
-	isize = MAX(ip->i_size, ip->i_new_size);
+	isize = MAX(i_size_read(VFS_I(ip)), ip->i_new_size);
 	isize = MIN(isize, bsize);
 	return isize > ip->i_d.di_size ? isize : 0;
 }
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 185a337..78a9cf4 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -320,7 +320,7 @@ xfs_file_aio_read(
 				mp->m_rtdev_targp : mp->m_ddev_targp;
 		if ((iocb->ki_pos & target->bt_smask) ||
 		    (size & target->bt_smask)) {
-			if (iocb->ki_pos == ip->i_size)
+			if (iocb->ki_pos == i_size_read(inode))
 				return 0;
 			return -XFS_ERROR(EINVAL);
 		}
@@ -405,30 +405,6 @@ xfs_file_splice_read(
 	return ret;
 }
 
-STATIC void
-xfs_aio_write_isize_update(
-	struct inode	*inode,
-	loff_t		*ppos,
-	ssize_t		bytes_written)
-{
-	struct xfs_inode	*ip = XFS_I(inode);
-	xfs_fsize_t		isize = i_size_read(inode);
-
-	if (bytes_written > 0)
-		XFS_STATS_ADD(xs_write_bytes, bytes_written);
-
-	if (unlikely(bytes_written < 0 && bytes_written != -EFAULT &&
-					*ppos > isize))
-		*ppos = isize;
-
-	if (*ppos > ip->i_size) {
-		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
-		if (*ppos > ip->i_size)
-			ip->i_size = *ppos;
-		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
-	}
-}
-
 /*
  * If this was a direct or synchronous I/O that failed (such as ENOSPC) then
  * part of the I/O may have been written to disk before the error occurred.  In
@@ -444,8 +420,8 @@ xfs_aio_write_newsize_update(
 		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
 		if (new_size == ip->i_new_size)
 			ip->i_new_size = 0;
-		if (ip->i_d.di_size > ip->i_size)
-			ip->i_d.di_size = ip->i_size;
+		if (ip->i_d.di_size > i_size_read(VFS_I(ip)))
+			ip->i_d.di_size = i_size_read(VFS_I(ip));
 		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
 	}
 }
@@ -485,15 +461,16 @@ xfs_file_splice_write(
 	new_size = *ppos + count;
 
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	if (new_size > ip->i_size)
+	if (new_size > i_size_read(inode))
 		ip->i_new_size = new_size;
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 
 	trace_xfs_file_splice_write(ip, count, *ppos, ioflags);
 
 	ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags);
+	if (ret > 0)
+		XFS_STATS_ADD(xs_write_bytes, ret);
 
-	xfs_aio_write_isize_update(inode, ppos, ret);
 	xfs_aio_write_newsize_update(ip, new_size);
 	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 	return ret;
@@ -723,14 +700,14 @@ restart:
 	 * values are still valid.
 	 */
 	if ((ip->i_new_size && *pos > ip->i_new_size) ||
-	    (!ip->i_new_size && *pos > ip->i_size)) {
+	    (!ip->i_new_size && *pos > i_size_read(inode))) {
 		if (*iolock == XFS_IOLOCK_SHARED) {
 			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
 			*iolock = XFS_IOLOCK_EXCL;
 			xfs_rw_ilock(ip, XFS_ILOCK_EXCL | *iolock);
 			goto restart;
 		}
-		error = -xfs_zero_eof(ip, *pos, ip->i_size);
+		error = -xfs_zero_eof(ip, *pos, i_size_read(inode));
 	}
 
 	/*
@@ -739,7 +716,7 @@ restart:
 	 * ip->i_new_size if this IO ends beyond any other in-flight writes.
 	 */
 	new_size = *pos + *count;
-	if (new_size > ip->i_size) {
+	if (new_size > i_size_read(inode)) {
 		if (new_size > ip->i_new_size)
 			ip->i_new_size = new_size;
 		*new_sizep = new_size;
@@ -952,11 +929,11 @@ xfs_file_aio_write(
 		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
 						ocount, &new_size, &iolock);
 
-	xfs_aio_write_isize_update(inode, &iocb->ki_pos, ret);
-
 	if (ret <= 0)
 		goto out_unlock;
 
+	XFS_STATS_ADD(xs_write_bytes, ret);
+
 	/* Handle various SYNC-type writes */
 	if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
 		loff_t end = pos + ret - 1;
diff --git a/fs/xfs/linux-2.6/xfs_fs_subr.c b/fs/xfs/linux-2.6/xfs_fs_subr.c
index ed88ed1..652b875 100644
--- a/fs/xfs/linux-2.6/xfs_fs_subr.c
+++ b/fs/xfs/linux-2.6/xfs_fs_subr.c
@@ -90,7 +90,7 @@ xfs_wait_on_pages(
 
 	if (mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK)) {
 		return -filemap_fdatawait_range(mapping, first,
-					last == -1 ? ip->i_size - 1 : last);
+					last == -1 ? XFS_ISIZE(ip) - 1 : last);
 	}
 	return 0;
 }
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 05ea68f..51b3489 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -4045,11 +4045,8 @@ xfs_bmap_one_block(
 	xfs_bmbt_irec_t	s;		/* internal version of extent */
 
 #ifndef DEBUG
-	if (whichfork == XFS_DATA_FORK) {
-		return ((ip->i_d.di_mode & S_IFMT) == S_IFREG) ?
-			(ip->i_size == ip->i_mount->m_sb.sb_blocksize) :
-			(ip->i_d.di_size == ip->i_mount->m_sb.sb_blocksize);
-	}
+	if (whichfork == XFS_DATA_FORK)
+		return XFS_ISIZE(ip) == ip->i_mount->m_sb.sb_blocksize;
 #endif	/* !DEBUG */
 	if (XFS_IFORK_NEXTENTS(ip, whichfork) != 1)
 		return 0;
@@ -4061,7 +4058,7 @@ xfs_bmap_one_block(
 	xfs_bmbt_get_all(ep, &s);
 	rval = s.br_startoff == 0 && s.br_blockcount == 1;
 	if (rval && whichfork == XFS_DATA_FORK)
-		ASSERT(ip->i_size == ip->i_mount->m_sb.sb_blocksize);
+		ASSERT(XFS_ISIZE(ip) == ip->i_mount->m_sb.sb_blocksize);
 	return rval;
 }
 
@@ -5346,7 +5343,7 @@ xfs_getbmapx_fix_eof_hole(
 	if (startblock == HOLESTARTBLOCK) {
 		mp = ip->i_mount;
 		out->bmv_block = -1;
-		fixlen = XFS_FSB_TO_BB(mp, XFS_B_TO_FSB(mp, ip->i_size));
+		fixlen = XFS_FSB_TO_BB(mp, XFS_B_TO_FSB(mp, XFS_ISIZE(ip)));
 		fixlen -= out->bmv_offset;
 		if (prealloced && out->bmv_offset + out->bmv_length == end) {
 			/* Came to hole at EOF. Trim it. */
@@ -5434,7 +5431,7 @@ xfs_getbmap(
 			fixlen = XFS_MAXIOFFSET(mp);
 		} else {
 			prealloced = 0;
-			fixlen = ip->i_size;
+			fixlen = XFS_ISIZE(ip);
 		}
 	}
 
@@ -5463,7 +5460,7 @@ xfs_getbmap(
 
 	xfs_ilock(ip, XFS_IOLOCK_SHARED);
 	if (whichfork == XFS_DATA_FORK && !(iflags & BMV_IF_DELALLOC)) {
-		if (ip->i_delayed_blks || ip->i_size > ip->i_d.di_size) {
+		if (ip->i_delayed_blks || XFS_ISIZE(ip) > ip->i_d.di_size) {
 			error = xfs_flush_pages(ip, 0, -1, 0, FI_REMAPF);
 			if (error)
 				goto out_unlock_iolock;
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index ca752f0..4c8df0b 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -96,7 +96,6 @@ xfs_inode_alloc(
 	ip->i_update_core = 0;
 	ip->i_delayed_blks = 0;
 	memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
-	ip->i_size = 0;
 	ip->i_new_size = 0;
 
 	return ip;
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 2966b4d..1c6788a 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -357,7 +357,6 @@ xfs_iformat(
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 		ip->i_d.di_size = 0;
-		ip->i_size = 0;
 		ip->i_df.if_u2.if_rdev = xfs_dinode_get_rdev(dip);
 		break;
 
@@ -868,7 +867,6 @@ xfs_iread(
 	}
 
 	ip->i_delayed_blks = 0;
-	ip->i_size = ip->i_d.di_size;
 
 	/*
 	 * Mark the buffer containing the inode as something to keep
@@ -1058,7 +1056,6 @@ xfs_ialloc(
 	}
 
 	ip->i_d.di_size = 0;
-	ip->i_size = 0;
 	ip->i_d.di_nextents = 0;
 	ASSERT(ip->i_d.di_nblocks == 0);
 
@@ -1257,7 +1254,7 @@ xfs_file_last_byte(
 	} else {
 		last_block = 0;
 	}
-	size_last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)ip->i_size);
+	size_last_block = XFS_B_TO_FSB(mp, XFS_ISIZE(ip));
 	last_block = XFS_FILEOFF_MAX(last_block, size_last_block);
 
 	last_byte = XFS_FSB_TO_B(mp, last_block);
@@ -1314,14 +1311,14 @@ xfs_itruncate_start(
 	int		error = 0;
 
 	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
-	ASSERT((new_size == 0) || (new_size <= ip->i_size));
+	ASSERT(new_size == 0 || new_size <= XFS_ISIZE(ip));
 	ASSERT((flags == XFS_ITRUNC_DEFINITE) ||
 	       (flags == XFS_ITRUNC_MAYBE));
 
 	mp = ip->i_mount;
 
 	/* wait for the completion of any pending DIOs */
-	if (new_size == 0 || new_size < ip->i_size)
+	if (new_size == 0 || new_size < XFS_ISIZE(ip))
 		xfs_ioend_wait(ip);
 
 	/*
@@ -1438,7 +1435,7 @@ xfs_itruncate_finish(
 	int		error;
 
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
-	ASSERT((new_size == 0) || (new_size <= ip->i_size));
+	ASSERT(new_size == 0 || new_size <= XFS_ISIZE(ip));
 	ASSERT(*tp != NULL);
 	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
 	ASSERT(ip->i_transp == *tp);
@@ -1522,9 +1519,8 @@ xfs_itruncate_finish(
 			 * flushed to disk then the files may be full of
 			 * holes (ie NULL files bug).
 			 */
-			if (ip->i_size != new_size) {
+			if (ip->i_d.di_size != new_size) {
 				ip->i_d.di_size = new_size;
-				ip->i_size = new_size;
 				xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
 			}
 		}
@@ -1648,9 +1644,8 @@ xfs_itruncate_finish(
 		 * flushed to disk then the files may be full of
 		 * holes (ie NULL files bug).
 		 */
-		if (ip->i_size != new_size) {
+		if (ip->i_d.di_size != new_size) {
 			ip->i_d.di_size = new_size;
-			ip->i_size = new_size;
 		}
 	}
 	xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
@@ -2085,8 +2080,7 @@ xfs_ifree(
 	ASSERT(ip->i_d.di_nlink == 0);
 	ASSERT(ip->i_d.di_nextents == 0);
 	ASSERT(ip->i_d.di_anextents == 0);
-	ASSERT((ip->i_d.di_size == 0 && ip->i_size == 0) ||
-	       ((ip->i_d.di_mode & S_IFMT) != S_IFREG));
+	ASSERT(ip->i_d.di_size == 0 || !S_ISREG(ip->i_d.di_mode));
 	ASSERT(ip->i_d.di_nblocks == 0);
 
 	/*
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 8a0979a..ed1c89e 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -256,7 +256,6 @@ typedef struct xfs_inode {
 
 	xfs_icdinode_t		i_d;		/* most of ondisk inode */
 
-	xfs_fsize_t		i_size;		/* in-memory size */
 	xfs_fsize_t		i_new_size;	/* size when write completes */
 	atomic_t		i_iocount;	/* outstanding I/O count */
 
@@ -264,9 +263,6 @@ typedef struct xfs_inode {
 	struct inode		i_vnode;	/* embedded VFS inode */
 } xfs_inode_t;
 
-#define XFS_ISIZE(ip)	(((ip)->i_d.di_mode & S_IFMT) == S_IFREG) ? \
-				(ip)->i_size : (ip)->i_d.di_size;
-
 /* Convert from vfs inode to xfs inode */
 static inline struct xfs_inode *XFS_I(struct inode *inode)
 {
@@ -280,6 +276,18 @@ static inline struct inode *VFS_I(struct xfs_inode *ip)
 }
 
 /*
+ * For regular files we only update the on-disk filesize when actually
+ * writing data back to disk.  Until then only the copy in the VFS inode
+ * is uptodate.
+ */
+static inline xfs_fsize_t XFS_ISIZE(struct xfs_inode *ip)
+{
+	if (S_ISREG(ip->i_d.di_mode))
+		return i_size_read(VFS_I(ip));
+	return ip->i_d.di_size;
+}
+
+/*
  * i_flags helper functions
  */
 static inline void
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 091d82b..4460beb 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -68,14 +68,14 @@ xfs_iomap_eof_align_last_fsb(
 	 * stripe width and we are allocating past the allocation eof.
 	 */
 	else if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC) &&
-	        (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_swidth)))
+	        (XFS_ISIZE(ip) >= XFS_FSB_TO_B(mp, mp->m_swidth)))
 		new_last_fsb = roundup_64(*last_fsb, mp->m_swidth);
 	/*
 	 * Roundup the allocation request to a stripe unit (m_dalign) boundary
 	 * if the file size is >= stripe unit size, and we are allocating past
 	 * the allocation eof.
 	 */
-	else if (mp->m_dalign && (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_dalign)))
+	else if (mp->m_dalign && (XFS_ISIZE(ip) >= XFS_FSB_TO_B(mp, mp->m_dalign)))
 		new_last_fsb = roundup_64(*last_fsb, mp->m_dalign);
 
 	/*
@@ -154,7 +154,7 @@ xfs_iomap_write_direct(
 
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 	last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
-	if ((offset + count) > ip->i_size) {
+	if ((offset + count) > XFS_ISIZE(ip)) {
 		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
 		if (error)
 			goto error_out;
@@ -211,7 +211,7 @@ xfs_iomap_write_direct(
 	xfs_trans_ijoin(tp, ip);
 
 	bmapi_flag = XFS_BMAPI_WRITE;
-	if (offset < ip->i_size || extsz)
+	if (offset < XFS_ISIZE(ip) || extsz)
 		bmapi_flag |= XFS_BMAPI_PREALLOC;
 
 	/*
@@ -288,7 +288,7 @@ xfs_iomap_eof_want_preallocate(
 	int		found_delalloc = 0;
 
 	*prealloc = 0;
-	if ((offset + count) <= ip->i_size)
+	if (offset + count <= XFS_ISIZE(ip))
 		return 0;
 
 	/*
@@ -342,7 +342,7 @@ xfs_iomap_prealloc_size(
 		 * if we pass in alloc_blocks = 0. Hence the "+ 1" to
 		 * ensure we always pass in a non-zero value.
 		 */
-		alloc_blocks = XFS_B_TO_FSB(mp, ip->i_size) + 1;
+		alloc_blocks = XFS_B_TO_FSB(mp, XFS_ISIZE(ip)) + 1;
 		alloc_blocks = XFS_FILEOFF_MIN(MAXEXTLEN,
 					rounddown_pow_of_two(alloc_blocks));
 
@@ -571,7 +571,7 @@ xfs_iomap_write_allocate(
 			 * back....
 			 */
 			nimaps = 1;
-			end_fsb = XFS_B_TO_FSB(mp, ip->i_size);
+			end_fsb = XFS_B_TO_FSB(mp, XFS_ISIZE(ip));
 			error = xfs_bmap_last_offset(NULL, ip, &last_block,
 							XFS_DATA_FORK);
 			if (error)
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 44f780c..23a25e6 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -179,9 +179,11 @@ xfs_setattr(
 	 * Truncate file.  Must have write permission and not be a directory.
 	 */
 	if (mask & ATTR_SIZE) {
+		loff_t	old_size = XFS_ISIZE(ip);
+
 		/* Short circuit the truncate case for zero length files */
 		if (iattr->ia_size == 0 &&
-		    ip->i_size == 0 && ip->i_d.di_nextents == 0) {
+		    old_size == 0 && ip->i_d.di_nextents == 0) {
 			xfs_iunlock(ip, XFS_ILOCK_EXCL);
 			lock_flags &= ~XFS_ILOCK_EXCL;
 			if (mask & ATTR_CTIME) {
@@ -216,14 +218,14 @@ xfs_setattr(
 		 * to the transaction, because the inode cannot be unlocked
 		 * once it is a part of the transaction.
 		 */
-		if (iattr->ia_size > ip->i_size) {
+		if (iattr->ia_size > old_size) {
 			/*
 			 * Do the first part of growing a file: zero any data
 			 * in the last block that is beyond the old EOF.  We
 			 * need to do this before the inode is joined to the
 			 * transaction to modify the i_size.
 			 */
-			code = xfs_zero_eof(ip, iattr->ia_size, ip->i_size);
+			code = xfs_zero_eof(ip, iattr->ia_size, old_size);
 			if (code)
 				goto error_return;
 		}
@@ -242,7 +244,7 @@ xfs_setattr(
 		 * really care about here and prevents waiting for other data
 		 * not within the range we care about here.
 		 */
-		if (ip->i_size != ip->i_d.di_size &&
+		if (old_size != ip->i_d.di_size &&
 		    iattr->ia_size > ip->i_d.di_size) {
 			code = xfs_flush_pages(ip,
 					ip->i_d.di_size, iattr->ia_size,
@@ -287,18 +289,17 @@ xfs_setattr(
 		 * VFS set these flags explicitly if it wants a timestamp
 		 * update.
 		 */
-		if (iattr->ia_size != ip->i_size &&
+		if (iattr->ia_size != old_size &&
 		    (!(mask & (ATTR_CTIME | ATTR_MTIME)))) {
 			iattr->ia_ctime = iattr->ia_mtime =
 				current_fs_time(inode->i_sb);
 			mask |= ATTR_CTIME | ATTR_MTIME;
 		}
 
-		if (iattr->ia_size > ip->i_size) {
+		if (iattr->ia_size > old_size) {
 			ip->i_d.di_size = iattr->ia_size;
-			ip->i_size = iattr->ia_size;
 			xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-		} else if (iattr->ia_size <= ip->i_size ||
+		} else if (iattr->ia_size <= old_size ||
 			   (iattr->ia_size == 0 && ip->i_d.di_nextents)) {
 			/*
 			 * signal a sync transaction unless
@@ -598,7 +599,7 @@ xfs_free_eofblocks(
 	 * Figure out if there are any blocks beyond the end
 	 * of the file.  If not, then there is nothing to do.
 	 */
-	end_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)ip->i_size));
+	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_ISIZE(ip));
 	last_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
 	if (last_fsb <= end_fsb)
 		return 0;
@@ -643,7 +644,7 @@ xfs_free_eofblocks(
 			xfs_ilock(ip, XFS_IOLOCK_EXCL);
 		}
 		error = xfs_itruncate_start(ip, XFS_ITRUNC_DEFINITE,
-				    ip->i_size);
+				    XFS_ISIZE(ip));
 		if (error) {
 			xfs_trans_cancel(tp, 0);
 			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
@@ -665,7 +666,7 @@ xfs_free_eofblocks(
 		xfs_trans_ijoin(tp, ip);
 
 		error = xfs_itruncate_finish(&tp, ip,
-					     ip->i_size,
+					     XFS_ISIZE(ip),
 					     XFS_DATA_FORK,
 					     0);
 		/*
@@ -981,7 +982,7 @@ xfs_release(
 		return 0;
 
 	if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
-	     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
+	     ((XFS_ISIZE(ip) > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
 	       ip->i_delayed_blks > 0)) &&
 	     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
 	    (!(ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
@@ -1059,7 +1060,7 @@ xfs_inactive(
 	 * only one with a reference to the inode.
 	 */
 	truncate = ((ip->i_d.di_nlink == 0) &&
-	    ((ip->i_d.di_size != 0) || (ip->i_size != 0) ||
+	    ((ip->i_d.di_size != 0) || XFS_ISIZE(ip) != 0 ||
 	     (ip->i_d.di_nextents > 0) || (ip->i_delayed_blks > 0)) &&
 	    ((ip->i_d.di_mode & S_IFMT) == S_IFREG));
 
@@ -1073,7 +1074,7 @@ xfs_inactive(
 
 	if (ip->i_d.di_nlink != 0) {
 		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
-                     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
+                     ((XFS_ISIZE(ip) > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
                        ip->i_delayed_blks > 0)) &&
 		      (ip->i_df.if_flags & XFS_IFEXTENTS) &&
 		     (!(ip->i_d.di_flags &
@@ -2431,11 +2432,11 @@ xfs_zero_remaining_bytes(
 	 * since nothing can read beyond eof.  The space will
 	 * be zeroed when the file is extended anyway.
 	 */
-	if (startoff >= ip->i_size)
+	if (startoff >= XFS_ISIZE(ip))
 		return 0;
 
-	if (endoff > ip->i_size)
-		endoff = ip->i_size;
+	if (endoff > XFS_ISIZE(ip))
+		endoff = XFS_ISIZE(ip);
 
 	bp = xfs_buf_get_uncached(XFS_IS_REALTIME_INODE(ip) ?
 					mp->m_rtdev_targp : mp->m_ddev_targp,
@@ -2729,7 +2730,7 @@ xfs_change_file_space(
 		bf->l_start += offset;
 		break;
 	case 2: /*SEEK_END*/
-		bf->l_start += ip->i_size;
+		bf->l_start += XFS_ISIZE(ip);
 		break;
 	default:
 		return XFS_ERROR(EINVAL);
@@ -2746,7 +2747,7 @@ xfs_change_file_space(
 	bf->l_whence = 0;
 
 	startoffset = bf->l_start;
-	fsize = ip->i_size;
+	fsize = XFS_ISIZE(ip);
 
 	/*
 	 * XFS_IOC_RESVSP and XFS_IOC_UNRESVSP will reserve or unreserve
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 030/102] xfs: remove the i_new_size field in struct xfs_inode

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 2813d682e8e6a278f94817429afd46b30875bb6e

Now that we use the VFS i_size field throughout XFS there is no need for the
i_new_size field any more given that the VFS i_size field gets updated
in ->write_end before unlocking the page, and thus is always uptodate when
writeback could see a page.  Removing i_new_size also has the advantage that
we will never have to trim back di_size during a failed buffered write,
given that it never gets updated past i_size.

Note that currently the generic direct I/O code only updates i_size after
calling our end_io handler, which requires a small workaround to make
sure di_size actually makes it to disk.  I hope to fix this properly in
the generic code.

A downside is that we lose the support for parallel non-overlapping O_DIRECT
appending writes that recently was added.  I don't think keeping the complex
and fragile i_new_size infrastructure for this is a good tradeoff - if we
really care about parallel appending writers we should investigate turning
the iolock into a range lock, which would also allow for parallel
non-overlapping buffered writers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c  |   29 +++++++++--------
 fs/xfs/linux-2.6/xfs_file.c  |   72 ++++++------------------------------------
 fs/xfs/linux-2.6/xfs_trace.h |   18 +++--------
 fs/xfs/xfs_iget.c            |    1 -
 fs/xfs/xfs_inode.h           |    1 -
 5 files changed, 30 insertions(+), 91 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index a284ea7..91aa080 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -145,8 +145,7 @@ xfs_ioend_new_eof(
 	xfs_fsize_t		bsize;
 
 	bsize = ioend->io_offset + ioend->io_size;
-	isize = MAX(i_size_read(VFS_I(ip)), ip->i_new_size);
-	isize = MIN(isize, bsize);
+	isize = MIN(i_size_read(VFS_I(ip)), bsize);
 	return isize > ip->i_d.di_size ? isize : 0;
 }
 
@@ -160,11 +159,7 @@ static inline bool xfs_ioend_is_append(struct xfs_ioend *ioend)
 }
 
 /*
- * Update on-disk file size now that data has been written to disk.  The
- * current in-memory file size is i_size.  If a write is beyond eof i_new_size
- * will be the intended file size until i_size is updated.  If this write does
- * not extend all the way to the valid file size then restrict this update to
- * the end of the write.
+ * Update on-disk file size now that data has been written to disk.
  *
  * This function does not block as blocking on the inode lock in IO completion
  * can lead to IO completion order dependency deadlocks.. If it can't get the
@@ -1325,6 +1320,15 @@ xfs_end_io_direct_write(
 	struct xfs_ioend	*ioend = iocb->private;
 
 	/*
+	 * While the generic direct I/O code updates the inode size, it does
+	 * so only after the end_io handler is called, which means our
+	 * end_io handler thinks the on-disk size is outside the in-core
+	 * size.  To prevent this just update it a little bit earlier here.
+	 */
+	if (offset + size > i_size_read(ioend->io_inode))
+		i_size_write(ioend->io_inode, offset + size);
+
+	/*
 	 * blockdev_direct_IO can return an error even after the I/O
 	 * completion handler was called.  Thus we need to protect
 	 * against double-freeing.
@@ -1386,12 +1390,11 @@ xfs_vm_write_failed(
 
 	if (to > inode->i_size) {
 		/*
-		 * punch out the delalloc blocks we have already allocated. We
-		 * don't call xfs_setattr() to do this as we may be in the
-		 * middle of a multi-iovec write and so the vfs inode->i_size
-		 * will not match the xfs ip->i_size and so it will zero too
-		 * much. Hence we jus truncate the page cache to zero what is
-		 * necessary and punch the delalloc blocks directly.
+		 * Punch out the delalloc blocks we have already allocated.
+		 *
+		 * Don't bother with xfs_setattr given that nothing can have
+		 * made it to disk yet as the page is still locked at this
+		 * point.
 		 */
 		struct xfs_inode	*ip = XFS_I(inode);
 		xfs_fileoff_t		start_fsb;
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 78a9cf4..e3076f0 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -406,27 +406,6 @@ xfs_file_splice_read(
 }
 
 /*
- * If this was a direct or synchronous I/O that failed (such as ENOSPC) then
- * part of the I/O may have been written to disk before the error occurred.  In
- * this case the on-disk file size may have been adjusted beyond the in-memory
- * file size and now needs to be truncated back.
- */
-STATIC void
-xfs_aio_write_newsize_update(
-	struct xfs_inode	*ip,
-	xfs_fsize_t		new_size)
-{
-	if (new_size == ip->i_new_size) {
-		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
-		if (new_size == ip->i_new_size)
-			ip->i_new_size = 0;
-		if (ip->i_d.di_size > i_size_read(VFS_I(ip)))
-			ip->i_d.di_size = i_size_read(VFS_I(ip));
-		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
-	}
-}
-
-/*
  * xfs_file_splice_write() does not use xfs_rw_ilock() because
  * generic_file_splice_write() takes the i_mutex itself. This, in theory,
  * couuld cause lock inversions between the aio_write path and the splice path
@@ -444,7 +423,6 @@ xfs_file_splice_write(
 {
 	struct inode		*inode = outfilp->f_mapping->host;
 	struct xfs_inode	*ip = XFS_I(inode);
-	xfs_fsize_t		new_size;
 	int			ioflags = 0;
 	ssize_t			ret;
 
@@ -458,20 +436,12 @@ xfs_file_splice_write(
 
 	xfs_ilock(ip, XFS_IOLOCK_EXCL);
 
-	new_size = *ppos + count;
-
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	if (new_size > i_size_read(inode))
-		ip->i_new_size = new_size;
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-
 	trace_xfs_file_splice_write(ip, count, *ppos, ioflags);
 
 	ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags);
 	if (ret > 0)
 		XFS_STATS_ADD(xs_write_bytes, ret);
 
-	xfs_aio_write_newsize_update(ip, new_size);
 	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 	return ret;
 }
@@ -668,16 +638,13 @@ xfs_file_aio_write_checks(
 	struct file		*file,
 	loff_t			*pos,
 	size_t			*count,
-	xfs_fsize_t		*new_sizep,
 	int			*iolock)
 {
 	struct inode		*inode = file->f_mapping->host;
 	struct xfs_inode	*ip = XFS_I(inode);
-	xfs_fsize_t		new_size;
 	int			error = 0;
 
 	xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
-	*new_sizep = 0;
 restart:
 	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
 	if (error) {
@@ -692,15 +659,13 @@ restart:
 	/*
 	 * If the offset is beyond the size of the file, we need to zero any
 	 * blocks that fall between the existing EOF and the start of this
-	 * write. There is no need to issue zeroing if another in-flght IO ends
-	 * at or before this one If zeronig is needed and we are currently
-	 * holding the iolock shared, we need to update it to exclusive which
-	 * involves dropping all locks and relocking to maintain correct locking
-	 * order. If we do this, restart the function to ensure all checks and
-	 * values are still valid.
+	 * write.  If zeroing is needed and we are currently holding the
+	 * iolock shared, we need to update it to exclusive which involves
+	 * dropping all locks and relocking to maintain correct locking order.
+	 * If we do this, restart the function to ensure all checks and values
+	 * are still valid.
 	 */
-	if ((ip->i_new_size && *pos > ip->i_new_size) ||
-	    (!ip->i_new_size && *pos > i_size_read(inode))) {
+	if (*pos > i_size_read(inode)) {
 		if (*iolock == XFS_IOLOCK_SHARED) {
 			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
 			*iolock = XFS_IOLOCK_EXCL;
@@ -709,19 +674,6 @@ restart:
 		}
 		error = -xfs_zero_eof(ip, *pos, i_size_read(inode));
 	}
-
-	/*
-	 * If this IO extends beyond EOF, we may need to update ip->i_new_size.
-	 * We have already zeroed space beyond EOF (if necessary).  Only update
-	 * ip->i_new_size if this IO ends beyond any other in-flight writes.
-	 */
-	new_size = *pos + *count;
-	if (new_size > i_size_read(inode)) {
-		if (new_size > ip->i_new_size)
-			ip->i_new_size = new_size;
-		*new_sizep = new_size;
-	}
-
 	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
 	if (error)
 		return error;
@@ -767,7 +719,6 @@ xfs_file_dio_aio_write(
 	unsigned long		nr_segs,
 	loff_t			pos,
 	size_t			ocount,
-	xfs_fsize_t		*new_size,
 	int			*iolock)
 {
 	struct file		*file = iocb->ki_filp;
@@ -801,7 +752,7 @@ xfs_file_dio_aio_write(
 		*iolock = XFS_IOLOCK_SHARED;
 	xfs_rw_ilock(ip, *iolock);
 
-	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
+	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
 	if (ret)
 		return ret;
 
@@ -850,7 +801,6 @@ xfs_file_buffered_aio_write(
 	unsigned long		nr_segs,
 	loff_t			pos,
 	size_t			ocount,
-	xfs_fsize_t		*new_size,
 	int			*iolock)
 {
 	struct file		*file = iocb->ki_filp;
@@ -864,7 +814,7 @@ xfs_file_buffered_aio_write(
 	*iolock = XFS_IOLOCK_EXCL;
 	xfs_rw_ilock(ip, *iolock);
 
-	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
+	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
 	if (ret)
 		return ret;
 
@@ -904,7 +854,6 @@ xfs_file_aio_write(
 	ssize_t			ret;
 	int			iolock;
 	size_t			ocount = 0;
-	xfs_fsize_t		new_size = 0;
 
 	XFS_STATS_INC(xs_write_calls);
 
@@ -924,10 +873,10 @@ xfs_file_aio_write(
 
 	if (unlikely(file->f_flags & O_DIRECT))
 		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &new_size, &iolock);
+						ocount, &iolock);
 	else
 		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &new_size, &iolock);
+						ocount, &iolock);
 
 	if (ret <= 0)
 		goto out_unlock;
@@ -952,7 +901,6 @@ xfs_file_aio_write(
 	}
 
 out_unlock:
-	xfs_aio_write_newsize_update(ip, new_size);
 	xfs_rw_iunlock(ip, iolock);
 	return ret;
 }
diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 59e8096..70c7739 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -894,7 +894,6 @@ DECLARE_EVENT_CLASS(xfs_file_class,
 		__field(dev_t, dev)
 		__field(xfs_ino_t, ino)
 		__field(xfs_fsize_t, size)
-		__field(xfs_fsize_t, new_size)
 		__field(loff_t, offset)
 		__field(size_t, count)
 		__field(int, flags)
@@ -903,17 +902,15 @@ DECLARE_EVENT_CLASS(xfs_file_class,
 		__entry->dev = VFS_I(ip)->i_sb->s_dev;
 		__entry->ino = ip->i_ino;
 		__entry->size = ip->i_d.di_size;
-		__entry->new_size = ip->i_new_size;
 		__entry->offset = offset;
 		__entry->count = count;
 		__entry->flags = flags;
 	),
-	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx "
 		  "offset 0x%llx count 0x%zx ioflags %s",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->ino,
 		  __entry->size,
-		  __entry->new_size,
 		  __entry->offset,
 		  __entry->count,
 		  __print_flags(__entry->flags, "|", XFS_IO_FLAGS))
@@ -981,7 +978,6 @@ DECLARE_EVENT_CLASS(xfs_imap_class,
 		__field(dev_t, dev)
 		__field(xfs_ino_t, ino)
 		__field(loff_t, size)
-		__field(loff_t, new_size)
 		__field(loff_t, offset)
 		__field(size_t, count)
 		__field(int, type)
@@ -993,7 +989,6 @@ DECLARE_EVENT_CLASS(xfs_imap_class,
 		__entry->dev = VFS_I(ip)->i_sb->s_dev;
 		__entry->ino = ip->i_ino;
 		__entry->size = ip->i_d.di_size;
-		__entry->new_size = ip->i_new_size;
 		__entry->offset = offset;
 		__entry->count = count;
 		__entry->type = type;
@@ -1001,13 +996,11 @@ DECLARE_EVENT_CLASS(xfs_imap_class,
 		__entry->startblock = irec ? irec->br_startblock : 0;
 		__entry->blockcount = irec ? irec->br_blockcount : 0;
 	),
-	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
-		  "offset 0x%llx count %zd type %s "
-		  "startoff 0x%llx startblock %lld blockcount 0x%llx",
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx offset 0x%llx count %zd "
+		  "type %s startoff 0x%llx startblock %lld blockcount 0x%llx",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->ino,
 		  __entry->size,
-		  __entry->new_size,
 		  __entry->offset,
 		  __entry->count,
 		  __print_symbolic(__entry->type, XFS_IO_TYPES),
@@ -1033,7 +1026,6 @@ DECLARE_EVENT_CLASS(xfs_simple_io_class,
 		__field(dev_t, dev)
 		__field(xfs_ino_t, ino)
 		__field(loff_t, size)
-		__field(loff_t, new_size)
 		__field(loff_t, offset)
 		__field(size_t, count)
 	),
@@ -1041,16 +1033,14 @@ DECLARE_EVENT_CLASS(xfs_simple_io_class,
 		__entry->dev = VFS_I(ip)->i_sb->s_dev;
 		__entry->ino = ip->i_ino;
 		__entry->size = ip->i_d.di_size;
-		__entry->new_size = ip->i_new_size;
 		__entry->offset = offset;
 		__entry->count = count;
 	),
-	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx "
 		  "offset 0x%llx count %zd",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->ino,
 		  __entry->size,
-		  __entry->new_size,
 		  __entry->offset,
 		  __entry->count)
 );
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index 4c8df0b..6364fc0 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -96,7 +96,6 @@ xfs_inode_alloc(
 	ip->i_update_core = 0;
 	ip->i_delayed_blks = 0;
 	memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
-	ip->i_new_size = 0;
 
 	return ip;
 }
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index ed1c89e..1a181d5 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -256,7 +256,6 @@ typedef struct xfs_inode {
 
 	xfs_icdinode_t		i_d;		/* most of ondisk inode */
 
-	xfs_fsize_t		i_new_size;	/* size when write completes */
 	atomic_t		i_iocount;	/* outstanding I/O count */
 
 	/* VFS inode */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 031/102] xfs: always return with the iolock held from

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 5bf1f26227a59b9634e95eb3c7c012b766e5e6a0
 xfs_file_aio_write_checks

While xfs_iunlock is fine with 0 lockflags the calling conventions are much
cleaner if xfs_file_aio_write_checks never returns without the iolock held.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index e3076f0..b9e3101 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -631,7 +631,9 @@ out_lock:
 /*
  * Common pre-write limit and setup checks.
  *
- * Returns with iolock held according to @iolock.
+ * Called with the iolocked held either shared and exclusive according to
+ * @iolock, and returns with it held.  Might upgrade the iolock to exclusive
+ * if called for a direct write beyond i_size.
  */
 STATIC ssize_t
 xfs_file_aio_write_checks(
@@ -648,8 +650,7 @@ xfs_file_aio_write_checks(
 restart:
 	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
 	if (error) {
-		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
-		*iolock = 0;
+		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
 		return error;
 	}
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 032/102] xfs: cleanup xfs_file_aio_write

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: d060646436233912178e6b9e3a7f30a41214220f

With all the size field updates out of the way xfs_file_aio_write can
be further simplified by pushing all iolock handling into
xfs_file_dio_aio_write and xfs_file_buffered_aio_write and using
the generic generic_write_sync helper for synchronous writes.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |   90 +++++++++++++++++++------------------------
 1 file changed, 39 insertions(+), 51 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index b9e3101..d9f5f9f 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -719,8 +719,7 @@ xfs_file_dio_aio_write(
 	const struct iovec	*iovp,
 	unsigned long		nr_segs,
 	loff_t			pos,
-	size_t			ocount,
-	int			*iolock)
+	size_t			ocount)
 {
 	struct file		*file = iocb->ki_filp;
 	struct address_space	*mapping = file->f_mapping;
@@ -730,10 +729,10 @@ xfs_file_dio_aio_write(
 	ssize_t			ret = 0;
 	size_t			count = ocount;
 	int			unaligned_io = 0;
+	int			iolock;
 	struct xfs_buftarg	*target = XFS_IS_REALTIME_INODE(ip) ?
 					mp->m_rtdev_targp : mp->m_ddev_targp;
 
-	*iolock = 0;
 	if ((pos & target->bt_smask) || (count & target->bt_smask))
 		return -XFS_ERROR(EINVAL);
 
@@ -748,31 +747,31 @@ xfs_file_dio_aio_write(
 	 * EOF zeroing cases and fill out the new inode size as appropriate.
 	 */
 	if (unaligned_io || mapping->nrpages)
-		*iolock = XFS_IOLOCK_EXCL;
+		iolock = XFS_IOLOCK_EXCL;
 	else
-		*iolock = XFS_IOLOCK_SHARED;
-	xfs_rw_ilock(ip, *iolock);
-
-	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
-	if (ret)
-		return ret;
+		iolock = XFS_IOLOCK_SHARED;
+	xfs_rw_ilock(ip, iolock);
 
 	/*
 	 * Recheck if there are cached pages that need invalidate after we got
 	 * the iolock to protect against other threads adding new pages while
 	 * we were waiting for the iolock.
 	 */
-	if (mapping->nrpages && *iolock == XFS_IOLOCK_SHARED) {
-		xfs_rw_iunlock(ip, *iolock);
-		*iolock = XFS_IOLOCK_EXCL;
-		xfs_rw_ilock(ip, *iolock);
+	if (mapping->nrpages && iolock == XFS_IOLOCK_SHARED) {
+		xfs_rw_iunlock(ip, iolock);
+		iolock = XFS_IOLOCK_EXCL;
+		xfs_rw_ilock(ip, iolock);
 	}
 
+	ret = xfs_file_aio_write_checks(file, &pos, &count, &iolock);
+	if (ret)
+		goto out;
+
 	if (mapping->nrpages) {
 		ret = -xfs_flushinval_pages(ip, (pos & PAGE_CACHE_MASK), -1,
 							FI_REMAPF_LOCKED);
 		if (ret)
-			return ret;
+			goto out;
 	}
 
 	/*
@@ -781,15 +780,18 @@ xfs_file_dio_aio_write(
 	 */
 	if (unaligned_io)
 		xfs_ioend_wait(ip);
-	else if (*iolock == XFS_IOLOCK_EXCL) {
+	else if (iolock == XFS_IOLOCK_EXCL) {
 		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
-		*iolock = XFS_IOLOCK_SHARED;
+		iolock = XFS_IOLOCK_SHARED;
 	}
 
 	trace_xfs_file_direct_write(ip, count, iocb->ki_pos, 0);
 	ret = generic_file_direct_write(iocb, iovp,
 			&nr_segs, pos, &iocb->ki_pos, count, ocount);
 
+out:
+	xfs_rw_iunlock(ip, iolock);
+
 	/* No fallback to buffered IO on errors for XFS. */
 	ASSERT(ret < 0 || ret == count);
 	return ret;
@@ -801,8 +803,7 @@ xfs_file_buffered_aio_write(
 	const struct iovec	*iovp,
 	unsigned long		nr_segs,
 	loff_t			pos,
-	size_t			ocount,
-	int			*iolock)
+	size_t			ocount)
 {
 	struct file		*file = iocb->ki_filp;
 	struct address_space	*mapping = file->f_mapping;
@@ -810,14 +811,14 @@ xfs_file_buffered_aio_write(
 	struct xfs_inode	*ip = XFS_I(inode);
 	ssize_t			ret;
 	int			enospc = 0;
+	int			iolock = XFS_IOLOCK_EXCL;
 	size_t			count = ocount;
 
-	*iolock = XFS_IOLOCK_EXCL;
-	xfs_rw_ilock(ip, *iolock);
+	xfs_rw_ilock(ip, iolock);
 
-	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
+	ret = xfs_file_aio_write_checks(file, &pos, &count, &iolock);
 	if (ret)
-		return ret;
+		goto out;
 
 	/* We can write back this queue in page reclaim */
 	current->backing_dev_info = mapping->backing_dev_info;
@@ -831,13 +832,15 @@ write_retry:
 	 * page locks and retry *once*
 	 */
 	if (ret == -ENOSPC && !enospc) {
-		ret = -xfs_flush_pages(ip, 0, -1, 0, FI_NONE);
-		if (ret)
-			return ret;
 		enospc = 1;
-		goto write_retry;
+		ret = -xfs_flush_pages(ip, 0, -1, 0, FI_NONE);
+		if (!ret)
+			goto write_retry;
 	}
+
 	current->backing_dev_info = NULL;
+out:
+	xfs_rw_iunlock(ip, iolock);
 	return ret;
 }
 
@@ -853,7 +856,6 @@ xfs_file_aio_write(
 	struct inode		*inode = mapping->host;
 	struct xfs_inode	*ip = XFS_I(inode);
 	ssize_t			ret;
-	int			iolock;
 	size_t			ocount = 0;
 
 	XFS_STATS_INC(xs_write_calls);
@@ -873,36 +875,22 @@ xfs_file_aio_write(
 		return -EIO;
 
 	if (unlikely(file->f_flags & O_DIRECT))
-		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &iolock);
+		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos, ocount);
 	else
 		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
-						ocount, &iolock);
-
-	if (ret <= 0)
-		goto out_unlock;
+						  ocount);
 
-	XFS_STATS_ADD(xs_write_bytes, ret);
+	if (ret > 0) {
+		ssize_t err;
 
-	/* Handle various SYNC-type writes */
-	if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
-		loff_t end = pos + ret - 1;
-		int error, error2;
-
-		xfs_rw_iunlock(ip, iolock);
-		error = filemap_write_and_wait_range(mapping, pos, end);
-		xfs_rw_ilock(ip, iolock);
+		XFS_STATS_ADD(xs_write_bytes, ret);
 
-		error2 = -xfs_file_fsync(file,
-					 (file->f_flags & __O_SYNC) ? 0 : 1);
-		if (error)
-			ret = error;
-		else if (error2)
-			ret = error2;
+		/* Handle various SYNC-type writes */
+		err = generic_write_sync(file, pos, ret);
+		if (err < 0)
+			ret = err;
 	}
 
-out_unlock:
-	xfs_rw_iunlock(ip, iolock);
 	return ret;
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 033/102] xfs: pass KM_SLEEP flag to kmem_realloc() in

[Dave Chinner]

From: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>

Upstream commit: 4505360376637832f79f84f352588b0a045ad113
 xlog_recover_add_to_cnt_trans()

The kmem_realloc() in xfs is given KM_* memory allocation flags. And it
allocates memory using kmalloc() after they are converted to gfp_mask
flags. In xlog_recover_add_to_cont_trans(), 0u is passed to kmem_realloc(),
instead of them. I guess it is preferred to use them, and here memory must
be allocated but don't have to be done with GFP_ATOMIC. So, this patch
changes it to KM_SLEEP.

Signed-off-by: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>
Cc: Ben Myers <bpm@sgi.com>
Cc: Alex Elder <elder@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log_recover.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index dfd872a..eead6dd 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -1488,7 +1488,7 @@ xlog_recover_add_to_cont_trans(
 	old_ptr = item->ri_buf[item->ri_cnt-1].i_addr;
 	old_len = item->ri_buf[item->ri_cnt-1].i_len;
 
-	ptr = kmem_realloc(old_ptr, len+old_len, old_len, 0u);
+	ptr = kmem_realloc(old_ptr, len+old_len, old_len, KM_SLEEP);
 	memcpy(&ptr[old_len], dp, len); /* d, s, l */
 	item->ri_buf[item->ri_cnt-1].i_len += len;
 	item->ri_buf[item->ri_cnt-1].i_addr = ptr;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 034/102] xfs: show uuid when mount fails due to duplicate uuid

[Dave Chinner]

From: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>

Upstream commit: 021000e59c0db2d3a8113e906bde3183c33fa84b

When a system tries to mount a filesystem (FS) using UUID, the xfs
returns -EINVAL and shows a message if a FS with the same UUID has
been already mounted. It is useful to output the duplicate UUID
with it.

Signed-off-by: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Ben Myers <bpm@sgi.com>
Cc: Alex Elder <elder@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_mount.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 3effa7f..2f011fd 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -158,7 +158,7 @@ xfs_uuid_mount(
 
  out_duplicate:
 	mutex_unlock(&xfs_uuid_table_mutex);
-	xfs_warn(mp, "Filesystem has duplicate UUID - can't mount");
+	xfs_warn(mp, "Filesystem has duplicate UUID %pU - can't mount", uuid);
 	return XFS_ERROR(EINVAL);
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 035/102] xfs: xfs_trans_add_item() - don't assign in ASSERT() when compare is intended

[Dave Chinner]

From: Jesper Juhl <jj@chaosbits.net>

Upstream commit: f65020a83ad570c1788f7d8ece67f3487166576b

It looks to me like the two ASSERT()s in xfs_trans_add_item() really
want to do a compare (==) rather than assignment (=).
This patch changes it from the latter to the former.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Ben Myers <bpm@sgi.com>

(cherry picked from commit 05293485a0b6b1f803e8a3c0ff188c38f6969985)
---
 fs/xfs/xfs_trans.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index efc147f..9b4437d 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -1151,8 +1151,8 @@ xfs_trans_add_item(
 {
 	struct xfs_log_item_desc *lidp;
 
-	ASSERT(lip->li_mountp = tp->t_mountp);
-	ASSERT(lip->li_ailp = tp->t_mountp->m_ail);
+	ASSERT(lip->li_mountp == tp->t_mountp);
+	ASSERT(lip->li_ailp == tp->t_mountp->m_ail);
 
 	lidp = kmem_zone_zalloc(xfs_log_item_desc_zone, KM_SLEEP | KM_NOFS);
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 036/102] xfs: split tail_lsn assignments from log space wakeups

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 09a423a3d6c70905f1090f01aadb8e6abff527ce

Currently xfs_log_move_tail has a tail_lsn argument that is horribly
overloaded: it may contain either an actual lsn to assign to the log tail,
0 as a special case to use the last sync LSN, or 1 to indicate that no tail
LSN assignment should be performed, and we should opportunisticly wake up
at one task waiting for log space even if we did not move the LSN.

Remove the tail lsn assigned from xfs_log_move_tail and make the two callers
use xlog_assign_tail_lsn instead of the current variant of partially using
the code in xfs_log_move_tail and partially opencoding it.  Note that means
we grow an addition lock roundtrip on the AIL lock for each bulk update
or delete, which is still far less than what we had before introducing the
bulk operations.  If this proves to be a problem we can still add a variant
of xlog_assign_tail_lsn that expects the lock to be held already.

Also rename the remainder of xfs_log_move_tail to xfs_log_space_wake as
that name describes its functionality much better.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c       |   74 ++++++++++++++++++++----------------------------
 fs/xfs/xfs_log.h       |    5 ++--
 fs/xfs/xfs_log_priv.h  |    1 -
 fs/xfs/xfs_trans_ail.c |   45 +++++++----------------------
 4 files changed, 45 insertions(+), 80 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 42d8a5d..6907c82 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -792,37 +792,35 @@ xfs_log_write(
 	return error;
 }
 
+/*
+ * Wake up processes waiting for log space after we have moved the log tail.
+ *
+ * If opportunistic is set wake up one waiter even if we do not have enough
+ * free space by our strict accounting.
+ */
 void
-xfs_log_move_tail(xfs_mount_t	*mp,
-		  xfs_lsn_t	tail_lsn)
+xfs_log_space_wake(
+	struct xfs_mount	*mp,
+	bool			opportunistic)
 {
-	xlog_ticket_t	*tic;
-	xlog_t		*log = mp->m_log;
-	int		need_bytes, free_bytes;
+	struct xlog_ticket	*tic;
+	struct log		*log = mp->m_log;
+	int			need_bytes, free_bytes;
 
 	if (XLOG_FORCED_SHUTDOWN(log))
 		return;
 
-	if (tail_lsn == 0)
-		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
-
-	/* tail_lsn == 1 implies that we weren't passed a valid value.  */
-	if (tail_lsn != 1)
-		atomic64_set(&log->l_tail_lsn, tail_lsn);
-
 	if (!list_empty_careful(&log->l_writeq)) {
-#ifdef DEBUG
-		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
-			panic("Recovery problem");
-#endif
+		ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
+
 		spin_lock(&log->l_grant_write_lock);
 		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
 		list_for_each_entry(tic, &log->l_writeq, t_queue) {
 			ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
 
-			if (free_bytes < tic->t_unit_res && tail_lsn != 1)
+			if (free_bytes < tic->t_unit_res && !opportunistic)
 				break;
-			tail_lsn = 0;
+			opportunistic = false;
 			free_bytes -= tic->t_unit_res;
 			trace_xfs_log_regrant_write_wake_up(log, tic);
 			wake_up(&tic->t_wait);
@@ -831,10 +829,8 @@ xfs_log_move_tail(xfs_mount_t	*mp,
 	}
 
 	if (!list_empty_careful(&log->l_reserveq)) {
-#ifdef DEBUG
-		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
-			panic("Recovery problem");
-#endif
+		ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
+
 		spin_lock(&log->l_grant_reserve_lock);
 		free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
 		list_for_each_entry(tic, &log->l_reserveq, t_queue) {
@@ -842,9 +838,9 @@ xfs_log_move_tail(xfs_mount_t	*mp,
 				need_bytes = tic->t_unit_res*tic->t_cnt;
 			else
 				need_bytes = tic->t_unit_res;
-			if (free_bytes < need_bytes && tail_lsn != 1)
+			if (free_bytes < need_bytes && !opportunistic)
 				break;
-			tail_lsn = 0;
+			opportunistic = false;
 			free_bytes -= need_bytes;
 			trace_xfs_log_grant_wake_up(log, tic);
 			wake_up(&tic->t_wait);
@@ -899,21 +895,7 @@ xfs_log_need_covered(xfs_mount_t *mp)
 	return needed;
 }
 
-/******************************************************************************
- *
- *	local routines
- *
- ******************************************************************************
- */
-
-/* xfs_trans_tail_ail returns 0 when there is nothing in the list.
- * The log manager must keep track of the last LR which was committed
- * to disk.  The lsn of this LR will become the new tail_lsn whenever
- * xfs_trans_tail_ail returns 0.  If we don't do this, we run into
- * the situation where stuff could be written into the log but nothing
- * was ever in the AIL when asked.  Eventually, we panic since the
- * tail hits the head.
- *
+/*
  * We may be holding the log iclog lock upon entering this routine.
  */
 xfs_lsn_t
@@ -923,10 +905,17 @@ xlog_assign_tail_lsn(
 	xfs_lsn_t		tail_lsn;
 	struct log		*log = mp->m_log;
 
+	/*
+	 * To make sure we always have a valid LSN for the log tail we keep
+	 * track of the last LSN which was committed in log->l_last_sync_lsn,
+	 * and use that when the AIL was empty and xfs_ail_min_lsn returns 0.
+	 *
+	 * If the AIL has been emptied we also need to wake any process
+	 * waiting for this condition.
+	 */
 	tail_lsn = xfs_ail_min_lsn(mp->m_ail);
 	if (!tail_lsn)
 		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
-
 	atomic64_set(&log->l_tail_lsn, tail_lsn);
 	return tail_lsn;
 }
@@ -2807,9 +2796,8 @@ xlog_ungrant_log_space(xlog_t	     *log,
 
 	trace_xfs_log_ungrant_exit(log, ticket);
 
-	xfs_log_move_tail(log->l_mp, 1);
-}	/* xlog_ungrant_log_space */
-
+	xfs_log_space_wake(log->l_mp, true);
+}
 
 /*
  * Flush iclog to disk if this is the last reference to the given iclog and
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index 78c9039..e9ef054 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -160,8 +160,9 @@ int	  xfs_log_mount(struct xfs_mount	*mp,
 			xfs_daddr_t		start_block,
 			int		 	num_bblocks);
 int	  xfs_log_mount_finish(struct xfs_mount *mp);
-void	  xfs_log_move_tail(struct xfs_mount	*mp,
-			    xfs_lsn_t		tail_lsn);
+xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
+void	  xfs_log_space_wake(struct xfs_mount	*mp,
+			     bool		opportunistic);
 int	  xfs_log_notify(struct xfs_mount	*mp,
 			 struct xlog_in_core	*iclog,
 			 xfs_log_callback_t	*callback_entry);
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 2d3b6a4..785905e 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -545,7 +545,6 @@ typedef struct log {
 #define XLOG_FORCED_SHUTDOWN(log)	((log)->l_flags & XLOG_IO_ERROR)
 
 /* common routines */
-extern xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
 extern int	 xlog_recover(xlog_t *log);
 extern int	 xlog_recover_finish(xlog_t *log);
 extern void	 xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index 409ac37..db5cd1b 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -694,15 +694,15 @@ xfs_trans_unlocked_item(
 	 * at the tail, it doesn't matter what result we get back.  This
 	 * is slightly racy because since we were just unlocked, we could
 	 * go to sleep between the call to xfs_ail_min and the call to
-	 * xfs_log_move_tail, have someone else lock us, commit to us disk,
+	 * xfs_log_space_wake, have someone else lock us, commit to us disk,
 	 * move us out of the tail of the AIL, and then we wake up.  However,
-	 * the call to xfs_log_move_tail() doesn't do anything if there's
+	 * the call to xfs_log_space_wake() doesn't do anything if there's
 	 * not enough free space to wake people up so we're safe calling it.
 	 */
 	min_lip = xfs_ail_min(ailp);
 
 	if (min_lip == lip)
-		xfs_log_move_tail(ailp->xa_mount, 1);
+		xfs_log_space_wake(ailp->xa_mount, true);
 }	/* xfs_trans_unlocked_item */
 
 /*
@@ -736,7 +736,6 @@ xfs_trans_ail_update_bulk(
 	xfs_lsn_t		lsn) __releases(ailp->xa_lock)
 {
 	xfs_log_item_t		*mlip;
-	xfs_lsn_t		tail_lsn;
 	int			mlip_changed = 0;
 	int			i;
 	LIST_HEAD(tmp);
@@ -761,22 +760,12 @@ xfs_trans_ail_update_bulk(
 	}
 
 	xfs_ail_splice(ailp, cur, &tmp, lsn);
+	spin_unlock(&ailp->xa_lock);
 
-	if (!mlip_changed) {
-		spin_unlock(&ailp->xa_lock);
-		return;
+	if (mlip_changed && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
+		xlog_assign_tail_lsn(ailp->xa_mount);
+		xfs_log_space_wake(ailp->xa_mount, false);
 	}
-
-	/*
-	 * It is not safe to access mlip after the AIL lock is dropped, so we
-	 * must get a copy of li_lsn before we do so.  This is especially
-	 * important on 32-bit platforms where accessing and updating 64-bit
-	 * values like li_lsn is not atomic.
-	 */
-	mlip = xfs_ail_min(ailp);
-	tail_lsn = mlip->li_lsn;
-	spin_unlock(&ailp->xa_lock);
-	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
 }
 
 /*
@@ -807,7 +796,6 @@ xfs_trans_ail_delete_bulk(
 	int			nr_items) __releases(ailp->xa_lock)
 {
 	xfs_log_item_t		*mlip;
-	xfs_lsn_t		tail_lsn;
 	int			mlip_changed = 0;
 	int			i;
 
@@ -834,23 +822,12 @@ xfs_trans_ail_delete_bulk(
 		if (mlip == lip)
 			mlip_changed = 1;
 	}
+	spin_unlock(&ailp->xa_lock);
 
-	if (!mlip_changed) {
-		spin_unlock(&ailp->xa_lock);
-		return;
+	if (mlip_changed && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
+		xlog_assign_tail_lsn(ailp->xa_mount);
+		xfs_log_space_wake(ailp->xa_mount, false);
 	}
-
-	/*
-	 * It is not safe to access mlip after the AIL lock is dropped, so we
-	 * must get a copy of li_lsn before we do so.  This is especially
-	 * important on 32-bit platforms where accessing and updating 64-bit
-	 * values like li_lsn is not atomic. It is possible we've emptied the
-	 * AIL here, so if that is the case, pass an LSN of 0 to the tail move.
-	 */
-	mlip = xfs_ail_min(ailp);
-	tail_lsn = mlip ? mlip->li_lsn : 0;
-	spin_unlock(&ailp->xa_lock);
-	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
 }
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 037/102] xfs: do exact log space wakeups in xlog_ungrant_log_space

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 3af1de753b3caf9fa3762b4b1b85d833c121847e

The only reason that xfs_log_space_wake had to do opportunistic wakeups
was that the old xfs_log_move_tail calling convention didn't allow for
exact wakeups when not updating the log tail LSN.  Since this issue has
been fixed we can do exact wakeups now.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 6907c82..ccb6199 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -2796,7 +2796,7 @@ xlog_ungrant_log_space(xlog_t	     *log,
 
 	trace_xfs_log_ungrant_exit(log, ticket);
 
-	xfs_log_space_wake(log->l_mp, true);
+	xfs_log_space_wake(log->l_mp, false);
 }
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 038/102] xfs: remove xfs_trans_unlocked_item

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 5b03ff1b2444ddf7b8084b7505101e97257aff5a

There is no reason to wake up log space waiters when unlocking inodes or
dquots, and the commit log has no explanation for this function either.

Given that we now have exact log space wakeups everywhere we can assume
the reason for this function was to paper over log space races in earlier
XFS versions.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/quota/xfs_dquot.c |    5 -----
 fs/xfs/xfs_iget.c        |   13 +------------
 fs/xfs/xfs_inode.h       |    4 +---
 fs/xfs/xfs_inode_item.c  |    6 +-----
 fs/xfs/xfs_trans_ail.c   |   44 --------------------------------------------
 fs/xfs/xfs_trans_buf.c   |   26 +-------------------------
 fs/xfs/xfs_trans_priv.h  |    3 ---
 7 files changed, 4 insertions(+), 97 deletions(-)

diff --git a/fs/xfs/quota/xfs_dquot.c b/fs/xfs/quota/xfs_dquot.c
index b8be8a0..85176d6 100644
--- a/fs/xfs/quota/xfs_dquot.c
+++ b/fs/xfs/quota/xfs_dquot.c
@@ -1278,11 +1278,6 @@ xfs_dqunlock(
 	xfs_dquot_t *dqp)
 {
 	mutex_unlock(&(dqp->q_qlock));
-	if (dqp->q_logitem.qli_dquot == dqp) {
-		/* Once was dqp->q_mount, but might just have been cleared */
-		xfs_trans_unlocked_item(dqp->q_logitem.qli_item.li_ailp,
-					(xfs_log_item_t*)&(dqp->q_logitem));
-	}
 }
 
 
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index 6364fc0..b8c110f 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -654,8 +654,7 @@ xfs_iunlock(
 	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
 	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
 	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
-	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_IUNLOCK_NONOTIFY |
-			XFS_LOCK_DEP_MASK)) == 0);
+	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_DEP_MASK)) == 0);
 	ASSERT(lock_flags != 0);
 
 	if (lock_flags & XFS_IOLOCK_EXCL)
@@ -668,16 +667,6 @@ xfs_iunlock(
 	else if (lock_flags & XFS_ILOCK_SHARED)
 		mrunlock_shared(&ip->i_lock);
 
-	if ((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) &&
-	    !(lock_flags & XFS_IUNLOCK_NONOTIFY) && ip->i_itemp) {
-		/*
-		 * Let the AIL know that this item has been unlocked in case
-		 * it is in the AIL and anyone is waiting on it.  Don't do
-		 * this if the caller has asked us not to.
-		 */
-		xfs_trans_unlocked_item(ip->i_itemp->ili_item.li_ailp,
-					(xfs_log_item_t*)(ip->i_itemp));
-	}
 	trace_xfs_iunlock(ip, lock_flags, _RET_IP_);
 }
 
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 1a181d5..d3bbc93 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -409,7 +409,6 @@ static inline void xfs_ifunlock(xfs_inode_t *ip)
 #define	XFS_IOLOCK_SHARED	(1<<1)
 #define	XFS_ILOCK_EXCL		(1<<2)
 #define	XFS_ILOCK_SHARED	(1<<3)
-#define	XFS_IUNLOCK_NONOTIFY	(1<<4)
 
 #define XFS_LOCK_MASK		(XFS_IOLOCK_EXCL | XFS_IOLOCK_SHARED \
 				| XFS_ILOCK_EXCL | XFS_ILOCK_SHARED)
@@ -418,8 +417,7 @@ static inline void xfs_ifunlock(xfs_inode_t *ip)
 	{ XFS_IOLOCK_EXCL,	"IOLOCK_EXCL" }, \
 	{ XFS_IOLOCK_SHARED,	"IOLOCK_SHARED" }, \
 	{ XFS_ILOCK_EXCL,	"ILOCK_EXCL" }, \
-	{ XFS_ILOCK_SHARED,	"ILOCK_SHARED" }, \
-	{ XFS_IUNLOCK_NONOTIFY,	"IUNLOCK_NONOTIFY" }
+	{ XFS_ILOCK_SHARED,	"ILOCK_SHARED" }
 
 
 /*
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 391044c..61fb163 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -600,11 +600,7 @@ xfs_inode_item_trylock(
 	/* Stale items should force out the iclog */
 	if (ip->i_flags & XFS_ISTALE) {
 		xfs_ifunlock(ip);
-		/*
-		 * we hold the AIL lock - notify the unlock routine of this
-		 * so it doesn't try to get the lock again.
-		 */
-		xfs_iunlock(ip, XFS_ILOCK_SHARED|XFS_IUNLOCK_NONOTIFY);
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
 		return XFS_ITEM_PINNED;
 	}
 
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index db5cd1b..3934d97 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -662,50 +662,6 @@ xfs_ail_push_all(
 }
 
 /*
- * This is to be called when an item is unlocked that may have
- * been in the AIL.  It will wake up the first member of the AIL
- * wait list if this item's unlocking might allow it to progress.
- * If the item is in the AIL, then we need to get the AIL lock
- * while doing our checking so we don't race with someone going
- * to sleep waiting for this event in xfs_trans_push_ail().
- */
-void
-xfs_trans_unlocked_item(
-	struct xfs_ail	*ailp,
-	xfs_log_item_t	*lip)
-{
-	xfs_log_item_t	*min_lip;
-
-	/*
-	 * If we're forcibly shutting down, we may have
-	 * unlocked log items arbitrarily. The last thing
-	 * we want to do is to move the tail of the log
-	 * over some potentially valid data.
-	 */
-	if (!(lip->li_flags & XFS_LI_IN_AIL) ||
-	    XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
-		return;
-	}
-
-	/*
-	 * This is the one case where we can call into xfs_ail_min()
-	 * without holding the AIL lock because we only care about the
-	 * case where we are at the tail of the AIL.  If the object isn't
-	 * at the tail, it doesn't matter what result we get back.  This
-	 * is slightly racy because since we were just unlocked, we could
-	 * go to sleep between the call to xfs_ail_min and the call to
-	 * xfs_log_space_wake, have someone else lock us, commit to us disk,
-	 * move us out of the tail of the AIL, and then we wake up.  However,
-	 * the call to xfs_log_space_wake() doesn't do anything if there's
-	 * not enough free space to wake people up so we're safe calling it.
-	 */
-	min_lip = xfs_ail_min(ailp);
-
-	if (min_lip == lip)
-		xfs_log_space_wake(ailp->xa_mount, true);
-}	/* xfs_trans_unlocked_item */
-
-/*
  * xfs_trans_ail_update - bulk AIL insertion operation.
  *
  * @xfs_trans_ail_update takes an array of log items that all need to be
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c
index 26abce5..d733bb8 100644
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -457,26 +457,12 @@ xfs_trans_brelse(xfs_trans_t	*tp,
 		 xfs_buf_t	*bp)
 {
 	xfs_buf_log_item_t	*bip;
-	xfs_log_item_t		*lip;
 
 	/*
 	 * Default to a normal brelse() call if the tp is NULL.
 	 */
 	if (tp == NULL) {
 		ASSERT(XFS_BUF_FSPRIVATE2(bp, void *) == NULL);
-		/*
-		 * If there's a buf log item attached to the buffer,
-		 * then let the AIL know that the buffer is being
-		 * unlocked.
-		 */
-		if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
-			lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
-			if (lip->li_type == XFS_LI_BUF) {
-				bip = XFS_BUF_FSPRIVATE(bp,xfs_buf_log_item_t*);
-				xfs_trans_unlocked_item(bip->bli_item.li_ailp,
-							lip);
-			}
-		}
 		xfs_buf_relse(bp);
 		return;
 	}
@@ -551,19 +537,9 @@ xfs_trans_brelse(xfs_trans_t	*tp,
 		ASSERT(!(bip->bli_item.li_flags & XFS_LI_IN_AIL));
 		ASSERT(!(bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF));
 		xfs_buf_item_relse(bp);
-		bip = NULL;
-	}
-	XFS_BUF_SET_FSPRIVATE2(bp, NULL);
-
-	/*
-	 * If we've still got a buf log item on the buffer, then
-	 * tell the AIL that the buffer is being unlocked.
-	 */
-	if (bip != NULL) {
-		xfs_trans_unlocked_item(bip->bli_item.li_ailp,
-					(xfs_log_item_t*)bip);
 	}
 
+	XFS_BUF_SET_FSPRIVATE2(bp, NULL);
 	xfs_buf_relse(bp);
 	return;
 }
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
index 8c0c465..0683936 100644
--- a/fs/xfs/xfs_trans_priv.h
+++ b/fs/xfs/xfs_trans_priv.h
@@ -104,9 +104,6 @@ void			xfs_ail_push(struct xfs_ail *, xfs_lsn_t);
 void			xfs_ail_push_all(struct xfs_ail *);
 xfs_lsn_t		xfs_ail_min_lsn(struct xfs_ail *ailp);
 
-void			xfs_trans_unlocked_item(struct xfs_ail *,
-					xfs_log_item_t *);
-
 struct xfs_log_item *	xfs_trans_ail_cursor_first(struct xfs_ail *ailp,
 					struct xfs_ail_cursor *cur,
 					xfs_lsn_t lsn);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 039/102] xfs: cleanup xfs_log_space_wake

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: cfb7cdca0aca5ee2e2ef491284bf1edc3b581885

Remove the now unused opportunistic parameter, and use the the
xlog_writeq_wake and xlog_reserveq_wake helpers now that we don't have
to care about the opportunistic wakeups.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c       |   35 +++++------------------------------
 fs/xfs/xfs_log.h       |    3 +--
 fs/xfs/xfs_trans_ail.c |    4 ++--
 3 files changed, 8 insertions(+), 34 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index ccb6199..d299223 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -794,18 +794,13 @@ xfs_log_write(
 
 /*
  * Wake up processes waiting for log space after we have moved the log tail.
- *
- * If opportunistic is set wake up one waiter even if we do not have enough
- * free space by our strict accounting.
  */
 void
 xfs_log_space_wake(
-	struct xfs_mount	*mp,
-	bool			opportunistic)
+	struct xfs_mount	*mp)
 {
-	struct xlog_ticket	*tic;
 	struct log		*log = mp->m_log;
-	int			need_bytes, free_bytes;
+	int			free_bytes;
 
 	if (XLOG_FORCED_SHUTDOWN(log))
 		return;
@@ -815,16 +810,7 @@ xfs_log_space_wake(
 
 		spin_lock(&log->l_grant_write_lock);
 		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
-		list_for_each_entry(tic, &log->l_writeq, t_queue) {
-			ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
-
-			if (free_bytes < tic->t_unit_res && !opportunistic)
-				break;
-			opportunistic = false;
-			free_bytes -= tic->t_unit_res;
-			trace_xfs_log_regrant_write_wake_up(log, tic);
-			wake_up(&tic->t_wait);
-		}
+		xlog_writeq_wake(log, &free_bytes);
 		spin_unlock(&log->l_grant_write_lock);
 	}
 
@@ -833,18 +819,7 @@ xfs_log_space_wake(
 
 		spin_lock(&log->l_grant_reserve_lock);
 		free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
-		list_for_each_entry(tic, &log->l_reserveq, t_queue) {
-			if (tic->t_flags & XLOG_TIC_PERM_RESERV)
-				need_bytes = tic->t_unit_res*tic->t_cnt;
-			else
-				need_bytes = tic->t_unit_res;
-			if (free_bytes < need_bytes && !opportunistic)
-				break;
-			opportunistic = false;
-			free_bytes -= need_bytes;
-			trace_xfs_log_grant_wake_up(log, tic);
-			wake_up(&tic->t_wait);
-		}
+		xlog_reserveq_wake(log, &free_bytes);
 		spin_unlock(&log->l_grant_reserve_lock);
 	}
 }
@@ -2796,7 +2771,7 @@ xlog_ungrant_log_space(xlog_t	     *log,
 
 	trace_xfs_log_ungrant_exit(log, ticket);
 
-	xfs_log_space_wake(log->l_mp, false);
+	xfs_log_space_wake(log->l_mp);
 }
 
 /*
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index e9ef054..953358b 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -161,8 +161,7 @@ int	  xfs_log_mount(struct xfs_mount	*mp,
 			int		 	num_bblocks);
 int	  xfs_log_mount_finish(struct xfs_mount *mp);
 xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
-void	  xfs_log_space_wake(struct xfs_mount	*mp,
-			     bool		opportunistic);
+void	  xfs_log_space_wake(struct xfs_mount *mp);
 int	  xfs_log_notify(struct xfs_mount	*mp,
 			 struct xlog_in_core	*iclog,
 			 xfs_log_callback_t	*callback_entry);
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index 3934d97..a0ee8c8 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -720,7 +720,7 @@ xfs_trans_ail_update_bulk(
 
 	if (mlip_changed && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
 		xlog_assign_tail_lsn(ailp->xa_mount);
-		xfs_log_space_wake(ailp->xa_mount, false);
+		xfs_log_space_wake(ailp->xa_mount);
 	}
 }
 
@@ -782,7 +782,7 @@ xfs_trans_ail_delete_bulk(
 
 	if (mlip_changed && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
 		xlog_assign_tail_lsn(ailp->xa_mount);
-		xfs_log_space_wake(ailp->xa_mount, false);
+		xfs_log_space_wake(ailp->xa_mount);
 	}
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 040/102] xfs: remove log space waitqueues

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 14a7235fba4302a82d61150eda92ec90d3ae9cfb

The tic->t_wait waitqueues can never have more than a single waiter
on them, so we can easily replace them with a task_struct pointer
and wake_up_process.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c      |   24 +++++++++++++++---------
 fs/xfs/xfs_log_priv.h |    2 +-
 2 files changed, 16 insertions(+), 10 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index d299223..fc2b84b 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -169,7 +169,7 @@ xlog_reserveq_wake(
 		*free_bytes -= need_bytes;
 
 		trace_xfs_log_grant_wake_up(log, tic);
-		wake_up(&tic->t_wait);
+		wake_up_process(tic->t_task);
 	}
 
 	return true;
@@ -193,7 +193,7 @@ xlog_writeq_wake(
 		*free_bytes -= need_bytes;
 
 		trace_xfs_log_regrant_write_wake_up(log, tic);
-		wake_up(&tic->t_wait);
+		wake_up_process(tic->t_task);
 	}
 
 	return true;
@@ -212,10 +212,13 @@ xlog_reserveq_wait(
 			goto shutdown;
 		xlog_grant_push_ail(log, need_bytes);
 
+		__set_current_state(TASK_UNINTERRUPTIBLE);
+		spin_unlock(&log->l_grant_reserve_lock);
+
 		XFS_STATS_INC(xs_sleep_logspace);
-		trace_xfs_log_grant_sleep(log, tic);
 
-		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
+		trace_xfs_log_grant_sleep(log, tic);
+		schedule();
 		trace_xfs_log_grant_wake(log, tic);
 
 		spin_lock(&log->l_grant_reserve_lock);
@@ -243,10 +246,13 @@ xlog_writeq_wait(
 			goto shutdown;
 		xlog_grant_push_ail(log, need_bytes);
 
+		__set_current_state(TASK_UNINTERRUPTIBLE);
+		spin_unlock(&log->l_grant_write_lock);
+
 		XFS_STATS_INC(xs_sleep_logspace);
-		trace_xfs_log_regrant_write_sleep(log, tic);
 
-		xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
+		trace_xfs_log_regrant_write_sleep(log, tic);
+		schedule();
 		trace_xfs_log_regrant_write_wake(log, tic);
 
 		spin_lock(&log->l_grant_write_lock);
@@ -3327,6 +3333,7 @@ xlog_ticket_alloc(
         }
 
 	atomic_set(&tic->t_ref, 1);
+	tic->t_task		= current;
 	INIT_LIST_HEAD(&tic->t_queue);
 	tic->t_unit_res		= unit_bytes;
 	tic->t_curr_res		= unit_bytes;
@@ -3338,7 +3345,6 @@ xlog_ticket_alloc(
 	tic->t_trans_type	= 0;
 	if (xflags & XFS_LOG_PERM_RESERV)
 		tic->t_flags |= XLOG_TIC_PERM_RESERV;
-	init_waitqueue_head(&tic->t_wait);
 
 	xlog_tic_reset_res(tic);
 
@@ -3666,12 +3672,12 @@ xfs_log_force_umount(
 	 */
 	spin_lock(&log->l_grant_reserve_lock);
 	list_for_each_entry(tic, &log->l_reserveq, t_queue)
-		wake_up(&tic->t_wait);
+		wake_up_process(tic->t_task);
 	spin_unlock(&log->l_grant_reserve_lock);
 
 	spin_lock(&log->l_grant_write_lock);
 	list_for_each_entry(tic, &log->l_writeq, t_queue)
-		wake_up(&tic->t_wait);
+		wake_up_process(tic->t_task);
 	spin_unlock(&log->l_grant_write_lock);
 
 	if (!(log->l_iclog->ic_state & XLOG_STATE_IOERROR)) {
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 785905e..d8c5e47 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -239,8 +239,8 @@ typedef struct xlog_res {
 } xlog_res_t;
 
 typedef struct xlog_ticket {
-	wait_queue_head_t  t_wait;	 /* ticket wait queue */
 	struct list_head   t_queue;	 /* reserve/write queue */
+	struct task_struct *t_task;	 /* task that owns this ticket */
 	xlog_tid_t	   t_tid;	 /* transaction identifier	 : 4  */
 	atomic_t	   t_ref;	 /* ticket reference count       : 4  */
 	int		   t_curr_res;	 /* current reservation in bytes : 4  */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 041/102] xfs: add the xlog_grant_head structure

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 28496968a6ac37c8b8c44b5156e633c581bb8378

Add a new data structure to allow sharing code between the log grant and
regrant code.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |    8 +--
 fs/xfs/xfs_log.c             |  112 +++++++++++++++++++++---------------------
 fs/xfs/xfs_log_priv.h        |   23 ++++-----
 fs/xfs/xfs_log_recover.c     |    4 +-
 4 files changed, 74 insertions(+), 73 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 70c7739..cc7311c 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -786,12 +786,12 @@ DECLARE_EVENT_CLASS(xfs_loggrant_class,
 		__entry->curr_res = tic->t_curr_res;
 		__entry->unit_res = tic->t_unit_res;
 		__entry->flags = tic->t_flags;
-		__entry->reserveq = list_empty(&log->l_reserveq);
-		__entry->writeq = list_empty(&log->l_writeq);
-		xlog_crack_grant_head(&log->l_grant_reserve_head,
+		__entry->reserveq = list_empty(&log->l_reserve_head.waiters);
+		__entry->writeq = list_empty(&log->l_write_head.waiters);
+		xlog_crack_grant_head(&log->l_reserve_head.grant,
 				&__entry->grant_reserve_cycle,
 				&__entry->grant_reserve_bytes);
-		xlog_crack_grant_head(&log->l_grant_write_head,
+		xlog_crack_grant_head(&log->l_write_head.grant,
 				&__entry->grant_write_cycle,
 				&__entry->grant_write_bytes);
 		__entry->curr_cycle = log->l_curr_cycle;
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index fc2b84b..5c5aa54 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -158,7 +158,7 @@ xlog_reserveq_wake(
 	struct xlog_ticket	*tic;
 	int			need_bytes;
 
-	list_for_each_entry(tic, &log->l_reserveq, t_queue) {
+	list_for_each_entry(tic, &log->l_reserve_head.waiters, t_queue) {
 		if (tic->t_flags & XLOG_TIC_PERM_RESERV)
 			need_bytes = tic->t_unit_res * tic->t_cnt;
 		else
@@ -183,7 +183,7 @@ xlog_writeq_wake(
 	struct xlog_ticket	*tic;
 	int			need_bytes;
 
-	list_for_each_entry(tic, &log->l_writeq, t_queue) {
+	list_for_each_entry(tic, &log->l_write_head.waiters, t_queue) {
 		ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
 
 		need_bytes = tic->t_unit_res;
@@ -205,7 +205,7 @@ xlog_reserveq_wait(
 	struct xlog_ticket	*tic,
 	int			need_bytes)
 {
-	list_add_tail(&tic->t_queue, &log->l_reserveq);
+	list_add_tail(&tic->t_queue, &log->l_reserve_head.waiters);
 
 	do {
 		if (XLOG_FORCED_SHUTDOWN(log))
@@ -213,7 +213,7 @@ xlog_reserveq_wait(
 		xlog_grant_push_ail(log, need_bytes);
 
 		__set_current_state(TASK_UNINTERRUPTIBLE);
-		spin_unlock(&log->l_grant_reserve_lock);
+		spin_unlock(&log->l_reserve_head.lock);
 
 		XFS_STATS_INC(xs_sleep_logspace);
 
@@ -221,10 +221,10 @@ xlog_reserveq_wait(
 		schedule();
 		trace_xfs_log_grant_wake(log, tic);
 
-		spin_lock(&log->l_grant_reserve_lock);
+		spin_lock(&log->l_reserve_head.lock);
 		if (XLOG_FORCED_SHUTDOWN(log))
 			goto shutdown;
-	} while (xlog_space_left(log, &log->l_grant_reserve_head) < need_bytes);
+	} while (xlog_space_left(log, &log->l_reserve_head.grant) < need_bytes);
 
 	list_del_init(&tic->t_queue);
 	return 0;
@@ -239,7 +239,7 @@ xlog_writeq_wait(
 	struct xlog_ticket	*tic,
 	int			need_bytes)
 {
-	list_add_tail(&tic->t_queue, &log->l_writeq);
+	list_add_tail(&tic->t_queue, &log->l_write_head.waiters);
 
 	do {
 		if (XLOG_FORCED_SHUTDOWN(log))
@@ -247,7 +247,7 @@ xlog_writeq_wait(
 		xlog_grant_push_ail(log, need_bytes);
 
 		__set_current_state(TASK_UNINTERRUPTIBLE);
-		spin_unlock(&log->l_grant_write_lock);
+		spin_unlock(&log->l_write_head.lock);
 
 		XFS_STATS_INC(xs_sleep_logspace);
 
@@ -255,10 +255,10 @@ xlog_writeq_wait(
 		schedule();
 		trace_xfs_log_regrant_write_wake(log, tic);
 
-		spin_lock(&log->l_grant_write_lock);
+		spin_lock(&log->l_write_head.lock);
 		if (XLOG_FORCED_SHUTDOWN(log))
 			goto shutdown;
-	} while (xlog_space_left(log, &log->l_grant_write_head) < need_bytes);
+	} while (xlog_space_left(log, &log->l_write_head.grant) < need_bytes);
 
 	list_del_init(&tic->t_queue);
 	return 0;
@@ -811,22 +811,22 @@ xfs_log_space_wake(
 	if (XLOG_FORCED_SHUTDOWN(log))
 		return;
 
-	if (!list_empty_careful(&log->l_writeq)) {
+	if (!list_empty_careful(&log->l_write_head.waiters)) {
 		ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
 
-		spin_lock(&log->l_grant_write_lock);
-		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+		spin_lock(&log->l_write_head.lock);
+		free_bytes = xlog_space_left(log, &log->l_write_head.grant);
 		xlog_writeq_wake(log, &free_bytes);
-		spin_unlock(&log->l_grant_write_lock);
+		spin_unlock(&log->l_write_head.lock);
 	}
 
-	if (!list_empty_careful(&log->l_reserveq)) {
+	if (!list_empty_careful(&log->l_reserve_head.waiters)) {
 		ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
 
-		spin_lock(&log->l_grant_reserve_lock);
-		free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+		spin_lock(&log->l_reserve_head.lock);
+		free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
 		xlog_reserveq_wake(log, &free_bytes);
-		spin_unlock(&log->l_grant_reserve_lock);
+		spin_unlock(&log->l_reserve_head.lock);
 	}
 }
 
@@ -1108,12 +1108,12 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	xlog_assign_atomic_lsn(&log->l_tail_lsn, 1, 0);
 	xlog_assign_atomic_lsn(&log->l_last_sync_lsn, 1, 0);
 	log->l_curr_cycle  = 1;	    /* 0 is bad since this is initial value */
-	xlog_assign_grant_head(&log->l_grant_reserve_head, 1, 0);
-	xlog_assign_grant_head(&log->l_grant_write_head, 1, 0);
-	INIT_LIST_HEAD(&log->l_reserveq);
-	INIT_LIST_HEAD(&log->l_writeq);
-	spin_lock_init(&log->l_grant_reserve_lock);
-	spin_lock_init(&log->l_grant_write_lock);
+	xlog_assign_grant_head(&log->l_reserve_head.grant, 1, 0);
+	xlog_assign_grant_head(&log->l_write_head.grant, 1, 0);
+	INIT_LIST_HEAD(&log->l_reserve_head.waiters);
+	INIT_LIST_HEAD(&log->l_write_head.waiters);
+	spin_lock_init(&log->l_reserve_head.lock);
+	spin_lock_init(&log->l_write_head.lock);
 
 	error = EFSCORRUPTED;
 	if (xfs_sb_version_hassector(&mp->m_sb)) {
@@ -1293,7 +1293,7 @@ xlog_grant_push_ail(
 
 	ASSERT(BTOBB(need_bytes) < log->l_logBBsize);
 
-	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+	free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
 	free_blocks = BTOBBT(free_bytes);
 
 	/*
@@ -1427,8 +1427,8 @@ xlog_sync(xlog_t		*log,
 		 roundoff < BBTOB(1)));
 
 	/* move grant heads by roundoff in sync */
-	xlog_grant_add_space(log, &log->l_grant_reserve_head, roundoff);
-	xlog_grant_add_space(log, &log->l_grant_write_head, roundoff);
+	xlog_grant_add_space(log, &log->l_reserve_head.grant, roundoff);
+	xlog_grant_add_space(log, &log->l_write_head.grant, roundoff);
 
 	/* put cycle number in every block */
 	xlog_pack_data(log, iclog, roundoff); 
@@ -2595,8 +2595,8 @@ restart:
  * path. Hence any lock will be globally hot if we take it unconditionally on
  * every pass.
  *
- * As tickets are only ever moved on and off the reserveq under the
- * l_grant_reserve_lock, we only need to take that lock if we are going to add
+ * As tickets are only ever moved on and off the l_reserve.waiters under the
+ * l_reserve.lock, we only need to take that lock if we are going to add
  * the ticket to the queue and sleep. We can avoid taking the lock if the ticket
  * was never added to the reserveq because the t_queue list head will be empty
  * and we hold the only reference to it so it can safely be checked unlocked.
@@ -2622,23 +2622,23 @@ xlog_grant_log_space(
 	need_bytes = tic->t_unit_res;
 	if (tic->t_flags & XFS_LOG_PERM_RESERV)
 		need_bytes *= tic->t_ocnt;
-	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
-	if (!list_empty_careful(&log->l_reserveq)) {
-		spin_lock(&log->l_grant_reserve_lock);
+	free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
+	if (!list_empty_careful(&log->l_reserve_head.waiters)) {
+		spin_lock(&log->l_reserve_head.lock);
 		if (!xlog_reserveq_wake(log, &free_bytes) ||
 		    free_bytes < need_bytes)
 			error = xlog_reserveq_wait(log, tic, need_bytes);
-		spin_unlock(&log->l_grant_reserve_lock);
+		spin_unlock(&log->l_reserve_head.lock);
 	} else if (free_bytes < need_bytes) {
-		spin_lock(&log->l_grant_reserve_lock);
+		spin_lock(&log->l_reserve_head.lock);
 		error = xlog_reserveq_wait(log, tic, need_bytes);
-		spin_unlock(&log->l_grant_reserve_lock);
+		spin_unlock(&log->l_reserve_head.lock);
 	}
 	if (error)
 		return error;
 
-	xlog_grant_add_space(log, &log->l_grant_reserve_head, need_bytes);
-	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
+	xlog_grant_add_space(log, &log->l_reserve_head.grant, need_bytes);
+	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
 	trace_xfs_log_grant_exit(log, tic);
 	xlog_verify_grant_tail(log);
 	return 0;
@@ -2675,23 +2675,23 @@ xlog_regrant_write_log_space(
 	 * otherwise try to get some space for this transaction.
 	 */
 	need_bytes = tic->t_unit_res;
-	free_bytes = xlog_space_left(log, &log->l_grant_write_head);
-	if (!list_empty_careful(&log->l_writeq)) {
-		spin_lock(&log->l_grant_write_lock);
+	free_bytes = xlog_space_left(log, &log->l_write_head.grant);
+	if (!list_empty_careful(&log->l_write_head.waiters)) {
+		spin_lock(&log->l_write_head.lock);
 		if (!xlog_writeq_wake(log, &free_bytes) ||
 		    free_bytes < need_bytes)
 			error = xlog_writeq_wait(log, tic, need_bytes);
-		spin_unlock(&log->l_grant_write_lock);
+		spin_unlock(&log->l_write_head.lock);
 	} else if (free_bytes < need_bytes) {
-		spin_lock(&log->l_grant_write_lock);
+		spin_lock(&log->l_write_head.lock);
 		error = xlog_writeq_wait(log, tic, need_bytes);
-		spin_unlock(&log->l_grant_write_lock);
+		spin_unlock(&log->l_write_head.lock);
 	}
 
 	if (error)
 		return error;
 
-	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
+	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
 	trace_xfs_log_regrant_write_exit(log, tic);
 	xlog_verify_grant_tail(log);
 	return 0;
@@ -2713,9 +2713,9 @@ xlog_regrant_reserve_log_space(xlog_t	     *log,
 	if (ticket->t_cnt > 0)
 		ticket->t_cnt--;
 
-	xlog_grant_sub_space(log, &log->l_grant_reserve_head,
+	xlog_grant_sub_space(log, &log->l_reserve_head.grant,
 					ticket->t_curr_res);
-	xlog_grant_sub_space(log, &log->l_grant_write_head,
+	xlog_grant_sub_space(log, &log->l_write_head.grant,
 					ticket->t_curr_res);
 	ticket->t_curr_res = ticket->t_unit_res;
 	xlog_tic_reset_res(ticket);
@@ -2726,7 +2726,7 @@ xlog_regrant_reserve_log_space(xlog_t	     *log,
 	if (ticket->t_cnt > 0)
 		return;
 
-	xlog_grant_add_space(log, &log->l_grant_reserve_head,
+	xlog_grant_add_space(log, &log->l_reserve_head.grant,
 					ticket->t_unit_res);
 
 	trace_xfs_log_regrant_reserve_exit(log, ticket);
@@ -2772,8 +2772,8 @@ xlog_ungrant_log_space(xlog_t	     *log,
 		bytes += ticket->t_unit_res*ticket->t_cnt;
 	}
 
-	xlog_grant_sub_space(log, &log->l_grant_reserve_head, bytes);
-	xlog_grant_sub_space(log, &log->l_grant_write_head, bytes);
+	xlog_grant_sub_space(log, &log->l_reserve_head.grant, bytes);
+	xlog_grant_sub_space(log, &log->l_write_head.grant, bytes);
 
 	trace_xfs_log_ungrant_exit(log, ticket);
 
@@ -3400,7 +3400,7 @@ xlog_verify_grant_tail(
 	int		tail_cycle, tail_blocks;
 	int		cycle, space;
 
-	xlog_crack_grant_head(&log->l_grant_write_head, &cycle, &space);
+	xlog_crack_grant_head(&log->l_write_head.grant, &cycle, &space);
 	xlog_crack_atomic_lsn(&log->l_tail_lsn, &tail_cycle, &tail_blocks);
 	if (tail_cycle != cycle) {
 		if (cycle - 1 != tail_cycle &&
@@ -3670,15 +3670,15 @@ xfs_log_force_umount(
 	 * we don't enqueue anything once the SHUTDOWN flag is set, and this
 	 * action is protected by the grant locks.
 	 */
-	spin_lock(&log->l_grant_reserve_lock);
-	list_for_each_entry(tic, &log->l_reserveq, t_queue)
+	spin_lock(&log->l_reserve_head.lock);
+	list_for_each_entry(tic, &log->l_reserve_head.waiters, t_queue)
 		wake_up_process(tic->t_task);
-	spin_unlock(&log->l_grant_reserve_lock);
+	spin_unlock(&log->l_reserve_head.lock);
 
-	spin_lock(&log->l_grant_write_lock);
-	list_for_each_entry(tic, &log->l_writeq, t_queue)
+	spin_lock(&log->l_write_head.lock);
+	list_for_each_entry(tic, &log->l_write_head.waiters, t_queue)
 		wake_up_process(tic->t_task);
-	spin_unlock(&log->l_grant_write_lock);
+	spin_unlock(&log->l_write_head.lock);
 
 	if (!(log->l_iclog->ic_state & XLOG_STATE_IOERROR)) {
 		ASSERT(!logerror);
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index d8c5e47..eba4ec9 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -470,6 +470,16 @@ struct xfs_cil {
 #define XLOG_CIL_HARD_SPACE_LIMIT(log)	(3 * (log->l_logsize >> 4))
 
 /*
+ * ticket grant locks, queues and accounting have their own cachlines
+ * as these are quite hot and can be operated on concurrently.
+ */
+struct xlog_grant_head {
+	spinlock_t		lock ____cacheline_aligned_in_smp;
+	struct list_head	waiters;
+	atomic64_t		grant;
+};
+
+/*
  * The reservation head lsn is not made up of a cycle number and block number.
  * Instead, it uses a cycle number and byte number.  Logs don't expect to
  * overflow 31 bits worth of byte offset, so using a byte number will mean
@@ -520,17 +530,8 @@ typedef struct log {
 	/* lsn of 1st LR with unflushed * buffers */
 	atomic64_t		l_tail_lsn ____cacheline_aligned_in_smp;
 
-	/*
-	 * ticket grant locks, queues and accounting have their own cachlines
-	 * as these are quite hot and can be operated on concurrently.
-	 */
-	spinlock_t		l_grant_reserve_lock ____cacheline_aligned_in_smp;
-	struct list_head	l_reserveq;
-	atomic64_t		l_grant_reserve_head;
-
-	spinlock_t		l_grant_write_lock ____cacheline_aligned_in_smp;
-	struct list_head	l_writeq;
-	atomic64_t		l_grant_write_head;
+	struct xlog_grant_head	l_reserve_head;
+	struct xlog_grant_head	l_write_head;
 
 	/* The following field are used for debugging; need to hold icloglock */
 #ifdef DEBUG
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index eead6dd..687bc9f 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -964,9 +964,9 @@ xlog_find_tail(
 		log->l_curr_cycle++;
 	atomic64_set(&log->l_tail_lsn, be64_to_cpu(rhead->h_tail_lsn));
 	atomic64_set(&log->l_last_sync_lsn, be64_to_cpu(rhead->h_lsn));
-	xlog_assign_grant_head(&log->l_grant_reserve_head, log->l_curr_cycle,
+	xlog_assign_grant_head(&log->l_reserve_head.grant, log->l_curr_cycle,
 					BBTOB(log->l_curr_block));
-	xlog_assign_grant_head(&log->l_grant_write_head, log->l_curr_cycle,
+	xlog_assign_grant_head(&log->l_write_head.grant, log->l_curr_cycle,
 					BBTOB(log->l_curr_block));
 
 	/*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 042/102] xfs: add xlog_grant_head_init

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: c303c5b8c3b8eace41c4fba26205b50c0f8e4ca0

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c |   18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 5c5aa54..6ef0264 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -150,6 +150,15 @@ xlog_grant_add_space(
 	} while (head_val != old);
 }
 
+STATIC void
+xlog_grant_head_init(
+	struct xlog_grant_head	*head)
+{
+	xlog_assign_grant_head(&head->grant, 1, 0);
+	INIT_LIST_HEAD(&head->waiters);
+	spin_lock_init(&head->lock);
+}
+
 STATIC bool
 xlog_reserveq_wake(
 	struct log		*log,
@@ -1108,12 +1117,9 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	xlog_assign_atomic_lsn(&log->l_tail_lsn, 1, 0);
 	xlog_assign_atomic_lsn(&log->l_last_sync_lsn, 1, 0);
 	log->l_curr_cycle  = 1;	    /* 0 is bad since this is initial value */
-	xlog_assign_grant_head(&log->l_reserve_head.grant, 1, 0);
-	xlog_assign_grant_head(&log->l_write_head.grant, 1, 0);
-	INIT_LIST_HEAD(&log->l_reserve_head.waiters);
-	INIT_LIST_HEAD(&log->l_write_head.waiters);
-	spin_lock_init(&log->l_reserve_head.lock);
-	spin_lock_init(&log->l_write_head.lock);
+
+	xlog_grant_head_init(&log->l_reserve_head);
+	xlog_grant_head_init(&log->l_write_head);
 
 	error = EFSCORRUPTED;
 	if (xfs_sb_version_hassector(&mp->m_sb)) {
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 043/102] xfs: add xlog_grant_head_wake_all

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: a79bf2d75b8f96bcdb6714138cd53cb3c358669c

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c |   24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 6ef0264..eb19abf 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -159,6 +159,18 @@ xlog_grant_head_init(
 	spin_lock_init(&head->lock);
 }
 
+STATIC void
+xlog_grant_head_wake_all(
+	struct xlog_grant_head	*head)
+{
+	struct xlog_ticket	*tic;
+
+	spin_lock(&head->lock);
+	list_for_each_entry(tic, &head->waiters, t_queue)
+		wake_up_process(tic->t_task);
+	spin_unlock(&head->lock);
+}
+
 STATIC bool
 xlog_reserveq_wake(
 	struct log		*log,
@@ -3608,7 +3620,6 @@ xfs_log_force_umount(
 	struct xfs_mount	*mp,
 	int			logerror)
 {
-	xlog_ticket_t	*tic;
 	xlog_t		*log;
 	int		retval;
 
@@ -3676,15 +3687,8 @@ xfs_log_force_umount(
 	 * we don't enqueue anything once the SHUTDOWN flag is set, and this
 	 * action is protected by the grant locks.
 	 */
-	spin_lock(&log->l_reserve_head.lock);
-	list_for_each_entry(tic, &log->l_reserve_head.waiters, t_queue)
-		wake_up_process(tic->t_task);
-	spin_unlock(&log->l_reserve_head.lock);
-
-	spin_lock(&log->l_write_head.lock);
-	list_for_each_entry(tic, &log->l_write_head.waiters, t_queue)
-		wake_up_process(tic->t_task);
-	spin_unlock(&log->l_write_head.lock);
+	xlog_grant_head_wake_all(&log->l_reserve_head);
+	xlog_grant_head_wake_all(&log->l_write_head);
 
 	if (!(log->l_iclog->ic_state & XLOG_STATE_IOERROR)) {
 		ASSERT(!logerror);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 044/102] xfs: share code for grant head waiting

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 23ee3df349b8b8fd153bd02fccf08b31aec5bce3

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |    2 --
 fs/xfs/xfs_log.c             |   63 ++++++++++++------------------------------
 2 files changed, 18 insertions(+), 47 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index cc7311c..0e6817c 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -841,8 +841,6 @@ DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index eb19abf..7955b1b 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -221,12 +221,13 @@ xlog_writeq_wake(
 }
 
 STATIC int
-xlog_reserveq_wait(
+xlog_grant_head_wait(
 	struct log		*log,
+	struct xlog_grant_head	*head,
 	struct xlog_ticket	*tic,
 	int			need_bytes)
 {
-	list_add_tail(&tic->t_queue, &log->l_reserve_head.waiters);
+	list_add_tail(&tic->t_queue, &head->waiters);
 
 	do {
 		if (XLOG_FORCED_SHUTDOWN(log))
@@ -234,7 +235,7 @@ xlog_reserveq_wait(
 		xlog_grant_push_ail(log, need_bytes);
 
 		__set_current_state(TASK_UNINTERRUPTIBLE);
-		spin_unlock(&log->l_reserve_head.lock);
+		spin_unlock(&head->lock);
 
 		XFS_STATS_INC(xs_sleep_logspace);
 
@@ -242,44 +243,10 @@ xlog_reserveq_wait(
 		schedule();
 		trace_xfs_log_grant_wake(log, tic);
 
-		spin_lock(&log->l_reserve_head.lock);
+		spin_lock(&head->lock);
 		if (XLOG_FORCED_SHUTDOWN(log))
 			goto shutdown;
-	} while (xlog_space_left(log, &log->l_reserve_head.grant) < need_bytes);
-
-	list_del_init(&tic->t_queue);
-	return 0;
-shutdown:
-	list_del_init(&tic->t_queue);
-	return XFS_ERROR(EIO);
-}
-
-STATIC int
-xlog_writeq_wait(
-	struct log		*log,
-	struct xlog_ticket	*tic,
-	int			need_bytes)
-{
-	list_add_tail(&tic->t_queue, &log->l_write_head.waiters);
-
-	do {
-		if (XLOG_FORCED_SHUTDOWN(log))
-			goto shutdown;
-		xlog_grant_push_ail(log, need_bytes);
-
-		__set_current_state(TASK_UNINTERRUPTIBLE);
-		spin_unlock(&log->l_write_head.lock);
-
-		XFS_STATS_INC(xs_sleep_logspace);
-
-		trace_xfs_log_regrant_write_sleep(log, tic);
-		schedule();
-		trace_xfs_log_regrant_write_wake(log, tic);
-
-		spin_lock(&log->l_write_head.lock);
-		if (XLOG_FORCED_SHUTDOWN(log))
-			goto shutdown;
-	} while (xlog_space_left(log, &log->l_write_head.grant) < need_bytes);
+	} while (xlog_space_left(log, &head->grant) < need_bytes);
 
 	list_del_init(&tic->t_queue);
 	return 0;
@@ -2644,12 +2611,15 @@ xlog_grant_log_space(
 	if (!list_empty_careful(&log->l_reserve_head.waiters)) {
 		spin_lock(&log->l_reserve_head.lock);
 		if (!xlog_reserveq_wake(log, &free_bytes) ||
-		    free_bytes < need_bytes)
-			error = xlog_reserveq_wait(log, tic, need_bytes);
+		    free_bytes < need_bytes) {
+			error = xlog_grant_head_wait(log, &log->l_reserve_head,
+						     tic, need_bytes);
+		}
 		spin_unlock(&log->l_reserve_head.lock);
 	} else if (free_bytes < need_bytes) {
 		spin_lock(&log->l_reserve_head.lock);
-		error = xlog_reserveq_wait(log, tic, need_bytes);
+		error = xlog_grant_head_wait(log, &log->l_reserve_head, tic,
+					     need_bytes);
 		spin_unlock(&log->l_reserve_head.lock);
 	}
 	if (error)
@@ -2697,12 +2667,15 @@ xlog_regrant_write_log_space(
 	if (!list_empty_careful(&log->l_write_head.waiters)) {
 		spin_lock(&log->l_write_head.lock);
 		if (!xlog_writeq_wake(log, &free_bytes) ||
-		    free_bytes < need_bytes)
-			error = xlog_writeq_wait(log, tic, need_bytes);
+		    free_bytes < need_bytes) {
+			error = xlog_grant_head_wait(log, &log->l_write_head,
+						     tic, need_bytes);
+		}
 		spin_unlock(&log->l_write_head.lock);
 	} else if (free_bytes < need_bytes) {
 		spin_lock(&log->l_write_head.lock);
-		error = xlog_writeq_wait(log, tic, need_bytes);
+		error = xlog_grant_head_wait(log, &log->l_write_head, tic,
+					     need_bytes);
 		spin_unlock(&log->l_write_head.lock);
 	}
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 045/102] xfs: share code for grant head wakeups

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: e179840d74606ab1935c83fe5ad9d93c95ddc956

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |    1 -
 fs/xfs/xfs_log.c             |   50 +++++++++++++++++-------------------------
 2 files changed, 20 insertions(+), 31 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 0e6817c..3617ba1 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -841,7 +841,6 @@ DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_sub);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 7955b1b..ad16114 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -171,49 +171,39 @@ xlog_grant_head_wake_all(
 	spin_unlock(&head->lock);
 }
 
-STATIC bool
-xlog_reserveq_wake(
+static inline int
+xlog_ticket_reservation(
 	struct log		*log,
-	int			*free_bytes)
+	struct xlog_grant_head	*head,
+	struct xlog_ticket	*tic)
 {
-	struct xlog_ticket	*tic;
-	int			need_bytes;
-
-	list_for_each_entry(tic, &log->l_reserve_head.waiters, t_queue) {
+	if (head == &log->l_write_head) {
+		ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
+		return tic->t_unit_res;
+	} else {
 		if (tic->t_flags & XLOG_TIC_PERM_RESERV)
-			need_bytes = tic->t_unit_res * tic->t_cnt;
+			return tic->t_unit_res * tic->t_cnt;
 		else
-			need_bytes = tic->t_unit_res;
-
-		if (*free_bytes < need_bytes)
-			return false;
-		*free_bytes -= need_bytes;
-
-		trace_xfs_log_grant_wake_up(log, tic);
-		wake_up_process(tic->t_task);
+			return tic->t_unit_res;
 	}
-
-	return true;
 }
 
 STATIC bool
-xlog_writeq_wake(
+xlog_grant_head_wake(
 	struct log		*log,
+	struct xlog_grant_head	*head,
 	int			*free_bytes)
 {
 	struct xlog_ticket	*tic;
 	int			need_bytes;
 
-	list_for_each_entry(tic, &log->l_write_head.waiters, t_queue) {
-		ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
-
-		need_bytes = tic->t_unit_res;
-
+	list_for_each_entry(tic, &head->waiters, t_queue) {
+		need_bytes = xlog_ticket_reservation(log, head, tic);
 		if (*free_bytes < need_bytes)
 			return false;
-		*free_bytes -= need_bytes;
 
-		trace_xfs_log_regrant_write_wake_up(log, tic);
+		*free_bytes -= need_bytes;
+		trace_xfs_log_grant_wake_up(log, tic);
 		wake_up_process(tic->t_task);
 	}
 
@@ -804,7 +794,7 @@ xfs_log_space_wake(
 
 		spin_lock(&log->l_write_head.lock);
 		free_bytes = xlog_space_left(log, &log->l_write_head.grant);
-		xlog_writeq_wake(log, &free_bytes);
+		xlog_grant_head_wake(log, &log->l_write_head, &free_bytes);
 		spin_unlock(&log->l_write_head.lock);
 	}
 
@@ -813,7 +803,7 @@ xfs_log_space_wake(
 
 		spin_lock(&log->l_reserve_head.lock);
 		free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
-		xlog_reserveq_wake(log, &free_bytes);
+		xlog_grant_head_wake(log, &log->l_reserve_head, &free_bytes);
 		spin_unlock(&log->l_reserve_head.lock);
 	}
 }
@@ -2610,7 +2600,7 @@ xlog_grant_log_space(
 	free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
 	if (!list_empty_careful(&log->l_reserve_head.waiters)) {
 		spin_lock(&log->l_reserve_head.lock);
-		if (!xlog_reserveq_wake(log, &free_bytes) ||
+		if (!xlog_grant_head_wake(log, &log->l_reserve_head, &free_bytes) ||
 		    free_bytes < need_bytes) {
 			error = xlog_grant_head_wait(log, &log->l_reserve_head,
 						     tic, need_bytes);
@@ -2666,7 +2656,7 @@ xlog_regrant_write_log_space(
 	free_bytes = xlog_space_left(log, &log->l_write_head.grant);
 	if (!list_empty_careful(&log->l_write_head.waiters)) {
 		spin_lock(&log->l_write_head.lock);
-		if (!xlog_writeq_wake(log, &free_bytes) ||
+		if (!xlog_grant_head_wake(log, &log->l_write_head, &free_bytes) ||
 		    free_bytes < need_bytes) {
 			error = xlog_grant_head_wait(log, &log->l_write_head,
 						     tic, need_bytes);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 046/102] xfs: share code for grant head availability checks

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 42ceedb3caffe67c4ec0dfbb78ce410832d429b9

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c |  133 ++++++++++++++++++++++++------------------------------
 1 file changed, 60 insertions(+), 73 deletions(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index ad16114..795c4c2 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -245,6 +245,60 @@ shutdown:
 	return XFS_ERROR(EIO);
 }
 
+/*
+ * Atomically get the log space required for a log ticket.
+ *
+ * Once a ticket gets put onto head->waiters, it will only return after the
+ * needed reservation is satisfied.
+ *
+ * This function is structured so that it has a lock free fast path. This is
+ * necessary because every new transaction reservation will come through this
+ * path. Hence any lock will be globally hot if we take it unconditionally on
+ * every pass.
+ *
+ * As tickets are only ever moved on and off head->waiters under head->lock, we
+ * only need to take that lock if we are going to add the ticket to the queue
+ * and sleep. We can avoid taking the lock if the ticket was never added to
+ * head->waiters because the t_queue list head will be empty and we hold the
+ * only reference to it so it can safely be checked unlocked.
+ */
+STATIC int
+xlog_grant_head_check(
+	struct log		*log,
+	struct xlog_grant_head	*head,
+	struct xlog_ticket	*tic,
+	int			*need_bytes)
+{
+	int			free_bytes;
+	int			error = 0;
+
+	ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
+
+	/*
+	 * If there are other waiters on the queue then give them a chance at
+	 * logspace before us.  Wake up the first waiters, if we do not wake
+	 * up all the waiters then go to sleep waiting for more free space,
+	 * otherwise try to get some space for this transaction.
+	 */
+	*need_bytes = xlog_ticket_reservation(log, head, tic);
+	free_bytes = xlog_space_left(log, &head->grant);
+	if (!list_empty_careful(&head->waiters)) {
+		spin_lock(&head->lock);
+		if (!xlog_grant_head_wake(log, head, &free_bytes) ||
+		    free_bytes < *need_bytes) {
+			error = xlog_grant_head_wait(log, head, tic,
+						     *need_bytes);
+		}
+		spin_unlock(&head->lock);
+	} else if (free_bytes < *need_bytes) {
+		spin_lock(&head->lock);
+		error = xlog_grant_head_wait(log, head, tic, *need_bytes);
+		spin_unlock(&head->lock);
+	}
+
+	return error;
+}
+
 static void
 xlog_tic_reset_res(xlog_ticket_t *tic)
 {
@@ -2559,59 +2613,18 @@ restart:
 	return 0;
 }	/* xlog_state_get_iclog_space */
 
-/*
- * Atomically get the log space required for a log ticket.
- *
- * Once a ticket gets put onto the reserveq, it will only return after the
- * needed reservation is satisfied.
- *
- * This function is structured so that it has a lock free fast path. This is
- * necessary because every new transaction reservation will come through this
- * path. Hence any lock will be globally hot if we take it unconditionally on
- * every pass.
- *
- * As tickets are only ever moved on and off the l_reserve.waiters under the
- * l_reserve.lock, we only need to take that lock if we are going to add
- * the ticket to the queue and sleep. We can avoid taking the lock if the ticket
- * was never added to the reserveq because the t_queue list head will be empty
- * and we hold the only reference to it so it can safely be checked unlocked.
- */
 STATIC int
 xlog_grant_log_space(
 	struct log		*log,
 	struct xlog_ticket	*tic)
 {
-	int			free_bytes, need_bytes;
+	int			need_bytes;
 	int			error = 0;
 
-	ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
-
 	trace_xfs_log_grant_enter(log, tic);
 
-	/*
-	 * If there are other waiters on the queue then give them a chance at
-	 * logspace before us.  Wake up the first waiters, if we do not wake
-	 * up all the waiters then go to sleep waiting for more free space,
-	 * otherwise try to get some space for this transaction.
-	 */
-	need_bytes = tic->t_unit_res;
-	if (tic->t_flags & XFS_LOG_PERM_RESERV)
-		need_bytes *= tic->t_ocnt;
-	free_bytes = xlog_space_left(log, &log->l_reserve_head.grant);
-	if (!list_empty_careful(&log->l_reserve_head.waiters)) {
-		spin_lock(&log->l_reserve_head.lock);
-		if (!xlog_grant_head_wake(log, &log->l_reserve_head, &free_bytes) ||
-		    free_bytes < need_bytes) {
-			error = xlog_grant_head_wait(log, &log->l_reserve_head,
-						     tic, need_bytes);
-		}
-		spin_unlock(&log->l_reserve_head.lock);
-	} else if (free_bytes < need_bytes) {
-		spin_lock(&log->l_reserve_head.lock);
-		error = xlog_grant_head_wait(log, &log->l_reserve_head, tic,
-					     need_bytes);
-		spin_unlock(&log->l_reserve_head.lock);
-	}
+	error = xlog_grant_head_check(log, &log->l_reserve_head, tic,
+				      &need_bytes);
 	if (error)
 		return error;
 
@@ -2624,16 +2637,13 @@ xlog_grant_log_space(
 
 /*
  * Replenish the byte reservation required by moving the grant write head.
- *
- * Similar to xlog_grant_log_space, the function is structured to have a lock
- * free fast path.
  */
 STATIC int
 xlog_regrant_write_log_space(
 	struct log		*log,
 	struct xlog_ticket	*tic)
 {
-	int			free_bytes, need_bytes;
+	int			need_bytes;
 	int			error = 0;
 
 	tic->t_curr_res = tic->t_unit_res;
@@ -2642,33 +2652,10 @@ xlog_regrant_write_log_space(
 	if (tic->t_cnt > 0)
 		return 0;
 
-	ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY));
-
 	trace_xfs_log_regrant_write_enter(log, tic);
 
-	/*
-	 * If there are other waiters on the queue then give them a chance at
-	 * logspace before us.  Wake up the first waiters, if we do not wake
-	 * up all the waiters then go to sleep waiting for more free space,
-	 * otherwise try to get some space for this transaction.
-	 */
-	need_bytes = tic->t_unit_res;
-	free_bytes = xlog_space_left(log, &log->l_write_head.grant);
-	if (!list_empty_careful(&log->l_write_head.waiters)) {
-		spin_lock(&log->l_write_head.lock);
-		if (!xlog_grant_head_wake(log, &log->l_write_head, &free_bytes) ||
-		    free_bytes < need_bytes) {
-			error = xlog_grant_head_wait(log, &log->l_write_head,
-						     tic, need_bytes);
-		}
-		spin_unlock(&log->l_write_head.lock);
-	} else if (free_bytes < need_bytes) {
-		spin_lock(&log->l_write_head.lock);
-		error = xlog_grant_head_wait(log, &log->l_write_head, tic,
-					     need_bytes);
-		spin_unlock(&log->l_write_head.lock);
-	}
-
+	error = xlog_grant_head_check(log, &log->l_write_head, tic,
+				      &need_bytes);
 	if (error)
 		return error;
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 047/102] xfs: split and cleanup xfs_log_reserve

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 9006fb91cfdf22812923f0536c7531c429c1aeab

Split the log regrant case out of xfs_log_reserve into a separate function,
and merge xlog_grant_log_space and xlog_regrant_write_log_space into their
respective callers.  Also replace the XFS_LOG_PERM_RESERV flag, which easily
got misused before the previous cleanups with a simple boolean parameter.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   11 +-
 fs/xfs/xfs_log.c             |  265 ++++++++++++++++++++----------------------
 fs/xfs/xfs_log.h             |   12 +-
 fs/xfs/xfs_log_priv.h        |    2 +-
 fs/xfs/xfs_trans.c           |   31 +++--
 5 files changed, 151 insertions(+), 170 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 3617ba1..117d3ae 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -830,17 +830,14 @@ DEFINE_EVENT(xfs_loggrant_class, name, \
 	TP_ARGS(log, tic))
 DEFINE_LOGGRANT_EVENT(xfs_log_done_nonperm);
 DEFINE_LOGGRANT_EVENT(xfs_log_done_perm);
-DEFINE_LOGGRANT_EVENT(xfs_log_reserve);
 DEFINE_LOGGRANT_EVENT(xfs_log_umount_write);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_enter);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_exit);
-DEFINE_LOGGRANT_EVENT(xfs_log_grant_error);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
-DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
+DEFINE_LOGGRANT_EVENT(xfs_log_reserve);
+DEFINE_LOGGRANT_EVENT(xfs_log_reserve_exit);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_sub);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 795c4c2..d92b442 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -67,15 +67,10 @@ STATIC void xlog_state_switch_iclogs(xlog_t		*log,
 				     int		eventual_size);
 STATIC void xlog_state_want_sync(xlog_t	*log, xlog_in_core_t *iclog);
 
-/* local functions to manipulate grant head */
-STATIC int  xlog_grant_log_space(xlog_t		*log,
-				 xlog_ticket_t	*xtic);
 STATIC void xlog_grant_push_ail(struct log	*log,
 				int		need_bytes);
 STATIC void xlog_regrant_reserve_log_space(xlog_t	 *log,
 					   xlog_ticket_t *ticket);
-STATIC int xlog_regrant_write_log_space(xlog_t		*log,
-					 xlog_ticket_t  *ticket);
 STATIC void xlog_ungrant_log_space(xlog_t	 *log,
 				   xlog_ticket_t *ticket);
 
@@ -324,6 +319,128 @@ xlog_tic_add_region(xlog_ticket_t *tic, uint len, uint type)
 }
 
 /*
+ * Replenish the byte reservation required by moving the grant write head.
+ */
+int
+xfs_log_regrant(
+	struct xfs_mount	*mp,
+	struct xlog_ticket	*tic)
+{
+	struct log		*log = mp->m_log;
+	int			need_bytes;
+	int			error = 0;
+
+	if (XLOG_FORCED_SHUTDOWN(log))
+		return XFS_ERROR(EIO);
+
+	XFS_STATS_INC(xs_try_logspace);
+
+	/*
+	 * This is a new transaction on the ticket, so we need to change the
+	 * transaction ID so that the next transaction has a different TID in
+	 * the log. Just add one to the existing tid so that we can see chains
+	 * of rolling transactions in the log easily.
+	 */
+	tic->t_tid++;
+
+	xlog_grant_push_ail(log, tic->t_unit_res);
+
+	tic->t_curr_res = tic->t_unit_res;
+	xlog_tic_reset_res(tic);
+
+	if (tic->t_cnt > 0)
+		return 0;
+
+	trace_xfs_log_regrant(log, tic);
+
+	error = xlog_grant_head_check(log, &log->l_write_head, tic,
+				      &need_bytes);
+	if (error)
+		goto out_error;
+
+	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
+	trace_xfs_log_regrant_exit(log, tic);
+	xlog_verify_grant_tail(log);
+	return 0;
+
+out_error:
+	/*
+	 * If we are failing, make sure the ticket doesn't have any current
+	 * reservations.  We don't want to add this back when the ticket/
+	 * transaction gets cancelled.
+	 */
+	tic->t_curr_res = 0;
+	tic->t_cnt = 0;	/* ungrant will give back unit_res * t_cnt. */
+	return error;
+}
+
+/*
+ * Reserve log space and return a ticket corresponding the reservation.
+ *
+ * Each reservation is going to reserve extra space for a log record header.
+ * When writes happen to the on-disk log, we don't subtract the length of the
+ * log record header from any reservation.  By wasting space in each
+ * reservation, we prevent over allocation problems.
+ */
+int
+xfs_log_reserve(
+	struct xfs_mount	*mp,
+	int		 	unit_bytes,
+	int		 	cnt,
+	struct xlog_ticket	**ticp,
+	__uint8_t	 	client,
+	bool			permanent,
+	uint		 	t_type)
+{
+	struct log		*log = mp->m_log;
+	struct xlog_ticket	*tic;
+	int			need_bytes;
+	int			error = 0;
+
+	ASSERT(client == XFS_TRANSACTION || client == XFS_LOG);
+
+	if (XLOG_FORCED_SHUTDOWN(log))
+		return XFS_ERROR(EIO);
+
+	XFS_STATS_INC(xs_try_logspace);
+
+	ASSERT(*ticp == NULL);
+	tic = xlog_ticket_alloc(log, unit_bytes, cnt, client, permanent,
+				KM_SLEEP | KM_MAYFAIL);
+	if (!tic)
+		return XFS_ERROR(ENOMEM);
+
+	tic->t_trans_type = t_type;
+	*ticp = tic;
+
+	xlog_grant_push_ail(log, tic->t_unit_res * tic->t_cnt);
+
+	trace_xfs_log_reserve(log, tic);
+
+	error = xlog_grant_head_check(log, &log->l_reserve_head, tic,
+				      &need_bytes);
+	if (error)
+		goto out_error;
+
+	xlog_grant_add_space(log, &log->l_reserve_head.grant, need_bytes);
+	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
+	trace_xfs_log_reserve_exit(log, tic);
+	xlog_verify_grant_tail(log);
+	return 0;
+
+out_error:
+	/*
+	 * If we are failing, make sure the ticket doesn't have any current
+	 * reservations.  We don't want to add this back when the ticket/
+	 * transaction gets cancelled.
+	 */
+	tic->t_curr_res = 0;
+	tic->t_cnt = 0;	/* ungrant will give back unit_res * t_cnt. */
+	return error;
+}
+
+
+/*
  * NOTES:
  *
  *	1. currblock field gets updated at startup and after in-core logs
@@ -433,88 +550,6 @@ xfs_log_release_iclog(
 }
 
 /*
- *  1. Reserve an amount of on-disk log space and return a ticket corresponding
- *	to the reservation.
- *  2. Potentially, push buffers at tail of log to disk.
- *
- * Each reservation is going to reserve extra space for a log record header.
- * When writes happen to the on-disk log, we don't subtract the length of the
- * log record header from any reservation.  By wasting space in each
- * reservation, we prevent over allocation problems.
- */
-int
-xfs_log_reserve(
-	struct xfs_mount	*mp,
-	int		 	unit_bytes,
-	int		 	cnt,
-	struct xlog_ticket	**ticket,
-	__uint8_t	 	client,
-	uint		 	flags,
-	uint		 	t_type)
-{
-	struct log		*log = mp->m_log;
-	struct xlog_ticket	*internal_ticket;
-	int			retval = 0;
-
-	ASSERT(client == XFS_TRANSACTION || client == XFS_LOG);
-
-	if (XLOG_FORCED_SHUTDOWN(log))
-		return XFS_ERROR(EIO);
-
-	XFS_STATS_INC(xs_try_logspace);
-
-
-	if (*ticket != NULL) {
-		ASSERT(flags & XFS_LOG_PERM_RESERV);
-		internal_ticket = *ticket;
-
-		/*
-		 * this is a new transaction on the ticket, so we need to
-		 * change the transaction ID so that the next transaction has a
-		 * different TID in the log. Just add one to the existing tid
-		 * so that we can see chains of rolling transactions in the log
-		 * easily.
-		 */
-		internal_ticket->t_tid++;
-
-		trace_xfs_log_reserve(log, internal_ticket);
-
-		xlog_grant_push_ail(log, internal_ticket->t_unit_res);
-		retval = xlog_regrant_write_log_space(log, internal_ticket);
-	} else {
-		/* may sleep if need to allocate more tickets */
-		internal_ticket = xlog_ticket_alloc(log, unit_bytes, cnt,
-						  client, flags,
-						  KM_SLEEP|KM_MAYFAIL);
-		if (!internal_ticket)
-			return XFS_ERROR(ENOMEM);
-		internal_ticket->t_trans_type = t_type;
-		*ticket = internal_ticket;
-
-		trace_xfs_log_reserve(log, internal_ticket);
-
-		xlog_grant_push_ail(log,
-				    (internal_ticket->t_unit_res *
-				     internal_ticket->t_cnt));
-		retval = xlog_grant_log_space(log, internal_ticket);
-	}
-
-	if (unlikely(retval)) {
-		/*
-		 * If we are failing, make sure the ticket doesn't have any
-		 * current reservations.  We don't want to add this back
-		 * when the ticket/ transaction gets cancelled.
-		 */
-		internal_ticket->t_curr_res = 0;
-		/* ungrant will give back unit_res * t_cnt. */
-		internal_ticket->t_cnt = 0;
-	}
-
-	return retval;
-}
-
-
-/*
  * Mount a log filesystem
  *
  * mp		- ubiquitous xfs mount point structure
@@ -2613,58 +2648,6 @@ restart:
 	return 0;
 }	/* xlog_state_get_iclog_space */
 
-STATIC int
-xlog_grant_log_space(
-	struct log		*log,
-	struct xlog_ticket	*tic)
-{
-	int			need_bytes;
-	int			error = 0;
-
-	trace_xfs_log_grant_enter(log, tic);
-
-	error = xlog_grant_head_check(log, &log->l_reserve_head, tic,
-				      &need_bytes);
-	if (error)
-		return error;
-
-	xlog_grant_add_space(log, &log->l_reserve_head.grant, need_bytes);
-	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
-	trace_xfs_log_grant_exit(log, tic);
-	xlog_verify_grant_tail(log);
-	return 0;
-}
-
-/*
- * Replenish the byte reservation required by moving the grant write head.
- */
-STATIC int
-xlog_regrant_write_log_space(
-	struct log		*log,
-	struct xlog_ticket	*tic)
-{
-	int			need_bytes;
-	int			error = 0;
-
-	tic->t_curr_res = tic->t_unit_res;
-	xlog_tic_reset_res(tic);
-
-	if (tic->t_cnt > 0)
-		return 0;
-
-	trace_xfs_log_regrant_write_enter(log, tic);
-
-	error = xlog_grant_head_check(log, &log->l_write_head, tic,
-				      &need_bytes);
-	if (error)
-		return error;
-
-	xlog_grant_add_space(log, &log->l_write_head.grant, need_bytes);
-	trace_xfs_log_regrant_write_exit(log, tic);
-	xlog_verify_grant_tail(log);
-	return 0;
-}
-
 /* The first cnt-1 times through here we don't need to
  * move the grant write head because the permanent
  * reservation has reserved cnt times the unit amount.
@@ -3207,7 +3190,7 @@ xlog_ticket_alloc(
 	int		unit_bytes,
 	int		cnt,
 	char		client,
-	uint		xflags,
+	bool		permanent,
 	int		alloc_flags)
 {
 	struct xlog_ticket *tic;
@@ -3311,7 +3294,7 @@ xlog_ticket_alloc(
 	tic->t_clientid		= client;
 	tic->t_flags		= XLOG_TIC_INITED;
 	tic->t_trans_type	= 0;
-	if (xflags & XFS_LOG_PERM_RESERV)
+	if (permanent)
 		tic->t_flags |= XLOG_TIC_PERM_RESERV;
 
 	xlog_tic_reset_res(tic);
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index 953358b..0694c85 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -53,15 +53,6 @@ static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
 #define XFS_LOG_REL_PERM_RESERV	0x1
 
 /*
- * Flags to xfs_log_reserve()
- *
- *	XFS_LOG_PERM_RESERV: Permanent reservation.  When writes are
- *		performed against this type of reservation, the reservation
- *		is not decreased.  Long running transactions should use this.
- */
-#define XFS_LOG_PERM_RESERV	0x2
-
-/*
  * Flags to xfs_log_force()
  *
  *	XFS_LOG_SYNC:	Synchronous force in-core log to disk
@@ -172,8 +163,9 @@ int	  xfs_log_reserve(struct xfs_mount *mp,
 			  int		   count,
 			  struct xlog_ticket **ticket,
 			  __uint8_t	   clientid,
-			  uint		   flags,
+			  bool		   permanent,
 			  uint		   t_type);
+int	  xfs_log_regrant(struct xfs_mount *mp, struct xlog_ticket *tic);
 int	  xfs_log_write(struct xfs_mount *mp,
 			xfs_log_iovec_t  region[],
 			int		 nentries,
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index eba4ec9..2152900 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -552,7 +552,7 @@ extern void	 xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
 
 extern kmem_zone_t *xfs_log_ticket_zone;
 struct xlog_ticket *xlog_ticket_alloc(struct log *log, int unit_bytes,
-				int count, char client, uint xflags,
+				int count, char client, bool permanent,
 				int alloc_flags);
 
 
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 9b4437d..cc3bc95 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -681,7 +681,6 @@ xfs_trans_reserve(
 	uint		flags,
 	uint		logcount)
 {
-	int		log_flags;
 	int		error = 0;
 	int		rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
 
@@ -707,24 +706,32 @@ xfs_trans_reserve(
 	 * Reserve the log space needed for this transaction.
 	 */
 	if (logspace > 0) {
-		ASSERT((tp->t_log_res == 0) || (tp->t_log_res == logspace));
-		ASSERT((tp->t_log_count == 0) ||
-			(tp->t_log_count == logcount));
+		bool	permanent = false;
+
+		ASSERT(tp->t_log_res == 0 || tp->t_log_res == logspace);
+		ASSERT(tp->t_log_count == 0 || tp->t_log_count == logcount);
+
 		if (flags & XFS_TRANS_PERM_LOG_RES) {
-			log_flags = XFS_LOG_PERM_RESERV;
 			tp->t_flags |= XFS_TRANS_PERM_LOG_RES;
+			permanent = true;
 		} else {
 			ASSERT(tp->t_ticket == NULL);
 			ASSERT(!(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
-			log_flags = 0;
 		}
 
-		error = xfs_log_reserve(tp->t_mountp, logspace, logcount,
-					&tp->t_ticket,
-					XFS_TRANSACTION, log_flags, tp->t_type);
-		if (error) {
-			goto undo_blocks;
+		if (tp->t_ticket != NULL) {
+			ASSERT(flags & XFS_TRANS_PERM_LOG_RES);
+			error = xfs_log_regrant(tp->t_mountp, tp->t_ticket);
+		} else {
+			error = xfs_log_reserve(tp->t_mountp, logspace,
+						logcount, &tp->t_ticket,
+						XFS_TRANSACTION, permanent,
+						tp->t_type);
 		}
+
+		if (error)
+			goto undo_blocks;
+
 		tp->t_log_res = logspace;
 		tp->t_log_count = logcount;
 	}
@@ -752,6 +759,8 @@ xfs_trans_reserve(
 	 */
 undo_log:
 	if (logspace > 0) {
+		int		log_flags;
+
 		if (flags & XFS_TRANS_PERM_LOG_RES) {
 			log_flags = XFS_LOG_REL_PERM_RESERV;
 		} else {
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 048/102] xfs: only take the ILOCK in xfs_reclaim_inode()

[Dave Chinner]

From: Alex Elder <elder@dreamhost.com>

Upstream commit: ad637a10f444fc66b1f6d4a28fe30d4c61ed0161

At the end of xfs_reclaim_inode(), the inode is locked in order to
we wait for a possible concurrent lookup to complete before the
inode is freed.  This synchronization step was taking both the ILOCK
and the IOLOCK, but the latter was causing lockdep to produce
reports of the possibility of deadlock.

It turns out that there's no need to acquire the IOLOCK at this
point anyway.  It may have been required in some earlier version of
the code, but there should be no need to take the IOLOCK in
xfs_iget(), so there's no (longer) any need to get it here for
synchronization.  Add an assertion in xfs_iget() as a reminder
of this assumption.

Dave Chinner diagnosed this on IRC, and Christoph Hellwig suggested
no longer including the IOLOCK.  I just put together the patch.

Signed-off-by: Alex Elder <elder@dreamhost.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_sync.c |   10 ++++------
 fs/xfs/xfs_iget.c           |    9 +++++++++
 2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 2f277a0..b90f3d8 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -921,17 +921,15 @@ reclaim:
 	 * can reference the inodes in the cache without taking references.
 	 *
 	 * We make that OK here by ensuring that we wait until the inode is
-	 * unlocked after the lookup before we go ahead and free it.  We get
-	 * both the ilock and the iolock because the code may need to drop the
-	 * ilock one but will still hold the iolock.
+	 * unlocked after the lookup before we go ahead and free it.
 	 */
-	xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	xfs_qm_dqdetach(ip);
-	xfs_iunlock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 
 	xfs_inode_free(ip);
-	return error;
 
+	return error;
 }
 
 /*
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index b8c110f..550b14e 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -428,6 +428,15 @@ xfs_iget(
 	xfs_perag_t	*pag;
 	xfs_agino_t	agino;
 
+	/*
+	 * xfs_reclaim_inode() uses the ILOCK to ensure an inode
+	 * doesn't get freed while it's being referenced during a
+	 * radix tree traversal here.  It assumes this function
+	 * aqcuires only the ILOCK (and therefore it has no need to
+	 * involve the IOLOCK in this synchronization).
+	 */
+	ASSERT((lock_flags & (XFS_IOLOCK_EXCL | XFS_IOLOCK_SHARED)) == 0);
+
 	/* reject inode numbers outside existing AGs */
 	if (!ino || XFS_INO_TO_AGNO(mp, ino) >= mp->m_sb.sb_agcount)
 		return EINVAL;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 049/102] xfs: use per-filesystem I/O completion workqueues

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: aa6bf01d391935a8929333bc2e243084ea0c58db

The new concurrency managed workqueues are cheap enough that we can create
per-filesystem instead of global workqueues.  This allows us to remove the
trylock or defer scheme on the ilock, which is not helpful once we have
outstanding log reservations until finishing a size update.

Also allow the default concurrency on this workqueues so that I/O completions
blocking on the ilock for one inode do not block process for another inode.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c  |   39 ++++++++-----------------------
 fs/xfs/linux-2.6/xfs_aops.h  |    2 --
 fs/xfs/linux-2.6/xfs_buf.c   |   17 --------------
 fs/xfs/linux-2.6/xfs_super.c |   53 +++++++++++++++++++++++++++++++++++++++++-
 fs/xfs/xfs_mount.h           |    5 ++++
 5 files changed, 67 insertions(+), 49 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 91aa080..6a3cafc 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -160,21 +160,15 @@ static inline bool xfs_ioend_is_append(struct xfs_ioend *ioend)
 
 /*
  * Update on-disk file size now that data has been written to disk.
- *
- * This function does not block as blocking on the inode lock in IO completion
- * can lead to IO completion order dependency deadlocks.. If it can't get the
- * inode ilock it will return EAGAIN. Callers must handle this.
  */
-STATIC int
+STATIC void
 xfs_setfilesize(
-	xfs_ioend_t		*ioend)
+	struct xfs_ioend	*ioend)
 {
-	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
+	struct xfs_inode	*ip = XFS_I(ioend->io_inode);
 	xfs_fsize_t		isize;
 
-	if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL))
-		return EAGAIN;
-
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	isize = xfs_ioend_new_eof(ioend);
 	if (isize) {
 		ip->i_d.di_size = isize;
@@ -182,7 +176,6 @@ xfs_setfilesize(
 	}
 
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-	return 0;
 }
 
 /*
@@ -196,10 +189,12 @@ xfs_finish_ioend(
 	struct xfs_ioend	*ioend)
 {
 	if (atomic_dec_and_test(&ioend->io_remaining)) {
+		struct xfs_mount	*mp = XFS_I(ioend->io_inode)->i_mount;
+
 		if (ioend->io_type == IO_UNWRITTEN)
-			queue_work(xfsconvertd_workqueue, &ioend->io_work);
+			queue_work(mp->m_unwritten_workqueue, &ioend->io_work);
 		else if (xfs_ioend_is_append(ioend))
-			queue_work(xfsdatad_workqueue, &ioend->io_work);
+			queue_work(mp->m_data_workqueue, &ioend->io_work);
 		else
 			xfs_destroy_ioend(ioend);
 	}
@@ -240,23 +235,9 @@ xfs_end_io(
 	 * We might have to update the on-disk file size after extending
 	 * writes.
 	 */
-	error = xfs_setfilesize(ioend);
-	ASSERT(!error || error == EAGAIN);
-
+	xfs_setfilesize(ioend);
 done:
-	/*
-	 * If we didn't complete processing of the ioend, requeue it to the
-	 * tail of the workqueue for another attempt later. Otherwise destroy
-	 * it.
-	 */
-	if (error == EAGAIN) {
-		atomic_inc(&ioend->io_remaining);
-		xfs_finish_ioend(ioend);
-		/* ensure we don't spin on blocked ioends */
-		delay(1);
-	} else {
-		xfs_destroy_ioend(ioend);
-	}
+	xfs_destroy_ioend(ioend);
 }
 
 /*
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
index ce3dcb5..4b48afc 100644
--- a/fs/xfs/linux-2.6/xfs_aops.h
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -18,8 +18,6 @@
 #ifndef __XFS_AOPS_H__
 #define __XFS_AOPS_H__
 
-extern struct workqueue_struct *xfsdatad_workqueue;
-extern struct workqueue_struct *xfsconvertd_workqueue;
 extern mempool_t *xfs_ioend_pool;
 
 /*
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 4508ec2..2df0b4a 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -46,8 +46,6 @@ STATIC int xfsbufd(void *);
 STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
 
 static struct workqueue_struct *xfslogd_workqueue;
-struct workqueue_struct *xfsdatad_workqueue;
-struct workqueue_struct *xfsconvertd_workqueue;
 
 #ifdef XFS_BUF_LOCK_TRACKING
 # define XB_SET_OWNER(bp)	((bp)->b_last_holder = current->pid)
@@ -1856,21 +1854,8 @@ xfs_buf_init(void)
 	if (!xfslogd_workqueue)
 		goto out_free_buf_zone;
 
-	xfsdatad_workqueue = alloc_workqueue("xfsdatad", WQ_MEM_RECLAIM, 1);
-	if (!xfsdatad_workqueue)
-		goto out_destroy_xfslogd_workqueue;
-
-	xfsconvertd_workqueue = alloc_workqueue("xfsconvertd",
-						WQ_MEM_RECLAIM, 1);
-	if (!xfsconvertd_workqueue)
-		goto out_destroy_xfsdatad_workqueue;
-
 	return 0;
 
- out_destroy_xfsdatad_workqueue:
-	destroy_workqueue(xfsdatad_workqueue);
- out_destroy_xfslogd_workqueue:
-	destroy_workqueue(xfslogd_workqueue);
  out_free_buf_zone:
 	kmem_zone_destroy(xfs_buf_zone);
  out:
@@ -1880,8 +1865,6 @@ xfs_buf_init(void)
 void
 xfs_buf_terminate(void)
 {
-	destroy_workqueue(xfsconvertd_workqueue);
-	destroy_workqueue(xfsdatad_workqueue);
 	destroy_workqueue(xfslogd_workqueue);
 	kmem_zone_destroy(xfs_buf_zone);
 }
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 04dd4d4..1a1c26e 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -768,6 +768,50 @@ xfs_setup_devices(
 	return 0;
 }
 
+STATIC int
+xfs_init_mount_workqueues(
+	struct xfs_mount	*mp)
+{
+	mp->m_data_wq_name = kmalloc(strlen(mp->m_fsname) +
+					strlen("xfs-data/") + 1, GFP_KERNEL);
+	if (!mp->m_data_wq_name)
+		goto out;
+	sprintf(mp->m_data_wq_name, "xfs-data/%s", mp->m_fsname);
+	mp->m_data_workqueue = alloc_workqueue(mp->m_data_wq_name,
+						WQ_MEM_RECLAIM, 0);
+	if (!mp->m_data_workqueue)
+		goto out;
+
+	mp->m_unwritten_wq_name = kmalloc(strlen(mp->m_fsname) +
+					strlen("xfs-conv/") + 1, GFP_KERNEL);
+	if (!mp->m_unwritten_wq_name)
+		goto out_destroy_data_iodone_queue;
+	sprintf(mp->m_unwritten_wq_name, "xfs-conv/%s", mp->m_fsname);
+	mp->m_unwritten_workqueue = alloc_workqueue(mp->m_unwritten_wq_name,
+						WQ_MEM_RECLAIM, 0);
+	if (!mp->m_unwritten_workqueue)
+		goto out_destroy_data_iodone_queue;
+
+	return 0;
+
+out_destroy_data_iodone_queue:
+	destroy_workqueue(mp->m_data_workqueue);
+out:
+	kfree(mp->m_data_wq_name);
+	kfree(mp->m_unwritten_wq_name);
+	return -ENOMEM;
+}
+
+STATIC void
+xfs_destroy_mount_workqueues(
+	struct xfs_mount	*mp)
+{
+	destroy_workqueue(mp->m_data_workqueue);
+	destroy_workqueue(mp->m_unwritten_workqueue);
+	kfree(mp->m_data_wq_name);
+	kfree(mp->m_unwritten_wq_name);
+}
+
 /* Catch misguided souls that try to use this interface on XFS */
 STATIC struct inode *
 xfs_fs_alloc_inode(
@@ -1008,6 +1052,7 @@ xfs_fs_put_super(
 	xfs_unmountfs(mp);
 	xfs_freesb(mp);
 	xfs_icsb_destroy_counters(mp);
+	xfs_destroy_mount_workqueues(mp);
 	xfs_close_devices(mp);
 	xfs_free_fsname(mp);
 	kfree(mp);
@@ -1341,10 +1386,14 @@ xfs_fs_fill_super(
 	if (error)
 		goto out_free_fsname;
 
-	error = xfs_icsb_init_counters(mp);
+	error = xfs_init_mount_workqueues(mp);
 	if (error)
 		goto out_close_devices;
 
+	error = xfs_icsb_init_counters(mp);
+	if (error)
+		goto out_destroy_workqueues;
+
 	error = xfs_readsb(mp, flags);
 	if (error)
 		goto out_destroy_counters;
@@ -1410,6 +1459,8 @@ xfs_fs_fill_super(
 	xfs_freesb(mp);
  out_destroy_counters:
 	xfs_icsb_destroy_counters(mp);
+out_destroy_workqueues:
+	xfs_destroy_mount_workqueues(mp);
  out_close_devices:
 	xfs_close_devices(mp);
  out_free_fsname:
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index 3d68bb2..fdcb826 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -211,6 +211,11 @@ typedef struct xfs_mount {
 	struct shrinker		m_inode_shrink;	/* inode reclaim shrinker */
 	int64_t			m_low_space[XFS_LOWSP_MAX];
 						/* low free space thresholds */
+
+	struct workqueue_struct	*m_data_workqueue;
+	struct workqueue_struct	*m_unwritten_workqueue;
+	char			*m_data_wq_name;
+	char			*m_unwritten_wq_name;
 } xfs_mount_t;
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 050/102] xfs: do not require an ioend for new EOF calculation

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 6923e686f19cb7017fc9777a10e06c2e2b2a2936

Replace xfs_ioend_new_eof with a new inline xfs_new_eof helper that
doesn't require and ioend, and is available also outside of xfs_aops.c.

Also make the code a bit more clear by using a normal if statement
instead of a slightly misleading MIN().

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   24 ++++--------------------
 fs/xfs/xfs_inode.h          |   14 ++++++++++++++
 2 files changed, 18 insertions(+), 20 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 6a3cafc..5297866 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -133,23 +133,6 @@ xfs_destroy_ioend(
 }
 
 /*
- * If the end of the current ioend is beyond the current EOF,
- * return the new EOF value, otherwise zero.
- */
-STATIC xfs_fsize_t
-xfs_ioend_new_eof(
-	xfs_ioend_t		*ioend)
-{
-	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
-	xfs_fsize_t		isize;
-	xfs_fsize_t		bsize;
-
-	bsize = ioend->io_offset + ioend->io_size;
-	isize = MIN(i_size_read(VFS_I(ip)), bsize);
-	return isize > ip->i_d.di_size ? isize : 0;
-}
-
-/*
  * Fast and loose check if this write could update the on-disk inode size.
  */
 static inline bool xfs_ioend_is_append(struct xfs_ioend *ioend)
@@ -169,7 +152,7 @@ xfs_setfilesize(
 	xfs_fsize_t		isize;
 
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	isize = xfs_ioend_new_eof(ioend);
+	isize = xfs_new_eof(ip, ioend->io_offset + ioend->io_size);
 	if (isize) {
 		ip->i_d.di_size = isize;
 		xfs_mark_inode_dirty(ip);
@@ -391,6 +374,7 @@ xfs_submit_ioend_bio(
 	xfs_ioend_t		*ioend,
 	struct bio		*bio)
 {
+	struct xfs_inode	*ip = XFS_I(ioend->io_inode);
 	atomic_inc(&ioend->io_remaining);
 	bio->bi_private = ioend;
 	bio->bi_end_io = xfs_end_bio;
@@ -399,8 +383,8 @@ xfs_submit_ioend_bio(
 	 * If the I/O is beyond EOF we mark the inode dirty immediately
 	 * but don't update the inode size until I/O completion.
 	 */
-	if (xfs_ioend_new_eof(ioend))
-		xfs_mark_inode_dirty(XFS_I(ioend->io_inode));
+	if (xfs_new_eof(ip, ioend->io_offset + ioend->io_size))
+		xfs_mark_inode_dirty(ip);
 
 	submit_bio(wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE, bio);
 }
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index d3bbc93..ec81926 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -287,6 +287,20 @@ static inline xfs_fsize_t XFS_ISIZE(struct xfs_inode *ip)
 }
 
 /*
+ * If this I/O goes past the on-disk inode size update it unless it would
+ * be past the current in-core inode size.
+ */
+static inline xfs_fsize_t
+xfs_new_eof(struct xfs_inode *ip, xfs_fsize_t new_size)
+{
+	xfs_fsize_t i_size = i_size_read(VFS_I(ip));
+
+	if (new_size > i_size)
+		new_size = i_size;
+	return new_size > ip->i_d.di_size ? new_size : 0;
+}
+
+/*
  * i_flags helper functions
  */
 static inline void
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 054/102] xfs: make xfs_inode_item_size idempotent

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 339a5f5dd9d3ac3d68a594d81507e1eab77ed223

Move all code messing with the inode log item flags into xfs_inode_item_format
to make sure xfs_inode_item_size really only calculates the the number of
vectors, but doesn't modify any state of the inode item.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_inode_item.c |  218 ++++++++++++++++++-----------------------------
 1 file changed, 83 insertions(+), 135 deletions(-)

diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 6170176..d5ea2d8 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -57,79 +57,28 @@ xfs_inode_item_size(
 	struct xfs_inode	*ip = iip->ili_inode;
 	uint			nvecs = 2;
 
-	/*
-	 * Only log the data/extents/b-tree root if there is something
-	 * left to log.
-	 */
-	iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
-
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
-		    (ip->i_d.di_nextents > 0) &&
-		    (ip->i_df.if_bytes > 0)) {
-			ASSERT(ip->i_df.if_u1.if_extents != NULL);
+		    ip->i_d.di_nextents > 0 &&
+		    ip->i_df.if_bytes > 0)
 			nvecs++;
-		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DEXT;
-		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		ASSERT(ip->i_df.if_ext_max ==
-		       XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t));
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
-		    (ip->i_df.if_broot_bytes > 0)) {
-			ASSERT(ip->i_df.if_broot != NULL);
+		    ip->i_df.if_broot_bytes > 0)
 			nvecs++;
-		} else {
-			ASSERT(!(iip->ili_format.ilf_fields &
-				 XFS_ILOG_DBROOT));
-#ifdef XFS_TRANS_DEBUG
-			if (iip->ili_root_size > 0) {
-				ASSERT(iip->ili_root_size ==
-				       ip->i_df.if_broot_bytes);
-				ASSERT(memcmp(iip->ili_orig_root,
-					    ip->i_df.if_broot,
-					    iip->ili_root_size) == 0);
-			} else {
-				ASSERT(ip->i_df.if_broot_bytes == 0);
-			}
-#endif
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DBROOT;
-		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
-		    (ip->i_df.if_bytes > 0)) {
-			ASSERT(ip->i_df.if_u1.if_data != NULL);
-			ASSERT(ip->i_d.di_size > 0);
+		    ip->i_df.if_bytes > 0)
 			nvecs++;
-		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DDATA;
-		}
 		break;
 
 	case XFS_DINODE_FMT_DEV:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
-			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
-		break;
-
 	case XFS_DINODE_FMT_UUID:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
-			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
 		break;
 
 	default:
@@ -137,56 +86,31 @@ xfs_inode_item_size(
 		break;
 	}
 
-	/*
-	 * If there are no attributes associated with this file,
-	 * then there cannot be anything more to log.
-	 * Clear all attribute-related log flags.
-	 */
-	if (!XFS_IFORK_Q(ip)) {
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
+	if (!XFS_IFORK_Q(ip))
 		return nvecs;
-	}
+
 
 	/*
 	 * Log any necessary attribute data.
 	 */
 	switch (ip->i_d.di_aformat) {
 	case XFS_DINODE_FMT_EXTENTS:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
-		    (ip->i_d.di_anextents > 0) &&
-		    (ip->i_afp->if_bytes > 0)) {
-			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
+		    ip->i_d.di_anextents > 0 &&
+		    ip->i_afp->if_bytes > 0)
 			nvecs++;
-		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_AEXT;
-		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
-		    (ip->i_afp->if_broot_bytes > 0)) {
-			ASSERT(ip->i_afp->if_broot != NULL);
+		    ip->i_afp->if_broot_bytes > 0)
 			nvecs++;
-		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_ABROOT;
-		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		iip->ili_format.ilf_fields &=
-			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);
 		if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
-		    (ip->i_afp->if_bytes > 0)) {
-			ASSERT(ip->i_afp->if_u1.if_data != NULL);
+		    ip->i_afp->if_bytes > 0)
 			nvecs++;
-		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_ADATA;
-		}
 		break;
 
 	default:
@@ -294,16 +218,17 @@ xfs_inode_item_format(
 
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_DEXT) {
-			ASSERT(ip->i_df.if_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+		    ip->i_d.di_nextents > 0 &&
+		    ip->i_df.if_bytes > 0) {
 			ASSERT(ip->i_df.if_u1.if_extents != NULL);
-			ASSERT(ip->i_d.di_nextents > 0);
+			ASSERT(ip->i_df.if_bytes / sizeof(xfs_bmbt_rec_t) > 0);
 			ASSERT(iip->ili_extents_buf == NULL);
-			ASSERT((ip->i_df.if_bytes /
-				(uint)sizeof(xfs_bmbt_rec_t)) > 0);
+
 #ifdef XFS_NATIVE_HOST
                        if (ip->i_d.di_nextents == ip->i_df.if_bytes /
                                                (uint)sizeof(xfs_bmbt_rec_t)) {
@@ -325,15 +250,18 @@ xfs_inode_item_format(
 			iip->ili_format.ilf_dsize = vecp->i_len;
 			vecp++;
 			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DEXT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_DDATA | XFS_ILOG_DEXT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) {
-			ASSERT(ip->i_df.if_broot_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+		    ip->i_df.if_broot_bytes > 0) {
 			ASSERT(ip->i_df.if_broot != NULL);
 			vecp->i_addr = ip->i_df.if_broot;
 			vecp->i_len = ip->i_df.if_broot_bytes;
@@ -341,15 +269,30 @@ xfs_inode_item_format(
 			vecp++;
 			nvecs++;
 			iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
+		} else {
+			ASSERT(!(iip->ili_format.ilf_fields &
+				 XFS_ILOG_DBROOT));
+#ifdef XFS_TRANS_DEBUG
+			if (iip->ili_root_size > 0) {
+				ASSERT(iip->ili_root_size ==
+				       ip->i_df.if_broot_bytes);
+				ASSERT(memcmp(iip->ili_orig_root,
+					    ip->i_df.if_broot,
+					    iip->ili_root_size) == 0);
+			} else {
+				ASSERT(ip->i_df.if_broot_bytes == 0);
+			}
+#endif
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DBROOT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
-			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_DDATA) {
-			ASSERT(ip->i_df.if_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+		    ip->i_df.if_bytes > 0) {
 			ASSERT(ip->i_df.if_u1.if_data != NULL);
 			ASSERT(ip->i_d.di_size > 0);
 
@@ -367,13 +310,15 @@ xfs_inode_item_format(
 			vecp++;
 			nvecs++;
 			iip->ili_format.ilf_dsize = (unsigned)data_bytes;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DDATA;
 		}
 		break;
 
 	case XFS_DINODE_FMT_DEV:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
-			  XFS_ILOG_DDATA | XFS_ILOG_UUID)));
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
 		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
 			iip->ili_format.ilf_u.ilfu_rdev =
 				ip->i_df.if_u2.if_rdev;
@@ -381,9 +326,9 @@ xfs_inode_item_format(
 		break;
 
 	case XFS_DINODE_FMT_UUID:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
-			  XFS_ILOG_DDATA | XFS_ILOG_DEV)));
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
 		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
 			iip->ili_format.ilf_u.ilfu_uuid =
 				ip->i_df.if_u2.if_uuid;
@@ -396,32 +341,26 @@ xfs_inode_item_format(
 	}
 
 	/*
-	 * If there are no attributes associated with the file,
-	 * then we're done.
-	 * Assert that no attribute-related log flags are set.
+	 * If there are no attributes associated with the file, then we're done.
 	 */
 	if (!XFS_IFORK_Q(ip)) {
-		ASSERT(nvecs == lip->li_desc->lid_size);
 		iip->ili_format.ilf_size = nvecs;
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
 		return;
 	}
 
 	switch (ip->i_d.di_aformat) {
 	case XFS_DINODE_FMT_EXTENTS:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_ADATA | XFS_ILOG_ABROOT)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_AEXT) {
-#ifdef DEBUG
-			int nrecs = ip->i_afp->if_bytes /
-				(uint)sizeof(xfs_bmbt_rec_t);
-			ASSERT(nrecs > 0);
-			ASSERT(nrecs == ip->i_d.di_anextents);
-			ASSERT(ip->i_afp->if_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);
+
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+		    ip->i_d.di_anextents > 0 &&
+		    ip->i_afp->if_bytes > 0) {
+			ASSERT(ip->i_afp->if_bytes / sizeof(xfs_bmbt_rec_t) ==
+				ip->i_d.di_anextents);
 			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
-			ASSERT(ip->i_d.di_anextents > 0);
-#endif
 #ifdef XFS_NATIVE_HOST
 			/*
 			 * There are not delayed allocation extents
@@ -438,29 +377,36 @@ xfs_inode_item_format(
 			iip->ili_format.ilf_asize = vecp->i_len;
 			vecp++;
 			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_AEXT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_ADATA | XFS_ILOG_AEXT)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) {
-			ASSERT(ip->i_afp->if_broot_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);
+
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+		    ip->i_afp->if_broot_bytes > 0) {
 			ASSERT(ip->i_afp->if_broot != NULL);
+
 			vecp->i_addr = ip->i_afp->if_broot;
 			vecp->i_len = ip->i_afp->if_broot_bytes;
 			vecp->i_type = XLOG_REG_TYPE_IATTR_BROOT;
 			vecp++;
 			nvecs++;
 			iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_ABROOT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		ASSERT(!(iip->ili_format.ilf_fields &
-			 (XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
-		if (iip->ili_format.ilf_fields & XFS_ILOG_ADATA) {
-			ASSERT(ip->i_afp->if_bytes > 0);
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);
+
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+		    ip->i_afp->if_bytes > 0) {
 			ASSERT(ip->i_afp->if_u1.if_data != NULL);
 
 			vecp->i_addr = ip->i_afp->if_u1.if_data;
@@ -477,6 +423,8 @@ xfs_inode_item_format(
 			vecp++;
 			nvecs++;
 			iip->ili_format.ilf_asize = (unsigned)data_bytes;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_ADATA;
 		}
 		break;
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 055/102] xfs: split in-core and on-disk inode log item fields

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: f5d8d5c4bf29c9f7754d9cbe5e27c785106ba872

Add a new ili_fields member to the inode log item to isolate the in-memory
flags from the ones that actually go to the log.  This will allow tracking
timestamp-only updates for fdatasync and O_DSYNC in the next patch and
prepares for divorcing the on-disk log format from the in-memory log item
a little further down the road.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_dfrag.c       |   24 ++++++--------
 fs/xfs/xfs_inode.c       |   69 ++++++++++++++++++--------------------
 fs/xfs/xfs_inode_item.c  |   83 ++++++++++++++++++++++++----------------------
 fs/xfs/xfs_inode_item.h  |    4 +--
 fs/xfs/xfs_trans_inode.c |    4 +--
 5 files changed, 92 insertions(+), 92 deletions(-)

diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 645387b..b2ecb56 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -206,7 +206,7 @@ xfs_swap_extents(
 	xfs_trans_t	*tp;
 	xfs_bstat_t	*sbp = &sxp->sx_stat;
 	xfs_ifork_t	*tempifp, *ifp, *tifp;
-	int		ilf_fields, tilf_fields;
+	int		src_log_flags, target_log_flags;
 	int		error = 0;
 	int		aforkblks = 0;
 	int		taforkblks = 0;
@@ -386,9 +386,8 @@ xfs_swap_extents(
 	tip->i_delayed_blks = ip->i_delayed_blks;
 	ip->i_delayed_blks = 0;
 
-	ilf_fields = XFS_ILOG_CORE;
-
-	switch(ip->i_d.di_format) {
+	src_log_flags = XFS_ILOG_CORE;
+	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
 		 * pointer.  Otherwise it's already NULL or
@@ -398,16 +397,15 @@ xfs_swap_extents(
 			ifp->if_u1.if_extents =
 				ifp->if_u2.if_inline_ext;
 		}
-		ilf_fields |= XFS_ILOG_DEXT;
+		src_log_flags |= XFS_ILOG_DEXT;
 		break;
 	case XFS_DINODE_FMT_BTREE:
-		ilf_fields |= XFS_ILOG_DBROOT;
+		src_log_flags |= XFS_ILOG_DBROOT;
 		break;
 	}
 
-	tilf_fields = XFS_ILOG_CORE;
-
-	switch(tip->i_d.di_format) {
+	target_log_flags = XFS_ILOG_CORE;
+	switch (tip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
 		 * pointer.  Otherwise it's already NULL or
@@ -417,10 +415,10 @@ xfs_swap_extents(
 			tifp->if_u1.if_extents =
 				tifp->if_u2.if_inline_ext;
 		}
-		tilf_fields |= XFS_ILOG_DEXT;
+		target_log_flags |= XFS_ILOG_DEXT;
 		break;
 	case XFS_DINODE_FMT_BTREE:
-		tilf_fields |= XFS_ILOG_DBROOT;
+		target_log_flags |= XFS_ILOG_DBROOT;
 		break;
 	}
 
@@ -428,8 +426,8 @@ xfs_swap_extents(
 	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 	xfs_trans_ijoin_ref(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 
-	xfs_trans_log_inode(tp, ip,  ilf_fields);
-	xfs_trans_log_inode(tp, tip, tilf_fields);
+	xfs_trans_log_inode(tp, ip,  src_log_flags);
+	xfs_trans_log_inode(tp, tip, target_log_flags);
 
 	/*
 	 * If this is a synchronous mount, make sure that the
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 3942560..4b278d6 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -2031,8 +2031,8 @@ retry:
 				continue;
 			}
 
-			iip->ili_last_fields = iip->ili_format.ilf_fields;
-			iip->ili_format.ilf_fields = 0;
+			iip->ili_last_fields = iip->ili_fields;
+			iip->ili_fields = 0;
 			iip->ili_logged = 1;
 			xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
 						&iip->ili_item.li_lsn);
@@ -2534,7 +2534,7 @@ xfs_iflush_fork(
 	mp = ip->i_mount;
 	switch (XFS_IFORK_FORMAT(ip, whichfork)) {
 	case XFS_DINODE_FMT_LOCAL:
-		if ((iip->ili_format.ilf_fields & dataflag[whichfork]) &&
+		if ((iip->ili_fields & dataflag[whichfork]) &&
 		    (ifp->if_bytes > 0)) {
 			ASSERT(ifp->if_u1.if_data != NULL);
 			ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
@@ -2544,8 +2544,8 @@ xfs_iflush_fork(
 
 	case XFS_DINODE_FMT_EXTENTS:
 		ASSERT((ifp->if_flags & XFS_IFEXTENTS) ||
-		       !(iip->ili_format.ilf_fields & extflag[whichfork]));
-		if ((iip->ili_format.ilf_fields & extflag[whichfork]) &&
+		       !(iip->ili_fields & extflag[whichfork]));
+		if ((iip->ili_fields & extflag[whichfork]) &&
 		    (ifp->if_bytes > 0)) {
 			ASSERT(xfs_iext_get_ext(ifp, 0));
 			ASSERT(XFS_IFORK_NEXTENTS(ip, whichfork) > 0);
@@ -2555,7 +2555,7 @@ xfs_iflush_fork(
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		if ((iip->ili_format.ilf_fields & brootflag[whichfork]) &&
+		if ((iip->ili_fields & brootflag[whichfork]) &&
 		    (ifp->if_broot_bytes > 0)) {
 			ASSERT(ifp->if_broot != NULL);
 			ASSERT(ifp->if_broot_bytes <=
@@ -2568,14 +2568,14 @@ xfs_iflush_fork(
 		break;
 
 	case XFS_DINODE_FMT_DEV:
-		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+		if (iip->ili_fields & XFS_ILOG_DEV) {
 			ASSERT(whichfork == XFS_DATA_FORK);
 			xfs_dinode_put_rdev(dip, ip->i_df.if_u2.if_rdev);
 		}
 		break;
 
 	case XFS_DINODE_FMT_UUID:
-		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+		if (iip->ili_fields & XFS_ILOG_UUID) {
 			ASSERT(whichfork == XFS_DATA_FORK);
 			memcpy(XFS_DFORK_DPTR(dip),
 			       &ip->i_df.if_u2.if_uuid,
@@ -2809,7 +2809,7 @@ xfs_iflush(
 	 */
 	if (XFS_FORCED_SHUTDOWN(mp)) {
 		if (iip)
-			iip->ili_format.ilf_fields = 0;
+			iip->ili_fields = 0;
 		xfs_ifunlock(ip);
 		return XFS_ERROR(EIO);
 	}
@@ -2997,36 +2997,33 @@ xfs_iflush_int(
 	xfs_inobp_check(mp, bp);
 
 	/*
-	 * We've recorded everything logged in the inode, so we'd
-	 * like to clear the ilf_fields bits so we don't log and
-	 * flush things unnecessarily.  However, we can't stop
-	 * logging all this information until the data we've copied
-	 * into the disk buffer is written to disk.  If we did we might
-	 * overwrite the copy of the inode in the log with all the
-	 * data after re-logging only part of it, and in the face of
-	 * a crash we wouldn't have all the data we need to recover.
+	 * We've recorded everything logged in the inode, so we'd like to clear
+	 * the ili_fields bits so we don't log and flush things unnecessarily.
+	 * However, we can't stop logging all this information until the data
+	 * we've copied into the disk buffer is written to disk.  If we did we
+	 * might overwrite the copy of the inode in the log with all the data
+	 * after re-logging only part of it, and in the face of a crash we
+	 * wouldn't have all the data we need to recover.
 	 *
-	 * What we do is move the bits to the ili_last_fields field.
-	 * When logging the inode, these bits are moved back to the
-	 * ilf_fields field.  In the xfs_iflush_done() routine we
-	 * clear ili_last_fields, since we know that the information
-	 * those bits represent is permanently on disk.  As long as
-	 * the flush completes before the inode is logged again, then
-	 * both ilf_fields and ili_last_fields will be cleared.
+	 * What we do is move the bits to the ili_last_fields field.  When
+	 * logging the inode, these bits are moved back to the ili_fields field.
+	 * In the xfs_iflush_done() routine we clear ili_last_fields, since we
+	 * know that the information those bits represent is permanently on
+	 * disk.  As long as the flush completes before the inode is logged
+	 * again, then both ili_fields and ili_last_fields will be cleared.
 	 *
-	 * We can play with the ilf_fields bits here, because the inode
-	 * lock must be held exclusively in order to set bits there
-	 * and the flush lock protects the ili_last_fields bits.
-	 * Set ili_logged so the flush done
-	 * routine can tell whether or not to look in the AIL.
-	 * Also, store the current LSN of the inode so that we can tell
-	 * whether the item has moved in the AIL from xfs_iflush_done().
-	 * In order to read the lsn we need the AIL lock, because
-	 * it is a 64 bit value that cannot be read atomically.
+	 * We can play with the ili_fields bits here, because the inode lock
+	 * must be held exclusively in order to set bits there and the flush
+	 * lock protects the ili_last_fields bits.  Set ili_logged so the flush
+	 * done routine can tell whether or not to look in the AIL.  Also, store
+	 * the current LSN of the inode so that we can tell whether the item has
+	 * moved in the AIL from xfs_iflush_done().  In order to read the lsn we
+	 * need the AIL lock, because it is a 64 bit value that cannot be read
+	 * atomically.
 	 */
-	if (iip != NULL && iip->ili_format.ilf_fields != 0) {
-		iip->ili_last_fields = iip->ili_format.ilf_fields;
-		iip->ili_format.ilf_fields = 0;
+	if (iip != NULL && iip->ili_fields != 0) {
+		iip->ili_last_fields = iip->ili_fields;
+		iip->ili_fields = 0;
 		iip->ili_logged = 1;
 
 		xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index d5ea2d8..45514c5 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -59,20 +59,20 @@ xfs_inode_item_size(
 
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
 		    ip->i_d.di_nextents > 0 &&
 		    ip->i_df.if_bytes > 0)
 			nvecs++;
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
 		    ip->i_df.if_broot_bytes > 0)
 			nvecs++;
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
 		    ip->i_df.if_bytes > 0)
 			nvecs++;
 		break;
@@ -95,20 +95,20 @@ xfs_inode_item_size(
 	 */
 	switch (ip->i_d.di_aformat) {
 	case XFS_DINODE_FMT_EXTENTS:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
 		    ip->i_d.di_anextents > 0 &&
 		    ip->i_afp->if_bytes > 0)
 			nvecs++;
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
 		    ip->i_afp->if_broot_bytes > 0)
 			nvecs++;
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
 		    ip->i_afp->if_bytes > 0)
 			nvecs++;
 		break;
@@ -185,7 +185,6 @@ xfs_inode_item_format(
 	vecp->i_type = XLOG_REG_TYPE_ICORE;
 	vecp++;
 	nvecs++;
-	iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
 
 	/*
 	 * If this is really an old format inode, then we need to
@@ -218,11 +217,11 @@ xfs_inode_item_format(
 
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
 			  XFS_ILOG_DEV | XFS_ILOG_UUID);
 
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
 		    ip->i_d.di_nextents > 0 &&
 		    ip->i_df.if_bytes > 0) {
 			ASSERT(ip->i_df.if_u1.if_extents != NULL);
@@ -251,16 +250,16 @@ xfs_inode_item_format(
 			vecp++;
 			nvecs++;
 		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DEXT;
+			iip->ili_fields &= ~XFS_ILOG_DEXT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
 			  XFS_ILOG_DEV | XFS_ILOG_UUID);
 
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
 		    ip->i_df.if_broot_bytes > 0) {
 			ASSERT(ip->i_df.if_broot != NULL);
 			vecp->i_addr = ip->i_df.if_broot;
@@ -270,7 +269,7 @@ xfs_inode_item_format(
 			nvecs++;
 			iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
 		} else {
-			ASSERT(!(iip->ili_format.ilf_fields &
+			ASSERT(!(iip->ili_fields &
 				 XFS_ILOG_DBROOT));
 #ifdef XFS_TRANS_DEBUG
 			if (iip->ili_root_size > 0) {
@@ -283,15 +282,15 @@ xfs_inode_item_format(
 				ASSERT(ip->i_df.if_broot_bytes == 0);
 			}
 #endif
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DBROOT;
+			iip->ili_fields &= ~XFS_ILOG_DBROOT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
 			  XFS_ILOG_DEV | XFS_ILOG_UUID);
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
 		    ip->i_df.if_bytes > 0) {
 			ASSERT(ip->i_df.if_u1.if_data != NULL);
 			ASSERT(ip->i_d.di_size > 0);
@@ -311,25 +310,25 @@ xfs_inode_item_format(
 			nvecs++;
 			iip->ili_format.ilf_dsize = (unsigned)data_bytes;
 		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_DDATA;
+			iip->ili_fields &= ~XFS_ILOG_DDATA;
 		}
 		break;
 
 	case XFS_DINODE_FMT_DEV:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
 			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
-		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+		if (iip->ili_fields & XFS_ILOG_DEV) {
 			iip->ili_format.ilf_u.ilfu_rdev =
 				ip->i_df.if_u2.if_rdev;
 		}
 		break;
 
 	case XFS_DINODE_FMT_UUID:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
 			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
-		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+		if (iip->ili_fields & XFS_ILOG_UUID) {
 			iip->ili_format.ilf_u.ilfu_uuid =
 				ip->i_df.if_u2.if_uuid;
 		}
@@ -344,18 +343,17 @@ xfs_inode_item_format(
 	 * If there are no attributes associated with the file, then we're done.
 	 */
 	if (!XFS_IFORK_Q(ip)) {
-		iip->ili_format.ilf_size = nvecs;
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
-		return;
+		goto out;
 	}
 
 	switch (ip->i_d.di_aformat) {
 	case XFS_DINODE_FMT_EXTENTS:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);
 
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
 		    ip->i_d.di_anextents > 0 &&
 		    ip->i_afp->if_bytes > 0) {
 			ASSERT(ip->i_afp->if_bytes / sizeof(xfs_bmbt_rec_t) ==
@@ -378,15 +376,15 @@ xfs_inode_item_format(
 			vecp++;
 			nvecs++;
 		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_AEXT;
+			iip->ili_fields &= ~XFS_ILOG_AEXT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_BTREE:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);
 
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
 		    ip->i_afp->if_broot_bytes > 0) {
 			ASSERT(ip->i_afp->if_broot != NULL);
 
@@ -397,15 +395,15 @@ xfs_inode_item_format(
 			nvecs++;
 			iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
 		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_ABROOT;
+			iip->ili_fields &= ~XFS_ILOG_ABROOT;
 		}
 		break;
 
 	case XFS_DINODE_FMT_LOCAL:
-		iip->ili_format.ilf_fields &=
+		iip->ili_fields &=
 			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);
 
-		if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
 		    ip->i_afp->if_bytes > 0) {
 			ASSERT(ip->i_afp->if_u1.if_data != NULL);
 
@@ -424,7 +422,7 @@ xfs_inode_item_format(
 			nvecs++;
 			iip->ili_format.ilf_asize = (unsigned)data_bytes;
 		} else {
-			iip->ili_format.ilf_fields &= ~XFS_ILOG_ADATA;
+			iip->ili_fields &= ~XFS_ILOG_ADATA;
 		}
 		break;
 
@@ -433,6 +431,14 @@ xfs_inode_item_format(
 		break;
 	}
 
+out:
+	/*
+	 * Now update the log format that goes out to disk from the in-core
+	 * values.  We always write the inode core to make the arithmetic
+	 * games in recovery easier, which isn't a big deal as just about any
+	 * transaction would dirty it anyway.
+	 */
+	iip->ili_format.ilf_fields = XFS_ILOG_CORE | iip->ili_fields;
 	ASSERT(nvecs == lip->li_desc->lid_size);
 	iip->ili_format.ilf_size = nvecs;
 }
@@ -518,7 +524,7 @@ xfs_inode_item_trylock(
 
 #ifdef DEBUG
 	if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-		ASSERT(iip->ili_format.ilf_fields != 0);
+		ASSERT(iip->ili_fields != 0);
 		ASSERT(iip->ili_logged == 0);
 		ASSERT(lip->li_flags & XFS_LI_IN_AIL);
 	}
@@ -555,7 +561,7 @@ xfs_inode_item_unlock(
 	if (iip->ili_extents_buf != NULL) {
 		ASSERT(ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS);
 		ASSERT(ip->i_d.di_nextents > 0);
-		ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_DEXT);
+		ASSERT(iip->ili_fields & XFS_ILOG_DEXT);
 		ASSERT(ip->i_df.if_bytes > 0);
 		kmem_free(iip->ili_extents_buf);
 		iip->ili_extents_buf = NULL;
@@ -563,7 +569,7 @@ xfs_inode_item_unlock(
 	if (iip->ili_aextents_buf != NULL) {
 		ASSERT(ip->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS);
 		ASSERT(ip->i_d.di_anextents > 0);
-		ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_AEXT);
+		ASSERT(iip->ili_fields & XFS_ILOG_AEXT);
 		ASSERT(ip->i_afp->if_bytes > 0);
 		kmem_free(iip->ili_aextents_buf);
 		iip->ili_aextents_buf = NULL;
@@ -680,8 +686,7 @@ xfs_inode_item_push(
 	 * lock without sleeping, then there must not have been
 	 * anyone in the process of flushing the inode.
 	 */
-	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) ||
-	       iip->ili_format.ilf_fields != 0);
+	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || iip->ili_fields != 0);
 
 	/*
 	 * Push the inode to it's backing buffer. This will not remove the
@@ -904,7 +909,7 @@ xfs_iflush_abort(
 		 * Clear the inode logging fields so no more flushes are
 		 * attempted.
 		 */
-		iip->ili_format.ilf_fields = 0;
+		iip->ili_fields = 0;
 	}
 	/*
 	 * Release the inode's flush lock since we're done with it.
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
index 25784b0..bc183d8 100644
--- a/fs/xfs/xfs_inode_item.h
+++ b/fs/xfs/xfs_inode_item.h
@@ -134,6 +134,7 @@ typedef struct xfs_inode_log_item {
 	unsigned short		ili_lock_flags;	   /* lock flags */
 	unsigned short		ili_logged;	   /* flushed logged data */
 	unsigned int		ili_last_fields;   /* fields when flushed */
+	unsigned int		ili_fields;	   /* fields to be logged */
 	struct xfs_bmbt_rec	*ili_extents_buf;  /* array of logged
 						      data exts */
 	struct xfs_bmbt_rec	*ili_aextents_buf; /* array of logged
@@ -148,8 +149,7 @@ typedef struct xfs_inode_log_item {
 
 static inline int xfs_inode_clean(xfs_inode_t *ip)
 {
-	return !ip->i_itemp ||
-		!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL);
+	return !ip->i_itemp || !(ip->i_itemp->ili_fields & XFS_ILOG_ALL);
 }
 
 extern void xfs_inode_item_init(struct xfs_inode *, struct xfs_mount *);
diff --git a/fs/xfs/xfs_trans_inode.c b/fs/xfs/xfs_trans_inode.c
index 47625a2..fc2e73e 100644
--- a/fs/xfs/xfs_trans_inode.c
+++ b/fs/xfs/xfs_trans_inode.c
@@ -154,12 +154,12 @@ xfs_trans_log_inode(
 	/*
 	 * Always OR in the bits from the ili_last_fields field.
 	 * This is to coordinate with the xfs_iflush() and xfs_iflush_done()
-	 * routines in the eventual clearing of the ilf_fields bits.
+	 * routines in the eventual clearing of the ili_fields bits.
 	 * See the big comment in xfs_iflush() for an explanation of
 	 * this coordination mechanism.
 	 */
 	flags |= ip->i_itemp->ili_last_fields;
-	ip->i_itemp->ili_format.ilf_fields |= flags;
+	ip->i_itemp->ili_fields |= flags;
 }
 
 #ifdef XFS_TRANS_DEBUG
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 056/102] xfs: reimplement fdatasync support

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 8f639ddea0c4978ae9b4e46ea041c9e5afe0ee8d

Add an in-memory only flag to say we logged timestamps only, and use it to
check if fdatasync can optimize away the log force.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c  |    7 +++++--
 fs/xfs/linux-2.6/xfs_super.c |    2 +-
 fs/xfs/xfs_inode_item.c      |    3 ++-
 fs/xfs/xfs_inode_item.h      |   11 ++++++++++-
 4 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index a575a63..1dbd1d8 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -191,8 +191,11 @@ xfs_file_fsync(
 	 * to flush the log up to the latest LSN that touched the inode.
 	 */
 	xfs_ilock(ip, XFS_ILOCK_SHARED);
-	if (xfs_ipincount(ip))
-		lsn = ip->i_itemp->ili_last_lsn;
+	if (xfs_ipincount(ip)) {
+		if (!datasync ||
+		    (ip->i_itemp->ili_fields & ~XFS_ILOG_TIMESTAMP))
+			lsn = ip->i_itemp->ili_last_lsn;
+	}
 	xfs_iunlock(ip, XFS_ILOCK_SHARED);
 
 	if (lsn)
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index c12df49..ec5ca7b 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -944,7 +944,7 @@ xfs_fs_dirty_inode(
 	ip->i_d.di_mtime.t_nsec = (__int32_t)inode->i_mtime.tv_nsec;
 
 	xfs_trans_ijoin(tp, ip);
-	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_TIMESTAMP);
 	error = xfs_trans_commit(tp, 0);
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	if (error)
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 45514c5..a76d1b9 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -438,7 +438,8 @@ out:
 	 * games in recovery easier, which isn't a big deal as just about any
 	 * transaction would dirty it anyway.
 	 */
-	iip->ili_format.ilf_fields = XFS_ILOG_CORE | iip->ili_fields;
+	iip->ili_format.ilf_fields = XFS_ILOG_CORE |
+		(iip->ili_fields & ~XFS_ILOG_TIMESTAMP);
 	ASSERT(nvecs == lip->li_desc->lid_size);
 	iip->ili_format.ilf_size = nvecs;
 }
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
index bc183d8..41d61c3 100644
--- a/fs/xfs/xfs_inode_item.h
+++ b/fs/xfs/xfs_inode_item.h
@@ -86,6 +86,15 @@ typedef struct xfs_inode_log_format_64 {
 #define	XFS_ILOG_AEXT	0x080	/* log i_af.if_extents */
 #define	XFS_ILOG_ABROOT	0x100	/* log i_af.i_broot */
 
+
+/*
+ * The timestamps are dirty, but not necessarily anything else in the inode
+ * core.  Unlike the other fields above this one must never make it to disk
+ * in the ilf_fields of the inode_log_format, but is purely store in-memory in
+ * ili_fields in the inode_log_item.
+ */
+#define XFS_ILOG_TIMESTAMP	0x4000
+
 #define	XFS_ILOG_NONCORE	(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
 				 XFS_ILOG_DBROOT | XFS_ILOG_DEV | \
 				 XFS_ILOG_UUID | XFS_ILOG_ADATA | \
@@ -101,7 +110,7 @@ typedef struct xfs_inode_log_format_64 {
 				 XFS_ILOG_DEXT | XFS_ILOG_DBROOT | \
 				 XFS_ILOG_DEV | XFS_ILOG_UUID | \
 				 XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
-				 XFS_ILOG_ABROOT)
+				 XFS_ILOG_ABROOT | XFS_ILOG_TIMESTAMP)
 
 static inline int xfs_ilog_fbroot(int w)
 {
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 057/102] xfs: fallback to vmalloc for large buffers in xfs_attrmulti_attr_get

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: ad650f5b27bc9858360b42aaa0d9204d16115316

xfsdump uses for a large buffer for extended attributes, which has a
kmalloc'd shadow buffer in the kernel. This can fail after the
system has been running for some time as it is a high order
allocation. Add a fallback to vmalloc so that it doesn't require
contiguous memory and so won't randomly fail while xfsdump is
running.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_ioctl.c |   14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index acca2c5..ada7da9 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -450,9 +450,12 @@ xfs_attrmulti_attr_get(
 
 	if (*len > XATTR_SIZE_MAX)
 		return EINVAL;
-	kbuf = kmalloc(*len, GFP_KERNEL);
-	if (!kbuf)
-		return ENOMEM;
+	kbuf = kmem_zalloc(*len, KM_SLEEP | KM_MAYFAIL);
+	if (!kbuf) {
+		kbuf = kmem_zalloc_large(*len);
+		if (!kbuf)
+			return ENOMEM;
+	}
 
 	error = xfs_attr_get(XFS_I(inode), name, kbuf, (int *)len, flags);
 	if (error)
@@ -462,7 +465,10 @@ xfs_attrmulti_attr_get(
 		error = EFAULT;
 
  out_kfree:
-	kfree(kbuf);
+	if (is_vmalloc_addr(kbuf))
+		kmem_free_large(kbuf);
+	else
+		kmem_free(kbuf);
 	return error;
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 058/102] xfs: fallback to vmalloc for large buffers in xfs_getbmap

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: f074211f6041305b645669464343d504f4e6a290

xfs_getbmap uses for a large buffer for extents, which is kmalloc'd.
This can fail after the system has been running for some time as it
is a high order allocation. Add a fallback to vmalloc so that it
doesn't require contiguous memory and so won't randomly fail on
files with large extent lists.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_bmap.c |   13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 51b3489..2af58c4 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5455,8 +5455,12 @@ xfs_getbmap(
 	if (bmv->bmv_count > ULONG_MAX / sizeof(struct getbmapx))
 		return XFS_ERROR(ENOMEM);
 	out = kmem_zalloc(bmv->bmv_count * sizeof(struct getbmapx), KM_MAYFAIL);
-	if (!out)
-		return XFS_ERROR(ENOMEM);
+	if (!out) {
+		out = kmem_zalloc_large(bmv->bmv_count *
+					sizeof(struct getbmapx));
+		if (!out)
+			return XFS_ERROR(ENOMEM);
+	}
 
 	xfs_ilock(ip, XFS_IOLOCK_SHARED);
 	if (whichfork == XFS_DATA_FORK && !(iflags & BMV_IF_DELALLOC)) {
@@ -5581,7 +5585,10 @@ xfs_getbmap(
 			break;
 	}
 
-	kmem_free(out);
+	if (is_vmalloc_addr(out))
+		kmem_free_large(out);
+	else
+		kmem_free(out);
 	return error;
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 059/102] xfs: fix deadlock in xfs_rtfree_extent

[Dave Chinner]

From: Kamal Dasu <kdasu.kdev@gmail.com>

Upstream commit: 5575acc7807595687288b3bbac15103f2a5462e1

To fix the deadlock caused by repeatedly calling xfs_rtfree_extent

 - removed xfs_ilock() and xfs_trans_ijoin() from xfs_rtfree_extent(),
   instead added asserts that the inode is locked and has an inode_item
   attached to it.
 - in xfs_bunmapi() when dealing with an inode with the rt flag
   call xfs_ilock() and xfs_trans_ijoin() so that the
   reference count is bumped on the inode and attached it to the
   transaction before calling into xfs_bmap_del_extent, similar to
   what we do in xfs_bmap_rtalloc.

Signed-off-by: Kamal Dasu <kdasu.kdev@gmail.com>
Reviewed-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_bmap.c    |    9 +++++++++
 fs/xfs/xfs_rtalloc.c |    9 ++++-----
 2 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 2af58c4..8f3b730 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5039,6 +5039,15 @@ xfs_bunmapi(
 		cur->bc_private.b.flags = 0;
 	} else
 		cur = NULL;
+
+	if (isrt) {
+		/*
+		 * Synchronize by locking the bitmap inode.
+		 */
+		xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, mp->m_rbmip);
+	}
+
 	extno = 0;
 	while (bno != (xfs_fileoff_t)-1 && bno >= start && lastx >= 0 &&
 	       (nexts == 0 || extno < nexts)) {
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
index 8f76fdf..7c62363 100644
--- a/fs/xfs/xfs_rtalloc.c
+++ b/fs/xfs/xfs_rtalloc.c
@@ -183,6 +183,7 @@ error_cancel:
 		oblocks = map.br_startoff + map.br_blockcount;
 	}
 	return 0;
+
 error:
 	return error;
 }
@@ -2149,11 +2150,9 @@ xfs_rtfree_extent(
 	xfs_buf_t	*sumbp;		/* summary file block buffer */
 
 	mp = tp->t_mountp;
-	/*
-	 * Synchronize by locking the bitmap inode.
-	 */
-	xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL);
-	xfs_trans_ijoin_ref(tp, mp->m_rbmip, XFS_ILOCK_EXCL);
+
+	ASSERT(mp->m_rbmip->i_itemp != NULL);
+	ASSERT(xfs_isilocked(mp->m_rbmip, XFS_ILOCK_EXCL));
 
 #if defined(__KERNEL__) && defined(DEBUG)
 	/*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 060/102] xfs: Fix open flag handling in open_by_handle code

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 1a1d772433d42aaff7315b3468fef5951604f5c6

Sparse identified some unsafe handling of open flags in the xfs open
by handle ioctl code. Update the code to use the correct access
macros to ensure that we handle the open flags correctly.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_ioctl.c |   14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index ada7da9..b27fb6d 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -209,6 +209,7 @@ xfs_open_by_handle(
 	struct file		*filp;
 	struct inode		*inode;
 	struct dentry		*dentry;
+	fmode_t			fmode;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -XFS_ERROR(EPERM);
@@ -228,26 +229,21 @@ xfs_open_by_handle(
 	hreq->oflags |= O_LARGEFILE;
 #endif
 
-	/* Put open permission in namei format. */
 	permflag = hreq->oflags;
-	if ((permflag+1) & O_ACCMODE)
-		permflag++;
-	if (permflag & O_TRUNC)
-		permflag |= 2;
-
+	fmode = OPEN_FMODE(permflag);
 	if ((!(permflag & O_APPEND) || (permflag & O_TRUNC)) &&
-	    (permflag & FMODE_WRITE) && IS_APPEND(inode)) {
+	    (fmode & FMODE_WRITE) && IS_APPEND(inode)) {
 		error = -XFS_ERROR(EPERM);
 		goto out_dput;
 	}
 
-	if ((permflag & FMODE_WRITE) && IS_IMMUTABLE(inode)) {
+	if ((fmode & FMODE_WRITE) && IS_IMMUTABLE(inode)) {
 		error = -XFS_ERROR(EACCES);
 		goto out_dput;
 	}
 
 	/* Can't write directories. */
-	if (S_ISDIR(inode->i_mode) && (permflag & FMODE_WRITE)) {
+	if (S_ISDIR(inode->i_mode) && (fmode & FMODE_WRITE)) {
 		error = -XFS_ERROR(EISDIR);
 		goto out_dput;
 	}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 061/102] xfs: introduce an allocation workqueue

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: c999a223c2f0d31c64ef7379814cea1378b2b800

We currently have significant issues with the amount of stack that
allocation in XFS uses, especially in the writeback path. We can
easily consume 4k of stack between mapping the page, manipulating
the bmap btree and allocating blocks from the free list. Not to
mention btree block readahead and other functionality that issues IO
in the allocation path.

As a result, we can no longer fit allocation in the writeback path
in the stack space provided on x86_64. To alleviate this problem,
introduce an allocation workqueue and move all allocations to a
seperate context. This can be easily added as an interposing layer
into xfs_alloc_vextent(), which takes a single argument structure
and does not return until the allocation is complete or has failed.

To do this, add a work structure and a completion to the allocation
args structure. This allows xfs_alloc_vextent to queue the args onto
the workqueue and wait for it to be completed by the worker. This
can be done completely transparently to the caller.

The worker function needs to ensure that it sets and clears the
PF_TRANS flag appropriately as it is being run in an active
transaction context. Work can also be queued in a memory reclaim
context, so a rescuer is needed for the workqueue.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_super.c |   16 ++++++++++++++++
 fs/xfs/xfs_alloc.c           |   34 +++++++++++++++++++++++++++++++++-
 fs/xfs/xfs_alloc.h           |    5 +++++
 3 files changed, 54 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index ec5ca7b..d687329 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1645,12 +1645,28 @@ xfs_init_workqueues(void)
 	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_NON_REENTRANT, 0);
 	if (!xfs_syncd_wq)
 		return -ENOMEM;
+
+	/*
+	 * The allocation workqueue can be used in memory reclaim situations
+	 * (writepage path), and parallelism is only limited by the number of
+	 * AGs in all the filesystems mounted. Hence use the default large
+	 * max_active value for this workqueue.
+	 */
+	xfs_alloc_wq = alloc_workqueue("xfsalloc", WQ_MEM_RECLAIM, 0);
+	if (!xfs_alloc_wq)
+		goto out_destroy_syncd;
+
 	return 0;
+
+out_destroy_syncd:
+	destroy_workqueue(xfs_syncd_wq);
+	return -ENOMEM;
 }
 
 STATIC void
 xfs_destroy_workqueues(void)
 {
+	destroy_workqueue(xfs_alloc_wq);
 	destroy_workqueue(xfs_syncd_wq);
 }
 
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 95862bb..2445e1a 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -35,6 +35,7 @@
 #include "xfs_error.h"
 #include "xfs_trace.h"
 
+struct workqueue_struct *xfs_alloc_wq;
 
 #define XFS_ABSDIFF(a,b)	(((a) <= (b)) ? ((b) - (a)) : ((a) - (b)))
 
@@ -2212,7 +2213,7 @@ xfs_alloc_read_agf(
  * group or loop over the allocation groups to find the result.
  */
 int				/* error */
-xfs_alloc_vextent(
+__xfs_alloc_vextent(
 	xfs_alloc_arg_t	*args)	/* allocation argument structure */
 {
 	xfs_agblock_t	agsize;	/* allocation group size */
@@ -2422,6 +2423,37 @@ error0:
 	return error;
 }
 
+static void
+xfs_alloc_vextent_worker(
+	struct work_struct	*work)
+{
+	struct xfs_alloc_arg	*args = container_of(work,
+						struct xfs_alloc_arg, work);
+	unsigned long		pflags;
+
+	/* we are in a transaction context here */
+	current_set_flags_nested(&pflags, PF_FSTRANS);
+
+	args->result = __xfs_alloc_vextent(args);
+	complete(args->done);
+
+	current_restore_flags_nested(&pflags, PF_FSTRANS);
+}
+
+
+int				/* error */
+xfs_alloc_vextent(
+	xfs_alloc_arg_t	*args)	/* allocation argument structure */
+{
+	DECLARE_COMPLETION_ONSTACK(done);
+
+	args->done = &done;
+	INIT_WORK(&args->work, xfs_alloc_vextent_worker);
+	queue_work(xfs_alloc_wq, &args->work);
+	wait_for_completion(&done);
+	return args->result;
+}
+
 /*
  * Free an extent.
  * Just break up the extent address and hand off to xfs_free_ag_extent
diff --git a/fs/xfs/xfs_alloc.h b/fs/xfs/xfs_alloc.h
index 2f52b92..ab5d0fd 100644
--- a/fs/xfs/xfs_alloc.h
+++ b/fs/xfs/xfs_alloc.h
@@ -25,6 +25,8 @@ struct xfs_perag;
 struct xfs_trans;
 struct xfs_busy_extent;
 
+extern struct workqueue_struct *xfs_alloc_wq;
+
 /*
  * Freespace allocation types.  Argument to xfs_alloc_[v]extent.
  */
@@ -119,6 +121,9 @@ typedef struct xfs_alloc_arg {
 	char		isfl;		/* set if is freelist blocks - !acctg */
 	char		userdata;	/* set if this is user data */
 	xfs_fsblock_t	firstblock;	/* io first block allocated */
+	struct completion *done;
+	struct work_struct work;
+	int		result;
 } xfs_alloc_arg_t;
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 062/102] xfs: trace xfs_name strings correctly

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: f616137519feb17b849894fcbe634a021d3fa7db

Strings store in an xfs_name structure are often not NUL terminated,
print them using the correct printf specifiers that make use of the
string length store in the xfs_name structure.

Reported-by: Brian Candler <B.Candler@pobox.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 3509faf..4e544af 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -628,16 +628,19 @@ DECLARE_EVENT_CLASS(xfs_namespace_class,
 	TP_STRUCT__entry(
 		__field(dev_t, dev)
 		__field(xfs_ino_t, dp_ino)
+		__field(int, namelen)
 		__dynamic_array(char, name, name->len)
 	),
 	TP_fast_assign(
 		__entry->dev = VFS_I(dp)->i_sb->s_dev;
 		__entry->dp_ino = dp->i_ino;
+		__entry->namelen = name->len;
 		memcpy(__get_str(name), name->name, name->len);
 	),
-	TP_printk("dev %d:%d dp ino 0x%llx name %s",
+	TP_printk("dev %d:%d dp ino 0x%llx name %.*s",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->dp_ino,
+		  __entry->namelen,
 		  __get_str(name))
 )
 
@@ -659,6 +662,8 @@ TRACE_EVENT(xfs_rename,
 		__field(dev_t, dev)
 		__field(xfs_ino_t, src_dp_ino)
 		__field(xfs_ino_t, target_dp_ino)
+		__field(int, src_namelen)
+		__field(int, target_namelen)
 		__dynamic_array(char, src_name, src_name->len)
 		__dynamic_array(char, target_name, target_name->len)
 	),
@@ -666,15 +671,20 @@ TRACE_EVENT(xfs_rename,
 		__entry->dev = VFS_I(src_dp)->i_sb->s_dev;
 		__entry->src_dp_ino = src_dp->i_ino;
 		__entry->target_dp_ino = target_dp->i_ino;
+		__entry->src_namelen = src_name->len;
+		__entry->target_namelen = target_name->len;
 		memcpy(__get_str(src_name), src_name->name, src_name->len);
-		memcpy(__get_str(target_name), target_name->name, target_name->len);
+		memcpy(__get_str(target_name), target_name->name,
+			target_name->len);
 	),
 	TP_printk("dev %d:%d src dp ino 0x%llx target dp ino 0x%llx"
-		  " src name %s target name %s",
+		  " src name %.*s target name %.*s",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->src_dp_ino,
 		  __entry->target_dp_ino,
+		  __entry->src_namelen,
 		  __get_str(src_name),
+		  __entry->target_namelen,
 		  __get_str(target_name))
 )
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 063/102] xfs: Account log unmount transaction correctly

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 3948659e30808fbaa7673bbe89de2ae9769e20a7

There have been a few reports of this warning appearing recently:

XFS (dm-4): xlog_space_left: head behind tail
 tail_cycle = 129, tail_bytes = 20163072
 GH   cycle = 129, GH   bytes = 20162880

The common cause appears to be lots of freeze and unfreeze cycles,
and the output from the warnings indicates that we are leaking
around 8 bytes of log space per freeze/unfreeze cycle.

When we freeze the filesystem, we write an unmount record and that
uses xlog_write directly - a special type of transaction,
effectively. What it doesn't do, however, is correctly account for
the log space it uses. The unmount record writes an 8 byte structure
with a special magic number into the log, and the space this
consumes is not accounted for in the log ticket tracking the
operation. Hence we leak 8 bytes every unmount record that is
written.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index d92b442..653cba8 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -726,8 +726,9 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 				.lv_iovecp = &reg,
 			};
 
-			/* remove inited flag */
+			/* remove inited flag, and account for space used */
 			tic->t_flags = 0;
+			tic->t_curr_res -= sizeof(magic);
 			error = xlog_write(log, &vec, tic, &lsn,
 					   NULL, XLOG_UNMOUNT_TRANS);
 			/*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 064/102] xfs: fix fstrim offset calculations

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: a66d636385d621e98a915233250356c394a437de

xfs_ioc_fstrim() doesn't treat the incoming offset and length
correctly. It treats them as a filesystem block address, rather than
a disk address. This is wrong because the range passed in is a
linear representation, while the filesystem block address notation
is a sparse representation. Hence we cannot convert the range direct
to filesystem block units and then use that for calculating the
range to trim.

While this sounds dangerous, the problem is limited to calculating
what AGs need to be trimmed. The code that calcuates the actual
ranges to trim gets the right result (i.e. only ever discards free
space), even though it uses the wrong ranges to limit what is
trimmed. Hence this is not a bug that endangers user data.

Fix this by treating the range as a disk address range and use the
appropriate functions to convert the range into the desired formats
for calculations.

Further, fix the first free extent lookup (the longest) to actually
find the largest free extent. Currently this lookup uses a <=
lookup, which results in finding the extent to the left of the
largest because we can never get an exact match on the largest
extent. This is due to the fact that while we know it's size, we
don't know it's location and so the exact match fails and we move
one record to the left to get the next largest extent. Instead, use
a >= search so that the lookup returns the largest extent regardless
of the fact we don't get an exact match on it.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_discard.c |   61 +++++++++++++++++++++++++---------------
 fs/xfs/xfs_alloc.c             |    2 +-
 fs/xfs/xfs_alloc.h             |    7 +++++
 3 files changed, 46 insertions(+), 24 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_discard.c b/fs/xfs/linux-2.6/xfs_discard.c
index 286a051..1ad3a4b 100644
--- a/fs/xfs/linux-2.6/xfs_discard.c
+++ b/fs/xfs/linux-2.6/xfs_discard.c
@@ -37,9 +37,9 @@ STATIC int
 xfs_trim_extents(
 	struct xfs_mount	*mp,
 	xfs_agnumber_t		agno,
-	xfs_fsblock_t		start,
-	xfs_fsblock_t		end,
-	xfs_fsblock_t		minlen,
+	xfs_daddr_t		start,
+	xfs_daddr_t		end,
+	xfs_daddr_t		minlen,
 	__uint64_t		*blocks_trimmed)
 {
 	struct block_device	*bdev = mp->m_ddev_targp->bt_bdev;
@@ -67,7 +67,7 @@ xfs_trim_extents(
 	/*
 	 * Look up the longest btree in the AGF and start with it.
 	 */
-	error = xfs_alloc_lookup_le(cur, 0,
+	error = xfs_alloc_lookup_ge(cur, 0,
 			    be32_to_cpu(XFS_BUF_TO_AGF(agbp)->agf_longest), &i);
 	if (error)
 		goto out_del_cursor;
@@ -77,8 +77,10 @@ xfs_trim_extents(
 	 * enough to be worth discarding.
 	 */
 	while (i) {
-		xfs_agblock_t fbno;
-		xfs_extlen_t flen;
+		xfs_agblock_t	fbno;
+		xfs_extlen_t	flen;
+		xfs_daddr_t	dbno;
+		xfs_extlen_t	dlen;
 
 		error = xfs_alloc_get_rec(cur, &fbno, &flen, &i);
 		if (error)
@@ -87,9 +89,17 @@ xfs_trim_extents(
 		ASSERT(flen <= be32_to_cpu(XFS_BUF_TO_AGF(agbp)->agf_longest));
 
 		/*
+		 * use daddr format for all range/len calculations as that is
+		 * the format the range/len variables are supplied in by
+		 * userspace.
+		 */
+		dbno = XFS_AGB_TO_DADDR(mp, agno, fbno);
+		dlen = XFS_FSB_TO_BB(mp, flen);
+
+		/*
 		 * Too small?  Give up.
 		 */
-		if (flen < minlen) {
+		if (dlen < minlen) {
 			trace_xfs_discard_toosmall(mp, agno, fbno, flen);
 			goto out_del_cursor;
 		}
@@ -99,8 +109,7 @@ xfs_trim_extents(
 		 * supposed to discard skip it.  Do not bother to trim
 		 * down partially overlapping ranges for now.
 		 */
-		if (XFS_AGB_TO_FSB(mp, agno, fbno) + flen < start ||
-		    XFS_AGB_TO_FSB(mp, agno, fbno) > end) {
+		if (dbno + dlen < start || dbno > end) {
 			trace_xfs_discard_exclude(mp, agno, fbno, flen);
 			goto next_extent;
 		}
@@ -115,10 +124,7 @@ xfs_trim_extents(
 		}
 
 		trace_xfs_discard_extent(mp, agno, fbno, flen);
-		error = -blkdev_issue_discard(bdev,
-				XFS_AGB_TO_DADDR(mp, agno, fbno),
-				XFS_FSB_TO_BB(mp, flen),
-				GFP_NOFS, 0);
+		error = -blkdev_issue_discard(bdev, dbno, dlen, GFP_NOFS, 0);
 		if (error)
 			goto out_del_cursor;
 		*blocks_trimmed += flen;
@@ -137,6 +143,15 @@ out_put_perag:
 	return error;
 }
 
+/*
+ * trim a range of the filesystem.
+ *
+ * Note: the parameters passed from userspace are byte ranges into the
+ * filesystem which does not match to the format we use for filesystem block
+ * addressing. FSB addressing is sparse (AGNO|AGBNO), while the incoming format
+ * is a linear address range. Hence we need to use DADDR based conversions and
+ * comparisons for determining the correct offset and regions to trim.
+ */
 int
 xfs_ioc_trim(
 	struct xfs_mount		*mp,
@@ -145,7 +160,7 @@ xfs_ioc_trim(
 	struct request_queue	*q = mp->m_ddev_targp->bt_bdev->bd_disk->queue;
 	unsigned int		granularity = q->limits.discard_granularity;
 	struct fstrim_range	range;
-	xfs_fsblock_t		start, end, minlen;
+	xfs_daddr_t		start, end, minlen;
 	xfs_agnumber_t		start_agno, end_agno, agno;
 	__uint64_t		blocks_trimmed = 0;
 	int			error, last_error = 0;
@@ -159,22 +174,22 @@ xfs_ioc_trim(
 
 	/*
 	 * Truncating down the len isn't actually quite correct, but using
-	 * XFS_B_TO_FSB would mean we trivially get overflows for values
+	 * BBTOB would mean we trivially get overflows for values
 	 * of ULLONG_MAX or slightly lower.  And ULLONG_MAX is the default
 	 * used by the fstrim application.  In the end it really doesn't
 	 * matter as trimming blocks is an advisory interface.
 	 */
-	start = XFS_B_TO_FSBT(mp, range.start);
-	end = start + XFS_B_TO_FSBT(mp, range.len) - 1;
-	minlen = XFS_B_TO_FSB(mp, max_t(u64, granularity, range.minlen));
+	start = BTOBB(range.start);
+	end = start + BTOBBT(range.len) - 1;
+	minlen = BTOBB(max_t(u64, granularity, range.minlen));
 
-	if (start >= mp->m_sb.sb_dblocks)
+	if (XFS_BB_TO_FSB(mp, start) >= mp->m_sb.sb_dblocks)
 		return -XFS_ERROR(EINVAL);
-	if (end > mp->m_sb.sb_dblocks - 1)
-		end = mp->m_sb.sb_dblocks - 1;
+	if (end > XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) - 1)
+		end = XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)- 1;
 
-	start_agno = XFS_FSB_TO_AGNO(mp, start);
-	end_agno = XFS_FSB_TO_AGNO(mp, end);
+	start_agno = xfs_daddr_to_agno(mp, start);
+	end_agno = xfs_daddr_to_agno(mp, end);
 
 	for (agno = start_agno; agno <= end_agno; agno++) {
 		error = -xfs_trim_extents(mp, agno, start, end, minlen,
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 2445e1a..dfa9454 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -69,7 +69,7 @@ xfs_alloc_lookup_eq(
  * Lookup the first record greater than or equal to [bno, len]
  * in the btree given by cur.
  */
-STATIC int				/* error */
+int				/* error */
 xfs_alloc_lookup_ge(
 	struct xfs_btree_cur	*cur,	/* btree cursor */
 	xfs_agblock_t		bno,	/* starting block of extent */
diff --git a/fs/xfs/xfs_alloc.h b/fs/xfs/xfs_alloc.h
index ab5d0fd..3a7e7d8 100644
--- a/fs/xfs/xfs_alloc.h
+++ b/fs/xfs/xfs_alloc.h
@@ -248,6 +248,13 @@ xfs_alloc_lookup_le(
 	xfs_extlen_t		len,	/* length of extent */
 	int			*stat);	/* success/failure */
 
+int				/* error */
+xfs_alloc_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat);	/* success/failure */
+
 int					/* error */
 xfs_alloc_get_rec(
 	struct xfs_btree_cur	*cur,	/* btree cursor */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 065/102] xfs: add lots of attribute trace points

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 5a5881cdeec2c019b5c9a307800218ee029f7f61

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   62 ++++++++++++++++++++++++++++++++++++++++--
 fs/xfs/xfs_attr.c            |   16 +++++++++++
 fs/xfs/xfs_attr_leaf.c       |   40 ++++++++++++++++++++++++---
 fs/xfs/xfs_da_btree.c        |   32 ++++++++++++++++++++++
 4 files changed, 144 insertions(+), 6 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 4e544af..1315931 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -1452,7 +1452,7 @@ DEFINE_ALLOC_EVENT(xfs_alloc_vextent_noagbp);
 DEFINE_ALLOC_EVENT(xfs_alloc_vextent_loopfailed);
 DEFINE_ALLOC_EVENT(xfs_alloc_vextent_allfailed);
 
-DECLARE_EVENT_CLASS(xfs_dir2_class,
+DECLARE_EVENT_CLASS(xfs_da_class,
 	TP_PROTO(struct xfs_da_args *args),
 	TP_ARGS(args),
 	TP_STRUCT__entry(
@@ -1487,7 +1487,7 @@ DECLARE_EVENT_CLASS(xfs_dir2_class,
 )
 
 #define DEFINE_DIR2_EVENT(name) \
-DEFINE_EVENT(xfs_dir2_class, name, \
+DEFINE_EVENT(xfs_da_class, name, \
 	TP_PROTO(struct xfs_da_args *args), \
 	TP_ARGS(args))
 DEFINE_DIR2_EVENT(xfs_dir2_sf_addname);
@@ -1516,6 +1516,64 @@ DEFINE_DIR2_EVENT(xfs_dir2_node_replace);
 DEFINE_DIR2_EVENT(xfs_dir2_node_removename);
 DEFINE_DIR2_EVENT(xfs_dir2_node_to_leaf);
 
+#define DEFINE_ATTR_EVENT(name) \
+DEFINE_EVENT(xfs_da_class, name, \
+	TP_PROTO(struct xfs_da_args *args), \
+	TP_ARGS(args))
+DEFINE_ATTR_EVENT(xfs_attr_sf_add);
+DEFINE_ATTR_EVENT(xfs_attr_sf_addname);
+DEFINE_ATTR_EVENT(xfs_attr_sf_create);
+DEFINE_ATTR_EVENT(xfs_attr_sf_lookup);
+DEFINE_ATTR_EVENT(xfs_attr_sf_remove);
+DEFINE_ATTR_EVENT(xfs_attr_sf_removename);
+DEFINE_ATTR_EVENT(xfs_attr_sf_to_leaf);
+
+DEFINE_ATTR_EVENT(xfs_attr_leaf_add);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_add_old);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_add_new);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_addname);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_create);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_lookup);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_replace);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_removename);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_split);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_split_before);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_split_after);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_clearflag);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_setflag);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_flipflags);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_to_sf);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_to_node);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_rebalance);
+DEFINE_ATTR_EVENT(xfs_attr_leaf_unbalance);
+
+DEFINE_ATTR_EVENT(xfs_attr_node_addname);
+DEFINE_ATTR_EVENT(xfs_attr_node_lookup);
+DEFINE_ATTR_EVENT(xfs_attr_node_replace);
+DEFINE_ATTR_EVENT(xfs_attr_node_removename);
+
+#define DEFINE_DA_EVENT(name) \
+DEFINE_EVENT(xfs_da_class, name, \
+	TP_PROTO(struct xfs_da_args *args), \
+	TP_ARGS(args))
+DEFINE_DA_EVENT(xfs_da_split);
+DEFINE_DA_EVENT(xfs_da_join);
+DEFINE_DA_EVENT(xfs_da_link_before);
+DEFINE_DA_EVENT(xfs_da_link_after);
+DEFINE_DA_EVENT(xfs_da_unlink_back);
+DEFINE_DA_EVENT(xfs_da_unlink_forward);
+DEFINE_DA_EVENT(xfs_da_root_split);
+DEFINE_DA_EVENT(xfs_da_root_join);
+DEFINE_DA_EVENT(xfs_da_node_add);
+DEFINE_DA_EVENT(xfs_da_node_create);
+DEFINE_DA_EVENT(xfs_da_node_split);
+DEFINE_DA_EVENT(xfs_da_node_remove);
+DEFINE_DA_EVENT(xfs_da_node_rebalance);
+DEFINE_DA_EVENT(xfs_da_node_unbalance);
+DEFINE_DA_EVENT(xfs_da_swap_lastblock);
+DEFINE_DA_EVENT(xfs_da_grow_inode);
+DEFINE_DA_EVENT(xfs_da_shrink_inode);
+
 DECLARE_EVENT_CLASS(xfs_dir2_space_class,
 	TP_PROTO(struct xfs_da_args *args, int idx),
 	TP_ARGS(args, idx),
diff --git a/fs/xfs/xfs_attr.c b/fs/xfs/xfs_attr.c
index fe01931..96e0367 100644
--- a/fs/xfs/xfs_attr.c
+++ b/fs/xfs/xfs_attr.c
@@ -857,6 +857,8 @@ xfs_attr_shortform_addname(xfs_da_args_t *args)
 {
 	int newsize, forkoff, retval;
 
+	trace_xfs_attr_sf_addname(args);
+
 	retval = xfs_attr_shortform_lookup(args);
 	if ((args->flags & ATTR_REPLACE) && (retval == ENOATTR)) {
 		return(retval);
@@ -900,6 +902,8 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
 	xfs_dabuf_t *bp;
 	int retval, error, committed, forkoff;
 
+	trace_xfs_attr_leaf_addname(args);
+
 	/*
 	 * Read the (only) block in the attribute list in.
 	 */
@@ -924,6 +928,9 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
 			xfs_da_brelse(args->trans, bp);
 			return(retval);
 		}
+
+		trace_xfs_attr_leaf_replace(args);
+
 		args->op_flags |= XFS_DA_OP_RENAME;	/* an atomic rename */
 		args->blkno2 = args->blkno;		/* set 2nd entry info*/
 		args->index2 = args->index;
@@ -1094,6 +1101,8 @@ xfs_attr_leaf_removename(xfs_da_args_t *args)
 	xfs_dabuf_t *bp;
 	int error, committed, forkoff;
 
+	trace_xfs_attr_leaf_removename(args);
+
 	/*
 	 * Remove the attribute.
 	 */
@@ -1227,6 +1236,8 @@ xfs_attr_node_addname(xfs_da_args_t *args)
 	xfs_mount_t *mp;
 	int committed, retval, error;
 
+	trace_xfs_attr_node_addname(args);
+
 	/*
 	 * Fill in bucket of arguments/results/context to carry around.
 	 */
@@ -1253,6 +1264,9 @@ restart:
 	} else if (retval == EEXIST) {
 		if (args->flags & ATTR_CREATE)
 			goto out;
+
+		trace_xfs_attr_node_replace(args);
+
 		args->op_flags |= XFS_DA_OP_RENAME;	/* atomic rename op */
 		args->blkno2 = args->blkno;		/* set 2nd entry info*/
 		args->index2 = args->index;
@@ -1484,6 +1498,8 @@ xfs_attr_node_removename(xfs_da_args_t *args)
 	xfs_dabuf_t *bp;
 	int retval, error, committed, forkoff;
 
+	trace_xfs_attr_node_removename(args);
+
 	/*
 	 * Tie a string around our finger to remind us where we are.
 	 */
diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c
index ec4f133..787f71f 100644
--- a/fs/xfs/xfs_attr_leaf.c
+++ b/fs/xfs/xfs_attr_leaf.c
@@ -235,6 +235,8 @@ xfs_attr_shortform_create(xfs_da_args_t *args)
 	xfs_inode_t *dp;
 	xfs_ifork_t *ifp;
 
+	trace_xfs_attr_sf_create(args);
+
 	dp = args->dp;
 	ASSERT(dp != NULL);
 	ifp = dp->i_afp;
@@ -268,6 +270,8 @@ xfs_attr_shortform_add(xfs_da_args_t *args, int forkoff)
 	xfs_inode_t *dp;
 	xfs_ifork_t *ifp;
 
+	trace_xfs_attr_sf_add(args);
+
 	dp = args->dp;
 	mp = dp->i_mount;
 	dp->i_d.di_forkoff = forkoff;
@@ -342,6 +346,8 @@ xfs_attr_shortform_remove(xfs_da_args_t *args)
 	xfs_mount_t *mp;
 	xfs_inode_t *dp;
 
+	trace_xfs_attr_sf_remove(args);
+
 	dp = args->dp;
 	mp = dp->i_mount;
 	base = sizeof(xfs_attr_sf_hdr_t);
@@ -414,6 +420,8 @@ xfs_attr_shortform_lookup(xfs_da_args_t *args)
 	int i;
 	xfs_ifork_t *ifp;
 
+	trace_xfs_attr_sf_lookup(args);
+
 	ifp = args->dp->i_afp;
 	ASSERT(ifp->if_flags & XFS_IFINLINE);
 	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
@@ -485,6 +493,8 @@ xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
 	xfs_dabuf_t *bp;
 	xfs_ifork_t *ifp;
 
+	trace_xfs_attr_sf_to_leaf(args);
+
 	dp = args->dp;
 	ifp = dp->i_afp;
 	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
@@ -784,6 +794,8 @@ xfs_attr_leaf_to_shortform(xfs_dabuf_t *bp, xfs_da_args_t *args, int forkoff)
 	char *tmpbuffer;
 	int error, i;
 
+	trace_xfs_attr_leaf_to_sf(args);
+
 	dp = args->dp;
 	tmpbuffer = kmem_alloc(XFS_LBSIZE(dp->i_mount), KM_SLEEP);
 	ASSERT(tmpbuffer != NULL);
@@ -857,6 +869,8 @@ xfs_attr_leaf_to_node(xfs_da_args_t *args)
 	xfs_dablk_t blkno;
 	int error;
 
+	trace_xfs_attr_leaf_to_node(args);
+
 	dp = args->dp;
 	bp1 = bp2 = NULL;
 	error = xfs_da_grow_inode(args, &blkno);
@@ -920,6 +934,8 @@ xfs_attr_leaf_create(xfs_da_args_t *args, xfs_dablk_t blkno, xfs_dabuf_t **bpp)
 	xfs_dabuf_t *bp;
 	int error;
 
+	trace_xfs_attr_leaf_create(args);
+
 	dp = args->dp;
 	ASSERT(dp != NULL);
 	error = xfs_da_get_buf(args->trans, args->dp, blkno, -1, &bp,
@@ -957,6 +973,8 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
 	xfs_dablk_t blkno;
 	int error;
 
+	trace_xfs_attr_leaf_split(state->args);
+
 	/*
 	 * Allocate space for a new leaf node.
 	 */
@@ -986,10 +1004,13 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
 	 *
 	 * Insert the "new" entry in the correct block.
 	 */
-	if (state->inleaf)
+	if (state->inleaf) {
+		trace_xfs_attr_leaf_add_old(state->args);
 		error = xfs_attr_leaf_add(oldblk->bp, state->args);
-	else
+	} else {
+		trace_xfs_attr_leaf_add_new(state->args);
 		error = xfs_attr_leaf_add(newblk->bp, state->args);
+	}
 
 	/*
 	 * Update last hashval in each block since we added the name.
@@ -1010,6 +1031,8 @@ xfs_attr_leaf_add(xfs_dabuf_t *bp, xfs_da_args_t *args)
 	xfs_attr_leaf_map_t *map;
 	int tablesize, entsize, sum, tmp, i;
 
+	trace_xfs_attr_leaf_add(args);
+
 	leaf = bp->data;
 	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
 	ASSERT((args->index >= 0)
@@ -1137,8 +1160,6 @@ xfs_attr_leaf_add_work(xfs_dabuf_t *bp, xfs_da_args_t *args, int mapindex)
 	       (be32_to_cpu(entry->hashval) <= be32_to_cpu((entry+1)->hashval)));
 
 	/*
-	 * Copy the attribute name and value into the new space.
-	 *
 	 * For "remote" attribute values, simply note that we need to
 	 * allocate space for the "remote" value.  We can't actually
 	 * allocate the extents in this transaction, and we can't decide
@@ -1274,6 +1295,8 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
 	ASSERT(be16_to_cpu(leaf2->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
 	args = state->args;
 
+	trace_xfs_attr_leaf_rebalance(args);
+
 	/*
 	 * Check ordering of blocks, reverse if it makes things simpler.
 	 *
@@ -1819,6 +1842,8 @@ xfs_attr_leaf_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
 	xfs_mount_t *mp;
 	char *tmpbuffer;
 
+	trace_xfs_attr_leaf_unbalance(state->args);
+
 	/*
 	 * Set up environment.
 	 */
@@ -1928,6 +1953,8 @@ xfs_attr_leaf_lookup_int(xfs_dabuf_t *bp, xfs_da_args_t *args)
 	int probe, span;
 	xfs_dahash_t hashval;
 
+	trace_xfs_attr_leaf_lookup(args);
+
 	leaf = bp->data;
 	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
 	ASSERT(be16_to_cpu(leaf->hdr.count)
@@ -2454,6 +2481,7 @@ xfs_attr_leaf_clearflag(xfs_da_args_t *args)
 	char *name;
 #endif /* DEBUG */
 
+	trace_xfs_attr_leaf_clearflag(args);
 	/*
 	 * Set up the operation.
 	 */
@@ -2518,6 +2546,8 @@ xfs_attr_leaf_setflag(xfs_da_args_t *args)
 	xfs_dabuf_t *bp;
 	int error;
 
+	trace_xfs_attr_leaf_setflag(args);
+
 	/*
 	 * Set up the operation.
 	 */
@@ -2574,6 +2604,8 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
 	char *name1, *name2;
 #endif /* DEBUG */
 
+	trace_xfs_attr_leaf_flipflags(args);
+
 	/*
 	 * Read the block containing the "old" attr
 	 */
diff --git a/fs/xfs/xfs_da_btree.c b/fs/xfs/xfs_da_btree.c
index 6102ac6..3624f78 100644
--- a/fs/xfs/xfs_da_btree.c
+++ b/fs/xfs/xfs_da_btree.c
@@ -111,6 +111,8 @@ xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
 	int error;
 	xfs_trans_t *tp;
 
+	trace_xfs_da_node_create(args);
+
 	tp = args->trans;
 	error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork);
 	if (error)
@@ -143,6 +145,8 @@ xfs_da_split(xfs_da_state_t *state)
 	xfs_dabuf_t *bp;
 	int max, action, error, i;
 
+	trace_xfs_da_split(state->args);
+
 	/*
 	 * Walk back up the tree splitting/inserting/adjusting as necessary.
 	 * If we need to insert and there isn't room, split the node, then
@@ -181,10 +185,12 @@ xfs_da_split(xfs_da_state_t *state)
 			state->extravalid = 1;
 			if (state->inleaf) {
 				state->extraafter = 0;	/* before newblk */
+				trace_xfs_attr_leaf_split_before(state->args);
 				error = xfs_attr_leaf_split(state, oldblk,
 							    &state->extrablk);
 			} else {
 				state->extraafter = 1;	/* after newblk */
+				trace_xfs_attr_leaf_split_after(state->args);
 				error = xfs_attr_leaf_split(state, newblk,
 							    &state->extrablk);
 			}
@@ -303,6 +309,8 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
 	xfs_mount_t *mp;
 	xfs_dir2_leaf_t *leaf;
 
+	trace_xfs_da_root_split(state->args);
+
 	/*
 	 * Copy the existing (incorrect) block from the root node position
 	 * to a free space somewhere.
@@ -383,6 +391,8 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
 	int newcount, error;
 	int useextra;
 
+	trace_xfs_da_node_split(state->args);
+
 	node = oldblk->bp->data;
 	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
 
@@ -469,6 +479,8 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
 	int count, tmp;
 	xfs_trans_t *tp;
 
+	trace_xfs_da_node_rebalance(state->args);
+
 	node1 = blk1->bp->data;
 	node2 = blk2->bp->data;
 	/*
@@ -577,6 +589,8 @@ xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
 	xfs_da_node_entry_t *btree;
 	int tmp;
 
+	trace_xfs_da_node_add(state->args);
+
 	node = oldblk->bp->data;
 	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
 	ASSERT((oldblk->index >= 0) && (oldblk->index <= be16_to_cpu(node->hdr.count)));
@@ -622,6 +636,8 @@ xfs_da_join(xfs_da_state_t *state)
 	xfs_da_state_blk_t *drop_blk, *save_blk;
 	int action, error;
 
+	trace_xfs_da_join(state->args);
+
 	action = 0;
 	drop_blk = &state->path.blk[ state->path.active-1 ];
 	save_blk = &state->altpath.blk[ state->path.active-1 ];
@@ -710,6 +726,8 @@ xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk)
 	xfs_dabuf_t *bp;
 	int error;
 
+	trace_xfs_da_root_join(state->args);
+
 	args = state->args;
 	ASSERT(args != NULL);
 	ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
@@ -934,6 +952,8 @@ xfs_da_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk)
 	xfs_da_node_entry_t *btree;
 	int tmp;
 
+	trace_xfs_da_node_remove(state->args);
+
 	node = drop_blk->bp->data;
 	ASSERT(drop_blk->index < be16_to_cpu(node->hdr.count));
 	ASSERT(drop_blk->index >= 0);
@@ -977,6 +997,8 @@ xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
 	int tmp;
 	xfs_trans_t *tp;
 
+	trace_xfs_da_node_unbalance(state->args);
+
 	drop_node = drop_blk->bp->data;
 	save_node = save_blk->bp->data;
 	ASSERT(be16_to_cpu(drop_node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
@@ -1223,6 +1245,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
 		/*
 		 * Link new block in before existing block.
 		 */
+		trace_xfs_da_link_before(args);
 		new_info->forw = cpu_to_be32(old_blk->blkno);
 		new_info->back = old_info->back;
 		if (old_info->back) {
@@ -1244,6 +1267,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
 		/*
 		 * Link new block in after existing block.
 		 */
+		trace_xfs_da_link_after(args);
 		new_info->forw = old_info->forw;
 		new_info->back = cpu_to_be32(old_blk->blkno);
 		if (old_info->forw) {
@@ -1341,6 +1365,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
 	 * Unlink the leaf block from the doubly linked chain of leaves.
 	 */
 	if (be32_to_cpu(save_info->back) == drop_blk->blkno) {
+		trace_xfs_da_unlink_back(args);
 		save_info->back = drop_info->back;
 		if (drop_info->back) {
 			error = xfs_da_read_buf(args->trans, args->dp,
@@ -1358,6 +1383,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
 			xfs_da_buf_done(bp);
 		}
 	} else {
+		trace_xfs_da_unlink_forward(args);
 		save_info->forw = drop_info->forw;
 		if (drop_info->forw) {
 			error = xfs_da_read_buf(args->trans, args->dp,
@@ -1562,6 +1588,8 @@ xfs_da_grow_inode(xfs_da_args_t *args, xfs_dablk_t *new_blkno)
 	xfs_mount_t *mp;
 	xfs_drfsbno_t	nblks;
 
+	trace_xfs_da_grow_inode(args);
+
 	dp = args->dp;
 	mp = dp->i_mount;
 	w = args->whichfork;
@@ -1673,6 +1701,8 @@ xfs_da_swap_lastblock(xfs_da_args_t *args, xfs_dablk_t *dead_blknop,
 	xfs_dir2_leaf_t *dead_leaf2;
 	xfs_dahash_t dead_hash;
 
+	trace_xfs_da_swap_lastblock(args);
+
 	dead_buf = *dead_bufp;
 	dead_blkno = *dead_blknop;
 	tp = args->trans;
@@ -1861,6 +1891,8 @@ xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
 	xfs_trans_t *tp;
 	xfs_mount_t *mp;
 
+	trace_xfs_da_shrink_inode(args);
+
 	dp = args->dp;
 	w = args->whichfork;
 	tp = args->trans;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 066/102] xfs: don't fill statvfs with project quota for a directory

[Dave Chinner]

From: Jie Liu <jeff.liu@oracle.com>

Upstream commit: da5bf95e3cdca348327c82568c2860229c0daaa2
 if it was not enabled.

Check if the project quota is running or not before performing
xfs_qm_statvfs(), just return if not.  Otherwise the ASSERT
XFS_IS_QUOTA_RUNNING in xfs_qm_dqget will be popped.

Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_super.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index d687329..b86c460 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1106,7 +1106,7 @@ xfs_fs_statfs(
 
 	spin_unlock(&mp->m_sb_lock);
 
-	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) ||
+	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
 	    ((mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_OQUOTA_ENFD))) ==
 			      (XFS_PQUOTA_ACCT|XFS_OQUOTA_ENFD))
 		xfs_qm_statvfs(ip, statp);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 067/102] xfs: Ensure inode reclaim can run during quotacheck

[Dave Chinner]

Upstream commit: 8a00ebe4cfc90eda9ecb575ba97e22021cd8cf70

Because the mount process can run a quotacheck and consume lots of
inodes, we need to be able to run periodic inode reclaim during the
mount process. This will prevent running the system out of memory
during quota checks.

This essentially reverts 2bcf6e97, but that is safe to do now that
the quota sync code that was causing problems during long quotacheck
executions is now gone.

The reclaim work is currently protected from running during the
unmount process by a check against MS_ACTIVE. Unfortunately, this
also means that the reclaim work cannot run during mount.  The
unmount process should stop the reclaim cleanly before freeing
anything that the reclaim work depends on, so there is no need to
have this guard in place.

Also, the inode reclaim work is demand driven, so there is no need
to start it immediately during mount. It will be started the moment
an inode is queued for reclaim, so qutoacheck will trigger it just
fine.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_super.c |   17 +++++++++--------
 fs/xfs/linux-2.6/xfs_sync.c  |   19 +++++++++----------
 2 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index b86c460..9c7f4b9 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1008,7 +1008,6 @@ xfs_fs_put_super(
 	 * structure so we don't have memory reclaim racing with us here.
 	 */
 	xfs_inode_shrinker_unregister(mp);
-	xfs_syncd_stop(mp);
 
 	/*
 	 * Blow away any referenced inode in the filestreams cache.
@@ -1020,6 +1019,7 @@ xfs_fs_put_super(
 	XFS_bflush(mp->m_ddev_targp);
 
 	xfs_unmountfs(mp);
+	xfs_syncd_stop(mp);
 	xfs_freesb(mp);
 	xfs_icsb_destroy_counters(mp);
 	xfs_destroy_mount_workqueues(mp);
@@ -1397,22 +1397,22 @@ xfs_fs_fill_super(
 
 	xfs_inode_shrinker_register(mp);
 
-	error = xfs_mountfs(mp);
+	error = xfs_syncd_init(mp);
 	if (error)
 		goto out_filestream_unmount;
 
-	error = xfs_syncd_init(mp);
+	error = xfs_mountfs(mp);
 	if (error)
-		goto out_unmount;
+		goto out_syncd_stop;
 
 	root = igrab(VFS_I(mp->m_rootip));
 	if (!root) {
 		error = ENOENT;
-		goto out_syncd_stop;
+		goto out_unmount;
 	}
 	if (is_bad_inode(root)) {
 		error = EINVAL;
-		goto out_syncd_stop;
+		goto out_unmount;
 	}
 	sb->s_root = d_alloc_root(root);
 	if (!sb->s_root) {
@@ -1422,6 +1422,8 @@ xfs_fs_fill_super(
 
 	return 0;
 
+ out_syncd_stop:
+	xfs_syncd_stop(mp);
  out_filestream_unmount:
 	xfs_inode_shrinker_unregister(mp);
 	xfs_filestream_unmount(mp);
@@ -1441,8 +1443,6 @@ out_destroy_workqueues:
 
  out_iput:
 	iput(root);
- out_syncd_stop:
-	xfs_syncd_stop(mp);
  out_unmount:
 	xfs_inode_shrinker_unregister(mp);
 
@@ -1456,6 +1456,7 @@ out_destroy_workqueues:
 	XFS_bflush(mp->m_ddev_targp);
 
 	xfs_unmountfs(mp);
+	xfs_syncd_stop(mp);
 	goto out_free_sb;
 }
 
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index cf26dd3..ccb77cf 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -466,7 +466,15 @@ xfs_sync_worker(
 					struct xfs_mount, m_sync_work);
 	int		error;
 
-	if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
+	/*
+	 * We shouldn't write/force the log if we are in the mount/unmount
+	 * process or on a read only filesystem. The workqueue still needs to be
+	 * active in both cases, however, because it is used for inode reclaim
+	 * during these times. hence use the MS_ACTIVE flag to avoid doing
+	 * anything in these periods.
+	 */
+	if (!(mp->m_super->s_flags & MS_ACTIVE) &&
+	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
 		/* dgc: errors ignored here */
 		if (mp->m_super->s_frozen == SB_UNFROZEN &&
 		    xfs_log_need_covered(mp))
@@ -495,14 +503,6 @@ xfs_syncd_queue_reclaim(
 	struct xfs_mount        *mp)
 {
 
-	/*
-	 * We can have inodes enter reclaim after we've shut down the syncd
-	 * workqueue during unmount, so don't allow reclaim work to be queued
-	 * during unmount.
-	 */
-	if (!(mp->m_super->s_flags & MS_ACTIVE))
-		return;
-
 	rcu_read_lock();
 	if (radix_tree_tagged(&mp->m_perag_tree, XFS_ICI_RECLAIM_TAG)) {
 		queue_delayed_work(xfs_syncd_wq, &mp->m_reclaim_work,
@@ -571,7 +571,6 @@ xfs_syncd_init(
 	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
 
 	xfs_syncd_queue_sync(mp);
-	xfs_syncd_queue_reclaim(mp);
 
 	return 0;
 }
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 068/102] xfs: avoid taking the ilock unnessecarily in xfs_qm_dqattach

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: b4d05e3019692fc5a8c573fbce60de2d48c5b7a1

Check if we actually need to attach a dquot before taking the ilock in
xfs_qm_dqattach.  This avoid superflous lock roundtrips for the common cases
of quota support compiled in but not activated on a filesystem and an
inode that already has the dquots attached.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/quota/xfs_qm.c |   26 +++++++++++++++++++++-----
 1 file changed, 21 insertions(+), 5 deletions(-)

diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index eacf7e0..fd8ddac 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -777,6 +777,23 @@ xfs_qm_dqattach_grouphint(
 	xfs_dqunlock(udq);
 }
 
+static bool
+xfs_qm_need_dqattach(
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp))
+		return false;
+	if (!XFS_IS_QUOTA_ON(mp))
+		return false;
+	if (!XFS_NOT_DQATTACHED(mp, ip))
+		return false;
+	if (ip->i_ino == mp->m_sb.sb_uquotino ||
+	    ip->i_ino == mp->m_sb.sb_gquotino)
+		return false;
+	return true;
+}
 
 /*
  * Given a locked inode, attach dquot(s) to it, taking U/G/P-QUOTAON
@@ -794,11 +811,7 @@ xfs_qm_dqattach_locked(
 	uint		nquotas = 0;
 	int		error = 0;
 
-	if (!XFS_IS_QUOTA_RUNNING(mp) ||
-	    !XFS_IS_QUOTA_ON(mp) ||
-	    !XFS_NOT_DQATTACHED(mp, ip) ||
-	    ip->i_ino == mp->m_sb.sb_uquotino ||
-	    ip->i_ino == mp->m_sb.sb_gquotino)
+	if (!xfs_qm_need_dqattach(ip))
 		return 0;
 
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
@@ -873,6 +886,9 @@ xfs_qm_dqattach(
 {
 	int			error;
 
+	if (!xfs_qm_need_dqattach(ip))
+		return 0;
+
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	error = xfs_qm_dqattach_locked(ip, flags);
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 069/102] xfs: reduce ilock hold times in xfs_file_aio_write_checks

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 467f78992a0743e0e71729e4faa20b67b0f25289

We do not need the ilock for generic_write_checks and the i_size_read,
which are protected by i_mutex and/or iolock, so reduce the ilock
critical section to just the call to xfs_zero_eof.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |   23 +++++++++--------------
 1 file changed, 9 insertions(+), 14 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 1dbd1d8..d0239cf 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -589,35 +589,31 @@ xfs_file_aio_write_checks(
 	struct xfs_inode	*ip = XFS_I(inode);
 	int			error = 0;
 
-	xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
 restart:
 	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
-	if (error) {
-		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
+	if (error)
 		return error;
-	}
 
 	/*
 	 * If the offset is beyond the size of the file, we need to zero any
 	 * blocks that fall between the existing EOF and the start of this
 	 * write.  If zeroing is needed and we are currently holding the
-	 * iolock shared, we need to update it to exclusive which involves
-	 * dropping all locks and relocking to maintain correct locking order.
-	 * If we do this, restart the function to ensure all checks and values
-	 * are still valid.
+	 * iolock shared, we need to update it to exclusive which implies
+	 * having to redo all checks before.
 	 */
 	if (*pos > i_size_read(inode)) {
 		if (*iolock == XFS_IOLOCK_SHARED) {
-			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
+			xfs_rw_iunlock(ip, *iolock);
 			*iolock = XFS_IOLOCK_EXCL;
-			xfs_rw_ilock(ip, XFS_ILOCK_EXCL | *iolock);
+			xfs_rw_ilock(ip, *iolock);
 			goto restart;
 		}
+		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
 		error = -xfs_zero_eof(ip, *pos, i_size_read(inode));
+		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
+		if (error)
+			return error;
 	}
-	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
-	if (error)
-		return error;
 
 	/*
 	 * Updating the timestamps will grab the ilock again from
@@ -634,7 +630,6 @@ restart:
 	 * people from modifying setuid and setgid binaries.
 	 */
 	return file_remove_suid(file);
-
 }
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 070/102] xfs: reduce ilock hold times in xfs_setattr_size

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: f38996f5768713fb60e1d2de66c097367d54bb6a

We do not need the ilock for most checks done in the beginning of
xfs_setattr_size.  Replace the long critical section before starting the
transaction with a smaller one around xfs_zero_eof and an optional one
inside xfs_qm_dqattach that isn't entered unless using quotas.  While
this isn't a big optimization for xfs_setattr_size itself it will allow
pushing the ilock into xfs_zero_eof itself later.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_vnodeops.c |   13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 0e94d59..b156e46 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -61,7 +61,7 @@ xfs_setattr(
 	int			mask = iattr->ia_valid;
 	xfs_trans_t		*tp;
 	int			code;
-	uint			lock_flags;
+	uint			lock_flags = 0;
 	uint			commit_flags=0;
 	uid_t			uid=0, iuid=0;
 	gid_t			gid=0, igid=0;
@@ -126,7 +126,7 @@ xfs_setattr(
 	 */
 	tp = NULL;
 restart:
-	lock_flags = XFS_ILOCK_EXCL;
+	lock_flags = 0;
 	if (flags & XFS_ATTR_NOLOCK)
 		need_iolock = 0;
 	if (!(mask & ATTR_SIZE)) {
@@ -139,6 +139,7 @@ restart:
 			lock_flags = 0;
 			goto error_return;
 		}
+		lock_flags = XFS_ILOCK_EXCL;
 	} else {
 		if (need_iolock)
 			lock_flags |= XFS_IOLOCK_EXCL;
@@ -186,8 +187,6 @@ restart:
 		/* Short circuit the truncate case for zero length files */
 		if (iattr->ia_size == 0 &&
 		    old_size == 0 && ip->i_d.di_nextents == 0) {
-			xfs_iunlock(ip, XFS_ILOCK_EXCL);
-			lock_flags &= ~XFS_ILOCK_EXCL;
 			if (mask & ATTR_CTIME) {
 				/* need to log timestamp changes */
 				iattr->ia_ctime = iattr->ia_mtime =
@@ -213,7 +212,7 @@ restart:
 		/*
 		 * Make sure that the dquots are attached to the inode.
 		 */
-		code = xfs_qm_dqattach_locked(ip, 0);
+		code = xfs_qm_dqattach(ip, 0);
 		if (code)
 			goto error_return;
 
@@ -232,12 +231,12 @@ restart:
 			 * need to do this before the inode is joined to the
 			 * transaction to modify the i_size.
 			 */
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
 			code = xfs_zero_eof(ip, iattr->ia_size, old_size);
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
 			if (code)
 				goto error_return;
 		}
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		lock_flags &= ~XFS_ILOCK_EXCL;
 
 		/*
 		 * We are going to log the inode size change in this
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 071/102] xfs: push the ilock into xfs_zero_eof

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 193aec10504e4c24521449c46317282141fb36e8

Instead of calling xfs_zero_eof with the ilock held only take it internally
for the minimall required critical section around xfs_bmapi_read.  This
also requires changing the calling convention for xfs_zero_last_block
slightly.  The actual zeroing operation is still serialized by the iolock,
which must be taken exclusively over the call to xfs_zero_eof.

We could in fact use a shared lock for the xfs_bmapi_read calls as long as
the extent list has been read in, but given that we already hold the iolock
exclusively there is little reason to micro optimize this further.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |  164 +++++++++++++++++--------------------------
 fs/xfs/xfs_vnodeops.c       |    2 -
 2 files changed, 64 insertions(+), 102 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index d0239cf..642554b8 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -390,116 +390,99 @@ xfs_file_splice_write(
 }
 
 /*
- * This routine is called to handle zeroing any space in the last
- * block of the file that is beyond the EOF.  We do this since the
- * size is being increased without writing anything to that block
- * and we don't want anyone to read the garbage on the disk.
+ * This routine is called to handle zeroing any space in the last block of the
+ * file that is beyond the EOF.  We do this since the size is being increased
+ * without writing anything to that block and we don't want to read the
+ * garbage on the disk.
  */
 STATIC int				/* error (positive) */
 xfs_zero_last_block(
-	xfs_inode_t	*ip,
-	xfs_fsize_t	offset,
-	xfs_fsize_t	isize)
+	struct xfs_inode	*ip,
+	xfs_fsize_t		offset,
+	xfs_fsize_t		isize)
 {
-	xfs_fileoff_t	last_fsb;
-	xfs_mount_t	*mp = ip->i_mount;
-	int		nimaps;
-	int		zero_offset;
-	int		zero_len;
-	int		error = 0;
-	xfs_bmbt_irec_t	imap;
-
-	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
-
-	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.
-		 */
-		return 0;
-	}
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_fileoff_t		last_fsb = XFS_B_TO_FSBT(mp, isize);
+	int			zero_offset = XFS_B_FSB_OFFSET(mp, isize);
+	int			zero_len;
+	int			nimaps = 1;
+	int			error = 0;
+	struct xfs_bmbt_irec	imap;
 
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	last_fsb = XFS_B_TO_FSBT(mp, isize);
 	nimaps = 1;
 	error = xfs_bmapi(NULL, ip, last_fsb, 1, 0, NULL, 0, &imap,
 			  &nimaps, NULL);
-	if (error) {
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	if (error)
 		return error;
-	}
+
 	ASSERT(nimaps > 0);
+
 	/*
 	 * If the block underlying isize is just a hole, then there
 	 * is nothing to zero.
 	 */
-	if (imap.br_startblock == HOLESTARTBLOCK) {
+	if (imap.br_startblock == HOLESTARTBLOCK)
 		return 0;
-	}
-	/*
-	 * Zero the part of the last block beyond the EOF, and write it
-	 * out sync.  We need to drop the ilock while we do this so we
-	 * don't deadlock when the buffer cache calls back to us.
-	 */
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 
 	zero_len = mp->m_sb.sb_blocksize - zero_offset;
 	if (isize + zero_len > offset)
 		zero_len = offset - isize;
-	error = xfs_iozero(ip, isize, zero_len);
-
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	ASSERT(error >= 0);
-	return error;
+	return xfs_iozero(ip, isize, zero_len);
 }
 
 /*
- * Zero any on disk space between the current EOF and the new,
- * larger EOF.  This handles the normal case of zeroing the remainder
- * of the last block in the file and the unusual case of zeroing blocks
- * out beyond the size of the file.  This second case only happens
- * with fixed size extents and when the system crashes before the inode
- * size was updated but after blocks were allocated.  If fill is set,
- * then any holes in the range are filled and zeroed.  If not, the holes
- * are left alone as holes.
+ * Zero any on disk space between the current EOF and the new, larger EOF.
+ *
+ * This handles the normal case of zeroing the remainder of the last block in
+ * the file and the unusual case of zeroing blocks out beyond the size of the
+ * file.  This second case only happens with fixed size extents and when the
+ * system crashes before the inode size was updated but after blocks were
+ * allocated.
+ *
+ * Expects the iolock to be held exclusive, and will take the ilock internally.
  */
-
 int					/* error (positive) */
 xfs_zero_eof(
-	xfs_inode_t	*ip,
-	xfs_off_t	offset,		/* starting I/O offset */
-	xfs_fsize_t	isize)		/* current inode size */
+	struct xfs_inode	*ip,
+	xfs_off_t		offset,		/* starting I/O offset */
+	xfs_fsize_t		isize)		/* current inode size */
 {
-	xfs_mount_t	*mp = ip->i_mount;
-	xfs_fileoff_t	start_zero_fsb;
-	xfs_fileoff_t	end_zero_fsb;
-	xfs_fileoff_t	zero_count_fsb;
-	xfs_fileoff_t	last_fsb;
-	xfs_fileoff_t	zero_off;
-	xfs_fsize_t	zero_len;
-	int		nimaps;
-	int		error = 0;
-	xfs_bmbt_irec_t	imap;
-
-	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_fileoff_t		start_zero_fsb;
+	xfs_fileoff_t		end_zero_fsb;
+	xfs_fileoff_t		zero_count_fsb;
+	xfs_fileoff_t		last_fsb;
+	xfs_fileoff_t		zero_off;
+	xfs_fsize_t		zero_len;
+	int			nimaps;
+	int			error = 0;
+	struct xfs_bmbt_irec	imap;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
 	ASSERT(offset > isize);
 
 	/*
 	 * 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, offset, isize);
-	if (error) {
-		ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
-		return error;
+	if (XFS_B_FSB_OFFSET(mp, isize) != 0) {
+		error = xfs_zero_last_block(ip, offset, isize);
+		if (error)
+			return error;
 	}
 
 	/*
-	 * Calculate the range between the new size and the old
-	 * where blocks needing to be zeroed may exist.  To get the
-	 * block where the last byte in the file currently resides,
-	 * we need to subtract one from the size and truncate back
-	 * to a block boundary.  We subtract 1 in case the size is
-	 * exactly on a block boundary.
+	 * Calculate the range between the new size and the old where blocks
+	 * needing to be zeroed may exist.
+	 *
+	 * To get the block where the last byte in the file currently resides,
+	 * we need to subtract one from the size and truncate back to a block
+	 * boundary.  We subtract 1 in case the size is exactly on a block
+	 * boundary.
 	 */
 	last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1;
 	start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
@@ -517,23 +500,18 @@ xfs_zero_eof(
 	while (start_zero_fsb <= end_zero_fsb) {
 		nimaps = 1;
 		zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
 		error = xfs_bmapi(NULL, ip, start_zero_fsb, zero_count_fsb,
 				  0, NULL, 0, &imap, &nimaps, NULL);
-		if (error) {
-			ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		if (error)
 			return error;
-		}
+
 		ASSERT(nimaps > 0);
 
 		if (imap.br_state == XFS_EXT_UNWRITTEN ||
 		    imap.br_startblock == HOLESTARTBLOCK) {
-			/*
-			 * This loop handles initializing pages that were
-			 * partially initialized by the code below this
-			 * loop. It basically zeroes the part of the page
-			 * that sits on a hole and sets the page as P_HOLE
-			 * and calls remapf if it is a mapped file.
-			 */
 			start_zero_fsb = imap.br_startoff + imap.br_blockcount;
 			ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
 			continue;
@@ -541,11 +519,7 @@ xfs_zero_eof(
 
 		/*
 		 * There are blocks we need to zero.
-		 * Drop the inode lock while we're doing the I/O.
-		 * We'll still have the iolock to protect us.
 		 */
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-
 		zero_off = XFS_FSB_TO_B(mp, start_zero_fsb);
 		zero_len = XFS_FSB_TO_B(mp, imap.br_blockcount);
 
@@ -553,22 +527,14 @@ xfs_zero_eof(
 			zero_len = offset - zero_off;
 
 		error = xfs_iozero(ip, zero_off, zero_len);
-		if (error) {
-			goto out_lock;
-		}
+		if (error)
+			return error;
 
 		start_zero_fsb = imap.br_startoff + imap.br_blockcount;
 		ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
-
-		xfs_ilock(ip, XFS_ILOCK_EXCL);
 	}
 
 	return 0;
-
-out_lock:
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	ASSERT(error >= 0);
-	return error;
 }
 
 /*
@@ -608,9 +574,7 @@ restart:
 			xfs_rw_ilock(ip, *iolock);
 			goto restart;
 		}
-		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
 		error = -xfs_zero_eof(ip, *pos, i_size_read(inode));
-		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
 		if (error)
 			return error;
 	}
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index b156e46..3988584 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -231,9 +231,7 @@ restart:
 			 * need to do this before the inode is joined to the
 			 * transaction to modify the i_size.
 			 */
-			xfs_ilock(ip, XFS_ILOCK_EXCL);
 			code = xfs_zero_eof(ip, iattr->ia_size, old_size);
-			xfs_iunlock(ip, XFS_ILOCK_EXCL);
 			if (code)
 				goto error_return;
 		}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 072/102] xfs: use shared ilock mode for direct IO writes by default

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 507630b29f13a3d8689895618b12015308402e22

For the direct IO write path, we only really need the ilock to be taken in
exclusive mode during IO submission if we need to do extent allocation
instead of all the time.

Change the block mapping code to take the ilock in shared mode for the
initial block mapping, and only retake it exclusively when we actually
have to perform extent allocations.  We were already dropping the ilock
for the transaction allocation, so this doesn't introduce new race windows.

Based on an earlier patch from Dave Chinner.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   30 +++++++++++++++++++++++++----
 fs/xfs/xfs_iomap.c          |   45 ++++++++++++++++++-------------------------
 2 files changed, 45 insertions(+), 30 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 390e511..2ce0e07 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1194,7 +1194,14 @@ __xfs_get_blocks(
 	if (!create && direct && offset >= i_size_read(inode))
 		return 0;
 
-	if (create) {
+	/*
+	 * Direct I/O is usually done on preallocated files, so try getting
+	 * a block mapping without an exclusive lock first.  For buffered
+	 * writes we already have the exclusive iolock anyway, so avoiding
+	 * a lock roundtrip here by taking the ilock exclusive from the
+	 * beginning is a useful micro optimization.
+	 */
+	if (create && !direct) {
 		lockmode = XFS_ILOCK_EXCL;
 		xfs_ilock(ip, lockmode);
 	} else {
@@ -1217,22 +1224,37 @@ __xfs_get_blocks(
 	     (imap.br_startblock == HOLESTARTBLOCK ||
 	      imap.br_startblock == DELAYSTARTBLOCK))) {
 		if (direct) {
+			/*
+			 * Drop the ilock in preparation for starting the block
+			 * allocation transaction.  It will be retaken
+			 * exclusively inside xfs_iomap_write_direct for the
+			 * actual allocation.
+			 */
+			xfs_iunlock(ip, lockmode);
 			error = xfs_iomap_write_direct(ip, offset, size,
 						       &imap, nimaps);
+			if (error)
+				return -error;
 		} else {
+			/*
+			 * Delalloc reservations do not require a transaction,
+			 * we can go on without dropping the lock here.
+			 */
 			error = xfs_iomap_write_delay(ip, offset, size, &imap);
+			if (error)
+				goto out_unlock;
+
+			xfs_iunlock(ip, lockmode);
 		}
-		if (error)
-			goto out_unlock;
 
 		trace_xfs_get_blocks_alloc(ip, offset, size, 0, &imap);
 	} else if (nimaps) {
 		trace_xfs_get_blocks_found(ip, offset, size, 0, &imap);
+		xfs_iunlock(ip, lockmode);
 	} else {
 		trace_xfs_get_blocks_notfound(ip, offset, size);
 		goto out_unlock;
 	}
-	xfs_iunlock(ip, lockmode);
 
 	if (imap.br_startblock != HOLESTARTBLOCK &&
 	    imap.br_startblock != DELAYSTARTBLOCK) {
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 9ee9467..29e2d85 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -142,11 +142,7 @@ xfs_iomap_write_direct(
 	int		committed;
 	int		error;
 
-	/*
-	 * Make sure that the dquots are there. This doesn't hold
-	 * the ilock across a disk read.
-	 */
-	error = xfs_qm_dqattach_locked(ip, 0);
+	error = xfs_qm_dqattach(ip, 0);
 	if (error)
 		return XFS_ERROR(error);
 
@@ -158,7 +154,7 @@ xfs_iomap_write_direct(
 	if ((offset + count) > XFS_ISIZE(ip)) {
 		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
 		if (error)
-			goto error_out;
+			return XFS_ERROR(error);
 	} else {
 		if (nmaps && (imap->br_startblock == HOLESTARTBLOCK))
 			last_fsb = MIN(last_fsb, (xfs_fileoff_t)
@@ -190,7 +186,6 @@ xfs_iomap_write_direct(
 	/*
 	 * Allocate and setup the transaction
 	 */
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
 	error = xfs_trans_reserve(tp, resblks,
 			XFS_WRITE_LOG_RES(mp), resrtextents,
@@ -199,15 +194,16 @@ xfs_iomap_write_direct(
 	/*
 	 * Check for running out of space, note: need lock to return
 	 */
-	if (error)
+	if (error) {
 		xfs_trans_cancel(tp, 0);
+		return XFS_ERROR(error);
+	}
+
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	if (error)
-		goto error_out;
 
 	error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
 	if (error)
-		goto error1;
+		goto out_trans_cancel;
 
 	xfs_trans_ijoin(tp, ip);
 
@@ -226,42 +222,39 @@ xfs_iomap_write_direct(
 	error = xfs_bmapi(tp, ip, offset_fsb, count_fsb, bmapi_flag,
 		&firstfsb, 0, imap, &nimaps, &free_list);
 	if (error)
-		goto error0;
+		goto out_bmap_cancel;
 
 	/*
 	 * Complete the transaction
 	 */
 	error = xfs_bmap_finish(&tp, &free_list, &committed);
 	if (error)
-		goto error0;
+		goto out_bmap_cancel;
 	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
 	if (error)
-		goto error_out;
+		goto out_unlock;
 
 	/*
 	 * Copy any maps to caller's array and return any error.
 	 */
 	if (nimaps == 0) {
-		error = ENOSPC;
-		goto error_out;
+		error = XFS_ERROR(ENOSPC);
+		goto out_unlock;
 	}
 
-	if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip))) {
+	if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip)))
 		error = xfs_alert_fsblock_zero(ip, imap);
-		goto error_out;
-	}
 
-	return 0;
+out_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
 
-error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
+out_bmap_cancel:
 	xfs_bmap_cancel(&free_list);
 	xfs_trans_unreserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
-
-error1:	/* Just cancel transaction */
+out_trans_cancel:
 	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
-
-error_out:
-	return XFS_ERROR(error);
+	goto out_unlock;
 }
 
 /*
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 073/102] xfs: punch all delalloc blocks beyond EOF on write failure.

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 01c84d2dc1311fb76ea217dadfd5b3a5f3cab563

I've been seeing regular ASSERT failures in xfstests when running
fsstress based tests over the past month. xfs_getbmap() has been
failing this test:

XFS: Assertion failed: ((iflags & BMV_IF_DELALLOC) != 0) ||
(map[i].br_startblock != DELAYSTARTBLOCK), file: fs/xfs/xfs_bmap.c,
line: 5650

where it is encountering a delayed allocation extent after writing
all the dirty data to disk and then walking the extent map
atomically by holding the XFS_IOLOCK_SHARED to prevent new delayed
allocation extents from being created.

Test 083 on a 512 byte block size filesystem was used to reproduce
the problem, because it only had a 5s run timeand would usually fail
every 3-4 runs. This test is exercising ENOSPC behaviour by running
fsstress on a nearly full filesystem. The following trace extract
shows the final few events on the inode that tripped the assert:

 xfs_ilock:             flags ILOCK_EXCL caller xfs_setfilesize
 xfs_setfilesize:       isize 0x180000 disize 0x12d400 offset 0x17e200 count 7680

file size updated to 0x180000 by IO completion

 xfs_ilock:             flags ILOCK_EXCL caller xfs_iomap_write_delay
 xfs_iext_insert:       state  idx 3 offset 3072 block 4503599627239432 count 1 flag 0 caller xfs_bmap_add_extent_hole_delay
 xfs_get_blocks_alloc:  size 0x180000 offset 0x180000 count 512 type  startoff 0xc00 startblock -1 blockcount 0x1
 xfs_ilock:             flags ILOCK_EXCL caller __xfs_get_blocks

delalloc write, adding a single block at offset 0x180000

 xfs_delalloc_enospc:   isize 0x180000 disize 0x180000 offset 0x180200 count 512

ENOSPC trying to allocate a dellalloc block at offset 0x180200

 xfs_ilock:             flags ILOCK_EXCL caller xfs_iomap_write_delay
 xfs_get_blocks_alloc:  size 0x180000 offset 0x180200 count 512 type  startoff 0xc00 startblock -1 blockcount 0x2

And succeeding on retry after flushing dirty inodes.

 xfs_ilock:             flags ILOCK_EXCL caller __xfs_get_blocks
 xfs_delalloc_enospc:   isize 0x180000 disize 0x180000 offset 0x180400 count 512

ENOSPC trying to allocate a dellalloc block at offset 0x180400

 xfs_ilock:             flags ILOCK_EXCL caller xfs_iomap_write_delay
 xfs_delalloc_enospc:   isize 0x180000 disize 0x180000 offset 0x180400 count 512

And failing the retry, giving a real ENOSPC error.

 xfs_ilock:             flags ILOCK_EXCL caller xfs_vm_write_failed
                                                ^^^^^^^^^^^^^^^^^^^
The smoking gun - the write being failed and cleaning up delalloc
blocks beyond EOF allocated by the failed write.

 xfs_getattr:
 xfs_ilock:             flags IOLOCK_SHARED caller xfs_getbmap
 xfs_ilock:             flags ILOCK_SHARED caller xfs_ilock_map_shared

And that's where we died almost immediately afterwards.
xfs_bmapi_read() found delalloc extent beyond current file in memory
file size. Some debug I added to xfs_getbmap() showed the state just
before the assert failure:

 ino 0x80e48: off 0xc00, fsb 0xffffffffffffffff, len 0x1, size 0x180000
 start_fsb 0x106, end_fsb 0x638
 ino flags 0x2 nex 0xd bmvcnt 0x555, len 0x3c58a6f23c0bf1, start 0xc00
 ext 0: off 0x1fc, fsb 0x24782, len 0x254
 ext 1: off 0x450, fsb 0x40851, len 0x30
 ext 2: off 0x480, fsb 0xd99, len 0x1b8
 ext 3: off 0x92f, fsb 0x4099a, len 0x3b
 ext 4: off 0x96d, fsb 0x41844, len 0x98
 ext 5: off 0xbf1, fsb 0x408ab, len 0xf

which shows that we found a single delalloc block beyond EOF (first
line of output) when we were returning the map for a length
somewhere around 10^16 bytes long (second line), and the on-disk
extents showed they didn't go past EOF (last lines).

Further debug added to xfs_vm_write_failed() showed this happened
when punching out delalloc blocks beyond the end of the file after
the failed write:

[  132.606693] ino 0x80e48: vwf to 0x181000, sze 0x180000
[  132.609573] start_fsb 0xc01, end_fsb 0xc08

It punched the range 0xc01 -> 0xc08, but the range we really need to
punch is 0xc00 -> 0xc07 (8 blocks from 0xc00) as this testing was
run on a 512 byte block size filesystem (8 blocks per page).
the punch from is 0xc00. So end_fsb is correct, but start_fsb is
wrong as we punch from start_fsb for (end_fsb - start_fsb) blocks.
Hence we are not punching the delalloc block beyond EOF in the case.

The fix is simple - it's a silly off-by-one mistake in calculating
the range. It's especially silly because the macro used to calculate
the start_fsb already takes into account the case where the inode
size is an exact multiple of the filesystem block size...

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 2ce0e07..d2d3c84 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1482,7 +1482,7 @@ xfs_vm_write_failed(
 		 * Check if there are any blocks that are outside of i_size
 		 * that need to be trimmed back.
 		 */
-		start_fsb = XFS_B_TO_FSB(ip->i_mount, inode->i_size) + 1;
+		start_fsb = XFS_B_TO_FSB(ip->i_mount, inode->i_size);
 		end_fsb = XFS_B_TO_FSB(ip->i_mount, to);
 		if (end_fsb <= start_fsb)
 			return;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 074/102] xfs: using GFP_NOFS for blkdev_issue_flush

[Dave Chinner]

From: Shaohua Li <shli@kernel.org>

Upstream commit: 7582df516c93046b8d2111a780c69de77f9882fb

Issuing a block device flush request in transaction context using GFP_KERNEL
directly can cause deadlocks due to memory reclaim recursion. Use GFP_NOFS to
avoid recursion from reclaim context.

Signed-off-by: Shaohua Li <shli@fusionio.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_super.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 9c7f4b9..2cf0967 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -631,7 +631,7 @@ void
 xfs_blkdev_issue_flush(
 	xfs_buftarg_t		*buftarg)
 {
-	blkdev_issue_flush(buftarg->bt_bdev, GFP_KERNEL, NULL);
+	blkdev_issue_flush(buftarg->bt_bdev, GFP_NOFS, NULL);
 }
 
 STATIC void
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 075/102] xfs: page type check in writeback only checks last buffer

[Dave Chinner]

Upstream commit: 6ffc4db5de61d36e969a26bc94509c59246c81f8

xfs_is_delayed_page() checks to see if a page has buffers matching
the given IO type passed in. It does so by walking the buffer heads
on the page and checking if the state flags match the IO type.

However, the "acceptable" variable that is calculated is overwritten
every time a new buffer is checked. Hence if the first buffer on the
page is of the right type, this state is lost if the second buffer
is not of the correct type. This means that xfs_aops_discard_page()
may not discard delalloc regions when it is supposed to, and
xfs_convert_page() may not cluster IO as efficiently as possible.

This problem only occurs on filesystems with a block size smaller
than page size.

Also, rename xfs_is_delayed_page() to xfs_check_page_type() to
better describe what it is doing - it is not delalloc specific
anymore.

The problem was first noticed by Peter Watkins.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index d2d3c84..51f7db1 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -660,7 +660,7 @@ xfs_map_at_offset(
  * or delayed allocate extent.
  */
 STATIC int
-xfs_is_delayed_page(
+xfs_check_page_type(
 	struct page		*page,
 	unsigned int		type)
 {
@@ -674,11 +674,11 @@ xfs_is_delayed_page(
 		bh = head = page_buffers(page);
 		do {
 			if (buffer_unwritten(bh))
-				acceptable = (type == IO_UNWRITTEN);
+				acceptable += (type == IO_UNWRITTEN);
 			else if (buffer_delay(bh))
-				acceptable = (type == IO_DELALLOC);
+				acceptable += (type == IO_DELALLOC);
 			else if (buffer_dirty(bh) && buffer_mapped(bh))
-				acceptable = (type == IO_OVERWRITE);
+				acceptable += (type == IO_OVERWRITE);
 			else
 				break;
 		} while ((bh = bh->b_this_page) != head);
@@ -721,7 +721,7 @@ xfs_convert_page(
 		goto fail_unlock_page;
 	if (page->mapping != inode->i_mapping)
 		goto fail_unlock_page;
-	if (!xfs_is_delayed_page(page, (*ioendp)->io_type))
+	if (!xfs_check_page_type(page, (*ioendp)->io_type))
 		goto fail_unlock_page;
 
 	/*
@@ -871,7 +871,7 @@ xfs_aops_discard_page(
 	struct buffer_head	*bh, *head;
 	loff_t			offset = page_offset(page);
 
-	if (!xfs_is_delayed_page(page, IO_DELALLOC))
+	if (!xfs_check_page_type(page, IO_DELALLOC))
 		goto out_invalidate;
 
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 076/102] xfs: punch new delalloc blocks out of failed writes inside

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: d3bc815afb549eecb3679a4b2f0df216e34df998
 EOF.

When a partial write inside EOF fails, it can leave delayed
allocation blocks lying around because they don't get punched back
out. This leads to assert failures like:

XFS: Assertion failed: XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0, file: fs/xfs/linux-2.6/xfs_super.c, line: 847

when evicting inodes from the cache. This can be trivially triggered
by xfstests 083, which takes between 5 and 15 executions on a 512
byte block size filesystem to trip over this. Debugging shows a
failed write due to ENOSPC calling xfs_vm_write_failed such as:

[ 5012.329024] ino 0xa0026: vwf to 0x17000, sze 0x1c85ae

and no action is taken on it. This leaves behind a delayed
allocation extent that has no page covering it and no data in it:

[ 5015.867162] ino 0xa0026: blks: 0x83 delay blocks 0x1, size 0x2538c0
[ 5015.868293] ext 0: off 0x4a, fsb 0x50306, len 0x1
[ 5015.869095] ext 1: off 0x4b, fsb 0x7899, len 0x6b
[ 5015.869900] ext 2: off 0xb6, fsb 0xffffffffe0008, len 0x1
                                    ^^^^^^^^^^^^^^^
[ 5015.871027] ext 3: off 0x36e, fsb 0x7a27, len 0xd
[ 5015.872206] ext 4: off 0x4cf, fsb 0x7a1d, len 0xa

So the delayed allocation extent is one block long at offset
0x16c00. Tracing shows that a bigger write:

xfs_file_buffered_write: size 0x1c85ae offset 0x959d count 0x1ca3f ioflags

allocates the block, and then fails with ENOSPC trying to allocate
the last block on the page, leading to a failed write with stale
delalloc blocks on it.

Because we've had an ENOSPC when trying to allocate 0x16e00, it
means that we are never goinge to call ->write_end on the page and
so the allocated new buffer will not get marked dirty or have the
buffer_new state cleared. In other works, what the above write is
supposed to end up with is this mapping for the page:

    +------+------+------+------+------+------+------+------+
      UMA    UMA    UMA    UMA    UMA    UMA    UND    FAIL

where:  U = uptodate
        M = mapped
        N = new
        A = allocated
        D = delalloc
        FAIL = block we ENOSPC'd on.

and the key point being the buffer_new() state for the newly
allocated delayed allocation block. Except it doesn't - we're not
marking buffers new correctly.

That buffer_new() problem goes back to the xfs_iomap removal days,
where xfs_iomap() used to return a "new" status for any map with
newly allocated blocks, so that __xfs_get_blocks() could call
set_buffer_new() on it. We still have the "new" variable and the
check for it in the set_buffer_new() logic - except we never set it
now!

Hence that newly allocated delalloc block doesn't have the new flag
set on it, so when the write fails we cannot tell which blocks we
are supposed to punch out. WHy do we need the buffer_new flag? Well,
that's because we can have this case:

    +------+------+------+------+------+------+------+------+
      UMD    UMD    UMD    UMD    UMD    UMD    UND    FAIL

where all the UMD buffers contain valid data from a previously
successful write() system call. We only want to punch the UND buffer
because that's the only one that we added in this write and it was
only this write that failed.

That implies that even the old buffer_new() logic was wrong -
because it would result in all those UMD buffers on the page having
set_buffer_new() called on them even though they aren't new. Hence
we shoul donly be calling set_buffer_new() for delalloc buffers that
were allocated (i.e. were a hole before xfs_iomap_write_delay() was
called).

So, fix this set_buffer_new logic according to how we need it to
work for handling failed writes correctly. Also, restore the new
buffer logic handling for blocks allocated via
xfs_iomap_write_direct(), because it should still set the buffer_new
flag appropriately for newly allocated blocks, too.

SO, now we have the buffer_new() being set appropriately in
__xfs_get_blocks(), we can detect the exact delalloc ranges that
we allocated in a failed write, and hence can now do a walk of the
buffers on a page to find them.

Except, it's not that easy. When block_write_begin() fails, it
unlocks and releases the page that we just had an error on, so we
can't use that page to handle errors anymore. We have to get access
to the page while it is still locked to walk the buffers. Hence we
have to open code block_write_begin() in xfs_vm_write_begin() to be
able to insert xfs_vm_write_failed() is the right place.

With that, we can pass the page and write range to
xfs_vm_write_failed() and walk the buffers on the page, looking for
delalloc buffers that are either new or beyond EOF and punch them
out. Handling buffers beyond EOF ensures we still handle the
existing case that xfs_vm_write_failed() handles.

Of special note is the truncate_pagecache() handling - that only
should be done for pages outside EOF - pages within EOF can still
contain valid, dirty data so we must not punch them out of the
cache.

That just leaves the xfs_vm_write_end() failure handling.
The only failure case here is that we didn't copy the entire range,
and generic_write_end() handles that by zeroing the region of the
page that wasn't copied, we don't have to punch out blocks within
the file because they are guaranteed to contain zeros. Hence we only
have to handle the existing "beyond EOF" case and don't need access
to the buffers on the page. Hence it remains largely unchanged.

Note that xfs_getbmap() can still trip over delalloc blocks beyond
EOF that are left there by speculative delayed allocation. Hence
this bug fix does not solve all known issues with bmap vs delalloc,
but it does fix all the the known accidental occurances of the
problem.

Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |  173 +++++++++++++++++++++++++++++++------------
 1 file changed, 127 insertions(+), 46 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 51f7db1..e4acb6b 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1235,11 +1235,18 @@ __xfs_get_blocks(
 						       &imap, nimaps);
 			if (error)
 				return -error;
+			new = 1;
 		} else {
 			/*
 			 * Delalloc reservations do not require a transaction,
-			 * we can go on without dropping the lock here.
+			 * we can go on without dropping the lock here. If we
+			 * are allocating a new delalloc block, make sure that
+			 * we set the new flag so that we mark the buffer new so
+			 * that we know that it is newly allocated if the write
+			 * fails.
 			 */
+			if (nimaps && imap.br_startblock == HOLESTARTBLOCK)
+				new = 1;
 			error = xfs_iomap_write_delay(ip, offset, size, &imap);
 			if (error)
 				goto out_unlock;
@@ -1456,52 +1463,91 @@ out_destroy_ioend:
 	return ret;
 }
 
+/*
+ * Punch out the delalloc blocks we have already allocated.
+ *
+ * Don't bother with xfs_setattr given that nothing can have made it to disk yet
+ * as the page is still locked at this point.
+ */
+STATIC void
+xfs_vm_kill_delalloc_range(
+	struct inode		*inode,
+	loff_t			start,
+	loff_t			end)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	xfs_fileoff_t		start_fsb;
+	xfs_fileoff_t		end_fsb;
+	int			error;
+
+	start_fsb = XFS_B_TO_FSB(ip->i_mount, start);
+	end_fsb = XFS_B_TO_FSB(ip->i_mount, end);
+	if (end_fsb <= start_fsb)
+		return;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	error = xfs_bmap_punch_delalloc_range(ip, start_fsb,
+						end_fsb - start_fsb);
+	if (error) {
+		/* something screwed, just bail */
+		if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+			xfs_alert(ip->i_mount,
+		"xfs_vm_write_failed: unable to clean up ino %lld",
+					ip->i_ino);
+		}
+	}
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+}
+
 STATIC void
 xfs_vm_write_failed(
-	struct address_space	*mapping,
-	loff_t			to)
+	struct inode		*inode,
+	struct page		*page,
+	loff_t			pos,
+	unsigned		len)
 {
-	struct inode		*inode = mapping->host;
+	loff_t			block_offset = pos & PAGE_MASK;
+	loff_t			block_start;
+	loff_t			block_end;
+	loff_t			from = pos & (PAGE_CACHE_SIZE - 1);
+	loff_t			to = from + len;
+	struct buffer_head	*bh, *head;
 
-	if (to > inode->i_size) {
-		/*
-		 * Punch out the delalloc blocks we have already allocated.
-		 *
-		 * Don't bother with xfs_setattr given that nothing can have
-		 * made it to disk yet as the page is still locked at this
-		 * point.
-		 */
-		struct xfs_inode	*ip = XFS_I(inode);
-		xfs_fileoff_t		start_fsb;
-		xfs_fileoff_t		end_fsb;
-		int			error;
+	ASSERT(block_offset + from == pos);
 
-		truncate_pagecache(inode, to, inode->i_size);
+	head = page_buffers(page);
+	block_start = 0;
+	for (bh = head; bh != head || !block_start;
+	     bh = bh->b_this_page, block_start = block_end,
+				   block_offset += bh->b_size) {
+		block_end = block_start + bh->b_size;
 
-		/*
-		 * Check if there are any blocks that are outside of i_size
-		 * that need to be trimmed back.
-		 */
-		start_fsb = XFS_B_TO_FSB(ip->i_mount, inode->i_size);
-		end_fsb = XFS_B_TO_FSB(ip->i_mount, to);
-		if (end_fsb <= start_fsb)
-			return;
-
-		xfs_ilock(ip, XFS_ILOCK_EXCL);
-		error = xfs_bmap_punch_delalloc_range(ip, start_fsb,
-							end_fsb - start_fsb);
-		if (error) {
-			/* something screwed, just bail */
-			if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-				xfs_alert(ip->i_mount,
-			"xfs_vm_write_failed: unable to clean up ino %lld",
-						ip->i_ino);
-			}
-		}
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		/* skip buffers before the write */
+		if (block_end <= from)
+			continue;
+
+		/* if the buffer is after the write, we're done */
+		if (block_start >= to)
+			break;
+
+		if (!buffer_delay(bh))
+			continue;
+
+		if (!buffer_new(bh) && block_offset < i_size_read(inode))
+			continue;
+
+		xfs_vm_kill_delalloc_range(inode, block_offset,
+					   block_offset + bh->b_size);
 	}
+
 }
 
+/*
+ * This used to call block_write_begin(), but it unlocks and releases the page
+ * on error, and we need that page to be able to punch stale delalloc blocks out
+ * on failure. hence we copy-n-waste it here and call xfs_vm_write_failed() at
+ * the appropriate point.
+ */
 STATIC int
 xfs_vm_write_begin(
 	struct file		*file,
@@ -1512,15 +1558,40 @@ xfs_vm_write_begin(
 	struct page		**pagep,
 	void			**fsdata)
 {
-	int			ret;
+	pgoff_t			index = pos >> PAGE_CACHE_SHIFT;
+	struct page		*page;
+	int			status;
 
-	ret = block_write_begin(mapping, pos, len, flags | AOP_FLAG_NOFS,
-				pagep, xfs_get_blocks);
-	if (unlikely(ret))
-		xfs_vm_write_failed(mapping, pos + len);
-	return ret;
+	ASSERT(len <= PAGE_CACHE_SIZE);
+
+	page = grab_cache_page_write_begin(mapping, index,
+					   flags | AOP_FLAG_NOFS);
+	if (!page)
+		return -ENOMEM;
+
+	status = __block_write_begin(page, pos, len, xfs_get_blocks);
+	if (unlikely(status)) {
+		struct inode	*inode = mapping->host;
+
+		xfs_vm_write_failed(inode, page, pos, len);
+		unlock_page(page);
+
+		if (pos + len > i_size_read(inode))
+			truncate_pagecache(inode, pos + len, i_size_read(inode));
+
+		page_cache_release(page);
+		page = NULL;
+	}
+
+	*pagep = page;
+	return status;
 }
 
+/*
+ * On failure, we only need to kill delalloc blocks beyond EOF because they
+ * will never be written. For blocks within EOF, generic_write_end() zeros them
+ * so they are safe to leave alone and be written with all the other valid data.
+ */
 STATIC int
 xfs_vm_write_end(
 	struct file		*file,
@@ -1533,9 +1604,19 @@ xfs_vm_write_end(
 {
 	int			ret;
 
+	ASSERT(len <= PAGE_CACHE_SIZE);
+
 	ret = generic_write_end(file, mapping, pos, len, copied, page, fsdata);
-	if (unlikely(ret < len))
-		xfs_vm_write_failed(mapping, pos + len);
+	if (unlikely(ret < len)) {
+		struct inode	*inode = mapping->host;
+		size_t		isize = i_size_read(inode);
+		loff_t		to = pos + len;
+
+		if (to > isize) {
+			truncate_pagecache(inode, to, isize);
+			xfs_vm_kill_delalloc_range(inode, isize, to);
+		}
+	}
 	return ret;
 }
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 077/102] xfs: prevent needless mount warning causing test failures

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 81158e0cecdf53b1f6d88a514c6c20e0ee18ec7b

Often mounting small filesystem with small logs will emit a warning
such as:

XFS (vdb): Invalid block length (0x2000) for buffer

during log recovery. This causes tests to randomly fail because this
output causes the clean filesystem checks on test completion to
think the filesystem is inconsistent.

The cause of the error is simply that log recovery is asking for a
buffer size that is larger than the log when zeroing the tail. This
is because the buffer size is rounded up, and if the right head and
tail conditions exist then the buffer size can be larger than the log.
Limit the variable size xlog_get_bp() callers to requesting buffers
smaller than the log.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_log_recover.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 687bc9f..82f7d18 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -439,6 +439,8 @@ xlog_find_verify_cycle(
 	 * a log sector, or we're out of luck.
 	 */
 	bufblks = 1 << ffs(nbblks);
+	while (bufblks > log->l_logBBsize)
+		bufblks >>= 1;
 	while (!(bp = xlog_get_bp(log, bufblks))) {
 		bufblks >>= 1;
 		if (bufblks < log->l_sectBBsize)
@@ -1224,6 +1226,8 @@ xlog_write_log_records(
 	 * log sector, or we're out of luck.
 	 */
 	bufblks = 1 << ffs(blocks);
+	while (bufblks > log->l_logBBsize)
+		bufblks >>= 1;
 	while (!(bp = xlog_get_bp(log, bufblks))) {
 		bufblks >>= 1;
 		if (bufblks < sectbb)
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 078/102] xfs: don't assert on delalloc regions beyond EOF

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 58e20770646932fe9b758c94e8c278ea9ec93878

When we are doing speculative delayed allocation beyond EOF,
conversion of the region allocated beyond EOF is dependent on the
largest free space extent available. If the largest free extent is
smaller than the delalloc range, then after allocation we leave
a delalloc extent that starts beyond EOF. This extent cannot *ever*
be converted by flushing data, and so will remain there until either
the EOF moves into the extent or it is truncated away.

Hence if xfs_getbmap() runs on such an inode and is asked to return
extents beyond EOF, it will assert fail on this extent even though
there is nothing xfs_getbmap() can do to convert it to a real
extent. Hence we should simply report these delalloc extents rather
than assert that there should be none.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_bmap.c |   16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 8f3b730..2b4e8d2 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5541,8 +5541,20 @@ xfs_getbmap(
 				XFS_FSB_TO_BB(mp, map[i].br_blockcount);
 			out[cur_ext].bmv_unused1 = 0;
 			out[cur_ext].bmv_unused2 = 0;
-			ASSERT(((iflags & BMV_IF_DELALLOC) != 0) ||
-			      (map[i].br_startblock != DELAYSTARTBLOCK));
+
+			/*
+			 * delayed allocation extents that start beyond EOF can
+			 * occur due to speculative EOF allocation when the
+			 * delalloc extent is larger than the largest freespace
+			 * extent at conversion time. These extents cannot be
+			 * converted by data writeback, so can exist here even
+			 * if we are not supposed to be finding delalloc
+			 * extents.
+			 */
+			if (map[i].br_startblock == DELAYSTARTBLOCK &&
+			    map[i].br_startoff <= XFS_B_TO_FSB(mp, XFS_ISIZE(ip)))
+				ASSERT((iflags & BMV_IF_DELALLOC) != 0);
+
                         if (map[i].br_startblock == HOLESTARTBLOCK &&
 			    whichfork == XFS_ATTR_FORK) {
 				/* came to the end of attribute fork */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 079/102] xfs: limit specualtive delalloc to maxioffset

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 3ed9116e8a3e9c0870b2076340b3da9b8f900f3b

Speculative delayed allocation beyond EOF near the maximum supported
file offset can result in creating delalloc extents beyond
mp->m_maxioffset (8EB). These can never be trimmed during
xfs_free_eof_blocks() because they are beyond mp->m_maxioffset, and
that results in assert failures in xfs_fs_destroy_inode() due to
delalloc blocks still being present. xfstests 071 exposes this
problem.

Limit speculative delalloc to mp->m_maxioffset to avoid this
problem.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_iomap.c |    9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 29e2d85..121309f 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -418,6 +418,15 @@ retry:
 			return error;
 	}
 
+	/*
+	 * Make sure preallocation does not create extents beyond the range we
+	 * actually support in this filesystem.
+	 */
+	if (last_fsb > XFS_B_TO_FSB(mp, mp->m_maxioffset))
+		last_fsb = XFS_B_TO_FSB(mp, mp->m_maxioffset);
+
+	ASSERT(last_fsb > offset_fsb);
+
 	nimaps = XFS_WRITE_IMAPS;
 	firstblock = NULLFSBLOCK;
 	error = xfs_bmapi(NULL, ip, offset_fsb,
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 080/102] xfs: Use preallocation for inodes with extsz hints

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: aff3a9edb7080f69f07fe76a8bd089b3dfa4cb5d

xfstest 229 exposes a problem with buffered IO, delayed allocation
and extent size hints. That is when we do delayed allocation during
buffered IO, we reserve space for the extent size hint alignment and
allocate the physical space to align the extent, but we do not zero
the regions of the extent that aren't written by the write(2)
syscall. The result is that we expose stale data in unwritten
regions of the extent size hints.

There are two ways to fix this. The first is to detect that we are
doing unaligned writes, check if there is already a mapping or data
over the extent size hint range, and if not zero the page cache
first before then doing the real write. This can be very expensive
for large extent size hints, especially if the subsequent writes
fill then entire extent size before the data is written to disk.

The second, and simpler way, is simply to turn off delayed
allocation when the extent size hint is set and use preallocation
instead. This results in unwritten extents being laid down on disk
and so only the written portions will be converted. This matches the
behaviour for direct IO, and will also work for the real time
device. The disadvantage of this approach is that for small extent
size hints we can get file fragmentation, but in general extent size
hints are fairly large (e.g. stripe width sized) so this isn't a big
deal.

Implement the second approach as it is simple and effective.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index e4acb6b..33abd52 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1223,7 +1223,7 @@ __xfs_get_blocks(
 	    (!nimaps ||
 	     (imap.br_startblock == HOLESTARTBLOCK ||
 	      imap.br_startblock == DELAYSTARTBLOCK))) {
-		if (direct) {
+		if (direct || xfs_get_extsz_hint(ip)) {
 			/*
 			 * Drop the ilock in preparation for starting the block
 			 * allocation transaction.  It will be retaken
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 081/102] xfs: fix buffer lookup race on allocation failure

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: fe2429b0966a7ec42b5fe3bf96f0f10de0a3b536

When memory allocation fails to add the page array or tht epages to
a buffer during xfs_buf_get(), the buffer is left in the cache in a
partially initialised state. There is enough state left for the next
lookup on that buffer to find the buffer, and for the buffer to then
be used without finishing the initialisation.  As a result, when an
attempt to do IO on the buffer occurs, it fails with EIO because
there are no pages attached to the buffer.

We cannot remove the buffer from the cache immediately and free it,
because there may already be a racing lookup that is blocked on the
buffer lock. Hence the moment we unlock the buffer to then free it,
the other user is woken and we have a use-after-free situation.

To avoid this race condition altogether, allocate the pages for the
buffer before we insert it into the cache.  This then means that we
don't have an allocation  failure case to deal after the buffer is
already present in the cache, and hence avoid the problem
altogether.  In most cases we won't have racing inserts for the same
buffer, and so won't increase the memory pressure allocation before
insertion may entail.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 2df0b4a..0858fe1 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -549,18 +549,20 @@ xfs_buf_get(
 	if (unlikely(!new_bp))
 		return NULL;
 
+	error = xfs_buf_allocate_memory(new_bp, flags);
+	if (error) {
+		kmem_zone_free(xfs_buf_zone, new_bp);
+		return NULL;
+	}
+
 	bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
 	if (!bp) {
-		kmem_zone_free(xfs_buf_zone, new_bp);
+		xfs_buf_free(new_bp);
 		return NULL;
 	}
 
-	if (bp == new_bp) {
-		error = xfs_buf_allocate_memory(bp, flags);
-		if (error)
-			goto no_buffer;
-	} else
-		kmem_zone_free(xfs_buf_zone, new_bp);
+	if (bp != new_bp)
+		xfs_buf_free(new_bp);
 
 	/*
 	 * Now we have a workable buffer, fill in the block number so
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 082/102] xfs: check for buffer errors before waiting

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 0e95f19ad983e72a9cb93a67b3290b58f0467b36

If we call xfs_buf_iowait() on a buffer that failed dispatch due to
an IO error, it will wait forever for an Io that does not exist.
This is hndled in xfs_buf_read, but there is other code that calls
xfs_buf_iowait directly that doesn't.

Rather than make the call sites have to handle checking for dispatch
errors and then checking for completion errors, make
xfs_buf_iowait() check for dispatch errors on the buffer before
waiting. This means we handle both dispatch and completion errors
with one set of error handling at the caller sites.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |   24 +++++++++++-------------
 fs/xfs/linux-2.6/xfs_buf.h |    2 +-
 fs/xfs/xfs_log_recover.c   |    2 ++
 3 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 0858fe1..5487eaa 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -597,8 +597,6 @@ _xfs_buf_read(
 	xfs_buf_t		*bp,
 	xfs_buf_flags_t		flags)
 {
-	int			status;
-
 	ASSERT(!(flags & (XBF_DELWRI|XBF_WRITE)));
 	ASSERT(bp->b_bn != XFS_BUF_DADDR_NULL);
 
@@ -607,9 +605,9 @@ _xfs_buf_read(
 	bp->b_flags |= flags & (XBF_READ | XBF_ASYNC | \
 			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
 
-	status = xfs_buf_iorequest(bp);
-	if (status || XFS_BUF_ISERROR(bp) || (flags & XBF_ASYNC))
-		return status;
+	xfs_buf_iorequest(bp);
+	if (flags & XBF_ASYNC)
+		return 0;
 	return xfs_buf_iowait(bp);
 }
 
@@ -696,7 +694,7 @@ xfs_buf_read_uncached(
 
 	xfsbdstrat(mp, bp);
 	error = xfs_buf_iowait(bp);
-	if (error || bp->b_error) {
+	if (error) {
 		xfs_buf_relse(bp);
 		return NULL;
 	}
@@ -1288,7 +1286,7 @@ next_chunk:
 	}
 }
 
-int
+void
 xfs_buf_iorequest(
 	xfs_buf_t		*bp)
 {
@@ -1296,7 +1294,7 @@ xfs_buf_iorequest(
 
 	if (bp->b_flags & XBF_DELWRI) {
 		xfs_buf_delwri_queue(bp, 1);
-		return 0;
+		return;
 	}
 
 	if (bp->b_flags & XBF_WRITE) {
@@ -1314,13 +1312,12 @@ xfs_buf_iorequest(
 	_xfs_buf_ioend(bp, 0);
 
 	xfs_buf_rele(bp);
-	return 0;
 }
 
 /*
- *	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.
+ * Waits for I/O to complete on the buffer supplied.  It returns immediately if
+ * no I/O is pending or there is already a pending error on the buffer.  It
+ * returns the I/O error code, if any, or 0 if there was no error.
  */
 int
 xfs_buf_iowait(
@@ -1328,7 +1325,8 @@ xfs_buf_iowait(
 {
 	trace_xfs_buf_iowait(bp, _RET_IP_);
 
-	wait_for_completion(&bp->b_iowait);
+	if (!bp->b_error)
+		wait_for_completion(&bp->b_iowait);
 
 	trace_xfs_buf_iowait_done(bp, _RET_IP_);
 	return bp->b_error;
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index 71d6698..edb8d7c 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -208,7 +208,7 @@ extern int xfs_bdstrat_cb(struct xfs_buf *);
 extern void xfs_buf_ioend(xfs_buf_t *,	int);
 extern void xfs_buf_ioerror(xfs_buf_t *, int);
 extern void xfs_buf_ioerror_alert(struct xfs_buf *, const char *func);
-extern int xfs_buf_iorequest(xfs_buf_t *);
+extern void 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, void *,
 				xfs_buf_rw_t);
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 82f7d18..13c057b 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -177,6 +177,7 @@ xlog_bread_noalign(
 	XFS_BUF_BUSY(bp);
 	XFS_BUF_SET_COUNT(bp, BBTOB(nbblks));
 	XFS_BUF_SET_TARGET(bp, log->l_mp->m_logdev_targp);
+	bp->b_error = 0;
 
 	xfsbdstrat(log->l_mp, bp);
 	error = xfs_buf_iowait(bp);
@@ -266,6 +267,7 @@ xlog_bwrite(
 	XFS_BUF_PSEMA(bp, PRIBIO);
 	XFS_BUF_SET_COUNT(bp, BBTOB(nbblks));
 	XFS_BUF_SET_TARGET(bp, log->l_mp->m_logdev_targp);
+	bp->b_error = 0;
 
 	error = xfs_bwrite(log->l_mp, bp);
 	if (error)
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 083/102] xfs: fix incorrect b_offset initialisation

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: ead360c50d33772f45943792893a58865adf3638

Because we no longer use the page cache for buffering, there is no
direct block number to page offset relationship anymore.
xfs_buf_get_pages is still setting up b_offset as if there was some
relationship, and that is leading to incorrectly setting up
*uncached* buffers that don't overwrite b_offset once they've had
pages allocated.

For cached buffers, the first block of the buffer is always at offset
zero into the allocated memory. This is true for sub-page sized
buffers, as well as for multiple-page buffers.

For uncached buffers, b_offset is only non-zero when we are
associating specific memory to the buffers, and that is set
correctly by the code setting up the buffer.

Hence remove the setting of b_offset in xfs_buf_get_pages, because
it is now always the wrong thing to do.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |    1 -
 1 file changed, 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 5487eaa..3e9946c 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -219,7 +219,6 @@ _xfs_buf_get_pages(
 {
 	/* Make sure that we have a page list */
 	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;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 084/102] xfs: use kmem_zone_zalloc for buffers

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: bf813cdddfb3a5bc88e1612e8f62a12367871213

To replace the alloc/memset pair.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 3e9946c..522ab83 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -170,7 +170,7 @@ xfs_buf_alloc(
 {
 	struct xfs_buf		*bp;
 
-	bp = kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags));
+	bp = kmem_zone_zalloc(xfs_buf_zone, xb_to_km(flags));
 	if (unlikely(!bp))
 		return NULL;
 
@@ -179,7 +179,6 @@ xfs_buf_alloc(
 	 */
 	flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);
 
-	memset(bp, 0, sizeof(xfs_buf_t));
 	atomic_set(&bp->b_hold, 1);
 	atomic_set(&bp->b_lru_ref, 1);
 	init_completion(&bp->b_iowait);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 085/102] xfs: use iolock on XFS_IOC_ALLOCSP calls

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: bc4010ecb8f4d4316e1a63a879a2715e49d113ad

fsstress has a particular effective way of stopping debug XFS
kernels. We keep seeing assert failures due finding delayed
allocation extents where there should be none. This shows up when
extracting extent maps and we are holding all the locks we should be
to prevent races, so this really makes no sense to see these errors.

After checking that fsstress does not use mmap, it occurred to me
that fsstress uses something that no sane application uses - the
XFS_IOC_ALLOCSP ioctl interfaces for preallocation. These interfaces
do allocation of blocks beyond EOF without using preallocation, and
then call setattr to extend and zero the allocated blocks.

THe problem here is this is a buffered write, and hence the
allocation is a delayed allocation. Unlike the buffered IO path, the
allocation and zeroing are not serialised using the IOLOCK. Hence
the ALLOCSP operation can race with operations holding the iolock to
prevent buffered IO operations from occurring.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_vnodeops.c |   21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 3988584..6ade3c6 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -2789,17 +2789,32 @@ xfs_change_file_space(
 	case XFS_IOC_ALLOCSP64:
 	case XFS_IOC_FREESP:
 	case XFS_IOC_FREESP64:
+		/*
+		 * These operations actually do IO when extending the file, but
+		 * the allocation is done seperately to the zeroing that is
+		 * done. This set of operations need to be serialised against
+		 * other IO operations, such as truncate and buffered IO. We
+		 * need to take the IOLOCK here to serialise the allocation and
+		 * zeroing IO to prevent other IOLOCK holders (e.g. getbmap,
+		 * truncate, direct IO) from racing against the transient
+		 * allocated but not written state we can have here.
+		 */
+		xfs_ilock(ip, XFS_IOLOCK_EXCL);
 		if (startoffset > fsize) {
 			error = xfs_alloc_file_space(ip, fsize,
-					startoffset - fsize, 0, attr_flags);
-			if (error)
+					startoffset - fsize, 0,
+					attr_flags | XFS_ATTR_NOLOCK);
+			if (error) {
+				xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 				break;
+			}
 		}
 
 		iattr.ia_valid = ATTR_SIZE;
 		iattr.ia_size = startoffset;
 
-		error = xfs_setattr(ip, &iattr, attr_flags);
+		error = xfs_setattr(ip, &iattr, attr_flags | XFS_ATTR_NOLOCK);
+		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 
 		if (error)
 			return error;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 086/102] xfs: Properly exclude IO type flags from buffer flags

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 12bcb3f7d4371f74bd25372e98e0d2da978e82b2

Recent event tracing during a debugging session showed that flags
that define the IO type for a buffer are leaking into the flags on
the buffer incorrectly. Fix the flag exclusion mask in
xfs_buf_alloc() to avoid problems that may be caused by such
leakage.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 522ab83..a87b93a 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -175,9 +175,11 @@ xfs_buf_alloc(
 		return NULL;
 
 	/*
-	 * We don't want certain flags to appear in b_flags.
+	 * We don't want certain flags to appear in b_flags unless they are
+	 * specifically set by later operations on the buffer.
 	 */
-	flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);
+	flags &= ~(XBF_LOCK | XBF_MAPPED | XBF_DONT_BLOCK |
+		   XBF_TRYLOCK | XBF_ASYNC | XBF_READ_AHEAD);
 
 	atomic_set(&bp->b_hold, 1);
 	atomic_set(&bp->b_lru_ref, 1);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 087/102] xfs: flush outstanding buffers on log mount failure

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: d4f3512b0891658b6b4d5fc99567242b3fc2d6b7

When we fail to mount the log in xfs_mountfs(), we tear down all the
infrastructure we have already allocated. However, the process of
mounting the log may have progressed to the point of reading,
caching and modifying buffers in memory. Hence before we can free
all the infrastructure, we have to flush and remove all the buffers
from memory.

Problem first reported by Eric Sandeen, later a different incarnation
was reported by Ben Myers.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_mount.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 2f011fd..beb9968 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -1297,7 +1297,7 @@ xfs_mountfs(
 			      XFS_FSB_TO_BB(mp, sbp->sb_logblocks));
 	if (error) {
 		xfs_warn(mp, "log mount failed");
-		goto out_free_perag;
+		goto out_fail_wait;
 	}
 
 	/*
@@ -1324,7 +1324,7 @@ xfs_mountfs(
 	     !mp->m_sb.sb_inprogress) {
 		error = xfs_initialize_perag_data(mp, sbp->sb_agcount);
 		if (error)
-			goto out_free_perag;
+			goto out_fail_wait;
 	}
 
 	/*
@@ -1448,6 +1448,10 @@ xfs_mountfs(
 	IRELE(rip);
  out_log_dealloc:
 	xfs_log_unmount(mp);
+ out_fail_wait:
+	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp)
+		xfs_wait_buftarg(mp->m_logdev_targp);
+	xfs_wait_buftarg(mp->m_ddev_targp);
  out_free_perag:
 	xfs_free_perag(mp);
  out_remove_uuid:
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 088/102] xfs: protect xfs_sync_worker with s_umount semaphore

[Dave Chinner]

From: Ben Myers <bpm@sgi.com>

Upstream commit: 1307bbd2af67283131728637e9489002adb26f10

xfs_sync_worker checks the MS_ACTIVE flag in s_flags to avoid doing
work during mount and unmount.  This flag can be cleared by unmount
after the xfs_sync_worker checks it but before the work is completed.
The has caused crashes in the completion handler for the dummy
transaction commited by xfs_sync_worker:

PID: 27544  TASK: ffff88013544e040  CPU: 3   COMMAND: "kworker/3:0"
 #0 [ffff88016fdff930] machine_kexec at ffffffff810244e9
 #1 [ffff88016fdff9a0] crash_kexec at ffffffff8108d053
 #2 [ffff88016fdffa70] oops_end at ffffffff813ad1b8
 #3 [ffff88016fdffaa0] no_context at ffffffff8102bd48
 #4 [ffff88016fdffaf0] __bad_area_nosemaphore at ffffffff8102c04d
 #5 [ffff88016fdffb40] bad_area_nosemaphore at ffffffff8102c12e
 #6 [ffff88016fdffb50] do_page_fault at ffffffff813afaee
 #7 [ffff88016fdffc60] page_fault at ffffffff813ac635
    [exception RIP: xlog_get_lowest_lsn+0x30]
    RIP: ffffffffa04a9910  RSP: ffff88016fdffd10  RFLAGS: 00010246
    RAX: ffffc90014e48000  RBX: ffff88014d879980  RCX: ffff88014d879980
    RDX: ffff8802214ee4c0  RSI: 0000000000000000  RDI: 0000000000000000
    RBP: ffff88016fdffd10   R8: ffff88014d879a80   R9: 0000000000000000
    R10: 0000000000000001  R11: 0000000000000000  R12: ffff8802214ee400
    R13: ffff88014d879980  R14: 0000000000000000  R15: ffff88022fd96605
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018
 #8 [ffff88016fdffd18] xlog_state_do_callback at ffffffffa04aa186 [xfs]
 #9 [ffff88016fdffd98] xlog_state_done_syncing at ffffffffa04aa568 [xfs]

Protect xfs_sync_worker by using the s_umount semaphore at the read
level to provide exclusion with unmount while work is progressing.

Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_sync.c |   31 +++++++++++++++++--------------
 1 file changed, 17 insertions(+), 14 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index ccb77cf..74ea9b1 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -470,21 +470,24 @@ xfs_sync_worker(
 	 * We shouldn't write/force the log if we are in the mount/unmount
 	 * process or on a read only filesystem. The workqueue still needs to be
 	 * active in both cases, however, because it is used for inode reclaim
-	 * during these times. hence use the MS_ACTIVE flag to avoid doing
-	 * anything in these periods.
+	 * during these times.  Use the s_umount semaphore to provide exclusion
+	 * with unmount.
 	 */
-	if (!(mp->m_super->s_flags & MS_ACTIVE) &&
-	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
-		/* dgc: errors ignored here */
-		if (mp->m_super->s_frozen == SB_UNFROZEN &&
-		    xfs_log_need_covered(mp))
-			error = xfs_fs_log_dummy(mp);
-		else
-			xfs_log_force(mp, 0);
-		error = xfs_qm_sync(mp, SYNC_TRYLOCK);
-
-		/* start pushing all the metadata that is currently dirty */
-		xfs_ail_push_all(mp->m_ail);
+	if (down_read_trylock(&mp->m_super->s_umount)) {
+		if (!(mp->m_super->s_flags & MS_ACTIVE) &&
+		    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+			/* dgc: errors ignored here */
+			if (mp->m_super->s_frozen == SB_UNFROZEN &&
+			    xfs_log_need_covered(mp))
+				error = xfs_fs_log_dummy(mp);
+			else
+				xfs_log_force(mp, 0);
+			error = xfs_qm_sync(mp, SYNC_TRYLOCK);
+
+			/* start pushing all the metadata that is currently dirty */
+			xfs_ail_push_all(mp->m_ail);
+		}
+		up_read(&mp->m_super->s_umount);
 	}
 
 	/* queue us up again */
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 089/102] xfs: fix memory reclaim deadlock on agi buffer

[Dave Chinner]

From: Peter Watkins <treestem@gmail.com>

Upstream commit: 3ba316037470bbf98c8a16c2179c02794fb8862e

Note xfs_iget can be called while holding a locked agi buffer. If
it goes into memory reclaim then inode teardown may try to lock the
same buffer. Prevent the deadlock by calling radix_tree_preload
with GFP_NOFS.

Signed-off-by: Peter Watkins <treestem@gmail.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_iget.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index 35b1f7b..a3663a8 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -337,9 +337,10 @@ xfs_iget_cache_miss(
 	/*
 	 * Preload the radix tree so we can insert safely under the
 	 * write spinlock. Note that we cannot sleep inside the preload
-	 * region.
+	 * region. Since we can be called from transaction context, don't
+	 * recurse into the file system.
 	 */
-	if (radix_tree_preload(GFP_KERNEL)) {
+	if (radix_tree_preload(GFP_NOFS)) {
 		error = EAGAIN;
 		goto out_destroy;
 	}
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 090/102] xfs: add trace points for log forces

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 14c26c6a05de138a4fd9a0c05ff8e7435a618324

To enable easy tracing of the location of log forces and the
frequency of them via perf, add a pair of trace points to the log
force functions.  This will help debug where excessive log forces
are being issued from by simple perf commands like:

# ~/perf/perf top -e xfs:xfs_log_force -G -U

Which gives this sort of output:

Events: 141  xfs:xfs_log_force
-  100.00%  [kernel]  [k] xfs_log_force
   - xfs_log_force
        87.04% xfsaild
           kthread
           kernel_thread_helper
      - 12.87% xfs_buf_lock
           _xfs_buf_find
           xfs_buf_get
           xfs_trans_get_buf
           xfs_da_do_buf
           xfs_da_get_buf
           xfs_dir2_data_init
           xfs_dir2_leaf_addname
           xfs_dir_createname
           xfs_create
           xfs_vn_mknod
           xfs_vn_create
           vfs_create
           do_last.isra.41
           path_openat
           do_filp_open
           do_sys_open
           sys_open
           system_call_fastpath

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sig.com>
---
 fs/xfs/linux-2.6/xfs_trace.h |   16 ++++++++++++++++
 fs/xfs/xfs_log.c             |    2 ++
 2 files changed, 18 insertions(+)

diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index 1315931..a163ae3 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -880,6 +880,22 @@ DECLARE_EVENT_CLASS(xfs_log_item_class,
 		  __print_flags(__entry->flags, "|", XFS_LI_FLAGS))
 )
 
+TRACE_EVENT(xfs_log_force,
+	TP_PROTO(struct xfs_mount *mp, xfs_lsn_t lsn),
+	TP_ARGS(mp, lsn),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_lsn_t, lsn)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->lsn = lsn;
+	),
+	TP_printk("dev %d:%d lsn 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->lsn)
+)
+
 #define DEFINE_LOG_ITEM_EVENT(name) \
 DEFINE_EVENT(xfs_log_item_class, name, \
 	TP_PROTO(struct xfs_log_item *lip), \
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 653cba8..9f80aa8 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -2981,6 +2981,7 @@ xfs_log_force(
 {
 	int	error;
 
+	trace_xfs_log_force(mp, 0);
 	error = _xfs_log_force(mp, flags, NULL);
 	if (error)
 		xfs_warn(mp, "%s: error %d returned.", __func__, error);
@@ -3131,6 +3132,7 @@ xfs_log_force_lsn(
 {
 	int	error;
 
+	trace_xfs_log_force(mp, lsn);
 	error = _xfs_log_force_lsn(mp, lsn, flags, NULL);
 	if (error)
 		xfs_warn(mp, "%s: error %d returned.", __func__, error);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 091/102] xfs: switch to proper __bitwise type for KM_... flags

[Dave Chinner]

From: Al Viro <viro@zeniv.linux.org.uk>

Upstream commit: 77ba78776e90e8de541f13b326e284c74286252f

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/xfs/linux-2.6/kmem.c |   10 +++++-----
 fs/xfs/linux-2.6/kmem.h |   21 +++++++++++----------
 fs/xfs/xfs_log.c        |    2 +-
 fs/xfs/xfs_log_priv.h   |    2 +-
 fs/xfs/xfs_trans.c      |    2 +-
 fs/xfs/xfs_trans.h      |    2 +-
 6 files changed, 20 insertions(+), 19 deletions(-)

diff --git a/fs/xfs/linux-2.6/kmem.c b/fs/xfs/linux-2.6/kmem.c
index a907de5..4a7286c 100644
--- a/fs/xfs/linux-2.6/kmem.c
+++ b/fs/xfs/linux-2.6/kmem.c
@@ -46,7 +46,7 @@ kmem_zalloc_greedy(size_t *size, size_t minsize, size_t maxsize)
 }
 
 void *
-kmem_alloc(size_t size, unsigned int __nocast flags)
+kmem_alloc(size_t size, xfs_km_flags_t flags)
 {
 	int	retries = 0;
 	gfp_t	lflags = kmem_flags_convert(flags);
@@ -65,7 +65,7 @@ kmem_alloc(size_t size, unsigned int __nocast flags)
 }
 
 void *
-kmem_zalloc(size_t size, unsigned int __nocast flags)
+kmem_zalloc(size_t size, xfs_km_flags_t flags)
 {
 	void	*ptr;
 
@@ -87,7 +87,7 @@ kmem_free(const void *ptr)
 
 void *
 kmem_realloc(const void *ptr, size_t newsize, size_t oldsize,
-	     unsigned int __nocast flags)
+	     xfs_km_flags_t flags)
 {
 	void	*new;
 
@@ -102,7 +102,7 @@ kmem_realloc(const void *ptr, size_t newsize, size_t oldsize,
 }
 
 void *
-kmem_zone_alloc(kmem_zone_t *zone, unsigned int __nocast flags)
+kmem_zone_alloc(kmem_zone_t *zone, xfs_km_flags_t flags)
 {
 	int	retries = 0;
 	gfp_t	lflags = kmem_flags_convert(flags);
@@ -121,7 +121,7 @@ kmem_zone_alloc(kmem_zone_t *zone, unsigned int __nocast flags)
 }
 
 void *
-kmem_zone_zalloc(kmem_zone_t *zone, unsigned int __nocast flags)
+kmem_zone_zalloc(kmem_zone_t *zone, xfs_km_flags_t flags)
 {
 	void	*ptr;
 
diff --git a/fs/xfs/linux-2.6/kmem.h b/fs/xfs/linux-2.6/kmem.h
index f7c8f7a..642f573 100644
--- a/fs/xfs/linux-2.6/kmem.h
+++ b/fs/xfs/linux-2.6/kmem.h
@@ -27,10 +27,11 @@
  * General memory allocation interfaces
  */
 
-#define KM_SLEEP	0x0001u
-#define KM_NOSLEEP	0x0002u
-#define KM_NOFS		0x0004u
-#define KM_MAYFAIL	0x0008u
+typedef unsigned __bitwise xfs_km_flags_t;
+#define KM_SLEEP	((__force xfs_km_flags_t)0x0001u)
+#define KM_NOSLEEP	((__force xfs_km_flags_t)0x0002u)
+#define KM_NOFS		((__force xfs_km_flags_t)0x0004u)
+#define KM_MAYFAIL	((__force xfs_km_flags_t)0x0008u)
 
 /*
  * We use a special process flag to avoid recursive callbacks into
@@ -38,7 +39,7 @@
  * warnings, so we explicitly skip any generic ones (silly of us).
  */
 static inline gfp_t
-kmem_flags_convert(unsigned int __nocast flags)
+kmem_flags_convert(xfs_km_flags_t flags)
 {
 	gfp_t	lflags;
 
@@ -54,9 +55,9 @@ kmem_flags_convert(unsigned int __nocast flags)
 	return lflags;
 }
 
-extern void *kmem_alloc(size_t, unsigned int __nocast);
-extern void *kmem_zalloc(size_t, unsigned int __nocast);
-extern void *kmem_realloc(const void *, size_t, size_t, unsigned int __nocast);
+extern void *kmem_alloc(size_t, xfs_km_flags_t);
+extern void *kmem_zalloc(size_t, xfs_km_flags_t);
+extern void *kmem_realloc(const void *, size_t, size_t, xfs_km_flags_t);
 extern void  kmem_free(const void *);
 
 static inline void *kmem_zalloc_large(size_t size)
@@ -112,8 +113,8 @@ kmem_zone_destroy(kmem_zone_t *zone)
 		kmem_cache_destroy(zone);
 }
 
-extern void *kmem_zone_alloc(kmem_zone_t *, unsigned int __nocast);
-extern void *kmem_zone_zalloc(kmem_zone_t *, unsigned int __nocast);
+extern void *kmem_zone_alloc(kmem_zone_t *, xfs_km_flags_t);
+extern void *kmem_zone_zalloc(kmem_zone_t *, xfs_km_flags_t);
 
 static inline int
 kmem_shake_allow(gfp_t gfp_mask)
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 9f80aa8..dc37cdb 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -3194,7 +3194,7 @@ xlog_ticket_alloc(
 	int		cnt,
 	char		client,
 	bool		permanent,
-	int		alloc_flags)
+	xfs_km_flags_t	alloc_flags)
 {
 	struct xlog_ticket *tic;
 	uint		num_headers;
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 2152900..b871b6e 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -553,7 +553,7 @@ extern void	 xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
 extern kmem_zone_t *xfs_log_ticket_zone;
 struct xlog_ticket *xlog_ticket_alloc(struct log *log, int unit_bytes,
 				int count, char client, bool permanent,
-				int alloc_flags);
+				xfs_km_flags_t alloc_flags);
 
 
 static inline void
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index cc3bc95..0ee9042 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -585,7 +585,7 @@ xfs_trans_t *
 _xfs_trans_alloc(
 	xfs_mount_t	*mp,
 	uint		type,
-	uint		memflags)
+	xfs_km_flags_t	memflags)
 {
 	xfs_trans_t	*tp;
 
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index 53597f4..0269047 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -448,7 +448,7 @@ typedef struct xfs_trans {
  * XFS transaction mechanism exported interfaces.
  */
 xfs_trans_t	*xfs_trans_alloc(struct xfs_mount *, uint);
-xfs_trans_t	*_xfs_trans_alloc(struct xfs_mount *, uint, uint);
+xfs_trans_t	*_xfs_trans_alloc(struct xfs_mount *, uint, xfs_km_flags_t);
 xfs_trans_t	*xfs_trans_dup(xfs_trans_t *);
 int		xfs_trans_reserve(xfs_trans_t *, uint, uint, uint,
 				  uint, uint);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 092/102] xfs: xfs_vm_writepage clear iomap_valid when

[Dave Chinner]

From: Alain Renaud <arenaud@sgi.com>

Upstream commit: 66f9311381b4772003d595fb6c518f1647450db0
 !buffer_uptodate (REV2)

On filesytems with a block size smaller than PAGE_SIZE we currently have
a problem with unwritten extents.  If a we have multi-block page for
which an unwritten extent has been allocated, and only some of the
buffers have been written to, and they are not contiguous, we can expose
stale data from disk in the blocks between the writes after extent
conversion.

Example of a page with unwritten and real data.
buffer  content
0       empty  b_state = 0
1       DATA   b_state = 0x1023 Uptodate,Dirty,Mapped,Unwritten
2       DATA   b_state = 0x1023 Uptodate,Dirty,Mapped,Unwritten
3       empty  b_state = 0
4       empty  b_state = 0
5       DATA   b_state = 0x1023 Uptodate,Dirty,Mapped,Unwritten
6       DATA   b_state = 0x1023 Uptodate,Dirty,Mapped,Unwritten
7       empty  b_state = 0

Buffers 1, 2, 5, and 6 have been written to, leaving 0, 3, 4, and 7
empty.  Currently buffers 1, 2, 5, and 6 are added to a single ioend,
and when IO has completed, extent conversion creates a real extent from
block 1 through block 6, leaving 0 and 7 unwritten.  However buffers 3
and 4 were not written to disk, so stale data is exposed from those
blocks on a subsequent read.

Fix this by setting iomap_valid = 0 when we find a buffer that is not
Uptodate.  This ensures that buffers 5 and 6 are not added to the same
ioend as buffers 1 and 2.  Later these blocks will be converted into two
separate real extents, leaving the blocks in between unwritten.

Signed-off-by: Alain Renaud <arenaud@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 33abd52..c16caf2 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1032,10 +1032,15 @@ xfs_vm_writepage(
 				imap_valid = 0;
 			}
 		} else {
-			if (PageUptodate(page)) {
+			if (PageUptodate(page))
 				ASSERT(buffer_mapped(bh));
-				imap_valid = 0;
-			}
+			/*
+			 * This buffer is not uptodate and will not be
+			 * written to disk.  Ensure that we will put any
+			 * subsequent writeable buffers into a new
+			 * ioend.
+			 */
+			imap_valid = 0;
 			continue;
 		}
 
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 093/102] xfs: fix debug_object WARN at xfs_alloc_vextent()

[Dave Chinner]

From: Jeff Liu <jeff.liu@oracle.com>

Upstream commit: 3b876c8f2a361ceeed3fed894980c69066f903a0

Fengguang reports:

[  780.529603] XFS (vdd): Ending clean mount
[  781.454590] ODEBUG: object is on stack, but not annotated
[  781.455433] ------------[ cut here ]------------
[  781.455433] WARNING: at /c/kernel-tests/sound/lib/debugobjects.c:301 __debug_object_init+0x173/0x1f1()
[  781.455433] Hardware name: Bochs
[  781.455433] Modules linked in:
[  781.455433] Pid: 26910, comm: kworker/0:2 Not tainted 3.4.0+ #51
[  781.455433] Call Trace:
[  781.455433]  [<ffffffff8106bc84>] warn_slowpath_common+0x83/0x9b
[  781.455433]  [<ffffffff8106bcb6>] warn_slowpath_null+0x1a/0x1c
[  781.455433]  [<ffffffff814919a5>] __debug_object_init+0x173/0x1f1
[  781.455433]  [<ffffffff81491c65>] debug_object_init+0x14/0x16
[  781.455433]  [<ffffffff8108842a>] __init_work+0x20/0x22
[  781.455433]  [<ffffffff8134ea56>] xfs_alloc_vextent+0x6c/0xd5

Use INIT_WORK_ONSTACK in xfs_alloc_vextent instead of INIT_WORK.

Reported-by: Wu Fengguang <wfg@linux.intel.com>
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_alloc.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index dfa9454..726418e 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -2448,7 +2448,7 @@ xfs_alloc_vextent(
 	DECLARE_COMPLETION_ONSTACK(done);
 
 	args->done = &done;
-	INIT_WORK(&args->work, xfs_alloc_vextent_worker);
+	INIT_WORK_ONSTACK(&args->work, xfs_alloc_vextent_worker);
 	queue_work(xfs_alloc_wq, &args->work);
 	wait_for_completion(&done);
 	return args->result;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 094/102] xfs: m_maxioffset is redundant

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: d2c2819117176e139dc761873c664aaa770c79c9

The m_maxioffset field in the struct xfs_mount contains the same
value as the superblock s_maxbytes field. There is no need to carry
two copies of this limit around, so use the VFS superblock version.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   12 ++++++------
 fs/xfs/xfs_iomap.c          |    4 ++--
 fs/xfs/xfs_mount.c          |    2 --
 fs/xfs/xfs_mount.h          |    3 +--
 4 files changed, 9 insertions(+), 12 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index c16caf2..42eaf00 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -362,10 +362,10 @@ xfs_map_blocks(
 
 	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
 	       (ip->i_df.if_flags & XFS_IFEXTENTS));
-	ASSERT(offset <= mp->m_maxioffset);
+	ASSERT(offset <= mp->m_super->s_maxbytes);
 
-	if (offset + count > mp->m_maxioffset)
-		count = mp->m_maxioffset - offset;
+	if (offset + count > mp->m_super->s_maxbytes)
+		count = mp->m_super->s_maxbytes - offset;
 	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
@@ -1213,9 +1213,9 @@ __xfs_get_blocks(
 		lockmode = xfs_ilock_map_shared(ip);
 	}
 
-	ASSERT(offset <= mp->m_maxioffset);
-	if (offset + size > mp->m_maxioffset)
-		size = mp->m_maxioffset - offset;
+	ASSERT(offset <= mp->m_super->s_maxbytes);
+	if (offset + size > mp->m_super->s_maxbytes)
+		size = mp->m_super->s_maxbytes - offset;
 	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + size);
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 121309f..1a7b2e7 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -422,8 +422,8 @@ retry:
 	 * Make sure preallocation does not create extents beyond the range we
 	 * actually support in this filesystem.
 	 */
-	if (last_fsb > XFS_B_TO_FSB(mp, mp->m_maxioffset))
-		last_fsb = XFS_B_TO_FSB(mp, mp->m_maxioffset);
+	if (last_fsb > XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes))
+		last_fsb = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
 
 	ASSERT(last_fsb > offset_fsb);
 
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index beb9968..2119389 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -1209,8 +1209,6 @@ xfs_mountfs(
 
 	xfs_set_maxicount(mp);
 
-	mp->m_maxioffset = xfs_max_file_offset(sbp->sb_blocklog);
-
 	error = xfs_uuid_mount(mp);
 	if (error)
 		goto out;
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index fdcb826..a666de7 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -176,7 +176,6 @@ typedef struct xfs_mount {
 	uint			m_qflags;	/* quota status flags */
 	xfs_trans_reservations_t m_reservations;/* precomputed res values */
 	__uint64_t		m_maxicount;	/* maximum inode count */
-	__uint64_t		m_maxioffset;	/* maximum inode offset */
 	__uint64_t		m_resblks;	/* total reserved blocks */
 	__uint64_t		m_resblks_avail;/* available reserved blocks */
 	__uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
@@ -299,7 +298,7 @@ xfs_preferred_iosize(xfs_mount_t *mp)
 			PAGE_CACHE_SIZE));
 }
 
-#define XFS_MAXIOFFSET(mp)	((mp)->m_maxioffset)
+#define XFS_MAXIOFFSET(mp)	((mp)->m_super->s_maxbytes)
 
 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)	\
 				((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 095/102] xfs: make largest supported offset less shouty

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 32972383ca46223aa2b129826b3789721ec147aa

XFS_MAXIOFFSET() is just a simple macro that resolves to
mp->m_maxioffset. It doesn't need to exist, and it just makes the
code unnecessarily loud and shouty.

Make it quiet and easy to read.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |    2 +-
 fs/xfs/quota/xfs_qm.c       |    2 +-
 fs/xfs/xfs_bmap.c           |    2 +-
 fs/xfs/xfs_inode.c          |   10 ++++------
 fs/xfs/xfs_iomap.c          |    2 +-
 fs/xfs/xfs_mount.h          |    2 --
 fs/xfs/xfs_vnodeops.c       |   10 +++++-----
 7 files changed, 13 insertions(+), 17 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 642554b8..28e3eed 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -269,7 +269,7 @@ xfs_file_aio_read(
 		}
 	}
 
-	n = XFS_MAXIOFFSET(mp) - iocb->ki_pos;
+	n = mp->m_super->s_maxbytes - iocb->ki_pos;
 	if (n <= 0 || size == 0)
 		return 0;
 
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index fd8ddac..3873d0e 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -1381,7 +1381,7 @@ xfs_qm_dqiterate(
 	map = kmem_alloc(XFS_DQITER_MAP_SIZE * sizeof(*map), KM_SLEEP);
 
 	lblkno = 0;
-	maxlblkcnt = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	maxlblkcnt = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
 	do {
 		nmaps = XFS_DQITER_MAP_SIZE;
 		/*
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 2b4e8d2..d2e5219 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5437,7 +5437,7 @@ xfs_getbmap(
 		if (xfs_get_extsz_hint(ip) ||
 		    ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC|XFS_DIFLAG_APPEND)){
 			prealloced = 1;
-			fixlen = XFS_MAXIOFFSET(mp);
+			fixlen = mp->m_super->s_maxbytes;
 		} else {
 			prealloced = 0;
 			fixlen = XFS_ISIZE(ip);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 4b278d6..591fca1 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1202,9 +1202,7 @@ xfs_isize_check(
 	 * an error.
 	 */
 	if (xfs_bmapi(NULL, ip, map_first,
-			 (XFS_B_TO_FSB(mp,
-				       (xfs_ufsize_t)XFS_MAXIOFFSET(mp)) -
-			  map_first),
+			 (XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes) - map_first),
 			 XFS_BMAPI_ENTIRE, NULL, 0, imaps, &nimaps,
 			 NULL))
 	    return;
@@ -1259,11 +1257,11 @@ xfs_file_last_byte(
 
 	last_byte = XFS_FSB_TO_B(mp, last_block);
 	if (last_byte < 0) {
-		return XFS_MAXIOFFSET(mp);
+		return mp->m_super->s_maxbytes;
 	}
 	last_byte += (1 << mp->m_writeio_log);
 	if (last_byte < 0) {
-		return XFS_MAXIOFFSET(mp);
+		return mp->m_super->s_maxbytes;
 	}
 	return last_byte;
 }
@@ -1535,7 +1533,7 @@ xfs_itruncate_finish(
 	 * beyond the maximum file size (ie it is the same as last_block),
 	 * then there is nothing to do.
 	 */
-	last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	last_block = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
 	ASSERT(first_unmap_block <= last_block);
 	done = 0;
 	if (last_block == first_unmap_block) {
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 1a7b2e7..388e744 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -290,7 +290,7 @@ xfs_iomap_eof_want_preallocate(
 	 * do any speculative allocation.
 	 */
 	start_fsb = XFS_B_TO_FSBT(mp, ((xfs_ufsize_t)(offset + count - 1)));
-	count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	count_fsb = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
 	while (count_fsb > 0) {
 		imaps = nimaps;
 		firstblock = NULLFSBLOCK;
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index a666de7..cbc8fe9 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -298,8 +298,6 @@ xfs_preferred_iosize(xfs_mount_t *mp)
 			PAGE_CACHE_SIZE));
 }
 
-#define XFS_MAXIOFFSET(mp)	((mp)->m_super->s_maxbytes)
-
 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)	\
 				((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
 #define XFS_FORCED_SHUTDOWN(mp)	((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 6ade3c6..e6135e9 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -601,7 +601,7 @@ xfs_free_eofblocks(
 	 * of the file.  If not, then there is nothing to do.
 	 */
 	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_ISIZE(ip));
-	last_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	last_fsb = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
 	if (last_fsb <= end_fsb)
 		return 0;
 	map_len = last_fsb - end_fsb;
@@ -2739,10 +2739,10 @@ xfs_change_file_space(
 
 	llen = bf->l_len > 0 ? bf->l_len - 1 : bf->l_len;
 
-	if (   (bf->l_start < 0)
-	    || (bf->l_start > XFS_MAXIOFFSET(mp))
-	    || (bf->l_start + llen < 0)
-	    || (bf->l_start + llen > XFS_MAXIOFFSET(mp)))
+	if (bf->l_start < 0 ||
+	    bf->l_start > mp->m_super->s_maxbytes ||
+	    bf->l_start + llen < 0 ||
+	    bf->l_start + llen > mp->m_super->s_maxbytes)
 		return XFS_ERROR(EINVAL);
 
 	bf->l_whence = 0;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 096/102] xfs: kill copy and paste segment checks in xfs_file_aio_read

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 5276432997feb2366ac1e77949e94fe86a394813

The generic segment check code now returns a count of the number of
bytes in the iovec, so we don't need to roll our own anymore.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_file.c |   17 +++--------------
 1 file changed, 3 insertions(+), 14 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 28e3eed..1b57ff9 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -232,7 +232,6 @@ xfs_file_aio_read(
 	ssize_t			ret = 0;
 	int			ioflags = 0;
 	xfs_fsize_t		n;
-	unsigned long		seg;
 
 	XFS_STATS_INC(xs_read_calls);
 
@@ -243,19 +242,9 @@ xfs_file_aio_read(
 	if (file->f_mode & FMODE_NOCMTIME)
 		ioflags |= IO_INVIS;
 
-	/* START copy & waste from filemap.c */
-	for (seg = 0; seg < nr_segs; seg++) {
-		const struct iovec *iv = &iovp[seg];
-
-		/*
-		 * If any segment has a negative length, or the cumulative
-		 * length ever wraps negative then return -EINVAL.
-		 */
-		size += iv->iov_len;
-		if (unlikely((ssize_t)(size|iv->iov_len) < 0))
-			return XFS_ERROR(-EINVAL);
-	}
-	/* END copy & waste from filemap.c */
+	ret = generic_segment_checks(iovp, &nr_segs, &size, VERIFY_WRITE);
+	if (ret < 0)
+		return ret;
 
 	if (unlikely(ioflags & IO_ISDIRECT)) {
 		xfs_buftarg_t	*target =
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 097/102] xfs: fix allocbt cursor leak in xfs_alloc_ag_vextent_near

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: 76d095388b040229ea1aad7dea45be0cfa20f589

When we fail to find an matching extent near the requested extent
specification during a left-right distance search in
xfs_alloc_ag_vextent_near, we fail to free the original cursor that
we used to look up the XFS_BTNUM_CNT tree and hence leak it.

Reported-by: Chris J Arges <chris.j.arges@canonical.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_alloc.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 726418e..679c049 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -1087,6 +1087,7 @@ restart:
 			goto restart;
 		}
 
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
 		trace_xfs_alloc_size_neither(args);
 		args->agbno = NULLAGBLOCK;
 		return 0;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 098/102] xfs: shutdown xfs_sync_worker before the log

[Dave Chinner]

From: Ben Myers <bpm@sgi.com>

Upstream commit: 8866fc6fa55e31b2bce931b7963ff16641b39dc7

Revert commit 1307bbd, which uses the s_umount semaphore to provide
exclusion between xfs_sync_worker and unmount, in favor of shutting down
the sync worker before freeing the log in xfs_log_unmount.  This is a
cleaner way of resolving the race between xfs_sync_worker and unmount
than using s_umount.

Signed-off-by: Ben Myers <bpm@sgi.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
---
 fs/xfs/linux-2.6/xfs_sync.c |   33 ++++++++++++++++-----------------
 fs/xfs/xfs_log.c            |    1 +
 2 files changed, 17 insertions(+), 17 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 74ea9b1..5b6f46b 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -470,24 +470,23 @@ xfs_sync_worker(
 	 * We shouldn't write/force the log if we are in the mount/unmount
 	 * process or on a read only filesystem. The workqueue still needs to be
 	 * active in both cases, however, because it is used for inode reclaim
-	 * during these times.  Use the s_umount semaphore to provide exclusion
-	 * with unmount.
+	 * during these times.  Use the MS_ACTIVE flag to avoid doing anything
+	 * during mount.  Doing work during unmount is avoided by calling
+	 * cancel_delayed_work_sync on this work queue before tearing down
+	 * the ail and the log in xfs_log_unmount.
 	 */
-	if (down_read_trylock(&mp->m_super->s_umount)) {
-		if (!(mp->m_super->s_flags & MS_ACTIVE) &&
-		    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
-			/* dgc: errors ignored here */
-			if (mp->m_super->s_frozen == SB_UNFROZEN &&
-			    xfs_log_need_covered(mp))
-				error = xfs_fs_log_dummy(mp);
-			else
-				xfs_log_force(mp, 0);
-			error = xfs_qm_sync(mp, SYNC_TRYLOCK);
-
-			/* start pushing all the metadata that is currently dirty */
-			xfs_ail_push_all(mp->m_ail);
-		}
-		up_read(&mp->m_super->s_umount);
+	if (!(mp->m_super->s_flags & MS_ACTIVE) &&
+	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		/* dgc: errors ignored here */
+		if (mp->m_super->s_frozen == SB_UNFROZEN &&
+		    xfs_log_need_covered(mp))
+			error = xfs_fs_log_dummy(mp);
+		else
+			xfs_log_force(mp, 0);
+		error = xfs_qm_sync(mp, SYNC_TRYLOCK);
+
+		/* start pushing all the metadata that is currently dirty */
+		xfs_ail_push_all(mp->m_ail);
 	}
 
 	/* queue us up again */
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index dc37cdb..3030618 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -813,6 +813,7 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 void
 xfs_log_unmount(xfs_mount_t *mp)
 {
+	cancel_delayed_work_sync(&mp->m_sync_work);
 	xfs_trans_ail_destroy(mp);
 	xlog_dealloc_log(mp->m_log);
 }
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 099/102] xfs: really fix the cursor leak in xfs_alloc_ag_vextent_near

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: e3a746f5aab71f2dd0a83116772922fb37ae29d6

The current cursor is reallocated when retrying the allocation, so
the existing cursor needs to be destroyed in both the restart and
the failure cases.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_alloc.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 679c049..de5a353 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -1081,13 +1081,13 @@ restart:
 	 * If we couldn't get anything, give up.
 	 */
 	if (bno_cur_lt == NULL && bno_cur_gt == NULL) {
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+
 		if (!forced++) {
 			trace_xfs_alloc_near_busy(args);
 			xfs_log_force(args->mp, XFS_LOG_SYNC);
 			goto restart;
 		}
-
-		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
 		trace_xfs_alloc_size_neither(args);
 		args->agbno = NULLAGBLOCK;
 		return 0;
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 100/102] xfs: don't defer metadata allocation to the workqueue

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Upstream commit: aa292847b9fc6e187547110de833a7d3131bbddf

Almost all metadata allocations come from shallow stack usage
situations. Avoid the overhead of switching the allocation to a
workqueue as we are not in danger of running out of stack when
making these allocations. Metadata allocations are already marked
through the args that are passed down, so this is trivial to do.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reported-by: Mel Gorman <mgorman@suse.de>
Tested-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/xfs_alloc.c |   15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index de5a353..5d830cf 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -2441,13 +2441,22 @@ xfs_alloc_vextent_worker(
 	current_restore_flags_nested(&pflags, PF_FSTRANS);
 }
 
-
-int				/* error */
+/*
+ * Data allocation requests often come in with little stack to work on. Push
+ * them off to a worker thread so there is lots of stack to use. Metadata
+ * requests, OTOH, are generally from low stack usage paths, so avoid the
+ * context switch overhead here.
+ */
+int
 xfs_alloc_vextent(
-	xfs_alloc_arg_t	*args)	/* allocation argument structure */
+	struct xfs_alloc_arg	*args)
 {
 	DECLARE_COMPLETION_ONSTACK(done);
 
+	if (!args->userdata)
+		return __xfs_alloc_vextent(args);
+
+
 	args->done = &done;
 	INIT_WORK_ONSTACK(&args->work, xfs_alloc_vextent_worker);
 	queue_work(xfs_alloc_wq, &args->work);
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 101/102] xfs: prevent recursion in xfs_buf_iorequest

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 40a9b7963df32e743c45d79a5f41445fe2476f15

If the b_iodone handler is run in calling context in xfs_buf_iorequest we
can run into a recursion where xfs_buf_iodone_callbacks keeps calling back
into xfs_buf_iorequest because an I/O error happened, which keeps calling
back into xfs_buf_iorequest.  This chain will usually not take long
because the filesystem gets shut down because of log I/O errors, but even
over a short time it can cause stack overflows if run on the same context.

As a short term workaround make sure we always call the iodone handler in
workqueue context.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_buf.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index a87b93a..c74d027 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -1309,7 +1309,7 @@ xfs_buf_iorequest(
 	 */
 	atomic_set(&bp->b_io_remaining, 1);
 	_xfs_buf_ioapply(bp);
-	_xfs_buf_ioend(bp, 0);
+	_xfs_buf_ioend(bp, 1);
 
 	xfs_buf_rele(bp);
 }
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 102/102] xfs: handle EOF correctly in xfs_vm_writepage

[Dave Chinner]

From: Christoph Hellwig <hch@infradead.org>

Upstream commit: 6b7a03f03a2f8b1629133e35729eba4727fae3cc

We need to zero out part of a page which beyond EOF before setting uptodate,
otherwise, mapread or write will see non-zero data beyond EOF.

Based on the code in fs/buffer.c and the following ext4 commit:

  ext4: handle EOF correctly in ext4_bio_write_page()

And yes, I wish we had a good test case for it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
---
 fs/xfs/linux-2.6/xfs_aops.c |   19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 42eaf00..8c2ed74 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -978,11 +978,26 @@ xfs_vm_writepage(
 	end_index = offset >> PAGE_CACHE_SHIFT;
 	last_index = (offset - 1) >> PAGE_CACHE_SHIFT;
 	if (page->index >= end_index) {
-		if ((page->index >= end_index + 1) ||
-		    !(i_size_read(inode) & (PAGE_CACHE_SIZE - 1))) {
+		unsigned offset_into_page = offset & (PAGE_CACHE_SIZE - 1);
+
+		/*
+		 * Just skip the page if it is fully outside i_size, e.g. due
+		 * to a truncate operation that is in progress.
+		 */
+		if (page->index >= end_index + 1 || offset_into_page == 0) {
 			unlock_page(page);
 			return 0;
 		}
+
+		/*
+		 * The page straddles i_size.  It must be zeroed out on each
+		 * and every writepage invocation because it may be mmapped.
+		 * "A file is mapped in multiples of the page size.  For a file
+		 * that is not a multiple of the  page size, the remaining
+		 * memory is zeroed when mapped, and writes to that region are
+		 * not written out to the file."
+		 */
+		zero_user_segment(page, offset_into_page, PAGE_CACHE_SIZE);
 	}
 
 	end_offset = min_t(unsigned long long,
-- 
1.7.10

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Dave Chinner]

Folks,

Can please tell me if you did or didn't get the entire series. I
didn't receive the entire series back in my inbox - about 20 are
missing from my inbox.  A quick check of the archives shows that all
102 patches reached the server and entered the archive, so I'm
curious to know how many people had delivery failures....

Cheers,

Dave.

On Thu, Aug 23, 2012 at 03:01:18PM +1000, Dave Chinner wrote:
> This series is a backport of all the major bug fixes in the current
> TOT kernel to the current 3.0.x stable tree.
> 
> I won't make any secret of this - the fixes and supporting patches
> have been selected as a result of the issues reported in the current
> RHEL XFS codebase which currently is 2 commits short of the 3.0 code
> base.  With that said, it doesn't take a brain surgeon to work out
> the motivations behind this work and eventual destination of the
> patch set. ;)
> 
> There's no guarantee I have caught every single bug fix that has
> been made since 3.0, but I've tried to grab all the bug fix commits
> as indicated by the commit headers (hence the importance of good one
> line bug summaries).
> 
> Over the past couple of weeks of testing and refining, I've had only
> three significant problems arise from QA and load testing:
> 
> 	1) An unreproducable log space hang
> 	2) An unmount panic due to buffers not being cleaned up
> 	before tearing down the perag tree
> 	3) A forced shutdown panic in block_invalidatepage()
> 	via xfs_aops_discard_page()
> 
> It's entirely possible that #1 was due to the CIL space hang we
> still haven't got to the bottom of, so i'm not not greatly concerned
> by that. #2 implies I haven't quite backported the shutdown ordering
> fixes correctly (or I missed one), so I have a bit more work to do
> there. And for #3 - I've never seen that before and I haven't been
> able to reproduce it, so I really don't know what potential cause or
> impact it has.
> 
> I've been beating on the series with xfstests, dbench, fsmark,
> postmark, compilebench and a few other load scripts that I've got,
> and it seems fairly resilient.  Hence, it's time to give the series
> wider testing and review to flush out any remaining issues.
> 
> For all the folks that run 3.0.x stable kernels, I'd appreciate it if
> you could give this a whirl on your test systems to see if there are
> any obvious, glaring problems that show up under your particular
> workloads. This woul dbe of great benefit to me before I submit the
> series to the stable kernel gurus - I'd prefer it there's more
> substantial testing than "i've done what I can" when sending them
> the series.
> 
> For all the XFS developers that have copious amounts of free time
> available, I'd appreciate an eye run over the patch list to see if
> there's any potential bug fixes that I missed or have made glaring
> errors in backporting. Some of the fixes are dependent on cleanups I
> haven't included, so some of the patches are a bit different to what
> is in mainline (e.g. anything that touches setattr). Most important
> to look at is probably the inode i_size changes and the logging of
> all metadata changes.
> 
> Enjoy!
> 
> Cheers,
> 
> Dave.
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 

-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Ben Myers]

Hi Dave,

On Fri, Aug 24, 2012 at 07:54:09AM +1000, Dave Chinner wrote:
> Can please tell me if you did or didn't get the entire series. I
> didn't receive the entire series back in my inbox - about 20 are
> missing from my inbox.  A quick check of the archives shows that all
> 102 patches reached the server and entered the archive, so I'm
> curious to know how many people had delivery failures....

I got the whole series.  Pretty excellent FWICS so far.

Regards,
Ben


> Cheers,
> 
> Dave.
> 
> On Thu, Aug 23, 2012 at 03:01:18PM +1000, Dave Chinner wrote:
> > This series is a backport of all the major bug fixes in the current
> > TOT kernel to the current 3.0.x stable tree.
> > 
> > I won't make any secret of this - the fixes and supporting patches
> > have been selected as a result of the issues reported in the current
> > RHEL XFS codebase which currently is 2 commits short of the 3.0 code
> > base.  With that said, it doesn't take a brain surgeon to work out
> > the motivations behind this work and eventual destination of the
> > patch set. ;)
> > 
> > There's no guarantee I have caught every single bug fix that has
> > been made since 3.0, but I've tried to grab all the bug fix commits
> > as indicated by the commit headers (hence the importance of good one
> > line bug summaries).
> > 
> > Over the past couple of weeks of testing and refining, I've had only
> > three significant problems arise from QA and load testing:
> > 
> > 	1) An unreproducable log space hang
> > 	2) An unmount panic due to buffers not being cleaned up
> > 	before tearing down the perag tree
> > 	3) A forced shutdown panic in block_invalidatepage()
> > 	via xfs_aops_discard_page()
> > 
> > It's entirely possible that #1 was due to the CIL space hang we
> > still haven't got to the bottom of, so i'm not not greatly concerned
> > by that. #2 implies I haven't quite backported the shutdown ordering
> > fixes correctly (or I missed one), so I have a bit more work to do
> > there. And for #3 - I've never seen that before and I haven't been
> > able to reproduce it, so I really don't know what potential cause or
> > impact it has.
> > 
> > I've been beating on the series with xfstests, dbench, fsmark,
> > postmark, compilebench and a few other load scripts that I've got,
> > and it seems fairly resilient.  Hence, it's time to give the series
> > wider testing and review to flush out any remaining issues.
> > 
> > For all the folks that run 3.0.x stable kernels, I'd appreciate it if
> > you could give this a whirl on your test systems to see if there are
> > any obvious, glaring problems that show up under your particular
> > workloads. This woul dbe of great benefit to me before I submit the
> > series to the stable kernel gurus - I'd prefer it there's more
> > substantial testing than "i've done what I can" when sending them
> > the series.
> > 
> > For all the XFS developers that have copious amounts of free time
> > available, I'd appreciate an eye run over the patch list to see if
> > there's any potential bug fixes that I missed or have made glaring
> > errors in backporting. Some of the fixes are dependent on cleanups I
> > haven't included, so some of the patches are a bit different to what
> > is in mainline (e.g. anything that touches setattr). Most important
> > to look at is probably the inode i_size changes and the logging of
> > all metadata changes.
> > 
> > Enjoy!
> > 
> > Cheers,
> > 
> > Dave.
> > 
> > _______________________________________________
> > xfs mailing list
> > xfs@oss.sgi.com
> > http://oss.sgi.com/mailman/listinfo/xfs
> > 
> > -- 
> > This message has been scanned for viruses and
> > dangerous content by MailScanner, and is
> > believed to be clean.
> > 
> > 
> 
> -- 
> Dave Chinner
> david@fromorbit.com
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Matthias Schniedermeyer]

On 24.08.2012 07:54, Dave Chinner wrote:
> Folks,
> 
> Can please tell me if you did or didn't get the entire series. I
> didn't receive the entire series back in my inbox - about 20 are
> missing from my inbox.  A quick check of the archives shows that all
> 102 patches reached the server and entered the archive, so I'm
> curious to know how many people had delivery failures....

The thread (currently) occupies slots 78 to 180 in my MUA, which is 103 
of 103 if i'm counting right.





Bis denn

-- 
Real Programmers consider "what you see is what you get" to be just as 
bad a concept in Text Editors as it is in women. No, the Real Programmer
wants a "you asked for it, you got it" text editor -- complicated, 
cryptic, powerful, unforgiving, dangerous.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Christoph Hellwig]

On Thu, Aug 23, 2012 at 03:01:38PM +1000, Dave Chinner wrote:
> From: Mel Gorman <mgorman@suse.de>
> 
> Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4
> 
> Direct reclaim should never writeback pages.  For now, handle the
> situation and warn about it.  Ultimately, this will be a BUG_ON.

Is this actually the case on 3.0-stable?

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Christoph Hellwig]

I've done a brief look over the patches this week and while I can't spot
anything wrong I'm defintively a bit concerned about the amount of churn
for a long term stable series.  A lot of this does not seem to fit the
strict -stable criteria, and given that I've not really seen any major
issues with the current 3.0-stable codebase I'm wondering what the
guranteed gain vs the status quo is.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Dave Chinner]

On Sat, Sep 01, 2012 at 07:10:19PM -0400, Christoph Hellwig wrote:
> I've done a brief look over the patches this week and while I can't spot
> anything wrong I'm defintively a bit concerned about the amount of churn
> for a long term stable series.  A lot of this does not seem to fit the
> strict -stable criteria, and given that I've not really seen any major
> issues with the current 3.0-stable codebase I'm wondering what the
> guranteed gain vs the status quo is.

You didn't troll the RH bugzilla ;)

The XFS code base in RHEL6 is sitting at 3.0, and several of the
problems that have workarounds in 3.0.x don't fix the problems
reported (e.g. the log space hangs), while the fixes in the more
recent mainline kernel do.

I simply figured that I've got to do this much work to fix all the
bugs reported in RHEL6 and given the code bases are almost identical
I'd do a community service and push it to 3.0.x first. I'm quite
happy not to push it to 3.0.x if the consensus is that it is too
much churn.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Ben Myers]

On Mon, Sep 03, 2012 at 04:04:06PM +1000, Dave Chinner wrote:
> On Sat, Sep 01, 2012 at 07:10:19PM -0400, Christoph Hellwig wrote:
> > I've done a brief look over the patches this week and while I can't spot
> > anything wrong I'm defintively a bit concerned about the amount of churn
> > for a long term stable series.  A lot of this does not seem to fit the
> > strict -stable criteria, and given that I've not really seen any major
> > issues with the current 3.0-stable codebase I'm wondering what the
> > guranteed gain vs the status quo is.
> 
> You didn't troll the RH bugzilla ;)

Hehe.  ;)

> The XFS code base in RHEL6 is sitting at 3.0, and several of the
> problems that have workarounds in 3.0.x don't fix the problems
> reported (e.g. the log space hangs), while the fixes in the more
> recent mainline kernel do.
> 
> I simply figured that I've got to do this much work to fix all the
> bugs reported in RHEL6 and given the code bases are almost identical
> I'd do a community service and push it to 3.0.x first. I'm quite
> happy not to push it to 3.0.x if the consensus is that it is too
> much churn.

FWIW I think it's great that you've done the community this service.
I'm just now (finally) getting this patchset onto a machine for some
testing.

Ok.. now that I've refreshed my memory of
Documentation/stable_kernel_rules.txt, I can see Christoph's point.
Some of these patches are hard to justify based upon those rules.  For
example:

Patch 2, 'remove dead ENODEV handling in xfs_destroy_ioend' seems to
just remove some dead code.  However, if it were clear that it fixes
some kind of problem by removing the dead code it could easily be
justified.

Patch 94, 'm_maxioffset is redundant' seems like a cleanup.  But maybe
it is required so that a subsequent patch will apply properly?

Patch 95, 'make largest supported offset less shouty' just cleans up a
macro.  Again... maybe there is a good reason to pull that in besides
the shouting.

So there are three examples that would probably require some kind of
justification to apply within the rules as stated.  Maybe -stable just
needs a link to the bug each one of them resolves.

It looks like there are plenty in the patchset that are very clearly
appropriate within the rules, no additional justification required.  We
certainly don't want to lose that.  We can go through the set and reply
to those which seem questionable...

-Ben

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Dave Chinner]

On Tue, Sep 04, 2012 at 04:13:57PM -0500, Ben Myers wrote:
> On Mon, Sep 03, 2012 at 04:04:06PM +1000, Dave Chinner wrote:
> > On Sat, Sep 01, 2012 at 07:10:19PM -0400, Christoph Hellwig wrote:
> > > I've done a brief look over the patches this week and while I can't spot
> > > anything wrong I'm defintively a bit concerned about the amount of churn
> > > for a long term stable series.  A lot of this does not seem to fit the
> > > strict -stable criteria, and given that I've not really seen any major
> > > issues with the current 3.0-stable codebase I'm wondering what the
> > > guranteed gain vs the status quo is.
> > 
> > You didn't troll the RH bugzilla ;)
> 
> Hehe.  ;)
> 
> > The XFS code base in RHEL6 is sitting at 3.0, and several of the
> > problems that have workarounds in 3.0.x don't fix the problems
> > reported (e.g. the log space hangs), while the fixes in the more
> > recent mainline kernel do.
> > 
> > I simply figured that I've got to do this much work to fix all the
> > bugs reported in RHEL6 and given the code bases are almost identical
> > I'd do a community service and push it to 3.0.x first. I'm quite
> > happy not to push it to 3.0.x if the consensus is that it is too
> > much churn.
> 
> FWIW I think it's great that you've done the community this service.
> I'm just now (finally) getting this patchset onto a machine for some
> testing.
> 
> Ok.. now that I've refreshed my memory of
> Documentation/stable_kernel_rules.txt, I can see Christoph's point.
> Some of these patches are hard to justify based upon those rules.  For
> example:
> 
> Patch 2, 'remove dead ENODEV handling in xfs_destroy_ioend' seems to
> just remove some dead code.  However, if it were clear that it fixes
> some kind of problem by removing the dead code it could easily be
> justified.

The are several patches in the series touch xfs_destroy_ioend().
Having to resolve all the conflicts by hand instead of including the
patch that removes the 5 lines of dead code means I didn't have to
modify the patches when applying them, and the result is verifiably
identical to the code in the current mainline tree...

Indeed, the very next patch would not apply without patch 2....

> Patch 94, 'm_maxioffset is redundant' seems like a cleanup.  But maybe
> it is required so that a subsequent patch will apply properly?
> 
> Patch 95, 'make largest supported offset less shouty' just cleans up a
> macro.  Again... maybe there is a good reason to pull that in besides
> the shouting.

Yeah, 94/95 can go. What probably happened there was that I was
considering a later patch that was dependent on these changes and I
ended up dropping that patch without realising I then didn't need
the m_maxioffset changes.

> So there are three examples that would probably require some kind of
> justification to apply within the rules as stated.  Maybe -stable just
> needs a link to the bug each one of them resolves.
> 
> It looks like there are plenty in the patchset that are very clearly
> appropriate within the rules, no additional justification required.  We
> certainly don't want to lose that.  We can go through the set and reply
> to those which seem questionable...

Sure, but remember that the dependent patches might be 50 patches
later in the series as the patches are applied in upstream commit
order....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Mel Gorman]

On Mon, Aug 27, 2012 at 02:17:42PM -0400, Christoph Hellwig wrote:
> On Thu, Aug 23, 2012 at 03:01:38PM +1000, Dave Chinner wrote:
> > From: Mel Gorman <mgorman@suse.de>
> > 
> > Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4
> > 
> > Direct reclaim should never writeback pages.  For now, handle the
> > situation and warn about it.  Ultimately, this will be a BUG_ON.
> 
> Is this actually the case on 3.0-stable?
> 

No, it is not. AFAIK, 3.0-stable does not contain [ee72886d: mm: vmscan:
do not writeback filesystem pages in direct reclaim] which is the absolute
minimum required for commit [94054fa3: xfs: warn if direct reclaim tries
to writeback pages] to make sense.

-- 
Mel Gorman
SUSE Labs

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Mark Tinguely]

On 08/23/12 00:01, Dave Chinner wrote:

> For all the XFS developers that have copious amounts of free time
> available, I'd appreciate an eye run over the patch list to see if
> there's any potential bug fixes that I missed or have made glaring
> errors in backporting. Some of the fixes are dependent on cleanups I
> haven't included, so some of the patches are a bit different to what
> is in mainline (e.g. anything that touches setattr). Most important
> to look at is probably the inode i_size changes and the logging of
> all metadata changes.


I think Brian Foster's log hang patch would be a good candidate:
xfs: check for stale inode before acquiring iflock on push
http://oss.sgi.com/archives/xfs/2012-06/msg00134.html

This hang can happen in Linux 2.6.x version of XFS.

Thanks for the work.

--Mark.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Mark Tinguely]

On 09/05/12 06:32, Mel Gorman wrote:
> On Mon, Aug 27, 2012 at 02:17:42PM -0400, Christoph Hellwig wrote:
>> On Thu, Aug 23, 2012 at 03:01:38PM +1000, Dave Chinner wrote:
>>> From: Mel Gorman<mgorman@suse.de>
>>>
>>> Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4
>>>
>>> Direct reclaim should never writeback pages.  For now, handle the
>>> situation and warn about it.  Ultimately, this will be a BUG_ON.
>>
>> Is this actually the case on 3.0-stable?
>>
>
> No, it is not. AFAIK, 3.0-stable does not contain [ee72886d: mm: vmscan:
> do not writeback filesystem pages in direct reclaim] which is the absolute
> minimum required for commit [94054fa3: xfs: warn if direct reclaim tries
> to writeback pages] to make sense.
>

I hit this warning testing on a linux-30.42 with a x86_64.

WARNING: at fs/xfs/linux-2.6/xfs_aops.c:961 xfs_vm_writepage+0x63c/0x6a0()
Hardware name: S2721-533 Thunder i7501 Pro
Modules linked in: ext4 jbd2 crc16
Pid: 12122, comm: cp Not tainted 3.0.42 #2
Call Trace:
  [<c1688bca>] ? printk+0x18/0x1a
  [<c103a3ad>] warn_slowpath_common+0x6d/0xa0
  [<c122afdc>] ? xfs_vm_writepage+0x63c/0x6a0
  [<c122afdc>] ? xfs_vm_writepage+0x63c/0x6a0
  [<c103a3fd>] warn_slowpath_null+0x1d/0x20
  [<c122afdc>] xfs_vm_writepage+0x63c/0x6a0
  [<c10901fd>] ? call_rcu_sched+0xd/0x10
  [<c128ac8c>] ? radix_tree_delete+0x1cc/0x280
  [<c110eb43>] ? free_buffer_head+0x23/0x30
  [<c110ebaa>] ? try_to_free_buffers+0x5a/0x90
  [<c10c2a82>] shrink_page_list+0x542/0x750
  [<c10ca984>] ? __mod_zone_page_state+0x54/0x60
  [<c10c3000>] shrink_inactive_list+0x190/0x360
  [<c10c35e0>] shrink_zone+0x410/0x580
  [<c10c40b0>] do_try_to_free_pages+0x100/0x360
  [<c10c4439>] try_to_free_pages+0x89/0x120
  [<c10baadb>] __alloc_pages_nodemask+0x3eb/0x650
  [<c10e423c>] new_slab+0x4c/0x1a0
  [<c168ad23>] __slab_alloc.constprop.65+0xed/0x1d5
  [<c10f21f6>] ? getname_flags+0x26/0x150
  [<c10e4f6b>] kmem_cache_alloc+0x11b/0x130
  [<c10675bf>] ? tick_dev_program_event+0x3f/0x160
  [<c10f21f6>] ? getname_flags+0x26/0x150
  [<c10f21f6>] getname_flags+0x26/0x150
  [<c10676fc>] ? tick_program_event+0x1c/0x30
  [<c10f60bf>] user_path_at_empty+0x1f/0x80
  [<c1090c98>] ? rcu_irq_exit+0x8/0x10
  [<c1040a07>] ? irq_exit+0x37/0x90
  [<c101a834>] ? smp_apic_timer_interrupt+0x54/0x90
  [<c10f613a>] user_path_at+0x1a/0x20
  [<c10ed12e>] vfs_fstatat+0x4e/0x80
  [<c10ed17b>] vfs_lstat+0x1b/0x20
  [<c10ed431>] sys_lstat64+0x11/0x30
  [<c10e74ff>] ? filp_close+0x4f/0x70
  [<c10e7577>] ? sys_close+0x57/0xa0
  [<c1690f25>] syscall_call+0x7/0xb

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Mel Gorman]

On Thu, Sep 13, 2012 at 03:12:16PM -0500, Mark Tinguely wrote:
> On 09/05/12 06:32, Mel Gorman wrote:
> >On Mon, Aug 27, 2012 at 02:17:42PM -0400, Christoph Hellwig wrote:
> >>On Thu, Aug 23, 2012 at 03:01:38PM +1000, Dave Chinner wrote:
> >>>From: Mel Gorman<mgorman@suse.de>
> >>>
> >>>Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4
> >>>
> >>>Direct reclaim should never writeback pages.  For now, handle the
> >>>situation and warn about it.  Ultimately, this will be a BUG_ON.
> >>
> >>Is this actually the case on 3.0-stable?
> >>
> >
> >No, it is not. AFAIK, 3.0-stable does not contain [ee72886d: mm: vmscan:
> >do not writeback filesystem pages in direct reclaim] which is the absolute
> >minimum required for commit [94054fa3: xfs: warn if direct reclaim tries
> >to writeback pages] to make sense.
> >
> 
> I hit this warning testing on a linux-30.42 with a x86_64.
> 

Just to be clear, this was linux 3.0.42 with this series of patches
applied on top, right?

> WARNING: at fs/xfs/linux-2.6/xfs_aops.c:961 xfs_vm_writepage+0x63c/0x6a0()
> Hardware name: S2721-533 Thunder i7501 Pro
> Modules linked in: ext4 jbd2 crc16
> Pid: 12122, comm: cp Not tainted 3.0.42 #2

I ask because on 3.0.42 this warning is not present. This patch should
not be merged to 3.0-stable.

-- 
Mel Gorman
SUSE Labs

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Mark Tinguely]

On 09/14/12 04:45, Mel Gorman wrote:
> On Thu, Sep 13, 2012 at 03:12:16PM -0500, Mark Tinguely wrote:
>> On 09/05/12 06:32, Mel Gorman wrote:
>>> On Mon, Aug 27, 2012 at 02:17:42PM -0400, Christoph Hellwig wrote:
>>>> On Thu, Aug 23, 2012 at 03:01:38PM +1000, Dave Chinner wrote:
>>>>> From: Mel Gorman<mgorman@suse.de>
>>>>>
>>>>> Upstream commit: 94054fa3fca1fd78db02cb3d68d5627120f0a1d4
>>>>>
>>>>> Direct reclaim should never writeback pages.  For now, handle the
>>>>> situation and warn about it.  Ultimately, this will be a BUG_ON.
>>>>
>>>> Is this actually the case on 3.0-stable?
>>>>
>>>
>>> No, it is not. AFAIK, 3.0-stable does not contain [ee72886d: mm: vmscan:
>>> do not writeback filesystem pages in direct reclaim] which is the absolute
>>> minimum required for commit [94054fa3: xfs: warn if direct reclaim tries
>>> to writeback pages] to make sense.
>>>
>>
>> I hit this warning testing on a linux-30.42 with a x86_64.
>>
>
> Just to be clear, this was linux 3.0.42 with this series of patches
> applied on top, right?
>
>> WARNING: at fs/xfs/linux-2.6/xfs_aops.c:961 xfs_vm_writepage+0x63c/0x6a0()
>> Hardware name: S2721-533 Thunder i7501 Pro
>> Modules linked in: ext4 jbd2 crc16
>> Pid: 12122, comm: cp Not tainted 3.0.42 #2
>
> I ask because on 3.0.42 this warning is not present. This patch should
> not be merged to 3.0-stable.
>

Correct. Linux-3.0.42 with Dave 102 patches, I removed patch 94 and 95 
per list comments. I added Brian Foster's patch as 103 <below>

I am testing xfstest 109 and 273 in a loop for shutdown worker races - 
none found yet. That message happened in the first loop, and never 
occurred again.

And the machine was a x86_32 not 64.

patch 103:


Subject: xfs: check for stale inode before acquiring iflock on push
From: Brian Foster <bfoster@redhat.com>
Upstream commit: 9a3a5dab63461b84213052888bf38a962b22d035

     An inode in the AIL can be flush locked and marked stale if
     a cluster free transaction occurs at the right time. The
     inode item is then marked as flushing, which causes xfsaild
     to spin and leaves the filesystem stalled. This is
     reproduced by running xfstests 273 in a loop for an
     extended period of time.

     Check for stale inodes before the flush lock. This marks
     the inode as pinned, leads to a log flush and allows the
     filesystem to proceed.

     Signed-off-by: Brian Foster <bfoster@redhat.com>
     Reviewed-by: Dave Chinner <dchinner@redhat.com>
     Reviewed-by: Christoph Hellwig <hch@lst.de>
     Reviewed-by: Mark Tinguely <tinguely@sgi.com>
     Signed-off-by: Ben Myers <bpm@sgi.com>


Index: linux-3.0.42/fs/xfs/xfs_inode_item.c
===================================================================
--- linux-3.0.42.orig/fs/xfs/xfs_inode_item.c
+++ linux-3.0.42/fs/xfs/xfs_inode_item.c
@@ -507,6 +507,21 @@ xfs_inode_item_trylock(
  	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
  		return XFS_ITEM_LOCKED;

+	/*
+	 * Re-check the pincount now that we stabilized the value by
+	 * taking the ilock.
+	 */
+	if (xfs_ipincount(ip) > 0) {
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+		return XFS_ITEM_PINNED;
+	}
+
+	/* Stale items should force out the iclog */
+	if (ip->i_flags & XFS_ISTALE) {
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+		return XFS_ITEM_PINNED;
+	}
+
  	if (!xfs_iflock_nowait(ip)) {
  		/*
  		 * inode has already been flushed to the backing buffer,
@@ -516,13 +531,6 @@ xfs_inode_item_trylock(
  		return XFS_ITEM_PUSHBUF;
  	}

-	/* Stale items should force out the iclog */
-	if (ip->i_flags & XFS_ISTALE) {
-		xfs_ifunlock(ip);
-		xfs_iunlock(ip, XFS_ILOCK_SHARED);
-		return XFS_ITEM_PINNED;
-	}
-
  #ifdef DEBUG
  	if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
  		ASSERT(iip->ili_fields != 0);

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 020/102] xfs: warn if direct reclaim tries to writeback pages

[Mark Tinguely]

On 09/14/12 07:56, Mark Tinguely wrote:
> On 09/14/12 04:45, Mel Gorman wrote:
>> On Thu, Sep 13, 2012 at 03:12:16PM -0500, Mark Tinguely wrote:

>>> WARNING: at fs/xfs/linux-2.6/xfs_aops.c:961
>>> xfs_vm_writepage+0x63c/0x6a0()
>>> Hardware name: S2721-533 Thunder i7501 Pro
>>> Modules linked in: ext4 jbd2 crc16
>>> Pid: 12122, comm: cp Not tainted 3.0.42 #2
>>
>> I ask because on 3.0.42 this warning is not present. This patch should
>> not be merged to 3.0-stable.
>>
>
> Correct. Linux-3.0.42 with Dave 102 patches, I removed patch 94 and 95
> per list comments. I added Brian Foster's patch as 103 <below>
>
> I am testing xfstest 109 and 273 in a loop for shutdown worker races -
> none found yet. That message happened in the first loop, and never
> occurred again.
>
> And the machine was a x86_32 not 64.

PS. of course it reports only once... that is what the patch does.
     reboot the machine and it displays on first call to test 273.

  Does not happen on the x86_64 machine. Until I can duplicate it on
  another x86_32, less us assume there is something weird with that
  machine.

--Mark.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Mark Tinguely]

On 08/23/12 00:01, Dave Chinner wrote:
> This series is a backport of all the major bug fixes in the current
> TOT kernel to the current 3.0.x stable tree.
>
> I won't make any secret of this - the fixes and supporting patches
> have been selected as a result of the issues reported in the current
> RHEL XFS codebase which currently is 2 commits short of the 3.0 code
> base.  With that said, it doesn't take a brain surgeon to work out
> the motivations behind this work and eventual destination of the
> patch set. ;)
>
> There's no guarantee I have caught every single bug fix that has
> been made since 3.0, but I've tried to grab all the bug fix commits
> as indicated by the commit headers (hence the importance of good one
> line bug summaries).
>
> Over the past couple of weeks of testing and refining, I've had only
> three significant problems arise from QA and load testing:
>
> 	1) An unreproducable log space hang
> 	2) An unmount panic due to buffers not being cleaned up
> 	before tearing down the perag tree
> 	3) A forced shutdown panic in block_invalidatepage()
> 	via xfs_aops_discard_page()
>
> It's entirely possible that #1 was due to the CIL space hang we
> still haven't got to the bottom of, so i'm not not greatly concerned
> by that. #2 implies I haven't quite backported the shutdown ordering
> fixes correctly (or I missed one), so I have a bit more work to do
> there. And for #3 - I've never seen that before and I haven't been
> able to reproduce it, so I really don't know what potential cause or
> impact it has.
>
> I've been beating on the series with xfstests, dbench, fsmark,
> postmark, compilebench and a few other load scripts that I've got,
> and it seems fairly resilient.  Hence, it's time to give the series
> wider testing and review to flush out any remaining issues.
>
> For all the folks that run 3.0.x stable kernels, I'd appreciate it if
> you could give this a whirl on your test systems to see if there are
> any obvious, glaring problems that show up under your particular
> workloads. This woul dbe of great benefit to me before I submit the
> series to the stable kernel gurus - I'd prefer it there's more
> substantial testing than "i've done what I can" when sending them
> the series.
>
> For all the XFS developers that have copious amounts of free time
> available, I'd appreciate an eye run over the patch list to see if
> there's any potential bug fixes that I missed or have made glaring
> errors in backporting. Some of the fixes are dependent on cleanups I
> haven't included, so some of the patches are a bit different to what
> is in mainline (e.g. anything that touches setattr). Most important
> to look at is probably the inode i_size changes and the logging of
> all metadata changes.
>
> Enjoy!
>
> Cheers,
>
> Dave.
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs

This looks great. Like I said earlier, I did not find Brian Foster's log 
patch:
   xfs: check for stale inode before acquiring iflock on push
   Upstream commit: 9a3a5dab63461b84213052888bf38a962b22d035
sample implementation listed on:
   http://oss.sgi.com/archives/xfs/2012-09/msg00188.html

Reviewed-by: Mark Tinguely <tinguely@sgi.com>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Dave Chinner]

On Tue, Sep 18, 2012 at 08:59:04AM -0500, Mark Tinguely wrote:
> This looks great. Like I said earlier, I did not find Brian Foster's
> log patch:
>   xfs: check for stale inode before acquiring iflock on push
>   Upstream commit: 9a3a5dab63461b84213052888bf38a962b22d035
> sample implementation listed on:
>   http://oss.sgi.com/archives/xfs/2012-09/msg00188.html
> 
> Reviewed-by: Mark Tinguely <tinguely@sgi.com>

Thatnks for looking over this, Mark.

The above patch does not directly apply to the 3.0.x branch because
the rework of the log item lock/push logic in the AIL was not
included in the series. Hence I'm not sure that backportingthis
patch is necessary because the problem only arose after we change
the locking/push logic...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC, PATCH 0/102]: xfs: 3.0.x stable kernel update

[Mark Tinguely]

On 09/18/12 18:50, Dave Chinner wrote:
> On Tue, Sep 18, 2012 at 08:59:04AM -0500, Mark Tinguely wrote:
>> This looks great. Like I said earlier, I did not find Brian Foster's
>> log patch:
>>    xfs: check for stale inode before acquiring iflock on push
>>    Upstream commit: 9a3a5dab63461b84213052888bf38a962b22d035
>> sample implementation listed on:
>>    http://oss.sgi.com/archives/xfs/2012-09/msg00188.html
>>
>> Reviewed-by: Mark Tinguely<tinguely@sgi.com>
>
> Thatnks for looking over this, Mark.
>
> The above patch does not directly apply to the 3.0.x branch because
> the rework of the log item lock/push logic in the AIL was not
> included in the series. Hence I'm not sure that backportingthis
> patch is necessary because the problem only arose after we change
> the locking/push logic...
>
> Cheers,
>
> Dave.

Hi Dave.

The original problem was in Linux 2.6.X.

The patch in the above link is a 3.0.42 port of Brian's top of tree
patch. This version places the tests in AIL trylock routine. We have
been using it on Linux 3.0.x for a couple months and it has avoided
the hangs that we used to get before.

Thank-you again for the work.

--Mark T.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs