All of lore.kernel.org
 help / color / mirror / Atom feed
From: Brian Foster <bfoster@redhat.com>
To: linux-xfs@vger.kernel.org
Subject: [PATCH v2 6/7] xfs: refactor inode chunk alloc/free tx reservation
Date: Thu, 30 Nov 2017 13:58:35 -0500	[thread overview]
Message-ID: <20171130185836.18481-7-bfoster@redhat.com> (raw)
In-Reply-To: <20171130185836.18481-1-bfoster@redhat.com>

The reservation for the various forms of inode allocation is
scattered across several different functions. This includes two
variants of chunk allocation (v5 icreate transactions vs. older
create transactions) and the inode free transaction.

To clean up some of this code and clarify the purpose of specific
allocfree reservations, continue the pattern of defining helper
functions for smaller operational units of broader transactions.
Refactor the reservation into an inode chunk alloc/free helper that
considers the various conditions based on filesystem format.

An inode chunk free involves an extent free and buffer
invalidations. The latter requires reservation for log headers only.
An inode chunk allocation modifies the free space btrees and logs
the chunk on v4 supers. v5 supers initialize the inode chunk using
ordered buffers and so do not log the chunk.

As a side effect of this refactoring, add one more allocfree res to
the ifree transaction. Technically this does not serve a specific
purpose because inode chunks are freed via deferred operations and
thus occur after a transaction roll. tr_ifree has a bit of a history
of tx overruns caused by too many agfl fixups during sustained file
deletion workloads, so add this extra reservation as a form of
padding nonetheless.

Signed-off-by: Brian Foster <bfoster@redhat.com>
---
 fs/xfs/libxfs/xfs_trans_resv.c | 67 ++++++++++++++++++++++++++++++++----------
 1 file changed, 52 insertions(+), 15 deletions(-)

diff --git a/fs/xfs/libxfs/xfs_trans_resv.c b/fs/xfs/libxfs/xfs_trans_resv.c
index 19f3a226a357..432dd7d7afea 100644
--- a/fs/xfs/libxfs/xfs_trans_resv.c
+++ b/fs/xfs/libxfs/xfs_trans_resv.c
@@ -34,6 +34,9 @@
 #include "xfs_trans_space.h"
 #include "xfs_trace.h"
 
+#define _ALLOC	true
+#define _FREE	false
+
 /*
  * A buffer has a format structure overhead in the log in addition
  * to the data, so we need to take this into account when reserving
@@ -172,6 +175,41 @@ xfs_calc_finobt_res(
 }
 
 /*
+ * Calculate the reservation required to allocate or free an inode chunk. This
+ * includes:
+ *
+ * the allocation btrees: 2 trees * (max depth - 1) * block size
+ * the inode chunk: m_ialloc_blks * N
+ *
+ * The size N of the inode chunk reservation depends on whether it is for
+ * allocation or free and which type of create transaction is in use. An inode
+ * chunk free always invalidates the buffers and only requires reservation for
+ * headers (N == 0). An inode chunk allocation requires a chunk sized
+ * reservation on v4 and older superblocks to initialize the chunk. No chunk
+ * reservation is required for allocation on v5 supers, which use ordered
+ * buffers to initialize.
+ */
+STATIC uint
+xfs_calc_inode_chunk_res(
+	struct xfs_mount	*mp,
+	bool			alloc)
+{
+	uint			res, size = 0;
+
+	res = xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
+			       XFS_FSB_TO_B(mp, 1));
+	if (alloc) {
+		/* icreate tx uses ordered buffers */
+		if (xfs_sb_version_hascrc(&mp->m_sb))
+			return res;
+		size = XFS_FSB_TO_B(mp, 1);
+	}
+
+	res += xfs_calc_buf_res(mp->m_ialloc_blks, size);
+	return res;
+}
+
+/*
  * Various log reservation values.
  *
  * These are based on the size of the file system block because that is what
@@ -379,8 +417,7 @@ xfs_calc_create_resv_modify(
  * For create we can allocate some inodes giving:
  *    the agi and agf of the ag getting the new inodes: 2 * sectorsize
  *    the superblock for the nlink flag: sector size
- *    the inode blocks allocated: mp->m_ialloc_blks * blocksize
- *    the allocation btrees: 2 trees * (max depth - 1) * block size
+ *    the inode chunk (allocation/init)
  *    the inode btree (record insertion)
  */
 STATIC uint
@@ -389,9 +426,7 @@ xfs_calc_create_resv_alloc(
 {
 	return xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
 		mp->m_sb.sb_sectsize +
-		xfs_calc_buf_res(mp->m_ialloc_blks, XFS_FSB_TO_B(mp, 1)) +
-		xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
-				 XFS_FSB_TO_B(mp, 1)) +
+		xfs_calc_inode_chunk_res(mp, _ALLOC) +
 		xfs_calc_inobt_res(mp);
 }
 
@@ -408,7 +443,7 @@ __xfs_calc_create_reservation(
  * For icreate we can allocate some inodes giving:
  *    the agi and agf of the ag getting the new inodes: 2 * sectorsize
  *    the superblock for the nlink flag: sector size
- *    the allocation btrees: 2 trees * (max depth - 1) * block size
+ *    the inode chunk (allocation, no init)
  *    the inobt (record insertion)
  *    the finobt (record insertion)
  */
@@ -418,8 +453,7 @@ xfs_calc_icreate_resv_alloc(
 {
 	return xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
 		mp->m_sb.sb_sectsize +
-		xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
-				 XFS_FSB_TO_B(mp, 1)) +
+		xfs_calc_inode_chunk_res(mp, _ALLOC) +
 		xfs_calc_inobt_res(mp) +
 		xfs_calc_finobt_res(mp);
 }
@@ -485,15 +519,15 @@ xfs_calc_symlink_reservation(
  *    the inode being freed: inode size
  *    the super block free inode counter, AGF and AGFL: sector size
  *    the on disk inode (agi unlinked list removal)
- *    the inode chunk is marked stale (headers only)
+ *    the inode chunk (invalidated, headers only)
  *    the inode btree
  *    the finobt (record insertion, removal or modification)
  *
- * Note that the allocfree res. for the inode chunk itself is not included
- * because the extent free occurs after a transaction roll. We could take the
- * maximum of the pre/post roll operations, but the pre-roll reservation already
- * includes at least one allocfree res. for the inobt and is thus guaranteed to
- * be larger.
+ * Note that the inode chunk res. includes an allocfree res. for freeing of the
+ * inode chunk. This is technically extraneous because the inode chunk free is
+ * deferred (it occurs after a transaction roll). Include the extra reservation
+ * anyways since we've had reports of ifree transaction overruns due to too many
+ * agfl fixups during inode chunk frees.
  */
 STATIC uint
 xfs_calc_ifree_reservation(
@@ -503,7 +537,7 @@ xfs_calc_ifree_reservation(
 		xfs_calc_inode_res(mp, 1) +
 		xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
 		xfs_calc_iunlink_remove_reservation(mp) +
-		xfs_calc_buf_res(mp->m_ialloc_blks, 0) +
+		xfs_calc_inode_chunk_res(mp, _FREE) +
 		xfs_calc_inobt_res(mp) +
 		xfs_calc_finobt_res(mp);
 }
@@ -795,6 +829,9 @@ xfs_calc_sb_reservation(
 	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize);
 }
 
+#undef _ALLOC
+#undef _FREE
+
 void
 xfs_trans_resv_calc(
 	struct xfs_mount	*mp,
-- 
2.13.6


  parent reply	other threads:[~2017-11-30 18:58 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-11-30 18:58 [PATCH v2 0/7] xfs: inode transaction reservation fixups Brian Foster
2017-11-30 18:58 ` [PATCH v2 1/7] xfs: print transaction log reservation on overrun Brian Foster
2017-12-07 21:34   ` Darrick J. Wong
2017-11-30 18:58 ` [PATCH v2 2/7] xfs: include inobt buffers in ifree tx log reservation Brian Foster
2017-12-03 21:44   ` Dave Chinner
2017-12-07 21:40   ` Darrick J. Wong
2017-11-30 18:58 ` [PATCH v2 3/7] xfs: fix up agi unlinked list reservations Brian Foster
2017-12-03 21:45   ` Dave Chinner
2017-12-07 21:41   ` Darrick J. Wong
2017-11-30 18:58 ` [PATCH v2 4/7] xfs: truncate transaction does not modify the inobt Brian Foster
2017-12-03 21:46   ` Dave Chinner
2017-12-07 21:44   ` Darrick J. Wong
2017-11-30 18:58 ` [PATCH v2 5/7] xfs: include an allocfree res for inobt modifications Brian Foster
2017-12-07 21:47   ` Darrick J. Wong
2017-11-30 18:58 ` Brian Foster [this message]
2017-12-03 21:52   ` [PATCH v2 6/7] xfs: refactor inode chunk alloc/free tx reservation Dave Chinner
2017-12-04 12:17     ` Brian Foster
2017-12-04 12:21   ` [PATCH v3 " Brian Foster
2017-12-07 21:53     ` Darrick J. Wong
2017-11-30 18:58 ` [PATCH v2 7/7] xfs: eliminate duplicate icreate tx reservation functions Brian Foster
2017-12-03 21:54   ` Dave Chinner
2017-12-07 21:57   ` Darrick J. Wong
2018-01-08 14:08 ` [PATCH v2 0/7] xfs: inode transaction reservation fixups Brian Foster
2018-01-08 18:06   ` Darrick J. Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20171130185836.18481-7-bfoster@redhat.com \
    --to=bfoster@redhat.com \
    --cc=linux-xfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.