From: Brian Foster <bfoster@redhat.com>
To: xfs@oss.sgi.com
Subject: [PATCH 15/18] xfs: only free allocated regions of inode chunks
Date: Thu, 24 Jul 2014 10:23:05 -0400 [thread overview]
Message-ID: <1406211788-63206-16-git-send-email-bfoster@redhat.com> (raw)
In-Reply-To: <1406211788-63206-1-git-send-email-bfoster@redhat.com>
An inode chunk is currently added to the transaction free list based on
a simple fsb conversion and hardcoded chunk length. The nature of sparse
chunks is such that the physical chunk of inodes on disk may consist of
one or more discontiguous parts. Blocks that reside in the holes of the
inode chunk are not inodes and could be allocated to any other use or
not allocated at all.
Refactor the existing xfs_bmap_add_free() call into the
xfs_difree_inode_chunk() helper. The new helper uses the existing
calculation if a chunk is not sparse. Otherwise, use the inobt record
holemask to free the contiguous regions of the chunk.
Signed-off-by: Brian Foster <bfoster@redhat.com>
---
fs/xfs/libxfs/xfs_ialloc.c | 64 +++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 61 insertions(+), 3 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index f75f191..1be57b1 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -1659,6 +1659,66 @@ out_error:
return error;
}
+/*
+ * Free the blocks of an inode chunk. We must consider that the inode chunk
+ * might be sparse and only free the regions that are allocated as part of the
+ * chunk.
+ */
+STATIC void
+xfs_difree_inode_chunk(
+ struct xfs_mount *mp,
+ xfs_agnumber_t agno,
+ struct xfs_inobt_rec_incore *rec,
+ struct xfs_bmap_free *flist)
+{
+ xfs_agblock_t sagbno = XFS_AGINO_TO_AGBNO(mp, rec->ir_startino);
+ xfs_agblock_t agbno;
+ int nextbit;
+ int contig, contigblk;
+ __uint16_t allocmask;
+ uint allocbitmap;
+
+ if (!xfs_inobt_issparse(rec)) {
+ /* not sparse, calculate extent info directly */
+ xfs_bmap_add_free(XFS_AGB_TO_FSB(mp, agno,
+ XFS_AGINO_TO_AGBNO(mp, rec->ir_startino)),
+ mp->m_ialloc_blks, flist, mp);
+ return;
+ }
+
+ /*
+ * The bit flip and type conversion are intentionally done separately
+ * here to zero-extend the bitmask.
+ */
+ allocmask = ~rec->ir_holemask;
+ allocbitmap = allocmask;
+
+ /*
+ * We now have an allocation bitmap in units of inodes at sparse chunk
+ * granularity (e.g., more than one inode per bit). Use the bitmask
+ * functions to find each contigious range of bits in the map. For each
+ * range, convert the start bit and count to block values and use that
+ * data to add the associated extent to the free list.
+ */
+ nextbit = xfs_next_bit(&allocbitmap, 1, 0);
+ while (nextbit != -1) {
+ agbno = (nextbit * XFS_INODES_PER_SPCHUNK) /
+ mp->m_sb.sb_inopblock;
+ agbno += sagbno;
+
+ contig = xfs_contig_bits(&allocbitmap, 1, nextbit);
+ contigblk = (contig * XFS_INODES_PER_SPCHUNK) /
+ mp->m_sb.sb_inopblock;
+
+ ASSERT(agbno % xfs_ialloc_cluster_alignment(mp) == 0);
+ ASSERT(contigblk % xfs_ialloc_cluster_alignment(mp) == 0);
+ xfs_bmap_add_free(XFS_AGB_TO_FSB(mp, agno, agbno), contigblk,
+ flist, mp);
+
+ nextbit = xfs_next_bit(&allocbitmap, 1, nextbit + contig + 1);
+ }
+}
+
STATIC int
xfs_difree_inobt(
struct xfs_mount *mp,
@@ -1750,9 +1810,7 @@ xfs_difree_inobt(
goto error0;
}
- xfs_bmap_add_free(XFS_AGB_TO_FSB(mp, agno,
- XFS_AGINO_TO_AGBNO(mp, rec.ir_startino)),
- mp->m_ialloc_blks, flist, mp);
+ xfs_difree_inode_chunk(mp, agno, &rec, flist);
} else {
*deleted = 0;
--
1.8.3.1
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
next prev parent reply other threads:[~2014-07-24 14:23 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-07-24 14:22 [PATCH RFC 00/18] xfs: sparse inode chunks Brian Foster
2014-07-24 14:22 ` [PATCH 01/18] xfs: refactor xfs_inobt_insert() to eliminate loop and support variable count Brian Foster
2014-07-24 22:10 ` Dave Chinner
2014-07-28 16:03 ` Brian Foster
2014-07-28 23:32 ` Dave Chinner
2014-07-29 14:43 ` Brian Foster
2014-07-24 14:22 ` [PATCH 02/18] xfs: pass xfs_mount directly to xfs_ialloc_cluster_alignment() Brian Foster
2014-07-24 14:22 ` [PATCH 03/18] xfs: define sparse inode chunks v5 sb feature bit and helper function Brian Foster
2014-07-24 17:08 ` Mark Tinguely
2014-07-24 17:37 ` Brian Foster
2014-07-24 18:38 ` Mark Tinguely
2014-07-24 19:38 ` Brian Foster
2014-07-24 23:35 ` Dave Chinner
2014-07-24 14:22 ` [PATCH 04/18] xfs: introduce inode record hole mask for sparse inode chunks Brian Foster
2014-07-24 22:14 ` Dave Chinner
2014-07-28 16:16 ` Brian Foster
2014-08-07 15:18 ` Brian Foster
2014-07-24 14:22 ` [PATCH 05/18] xfs: create macros/helpers for dealing with " Brian Foster
2014-07-24 22:13 ` Dave Chinner
2014-07-24 14:22 ` [PATCH 06/18] xfs: pass inode count through ordered icreate log item Brian Foster
2014-07-24 14:22 ` [PATCH 07/18] xfs: handle sparse inode chunks in icreate log recovery Brian Foster
2014-07-24 14:22 ` [PATCH 08/18] xfs: create helper to manage record overlap for sparse inode chunks Brian Foster
2014-07-24 22:41 ` Dave Chinner
2014-07-28 16:19 ` Brian Foster
2014-07-29 0:07 ` Dave Chinner
2014-07-29 15:10 ` Brian Foster
2014-07-24 14:22 ` [PATCH 09/18] xfs: allocate sparse inode chunks on full chunk allocation failure Brian Foster
2014-07-24 14:23 ` [PATCH 10/18] xfs: set sparse inodes feature bit when a sparse chunk is allocated Brian Foster
2014-07-24 22:46 ` Dave Chinner
2014-07-28 16:23 ` Brian Foster
2014-07-24 14:23 ` [PATCH 11/18] xfs: reduce min. inode allocation space requirement for sparse inode chunks Brian Foster
2014-07-24 22:50 ` Dave Chinner
2014-07-24 14:23 ` [PATCH 12/18] xfs: helper to convert inobt record holemask to inode alloc. bitmap Brian Foster
2014-07-24 23:21 ` Dave Chinner
2014-07-24 14:23 ` [PATCH 13/18] xfs: filter out sparse regions from individual inode allocation Brian Foster
2014-07-24 14:23 ` [PATCH 14/18] xfs: update free inode record logic to support sparse inode records Brian Foster
2014-07-24 14:23 ` Brian Foster [this message]
2014-07-24 23:24 ` [PATCH 15/18] xfs: only free allocated regions of inode chunks Dave Chinner
2014-07-24 14:23 ` [PATCH 16/18] xfs: skip unallocated regions of inode chunks in xfs_ifree_cluster() Brian Foster
2014-07-24 14:23 ` [PATCH 17/18] xfs: use actual inode count for sparse records in bulkstat/inumbers Brian Foster
2014-07-24 23:29 ` Dave Chinner
2014-07-24 14:23 ` [PATCH 18/18] xfs: enable sparse inode chunks for v5 superblocks Brian Foster
2014-07-24 23:34 ` Dave Chinner
2014-07-24 16:28 ` [PATCH RFC 00/18] xfs: sparse inode chunks Brian Foster
2014-07-24 22:32 ` Dave Chinner
2014-07-25 16:30 ` Brian Foster
2014-07-26 0:03 ` Dave Chinner
2014-07-28 12:14 ` Brian Foster
2014-07-29 0:26 ` Dave Chinner
2014-07-29 15:25 ` Brian Foster
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=1406211788-63206-16-git-send-email-bfoster@redhat.com \
--to=bfoster@redhat.com \
--cc=xfs@oss.sgi.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox