From: Brian Foster <bfoster@redhat.com>
To: linux-xfs@vger.kernel.org
Subject: [PATCH v2 3/5] xfs: reuse iomap delalloc code for COW fork reservation
Date: Wed, 11 Jan 2017 12:54:07 -0500 [thread overview]
Message-ID: <1484157249-464-4-git-send-email-bfoster@redhat.com> (raw)
In-Reply-To: <1484157249-464-1-git-send-email-bfoster@redhat.com>
COW fork reservation (delayed allocation) is implemented in
xfs_reflink_reserve_cow() and is generally based on the traditional data
fork delalloc logic in xfs_file_iomap_begin_delay(). In preparation for
further changes to implement more aggressive COW fork preallocation,
refactor the COW reservation code to reuse xfs_file_iomap_begin_delay()
for data fork allocation as well as COW fork reservation. This patch
does not change behavior.
Signed-off-by: Brian Foster <bfoster@redhat.com>
---
fs/xfs/xfs_iomap.c | 74 ++++++++++++++++++++++++++++++++++++++++--------------
1 file changed, 55 insertions(+), 19 deletions(-)
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 8791ed5..19b7eb0 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -527,9 +527,8 @@ xfs_iomap_prealloc_size(
* inode data and COW forks. If an existing data extent is shared, determine
* whether COW fork reservation is necessary.
*
- * The 'found' parameter indicates whether a writable mapping was found. If the
- * mapping is shared, a COW reservation is performed for the corresponding
- * range.
+ * The 'found' parameter indicates whether a writable mapping was found. If a
+ * shared mapping exists, found is set only if COW reservation exists as well.
*/
static int
xfs_iomap_search_extents(
@@ -540,12 +539,14 @@ xfs_iomap_search_extents(
int *eof,
int *idx,
struct xfs_bmbt_irec *got,
+ bool *shared,
bool *found) /* found usable extent */
{
struct xfs_ifork *ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
int error = 0;
+ bool trimmed;
- *found = false;
+ *shared = *found = false;
/*
* Look up a preexisting extent directly into imap. Set got for the
@@ -556,17 +557,37 @@ xfs_iomap_search_extents(
*got = *imap;
return 0;
}
+ if (!xfs_is_reflink_inode(ip)) {
+ *found = true;
+ return 0;
+ }
- if (xfs_is_reflink_inode(ip)) {
- bool shared;
-
- xfs_trim_extent(imap, offset_fsb, end_fsb - offset_fsb);
- error = xfs_reflink_reserve_cow(ip, imap, &shared);
- if (error)
- return error;
+ /*
+ * Found a data extent but we don't know if it is shared. If an extent
+ * exists in the cow fork, assume that it is. Use got for this lookup as
+ * imap must retain the data mapping.
+ */
+ xfs_trim_extent(imap, offset_fsb, end_fsb - offset_fsb);
+ ifp = XFS_IFORK_PTR(ip, XFS_COW_FORK);
+ *eof = !xfs_iext_lookup_extent(ip, ifp, offset_fsb, idx, got);
+ if (!*eof && got->br_startoff <= imap->br_startoff) {
+ trace_xfs_reflink_cow_found(ip, got);
+ xfs_trim_extent(imap, got->br_startoff, got->br_blockcount);
+ *found = true;
+ return 0;
}
- *found = true;
+ /*
+ * There is no existing cow fork extent. We have to explicitly check if
+ * the data extent is shared to determine whether COW fork reservation
+ * is required to map the data extent. Trim the mapping to the next
+ * (un)shared boundary at the same time.
+ */
+ error = xfs_reflink_trim_around_shared(ip, imap, shared, &trimmed);
+ if (error)
+ return error;
+ if (!*shared)
+ *found = true;
return error;
}
@@ -587,11 +608,13 @@ xfs_file_iomap_begin_delay(
xfs_fileoff_t maxbytes_fsb =
XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
int error = 0, eof = 0;
+ int fork = XFS_DATA_FORK;
struct xfs_bmbt_irec imap;
struct xfs_bmbt_irec got;
xfs_extnum_t idx;
xfs_fsblock_t prealloc_blocks = 0;
bool found;
+ bool shared;
ASSERT(!XFS_IS_REALTIME_INODE(ip));
ASSERT(!xfs_get_extsz_hint(ip));
@@ -620,16 +643,20 @@ xfs_file_iomap_begin_delay(
/*
* Search for preexisting extents. If an existing data extent is shared,
- * this will perform COW fork reservation.
+ * switch to the COW fork for COW reservation.
*/
error = xfs_iomap_search_extents(ip, offset_fsb, end_fsb, &imap, &eof,
- &idx, &got, &found);
+ &idx, &got, &shared, &found);
if (error)
goto out_unlock;
if (found) {
trace_xfs_iomap_found(ip, offset, count, 0, &imap);
goto done;
}
+ if (shared) {
+ end_fsb = imap.br_startoff + imap.br_blockcount;
+ fork = XFS_COW_FORK;
+ }
error = xfs_qm_dqattach_locked(ip, 0);
if (error)
@@ -645,9 +672,10 @@ xfs_file_iomap_begin_delay(
* the lower level functions are updated.
*/
count = min_t(loff_t, count, 1024 * PAGE_SIZE);
- end_fsb = min(XFS_B_TO_FSB(mp, offset + count), maxbytes_fsb);
+ end_fsb = min(end_fsb, XFS_B_TO_FSB(mp, offset + count));
+ xfs_trim_extent(&imap, offset_fsb, end_fsb - offset_fsb);
- if (eof) {
+ if (eof && fork == XFS_DATA_FORK) {
prealloc_blocks = xfs_iomap_prealloc_size(ip, offset, count, idx);
if (prealloc_blocks) {
xfs_extlen_t align;
@@ -669,7 +697,7 @@ xfs_file_iomap_begin_delay(
}
retry:
- error = xfs_bmapi_reserve_delalloc(ip, XFS_DATA_FORK, offset_fsb,
+ error = xfs_bmapi_reserve_delalloc(ip, fork, offset_fsb,
end_fsb - offset_fsb, prealloc_blocks, &got, &idx, eof);
switch (error) {
case 0:
@@ -687,8 +715,16 @@ xfs_file_iomap_begin_delay(
goto out_unlock;
}
- trace_xfs_iomap_alloc(ip, offset, count, 0, &got);
- imap = got;
+ /*
+ * For the data fork, the iomap mapping is simply the extent that was
+ * returned from the delalloc request. Otherwise, use imap as it refers
+ * to the data fork but is trimmed according to the shared state.
+ */
+ if (fork == XFS_DATA_FORK) {
+ trace_xfs_iomap_alloc(ip, offset, count, 0, &got);
+ imap = got;
+ } else
+ trace_xfs_reflink_cow_alloc(ip, &got);
done:
if (isnullstartblock(imap.br_startblock))
imap.br_startblock = DELAYSTARTBLOCK;
--
2.7.4
next prev parent reply other threads:[~2017-01-11 17:54 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-01-11 17:54 [PATCH v2 0/5] xfs: basic cow fork speculative preallocation Brian Foster
2017-01-11 17:54 ` [PATCH v2 1/5] xfs: refactor iomap delalloc existing extent search into helper Brian Foster
2017-01-11 17:54 ` [PATCH v2 2/5] xfs: logically separate iomap range from allocation range Brian Foster
2017-01-11 17:54 ` Brian Foster [this message]
2017-01-13 17:13 ` [PATCH v2 3/5] xfs: reuse iomap delalloc code for COW fork reservation Christoph Hellwig
2017-01-16 16:26 ` Brian Foster
2017-01-16 17:37 ` Christoph Hellwig
2017-01-16 19:04 ` Brian Foster
2017-01-11 17:54 ` [PATCH v2 4/5] xfs: free cowblocks and retry on buffered write ENOSPC Brian Foster
2017-01-13 17:12 ` Christoph Hellwig
2017-01-11 17:54 ` [PATCH v2 5/5] xfs: implement basic COW fork speculative preallocation Brian Foster
2017-01-13 17:13 ` [PATCH v2 0/5] xfs: basic cow " Christoph Hellwig
2017-01-16 16:26 ` 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=1484157249-464-4-git-send-email-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.