From: Steven Whitehouse <swhiteho@redhat.com>
To: cluster-devel@redhat.com, linux-kernel@vger.kernel.org
Cc: Benjamin Marzinski <bmarzins@redhat.com>,
Steven Whitehouse <swhiteho@redhat.com>
Subject: [PATCH 11/15] GFS2: fix block allocation check for fallocate
Date: Tue, 15 Mar 2011 09:11:41 +0000 [thread overview]
Message-ID: <1300180305-5993-12-git-send-email-swhiteho@redhat.com> (raw)
In-Reply-To: <1300180305-5993-1-git-send-email-swhiteho@redhat.com>
From: Benjamin Marzinski <bmarzins@redhat.com>
GFS2 fallocate wasn't properly checking if a blocks were already allocated.
In write_empty_blocks(), if a page didn't have buffer_heads attached, GFS2
was always treating it as if there were no blocks allocated for that page.
GFS2 now calls gfs2_block_map() to check if the blocks are allocated before
writing them out.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 2878481..4074b95 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -622,8 +622,7 @@ static void empty_write_end(struct page *page, unsigned from,
{
struct gfs2_inode *ip = GFS2_I(page->mapping->host);
- page_zero_new_buffers(page, from, to);
- flush_dcache_page(page);
+ zero_user(page, from, to-from);
mark_page_accessed(page);
if (!gfs2_is_writeback(ip))
@@ -632,36 +631,43 @@ static void empty_write_end(struct page *page, unsigned from,
block_commit_write(page, from, to);
}
-static int write_empty_blocks(struct page *page, unsigned from, unsigned to)
+static int needs_empty_write(sector_t block, struct inode *inode)
{
- unsigned start, end, next;
- struct buffer_head *bh, *head;
int error;
+ struct buffer_head bh_map = { .b_state = 0, .b_blocknr = 0 };
- if (!page_has_buffers(page)) {
- error = __block_write_begin(page, from, to - from, gfs2_block_map);
- if (unlikely(error))
- return error;
+ bh_map.b_size = 1 << inode->i_blkbits;
+ error = gfs2_block_map(inode, block, &bh_map, 0);
+ if (unlikely(error))
+ return error;
+ return !buffer_mapped(&bh_map);
+}
- empty_write_end(page, from, to);
- return 0;
- }
+static int write_empty_blocks(struct page *page, unsigned from, unsigned to)
+{
+ struct inode *inode = page->mapping->host;
+ unsigned start, end, next, blksize;
+ sector_t block = page->index << (PAGE_CACHE_SHIFT - inode->i_blkbits);
+ int ret;
- bh = head = page_buffers(page);
+ blksize = 1 << inode->i_blkbits;
next = end = 0;
while (next < from) {
- next += bh->b_size;
- bh = bh->b_this_page;
+ next += blksize;
+ block++;
}
start = next;
do {
- next += bh->b_size;
- if (buffer_mapped(bh)) {
+ next += blksize;
+ ret = needs_empty_write(block, inode);
+ if (unlikely(ret < 0))
+ return ret;
+ if (ret == 0) {
if (end) {
- error = __block_write_begin(page, start, end - start,
- gfs2_block_map);
- if (unlikely(error))
- return error;
+ ret = __block_write_begin(page, start, end - start,
+ gfs2_block_map);
+ if (unlikely(ret))
+ return ret;
empty_write_end(page, start, end);
end = 0;
}
@@ -669,13 +675,13 @@ static int write_empty_blocks(struct page *page, unsigned from, unsigned to)
}
else
end = next;
- bh = bh->b_this_page;
+ block++;
} while (next < to);
if (end) {
- error = __block_write_begin(page, start, end - start, gfs2_block_map);
- if (unlikely(error))
- return error;
+ ret = __block_write_begin(page, start, end - start, gfs2_block_map);
+ if (unlikely(ret))
+ return ret;
empty_write_end(page, start, end);
}
--
1.7.4
next prev parent reply other threads:[~2011-03-15 9:34 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-15 9:11 GFS2: Pre-pull patch posting (merge window) Steven Whitehouse
2011-03-15 9:11 ` [PATCH 01/15] GFS2: Use RCU for glock hash table Steven Whitehouse
2011-03-15 9:11 ` [PATCH 02/15] GFS2: Post-VFS scale update for RCU path walk Steven Whitehouse
2011-03-15 9:11 ` [PATCH 03/15] GFS2: Fix glock queue trace point Steven Whitehouse
2011-03-15 9:11 ` [PATCH 04/15] GFS2: Improve cluster mmap scalability Steven Whitehouse
2011-03-15 9:11 ` [PATCH 05/15] GFS2: panics on quotacheck update Steven Whitehouse
2011-03-15 9:11 ` [PATCH 06/15] GFS2: deallocation performance patch Steven Whitehouse
2011-03-15 9:11 ` [PATCH 07/15] GFS2: quota allows exceeding hard limit Steven Whitehouse
2011-03-15 9:11 ` [PATCH 08/15] GFS2: Fix glock deallocation race Steven Whitehouse
2011-03-15 9:11 ` [PATCH 09/15] GFS2: Remove potential race in flock code Steven Whitehouse
2011-03-15 9:11 ` [PATCH 10/15] GFS2: Optimize glock multiple-dequeue code Steven Whitehouse
2011-03-15 9:11 ` Steven Whitehouse [this message]
2011-03-15 9:11 ` [PATCH 12/15] GFS2: introduce AIL lock Steven Whitehouse
2011-03-15 9:11 ` [PATCH 13/15] GFS2: Update to AIL list locking Steven Whitehouse
2011-03-15 9:11 ` [PATCH 14/15] GFS2: Adding missing unlock_page() Steven Whitehouse
2011-03-15 9:11 ` [PATCH 15/15] GFS2: Don't use _raw version of RCU dereference Steven Whitehouse
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=1300180305-5993-12-git-send-email-swhiteho@redhat.com \
--to=swhiteho@redhat.com \
--cc=bmarzins@redhat.com \
--cc=cluster-devel@redhat.com \
--cc=linux-kernel@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®