From: Ted Ts'o <tytso@mit.edu>
To: Sami Liedes <sami.liedes@iki.fi>
Cc: linux-ext4@vger.kernel.org
Subject: Re: [PATCH 4/5] libext2fs: Implement ext2fs_find_first_zero_generic_bmap().
Date: Sun, 25 Mar 2012 22:39:07 -0400 [thread overview]
Message-ID: <20120326023907.GB16642@thunk.org> (raw)
In-Reply-To: <20120310213740.GO6961@sli.dy.fi>
On Sat, Mar 10, 2012 at 11:37:40PM +0200, Sami Liedes wrote:
> @@ -128,16 +129,27 @@ errcode_t ext2fs_new_inode(ext2_filsys fs, ext2_ino_t dir,
> return EXT2_ET_INODE_ALLOC_FAIL;
> i = start_inode;
> modulo = (i - 1) % EXT2_INODES_PER_GROUP(fs->super);
> -
> do {
> if (modulo == 0)
> check_inode_uninit(fs, map, (i - 1) /
> EXT2_INODES_PER_GROUP(fs->super));
>
> - if (!ext2fs_fast_test_inode_bitmap2(map, i))
> + upto = i + (EXT2_INODES_PER_GROUP(fs->super) - modulo);
> + if (i < start_inode && upto >= start_inode)
> + upto = start_inode - 1;
> + if (upto > fs->super->s_inodes_count)
> + upto = fs->super->s_inodes_count;
> +
> + err = ext2fs_find_first_zero_inode_bitmap2(map, i, upto, &first_zero);
> + if (!err) {
> + i = first_zero;
> break;
> - if (++modulo == EXT2_INODES_PER_GROUP(fs->super))
> - modulo = 0;
> + } else {
> + if (err != ENOENT)
> + return EXT2_ET_INODE_ALLOC_FAIL;
> + i = upto;
> + }
> +
> if (++i > fs->super->s_inodes_count) {
> i = EXT2_FIRST_INODE(fs->super);
> modulo = (i - 1) % EXT2_INODES_PER_GROUP(fs->super);
There a bug in this code. modulo (which I've renamed to ino_in_group
per Andreas's suggestion) isn't getting updated, and after the first
pass through the loop modulo needs to be zero. Otherwise the code
will end up spanning group boundaries and check_inode_uninit won't get
called at the right times.
I've fixed this up and applied it, with some changes to make sure
ext2fs_new_inode() is easier to understand, and hence audit for
correctness. It now looks like this:
errcode_t ext2fs_new_inode(ext2_filsys fs, ext2_ino_t dir,
int mode EXT2FS_ATTR((unused)),
ext2fs_inode_bitmap map, ext2_ino_t *ret)
{
ext2_ino_t start_inode = 0;
ext2_ino_t i, ino_in_group, upto, first_zero;
errcode_t retval;
dgrp_t group;
EXT2_CHECK_MAGIC(fs, EXT2_ET_MAGIC_EXT2FS_FILSYS);
if (!map)
map = fs->inode_map;
if (!map)
return EXT2_ET_NO_INODE_BITMAP;
if (dir > 0) {
group = (dir - 1) / EXT2_INODES_PER_GROUP(fs->super);
start_inode = (group * EXT2_INODES_PER_GROUP(fs->super)) + 1;
}
if (start_inode < EXT2_FIRST_INODE(fs->super))
start_inode = EXT2_FIRST_INODE(fs->super);
if (start_inode > fs->super->s_inodes_count)
return EXT2_ET_INODE_ALLOC_FAIL;
i = start_inode;
do {
ino_in_group = (i - 1) % EXT2_INODES_PER_GROUP(fs->super);
group = (i - 1) / EXT2_INODES_PER_GROUP(fs->super);
check_inode_uninit(fs, map, group);
upto = i + (EXT2_INODES_PER_GROUP(fs->super) - ino_in_group);
if (i < start_inode && upto >= start_inode)
upto = start_inode - 1;
if (upto > fs->super->s_inodes_count)
upto = fs->super->s_inodes_count;
retval = ext2fs_find_first_zero_inode_bitmap2(map, i, upto,
&first_zero);
if (retval == 0) {
i = first_zero;
break;
}
if (retval != ENOENT)
return EXT2_ET_INODE_ALLOC_FAIL;
i = upto + 1;
if (i > fs->super->s_inodes_count)
i = EXT2_FIRST_INODE(fs->super);
} while (i != start_inode);
if (ext2fs_test_inode_bitmap2(map, i))
return EXT2_ET_INODE_ALLOC_FAIL;
*ret = i;
return 0;
}
- Ted
next prev parent reply other threads:[~2012-03-26 3:18 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-03-10 21:33 [PATCH 0/5] Make filesystem shrinking faster and less CPU-intensive Sami Liedes
2012-03-10 21:34 ` [PATCH 1/5] libext2fs: Move a modulo operation out of a hot loop Sami Liedes
2012-03-11 9:50 ` Andreas Dilger
2012-03-10 21:35 ` [PATCH 2/5] resize2fs: Use EXT2_FLAG_64BITS Sami Liedes
2012-03-10 21:36 ` [PATCH 3/5] libext2fs: Document EXT2_FLAG_64BITS in ext2fs_open2() Sami Liedes
2012-03-10 21:37 ` [PATCH 4/5] libext2fs: Implement ext2fs_find_first_zero_generic_bmap() Sami Liedes
2012-03-11 9:51 ` Andreas Dilger
2012-03-12 19:15 ` Sami Liedes
2012-03-12 23:09 ` Andreas Dilger
2012-03-23 22:33 ` Ted Ts'o
2012-03-26 13:53 ` Sami Liedes
2012-03-26 15:34 ` Ted Ts'o
2012-03-26 2:39 ` Ted Ts'o [this message]
2012-03-10 21:38 ` [PATCH 5/5] libext2fs: Implement fast find_first_zero() for bitarray bitmaps Sami Liedes
2012-03-26 2:34 ` Ted Ts'o
2012-03-26 13:22 ` Sami Liedes
2012-03-26 15:26 ` Ted Ts'o
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=20120326023907.GB16642@thunk.org \
--to=tytso@mit.edu \
--cc=linux-ext4@vger.kernel.org \
--cc=sami.liedes@iki.fi \
/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;
as well as URLs for NNTP newsgroup(s).