2009-02-02 10:04:29

by Manish Katiyar

Subject: [PATCH] ext2: Mark the buffer as unmapped before reading the next quota block

Hi Jan,

This is in continuation with the previous patch
(http://patchwork.ozlabs.org/patch/20275/) . We also need to set the
buffer as unmapped before reading the next block. I don't think quota
files can have holes, but atleast for correctness, otherwise once we
have read a block, tmp_bh is always mapped even for holes.


Signed-off-by: Manish Katiyar <[email protected]>
---
fs/ext2/super.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index da8bdea..b4e7633 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -1328,6 +1328,7 @@ static ssize_t ext2_quota_read(struct
super_block *sb, int type, char *data,
sb->s_blocksize - offset : toread;

tmp_bh.b_state = 0;
+ clear_buffer_mapped(&tmp_bh);
err = ext2_get_block(inode, blk, &tmp_bh, 0);
if (err < 0)
return err;
--
1.5.4.3


Thanks -
Manish

2009-02-02 10:34:31

by Jan Kara

Subject: Re: [PATCH] ext2: Mark the buffer as unmapped before reading the next quota block

On Mon 02-02-09 15:34:27, Manish Katiyar wrote:
> Hi Jan,
>
> This is in continuation with the previous patch
> (http://patchwork.ozlabs.org/patch/20275/) . We also need to set the
> buffer as unmapped before reading the next block. I don't think quota
> files can have holes, but atleast for correctness, otherwise once we
> have read a block, tmp_bh is always mapped even for holes.
>
>
> Signed-off-by: Manish Katiyar <[email protected]>
> ---
> fs/ext2/super.c | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/fs/ext2/super.c b/fs/ext2/super.c
> index da8bdea..b4e7633 100644
> --- a/fs/ext2/super.c
> +++ b/fs/ext2/super.c
> @@ -1328,6 +1328,7 @@ static ssize_t ext2_quota_read(struct
> super_block *sb, int type, char *data,
> sb->s_blocksize - offset : toread;
>
> tmp_bh.b_state = 0;
> + clear_buffer_mapped(&tmp_bh);
I don't get this. When b_state is 0, then in particuler the buffer is not
mapped. So why clear the mapped bit explicitely?

> err = ext2_get_block(inode, blk, &tmp_bh, 0);
> if (err < 0)
> return err;

Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR

2009-02-02 11:05:23

by Manish Katiyar

Subject: Re: [PATCH] ext2: Mark the buffer as unmapped before reading the next quota block

On Mon, Feb 2, 2009 at 4:04 PM, Jan Kara <[email protected]> wrote:
> On Mon 02-02-09 15:34:27, Manish Katiyar wrote:
>> Hi Jan,
>>
>> This is in continuation with the previous patch
>> (http://patchwork.ozlabs.org/patch/20275/) . We also need to set the
>> buffer as unmapped before reading the next block. I don't think quota
>> files can have holes, but atleast for correctness, otherwise once we
>> have read a block, tmp_bh is always mapped even for holes.
>>
>>
>> Signed-off-by: Manish Katiyar <[email protected]>
>> ---
>> fs/ext2/super.c | 1 +
>> 1 files changed, 1 insertions(+), 0 deletions(-)
>>
>> diff --git a/fs/ext2/super.c b/fs/ext2/super.c
>> index da8bdea..b4e7633 100644
>> --- a/fs/ext2/super.c
>> +++ b/fs/ext2/super.c
>> @@ -1328,6 +1328,7 @@ static ssize_t ext2_quota_read(struct
>> super_block *sb, int type, char *data,
>> sb->s_blocksize - offset : toread;
>>
>> tmp_bh.b_state = 0;
>> + clear_buffer_mapped(&tmp_bh);
> I don't get this. When b_state is 0, then in particuler the buffer is not
> mapped. So why clear the mapped bit explicitely?

Sorry for the noise.......my bad :-( !!

Thanks -
Manish

>
>> err = ext2_get_block(inode, blk, &tmp_bh, 0);
>> if (err < 0)
>> return err;
>
> Honza
> --
> Jan Kara <[email protected]>
> SUSE Labs, CR
>