linux-ext4 - Re: [RFCv2 4/4] ext4: Move ext4_fiemap to use iomap infrastructure

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 30 Jan 2020 05:24:14 +0530
From:   Ritesh Harjani <riteshh@...ux.ibm.com>
To:     "Darrick J. Wong" <darrick.wong@...cle.com>
Cc:     jack@...e.cz, tytso@....edu, adilger.kernel@...ger.ca,
        linux-ext4@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        hch@...radead.org, cmaiolino@...hat.com
Subject: Re: [RFCv2 4/4] ext4: Move ext4_fiemap to use iomap infrastructure



On 1/29/20 9:48 PM, Darrick J. Wong wrote:
> On Wed, Jan 29, 2020 at 11:49:38AM +0530, Ritesh Harjani wrote:
>> Hello Darrick,
>>
>> On 1/28/20 8:58 PM, Darrick J. Wong wrote:
>>> On Tue, Jan 28, 2020 at 03:48:28PM +0530, Ritesh Harjani wrote:
>>>> Since ext4 already defines necessary iomap_ops required to move to iomap
>>>> for fiemap, so this patch makes those changes to use existing iomap_ops
>>>> for ext4_fiemap and thus removes all unwanted code.
>>>>
>>>> Signed-off-by: Ritesh Harjani <riteshh@...ux.ibm.com>
>>>> ---
>>>>    fs/ext4/extents.c | 279 +++++++---------------------------------------
>>>>    fs/ext4/inline.c  |  41 -------
>>>>    2 files changed, 38 insertions(+), 282 deletions(-)
>>>>
>>>> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
>>>> index 0de548bb3c90..901caee2fcb1 100644
>>>> --- a/fs/ext4/extents.c
>>>> +++ b/fs/ext4/extents.c
>>>
>>> <snip> Just a cursory glance...
>>>
>>>> @@ -5130,40 +4927,42 @@ static int ext4_xattr_fiemap(struct inode *inode,
>>>>    				EXT4_I(inode)->i_extra_isize;
>>>>    		physical += offset;
>>>>    		length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
>>>> -		flags |= FIEMAP_EXTENT_DATA_INLINE;
>>>>    		brelse(iloc.bh);
>>>>    	} else { /* external block */
>>>>    		physical = (__u64)EXT4_I(inode)->i_file_acl << blockbits;
>>>>    		length = inode->i_sb->s_blocksize;
>>>>    	}
>>>> -	if (physical)
>>>> -		error = fiemap_fill_next_extent(fieinfo, 0, physical,
>>>> -						length, flags);
>>>> -	return (error < 0 ? error : 0);
>>>> +	iomap->addr = physical;
>>>> +	iomap->offset = 0;
>>>> +	iomap->length = length;
>>>> +	iomap->type = IOMAP_INLINE;
>>>> +	iomap->flags = 0;
>>>
>>> Er... external "ACL" blocks aren't IOMAP_INLINE.
>>
>> Sorry, I should have mentioned about this too in the cover letter.
>> So current patchset is mainly only converting bmap & fiemap to use iomap
>> APIs. Even the original implementation does not have external ACL block
>> implemented for xattr_fiemap.
> 
> Er ... yes it did.  The "} else { /* external block */" block sets
> physical to i_file_acl.

Oops.. my bad. I got it confused with EA inode feature.
Urghh... I should remove my bias while looking at a review comment.

so I think for i_file_acl (external block) we should set
iomap->type = IOMAP_MAPPED.

Will fix this and submit in the same thread. Thanks for catching it.

> 
>> Let me spend some time to implement it. But I would still like to keep
>> that as a separate patch.
>>
>> But thanks for looking into it. There's this point 2.a & 2.b mentioned in
>> the cover letter where I could really use your help in understanding
>> if all of that is a known behavior from iomap_fiemap side
>> (whenever you have some time of course :) )
> 
> i'll go have a look.

Thanks.


> 
> --D
> 
>> -ritesh
>>