linux-ext4 - Re: [PATCH 3/4] ext4: Add batched discard support for ext4

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:	Tue, 26 Oct 2010 14:43:05 +0200 (CEST)
From:	Lukas Czerner <lczerner@...hat.com>
To:	"Ted Ts'o" <tytso@....edu>
cc:	Lukas Czerner <lczerner@...hat.com>, linux-ext4@...r.kernel.org,
	rwheeler@...hat.com, sandeen@...hat.com, adilger@...ger.ca,
	Dmitry Monakhov <dmonakhov@...nvz.org>
Subject: Re: [PATCH 3/4] ext4: Add batched discard support for ext4

On Mon, 25 Oct 2010, Ted Ts'o wrote:

> On Mon, Sep 27, 2010 at 04:09:59PM +0200, Lukas Czerner wrote:
> > diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> > index 93eb6c2..80a5139 100644
> > --- a/fs/ext4/mballoc.c
> > +++ b/fs/ext4/mballoc.c
>       ....
> > +	struct ext4_buddy e4b;
> > +	ext4_fsblk_t first_group, last_group;
> 
> This should be ext4_group_t, shouldn't it?  

Right, it should be. Sorry about that.

> 
> > +	/* Determine first and last group to examine based on start and len */
> > +	first_group = (start - le32_to_cpu(es->s_first_data_block)) /
> > +		      EXT4_BLOCKS_PER_GROUP(sb);
> > +	last_group = (start + len - le32_to_cpu(es->s_first_data_block)) /
> > +		     EXT4_BLOCKS_PER_GROUP(sb);
> 
> I've tried compiling this for 32-bit x86, and this blows up because
> you can't divide long long's in the kernel.  (This is what do_div is
> for, and it's why ext4_get_group_no_and_offset() exists.)

I am ashamed, I probably should test patches on different architectures.
Thanks.

> 
> > +	first_block = (start - le32_to_cpu(es->s_first_data_block)) %
> > +			EXT4_BLOCKS_PER_GROUP(sb);
> > +	last_block = EXT4_BLOCKS_PER_GROUP(sb);
> 
> This means that the ext4 FITRIM ioctl will trim to the end of the
> blockgroup, and not just to the last block specified by the user.  Is
> this intentional?

If the group is NOT last group, or (start+len) is aligned to the
EXT4_BLOCK_PER_GROUP() boundary we will trim all blocks in this
particular block group. Otherwise we will know how much we need to trim
in this group to satisfy user request

	if (len >= EXT4_BLOCKS_PER_GROUP(sb))
		len -= (EXT4_BLOCKS_PER_GROUP(sb) - first_block);
	else
		last_block = len;

because we do keep track of how many block we need to trim by
decreasing len.

> 
> Also, it looks like there's nothing to check for the last blockgroup,
> where the last block might be less than a grpblk_t offset of
> EXT4_BLOCKS_PER_GROUP()?

This is not a problem, because when traversing the bitmap we will hit
the end of the group anyway, because those blocks (out of filesystem) are
marked as used in the bitmap and hence:

	start = mb_find_next_zero_bit(bitmap, max, start);
		if (start >= max)
			break;

will end the traversing without any attempt to trim blocks out of
filesystem boundary.

> 
> 						- Ted
> 

-Lukas
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html