linux-ext4 - Re: jbd2-abort-instead-of-waiting-for-nonexistent-transactions.patch

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]

Message-ID: <e9e943910809230956n3f36e544hc46a80c00ff9804a@mail.gmail.com>
Date:	Tue, 23 Sep 2008 17:56:27 +0100
From:	"Duane Griffin" <duaneg@...da.com>
To:	"Andrew Morton" <akpm@...ux-foundation.org>
Cc:	"Theodore Ts'o" <tytso@....edu>,
	"Stephen C. Tweedie" <sct@...hat.com>, linux-ext4@...r.kernel.org
Subject: Re: jbd2-abort-instead-of-waiting-for-nonexistent-transactions.patch

2008/9/22 Andrew Morton <akpm@...ux-foundation.org>:
> Guys, I have a note here that this might be needed in 2.6.27.
>
> I also have a note that Stephen had issues with it, but I
> don't recall what they were.

Stephen suggested that it would be better to sanity check the journal
start/end pointers on mount, rather than catching the error later like
this. I never quite convinced myself I'd worked out the right way to
do that, sorry. Perhaps someone would like to confirm (or otherwise)
whether or not the following is correct:

In journal_reset (?) check that:

journal->j_first == 1 (this seems to be the only valid value)

and

journal->j_last >= JFS_MIN_JOURNAL_BLOCKS

Additionally, it should be possible to check the journal->j_last more
precisely. For internal journals it seems straight-forward, we can
just check that journal->j_last == inode->i_size >>
inode->i_sb->s_blocksize_bits. For external journals we'd need to load
the device's superblock and check journal->j_last == s_blocks_count.

> Can we get this sorted out please?

If the above is confirmed I'll send a patch to that effect for jdb,
jdb2 and for e2fsprogs (fsck doesn't check j_first/j_last either).

Regardless, I think the original patch may be a good idea. It improves
robustness and matches the other locations where we call
jbd2_log_do_checkpoint. They are all in loops that test that
journal->j_checkpoint_transactions != NULL.

Cheers,
Duane.

-- 
"I never could learn to drink that blood and call it wine" - Bob Dylan
--
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