linux-kernel - Re: [Bisected] linux-next 20141205 fails to mount ext4 fs on arm64 with CONFIG_ARM64_64K

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] [day] [month] [year] [list]

Message-ID: <20141205205800.GA10309@thunk.org>
Date:	Fri, 5 Dec 2014 15:58:00 -0500
From:	Theodore Ts'o <tytso@....edu>
To:	Dmitry Monakhov <dmonakhov@...nvz.org>
Cc:	Liviu Dudau <Liviu.Dudau@....com>,
	LKML <linux-kernel@...r.kernel.org>,
	Will Deacon <will.deacon@....com>,
	Catalin Marinas <catalin.marinas@....com>
Subject: Re: [Bisected] linux-next 20141205 fails to mount ext4 fs on arm64
 with CONFIG_ARM64_64K_PAGES=y

On Fri, Dec 05, 2014 at 10:54:20PM +0400, Dmitry Monakhov wrote:
> Liviu Dudau <Liviu.Dudau@....com> writes:
> 
> Yes. This is my crap.
> - sbi->s_group_desc[group_desc]
> + gd_bh = *rcu_dereference(sbi->s_group_desc) + group_desc;
> 
> Ted I have to admit that I'm an idiot.
> I already have fix where I've fixed the typo and complain from sparse
> which was reported by kbuild01. In order to prevent similar shame in future I've wrote several
> tests for resize2fs and unfortunately they are failed even w/o my
> patches. It is appeared that resize2fs is broken for big disks.
> So feel free to drop the patch, and give me some tome to investigate the
> issues. I'll back with the patch-set in 24hrs.

I had fixed the typo locally, but I'll drop this for now pending your
further investigations.  BTW, I think the reason why we didn't notice
is because for smaller disks, it just worked out that

	*rcu_dereference(sbi->s_group_desc) + group_desc == 
	*(rcu_dereference(sbi->s_group_desc) + group_desc)

by pure luck based on how the allocator works, since all of the group
descriptor bh's get allocated back to back.

Let me know about your resize2fs failures on big disks; I had found
fixed a large number of problems with online and offline resizes with
huge disks, so I had done quite a bit of testing in that area.  I
thought I had caught all of them, but if there are some additional
e2fsprogs or online resize bugs in the kernel bugs, I'd really like to
know.   Thanks!!

						- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/