linux-ext4 - Re: [PATCH -v2 2/2] ext4: simulate various I/O and checksum errors when reading metadata

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: <8504AF0E-39F8-4C56-86EE-9945E15C1A16@dilger.ca>
Date:   Thu, 5 Dec 2019 12:05:30 -0700
From:   Andreas Dilger <adilger@...ger.ca>
To:     Theodore Ts'o <tytso@....edu>
Cc:     Ext4 Developers List <linux-ext4@...r.kernel.org>
Subject: Re: [PATCH -v2 2/2] ext4: simulate various I/O and checksum errors
 when reading metadata

On Dec 3, 2019, at 8:23 PM, Theodore Ts'o <tytso@....edu> wrote:
> 
> This allows us to test various error handling code paths
> 
> Previous-Version-Link: https://lore.kernel.org/r/20191121183036.29385-2-tytso@mit.edu
> Signed-off-by: Theodore Ts'o <tytso@....edu>
> ---
> 
> +/*
> + * Simulate_fail flags
> + */
> +#define EXT4_SIM_BBITMAP_EIO	0x00000001
> +#define EXT4_SIM_BBITMAP_CRC	0x00000002
> +#define EXT4_SIM_IBITMAP_EIO	0x00000004
> +#define EXT4_SIM_IBITMAP_CRC	0x00000008
> +#define EXT4_SIM_INODE_EIO	0x00000010
> +#define EXT4_SIM_INODE_CRC	0x00000020
> +#define EXT4_SIM_DIRBLOCK_EIO	0x00000040
> +#define EXT4_SIM_DIRBLOCK_CRC	0x00000080

Do we really need to have the ability to inject several different failures
at the same time?  This will only allow 32 or 64 different failure hooks.

IMHO it would be much more flexible to have these as an enum of independent
values.  That does somewhat limit the ability to inject multiple failures,
but allows for many more different/specific fault injection points to be
added in the future.  We have hundreds of different fault points in Lustre
to simulate hard-to-hit race conditions and trigger specific error paths.

If this patch has already been landed, it would still be possible to change
ext4_simulate_fail() to check "if (likely(old != flag))" rather than just
the bit.  That allows making this into an enum that just happens to have an
non-consecutive sequence of values assigned rather than a bitmask.

Cheers, Andreas

Download attachment "signature.asc" of type "application/pgp-signature" (874 bytes)