linux-kernel - Re: module: add debugging alias parsing support

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: <20171130131710.ccccf4alzrnvmlp3@redbean>
Date:   Thu, 30 Nov 2017 14:17:11 +0100
From:   Jessica Yu <jeyu@...nel.org>
To:     "Luis R. Rodriguez" <mcgrof@...nel.org>
Cc:     jeyu@...hat.com, rusty@...tcorp.com.au, keescook@...omium.org,
        tixxdz@...il.com, mbenes@...e.cz, atomlin@...hat.com,
        pmladek@...e.com, hare@...e.com, james.l.morris@...cle.com,
        ebiederm@...ssion.com, davem@...emloft.net,
        akpm@...ux-foundation.org, torvalds@...ux-foundation.org,
        linux-kernel@...r.kernel.org
Subject: Re: module: add debugging alias parsing support

+++ Luis R. Rodriguez [29/11/17 18:36 -0800]:
>Debugging modules can often lead to an alias question. We purposely
>don't have alias parsing support upstream as this is all dealt with
>in userpace with the assumption that in-kernel we just process aliases
>and userspace Does The Right Thing (TM) about aliases.
>
>Obviously userspace can be buggy though, and it can lie to us. We
>currently have no easy way to determine this. Parsing aliases is
>an example debugging facility we can use to help with these sorts
>of problems.
>
>You can debug by adding to your dynamic debug:
>
>GRUB_CMDLINE_LINUX_DEFAULT="dyndbg=\"func module_process_aliases +p;\" "
>
>Upon boot for example here a few entries:
>
>module ext4 num_aliases: 5
>alias[0] = fs-ext4
>alias[1] = ext3
>alias[2] = fs-ext3
>alias[3] = ext2
>alias[4] = fs-ext2
>
>module xfs num_aliases: 1
>alias[0] = fs-xfs
>
>module floppy num_aliases: 3
>alias[0] = block-major-2-*
>alias[1] = acpi*:PNP0700:*
>alias[2] = pnp:dPNP0700*
>
>module ata_piix num_aliases: 89
>alias[0] = pci:v00008086d00008C81sv*sd*bc*sc*i*
>alias[1] = pci:v00008086d00008C80sv*sd*bc*sc*i*
>alias[2] = pci:v00008086d00008C89sv*sd*bc*sc*i*
>alias[3] = pci:v00008086d00008C88sv*sd*bc*sc*i*
>... etc ...
>
>Suggested-by: Linus Torvalds <torvalds@...ux-foundation.org>
>Signed-off-by: Luis R. Rodriguez <mcgrof@...nel.org>

Hi Luis,

Just some quick questions - are there any plans to use these in-kernel
module aliases anywhere else? Or are you using them just for debugging?

And if it's only for debugging, what's the benefit of printing the
aliases in-kernel when one could already use modinfo to print the
exact same information (namely, in-kernel module name, which was
recently added by Kees, + aliases)? We're essentially parsing the
.modinfo section of the module binary in two different places, just
that this patchset does it in-kernel and modinfo does it from
userspace, so I'm curious why we want to bring this into the kernel as
well without any additional users..

Thanks!

Jessica