linux-kernel - As of kernel 4.4-rc1 segmentation fault if msr module not loaded. [REGRESSION][BISECTED]

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-next>] [day] [month] [year] [list]

Message-ID: <000301d12867$19ba3030$4d2e9090$@net>
Date:	Thu, 26 Nov 2015 08:25:43 -0800
From:	"Doug Smythies" <dsmythies@...us.net>
To:	"'Borislav Petkov'" <bp@...en8.de>
Cc:	"Doug Smythies" <dsmythies@...us.net>,
	"'Borislav Petkov'" <bp@...e.de>, <linux-kernel@...r.kernel.org>
Subject: As of kernel 4.4-rc1 segmentation fault if msr module not loaded. [REGRESSION][BISECTED]

Hi,

This might be considered a minor issue, but I submit my
findings anyhow.

In some distros, the msr module is not loaded by default.
(I guess not built in by default, would be more correct.)
I often forget to load it before using "rdmsr" or "wrmsr".
It was never an issue before, but as of Kernel 4.4-rc1
it causes a segmentation fault crash. Example:

# rdmsr -a 0x19a
Segmentation fault (core dumped)

It could be argued that the msr-tools commands should
check if the module is loaded and abort with some error
message if not (or, and as turbostat now does, just load
the module). However, what I do not know is if there
might be other effected use cases.

The kernel was bisected, and this is the result:

9a2bc335f100a0f6ee6392b9f97ac4188d84db1d is the first bad commit
commit 9a2bc335f100a0f6ee6392b9f97ac4188d84db1d
Author: Borislav Petkov <bp@...e.de>
Date:   Tue Oct 20 11:54:44 2015 +0200

    x86/microcode: Unmodularize the microcode driver

    Make CONFIG_MICROCODE a bool. It was practically a bool already anyway,
    since early loader was forcing it to =y.
    ...

Since I sometimes mess up using git bisect, and end up at some random result,
the above was double checked manually:

9a2bc335f100a0f6ee6392b9f97ac4188d84db1d has the issue
81ffdcdd97d94110627caa81c23d5d780083731d (the previous commit) does not have the issue.

As a potentially interesting side note, the segmentation fault
does not occur if I use sudo instead. Example:

$ sudo rdmsr -a 0x19a
$

Of course, if I load the module then things work as expected:

# modprobe msr
# rdmsr -a 0x19a
0
0
0
0
0
0
0
0

--
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/