linux-kernel - Re: [PATCH] arch/x86/kernel/microcode_core.c: add missing platform_device

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: <20110824163354.GC21066@gere.osrc.amd.com>
Date:	Wed, 24 Aug 2011 18:33:54 +0200
From:	Borislav Petkov <bp@...64.org>
To:	Julia Lawall <julia@...u.dk>
CC:	Tigran Aivazian <tigran@...azian.fsnet.co.uk>,
	<kernel-janitors@...r.kernel.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	"H. Peter Anvin" <hpa@...or.com>, <x86@...nel.org>,
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] arch/x86/kernel/microcode_core.c: add missing
 platform_device_unregister

On Wed, Aug 24, 2011 at 05:44:08PM +0200, Julia Lawall wrote:
> That seems like a fairly big change.  Is it desirable to move the calls to 
> sysdev_driver_register and microcode_dev_init up over the get_online_cpus 
> ... put_online_cpus sequence?

Well, you need microcode_pdev for request_firmware() which happens
when you do sysdev_driver_register() so it has to run after
platform_device_register_simple().

And the microcode_dev_init() thing got moved further
down in 871b72dd1e12afc3f024479531d25a9339d2e3f9 but its
counterpart microcode_dev_exit() is mistakenly called if
platform_device_register_simple() fails which is clearly another bug.

And frankly, microcode_dev_init() is strictly needed only for the
CONFIG_MICROCODE_OLD_INTERFACE. And on the driver destroy path
microcode_exit() is calling microcode_dev_exit() first so it could be
that 871b72dd1e12afc3f024479531d25a9339d2e3f9 mistakenly moved the
microcode_dev_init() call down.

I guess you could move the microcode_dev_init() call up, before
platform_device_register_simple() but this needs testing on Intel
because they still can do microcode loading over the old interface using
microcode_ctl userspace helper...

Or, you can do a bit cleaner version of the goto-with-labels thing by
adding a helper like this:

static void __microcode_sysdev_unregister(void)
{
        get_online_cpus();
        mutex_lock(&microcode_mutex);

        sysdev_driver_unregister(&cpu_sysdev_class, &mc_sysdev_driver);

        mutex_unlock(&microcode_mutex);
        put_online_cpus();
}

and call it both on the error-out path in microcode_init(), and in
microcode_exit().

Oh boy, the rabbit hole turns out to be pretty deep once you go down
into it :-).

Btw, I totally understand if you've lost all desire to fix that properly
after all this :-).

Thanks.

-- 
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551

--
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/