linux-kernel - Re: [PATCH 07/10] cpufreq: Add support for x86 cpuinfo auto loading

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]

Date:	Thu, 8 Dec 2011 02:13:56 +0100
From:	Andi Kleen <andi@...stfloor.org>
To:	Dave Jones <davej@...hat.com>, Andi Kleen <andi@...stfloor.org>,
	linux-kernel@...r.kernel.org, kay.sievers@...y.org, trenn@...e.de,
	Andi Kleen <ak@...ux.intel.com>, hpa@...or.com
Subject: Re: [PATCH 07/10] cpufreq: Add support for x86 cpuinfo auto loading

On Wed, Dec 07, 2011 at 08:07:31PM -0500, Dave Jones wrote:
>  > +MODULE_DEVICE_TABLE(x86cpu, eps_cpu_id);
>  > +
>  >  static int __init eps_init(void)
>  >  {
>  > -	struct cpuinfo_x86 *c = &cpu_data(0);
>  > -
>  > -	/* This driver will work only on Centaur C7 processors with
>  > -	 * Enhanced SpeedStep/PowerSaver registers */
>  > -	if (c->x86_vendor != X86_VENDOR_CENTAUR
>  > -	    || c->x86 != 6 || c->x86_model < 10)
>  > -		return -ENODEV;
> 
> This makes e_powersaver bind to every family 6 VIA cpu.
> But the old logic only bound to certain models.
> Won't this will clash with this other driver if both are built ?

The code does

static int __init eps_init(void)
{
        if (!x86_match_cpu(eps_cpu_id) || boot_cpu_data.x86_model < 10)
                return -ENODEV;

So modprobe will load it, but if the CPU is too old it will just error out
again. I think that's reasonable.  There's no direct way current
to express a >= in the matches because modprobe uses fnmatch()

Also most likely the old CPUs won't have the EST bit anyways, then
it won't even be loaded.

> 
>  > diff --git a/drivers/cpufreq/longhaul.c b/drivers/cpufreq/longhaul.c
>  > index f47d26e..b4263ce 100644
>  > --- a/drivers/cpufreq/longhaul.c
>  > +++ b/drivers/cpufreq/longhaul.c
>  > 
>  > +static struct x86_cpu_id longhaul_id[] = {
>  > +	{ X86_VENDOR_CENTAUR, 6 },
>  > +	{}
>  > +};
>  > +MODULE_DEVICE_TABLE(x86cpu, longhaul_id);
> 
> iirc, the intention here was longhaul on cpus that don't have EST,
> and e_powersaver on those that do. Maybe an additional check for the
> absense of EST in longhaul's init code would do the trick.
> (sidenote: I don't recall why we even have e-powersaver, instead of them
>  just using acpi-cpufreq).

It's not done today, but I could add it. 

But I tried to keep the existing behaviour. AFAIK distros just load them
all right? This matches this. I have no way to test these CPUs so I would 
prefer to be as compatible as possible.
> 
>  > +static struct x86_cpu_id powernow_k8_ids[] = {
>  > +	{ X86_VENDOR_AMD, 0xf, },
>  > +	/* RED-PEN If HW PSTATE was a normal feature bit it could be matched here
>  > +	 * instead of a (limited) model list.
>  > +	 */
> 
> You mean make fake one in the identify code ? Do-able I guess.
> Though that would just be moving this list from this driver to a different place,
> and as this is the only place that cares..

Ok. 

Thanks for the review. I'll remove the comment.

-Andi

--
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/