Re: [PATCH v4 7/7] mtrr, mm, x86: Enhance MTRR checks for KVA huge page mapping

[Toshi Kani]

On Mon, 2015-05-11 at 23:42 +0200, Borislav Petkov wrote:
> On Mon, May 11, 2015 at 02:38:46PM -0600, Toshi Kani wrote:
> > MTRRs disabled is not an error case as it could be a normal
> > configuration on some platforms / BIOS setups.
> 
> Normal how? PAT-only systems? Examples please...

BIOS initializes and enables MTRRs at POST.  While the most (if not all)
BIOSes do it today, I do not think the x86 arch requires BIOS to enable
them.

Here is a quote from Intel SDM:
===
11.11.5 MTRR Initialization

On a hardware reset, the P6 and more recent processors clear the valid
flags in variable-range MTRRs and clear the E flag in the
IA32_MTRR_DEF_TYPE MSR to disable all MTRRs. All other bits in the MTRRs
are undefined.

Prior to initializing the MTRRs, software (normally the system BIOS)
must initialize all fixed-range and variablerange MTRR register fields
to 0. Software can then initialize the MTRRs according to known types of
memory, including memory on devices that it auto-configures.
Initialization is expected to occur prior to booting the operating
system.
===

> > I clarified it in the above comment that uniform is set for any return
> > value.
> 
> Hell no!
> 
> u8 mtrr_type_lookup(u64 start, u64 end, u8 *uniform)
> {
> 
> 	...
> 
>         *uniform = 1;
> 
>         if (!mtrr_state_set)
>                 return MTRR_TYPE_INVALID;
> 
>         if (!(mtrr_state.enabled & MTRR_STATE_MTRR_ENABLED))
>                 return MTRR_TYPE_INVALID;
> 
> 
> This is wrong and the fact that I still need to persuade you about it
> says a lot.
> 
> If you want to be able to state that a type is uniform even if MTRRs are
> disabled, you need to define another retval which means exactly that.

There may not be any type conflict with MTRR_TYPE_INVALID. 

> Or add an inline function called mtrr_enabled() and call it in the
> mtrr_type_lookup() callers.
> 
> Or whatever.
> 
> I don't want any confusing states with two return types and people
> having to figure out what it exactly means and digging into the code
> and scratching heads WTF is that supposed to mean.

I will change the caller to check MTRR_TYPE_INVALID, and treat it as a
uniform case.

Thanks,
-Toshi



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/