[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ARM LPAE flavour (Was: Re: [PATCH] udebs for armmp flavour)



On Sat, 2013-08-31 at 10:10 +0100, Ian Campbell wrote:
[...]
> > > @@ -19,6 +20,18 @@ image-file: arch/arm/boot/zImage
> > >  hardware: ARMv7 multiplatform compatible SoCs
> > >  hardware-long: ARMv7 multiplatform kernel for Marvell Armada 370/xp, Freescale iMX5x/iMX6
> > >  
> > > +[armmp-lpae_build]
> > > +image-file: arch/arm/boot/zImage
> > > +
> > > +[armmp-lpae_description]
> > > +hardware: ARMv7 multiplatform compatible SoCs
> > [...]
> > 
> > The short description should be different from that for armmp as well.
> 
> append "... supporting LPAE"? (that's roughly what x86 does)

Right.

> That reminds me, the armmp description is:
>         
>         Description: Linux 3.11-rc5 for ARMv7 multiplatform compatible SoCs
>          The Linux kernel 3.11-rc5 and modules for use on ARMv7 multiplatform
>          kernel for Marvell Armada 370/xp, Freescale iMX5x/iMX6.
>         
> The list of hardware there is already well out of date, I think. Are we
> intending to strive to keep it accurate and enumerate them all or do we
> want to make it generic enough to not need further maintenance?

It currently lists the platforms that Arnaud thought were working well
in 3.9, and probably should be updated for 3.11.  It doesn't look like
this is going to be truly generic any time soon, since we need to
provide DTBs and build-in some platform-specific drivers.

[...]
> The lists are the same for regular vs. LPAE, however unlike on x86 XN
> ( == NX on x86) support is available even in the short page table
> format, so it doesn't make much sense to use LPAE on such systems with
> <4GB (or more likely less than ~3.5GB to allow for MMIO space) of RAM.
> I'm not sure if the overhead of managing that (i.e. remembering to
> disabling a platform for LPAE when enabling it for regular) is worth it
> though?
[...]

I'll leave that question to the ARM porters.

Ben.

-- 
Ben Hutchings
If God had intended Man to program,
we'd have been born with serial I/O ports.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: