| Date | Wed, 11 May 2022 09:02:11 +0200 | From | Peter Zijlstra <> | Subject | Re: [RFCv2 04/10] x86/mm: Introduce X86_THREAD_LAM_U48 and X86_THREAD_LAM_U57 |
| |
On Wed, May 11, 2022 at 05:27:45AM +0300, Kirill A. Shutemov wrote:
> +#define LAM_NONE 0 > +#define LAM_U57 1 > +#define LAM_U48 2
> +#define X86_THREAD_LAM_U48 0x1 > +#define X86_THREAD_LAM_U57 0x2
Seriously pick an order and stick with it. I would suggest keeping the hardware order and then you can do:
> +static inline unsigned long lam_to_cr3(u8 lam) > +{ > + switch (lam) { > + case LAM_NONE: > + return 0; > + case LAM_U57: > + return X86_CR3_LAM_U57; > + case LAM_U48: > + return X86_CR3_LAM_U48; > + default: > + WARN_ON_ONCE(1); > + return 0; > + }
return (lam & 0x3) << X86_CR3_LAM_U57;
> +} > + > +static inline u8 cr3_to_lam(unsigned long cr3) > +{ > + if (cr3 & X86_CR3_LAM_U57) > + return LAM_U57; > + if (cr3 & X86_CR3_LAM_U48) > + return LAM_U48; > + return 0;
return (cr3 >> X86_CR3_LAM_U57) & 0x3;
> +}
and call it a day, or something.
I'm still not liking LAM(e), I'm thikning it's going to create more problems than it solves.
|