[PATCH] Accurate task and cpu time accounting

[Olof Johansson]

Hi,

On Wed, Feb 22, 2006 at 10:35:30PM +1100, Paul Mackerras wrote:

>  #ifndef __ASSEMBLY__
> @@ -313,7 +315,7 @@ enum {
>  	    CPU_FTR_HPTE_TABLE | CPU_FTR_PPCAS_ARCH_V2 |
>  	    CPU_FTR_MMCRA | CPU_FTR_SMT |
>  	    CPU_FTR_COHERENT_ICACHE | CPU_FTR_LOCKLESS_TLBIE |
> -	    CPU_FTR_MMCRA_SIHV,
> +	    CPU_FTR_MMCRA_SIHV | CPU_FTR_PURR,
>  	CPU_FTRS_CELL = CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_USE_TB |
>  	    CPU_FTR_HPTE_TABLE | CPU_FTR_PPCAS_ARCH_V2 |
>  	    CPU_FTR_ALTIVEC_COMP | CPU_FTR_MMCRA | CPU_FTR_SMT |
> @@ -326,7 +328,7 @@ enum {
>  #ifdef __powerpc64__
>  	    CPU_FTRS_POWER3 | CPU_FTRS_RS64 | CPU_FTRS_POWER4 |
>  	    CPU_FTRS_PPC970 | CPU_FTRS_POWER5 | CPU_FTRS_CELL |
> -            CPU_FTR_CI_LARGE_PAGE |
> +            CPU_FTR_CI_LARGE_PAGE | CPU_FTR_PURR |

Is this second change really needed (this is the setting of
CPU_FTRS_POSSIBLE)? It already includes CPU_FTRS_POWER5, which has the
bit set by the first change.

Only case I can see where it's mandated to include there is when the bit
is set based on device-tree contents, right?

CPU_FTR_CI_LARGE_PAGE seems to be a weird case, it's checked only in
one location (hash code that enables 64K pages) but never actually set
anywhere in current sources. It never seems to have been.


-Olof