Re: [V2][PATCH] New MMU feature for new hpte format in ISA3

On 15/06/16 15:42, Aneesh Kumar K.V wrote:

Balbir Singh [off-list ref] writes:

quoted

On 15/06/16 10:38, Balbir Singh wrote:

quoted

Thanks for a review, I'll do a V2 with some changes

Balbir

Michael, please review and check if you like the #ifdef,
since we are crossing the 32 bit boundary, I've used a
conditional define to select the feature.

We really don't need to make it unsigned long. Take a look at

http://mid.gmane.org/1465395958-21349-9-git-send-email-aneesh.kumar@linux.vnet.ibm.com

The comment in the file states that


/*
 * First half is MMU families
 */

So effectively the first 16 bits are for families. Unless we decide to override
that decision. One can also debate the description of what a MMU family is :)

But I still am not sure why we need an mmu feature for selecting new hpte
format. Why would the exiting ISA 3.0 check won't work?

I thought I answered it in my previous email. The check for CPU_FTR_ARCH_300 is
too large and we don't want to use it unless advertised via ibm,pa-features.

For the pa-feature 58 byte look at
http://mid.gmane.org/1465397003-26812-17-git-send-email-aneesh.kumar@linux.vnet.ibm.com