Re: [XenPPC] [PATCH 4 of 6] [PATCH] xen: implement guest_physmap_{add/remove}_page for ppc

[Hollis Blanchard <hollisb@xxxxxxxxxx>]

On Wed, 2007-02-21 at 18:17 -0500, Ryan Harper wrote:
> @@ -504,17 +508,15 @@ unsigned long mfn_to_gmfn(struct domain 
>          mfn < (rma_mfn + (1 << d->arch.rma_order)))
>          return mfn - rma_mfn;
> 
> -    /* Extent? */
> -    cur_pfn = 1UL << d->arch.rma_order;
> -    list_for_each_entry (pe, &d->arch.extent_list, pe_list) {
> -        uint pe_pages = 1UL << pe->order;
> -        uint b_mfn = page_to_mfn(pe->pg);
> -        uint e_mfn = b_mfn + pe_pages;
> -
> -        if (mfn >= b_mfn && mfn < e_mfn) {
> +    /* check extents (cpu-defined contiguous chunks after RMA) */
> +    cur_pfn = 1UL << d->arch.rma_order; /* start looking after RMA */
> +    for ( ; cur_pfn < d->max_pages; cur_pfn += ext_nrpages )
> +    {
> +        uint b_mfn = d->arch.p2m[cur_pfn];
> +        uint e_mfn = b_mfn + ext_nrpages;
> +
> +        if (mfn >= b_mfn && mfn < e_mfn)
>              return cur_pfn + (mfn - b_mfn);
> -        }
> -        cur_pfn += pe_pages;
>      }
>      return INVALID_M2P_ENTRY;
>  } 

I think you're splitting these patches up a lot more than necessary (to
the point I've having a hard time understanding them). Also, the above
code is just removed by the next patch! If you combine 4 and 5 I think
it will actually be smaller and easier to understand.

I didn't realize these were just RFC. When you resubmit, could you put a
little more description in each commit message?

-- 
Hollis Blanchard
IBM Linux Technology Center


_______________________________________________
Xen-ppc-devel mailing list
Xen-ppc-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ppc-devel