On May 18, 2005, at 9:14 AM, Dan Malek wrote:

On May 18, 2005, at 6:21 AM, Eugene Surovegin wrote:

quoted

This patch fixed CONFIG_TASK_SIZE handling on 44x.

Paul suggested to me and I've been contemplating changing
 the page table management on PowerPC to something similar
 to other platforms.=A0 At first, I was only going to do this on 8xx,
 but I think I'll do it for PowerPC in general so all boards can
 take advantage if they choose to do so.

Other ports populate the user page tables with the kernel pmd
 entries upon kernel faults, then we don't have to distinguish
 among page tables during the TLB exceptions.=A0 I'm hoping we
 can populate the init page table entries at boot time, and then
 everyone will inherit them, eliminating all faults unless the kernel
 vmalloc space grows to need a new pmd entry.

I'm a bit concerned about this in the future.  On Book-E parts we may=20
end up actually utilizing the address space bit to provide a true 4G/4G=20=

split.  I understand the current pain in doing this today.  I just want=20=

you to be aware of what the future might bring.

 This would be a generic solution to selecting task size, among
 other advanced memory mapping options.

I'll keep you posted.

While we wait on this I think getting Eugene's patch is worth while.

- kumar