Hi Russell,
On 24 January 2011 21:42, Russell King - ARM Linux
<[email protected]> wrote:
> On Tue, Jan 25, 2011 at 08:26:42AM +1100, Nick Piggin wrote:
>> Too bad about the PAE thing; my condolences.
>>
>>
>> On Tue, Jan 25, 2011 at 4:55 AM, Catalin Marinas
>> <[email protected]> wrote:
>> > This patch introduces the pgtable-3level*.h files with definitions
>> > specific to the LPAE page table format (3 levels of page tables).
>>
>> Seeing as you're shaking up these definitions, what do you think about
>> switching from 4level-fixup.h to pgtable-nopud.h / pgtable-nopmd.h headers?
>> One day eventually it would be nice to get rid of the fixup mode.
>
> I have patches to do this which I de-queued for the last merge window.
> It's not entirely trivial and without problem. You can find the patches
> in linux-next now.
>
> I was waiting for the new set of patches from Catalin before going back
> and working out the solutions to some of those problems.
Any plans for the nopmd patches? I haven't seen them in -next or on the list.
Thanks.
--
Catalin
On Mon, Mar 21, 2011 at 12:36:55PM +0000, Catalin Marinas wrote:
> Any plans for the nopmd patches? I haven't seen them in -next or on the list.
I dropped them again because of those pesky warnings, so again I'm not
planning to push them this window as I don't wish to be deluged in
people reporting the warnings.
They really need fixing once we know how the LPAE stuff interacts with
the change. At the moment I've no idea whether the existing section
support ends up at pgd or pmd level with LPAE.
Obviously that matters as with LPAE, pgd and pmd are different hardware
levels, but without LPAE they're the same hardware level.
On 21 March 2011 12:56, Russell King - ARM Linux <[email protected]> wrote:
> On Mon, Mar 21, 2011 at 12:36:55PM +0000, Catalin Marinas wrote:
>> Any plans for the nopmd patches? I haven't seen them in -next or on the list.
>
> I dropped them again because of those pesky warnings, so again I'm not
> planning to push them this window as I don't wish to be deluged in
> people reporting the warnings.
>
> They really need fixing once we know how the LPAE stuff interacts with
> the change. At the moment I've no idea whether the existing section
> support ends up at pgd or pmd level with LPAE.
With LPAE, the section should be at the pmd level (2nd level page
table) as the PGDIR_SHIFT is 30. But if we standardise on using the
pmd in both cases, macros like pmd_val() would expand to the right
level with classic page tables (it goes up to pgd_val).
PMD_SHIFT also gets defined as PGDIR_SHIFT for the classic page
tables. One of my patches in the series converts the existing code to
PMD_SHIFT from PGDIR_SHIFT.
I'm happy to give this a try if you have some existing patches for the
classic page tables (I can even start from scratch but I don't want to
duplicate work).
--
Catalin