Re: [PATCH 15/16] mm: use vm_unmapped_area() on sparc32 architecture

[Michel Lespinasse <walken@google.com>]

On Mon, Nov 5, 2012 at 5:25 PM, David Miller <davem@davemloft.net> wrote:
> From: Michel Lespinasse <walken@google.com>
> Date: Mon,  5 Nov 2012 14:47:12 -0800
>
>> Update the sparc32 arch_get_unmapped_area function to make use of
>> vm_unmapped_area() instead of implementing a brute force search.
>>
>> Signed-off-by: Michel Lespinasse <walken@google.com>
>
> Hmmm...
>
>> -     if (flags & MAP_SHARED)
>> -             addr = COLOUR_ALIGN(addr);
>> -     else
>> -             addr = PAGE_ALIGN(addr);
>
> What part of vm_unmapped_area() is going to duplicate this special
> aligning logic we need on sparc?

The idea there is that you can specify the desired alignment mask and
offset using info.align_mask and info.align_offset.

Now, I just noticed that the old code actually always uses an
alignment offset of 0 instead of basing it on pgoff. I'm not sure why
that is, but it looks like this may be an issue ?

-- 
Michel "Walken" Lespinasse
A program is never fully debugged until the last user dies.