Cortex-A15 MM patches

Cortex-A15 MM patches

[Simon Horman]

Hi Catalin, Hi All,

I would like to enquire about the status of two patches which I believe
were part of TC2 Evaluation Build which uses commit
7d80417ef67d51db15d5e72ac336aa6e0586014f of linaro-tracking.git.

  http://releases.linaro.org/12.10/android/vexpress

1. ARM: LPAE: Invalidate the TLB for module addresses during translation fault

   http://lists.infradead.org/pipermail/linux-arm-kernel/2012-February/086036.html

2. ARMv7: Do not clean the PTE if coherent page table walk is supported

   http://lists.infradead.org/pipermail/linux-arm-kernel/2011-July/055379.html

As per the links above the patches have been posted upstream some time ago.
In the case of the first patch an alternate approach was discussed.

   http://lists.infradead.org/pipermail/linux-arm-kernel/2012-February/086147.html

However, I do not see these changes upstream as of next-20120212.
I am wondering if an alternate approach has been taken upstream
or if these patches are still required upstream.

Thanks

Cortex-A15 MM patches

[Catalin Marinas]

Hi Simon,

On Wed, Feb 13, 2013 at 08:41:17AM +0000, Simon Horman wrote:
> I would like to enquire about the status of two patches which I believe
> were part of TC2 Evaluation Build which uses commit
> 7d80417ef67d51db15d5e72ac336aa6e0586014f of linaro-tracking.git.
> 
>   http://releases.linaro.org/12.10/android/vexpress
> 
> 1. ARM: LPAE: Invalidate the TLB for module addresses during translation fault
> 
>    http://lists.infradead.org/pipermail/linux-arm-kernel/2012-February/086036.html

Another variant has been proposed:

http://marc.info/?l=linux-arch&m=134858482615574&w=2

Unfortunately I haven't got any feedback.

Linaro should use the new patch instead of the old one so that it gets
wider test coverage.

> 2. ARMv7: Do not clean the PTE if coherent page table walk is supported
> 
>    http://lists.infradead.org/pipermail/linux-arm-kernel/2011-July/055379.html

That's a valid patch but IIRC Russell wanted to cover the pmd as well. I
think it's better to include such patch if it makes a significant
difference in performance. I haven't done any benchmarks.

-- 
Catalin