R-PI fixes for dhclient/sshd?

[Zbigniew Bodek]

2013/12/20 Adrian Chadd <adrian at freebsd.org>:
> On 20 December 2013 10:34, Zbigniew Bodek <zbb at freebsd.org> wrote:
>> 2013/12/20 Adrian Chadd <adrian at freebsd.org>:
>>> On 20 December 2013 09:29, Warner Losh <imp at bsdimp.com> wrote:
>>>>
>>>> On Dec 20, 2013, at 10:13 AM, Adrian Chadd wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> Did the fixes to the raspberry pi SoC support go into the tree? The
>>>>> stuff that was affecting the allocator behaviour for things like
>>>>> sshd/dhclient?
>>>>
>>>> I thought all armv6 builds were affected by this...
>>>
>>> Right. Well, I thought it had at least been diagnosed.
>>>
>>>
>> Hello.
>>
>> If you are referring to bug related to jemalloc and pmap stuff then:
>> - I was not able to reproduce this on my own (I mean on "not-rpi" machine)
>> - The commit that was affecting this problem was found in pmap however
>> there was no solution.
>> - I'll try to post some pmap patches in the near future (sorry I was
>> not able to do any concrete work lately).
>> - This issue is on my mind but I since I can't touch this directly I
>> can only try to find general improvements and post what have been
>> found in Semihalf.
>
> Thanks for the update.
>
> What would be the harm in reverting the patch?
>
> http://svnweb.freebsd.org/base/head/sys/arm/arm/pmap-v6.c?r1=251370&r2=252694&pathrev=252694
>

Hello again.

I guess it should be some performance impact since we will need to
write all pages marked as RW to backing
storage on page-out. This is regardless of its actual dirty state (no
modified emulation).
But this was how it behaved earlier and nothing bad happened so it
might be worth to temporary revert it and
debug the problem without the negative influence on the users. After
proper fix we should apply it again.

It's up to you. I have no objections to that (I have few other patches
that need to wait for pmap problems
resolution anyway like pmap_copy() + SP).

Best regards
zbb