A wee while ago,
USB support on FreeBSD6.2 SMP systems was discouraged.
Is that still the case?
Brett Davidson
brett at net24.co.nz
Wed Aug 1 01:13:27 UTC 2007
youshi10 at u.washington.edu wrote:
> On Wed, 1 Aug 2007, Brett Davidson wrote:
>
>> youshi10 at u.washington.edu wrote:
>>> On Wed, 1 Aug 2007, Brett Davidson wrote:
>>>
>>>> If there are patches/updates to this issue I would REALLY like to
>>>> know about them! :-)
>>>>
>>>> --
>>>> Brett Davidson
>>>> Systems Engineer
>>>> -- Net24 Limited
>>>> Web: www.net24.co.nz
>>>> Phone: 0800 5000 24 | DDI: +64 3 962 9518
>>>> -- // web hosting / email hosting / data backup // our reputation
>>>> for reliability precedes us
>>>>
>>>> This transmission is for the intended addressee only and is
>>>> confidential
>>>> information. If you have received this transmission in error, please
>>>> delete it and notify the sender.
>>>
>>> I may be wrong, but USB support has always been giant-locked, and
>>> continues to be giant-locked, at least until Hans work maybe gets
>>> integrated into 8-CURRENT after a major code review.
>>>
>>> -Garrett
>>>
>> Thanks Garrett.
>>
>> By the way, my colleague informed me incorrectly that USB and SMP
>> support was an issue. (Although I am keen for this to work!) :-)
>
> Heh :).
>
>> From the latest source, the Kernel PAE conf file mentions that USB
>> and PAE are not known to play well together. This is my immediate
>> concern.
>
> Not sure if that's a problem. PAE is just an ugly ugly hack, but you
> probably know that by now... should have gone with amd64 ;).
>
Indeed. I'd even say you're being a bit too polite in that indictment of
PAE. However, a few of the apps we run here don't support 64 bit FreeBSD
yet. We now require more memory than 4Gb, which in 32 bit land, means
that the big ugly hack (still keeping with your terminology here - this
is a "family" mailing list, after all) is necessary.
I only really want USB so that the Intel Remote Management module we are
running in our servers can actually control the system. (Yes, using USB
to allow a daughterboard to control a server is a particular kludge that
only Intel use but let's leave any discussion on that for another time,
shall we?).
Kinda important for me to have both working. Rock -> Me <- Hard Place.
>Uhm, it's sorta like that. Many drivers are still using giant-locks,
but there's a lot of effort going >into removing those giant locks for
creating more SMP'goodness :) .
That may be (and it sounds wonderful - I'll take three) but how does
this affect PAE again? The PAE kernel notes tell me that USB and PAE may
have issues with each other, but that's about it.
What is the problem? Has anyone found a problem?
Brett.
More information about the freebsd-questions
mailing list