torios-dev team mailing list archive

[KI7MT <ki7mt@xxxxxxxxx>]

Hello All,

Just my $.0.02 worth.

Having used Linux for the better part of 20yrs, the least amount of work
one does, with respect to customizing the mainline kernel, the better
off you'll be in the long run.

In the black-magic world Kernel development, there is little interest in
maintaining non-PAE patches. If you frequent the Kernel devel mailing
lists, there is very little discussion, if any, on the subject.

Maintaining one custom kernel will be a challenge for a small team, let
alone catering to fairly rare hardware (just look at the number of
patches released per iteration). I guess one has to weight the benefits
against the amount of work required to maintain support for it. Is it
really worth the effort ?

best regards,

Greg

On 06/02/2015 06:58 AM, Nio Wiklund wrote:
> Den 2015-06-02 14:29, Nio Wiklund skrev:
>> Den 2015-06-02 14:12, Israel skrev:
>>> On 06/02/2015 06:55 AM, Nio Wiklund wrote:
>>>> Den 2015-06-02 13:45, Israel skrev:
>>>>> On 06/02/2015 12:04 AM, Nio Wiklund wrote:
>>>>>> Den 2015-06-02 03:59, Israel skrev:
> ...
>>> Hi Nio,
>>> You didn't link me to a kernel package... this is the ISO file.
>>> Or did you mean for me to just download an run the ISO?  I unfortunately
>>> don't have a true non-PAE computer...  So I cannot really test it out.
>>>
>>> My intention was to get ready to start building ToriOS 2.0 so we can
>>> release the alpha for that almost as soon as ToriOS 1.0 is out.  :)
>>>
>>> I expect we will have a lot of work for ToriOS 3.0, since it will be
>>> using systemd, and I know nothing about that yet :)
>>>
>>
>> Hi Israel,
>>
>> I haven't got the kernel package or the system for getting it. Ask Phill
>> about that. What I have is an application with it. I might look for and
>> find the URL where I downloaded it. If Phill does not reply soon, just
>> ask me, and I'll look for it!
>>
>> And yes, I meant that you can install it and test it. If nothing else,
>> hands on experience that Phill's trusty non-pae kernel really works,
>> gives you a feeling that it is not only feasible, it is a real thing,
>> without much effort from you.
>>
>> You can set a virtual machine (e.g. in VirtualBox) to be non-pae, and
>> test that it works (like Jack Trice is testing ToriOS). I have tested it
>> in my IBM Thinkpad T42 with Pentium M. It works without fake-pae and
>> without forcepae (and without complaints) :-)
>>
>> The next step is to modify it, if you want to, but I remember that Phill
>> had to work quite hard to make it work at all, and after that to make it
>> work in live systems.
>>
>> Best regards
>> Nio
>>
> 
> Hi again Israel,
> 
> On the other hand, I think there are so few non-pae machines that you
> might skip them and cater for the Pentium M and Celeron M processors
> with trusty's built-in forcepae boot option.
> 
> I have searched for real non-pae machines via the Ubuntu Forums and it
> is almost impossible. But there is a fair number of 'fake-pae' machines
> (with Pentium M and Celeron M processors that have pae capability but
> lack a pae flag), even though they are also getting scarce as years pass by.
> 
> -o-
> 
> https://en.wikipedia.org/wiki/Physical_Address_Extension
> 
> 'PAE was first implemented in the Intel Pentium Pro in 1995,[1] although
> the accompanying chipsets usually lacked support for the required extra
> address bits.[2]'
> 
> So I think it is enough to have a non-pae kernel in ToriOS 1, which will
> get updated packages until April 2017 (22 years after pae was introduced).
> 
> -o-
> 
> Finally, I think it is very important to discuss this item at the ToriOS
> Saturday meetings, so that the active members at the meeting agree what
> to do. So I cc this mail to torios-dev. Now you know what I think about
> it :-)
> 
> Best regards
> Nio
>