Messages in this thread | | | Date | Thu, 29 Jan 2009 23:14:02 +0100 | From | Ingo Molnar <> | Subject | Re: x86: unify genapic code, unify subarchitectures, remove old subarchitecture code |
| |
* Tim Pepper <lnxninja@linux.vnet.ibm.com> wrote:
> The machines (yes plural!) on which we had regular testing happening > have been offline for a few months due to some broader dependencies (eg: > console server in the old product used an old version of windows that > doesn't meet the corporate security standards) and we've just not > managed to get past that. That in turn is in the way of debugging boot > issues with newer kernels. At this point the most passionate > Sequent/IBMers caring about the NUMAQ support are resigned to it being a > lost cause. > > The main benefit to carrying NUMAQ support along this long was these > machines had a knack for triggering real bugs. But since there's no > active bug testing happening on them now and it doesn't look like there > will be...
At this point there's no technical need to kill it - it's a zero-overhead thing tucked away into a single .c module in arch/x86/kernel/numaq_32.c. I even consolidated most of its headers in that file, to reduce its cross section.
That was the main point of the restructuring i did - subarchitectures were causing ongoing maintenance overhead before. That overhead is practically zero now.
So two days ago it might have made some sense to rip out this code - now there's no pressing need really. We can remove it in a few years once the last beep has been received from those platforms. The numaq code is now quite similar to an old driver in drivers/* - there's no real downside from having it around.
Ingo
| |