[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Xen-ia64-devel] [8/9] add mca.c for supporting INIT handler
I used RHEL4's INIT handlers code.
I almost don't change it.
The reasons why I use RHEL4 code are:
1. Macro used by INIT handler in xen-ia64 is likely RHEL4's one.
2. RHEL4's code is stable (I think)
The purpose of making INIT handler is to use INIT for debug hypervior.
1. To show call trace when xen infinity-loop.
2. To dump safely if kexec/kdump is implemented.
We're making MCA handler now.
(MCA test is difficult. we're now making MCA test program
with having Seto-san teach linux-MCA hadler.)
As you mentioned , my init handler is a little old.
If MCA handler use new linux code and require changing INIT hanler,
I'll update INIT handler. I promise it.
>On Fri, 2006-06-23 at 01:50 +0900, Akio Takebe wrote:
>> This is main patch for supporting INIT handler.
>> I update this patch.
>> - into xen/arch/ia64/xen
>> - make smaller mca.c.
>> (non-ifdef code)
> Maybe the question I should have asked the first time around was why
>is init handling so different between Xen/ia64 and Linux/ia64? I know I
>mentioned splitting this out into a separate file as a possibility, but
>I still have concerns about how we maintain this long term, especially
>as we add more components of MCA handling in the (near) future. I'm
>certainly willing to live with large chunks of CPE/CMC/MCA specific code
>#ifdef'd out. Would it make things any better if we pulled in updates
>to the files from newer Linux/ia64 source? MCA handling has undergone
>some overhauling in upstream kernels that may help. Is this separate
>file approach really the right answer? Thanks,
>Alex Williamson HP Open Source & Linux Org.
Xen-ia64-devel mailing list