Re: [Xen-merge] i386 subarch

[Chris Wright <chrisw@xxxxxxxx>]

* Martin J. Bligh (mbligh@xxxxxxxxxx) wrote:
> One quick question, are you working against virgin 2.6.12, or bleeding
> edge git tree? Is important we all work against the same thing, whatever
> that is ... will follow whatever you did, just need to know.

I'm using linux-2.6.12-xen0 as the base (from a normal xen build) using
a tip of 501a70f3ae96.  Will need to roll forward to current Xen soon,
but the patch is still in too much flux to move the base.

> I took a quick look through it, am reluctant to criticise anything
> under the circumstances, but still ... some specifics ... 

Thanks for looking it out.

> Some of the files look very surgically cut out; some of them look to
> have been copied over, and then a few bits edited ... I presume you
> actually know this already, but for what it's worth - seems to be 
> quite a bit of duplication (whole duplicated functions) in: 

Yup.  That's from stock Xen, and is under the section 4) on the todo
list earlier in this thread.

> arch/i386/mach-xen/kernel/cpu/common.c
> arch/i386/mach-xen/kernel/i386_ksyms.c
> arch/i386/mach-xen/kernel/io_apic.c
> arch/i386/mach-xen/kernel/ioport.c
> arch/i386/mach-xen/kernel/irq.c
> arch/i386/mach-xen/kernel/ldt.c
> arch/i386/mach-xen/kernel/mpparse.c
> arch/i386/mach-xen/kernel/pci-dma.c
> arch/i386/mach-xen/kernel/process.c
> arch/i386/mach-xen/kernel/setup.c
> arch/i386/mach-xen/kernel/signal.c
> arch/i386/mach-xen/kernel/syscall_table.S 
> arch/i386/mach-xen/kernel/time.c 
> arch/i386/mach-xen/kernel/traps.c
> arch/i386/mach-xen/mm/fault.c
> arch/i386/mach-xen/mm/highmem.c
> arch/i386/mach-xen/mm/init.c
> arch/i386/mach-xen/mm/ioremap.c
> arch/i386/mach-xen/mm/pgtable.c
> arch/i386/mach-xen/pci/irq.c
> 
> Might be useful to run diff in reverse somehow, and find all the
> commonalities between the two sections of code. I'll see if I can
> figure a way to do that. 

I have a script that has the effect of something like:

diff -Nuar arch/i386 arch/i386/mach-xen/

Much of the duplication is just how Xen has been developed.  The
Makefiles show where there's pure dup, and the sparse tree shows you
what's actually changed.

> So what do we want to do? Break it up by filename, and see if we
> can get several people to clean up a few files each? Tis a fair 
> amount of work to do. Will help if we can get the patch to as small
> as possible, and show as little duplication as possible to miminize
> ongoing maintainance for everyone (including, but not limited to
> the Xen people).

I think that's the best way to go forward.  I'd like to get it to a
point where the patch is minimal before we start breaking down the dups.

thanks,
-chris

_______________________________________________
Xen-merge mailing list
Xen-merge@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-merge