RE: [Xen-ia64-devel] CONFIG_DOMAIN0_CONTIGUOUS in domain.c - Magenheimer, Dan (HP Labs Fort Collins) - com.xensource.lists.xen-ia64-devel

10 messages in com.xensource.lists.xen-ia64-develRE: [Xen-ia64-devel] CONFIG_DOMAIN0_C...

From	Sent On	Attachments
Dong, Eddie	26 Feb 2006 23:53
Zhang, Xiantao	27 Feb 2006 20:31
Isaku Yamahata	27 Feb 2006 21:46
Dong, Eddie	27 Feb 2006 22:04
Tian, Kevin	27 Feb 2006 22:14
Magenheimer, Dan (HP Labs Fort Collins)	28 Feb 2006 08:52
Dong, Eddie	28 Feb 2006 15:18
Magenheimer, Dan (HP Labs Fort Collins)	28 Feb 2006 15:35
Yang, Fred	28 Feb 2006 16:51
Dong, Eddie	28 Feb 2006 16:56

Subject:	RE: [Xen-ia64-devel] CONFIG_DOMAIN0_CONTIGUOUS in domain.c
From:	Magenheimer, Dan (HP Labs Fort Collins) (dan....@hp.com)
Date:	02/28/2006 08:52:56 AM
List:	com.xensource.lists.xen-ia64-devel

Sorry for the delay in responding to this...

First note that on the current (P==M) xen-ia64-unstable tree, CONFIG_DOMAIN0_CONTIGUOUS is #define'd in domain.c. It has been turned on since the alpha release in December 2004.

When I was first bringing up domain0 (in mid-2004 on ski), domain0 was using VP. When I started bringing up I/O for domain0, I realized that if P==M, all domain0 I/O would work without change, so added ifdef's for CONFIG_DOMAIN0_CONTIGUOUS to test with P==M. Since it worked, CONFIG_DOMAIN0_CONTIGUOUS has been define'd on and any domain0 code with CONFIG_DOMAIN0_CONTIGUOUS off has likely bitrotted.

So I don't think current (P==M) xen-ia64-unstable will work with CONFIG_DOMAIN0_CONTIGUOUS undef'd. It should be undef'd as part of Isaku's work to convert domain0 to VP. HOWEVER... it may be possible and desirable for much of Isaku's work to support both VP and P==M. For non-I/O code, CONFIG_DOMAIN0_CONTIGUOUS could be used (or possibly renamed) to select VP or P==M at compile-time, at least until the conversion to VP+DMA is complete. This would allow at least some of Isaku's VP code to coexist in the main xen-ia64-unstable tree without breaking the current P==M model.

Dan

-----Original Message----- From: xen-...@lists.xensource.com [mailto:xen-...@lists.xensource.com] On Behalf Of Tian, Kevin Sent: Monday, February 27, 2006 11:15 PM To: Isaku Yamahata; Dong, Eddie Cc: xen-...@lists.xensource.com Subject: RE: [Xen-ia64-devel] CONFIG_DOMAIN0_CONTIGUOUS in domain.c

From: Isaku Yamahata

Sent: 2006年2月28日 13:47 Hi.

I think that construct_dom0() is broken for CONFIG_DOMAIN0_CONTIGUOUS. I had to modify it heavily to boot dom0 with P2M/VP model.

For example construct_dom0() ... memcpy(__va(pinitrd_start),initrd_start,initrd_len); This memcpy() assumes p==m.

I thought that CONFIG_DOMAIN0_CONTIGUOUS option was introduced at the eary develpment stage and it remained just because no one removed it. However I don't know its history.

That option might be introduced because dom0 can have more memories spanning machine holes, even when p==m was the model that time. To simplify the early design, thus limitation was added to ensure constraints within one continuous trunk. So it should be removed now as a step of your P2M effort.