On Fri, 2013-02-08 at 17:14 -0500, Kevin O'Connor wrote:
On Fri, Feb 08, 2013 at 02:50:35PM +0000, Ian Campbell wrote:
On Fri, 2013-02-08 at 14:45 +0000, Ian Campbell wrote:
I don't see this with rel-1.7.2. Bisecting fingers commit 6ca0460fbb8ecfa5d42c8928b7ee71f20d0cffdb Author: Kevin O'Connor email@example.com Date: Mon Jan 21 11:38:49 2013 -0500
POST: Reorganize post entry and "preinit" functions.
But it doesn't look like this changes the order of anything,
Except Xen is entering via handle_post not handle_elf, so it does change something, specifically xen_preinit() isn't called before make_bios_writable so the usingXen in there doesn't trigger. Xen (actually hvmloader) launches SeaBIOS with "ljmp $0xf000,$0xfff0" so I think handle_post is the expected landing point.
Thanks. Sorry about that. I was under the impression that Xen used the elf entry point from our discussions long ago - but I guess that was prior to the decision to use hvmloader.
I think that was the case, yes.
I'll revert the recent change that broke Xen.
Would you have a short description of how one can pull down the Xen code and run a custom compiled SeaBIOS image on it?
The "short" version would feature "download, build, install and configure Xen" I'm afraid :-/ I'd be more than happy to help you do that if you want though and if you can keep a suitably configured system around (it needn't be dedicated to Xen other than when testing Xen specifically) you won't need to do the majority of the setup again.