John wrote:
> On Tue, Jan 29, 2008 at 06:44:58AM -0800, J??rgen Keil wrote:
>
> > > > IMO, under the hypervisor we shouldn't panic here.
> > >
> > > Did you file an RFE for this?
> >
> > No, but there already is bug 6654995
> > "need clearer message trying to boot xVM on a non-PAE machine":
> >
> > http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6654995
> >
> > And idea who has filed that bug 6654995?
>
> It was an internal tester. If we can indeed dynamically adapt in both
> Xen and Solaris to non-PAE, then obviously fixing that would make this
> bug irrelevant.
Could the internal tester try my kmdb patch on that Tecra?
Or build/install a custom kernel with that panic removed in the
"__xpv" case, like this:
http://www.opensolaris.org/jive/thread.jspa?messageID=166513𨤀
http://mail.opensolaris.org/pipermail/xen-discuss/2007-October/001538.html
diff -r 7aad25c79a0f usr/src/uts/i86pc/vm/hat_i86.c
--- a/usr/src/uts/i86pc/vm/hat_i86.c Fri Oct 26 11:18:55 2007 -0700
+++ b/usr/src/uts/i86pc/vm/hat_i86.c Sun Oct 28 01:58:29 2007 +0200
@@ -516,7 +516,11 @@ mmu_init(void)
}
if (mmu.pae_hat && (x86_feature & X86_PAE) == 0)
+#if defined(__xpv)
+ cmn_err(CE_NOTE, "Processor does not support PAE");
+#else
panic("Processor does not support PAE");
+#endif
if ((x86_feature & X86_CX8) == 0)
panic("Processor does not support cmpxchg8b instruction");
How far do we get when the "x86_feature & X86_PAE"
test is disabled?
This message posted from opensolaris.org
_______________________________________________
xen-discuss mailing list
[email protected]
