So, I'm out of ideas.... and I don't think I have enough experience
with debugging Xen, nor with Xen internals, to go any further.
No matter what I try, the netbsd-5 XEN3PAE_DOMU i386 kernels I build all
fail to boot in exactly this same way, before xenconsoled even gets a
chance to attach a console connection. However the two 5.2.3 i386
kernels I've downloaded from nyftp boot just fine.
If anyone has any suggestions about what else I can try or examine,
please do let me know!
(XEN) d47v0: unhandled page fault (ec=0000)
(XEN) Pagetable walk from 00000000bf800000:
(XEN) L4[0x000] = 00000004821d7027 0000000000000603
(XEN) L3[0x002] = 00000004821d4027 0000000000000606
(XEN) L2[0x1fc] = 00000004821d6001 0000000000000604
(XEN) L1[0x000] = 0000000000000000 ffffffffffffffff
(XEN) domain_crash_sync called from entry.S: fault at ffff82d080214560 compat_create_bounce_frame+0xc6/0xde
(XEN) Domain 47 (vcpu#0) crashed on cpu#5:
(XEN) ----[ Xen-4.5.0 x86_64 debug=n Not tainted ]----
(XEN) CPU: 5
(XEN) RIP: 0009:[<00000000c0397055>]
(XEN) RFLAGS: 0000000000000297 EM: 1 CONTEXT: pv guest
(XEN) rax: 0000000000000000 rbx: 0000000000000000 rcx: 0000000000000003
(XEN) rdx: 00000000bf800000 rsi: 0000000052075001 rdi: 0000000000000008
(XEN) rbp: 00000000c0610cac rsp: 00000000c0610c54 r8: 00000037fedf3f00
(XEN) r9: 00000037fedf3f00 r10: 5000000000000000 r11: 0000000000000000
(XEN) r12: ffff83007f8fa000 r13: 0000000000000005 r14: ffff82d0802e2620
(XEN) r15: ffff830075fcb000 cr0: 000000008005003b cr4: 00000000000026f4
(XEN) cr3: 0000000851838000 cr2: 00000000bf800000
(XEN) ds: 0011 es: 0011 fs: 0031 gs: 0011 ss: 0011 cs: 0009
(XEN) Guest stack trace from esp=c0610c54:
(XEN) 00000000 c0397055 00010009 00010097 00000006 20202000 45202020 30343435
(XEN) 00014000 00000000 00000612 00000003 00000000 bf800000 00000612 00000000
(XEN) 005657f0 00000000 00000000 00000009 00000000 00001000 c0610d3c c03934c7
(XEN) 00000000 00000000 00000000 00000003 00000612 00000000 00014000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000008 00000000
(XEN) 00000000 00000000 000139ee 00000000 00008000 00000000 00000612 00000000
(XEN) 00000000 00000000 00000000 00000001 00008000 00000000 00000000 756e6547
(XEN) 00611000 c04db200 00000000 c0100063 00611000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00604000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
(XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
BTW, I tried setting XENCONSOLED_TRACE=guest (to add --log=guest to its
parameters), and it revealed this change (or equiv.) is needed:
--- sysutils/xentools45/files/xencommons.sh.~1.1.~ Tue Jan 20 08:42:13 2015
+++ sysutils/xentools45/files/xencommons.sh Mon Jun 1 19:21:44 2015
@@ -24,6 +24,8 @@
{
mkdir -p /var/run/xen || exit 1
mkdir -p /var/run/xenstored || exit 1
+ mkdir -p /var/log/xen/console || exit 1
+ mkdir -p /var/xen/dump || exit 1
}
xen_startcmd()
Searching on the Xen lists reveals a thread in which Andrew Cooper says
this about a "crash" very similar to the crash I'm seeing "This means
that Xen was unable to context switch back to dom0" [1]
(interestingly "rax" is always zero for me, but it seems to always have
a non-zero value in other similar reports, at least for Linux users)
Just today someone posted a message with a very similar crash report to
mine [2], with some similarities in their configuration (64-bit host,
32-bit domU), but with one main difference that I think is major enough
to indicate it can't be related (their report shows ec=0002), but I'll
mention it here just in case.
I guess I'll try sending a report to xen-users asking for help too.
[1] http://xen.markmail.org/message/bksienrx2lqmjt7v
[2] http://xen.markmail.org/message/nps2sce7yb2camor
--
Greg A. Woods
Planix, Inc.
<woods%planix.com@localhost> +1 250 762-7675 http://www.planix.com/
Attachment:
pgpGpBwaK7O3_.pgp
Description: PGP signature