Re: [Xen-devel] Dom0 LiveCD

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Sat, 2011-12-24 at 11:08 +0000, Roger Pau Monnà wrote:
> Hello,
> 
> I'm trying to create a Xen Dom0 LiveCD/usb, based on Alpine Linux,
> which has great support for this kind of stuff.

Cool!

> I've been able to
> successfully install and boot Xen kernel and tools from an HDD using
> the same configuration, but when using a tmpfs based boot, the Linux
> kernel 3.0.10 crashes while loading some modules (Xen kernel seems to
> boot fine). Xen version is stable 4.1.2 and Linux kernel is
> 3.0.10-grsec. Here is the full output of the boot process:
[...]
>  * Populating /dev with existing devices through uevents ... [ ok ]
>  * Waiting for uevents to be processed ... [ ok ]
>  * Mounting modloop /media/usb/boot/grsec.modloop.squashfs ... [ ok ]
>  * Loading hardware drivers ...Loading acpi:ACPI0003:
> Loading acpi:LNXCPU:
> Loading acpi:LNXPWRBN:
> [   23.080837] BUG: unable to handle kernel paging request at ffffe8ffffd3bfc4
> [   23.087991] IP: [<ffffffff81062026>] module_refcount+0x2e/0x99
> [   23.091743] PGD 351d6067 PUD 351d7067 PMD 35288067 PTE 0
> [   23.094530] Oops: 0000 [#1] SMP
> [   23.095852] CPU 0
> [   23.096561] Modules linked in: processor ac nls_iso8859_1 nls_cp437
> vfat fat sr_mod cdrom ata_generic ata_piix pata_acpi libata uhci_hcd
> ehci_hcd mptspi mptscsih mptbase scsi_transport_spi floppy usb_storage
> usb_libusual usbcore sd_mod scsi_mod squashfs loop
> [   23.121165]
> [   23.121801] Pid: 1626, comm: modprobe Not tainted 3.0.10-grsec
> #1-Alpine VMware, Inc. VMware Virtual Platform/440BX Desktop Reference
> Platform
> [   23.128242] RIP: e030:[<ffffffff81062026>]  [<ffffffff81062026>]
> module_refcount+0x2e/0x99
> [   23.134994] RSP: e02b:ffff8800351ddd88  EFLAGS: 00010287
> [   23.136997] RAX: 0000000000000001 RBX: 0000000000000001 RCX: 
> ffff88003aea8000
> [   23.140113] RDX: 000060ffc4e93fc0 RSI: 0000000000000020 RDI: 
> 0000000000000020
> [   23.143939] RBP: ffff8800351ddda8 R08: 0000000000000000 R09: 
> ffffffff814aa520
> [   23.146656] R10: ffff8800345d5000 R11: 0000000000000000 R12: 
> ffffffff814aa520
> [   23.149354] R13: ffffffffa014b7a0 R14: 0000000000001000 R15: 
> ffffffffa014b968
> [   23.157673] FS:  00005fb0175646a0(0000) GS:ffff88003ae91000(0000)
> knlGS:0000000000000000
> [   23.162354] CS:  e033 DS: 0000 ES: 0000 CR0: 000000008005003b
> [   23.164736] CR2: ffffe8ffffd3bfc4 CR3: 00000000339b3000 CR4: 
> 0000000000000660
> [   23.167954] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 
> 0000000000000000
> [   23.170719] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 
> 0000000000000400
> [   23.174109] Process modprobe (pid: 1626, threadinfo
> ffff88003477ab58, task ffff88003477a760)
> [   23.178661] Stack:
> [   23.179466]  ffffffffa014b7a8 ffff880034334300 ffffffffa014b7a0
> 0000000000001000
> [   23.182787]  ffff8800351dddf8 ffffffff8106211c ffff880035300420
> 00005fb016cffd10
> [   23.189494]  ffff8800351dddf8 ffff880034334300 0000000000000000
> 00000fecda0bcd90
> [   23.198618] Call Trace:
> [   23.200624]  [<ffffffff8106211c>] m_show+0x54/0x16b
> [   23.202512]  [<ffffffff810d1b5c>] seq_read+0x247/0x51f
> [   23.204483]  [<ffffffff810fccc2>] proc_reg_read+0x6c/0x85
> [   23.206543]  [<ffffffff810b614f>] vfs_read+0x104/0x198
> [   23.208909]  [<ffffffff810b6228>] sys_read+0x45/0x6c
> [   23.211075]  [<ffffffff812aa14b>] system_call_fastpath+0x16/0x1b
> [   23.215796]  [<ffffffff812aa0e9>] ? system_call_after_swapgs+0x19/0x65
> [   23.219196] Code: ff 48 89 e5 41 56 41 55 49 89 fd 41 54 4c 8b 25
> 79 12 29 00 53 31 db eb 16 48 63 c8 49 8b 95 f8 01 00 00 48 8b 0c cd
> 50 60 3a 81
> [   23.228774]  5c 0a 04 ff c0 be 20 00 00 00 4c 89 e7 48 63 d0 e8 34 24 0e
> [   23.235223] RIP  [<ffffffff81062026>] module_refcount+0x2e/0x99
> [   23.240718]  RSP <ffff8800351ddd88>
> [   23.243912] CR2: ffffe8ffffd3bfc4
> [   23.245256] ---[ end trace 4763addd9fb226e5 ]---
> Killed
> Loading acpi:LNXSYSTM:
> 
> I've changed the init script to load each module separatedly, instead
> of doing a modprobe -a <all modules>, if I loaded all the modules at
> once the system crashed instead of freezing. The problem seems to be
> related to acpi, and the modules are loaded from a squashfs image
> mounted on loop0. Does anyone know what might be causing this issue or
> if there is a possible fix?

Nothing springs to mind. On the face of it this doesn't look all that
Xen specific -- if you boot the same Linux kernel without Xen underneath
(but otherwise in the same configuration) does it work?

You are using the -grsec patches version of Linux, does a vanilla
mainline have the same issue?

Ian.



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