Re: [Xen-devel] [XenARM] Kernel Panic in booting domU

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Wed, 24 Apr 2013, Ian Campbell wrote:
> On Wed, 2013-04-24 at 10:12 +0100, Khandelwal, Shubham wrote:
> > I am running xen 4.3.3-unstable on ARM Fast Model for Versatile
> > Express platform.
> 
> The right list for Xen on ARM with virtualization extensions[0] is
> xen-devel@. The xen-arm@ list is for the Xen ARM PV port[1] (for ARMv6
> and earlier). I've moved xen-arm to bcc and added xen-devel to cc. (I've
> left quotes mostly untrimmed for the benefit of xen-devel)
> 
> Xen 4.3 has not been released and "4.3.3-unstable" isn't a real version
> and doesn't contain enough information to determine what you are
> running. When running development versions a reference to the git commit
> ID (and if necessary the tree+branch containing it) is usually better.
> 
> [0] http://wiki.xen.org/wiki/Xen_ARM_with_Virtualization_Extensions
> [1] http://wiki.xen.org/wiki/Xen_ARM_%28PV%29
> 
> >  I am getting a kernel panic when I am trying to create a domU using
> > âxl createâ command. My linux kernel version for dom0 and domU is
> > 3.9.0-rc7. The error is :
> > ...
> > 
> > (XEN) DOM1: Booting Linux on physical CPU 0x0
> > (XEN) DOM1: Linux version 3.9.0-rc7 (phoenix@ubuntu) (gcc version 4.6.3 
> > (GCC) ) #8 Tue Apr 23 17:35:45 IST 2013
> > (XEN) DOM1: CPU: ARMv7 Processor [412fc0f0] revision 0 (ARMv7), cr=10c53c7d
> > (XEN) DOM1: CPU: PIPT / VIPT nonaliasing data cache, PIPT instruction cache
> > (XEN) DOM1: Machine: ARM-Versatile Express, model: XENVM-4.2
> > (XEN) DOM1: bootconsole [earlycon0] enabled
> > (XEN) DOM1: Memory policy: ECC disabled, Data cache writeback
> > (XEN) DOM1: On node 0 totalpages: 32768
> > (XEN) DOM1: free_area_init_node: node 0, pgdat c0479040, node_mem_map 
> > c049c000
> > (XEN) DOM1:   Normal zone: 256 pages used for memmap
> > (XEN) DOM1:   Normal zone: 0 pages reserved
> > (XEN) DOM1:   Normal zone: 32768 pages, LIFO batch:7
> > (XEN) DOM1: CPU: All CPU(s) started in SVC mode.
> > (XEN) DOM1: pcpu-alloc: s0 r0 d32768 u32768 alloc=1*32768
> > (XEN) DOM1: pcpu-alloc: [0] 0
> > (XEN) DOM1: Built 1 zonelists in Zone order, mobility grouping on.  Total 
> > pages: 32512
> > (XEN) DOM1: Kernel command line: earlyprintk=xenboot console=hvc0 
> > root=/dev/xvda debug rw init=/bin/sh
> > (XEN) DOM1: PID hash table entries: 512 (order: -1, 2048 bytes)
> > (XEN) DOM1: Dentry cache hash table entries: 16384 (order: 4, 65536 bytes)
> > (XEN) DOM1: Inode-cache hash table entries: 8192 (order: 3, 32768 bytes)
> > (XEN) DOM1: __ex_table already sorted, skipping sort
> > (XEN) DOM1: Memory: 128MB = 128MB total
> > (XEN) DOM1: Memory: 125180k/125180k available, 5892k reserved, 0K highmem
> > (XEN) DOM1: Virtual kernel memory layout:
> > (XEN) DOM1:     vector  : 0xffff0000 - 0xffff1000   (   4 kB)
> > (XEN) DOM1:     fixmap  : 0xfff00000 - 0xfffe0000   ( 896 kB)
> > (XEN) DOM1:     vmalloc : 0xc8800000 - 0xff000000   ( 872 MB)
> > (XEN) DOM1:     lowmem  : 0xc0000000 - 0xc8000000   ( 128 MB)
> > (XEN) DOM1:     pkmap   : 0xbfe00000 - 0xc0000000   (   2 MB)
> > (XEN) DOM1:       .text : 0xc0008000 - 0xc042e138   (4249 kB)
> > (XEN) DOM1:       .init : 0xc042f000 - 0xc0450d34   ( 136 kB)
> > (XEN) DOM1:       .data : 0xc0452000 - 0xc04799c0   ( 159 kB)
> > (XEN) DOM1:        .bss : 0xc04799c0 - 0xc049b62c   ( 136 kB)
> > (XEN) DOM1: NR_IRQS:16 nr_irqs:16 16
> > (XEN) DOM1: ------------[ cut here ]------------
> > (XEN) DOM1: WARNING: at drivers/clk/versatile/clk-vexpress.c:29 
> > vexpress_sp810_init+0xa0/0xcc()
> > (XEN) DOM1: [<c0011b24>] (unwind_backtrace+0x0/0xe0) from [<c00184f8>] 
> > (warn_slowpath_common+0x48/0x64)
> > (XEN) DOM1: [<c00184f8>] (warn_slowpath_common+0x48/0x64) from [<c00185cc>] 
> > (warn_slowpath_null+0x18/0x1c)
> > (XEN) DOM1: [<c00185cc>] (warn_slowpath_null+0x18/0x1c) from [<c04436a8>] 
> > (vexpress_sp810_init+0xa0/0xcc)
> > (XEN) DOM1: [<c04436a8>] (vexpress_sp810_init+0xa0/0xcc) from [<c04438cc>] 
> > (vexpress_clk_of_init+0x2c/0x140)
> > (XEN) DOM1: [<c04438cc>] (vexpress_clk_of_init+0x2c/0x140) from 
> > [<c0435a78>] (v2m_dt_timer_init+0x8/0x90)
> > (XEN) DOM1: [<c0435a78>] (v2m_dt_timer_init+0x8/0x90) from [<c0432ed4>] 
> > (time_init+0x14/0x20)
> > (XEN) DOM1: [<c0432ed4>] (time_init+0x14/0x20) from [<c042f610>] 
> > (start_kernel+0x184/0x2bc)
> > (XEN) DOM1: [<c042f610>] (start_kernel+0x184/0x2bc) from [<80008070>] 
> > (0x80008070)
> > (XEN) DOM1: ---[ end trace 1b75b31a2719ed1c ]---
> > (XEN) DOM1: ------------[ cut here ]------------
> > (XEN) DOM1: WARNING: at drivers/clk/versatile/clk-vexpress.c:116 
> > vexpress_clk_of_init+0x78/0x140()
> > (XEN) DOM1: [<c0011b24>] (unwind_backtrace+0x0/0xe0) from [<c00184f8>] 
> > (warn_slowpath_common+0x48/0x64)
> > (XEN) DOM1: [<c00184f8>] (warn_slowpath_common+0x48/0x64) from [<c00185cc>] 
> > (warn_slowpath_null+0x18/0x1c)
> > (XEN) DOM1: [<c00185cc>] (warn_slowpath_null+0x18/0x1c) from [<c0443918>] 
> > (vexpress_clk_of_init+0x78/0x140)
> > (XEN) DOM1: [<c0443918>] (vexpress_clk_of_init+0x78/0x140) from 
> > [<c0435a78>] (v2m_dt_timer_init+0x8/0x90)
> > (XEN) DOM1: [<c0435a78>] (v2m_dt_timer_init+0x8/0x90) from [<c0432ed4>] 
> > (time_init+0x14/0x20)
> > (XEN) DOM1: [<c0432ed4>] (time_init+0x14/0x20) from [<c042f610>] 
> > (start_kernel+0x184/0x2bc)
> > (XEN) DOM1: [<c042f610>] (start_kernel+0x184/0x2bc) from [<80008070>] 
> > (0x80008070)
> > (XEN) DOM1: ---[ end trace 1b75b31a2719ed1d ]---
> > (XEN) DOM1: sched_clock: 32 bits at 24MHz, resolution 41ns, wraps every 
> > 178956ms
> > (XEN) DOM1: Unable to handle kernel NULL pointer dereference at virtual 
> > address 0000005c
> > (XEN) DOM1: pgd = c0004000
> > (XEN) DOM1: [0000005c] *pgd=00000000
> > (XEN) DOM1: Internal error: Oops: 5 [#1] ARM
> > (XEN) DOM1: CPU: 0    Tainted: G        W     (3.9.0-rc7 #8)
> > (XEN) DOM1: PC is at versatile_read_sched_clock+0x10/0x28
> 
> This should never be being called in a domU AFAICT.
> 
> What device tree are you passing/appending to the domU kernel?
> 
> This function should only be called if arch_timer_sched_clock_init()
> failed for some reason, which looks like it can only happen if
> arch_timer_get_rate() returns 0, which in turn only happens if
> arch_timer_available == 0.
> 
> Which all suggests that arch_timer_of_register() must have failed,
> however I don't see an error path which doesn't log and I don't see
> anything in your logs.
> 
> Perhaps you could instrument up that callchain and see what's going on?
> 
> Stefano's patches to switch to using mach-virt instead of mach-vexpress
> probably side step this issue too.

I would recommend trying this patch series:

git://git.kernel.org/pub/scm/linux/kernel/git/sstabellini/xen.git 
3.9-rc3-smp-3-tag

You might want to edit the xenvm DTS, removing the additional cpu from
it, because without all the appropriate Xen patches guest multicpu won't
work.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel