Re: 2.6.27-rc xen pvops regression?

[Jeremy Fitzhardinge <jeremy@goop.org>]

Alex Nixon wrote:
> Bernhard Schmidt wrote:
>> Hello,
>>
>> I've been a happy user of vanilla 2.6.26 with pvops in my Xen domU for
>> a couple of months. I wanted to test 2.6.27-rc (-rc3 a couple of weeks
>> ago, -rc5 just now) to verify another issue and ran into the following
>> error
>>   
> Thanks Bernhard - I'm able to reproduce it with your config.

Is this the crash with the UP config?

> I've had a look, and it seems the register_vcpu_info hypercall is
> being made twice, failing on the second time.  In fact,
> paravirt_pagetable_setup_{start,done} are both being called twice -
> once from arch/x86/kernel/setup.c, and then again from
> arch/x86/mm/init_32.c

Hm, yes.  Looks like a unifcation side-effect, though I thought I'd run
a 32-bit kernel recently enough to notice this.  Anyway, I think the
init_32.c calls should be redundant.

> I suspect one of them should go, but I'm not familiar enough with this
> code to suggest which one.
>
> But still with that crudely resolved, the kernel BUGs (see below). 
> Jeremy - any ideas?
>
> - Alex
>
>
> `xm dmesg':
>
> (XEN) mm.c:2249:d168 Mfn 24384 not pinned

Bit hard to know what this means without knowing what mfn 24384 is.

    J

>
> dmesg:
>
> EXT3-fs: mounted filesystem with ordered data mode.
> VFS: Mounted root (ext3 filesystem) readonly.
> Freeing unused kernel memory: 160k freed
> ------------[ cut here ]------------
> kernel BUG at
> /local/scratch/hotplug.linux.trees.git/arch/x86/xen/enlighten.c:850!
> invalid opcode: 0000 [#1]
>
> Pid: 1, comm: init Tainted: G        W
> (2.6.27-rc5-tip-00672-ge5c5407-dirty #52)
> EIP: 0061:[<c0102b04>] EFLAGS: 00010282 CPU: 0
> EIP is at pin_pagetable_pfn+0x3f/0x4b
> EAX: ffffffea EBX: e242dd74 ECX: 00000001 EDX: 00000000
> ESI: 00007ff0 EDI: 00000000 EBP: e242dd8c ESP: e242dd74
> DS: 007b ES: 007b FS: 0000 GS: 0000 SS: e021
> Process init (pid: 1, ti=e242c000 task=e242a000 task.ti=e242c000)
> Stack: 00000004 00024384 00000000 c144e8e0 00022747 00000000 e242dd9c
> c0102ee8
>       c02fe5d4 c144e8e0 e242dda4 c0102f23 e242ddbc c0116722 00000000
> 00000000
>       00000000 00000000 e242de2c c01467a8 00000000 c1000000 00000000
> 22747067
> Call Trace:
> [<c0102ee8>] ? xen_release_ptpage+0x52/0x71
> [<c0102f23>] ? xen_release_pte+0xd/0xf
> [<c0116722>] ? __pte_free_tlb+0x35/0x4b
> [<c01467a8>] ? free_pgd_range+0x377/0x38c
> [<c015a4bf>] ? setup_arg_pages+0x1b8/0x22b
> [<c017a80f>] ? load_elf_binary+0x3f1/0x10a4
> [<c0104ffa>] ? check_events+0x8/0xe
> [<c0104f71>] ? xen_restore_fl_direct_end+0x0/0x3
> [<c0104937>] ? xen_force_evtchn_callback+0xf/0x14
> [<c0104f71>] ? xen_restore_fl_direct_end+0x0/0x3
> [<c014298a>] ? page_address+0x85/0xa2
> [<c0159c51>] ? put_arg_page+0x8/0xa
> [<c0159e27>] ? copy_strings+0x156/0x160
> [<c0159ea8>] ? search_binary_handler+0x42/0x10a
> [<c015ad64>] ? do_execve+0x129/0x192
> [<c0105354>] ? sys_execve+0x29/0x50
> [<c0106b1e>] ? syscall_call+0x7/0xb
> [<c0109b2f>] ? kernel_execve+0x17/0x1c
> [<c0102140>] ? run_init_process+0x17/0x19
> [<c01021e3>] ? init_post+0xa1/0xf2
> [<c01077df>] ? kernel_thread_helper+0x7/0x10
>