I've tried running Alexander's kernel as a guest on x86. It just hangs without printing anything, even with earlyprintk on.
On Apr 24, 2013 6:16 PM, "Christoffer Dall" <cdall@xxxxxxxxxxxxxxx> wrote:
On Wed, Apr 24, 2013 at 12:06 PM, Howard Mao <zhehao.mao@xxxxxxxxx> wrote:
> I'm trying to use the pre-compiled images from the arndale website, but now
> on the host I keep getting a bunch of traces like this post-boot
>
> [00000030] *pgd=00000000
>
> Pid: 2368, comm: system_server
> CPU: 1 Not tainted (3.9.0-rc2-12898-ge1227ce #18)
> PC is at 0x9adf9bc4
> LR is at 0x9adf9b68
> pc : [<9adf9bc4>] lr : [<9adf9b68>] psr: 20000010
> sp : bed263a8 ip : 9ae2df20 fp : bed263d4
> r10: b886db98 r9 : b886dba0 r8 : ffffffff
> r7 : b886d9f8 r6 : 00000005 r5 : b886db88 r4 : 9ae231e8
> r3 : 00000000 r2 : ffffffb0 r1 : bed25f90 r0 : 00000000
> Flags: nzCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user
> Control: 30c5387d Table: 6e8b2e00 DAC: 55555555
> [<c0027058>] (unwind_backtrace+0x0/0x104) from [<c03f4848>]
> (dump_stack+0x20/0x24)
> [<c03f4848>] (dump_stack+0x20/0x24) from [<c0020ba4>] (show_regs+0x54/0x58)
> [<c0020ba4>] (show_regs+0x54/0x58) from [<c002c350>]
> (__do_user_fault+0xbc/0xc8)
> [<c002c350>] (__do_user_fault+0xbc/0xc8) from [<c002c554>]
> (do_page_fault+0x1f8/0x38c)
> [<c002c554>] (do_page_fault+0x1f8/0x38c) from [<c002c888>]
> (do_translation_fault+0xec/0xf4)
> [<c002c888>] (do_translation_fault+0xec/0xf4) from [<c0008494>]
> (do_DataAbort+0x3c/0xa0)
> [<c0008494>] (do_DataAbort+0x3c/0xa0) from [<c001f138>]
> (__dabt_usr+0x38/0x40)
> Exception stack(0xeeb5bfb0 to 0xeeb5bff8)
>
> It still boots okay and I can still use the shell. But getting spammed with
> messages like this is quite frustrating, and I didn't use to get them
> before. Does anyone know why system_server would be causing these page
> faults? Since the pgd is NULL, there's probably some mmap operation that
> went wrong.
>
can you all you try in QEMU TCG on your host first?
>
> On Tue, Apr 23, 2013 at 11:33 PM, Christoffer Dall <cdall@xxxxxxxxxxxxxxx>
> wrote:
>>
>> Did you use a self-compiled guest AOSP image for the user space
>> utilities that crash or the Arndale android dist.? Can you see if you
>> can grab a pre-compiled one from somewhere? I have a feeling there's
>> something wrong with the user space build...
>>
>> On Tue, Apr 23, 2013 at 5:07 PM, Howard Mao <zhehao.mao@xxxxxxxxx> wrote:
>> > I have VFP support enabled in the kernel config already. I will try
>> > using
>> > Alexander's kernel and qemu and see if that works. Thanks to both of you
>> > for
>> > your help.
>> >
>> >
>> > On Fri, Apr 19, 2013 at 5:36 AM, Marc Zyngier <marc.zyngier@xxxxxxx>
>> > wrote:
>> >>
>> >> On 19/04/13 03:33, Howard Mao wrote:
>> >> > Christoffer and I are attempting to boot an Android image on a Linux
>> >> > 3.9
>> >> > kernel with KVM-ARM enabled in QEMU, but on boot we keep getting a
>> >> > kernel panic complaining about an illegal instruction in init.
>> >> >
>> >> > init (1): undefined instruction: pc=00021a26
>> >> > Code: b5070000 46692001 ed3cf7f5 9b004a0d (0a01ed9d)
>> >> > Kernel panic - not syncing: Attempted to kill init!
>> >> > exitcode=0x00000004
>> >> >
>> >> > [<8001fe24>] (unwind_backtrace+0x0/0xf8) from [<803b4a5c>]
>> >> > (panic+0x90/0x1e4)
>> >> > [<803b4a5c>] (panic+0x90/0x1e4) from [<8002fe3c>]
>> >> > (do_exit+0x848/0x8b0)
>> >> > [<8002fe3c>] (do_exit+0x848/0x8b0) from [<80030008>]
>> >> > (do_group_exit+0x3c/0xc4)
>> >> > [<80030008>] (do_group_exit+0x3c/0xc4) from [<8003a398>]
>> >> > (get_signal_to_deliver+0x170/0x56c)
>> >> > [<8003a398>] (get_signal_to_deliver+0x170/0x56c) from [<8001c82c>]
>> >> > (do_signal+0xf0/0x3fc)
>> >> > [<8001c82c>] (do_signal+0xf0/0x3fc) from [<8001ce6c>]
>> >> > (do_work_pending+0x68/0xa8)
>> >> > [<8001ce6c>] (do_work_pending+0x68/0xa8) from [<80019900>]
>> >> > (work_pending+0xc/0x20)
>> >> >
>> >> > Has anyone seen something like this before when running Android in
>> >> > QEMU?
>> >> >
>> >> > We are using the kernel
>> >> > from
>> >> > https://github.com/columbia/linux-kvm-arm/tree/arndale-kvm-3.9-rc2
>> >> > QEMU from https://github.com/columbia/qemu/tree/kvm-arm-virtio
>> >> > Android from http://source.android.com/source/downloading.html
>> >> >
>> >> > This has been tested on an Arndale board with QEMU-KVM and QEMU-TCG
>> >> > and
>> >> > on an x86_64 machine with QEMU-TCG. All three platforms give the same
>> >> > error.
>> >>
>> >> This last sentence would imply that something is wrong in your Android
>> >> build, and has very little to do with KVM.
>> >>
>> >> If you look at the failing instruction, it disassemble as
>> >> VLDR s0, [sp, #4]
>> >>
>> >> My guess is that your guest kernel doesn't have VFP support enabled,
>> >> hence trapping on an UNDEF.
>> >>
>> >> M.
>> >> --
>> >> Jazz is not dead. It just smells funny...
>> >>
>> >
>> >
>> > _______________________________________________
>> > kvmarm mailing list
>> > kvmarm@xxxxxxxxxxxxxxxxxxxxx
>> > https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
>
>
_______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
