Linux Virtualization - Jul 2015 - I can read/write in virtio BLK Device, but I can't run a hello world program in it

Hi Linux Virtualization Mailing List Members,

I am trying to develop a driver for virtio blk device in guest. I can safely
mount the virtual blk device with: mount /dev/vda /mnt/
in /mnt/ I can read & write the device. But when I run a simple hello world
program in /mnt/root/, this error happens:

root at linux_guest:/mnt/root# ./helloworld_static 
[   23.003459] CPU: 0 PID: 491 Comm: helloworld_stat Not tainted 3.19.1+ #101
[   23.007030] Hardware name: ARM-Versatile Express
[   23.010102] task: 970f2400 ti: 9184e000 task.ti: 9184e000
[   23.011645] PC is at 0x11d6a
[   23.012801] LR is at 0x89b3
[   23.013627] pc : [<00011d6a>]    lr : [<000089b3>]    psr:
80000030
[   23.013627] sp : 7e932c10  ip : 00008e45  fp : 00000000
[   23.017108] r10: 00000000  r9 : 00000000  r8 : 00000000
[   23.023165] r7 : 00000000  r6 : 00053794  r5 : 7e932dd4  r4 : 0006a9b4
[   23.026854] r3 : 0000022c  r2 : 00000000  r1 : ffffffd0  r0 : 00000021
[   23.031814] Flags: Nzcv  IRQs on  FIQs on  Mode USER_32  ISA Thumb  Segment
user
[   23.036225] Control: 10c5387d  Table: 916f4059  DAC: 00000015
[   23.040885] CPU: 0 PID: 491 Comm: helloworld_stat Not tainted 3.19.1+ #101
[   23.043224] Hardware name: ARM-Versatile Express
[   23.045168] [<80012c68>] (unwind_backtrace) from [<80010dc8>]
(show_stack+0x10/0x14)
[   23.050392] [<80010dc8>] (show_stack) from [<8030c9d0>]
(__do_user_fault.isra.10+0x98/0xa8)
[   23.052687] [<8030c9d0>] (__do_user_fault.isra.10) from
[<80017c6c>] (do_page_fault+0x35c/0x3b8)
[   23.055358] [<80017c6c>] (do_page_fault) from [<80008488>]
(do_DataAbort+0x34/0x98)
[   23.057510] [<80008488>] (do_DataAbort) from [<80011a5c>]
(__dabt_usr+0x3c/0x40)
[   23.065243] Exception stack(0x9184ffb0 to 0x9184fff8)
[   23.069425] ffa0:                                     00000021 ffffffd0
00000000 0000022c
[   23.073779] ffc0: 0006a9b4 7e932dd4 00053794 00000000 00000000 00000000
00000000 00000000
[   23.079985] ffe0: 00008e45 7e932c10 000089b3 00011d6a 80000030 ffffffff

Can you tell me what I could probably miss? Thank you very much, I forever in
debt for your time and effort to read this. I am writing the BLK Device based on
Lguest's code.

some information that may help:

root at linux_guest:/mnt/root# file helloworld_static 
helloworld_static: ELF 32-bit LSB executable, ARM, version 1 (SYSV), statically
linked, for GNU/Linux 2.6.31,
BuildID[sha1]=0xc63622efd65f635f2b93c05caa8a1302d60a8caa, not stripped

root at linux_guest:/mnt/root# uname -a
Linux linux_guest 3.19.1+ #101 Tue Jul 7 09:35:23 KST 2015 armv7l GNU/Linux

root at linux_guest:/mnt/root# ls -l helloworld_static
-rwxr-xr-x    1 root     root        456218 Jul  7  2015 helloworld_static

during booting:

[    0.202185] brd: module loaded
[    0.262288] loop: module loaded
[    0.304868]  vda: unknown partition table
AtLFprofile: no performance counters
[    0.399451] oprofile: using timer interrupt.
CP: cubic registered
[    0.403361] NET: Registered protocol family 17
[    0.405039] Registering SWP/SWPB emulation handler
[    0.409375] RAMDISK: gzip image found at block 0
[    0.572643] EXT2-fs (ram0): warning: mounting unchecked fs, running e2fsck is
recommended
[    0.620201] VFS: Mounted root (ext2 filesystem) on device 1:0.
[    0.622113] devtmpfs: mounted
[    0.623328] Freeing unused kernel memory: 172K (80417000 - 80442000)

Starting logging: OK
Initializing random number generator... 
[    0.691483] random: dd urandom read with 0 bits of entropy available
done.
Starting network...
Starting dropbear sshd: OK

linux_guest login:
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.linuxfoundation.org/pipermail/virtualization/attachments/20150707/01d9061e/attachment.html>

On Tue, 7 Jul 2015 10:52:01 +0900
Ganis Zulfa Santoso <ganis.zulfa at gmail.com> wrote:
> Hi Linux Virtualization Mailing List Members,
> 
> I am trying to develop a driver for virtio blk device in guest.
Why don't you simply use the driver that is already available in the
kernel?
> I can safely mount the virtual blk device with: mount /dev/vda /mnt/
> in /mnt/ I can read & write the device. But when I run a simple hello
world program in /mnt/root/, this error happens:
Have you already verified that you can successfully read the _right_
data from your block device? I.e. something like

mount /dev/xxx /mnt
cp helloworld_static /mnt/root/helloworld_static
umount /mnt  # just to make sure that it is not cached
mount /dev/xxx /mnt
md5sum helloworld_static
md5sum /mnt/root/helloworld_static

... and then compare the two md5sums whether they are the same?

 Thomas

> Thomas Huth <huth at tuxfamily.org> wrote:
> 
> Why don't you simply use the driver that is already available in the
> kernel?
I am using a custom hypervisor, the available driver is not suitable for it.
> 
> Have you already verified that you can successfully read the _right_
> data from your block device? I.e. something like
> 
> mount /dev/xxx /mnt
> cp helloworld_static /mnt/root/helloworld_static
> umount /mnt  # just to make sure that it is not cached
> mount /dev/xxx /mnt
> md5sum helloworld_static
> md5sum /mnt/root/helloworld_static
> 
> ... and then compare the two md5sums whether they are the same?
They are different. Thank you so much. I am truly thankful. 
> 
> Thomas