[Qemu-devel] Sparc32 bash problem solved

[Qemu-devel] Sparc32 bash problem solved

[Blue Swirl]

[-- Attachment #1: Type: text/plain, Size: 695 bytes --]

For some reason, bash and strace didn't work inside Sparc32 system emulator. 
I finally found the reason by observing strange Linux behaviour relating to 
MMU no-fault mode. No fault mode seems to apply only to supervisor accesses, 
not user ones. The logic was not described like this in the manual, thank 
you very much.

Patched Qemu version can run several distro installers which crashed very 
early before. Bash and strace are okay.

There are still some bugs in the SCSI code, some installers didn't detect 
the drives or got the size wrong.

_________________________________________________________________
Don't just search. Find. Check out the new MSN Search! 
http://search.msn.com/

[-- Attachment #2: qemu-sparc.patch-56.bz2 --]
[-- Type: application/octet-stream, Size: 367 bytes --]

Re: [Qemu-devel] Sparc32 bash problem solved

[Joerg Platte]

Am Samstag, 27. Mai 2006 21:30 schrieb Blue Swirl:
> For some reason, bash and strace didn't work inside Sparc32 system
> emulator. I finally found the reason by observing strange Linux behaviour
> relating to MMU no-fault mode. No fault mode seems to apply only to
> supervisor accesses, not user ones. The logic was not described like this
> in the manual, thank you very much.

How did you find it?

> Patched Qemu version can run several distro installers which crashed very
> early before. Bash and strace are okay.

Yes, booting a small system with udev works! qemu in graphical mode boots up 
to the login prompt. But unfortunately each character i tried to type in is 
corrupt. Hence, its not possible to log in. In console mode getty causes a 
kernel oops. But it works now much better than before. Many thanks!

regards,
Jörg


INIT: version 2.86 booting
Mounting a tmpfs over /dev...done.
Creating initial device nodes...done.
Activating swap.
System time was Mon May 27 22:12:09 UTC 1974.
Setting the System Clock using the Hardware Clock as reference...
System Clock set. System local time is now Mon May 27 22:12:12 UTC 1974.
Checking all file systems...
fsck 1.37 (21-Mar-2005)
WARNING: Your /etc/fstab does not contain the fsck passno
        field.  I will kludge around things for you, but you
        should fix your /etc/fstab file as soon as you can.

Setting kernel variables ...
... done.
Mounting local filesystems...
Cleaning /tmp /var/run /var/lock.
Starting hotplug subsystem:
   pci
cat: /sys/bus/pci/devices/*/class: No such file or directory
cat: /sys/bus/pci/devices/*/vendor: No such file or directory
cat: /sys/bus/pci/devices/*/device: No such file or directory
cat: /sys/bus/pci/devices/*/subsystem_vendor: No such file or directory
cat: /sys/bus/pci/devices/*/subsystem_device: No such file or directory
Bad PCI agent invocation
   pci      [success]
   usb
   usb      [success]
   isapnp
   isapnp   [success]
   ide
   ide      [success]
   input
   input    [success]
   scsi
modprobe: Can't open dependencies file /lib/modules/2.6.13-sam/modules.dep (No 
such file or directory)
     sd_mod: can't be loaded (for disk)
modprobe: Can't open dependencies file /lib/modules/2.6.13-sam/modules.dep (No 
such file or directory)
     sd_mod: can't be loaded (for disk)
modprobe: Can't open dependencies file /lib/modules/2.6.13-sam/modules.dep (No 
such file or directory)
     sr_mod: can't be loaded (for cdrom)
modprobe: Can't open dependencies file /lib/modules/2.6.13-sam/modules.dep (No 
such file or directory)
     sg: can't be loaded (for cdrom)
   scsi     [success]
done.

Setting the System Clock using the Hardware Clock as reference...
System Clock set. Local time: Mon May 27 22:12:19 UTC 1974

Initializing random number generator...done.
Setting audio parameters...none.
INIT: Entering runlevel: 2
 Unable to handle kernel NULL pointer dereference
tsk->{mm,active_mm}->context = 00000085
tsk->{mm,active_mm}->pgd = fc00bc00
              \|/ ____ \|/
              "@'/ ,. \`@"
              /_| \__/ |_\
                 \__U_/
getty(1041): Oops [#1]
PSR: 040000c4 PC: f01290f0 NPC: f01290f4 Y: 00000000    Not tainted
PC: <tty_wakeup+0x4/0x64>
%G: 00000002 f0036e18  00000001 044000e6  f0036dbc 00000000  fb820000 00000001
%O: ffffffff 00000000  00000000 fb821bf8  00000018 00000007  fb8217c0 f001bb5c
RPC: <__udelay+0x1c/0x24>
%L: ffd0d000 ffd01754  ffd01758 00000020  00000000 ffffffff  fb820000 80028668
%I: 00000000 00066400  fb821924 fb821c10  ff3b373b 00000001  fb821828 f0036e18
Caller[f0036e18]: tasklet_action+0x6c/0xb8
Caller[f0036acc]: __do_softirq+0xa0/0xc4
Caller[f0036b30]: do_softirq+0x40/0x54
Caller[f0005e0c]: t_sec_trap_start_interrupt+0x38/0x5c
Caller[f01096c4]: prom_nbputchar+0xa0/0xb0
Caller[f0109700]: prom_putchar+0x8/0x24
Caller[f0109f6c]: prom_write+0x2c/0x58
Caller[f0109fcc]: prom_printf+0x34/0x48
Caller[f011a298]: promcon_cursor+0xa0/0xdc
Caller[f0138fc8]: set_cursor+0x84/0xb0
Caller[f013c5c8]: con_flush_chars+0x3c/0x50
Caller[f013c3e0]: con_write+0x1c/0x2c
Caller[f012f578]: write_chan+0x228/0x268
Caller[f0129a88]: tty_write+0x16c/0x258
Caller[f006d130]: vfs_write+0xa8/0x11c
Caller[f006d238]: sys_write+0x30/0x64
Caller[f001b83c]: syscall_is_too_hard+0x34/0x40
Caller[00011cd8]: 0x11cd8
Instruction DUMP: 81c3e008  90023fe7  9de3bf98 <c20620b8> 80886020  12800006  
b2102001  b0062934  b4102001
Kernel panic - not syncing: Aiee, killing interrupt handler!
 <0>Rebooting in 10 seconds..<1>Unable to handle kernel NULL pointer 
dereference

Re: [Qemu-devel] Sparc32 bash problem solved

[Joerg Platte]

Am Samstag, 27. Mai 2006 22:21 schrieb Joerg Platte:
> Yes, booting a small system with udev works! qemu in graphical mode boots
> up to the login prompt. But unfortunately each character i tried to type in
> is corrupt. Hence, its not possible to log in. In console mode getty causes
> a kernel oops. But it works now much better than before. Many thanks!

The oops and the corrupted characters are caused by the console-tools. After 
removing them everything works as expected! I was able to boot from the same 
image I used on a real 32 bit SPARC! I can log in and even log in again via 
ssh to localhost :-). Great work!

regards,
Jörg

Re: [Qemu-devel] Sparc32 bash problem solved

[Blue Swirl]

> > emulator. I finally found the reason by observing strange Linux 
>behaviour
> > relating to MMU no-fault mode. No fault mode seems to apply only to
> > supervisor accesses, not user ones. The logic was not described like 
>this
> > in the manual, thank you very much.
>
>How did you find it?

I looked at qemu.log to find the faulting location in bash. I enabled 
DEBUG_MMU in helper.c and op_helper.c to see if the MMU mappings are 
correct, generating some 400 megs of logs. The instruction was load and the 
data being accessed was not mapped. Then I attached gdb to emulator and set 
a breakpoint to the faulting address and also to Linux fault handler. I 
noticed that the fault kept repeating forever, as if Linux didn't realise 
that any fault has happened. Then I got the idea to try this change. I think 
the logic is that it is not safe to allow user to override the MMU mappings. 
What an exciting story to tell your grandchildren, don't you think?

>Caller[f0005e0c]: t_sec_trap_start_interrupt+0x38/0x5c
>Caller[f01096c4]: prom_nbputchar+0xa0/0xb0

Maybe there is something wrong with Proll putchar function, or Proll's data 
areas are overwritten.

_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar - get it now! 
http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/