[OpenBIOS] [Qemu-devel] solaris 8 on sparc, webstart launcher crashing

[Mark Cave-Ayland]

On 28/05/11 07:40, Blue Swirl wrote:

(lots cut)

>> I was able to get the exact same error&  stack trace as the sig11 one
>> from (2) above.  Addresses were the same, only diferences were the
>> PID, LWP, and sp (330, 9, 0xee752c78)
>>
>>
>> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
>> Starting Web Start Launcher in Command Line Mode.
>> Fri May 27 15:29:19 2011 fatal: mounting of "/vol" failed
>> signal fault in critical section
>> signal number: 11, signal code: 1,
>>          fault address: 0xee780de8, pc: 0xef40d4a8, sp: 0xeb1b2c78
>> libthread panic: fault in libthread critical section : dumping core
>> (PID: 330 LWP 12)
>> stacktrace:
>>        ef40d49c
>>        ef40f134
>>        ef408c48
>>        0
>> Abort - core dumped
>> (blahdy blah)
>> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
>>
>> For reference, I'm using a 3g qcow2 image.  I originally formatted it
>> as a "SUN2.9G" in the format utility, but I let it autoconfigure the
>> partitions.  partition '0' is root (1.18G), '1' is swap (149m), '2' is
>> backup (2.71g), and '7' is home (1.39g).  There's overlap in the
>> cylinder ranges specified and the total size between those partitions
>> is larger than 3g ... but, I have no idea whether that's just a
>> Sun-ism, or whether I need to partition manually.
>>
>> I don't think it's my partitions, though.  The 'vol' directory doesn't
>> exist in my root before I try to get the webstart stuff to run, so I
>> think it's creating that directory and trying to mount an image to it.
>>
>>
>> Additionally, I'm starting qemu with:
>>
>> ./qemu_system_sparc -nographic -bios
>> /imgpath/openbios-sparc32_artyom.bin -hda
>> /imgpath/solaris2.8-img3.qcow2 -m 256 -net nic -net user -cdrom
>> /imgpath/solaris2.8/software_1of2.iso prom-env 'auto-boot?=false'
>> -snapshot
>>
>>
>> If anyone has a good idea, I'm all ears ... (or is it eyes?)
>
> I suppose this is a problem with QEMU since the OS has started up
> successfully. The way to verify this would be to check if this happens
> also with OBP.
>
> If the problem is on the QEMU side, getting relevant QEMU debug logs
> (-d in_asm,int which needs #define DEBUG_PCALL enabled in
> target-sparc/op_helper.c) would be needed. Those will be quite large
> and the relevant info is only near the end.

Yeah - at this point the kernel should have taken over completely and so 
I expect that you're hitting an emulation bug (probably the Solaris 
compiler emits certain instruction sequences not used by gcc which is 
why this has only just come to light).

In order for someone to fix this, you'll need to supply the information 
requested by Blue above. Also which version of QEMU are you running?


ATB,

Mark.

-- 
Mark Cave-Ayland - Senior Technical Architect
PostgreSQL - PostGIS
Sirius Corporation plc - control through freedom
http://www.siriusit.co.uk
t: +44 870 608 0063

Sirius Labs: http://www.siriusit.co.uk/labs