Re: Bug in master branch: IbmPrep40pMachine.test_factory_firmware_and_netbsd broken

[Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>]

On 02/03/2023 10:27, Philippe Mathieu-Daudé wrote:

> On 2/3/23 10:48, Cédric Le Goater wrote:
>> On 3/2/23 08:47, Thomas Huth wrote:
>>>
>>>   Hi all,
>>>
>>> seems like we've got another bug that sneaked in during the CI minutes blackout: 
>>> The avocado test IbmPrep40pMachine.test_factory_firmware_and_netbsd is now broken 
>>> reliably, see:
>>>
>>>   https://gitlab.com/qemu-project/qemu/-/jobs/3858833617#L300
>>>   https://gitlab.com/thuth/qemu/-/jobs/3858727901#L300
>>>   https://gitlab.com/thuth/qemu/-/jobs/3857804517#L300
>>>
>>> Is anybody already looking into this?
>>>
>>>   Thomas
>>
>> I tried manually and QEMU reached :
>>        >> (builds@wb29, Tue Oct  7 23:59:44 PDT 2008)
>>      Boot:
>>      Loading in()
>>      3558080+232036 [197696+188133]=0x3fbbf0
>>      start=0x100000
>>      Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
>>          2006, 2007
>>          The NetBSD Foundation, Inc.  All rights reserved.
>>      Copyright (c) 1982, 1986, 1989, 1991, 1993
>>          The Regents of the University of California.  All rights reserved.
>>      NetBSD 4.0.1 (GENERIC) #0: Wed Oct  8 00:46:56 PDT 2008
>> builds@wb29:/home/builds/ab/netbsd-4-0-1-RELEASE/prep/200810080053Z-obj/home/builds/ab/netbsd-4-0-1-RELEASE/src/sys/arch/prep/compile/GENERIC 
>>
>>      Model: IBM PPS Model 6015
>>      total memory = 128 MB
>>      avail memory = 119 MB
>>      trap: kernel read DSI trap @ 0x7c3042a6 by 0x3e4cf8 (DSISR 0x40000000, 
>> err=14), lr 0x377eac
>>      panic: trap
>>      Stopped in pid 0.1 () at        netbsd:cpu_Debugger+0x10: lwz     r0, r1, 0
>>      x14
> 
> Bisecting using:
> 
>       make print-V \
>    && ninja qemu-system-ppc \
>    && AVOCADO_ALLOW_UNTRUSTED_CODE=1 \
>       avocado --show=app,console \
>         run --test-runner=nrunner -t slowness:high \
>           tests/avocado/ppc_prep_40p.py
> 
> commit cef2e7148e32d61338de0220619d308bf42af770
> Author: Philippe Mathieu-Daudé <f4bug@amsat.org>
> Date:   Fri Feb 10 16:05:34 2023 +0100
> 
>      hw/isa/i82378: Remove intermediate IRQ forwarder
> 
>      When the i82378 model was added in commit a04ff940974 ("prep:
>      Add i82378 PCI-to-ISA bridge emulation") the i8259 model was
>      not yet QOM'ified. This happened later in commit 747c70af78f
>      ("i8259: Convert to qdev").
> 
>      Directly dispatch ISA IRQs to 'cpu_intr' output IRQ
>      by removing the intermediate i82378_request_out0_irq()
>      handler. Rename ISA IRQs array as 'isa_irqs_in' to
>      emphasize these are input IRQs.
> 
> Sorry :/ I'll look at this.

Zoltan noticed a similar problem for vt82c686 here: 
https://lists.gnu.org/archive/html/qemu-devel/2023-02/msg08767.html.

I think possibly you need to make use of qdev_connect_gpio_out() rather than using 
s->cpu_intr directly?


ATB,

Mark.