Re: [hatari-devel] Issues after recent WinUAE CPU core changes

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 01/06/2020 à 15:44, Eero Tamminen a écrit :
> Hi,
>
> Latest Hatari Git sources break now Linux reliably, when trying to mount 
> /proc at start, or run any other (Busybox) user-space program than "cd" 
> or "echo" ("echo *" can be used to list files):
> -----------------------------------------------
> Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
> CPU: 0 PID: 1 Comm: sh Not tainted 5.6.0hatari-g515609e7b #3
> Stack from 00821de8:
>          00821de8 002d0816 00821e14 0002777e 0000000b 000000ff 400004d8 
> 0002ff16
>          0081ea50 00820000 00824d30 00821e54 00028f82 002ca018 0000000b 
> 0000000b
>          00000001 400004d8 0002ff16 0002dd80 00823af8 00821f18 00821f2c 
> 00824d30
>          00000000 00000000 000000cd 00821e64 00029554 0000000b 0000000a 
> 00821e9c
>          00030882 0000000b 00000000 00000002 00000000 00000000 00000000 
> 8017705e
>          00821f78 00821f18 00820000 00823a30 00821f78 00821f50 000041da 
> 00821f18
> Call Trace: [<0002777e>] panic+0xc2/0x27e
>   [<0002ff16>] do_signal_stop+0x0/0x14c
>   [<00028f82>] do_exit+0x126/0x676
>   [<0002ff16>] do_signal_stop+0x0/0x14c
>   [<0002dd80>] __list_del_entry+0x0/0x2c
>   [<00029554>] do_group_exit+0x4c/0x84
>   [<00030882>] get_signal+0x104/0x464
>   [<000041da>] do_notify_resume+0x32/0x460
>   [<00005cc4>] send_fault_sig+0x28/0x8c
>   [<00005ec6>] do_page_fault+0x19e/0x204
>   [<00002ac8>] do_signal_return+0x10/0x1a
>   [<000066f6>] get_pointer_table+0x104/0x16a
> -----------------------------------------------
>
> Because it says "Attempted to kill init", it means that the shell 
> running the command dies, not necessarily the command itself.
>
>
> Issue goes away only when I disable *both* 030 cache emulation and 
> prefetch with:
> --cpu-exact off --compatible off
>
> Disabling just one of them isn't enough.
>
>
> Hatari command line I'm using is following:
> hatari --machine falcon --dsp none --fpu 68882 --cpuclock 32 --mmu on -s 
> 14 --ttram 64 --addr24 off -c lilo.cfg --natfeats on --ide-master 
> bb-rootfs.img --lilo "debug=nfcon root=/dev/sda ro init=/bin/sh
>
> (Using e.g. parport instead of natfeats for debugging didn't change 
> anything.)
>
>
> Any ideas what could have changed, or should I just bisect it blindly? :-)

Hi

things are changing quite often recently.

I pushed some changes yesterday, did they break your test ?

Or toni fixed a week ago some problems that caused Amix not to boot 
under Amiga ; maybe that's the ones related to 68030 that could explain 
your problem.

In all cases, if you can bisect, I think it will be much faster to fix 
as WinUAE commits are rather short most of the time, so there should't 
be hundreds of lines to check.

Nicolas