Re: [hatari-devel] Bug report: AHCC not working anymore with GEMDOS emulation

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On keskiviikko 25 tammikuu 2012, Matthias Arndt wrote:
> Am Mittwoch, den 25.01.2012, 00:19 +0200 schrieb Eero Tamminen:
> > > Matthias, there's a new bugfixed AHCC v4.9 release, can you reproduce
> > > the issue also with that?
> 
> Same behavior as v4.8 here, AHCC TTP borks out with 2 bombs on my
> example project. (This project compiles if I use an ACSI image with the
> same contents!)
> 
> > And if it still happens, could you give "--trace gemdos" output for it?
> 
> I have attached my trace,

Hmm, you have lots of AUTO programs.   I wonder why TOS & XCONTROL open
all ACCs and CPXes twice...

What HDDRCONF.CPX does?


> please note that my Hatari build is a few days old.

That's OK.


> > I started to write a gemdos tester as part of a major tos bootup
> > tester update, and just found out that GEMDOS HD emulation breaks
> > the GEMDOS Fopen() special device ("AUX:", "PRN:", maybe also "CON:")
> > handing.  They're handled as regular files with it.

I commited a fix to this, but as such names were not listed
in your trace, it cannot affect the issue.


> > I think that should be easy to fix, I'll try to look at it tomorrow.
> 
> I'm looking forward. I more and more stronges suspect some GEMDOS
> related bugs here.

Are all your files read-only, including the AHCC intermediate files
or are you using Hatari's GEMDOS dir --protect-hd option (there's
also a checkbox for that in GUI)?

I remember there being some earlier AHCC crasher bug related to
read-only files.

--------------------
GEMDOS Fopen("C:\CODING\C\WIZZCAT\WIZZCAT.PRJ", 0x0)
GEMDOS: C:\CODING\C\WIZZCAT\WIZZCAT.PRJ -> host: 
/emulators/AtariST/SatanDisk/CODING/c/wizzcat/wizzcat.prj
-> FD 0 (read-only)
GEMDOS Fread(64, 1024, 0x118e8e)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x48 (Malloc)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS Fread(64, 1024, 0x118e8e)
GEMDOS Fclose(64)
GEMDOS Fwrite(1, 1, 0x1519f0)
GEMDOS Fopen("C:\CODING\C\WIZZCAT\WIZZCAT.TOS", 0x0)
GEMDOS didn't find filename 
/emulators/AtariST/SatanDisk/CODING/c/wizzcat/WIZZCAT.TOS
GEMDOS Fdatime(0x16bdbc, 65503, 0)
GEMDOS Fclose(65503)
GEMDOS Fclose(65503)
GEMDOS call 0x124 (MiNT?)
GEMDOS call 0x124 (MiNT?)
GEMDOS Fopen("C:\CODING\C\WIZZCAT\MAIN.O", 0x0)
GEMDOS didn't find filename 
/emulators/AtariST/SatanDisk/CODING/c/wizzcat/MAIN.O
GEMDOS call 0x124 (MiNT?)
GEMDOS Fwrite(1, 1, 0x151893)
GEMDOS Fwrite(1, 14, 0x1523d2)
GEMDOS call 0x48 (Malloc)
GEMDOS call 0x124 (MiNT?)
GEMDOS Fopen("C:\CODING\C\WIZZCAT\MAIN.C", 0x0)
GEMDOS: C:\CODING\C\WIZZCAT\MAIN.C -> host: 
/emulators/AtariST/SatanDisk/CODING/c/wizzcat/main.c
-> FD 0 (read-only)
GEMDOS call 0x124 (MiNT?)
GEMDOS Fseek(0, 64, 2)
GEMDOS Fseek(0, 64, 0)
GEMDOS call 0x48 (Malloc)
GEMDOS Fread(64, 863, 0x16bf5a)
GEMDOS Fclose(64)
GEMDOS call 0x48 (Malloc)
M68000 Bus Error reading at address $0.
--------------------

This doesn't show anything failing, so next thing would
be to get trace of all OS calls.

There's a bit less data if you set breakpoint on program
start entry point in the debugger:
	b pc = TEXT :once
	c
And then drag the PRJ file on AHCC.

After the breakpoint triggers, ask everything to be traced:
	trace os_all
	c

You can disable the trace when you get the bus error:
	trace none
	c


	- Eero