Re: [B2-devel] Tiger Display Problems

[Gwenole B. <gb....@fr...>]

Le vendredi, 1 jul 2005, =E0 00:00 Europe/Paris, Michael Dickison a =
=E9crit=20
:

> Yup, paranoia works! But I tried compiling the CVS today and I get the=20=

> following error:
>
> g++ -I../include -I. -DHAVE_CONFIG_H -D_REENTRANT=20
> -DDATADIR=3D\"/usr/local/share/SheepShaver\" -g -O2 -I/sw/include/SDL=20=

> -D_THREAD_SAFE -c main_unix.cpp -o obj/main_unix.o
> main_unix.cpp: In function `void TriggerInterrupt()':
> main_unix.cpp:1517: error: 'idle_resume' was not declared in this =
scope
> make: *** [obj/main_unix.o] Error 1

You generally need to update the Basilisk II tree as well. ;-)

>> Please tell me your screen depth. Since you could build from sources,=20=

>> I'd also appreciate to know the video modes reported by SDL (#define=20=

>> DEBUG 1 in video_sdl.cpp) + in Unix/video_blit.cpp, around line 570,=20=

>> remove the test for !blitter_found and the abort(), this will output=20=

>> the right RGB mask to use for your screen.
>
> I made the changes to the March snapshot, and in the Console I get=20
> three groups of messages, once at startup, second when the Mac OS=20
> desktop loads (startup is done), and lastly when I shutdown/quit (and=20=

> my screen is 1024x768@Thousands of colors):

[...]

> Current video mode:
> 640x480 (ID 81), 32 bpp
> the_buffer =3D 0x47c5000, the_buffer_copy =3D 0x465b000
> ### No appropriate blitter found
>     R/G/B mask values  : 0xff0000, 0x00ff00, 0x0000ff (depth =3D 32)
>     R/G/B shift values : 16/8/0
> monitor.mac_frame_base =3D 047c5000

Seems like SDL guessed a best mode with depth greater than what's is=20
available on your system, or I need some other heuristics to determine=20=

the current screen depth.

Please try:
<http://gwenole.beauchesne.free.fr/sheepshaver/sdl_video_info.c>

bash-2.05a$ gcc sdl_video_info.c $(sdl-config --cflags --libs)
bash-2.05a$ ./a.out
Max screen dimensions: 1152x768
Current screen depth:  16 bpp
SDL supports 1152x768x16 with a HWSURFACE (windowed)
SDL supports 1152x768x16 with a HWSURFACE (fullscreen)

> By gdb I think you mean the CrashReporter log? Anyway, this is the=20
> long message I get per crash, I cannot make any sense of it but=20
> hopefully it helps:
>
> Thread: 0
>
> Exception:  EXC_BAD_ACCESS (0x0001)
> Codes:      KERN_PROTECTION_FAILURE (0x0002) at 0x40810000
>
> Thread 0 crashed with PPC Thread State:
>   srr0: 0x40c6624c srr1: 0x0200f930                vrsave: 0x00000000
>     cr: 0x4010e0f1  xer: 0x00000000   lr: 0x40ca7388  ctr: 0x40c60000
>     r0: 0x00000000   r1: 0x203fffea   r2: 0x00000000   r3: 0x40810000
>     r4: 0x2054500c   r5: 0x00000000   r6: 0x00002070   r7: 0x00000000
>     r8: 0x00000000   r9: 0x2054500c  r10: 0x00000006  r11: 0x000002f4
>    r12: 0x00000861  r13: 0x00000000  r14: 0x00000000  r15: 0x0002003d
>    r16: 0x40810000  r17: 0x00000000  r18: 0x408153de  r19: 0x4085975c
>    r20: 0x40814c0e  r21: 0x23ffff88  r22: 0x4080020a  r23: 0x00000000
>    r24: 0x40815532  r25: 0x00000027  r26: 0x00000000  r27: 0x00006054
>    r28: 0x00000000  r29: 0x40ca7388  r30: 0x40c60000  r31: 0x68fff000

One of the 68k ROM routine tried to write into the ROM... The weird=20
thing is that we normally catch those kind of errors exactly. Is=20
CrashReporter overriding our Mach exception handler? Let's see with:
<http://gwenole.beauchesne.free.fr/sheepshaver/sigsegv_standalone.cpp>

1) In default mode with: g++ sigsegv_standalone.cpp, you must have this=20=

file in the BasiliskII/src/Unix/ directory
2) With SDL: g++ sigsegv_standalone.cpp  -DUSE_SDL $(sdl-config=20
--cflags --libs)

If CrashReporter appears in SDL mode, try to uncomment the=20
SDL_INIT_NOPARACHUTE part. You normally should get on the console=20
something like:

bash-2.05a$ ./a.out
Checking with SDL
sigsegv_test_handler(0x185207b, 0x3a7c)
expected fault at 0x185207b
expected instruction address range: 0x3a68-0x3ac0
sigsegv_insn_handler(0x1852001, 0x3c28)
00003c28: byte read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852002, 0x3c80)
00003c80: word read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852004, 0x3cd8)
00003cd8: long read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852004, 0x3d2c)
00003d2c: long read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852001, 0x3d80)
00003d80: byte read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852002, 0x3dd8)
00003dd8: word read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852004, 0x3e30)
00003e30: long read access r0 (rd =3D 0)
sigsegv_insn_handler(0x1852004, 0x3e84)
00003e84: long read access r0 (rd =3D 0)
do_test() returned 0

Bye,
Gwenol=E9.=