FS-UAE 1.3 Development Series

[FrodeSolheim]

Originally Posted by Foul

is there a way to disable floppy overlay ? i'm making some custom .adf floppies... but when i want to copie them, it's just the Original floppy, not modified...

I should make this optional, but right now it isn't. (I suppose you can "cheat" by using an ADF tool inside the Amiga and saving the ADF read from a floppy drive to a mounted hard drive folder).

[Foul]

not a bad idea

[gibs]

Hi

Somebody manage to boot on a HDF file with a CPU better than a 68020 ?

[Foul]

I'm using a copy of my real sdcard system and boot without prob on it. (68060 based system)

My old 040 hdf based system works too...

What is your prob ?

[grimi]

Hi Frode. Can you upload source code tarball 1.3.17 please? Thanks

[FrodeSolheim]

Originally Posted by grimi

Hi Frode. Can you upload source code tarball 1.3.17 please? Thanks

Uploaded!

[FrodeSolheim]

Version 1.3.18:

• Added option audio_buffer_target_bytes (defaults to 8192).
• Reduced internal uae audio buffer from 1024 to 512 bytes.
• Added logitech_r_precision_tm_gamepad_10_2_0_0_windows.conf [Paul].
• (Launcher) Search field allows to search for multiple (AND-ed) terms.
• (Launcher) Also index .adz files in the database.
• (Launcher) Fixed bug where paths relative to Floppies dir were not found by the launcher.

http://fengestad.no/wp/fs-uae/download-beta

Documentation for the new option is available here: http://fengestad.no/wp/fs-uae/audio-options

(There have been several releases which were only announced in the FS-UAE launcher thread since these mainly releases mainly updated the launcher).

[krutknut]

I'm logging on to say that this works really great
Thank's for sharing

Now, if there was also a way to convert the old gamebase amiga database to work with your frontend...

[FrodeSolheim]

Originally Posted by krutknut

I'm logging on to say that this works really great
Thank's for sharing

And thank you for dropping by and mentioning it

Originally Posted by krutknut

Now, if there was also a way to convert the old gamebase amiga database to work with your frontend...

Well, the licensing terms of GameBase Amiga forbids such use of the database anyway...

One alternative is to assemble a new database with a permissive license, see the threads http://eab.abime.net/showthread.php?t=65057 ("[FS-UAE] Amiga Game Database for FS-UAE Launcher") and http://eab.abime.net/showthread.php?t=64531 ("FS-UAE Launcher").

[Foul]

just installed BitchX (http://aminet.net/package/comm/irc/BitchX-11) and fs-uae crash at launch .. exactly like Netsurf... :\

[FrodeSolheim]

Originally Posted by Foul

just installed BitchX (http://aminet.net/package/comm/irc/BitchX-11) and fs-uae crash at launch .. exactly like Netsurf... :\

I installed ixemul and then BitchX on a A4000/040 setup (Linux x86-64), and it did not crash for me. It did not connect to anything however (got "unimplemented function" errors in the BitchX console, probably some unimplement socket function -haven't checked which one yet).

I ran with the default '040 CPU and installed the library "ixemul-040.library" -perhaps you can try with 040 instead of 060? Also check that the ixemul/ixnet libraries in LIBS corresponds to the CPU? Also perhaps use A4000 kickstart instead of A1200 kickstart (don't know if it should matter -of course, FS-UAE shouldn't crash anyway..)

If you can compile your own version of FS-UAE, then you could run in through gdb and get a stack trace when it crashes...

[Foul]

all my libs are 060 optimized, if i try with A4000/040 without uae_cpu_model=68060, some progies don't work/crash (but not fs-uae !). i'm using HSMathLibs too (http://www.hsmathlibs.de/ )

tryed ixemul 48.3 and lastest 63+.. Netsurf want 63+ libs .. and Bitch-X crash fs-uae with 48.3 libs (all 060/fpu version)

best way is to compile myself and do a trace like you said

launched fs-uae under gdb .. click on Netsurf and ... My Whole system completly frozen !!! --> Hard Reboot :\

[FrodeSolheim]

Originally Posted by Foul

all my libs are 060 optimized, if i try with A4000/040 without uae_cpu_model=68060, some progies don't work/crash (but not fs-uae !). i'm using HSMathLibs too (http://www.hsmathlibs.de/ )

tryed ixemul 48.3 and lastest 63+.. Netsurf want 63+ libs .. and Bitch-X crash fs-uae with 48.3 libs (all 060/fpu version)

best way is to compile myself and do a trace like you said

now : how to use gdb ?

I was under the impression that there is no real advantage to running 68060 in emulation (and the cycle speed is dependent on the host CPU, not 040 vs 060).

Well, anyway, to run in gdb:

cd fs-uae
make
gdb --args out/fs-uae [your additional fs-uae arguments if any]

You should run in window mode and release the mouse/input grab before it crashes, or the input will still be grabbed when the process is halted through gdb (unfortunately). So when it crashes, you can type "bt" in the gdb terminal to get the stack trace. Optimizations may make debugging difficult, so you may need to compile with make optimize=0 instead.

[FrodeSolheim]

Originally Posted by Foul

My Whole system completly frozen !!! --> Hard Reboot :\

Ah, I just mentioned it in the reply -the system did not freeze but the input was still captured and you probably run FS-UAE fullscreen which made the syste appear frozen. In this case, you could have switched to a virtual console (Ctrl+Alt+F2 for instance) and killed the gdb process.

[Foul]

[New Thread 0x7fffb5ffb700 (LWP 3335)]
[Thread 0x7fffb5ffb700 (LWP 3335) exited]
[New Thread 0x7fffb5ffb700 (LWP 3336)]
[Thread 0x7fffb5ffb700 (LWP 3336) exited]
[New Thread 0x7fffb5ffb700 (LWP 3337)]
[New Thread 0x7fffb4ff9700 (LWP 3338)]
[Thread 0x7fffb5ffb700 (LWP 3337) exited]
[New Thread 0x7fffb5ffb700 (LWP 3339)]
[Thread 0x7fffb5ffb700 (LWP 3339) exited]
FIXME: not calling mman_ResetWatch (p96ram_start + natmem_offset, allocated_gfxmem);
FIXME: not calling mman_ResetWatch (p96ram_start + natmem_offset, allocated_gfxmem);
FIXME: not calling mman_ResetWatch (p96ram_start + natmem_offset, allocated_gfxmem);
>   0   0 800 600
FIXME: adding bit 5 of write to BEAMCON0
[New Thread 0x7fffb5ffb700 (LWP 3341)]
[Thread 0x7fffb5ffb700 (LWP 3341) exited]
[New Thread 0x7fffb5ffb700 (LWP 3342)]
[Thread 0x7fffb5ffb700 (LWP 3342) exited]
[New Thread 0x7fffb5ffb700 (LWP 3343)]
[Thread 0x7fffb5ffb700 (LWP 3343) exited]
[New Thread 0x7fffb5ffb700 (LWP 3344)]

Program received signal SIGBUS, Bus error.
[Switching to Thread 0x7fffb5ffb700 (LWP 3344)]
0x00000000005277bf in host_IoctlSocket (context=<optimized out>, sb=0x7fffa4001500, sd=<optimized out>, request=2147772030, arg=283439724)
    at src/od-fs/bsdsocket_host.cpp:1633
1633                sb->ftable[sd-1] &= ~SF_BLOCKING;
(gdb) bt
#0  0x00000000005277bf in host_IoctlSocket (context=<optimized out>, sb=0x7fffa4001500, sd=<optimized out>, request=2147772030, 
    arg=283439724) at src/od-fs/bsdsocket_host.cpp:1633
#1  0x00000000004ef74e in trap_thread (arg=0x7fffd035fff0) at src/traps.cpp:235
#2  0x00007ffff74ab9e5 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#3  0x00007ffff58cfe9a in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
#4  0x00007ffff55fd4bd in clone () from /lib/x86_64-linux-gnu/libc.so.6
#5  0x0000000000000000 in ?? ()
(gdb)

May it help ?

[FrodeSolheim]

Originally Posted by Foul

... May it help ?

Possibly If you unzip the attached patch, you can do:

cd fs-uae
patch -p1 < /path/to/debug-socket-problem.patch
make

and then run fs-uae again -the output may give some more hints as to what the problem is.

Btw, I forgot to enable bsdsocket_library option when I tested, but I now only get a request to insert AmiTCP volume when I try to start BitchX (?)

[Foul]

where is the file ?

[FrodeSolheim]

Originally Posted by Foul

where is the file ?

Strange, I thought I attached it (probably forgot to press the upload button).

[Foul]

trying...

FIONBIO sb (0x7fffa4001500)->ftable[-1] == 
ERROR: sd - 1 (-1) >= sb->dtablesize (64)

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffb5ffb700 (LWP 4606)]
0x000000000052781e in host_IoctlSocket (context=<optimized out>, sb=0x7fffa4001500, sd=<optimized out>, request=2147772030, arg=283419804) at src/od-fs/bsdsocket_host.cpp:1638
1638                sb->ftable[sd-1] &= ~SF_BLOCKING;
(gdb) bt
#0  0x000000000052781e in host_IoctlSocket (context=<optimized out>, sb=0x7fffa4001500, sd=<optimized out>, request=2147772030, arg=283419804) at src/od-fs/bsdsocket_host.cpp:1638
#1  0x00000000004ef74e in trap_thread (arg=0x7fffd03603b0) at src/traps.cpp:235
#2  0x00007ffff74ab9e5 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#3  0x00007ffff58cfe9a in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
#4  0x00007ffff55fd4bd in clone () from /lib/x86_64-linux-gnu/libc.so.6
#5  0x0000000000000000 in ?? ()
(gdb)

Still crashing...

[FrodeSolheim]

Originally Posted by Foul

Still crashing...

It wasn't supposed to fix it ;-) -But at least we can see the direct cause of the crash (assuming the sb pointer is valid), sd is 0 which causes sd - 1 to overflow to a large integer and the program reads from invalid memory. Anyway, it is easy to fix the crash -the code assumes that sb >= 1, and we can check for this and don't do anything if sd is 0. This will fix the crash, but it's likely this is just a symptom -the real bug/question is why sd is 0 in the first place which it clearly should not be.

But you can test anyway : in src/od-fs/bsdsocket_host.cpp, after the line:

case 0x8004667E: /* FIONBIO */

add five lines so you end up with:

case 0x8004667E: /* FIONBIO */
    if (sd == 0) {
        printf("WARNING: sd was 0 ???\n");
        bsdsocklib_seterrno(sb, EINVAL); // ?
        return -1;
    }