Bug: rogomatic crashes on launch

[lcn2]

As of 2026 February 22, rogomatic cannot be start as it crashes on launch:

./rogomatic
Error opening terminal: rg.
                           zsh: bus error  ./rogomatic

[lcn2]

Using lldb:

lldb -- ./rogomatic

Setting a breakpoint in main:

b main

we trace the progress until we execute the player program:

Process 64524 stopped
* thread #1, queue = 'com.apple.main-thread', stop reason = step over
    frame #0: 0x00000001000026e0 rogomatic`main(argc=0, argv=0x000000016fdfe978) at setup.c:156:5
   153
   154       if (!author ()) nice (4);
   155
-> 156       execl ("player", "player", ft, rp, options, roguename, NULL);
   157   # ifdef PLAYER
   158       execl (PLAYER, "player", ft, rp, options, roguename, NULL);
   159   # endif
Target 0: (rogomatic) stopped.
(lldb) p ft
(char[3]) "fe"
(lldb) p rp
(char[257]) "64529"
(lldb) p options
(char[257]) "0,1,0,0,0,0,0,-1"
(lldb) p ro
Available completions:
        ropts
        roguename
(lldb) p roguename
(char[257]) "Rog-O-Matic XIV for chongo"]

It is the player program that crashes.

We need to figure out how to get lldb to set a breakpoint in the player code (set a breakpoint across an excel(3)` call.

[lcn2]

We compile with COPT= -O0 and DEBUG= -gdb3. You can either modify the Makefile, or just the command line:

make clobber all COPT="-O0" DEBUG="-gdb3"

We have these commands in the file: .lldbinit

b execl
b setup.c:350
b main.c:350

We run ./rogomatic via the lldb(1) debugger:

lldb -- ./rogomatic
Breakpoint 1: no locations (pending).
Breakpoint set in dummy target, will get copied into future targets.
Breakpoint 2: no locations (pending).
Breakpoint set in dummy target, will get copied into future targets.
Breakpoint 3: no locations (pending).
Breakpoint set in dummy target, will get copied into future targets.
(lldb) target create "./rogomatic"
Current executable set to '/usr/local/src/games/rogomatic/rogomatic' (arm64).

Then we do:

(lldb) b io.c:477
Breakpoint 4: no locations (pending).
WARNING:  Unable to resolve breakpoint to any actual locations.
(lldb) run
Process 6330 launched: '/usr/local/src/games/rogomatic/rogomatic' (arm64)
Process 6330 stopped
* thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.1
    frame #0: 0x0000000197ce57f4 libsystem_c.dylib`execl
libsystem_c.dylib`execl:
->  0x197ce57f4 <+0>:  pacibsp
    0x197ce57f8 <+4>:  stp    x22, x21, [sp, #-0x30]!
    0x197ce57fc <+8>:  stp    x20, x19, [sp, #0x10]
    0x197ce5800 <+12>: stp    x29, x30, [sp, #0x20]
Target 0: (rogomatic) stopped.
(lldb) c
Process 6330 resuming
Process 6330 stopped
* thread #2, stop reason = exec
    frame #0: 0x00000001002449c0 dyld`_dyld_start
dyld`_dyld_start:
->  0x1002449c0 <+0>:  mov    x0, sp
    0x1002449c4 <+4>:  and    sp, x0, #0xfffffffffffffff0
    0x1002449c8 <+8>:  mov    x29, #0x0 ; =0
    0x1002449cc <+12>: mov    x30, #0x0 ; =0
Target 0: (player) stopped.
1 location added to breakpoint 3
1 location added to breakpoint 4
(lldb) c
Process 6330 resuming
Process 6330 stopped
* thread #2, queue = 'com.apple.main-thread', stop reason = breakpoint 3.1
    frame #0: 0x000000010009273c player`main(argc=5, argv=0x000000016fdfe9c0) at main.c:350:3
   347 	int
   348 	main (int argc, char *argv[])
   349 	{ char  ch, *s;
-> 350 	  char msg[SM_BUF + 1];	/* message buffer, +1 for paranoia */
   351 	  int startingup = 1;
   352 	  int i;
   353
Target 0: (player) stopped.
(lldb) c
Process 6330 resuming
Process 6330 stopped
* thread #2, queue = 'com.apple.main-thread', stop reason = breakpoint 4.1
    frame #0: 0x0000000100070ab0 player`sendcnow(c=118) at io.c:477:12
   474 	    else
   475 	      fprintf (fecho, "%c", c);
   476 	  }
-> 477 	  fprintf (trogue, "%c", c);
   478 	}
   479
   480 	/*
Target 0: (player) stopped.
(lldb) p trogue
(FILE *) 0x00000002040908e8
(lldb) n
Process 6330 exited with status = 13 (0x0000000d) Terminated due to signal 13

[lcn2]

With commit d0573bd, rogomatic no longer crashes with a "~ bus error ./rogomatic~".

Under MacOS, rogomatic now will start to drive rogue 5.4.5, only to hang after a brief period of time.

Under RHEL 9.7 Linux, rogomatic now will exit with a "Error opening terminal: rg." error.

[lcn2]

With commit ce667fe, commit 7c3d2c8, and commit 916976d:

Under macOS, rogomatic will run for a second or two and then hang:

your hands begin to glow red
 ---------------
 |.............|                         --------        --------------------
 |.............|                         |......+#####   |..................|
 |.............|          ###############+......|    #   |..................|
 |.............+###########              |......|    ####+..................|
 ---------------                         --------        -------------+------
                                                                      #
                                                                      #
                                                                      #######
                           -----------                                      #
                       ####+.........|                                      #
                ########   |@........+################                      #
                #          -+---------               #                      #
                #           #                        ########################
                #           #                                    ############
                ####        #                        ------------+-------
                   #        #                      ##+..................|
                   #        #########              # |..................|
                   #            ----+-             # |..................|
                   #            |....|             # |..................|
                   #            |....+############## |..................|
                   #            ------               --------------------
Level: 2  Gold: 7      Hp: 22(22)  Str: 16(16)  Arm: 4   Exp: 3/21

Under RHEL 9.7 Linux, rogomatic now will exit with a "Error opening terminal: rg." error but not crash.

UPDATE 0

With commit d71b44c and macOS Address Sanitizer (ASAN) enabled under macOS, detects a problem:

==15616==ERROR: AddressSanitizer: global-buffer-overflow on address 0x000100e38020 at pc 0x0001016b293c bp 0x00016f197db0 sp 0x00016f197570
READ of size 81 at 0x000100e38020 thread T0
     #0 0x0001016b2938 in strlen+0x1b4 (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x7a938)
     #1 0x000197ca0970 in vsscanf_l+0x118 (libsystem_c.dylib:arm64e+0x9970)
     #2 0x0001016530b8 in sscanf+0x58 (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x1b0b8)
     #3 0x000100cd2100 in terpbot io.c:343
     #4 0x000100ae42c8 in getrogue io.c:244
     #5 0x000100af152c in getrogver io.c:911
     #6 0x000100b0fcfc in main main.c:427
     #7 0x000197a41d50  (<unknown module>)

0x000100e38020 is located 0 bytes after global variable 'screen' defined in 'main.c' (0x000100e378a0) of size 1920

SUMMARY: AddressSanitizer: global-buffer-overflow (libsystem_c.dylib:arm64e+0x9970) in vsscanf_l+0x118

[jenoola]

pull the latest version of rogue and rogomatic, the crash is still observed on macOS

[lcn2]

I winder of setting the recommend rogue options:

export ROGUEOPTS=noterse,flush,jump,seefloor,passgo,tombstone,inven=overwrite,fruit=apple

Is a good idea?

UPDATE 0

Doesn't matter as rogomatic will assign the ROGUEOPTS environment variable a specific value.

[lcn2]

Well @jenoola ,

It turns out that there is NO portable way to supply a terminfo in the environment. The only portable way is to have a terminfo tree. One cannot supply a compiled terminfo contents as the compiled terminfo stuff is not portable. You can only start with the rterm.tic file and compile the rt terminfo under a directory tree and then point $TERMINFO at the top of that tree and set '$TERM'.

The "rt" terminal is the major complication.

The old code for around this by using the vt100 terminal.

The vt100 terminal is "nearly" universal. So the old code got away that approach because most systems have a vt100 terminal description.

... stay tuned ...

UPDATE 0

I see 3+ approaches to a portable solution:

0. compile the term.tic into a local terminfo tree

1. change the rogomatic code to parse the vt100 terminal

2. revert to using old rogomatic code with the idea of undoing the GNU autoconf stuff

3. form a vt100 branch, copy over old rogomatic code and get that code to compile cleanly using the existing Makefile

UPDATE 1a

I am inclined to try options (0) and (3).

[lcn2]

Hello @jenoola ,

Good news, we have in the vt100 brach, code that is able to drive rogue 5.4.5 for a bit .. until rogomatic gets confused and hangs. It does this on both macOS and Linux. I believe this will be a much better code base to debug and improve.

The vt100 branch assumes a vt100 terminal, a generic terminal type that is nearly universal. Thus we avoid the whole "rg terminal" stuff as the vt100 branch parses the vt100 terminal codes.

I have created a branch called old-codebase which is the current contents of master (with a few fixes). I will keep this branch around for a while in case we want to look at the code we were debugging on Friday.

Given that a copy of master is in old-codebase, I plan to now replace master with the contents of vt100.

UPDATE 0

After making sure that the master (the code were debugging on Friday), old-codebase (a copy of master), and vt100 (the code that is working better) branches are up to date, I will now replace master with the contents of vt100:

git checkout vt100
git merge -s ours master

Then fast forward the master branch:

git checkout master
git merge vt100

Then push the changes to the remote:

git push --force origin master

Well that worked. :-)

UPDATE 1

Please, @jenoola , refetch the master branch and give it a try.

The code should let you start to run rogue, but depending on the random number generator, it will hand eventually: sometimes after a short run, sometimes after a longer run.

If you are able to build with the updated master branch and not "crash" but drive rogue for a bit, then we can close the bug as fixed.

.. of course there is more work to fix the hanging problems! :-)