[Stabs]

Got one of the bastards with GDB!

Iam a total rube with programing so i couldnt grab that text out of myss to copy
so here is a picture, its PNG at least



this is related to it crashing from 2d/3d mode in rendermode 4

is there a way to run mapsters command line option through GDB?

[Plagman]

Yeah, using gdb --args <command line here>. However you won't get much information from a release build like that, especially with gdb. Try using this:

http://forums.duke4.net/index.php?s=&s...ost&p=59750

You can use gdb on the debugging binaries inside the latest snapshot or use the automatically-generated crashlog.

[Stabs]

here is the crash i frequently get, also

the mouse wheel when moved fast in 3d to move objects up and down needs a slight throttle because it just moves up and down randomly

copying maps with alt into other maps removes a sprites extra value.

gltexinvalidateall()
gltexinvalidate8()
Enabling desktop composition...
Setting video mode 1280x1024 (8-bit windowed)
Board autosaved to AUTOSAVE.MAP
Frame still locked 1 times when showframe() called.
Disabling desktop composition...
Setting video mode 1280x1024 (32-bit windowed)
Initializing Polymer subsystem...
PR : Board loaded.
PR : FBO #0 initialization failed.
PR : FBO #1 initialization failed.
PR : FBO #2 initialization failed.
PR : FBO #3 initialization failed.
PR : FBO #4 initialization failed.
PR : FBO #5 initialization failed.
PR : Initialization complete.
gltexinvalidateall()
gltexinvalidate8()

i hate that bug so much

attached is the whole log

 mapster32.log (55.23K)
Number of downloads: 350

[Stabs]

another crash log


 eduke32.txt (1.46K)
Number of downloads: 358

[Plagman]

The first one looks the same as the one The Commander was having, it looks like the hightile linked list is getting corrupted somehow.

The second one is also concerning, what exactly were you doing at the time?

[Stabs]

moving around in 2d and zooming i think

[Helixhorned]

Hm, looks like pixels are drawn where they shouldn't. Could you try the revision I just checked in and see whether the crashes disappear? There's a diagnostic white box in the lower left corner, does it light up often?

[Stabs]

will do, ill check when i get home, cheers for fixing that extra thing that makes life alot easier, was resorting to LEbuild for that , has the mouse wheel in 3d mode been fixed, its kinda annoying

[Stabs]

another log, this time art failed to load and then next time i went from 2d to 3d it crashed.

 mapster32.log (36.36K)
Number of downloads: 385

[Helixhorned]

What exactly is wrong with the mouse wheel? I could imagine that if you rotate it faster than than the time it take to do one main loop iteration, the surplus events are "eaten", but other than that I don't see anything erratic or random with it.

[Stabs]

if i scroll rapidly up in 3d mode it will actually go down, or just sit there going up & down.

[Stabs]

could you look into alt + tabbing out of mapster, most time it just crashes mapster now and if it dosnt it will probably hijack my mouse

I run it in windowed mode

edit : also when i copy a dummy tile sprite and press s its always tile 0

[Helixhorned]

Do you have a crashlog of that?

What do you mean by "dummy tiles"? Which picnum a newly inserted sprite gets is determined in this order:
1. (2D mode only) If you pressed a prefix key [1]-[0] on the top row, insert the picnum bound to it.
2. If the clipboard holds a sprite, insert its picnum
3. Insert the picnum that is most frequent

[Stabs]

well i use this in def

dummytilerange 5240 5450 xscale 2.0 yscale 2.0

and then i can keep importing models without having to assign an actual tile image, its those tiles that will paste as tile 0

i have no crashlog of those errors, i dont seem to get many of those

[Helixhorned]

That should be
dummytilerange 5240 5450 <x-dimension> <y-dimension>
Not sure what the dummy dimensions will effect but they have to be there because that's how the mere existence of a tile is determined. Also, aren't the scale values for the models?

If you don't have a mapster32.crashlog, you're probably running the release version and not mapster32.debug.exe.

[bioman]

Hello,

Found an issue with the _clipshape0.map file, when placed in same directory that the eduke32 binary or in a search path and have the Polymer activated, EDuke32 crash while running.
EDuke32 rev 1724 / Linux 64 bits.

Terminal output:

./eduke32[0x5d4bb7]
/lib/libc.so.6(+0x33c20)[0x7f2b9bb2cc20]
./eduke32(mspace_free+0x5c)[0x5ce856]
./eduke32[0x5d04ce]
./eduke32[0x5d0c77]
./eduke32[0x5d0e3b]
./eduke32(nedpfree+0x188)[0x5d371f]
./eduke32(nedfree+0x1d)[0x5d025a]
./eduke32(gloadtile_hi+0xd26)[0x571d20]
./eduke32(gltexcache+0x3b2)[0x56d19e]
./eduke32[0x5c4102]
./eduke32[0x5bc967]
./eduke32(polymer_loadboard+0x5c)[0x5b4a45]
./eduke32(polymer_init+0x10e)[0x5b45a3]
./eduke32(setgamemode+0x2d5)[0x55829a]
./eduke32(app_main+0x1322)[0x44cfed]
./eduke32(main+0x16d)[0x5d4ae4]
/lib/libc.so.6(__libc_start_main+0xfe)[0x7f2b9bb17d8e]
./eduke32[0x4257f9]
Segmentation fault

With gdb :

Program received signal SIGSEGV, Segmentation fault.
0x00000000005d03ef in RemoveCacheEntries (p=0x1ab4ca0, tc=0x7ffff7ec6428, age=32) at src/nedmalloc.c:699
699						(*tcb)->next=0;
(gdb) bt
#0  0x00000000005d03ef in RemoveCacheEntries (p=0x1ab4ca0, tc=0x7ffff7ec6428, age=32) at src/nedmalloc.c:699
#1  0x00000000005d0c77 in ReleaseFreeInCache (p=0x1ab4ca0, tc=0x7ffff7ec6428, mymspace=0) at src/nedmalloc.c:872
#2  0x00000000005d0e3b in threadcache_free (p=0x1ab4ca0, tc=0x7ffff7ec6428, mymspace=0, mem=0x7fffe23e2a78, size=131072) at src/nedmalloc.c:939
#3  0x00000000005d371f in nedpfree (p=0x1ab4ca0, mem=0x7fffe23e2a78) at src/nedmalloc.c:1423
#4  0x00000000005d025a in nedfree (mem=0x7fffe23e2a78) at src/nedmalloc.c:544
#5  0x000000000056fad3 in gloadtile_art (dapic=96, dapal=0, dameth=0, pth=0x7fffe23e2a10, doalloc=1) at src/polymost.c:1231
#6  0x000000000056d315 in gltexcache (dapicnum=96, dapalnum=0, dameth=0) at src/polymost.c:407
#7  0x00000000005c4102 in polymer_getbuildmaterial (material=0x7fffe23e1078, tilenum=96, pal=0 '\000', shade=0 '\000', cmeth=0) at src/polymer.c:4070
#8  0x00000000005bc967 in polymer_updatewall (wallnum=111) at src/polymer.c:2894
#9  0x00000000005b4a45 in polymer_loadboard () at src/polymer.c:775
#10 0x00000000005b45a3 in polymer_init () at src/polymer.c:618
#11 0x000000000055829a in setgamemode (davidoption=0 '\000', daxdim=1024, daydim=768, dabpp=32) at src/engine.c:8239
#12 0x000000000044cfed in app_main (argc=1, argv=0x7fffffffe218) at source/game.c:9870
#13 0x00000000005d4ae4 in main (argc=1, argv=0x7fffffffe218) at src/sdlayer.c:208

[Helixhorned]

Arr, more of them weird memory problems.
Could you try compiling with NEDMALLOC set to 0 in Makefile and build/Makefile? And if you're inclined, help me to test a hypothesis by entering the following commands in GDB after the crash (they just print the values of some variables):

frame 5
print xsiz
print ysiz

[bioman]

Failed to build with NEDMALLOC set to 0

Program received signal SIGSEGV, Segmentation fault.
0x00000000005d03ef in RemoveCacheEntries (p=0x1ab4ca0, tc=0x7ffff7ec6428, age=32) at src/nedmalloc.c:699
699						(*tcb)->next=0;
(gdb) frame 5
#5  0x000000000056fad3 in gloadtile_art (dapic=96, dapal=0, dameth=0, pth=0x7fffe1b178e8, doalloc=1) at src/polymost.c:1231
1231	    if (pic) Bfree(pic);
(gdb) print xsiz
$1 = 128
(gdb) print ysiz
$2 = 200

[Helixhorned]

Forgot to tell you to svn update, it should compile fine then.

[bioman]

Tried with rev 1726 and NEDMALLOC set to 0, build but crashs at startup without _clipshape0.map :

Program received signal SIGABRT, Aborted.
0x00007ffff68c7ba5 in raise () from /lib/libc.so.6
(gdb) bt
#0  0x00007ffff68c7ba5 in raise () from /lib/libc.so.6
#1  0x00007ffff68cb6b0 in abort () from /lib/libc.so.6
#2  0x00007ffff690143b in ?? () from /lib/libc.so.6
#3  0x00007ffff690b4b6 in ?? () from /lib/libc.so.6
#4  0x00007ffff6911c83 in free () from /lib/libc.so.6
#5  0x000000000044c090 in app_main (argc=1, argv=0x7fffffffe218) at source/game.c:9414
#6  0x00000000005d4b68 in main (argc=1, argv=0x7fffffffe218) at src/sdlayer.c:208

[Helixhorned]

Now this looks like a linking problem or there's something wrong with the configuration, because it's quite unlikely that the malloc/free code in two different places is wrong. Just to be sure, does

nm ./eduke32 | grep ned

display anything for the non-NEDMALLOC version?

[Plagman]

did you rebuild from scratch when changing NEDMALLOC? eg make veryclean && make NEDMALLOC=0

[bioman]

I set NEDMALLOC to 0 in the makefile, it builds a nedmalloc object file (compiled from scratch to be sure)

00000000005cfc07 t nedblkmstate
00000000005cfc51 T nedblksize
00000000005cfd0f T nedcalloc
00000000005d1309 T nedcreatepool
00000000005d169b T neddestroypool
00000000005d1a0e T neddestroysyspool
00000000005d2027 T neddisablethreadcache
00000000005cfd5d T nedfree
00000000005d1db6 T nedgetvalue
00000000005cfe27 T nedindependent_calloc
00000000005cfe56 T nedindependent_comalloc
00000000005cfda3 T nedmallinfo
00000000005cfcf0 T nedmalloc
00000000005cfe17 T nedmalloc_footprint
00000000005d38de T nedmalloc_internals
00000000005cfe07 T nedmalloc_stats
00000000005cfde8 T nedmalloc_trim
00000000005cfdc6 T nedmallopt
00000000005cfd7c T nedmemalign
00000000005d257a T nedpcalloc
00000000005d30b7 T nedpfree
00000000005d3ac6 T nedpindependent_calloc
00000000005d3d92 T nedpindependent_comalloc
00000000005d3723 T nedpmallinfo
00000000005d211b T nedpmalloc
00000000005d3a35 T nedpmalloc_footprint
00000000005d39b4 T nedpmalloc_stats
00000000005d391b T nedpmalloc_trim
00000000005d38bb T nedpmallopt
00000000005d3306 T nedpmemalign
00000000005d1bfc T nedpoollist
00000000005d29c3 T nedprealloc
00000000005d1d6b T nedpsetvalue
00000000005cfd36 T nedrealloc
00000000005cfcd1 T nedsetvalue
00000000005d4113 T nedstrdup
00000000005d1e27 T nedtrimthreadcache

I set NEDMALLOC=0 in the CLI, it does not build the object file but crash with _clipshape0.map :

(gdb) bt
#0  0x00007ffff68c7ba5 in raise () from /lib/libc.so.6
#1  0x00007ffff68cb6b0 in abort () from /lib/libc.so.6
#2  0x00007ffff690143b in ?? () from /lib/libc.so.6
#3  0x00007ffff690b4b6 in ?? () from /lib/libc.so.6
#4  0x00007ffff690f55f in ?? () from /lib/libc.so.6
#5  0x00007ffff6912254 in calloc () from /lib/libc.so.6
#6  0x00007fffe7aba242 in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#7  0x00007fffe7aba947 in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#8  0x00007fffe794f74e in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#9  0x00007fffe7952fd5 in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#10 0x00007fffe79557f1 in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#11 0x00007fffe7955e4f in ?? () from /usr/lib/nvidia-current/libnvidia-glcore.so.260.19.06
#12 0x000000000056e1fd in uploadtexture (doalloc=1, xsiz=23632, ysiz=0, intexfmt=6408, texfmt=6408, 
    pic=0x35a3bd0, tsizx=1, tsizy=1, dameth=0) at src/polymost.c:1104
#13 0x000000000056e6b6 in gloadtile_art (dapic=-32, dapal=0, dameth=0, pth=0x35a3b90, doalloc=1)
    at src/polymost.c:1206
#14 0x000000000056c085 in gltexcache (dapicnum=-32, dapalnum=0, dameth=0) at src/polymost.c:407
#15 0x00000000005c2e72 in polymer_getbuildmaterial (material=0x35a0318, tilenum=-32, pal=0 '\000', 
    shade=0 '\000', cmeth=0) at src/polymer.c:4070
#16 0x00000000005bb6d7 in polymer_updatewall (wallnum=0) at src/polymer.c:2894
#17 0x00000000005b37b5 in polymer_loadboard () at src/polymer.c:775
---Type <return> to continue, or q <return> to quit---
#18 0x00000000005b3313 in polymer_init () at src/polymer.c:618
#19 0x0000000000557197 in setgamemode (davidoption=0 '\000', daxdim=1024, daydim=768, dabpp=32)
    at src/engine.c:8239
#20 0x000000000044bf4d in app_main (argc=1, argv=0x7fffffffe218) at source/game.c:9870
#21 0x00000000005c75a9 in main (argc=1, argv=0x7fffffffe218) at src/sdlayer.c:208
(gdb) frame 5
#5  0x00007ffff6912254 in calloc () from /lib/libc.so.6
(gdb) print xsiz
$1 = 0
(gdb) print ysiz
$2 = 0
(gdb)

[Helixhorned]

bioman, on Nov 12 2010, 01:55 AM, said:

#15 0x00000000005c2e72 in polymer_getbuildmaterial (material=0x35a0318, tilenum=-32, pal=0 '\000',
shade=0 '\000', cmeth=0) at src/polymer.c:4070

Plagman?

bioman: it was necessary to set it both in Makefile and build/Makefile but that's fixed now.

[DavoX]

I'm using the latest snapshot and it mapster32 keeps crashing after switching from 2D to 3D after a few times... also it eats up a lot of memory over time.

[Plagman]

It must be reading that -32 value from somewhere. What map were you trying to run? Could it be that it's just corrupt?

[bioman]

The map is _clipshape0.map from the samples directory.
This map is automatically loaded at startup and crash at this place, otherwise no problems if i rename it and load the map from the game.

[Helixhorned]

Oh, I think then everything is clear. Polymer should not have gotten hold of the clipshape map because it's not rendered unless you play it. Can you verify that the newest build fixes the crash? Thanks for the patience!

[bioman]

Yes, the last rev fixes the crash, thanks.

[William Gee]

How do I turn polymer on in mapster32? and where do I go to learn all the new polymer effects for mapster32?