crash at revision e1846008ed75

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Build wmii from hg at e1846008ed75 or newer.
2. Launch xterm and then run wmii in your xinitrc.
3. Watch it crash instantly.

What is the expected output? What do you see instead?

Here is the first revision this problem occurs with:
http://hg.suckless.org/wmii/rev/e1846008ed75

Here is the full GDB trace of the problem:

[New Thread 19556]
Core was generated by `wmii'.
Program terminated with signal 11, Segmentation fault.
#0  0xb76412c0 in XCreateWindow () from /usr/lib/libX11.so.6
#0  0xb76412c0 in XCreateWindow () from /usr/lib/libX11.so.6
No symbol table info available.
#1  0x08070f88 in createwindow_visual (parent=0x8085a64, r=..., depth=24, 
    vis=0xe0002d, class=1, wa=0xbfd1fb74, valmask=2560)
    at x11/windows/createwindow_visual.c:32
        w = 0x894eef0
        wa_empty = {background_pixmap = 143881096, background_pixel = 1, 
          border_pixmap = 2097151, border_pixel = 14, bit_gravity = 128, 
          win_gravity = 14, backing_store = 104, backing_planes = 3077701664, 
          backing_pixel = 0, save_under = -1219502354, 
          event_mask = -1076757768, do_not_propagate_mask = -1218137904, 
          override_redirect = 143860272, colormap = 15, cursor = 112}
        __PRETTY_FUNCTION__ = "createwindow_visual"
#2  0x08070e6a in createwindow (parent=0x8085a64, r=..., depth=24, class=1, 
    wa=0xbfd1fb74, valmask=2560) at x11/windows/createwindow.c:8
No locals.
#3  0x0804c3b4 in bar_init (s=0x894ea10) at bar.c:30
        wa = {background_pixmap = 0, background_pixel = 3218209720, 
          border_pixmap = 143976976, border_pixel = 143857720, 
          bit_gravity = 14680112, win_gravity = 0, backing_store = 134664834, 
          backing_planes = 143976976, backing_pixel = 0, save_under = 56, 
          event_mask = 2129932, do_not_propagate_mask = -1076757480, 
          override_redirect = 1, colormap = 143923056, cursor = 143976976}
#4  0x0805bb93 in init_screens () at main.c:213
        rects = 0x894e838
        v = 0x0
        i = 0
        n = 1
        m = 0
#5  0x0805c587 in main (argc=0, argv=0xbfd1fd98) at main.c:423
        oargv = 0xbfd1fd94
        wmiirc = 0x807a7f0 "wmiirc"
        i = -1218609164
        _argtmp = 0
        _inargv = 0
        _argv = 0x0

Original issue reported on code.google.com by sunaku on 26 Jun 2010 at 12:18

[GoogleCodeExporter]

Is this from a clean build? I can't reproduce it. If a clean build doesn't fix 
it, can you try running wmii from valgrind?

Thanks.

Original comment by maglion...@gmail.com on 26 Jun 2010 at 12:31

[GoogleCodeExporter]

Yes, it still occurs with a completely clean build.  Attached is valgrind 
output.

Original comment by sunaku on 26 Jun 2010 at 2:00

Attachments:

• wmii-hg2743.log

[GoogleCodeExporter]

It doesn't seem to indicate a segfault. Or any other memory errors. Did it 
crash in that instance?

Also, leak-check=full adds a lot of noise without providing any useful 
information. It's possible that I've missed something.

Original comment by maglion...@gmail.com on 26 Jun 2010 at 2:23

[GoogleCodeExporter]

Argh, sorry.  I did a clean builds & installs again, testing one commit at a 
time, and found that it was the bar file changes causing problems in my wmiirc:

http://hg.suckless.org/wmii/rev/9d31d13386b8

I'll do completely clean builds & installs from here onwards.

Original comment by sunaku on 26 Jun 2010 at 5:05

• Changed state: Invalid

[GoogleCodeExporter]

I think you'll find that running 'make depend' every so often prevents most of 
the same problems fixed by clean installs. I guess since that's not part of the 
GNU toolchain, only BSD people do it, though.

Also, writing revision 9d31d13386b8 suffices to link a revision. It doesn't 
make a difference to me, but it is a bit easier to type.

Original comment by maglion...@gmail.com on 26 Jun 2010 at 5:12

[GoogleCodeExporter]

Oh, and I notice that the Google people finally decided to make their diff 
display colourblind friendly... Nice to know.

Original comment by maglion...@gmail.com on 26 Jun 2010 at 5:14