NVIZ Clear Scaled difference causes segfault

[marisn]

Start nviz -> Visualize -> Scaled Difference -> Clear -> KaBOOM!

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb70ddb50 (LWP 19340)]
0xb7f3900b in gs_get_att_typbuff (gs=0x0, desc=1, to_write=0) at gs.c:686
686         G_debug(5, "gs_get_att_typbuff(): id=%d desc=%d to_write=%d",
(gdb) bt
#0  0xb7f3900b in gs_get_att_typbuff (gs=0x0, desc=1, to_write=0) at gs.c:686
#1  0xb7f3af03 in gsdiff_set_SDref (gsref=0x0) at gsdiff.c:67
#2  0xb7f33b1e in GS_unset_SDsurf () at GS2.c:1016
#3  0x0805b855 in Nunset_SDsurf_cmd (data=0x8071ce0, interp=0x8077d50, argc=1, argv=0xbff9b4cc) at misc.c:58
#4  0xb7c55b24 in TclInvokeStringCommand () from /usr/lib/libtcl8.4.so
#5  0xb7c57566 in TclEvalObjvInternal () from /usr/lib/libtcl8.4.so
#6  0xb7c81578 in ?? () from /usr/lib/libtcl8.4.so
#7  0x08077d50 in ?? ()
#8  0x00000001 in ?? ()
#9  0x08079b58 in ?? ()
#10 0x00000000 in ?? ()

[cmbarton]

I have no problem on the Mac with develbranch_6. A Linux or perhaps particular flavor of Linux issue?

[cmbarton]

Found it.

This seems to be a bug in ./nviz/src/misc.c somewhere in lines 48-61 (or one of the functions called here). I don't know C, and haven't been able to debug it.

[marisn]

Some G_debug calls in lib/ogsf where not checking for pointer existence before it's usage. Fixed in r38672. Candidate for backporting to 6.4.0.

[marisn]

Fix backported to releasebranch (r38737) and trunk (r38735) by martinl.