segfault on big endian machine

[Link4Electronics]

Hello,

Running Debian Linux 6.16.12 with 64kb pagesize PowerPC 64-bit on a Power Mac G5 with a Radeon R5 230 OpenGL 3.2, compiled lastest master git branch, when trying to launch with -halflife argument, it segfault. Built a debug, it stops at line 1475 in file engine/gl/gl_font.c. iqmtool and imgtool complains about ShortSwap LongSwap I64Swap inside engine/client/image.c around line 2704

Best regards,
Link

[erysdren]

looks like it's failing on an alloca() call, that's interesting. i can patch the code to use some other memory allocation instead, which may fix your issue. what do you mean about imgtool complaining about those functions? is it a linker failure? let me know. thanks!

[Link4Electronics]

Hello, sorry for taking a while to reply.

Yes iqmtool and imgtool fails when trying to link and gives undefined reference to that type of variables. qtv is failing to compile, on both little and big endian machines, on line 302 fteqtv/qtv.h cannot use keyword 'false' as enumeration constant.

Best regards,
Link.

[erysdren]

as for imgtool, it's probably not including the proper file that contains the *Swap functions. FTEQW is rarely tested on big endian machines.

as for qtv, i've fixed that in this patch: #381

[Link4Electronics]

Had an interesting behavior, using package_valve.pk4 from fnvalve instead of package_valve.pk3 from github, had this:

Menu sound works, if try to start a new game:

Mod_LoadBrushModel: maps/c0a0.bsp appears truncated Host_EndGame: Mod_NumForName: maps/c0a0.bsp not found or couldn't load

and then hangs. Tried to replace glTexImage2D GL_UNSIGNED_BYTE to GL_UNSIGNED_INT_8_8_8_8_REV everywhere to see if it would fix the inverted RGB colours from menu but nothing.

Best regards,
Link.