MOL/VNC crashes

Brian Ruth mol-general@lists.maconlinux.org
Tue, 26 Aug 2003 17:13:43 +0100


Hi,

I've just been trying out using MOL with VNC as an alternative to using the
X video driver. 

Whilst it seems to work mostly okay, there does appear to be a problem with
translation between pixel depths - changing the pixel depth on the client
appears to result in a major crash in the VNC thread in MOL.

I have 16bpp configured in molrc.video; however, OSX is configured for 8bpp.
MOL starts up, and displays on the VNC display (which is configured for
16bpp). However, as soon as the MOL video driver starts (marking the point
at which the OSX 'starting up' window appears, the VNC video driver in MOL
tries to call 'init_translationtable 8 16'. At this point, a big crash 
occurs:

+ Video Driver v1.12
vncvideo: vopen called
vncvideo: init_translationtable 8 16
vncvideo: init_translationtable 8 16
vncvideo: init_translationtable 8 16
***** SIGNAL 11 [Segmentation fault] in thread VNC-thread *****
   si_signo = 11, si_errno 0, si_code 00030001, si_addr 0xd8584a55
   Last RVEC: 0x0 (0), last OSI: 61, mac_nip 900049E0
   NIP translate + 0x5c
***** Backtrace *****
   7f3ff144: backtrace + 0x3c
   7f3ff164: signal_handler + 0x184
   7f3ff194: 0x0fda58d8
   7f3ff1c4: 0x7f3ff498
   7f3ff580: 0x0fda35b4
   7f3ff5a0: sendHextiles8 + 0x1d0
   7f3ff730: send_rect_hextile + 0xb8
   7f3ff760: send_rect + 0x64
   7f3ff7b0: send_update + 0x160
   7f3ff880: read_exact + 0xf8
   7f3ff8b0: vnc_thread + 0x26c
   7f3ffae0: thread_entry + 0x7c
   7f3ffb20: 0x0fda2428
   7f3ffbf0: 0x0fc8be68
Exiting due to a crashed thread

..and that's that! It only happens if there's a mismatch between the colour
depth on the Mac OS (both 9 and X) and that on the VNC client - presumably,
there's some issue in the 'init_translationtable' function. I haven't looked
at this, and I doubt my knowledge of VNC would be enough to fix it anyway,
but has anyone a quick fix for it?

Brian.