bug with GCSTAT

Patrick Tullmann tullmann at cs.utah.edu
Sat Sep 9 01:29:00 PDT 2000


I was playing with Kaffe's '-vmdebug GCSTAT' today and tripped across
a bug in it(?).  (Its very cool, btw, and I only just noticed it.)

I can reliably get a seg-fault at gc-incremental.c:205 (in a GCSTAT
macro).  

If I turn on GCWALK too, I get this right before the fault:
	marking @0x8ffef58: newly born ref-arrays

Here are the values of the relevant gc header fields:

info.funcs[idx] == 4
info.state[idx] == 8 // GC_STATE_NORMAL|GC_COLOUR_INUSE

I'm hoping someone (well, Jason or Godmar, actually :) will recognize
this problem immed.  Otherwise I can look into it some more.

-Pat

----- ----- ---- ---  ---  --   -    -      -         -               -
Pat Tullmann                                       tullmann at cs.utah.edu
		   ${HOME} is where the .emacs is.


More information about the kaffe mailing list