From: Yukihiro Matsumoto <matz@...>
Date: 2006-02-16T01:19:55+09:00
Subject: Re: Ruby 1.8 CVS miniruby core dumps

Hi,

In message "Re: Ruby 1.8 CVS miniruby core dumps"
    on Wed, 15 Feb 2006 18:27:08 +0900, ville.mattila@stonesoft.com writes:

|>|current cvs dumps. The dumps appeared with changes on date
|>|cvs update -rruby_1_8 -D'Tue Feb 14 13:07:02 2006 UTC'.
|>|This happens both 32 and 64 bit versions.
|>
|>Couldn't reproduce your problem on my machine.  Probably Solaris
|>specific problem.  Could you get the following value in the function
|>backtrace in the debugger session for us?
|>
|>*frame
|>*frame->prev
|
|I didn't found frame, did you mean ruby_frame?

The compiler might optimized out the local variable.  But it's OK.

|(dbx) print *ruby_frame->prev
|*ruby_frame->prev = {
|    self       = 6422320U
|    argc       = 0
|    last_func  = 0
|    orig_func  = 18446741324928380832U
|    last_class = 0
|    prev       = 0xfffffd7fffdff350
|    tmp        = 0xfffffd7fffdfef58
|    node       = 0x6008b0
|    iter       = 0
|    flags      = 0
|    uniq       = 196U
|}

Thank you.  This helped me a lot.  An uninitialized member is
accessed.  I just committed the fix.  Could you try the latest?

							matz.