Gary Benson gbenson at redhat.com
Tue May 26 08:12:02 PDT 2009

Hi all,

I've recently started seeing a crash on amd64 using the zero-assembler
port which I'm trying to debug.  A lot of the local variables in the
crash dump seem to have their top half replaced with 0xdeaddeaf, like

  0x7fab7ece2338: istate->_self_link    = 0x00007fab7ece22b8
  0x7fab7ece2340: frame_type            = INTERPRETER_FRAME
  0x7fab7ece2348: next_frame            = 0x00007fab7ece23f0

  0x7fab7ece2350: local[1]              = 0xdeaddeaf020c1370
  0x7fab7ece2358: local[0]              = 0xdeaddeaf7ece2380
  0x7fab7ece2360: istate->_thread       = 0x0000000001acf660

Is this something new?  It's not something I've seen before.  In what
circumstances would the VM write 0xdeaddeaf into stack slots?



More information about the hotspot-dev mailing list