Thanks for pointing this out; it does look like a potential problem.
-Original Message-
From: Ding-Kai Chen [mailto:[email protected]]
Sent: Friday, September 24, 2010 10:33 AM
To: open64-devel
Subject: [Open64-devel] Potential stale pointer in cgemit.cxx
I found a potential issue
I found a potential issue with vstring usage in r_assemble_list() in
cgemit.cxx.
The opnd[] and result[] are storing addresses off a vstring 'buf' while
it is being appended. If one of the 'append' triggers a realloc and new
block has to be alloc'ed, the old pointers in opnd[] and result[] bec
