On Thu, Feb 16, 2017, at 07:11 AM, Peter Stephenson wrote:
> Yes, it doesn't fix the problem, and it looks like it's a bit more
> subtle than simply a wrong length.
> 
> I've managed to get valgrind reporting errors but with and without
> --enable-zsh-mem.  Here's one without: --enable-zsh-debug is on
> but that's all.
> 
> I'm afraid that's all I'm going to have time to do.
> 
> pws
> 
> % zmodload zsh/db/gdbm
> % ztie -d db/gdbm -f db.gdbm dbase
> % dbase[testkey]=somewhat
> ==25034== Invalid write of size 1
> ==25034==    at 0x80E0B78: metafy (utils.c:4652)

I've used HEAP_ALLOC instead of HEAP_DUP in argument for metafy. Fixing
this resolves valgrind report. Important catch.. Hopefully it will
resolve enable-zsh-mem problem (I cannot reproduce that). Sending as
complete patch plus db_gdbm.c file.

-- 
  Sebastian Gniazdowski
  psprint2@fastmail.com