ARM code generation bug with -O2...

[Jeffrey A Law]

  In message < E137QAG-0000ys-00@kings-cross.london.uk.eu.org >you write:
  > >Should the compiler be generating this assembler construct?
  > 
  > No.  The generated code you quoted derives no advantage from the post-inc 
  > and this construct is fairly clearly bogus.  Do you have a testcase?
And where does the testcase come from ;-)  I'd like to see it added to
our regression testsuite.
jeff