Re: RFC/RFA: patch for PR 22156: improve SRA for bit-fields

[Alexandre Oliva <aoliva at redhat dot com>]

I'm sorry, I'd mistaken your report as a miscompilation, that I hoped
would be fixed with my revised patch.  Now I see it's just an
optimization issue.

On Apr 22, 2007, Eric Botcazou <ebotcazou@adacore.com> wrote:

> I find it a little strange that your patch doesn't contain a single instance 
> of DECL_BIT_FIELD:

That's because it's not focused specifically on bit-fields.

It tries to represent in a single gimple register all adjacent fields
of a struct that are not accessed individually.

  lib_opts.0 = VIEW_CONVERT_EXPR<struct gnat__strings__string_access>(D.746);
  lib_opts = lib_opts.0;

So you see, lib_opts, whose members presumably get accessed directly,
gets its members represented as separate variables.  lib_opts.0, that
is only accessed as a block, gets represented as a single variable.


Given the explanation above, does this request still hold?

> We definitely don't want to handle P_ARRAY and P_BOUNDS as a block,

-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
FSF Latin America Board Member         http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}