[Bug target/61656] Undefined behavior in classify_argument

["ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61656

--- Comment #4 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Jakub Jelinek from comment #3)
> So, are you ok with the #c0 pseudo patch?  I guess I can bootstrap/regtest
> it today.

If we are narrowing the load from movq to movl for a SImode value, then the
patch is OK.
>From gcc-bugs-return-456067-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Jul 10 15:50:39 2014
Return-Path: <gcc-bugs-return-456067-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 8106 invoked by alias); 10 Jul 2014 15:50:38 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 7794 invoked by uid 48); 10 Jul 2014 15:50:30 -0000
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/61656] Undefined behavior in classify_argument
Date: Thu, 10 Jul 2014 15:50:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 4.10.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-61656-4-yLNTYhDXDg@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-61656-4@http.gcc.gnu.org/bugzilla/>
References: <bug-61656-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-07/txt/msg00658.txt.bz2
Content-length: 803

https://gcc.gnu.org/bugzilla/show_bug.cgi?ida656

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Depends on what testcase.  In pr42025-2.c, it clearly uses movl %edi, %eax,
with and without the patch.  If I write:
typedef struct
{
  int p;
} Ptr;

struct A
{
  long int i;
  union
  {
    Ptr p;
    int q;
  } u;
};

extern Ptr get_stuff (void);
extern void use_stuff (int);

__attribute__((noinline, noclone))
static int foo(struct A p, int q)
{
  int ret = p.u.q;
  if (p.i)
    p.u.p = get_stuff ();
  else
    p.u.q = q;

  use_stuff (p.u.q);
  return ret;
}

void bar(struct A *p, int q)
{
  foo(*p, q);
}

then it uses movq %rsi, ... and the patch doesn't change anything on that.
If you want to use movl %esi, ... in foo in this testcase, more changes are
needed of course.