From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-408645-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 25565 invoked by alias); 29 Sep 2015 21:24:34 -0000
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
Received: (qmail 25529 invoked by uid 89); 29 Sep 2015 21:24:30 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,SPF_PASS,T_RP_MATCHES_RCVD autolearn=ham version=3.3.2
X-HELO: resqmta-po-05v.sys.comcast.net
Received: from resqmta-po-05v.sys.comcast.net (HELO resqmta-po-05v.sys.comcast.net) (96.114.154.164) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with (AES128-SHA encrypted) ESMTPS; Tue, 29 Sep 2015 21:24:29 +0000
Received: from resomta-po-16v.sys.comcast.net ([96.114.154.240])	by resqmta-po-05v.sys.comcast.net with comcast	id P9QP1r0075BUCh4019QUGH; Tue, 29 Sep 2015 21:24:28 +0000
Received: from [IPv6:2001:558:6045:a4:40c6:7199:cd03:b02d] ([IPv6:2001:558:6045:a4:40c6:7199:cd03:b02d])	by resomta-po-16v.sys.comcast.net with comcast	id P9QS1r00J2ztT3H019QTqX; Tue, 29 Sep 2015 21:24:27 +0000
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.3 \(1878.6\))
Subject: Re: [PATCH] x86 interrupt attribute
From: Mike Stump <mikestump@comcast.net>
In-Reply-To: <CAMe9rOoeorsg06XxtNr1-iNX_m-Fe8gtsW2uXs_M2+EOw+gLqg@mail.gmail.com>
Date: Tue, 29 Sep 2015 22:10:00 -0000
Cc: Yulia Koval <vaalfreja@gmail.com>, GCC Patches <gcc-patches@gcc.gnu.org>, Uros Bizjak <ubizjak@gmail.com>
Content-Transfer-Encoding: quoted-printable
Message-Id: <F096FB33-4A00-4D7B-8C9B-3FE8ACE7FEF9@comcast.net>
References: <CAE0nxxi-T8da1=dXkHem2m6d_2RramotAU3HY=UqFkmT0kW=RQ@mail.gmail.com> <2AB9494F-91EC-4CF6-945B-0CB03A583513@comcast.net> <CAMe9rOpMB9fnj_ZS7h122BuODsf9TZNOUvWVzrYvLN8VxGiFkg@mail.gmail.com> <CAMe9rOoeorsg06XxtNr1-iNX_m-Fe8gtsW2uXs_M2+EOw+gLqg@mail.gmail.com>
To: "H.J. Lu" <hjl.tools@gmail.com>
X-IsSubscribed: yes
X-SW-Source: 2015-09/txt/msg02270.txt.bz2

On Sep 29, 2015, at 1:59 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> commit f3a6675a8d69d810d2cad0c090a762094a0a8622
> Author: H.J. Lu <hjl.tools@gmail.com>
> Date:   Tue Sep 29 13:47:18 2015 -0700
>=20
>    Define EPILOGUE_USES in i386 so that all preserved registers are used
>    by the epilogue of interrupt handler.  Don't explicitly mark BP and SP
>    registers as used since they are always used in epilogue.
>=20
> Please take a look.

Oh, too bad you didn=92t copy it here.  The easiest thing to blow is the ad=
dition of reload_completed && on the condition:

  /* An interrupt handler must preserve some registers that are=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20
     ordinarily call-clobbered.  */
  if (reload_completed
      && myarch_interrupt_func (current_function_decl)
      && save_reg_p (regno))
    return true;

without it, the optimizer will blow chunks all over the place and code-gen =
will not be very good, if it doesn=92t.  I=92d love this to be shared acros=
s all ports, it it is cryptic and usually test cases are not elaborate enou=
gh to find the problem.  When we ported a large library to our system that =
made extensive uses of complex interrupt routines, the compiler blew chunks=
.  With lessor code, we never even noticed a problem.