From mboxrd@z Thu Jan 1 00:00:00 1970 X-Msuck: nntp://news.gmane.org/gmane.linux.lib.musl.general/5115 Path: news.gmane.org!not-for-mail From: Stephen Thomas Newsgroups: gmane.linux.lib.musl.general Subject: RE: Broken GCC versions: 4.8.2 and 4.9.0 Date: Mon, 12 May 2014 02:23:19 +0100 Message-ID: References: <20140511010503.GA6502@brightrain.aerifal.cx>,<20140511181020.0a8b66f1@free-electrons.com>,<20140511161943.GR26358@brightrain.aerifal.cx>,,<20140511190320.GV26358@brightrain.aerifal.cx>,,<20140511212030.GY26358@brightrain.aerifal.cx> Reply-To: musl@lists.openwall.com NNTP-Posting-Host: plane.gmane.org Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="_a5ee53a8-37b1-4c3b-b25c-b0941420102b_" X-Trace: ger.gmane.org 1399857823 27765 80.91.229.3 (12 May 2014 01:23:43 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Mon, 12 May 2014 01:23:43 +0000 (UTC) Cc: Thomas Petazzoni To: "musl@lists.openwall.com" , James Cloos Original-X-From: musl-return-5120-gllmg-musl=m.gmane.org@lists.openwall.com Mon May 12 03:23:36 2014 Return-path: Envelope-to: gllmg-musl@plane.gmane.org Original-Received: from mother.openwall.net ([195.42.179.200]) by plane.gmane.org with smtp (Exim 4.69) (envelope-from ) id 1Wjexq-0006NY-3p for gllmg-musl@plane.gmane.org; Mon, 12 May 2014 03:23:34 +0200 Original-Received: (qmail 22404 invoked by uid 550); 12 May 2014 01:23:31 -0000 Mailing-List: contact musl-help@lists.openwall.com; run by ezmlm Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: Original-Received: (qmail 22393 invoked from network); 12 May 2014 01:23:31 -0000 X-TMN: [Y6R8QvislWp5evTDtYgmtgMyPgvftfJe] X-Originating-Email: [scjthm@live.com] Importance: Normal In-Reply-To: <20140511212030.GY26358@brightrain.aerifal.cx> X-OriginalArrivalTime: 12 May 2014 01:23:19.0285 (UTC) FILETIME=[C1DE5A50:01CF6D80] Xref: news.gmane.org gmane.linux.lib.musl.general:5115 Archived-At: --_a5ee53a8-37b1-4c3b-b25c-b0941420102b_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable > On Sun=2C May 11=2C 2014 at 04:08:37PM -0400=2C James Cloos wrote: > > >>>>> "RF" =3D=3D Rich Felker writes: > >=20 > > RF> I've tested it on gcc.godbolt.org and others have tested with local > > RF> gcc 4.8.2 and 4.9.0=2C probably distro-provided (I didn't ask). > >=20 > > I just tried debian-sid's 4.9. That does show the bug. As does their > > gcc-snapshot version (a 4.10 pre). > >=20 > > Perhaps the distro(s) which have the bug with 4.8.2 backported the comm= it > > where the bug first occurs? >=20 > Sounds plausible. Maybe that will help find the offending commit. >=20 > > RF> I wonder if the broken GCC is using isl/cloog > >=20 > > I use the graphite optimizer on my Gentoo box (which generated the > > assembly I quoted). The graphite code is only used when one or more > > of -floop-interchange -floop-strip-mine -floop-block are specified. > >=20 > > Testing with those options doesn't change the results. >=20 > OK=2C ruled that out then. FWIW=2C I *think* noticed that this problem only appeared in buildroot=2C w= ithout enabling c++ in the languages. Actually=2C when I decided to evaluat= e the C++ changes and therefore enabled C++ in buildroot (I am not sure wha= t it does in addition to --enable-languages=3Dc=2Cc++). I cannot be certain= and I have not had the time to retest this.=20 Cheers Thomo = --_a5ee53a8-37b1-4c3b-b25c-b0941420102b_ Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
>=3B On Sun=2C May 11=2C 2014 = at 04:08:37PM -0400=2C James Cloos wrote:
>=3B >=3B >=3B>=3B>= =3B>=3B>=3B "RF" =3D=3D Rich Felker <=3Bdalias@libc.org>=3B writes:=
>=3B >=3B
>=3B >=3B RF>=3B I've tested it on gcc.godbolt.= org and others have tested with local
>=3B >=3B RF>=3B gcc 4.8.2 a= nd 4.9.0=2C probably distro-provided (I didn't ask).
>=3B >=3B
&= gt=3B >=3B I just tried debian-sid's 4.9. That does show the bug. As do= es their
>=3B >=3B gcc-snapshot version (a 4.10 pre).
>=3B >= =3B
>=3B >=3B Perhaps the distro(s) which have the bug with 4.8.2 b= ackported the commit
>=3B >=3B where the bug first occurs?
>=3B=
>=3B Sounds plausible. Maybe that will help find the offending commi= t.
>=3B
>=3B >=3B RF>=3B I wonder if the broken GCC is using= isl/cloog
>=3B >=3B
>=3B >=3B I use the graphite optimizer = on my Gentoo box (which generated the
>=3B >=3B assembly I quoted). = The graphite code is only used when one or more
>=3B >=3B of -floop= -interchange -floop-strip-mine -floop-block are specified.
>=3B >=3B=
>=3B >=3B Testing with those options doesn't change the results.>=3B
>=3B OK=2C ruled that out then.

FWIW=2C I *think= * noticed that this problem only appeared in buildroot=2C without enabling = c++ in the languages. Actually=2C when I decided to evaluate the C++ change= s and therefore enabled C++ in buildroot (I am not sure what it does in add= ition to --enable-languages=3Dc=2Cc++). I cannot be certain and I have not = had the time to retest this. =3B

Cheers
<= div>
Thomo
= --_a5ee53a8-37b1-4c3b-b25c-b0941420102b_--