From mint-bounce@lists.fishpool.fi Wed Aug 18 10:03:30 2010 Message-ID: Content-Type: multipart/alternative; boundary="_dda788c7-3d6a-4a36-8b30-b89519623cec_" X-Originating-IP: [193.62.204.18] From: Matt Waller To: "Miro Kropacek (http://mikro.umpc.sk/)" CC: mint Subject: Re: [MiNT] automake & configure (m4-1.4.1) Date: Wed, 18 Aug 2010 15:00:17 +0100 Importance: Normal In-Reply-To: References: <0001c724.01b432629a82@smtp.freeola.net>,<0012797b.01b8d262c965@smtp.freeola.net>,,,, MIME-Version: 1.0 X-OriginalArrivalTime: 18 Aug 2010 14:00:17.0335 (UTC) FILETIME=[B0198C70:01CB3EDD] X-ecartis-version: Ecartis v1.0.0 Sender: mint-bounce@lists.fishpool.fi Errors-to: mint-bounce@lists.fishpool.fi X-original-sender: mw333@hotmail.com Precedence: bulk List-help: List-unsubscribe: List-Id: X-List-ID: List-subscribe: List-owner: List-post: --_dda788c7-3d6a-4a36-8b30-b89519623cec_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable ok.. a minimal test script:In a fresh m4-1.4.1 download:tar -zxvf m4-1.4.1.= tar.gz=3B cd m4-1.4.1.tar.gz the following command gives a bus error after reading line 86 of configure.= in:/usr/bin/m4 \--nesting-limit=3D9 \--debug=3Daflq \--trace=3DAC_SUBST_TRA= CE \--reload-state=3D/usr/local/share/autoconf/autoconf/autoconf.m4f \confi= gure.in ------ > Date: Wed=2C 18 Aug 2010 14:00:40 +0200 > Subject: Re: [MiNT] automake & configure (m4-1.4.1) > From: miro.kropacek@gmail.com > To: mw333@hotmail.com > CC: mint@lists.fishpool.fi >=20 > > Back to my original problem.. The m4-1.4.1 that I built likes to crash = 'Bus > > Error' when used by my system's autom4te=2C HOWEVER the original system > > m4-1.4.1 doesn't have this problem. > There could be number of reasons -- mintlib has changed=2C new gcc > produces different (more stack hungry for example) code=2C ... or even > it might be a hidden bug in mintlib which popped out accidentally. >=20 > > An example command that crashes it is: > > autoconf --verbose (which runs autom4te which runs m4) > And when you run m4 standalone? If standalone works=2C try to catch m4 > arguments used in autoconf (in 'ps' output for example). >=20 > > - What's an easy way to isolate mintlib daily from the existing system > > libraries? gcc by default wants to look in /usr/local/include and > > /usr/include & /usr/local/lib > make install prefix=3D/usr/local in mintlib root. but you need to build > it by yourself of course. Or=2C you can copy stuff from tar.bz2 archive > not into /usr but into /usr/local (file tree is the same). But be sure > gcc really uses it. >=20 > > - Apart from gcc and ld=2C are there other binaries that I should check= ? Is > > there a better ld? > It's most probably some stack / overflow problem. gcc=2C ld etc should be= fine. >=20 > > - What are your favourite gcc options? I'm using CFLAGS=3D"-m68020-60 -= Wall" > As do I :) When not sure what's linked in=2C I use Vincent's -Wl=2C-t opt= ion=2C too. >=20 > --=20 > MiKRO / Mystic Bytes > http://mikro.atari.org >=20 >=20 = --_dda788c7-3d6a-4a36-8b30-b89519623cec_ Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable ok.. a minimal test script:

In a fresh m4-1.4.1 download:

tar= -zxvf m4-1.4.1.tar.gz=3B =3Bcd m4-1.4.1.tar.gz

the following command gives a bus error after reading line 86 of configur= e.in:

/usr/bin/m4 \

--nesting-limit=3D9 \

--debug=3Daflq \

--trace=3DAC_SUBST_TRACE \

--reload-= state=3D/usr/local/share/autoconf/autoconf/autoconf.m4f \

configu= re.in

------

>= =3B Date: Wed=2C 18 Aug 2010 14:00:40 +0200
>=3B Subject: Re: [MiNT] a= utomake &=3B configure (m4-1.4.1)
>=3B From: miro.kropacek@gmail.co= m
>=3B To: mw333@hotmail.com
>=3B CC: mint@lists.fishpool.fi
&= gt=3B
>=3B >=3B Back to my original problem.. The m4-1.4.1 that I b= uilt likes to crash 'Bus
>=3B >=3B Error' when used by my system's a= utom4te=2C HOWEVER the original system
>=3B >=3B m4-1.4.1 doesn't ha= ve this problem.
>=3B There could be number of reasons -- mintlib has = changed=2C new gcc
>=3B produces different (more stack hungry for exam= ple) code=2C ... or even
>=3B it might be a hidden bug in mintlib whic= h popped out accidentally.
>=3B
>=3B >=3B An example command t= hat crashes it is:
>=3B >=3B autoconf --verbose (which runs autom4te= which runs m4)
>=3B And when you run m4 standalone? If standalone wor= ks=2C try to catch m4
>=3B arguments used in autoconf (in 'ps' output = for example).
>=3B
>=3B >=3B - What's an easy way to isolate m= intlib daily from the existing system
>=3B >=3B libraries? gcc by de= fault wants to look in /usr/local/include and
>=3B >=3B /usr/include= &=3B /usr/local/lib
>=3B make install prefix=3D/usr/local in mintl= ib root. but you need to build
>=3B it by yourself of course. Or=2C yo= u can copy stuff from tar.bz2 archive
>=3B not into /usr but into /usr= /local (file tree is the same). But be sure
>=3B gcc really uses it.>=3B
>=3B >=3B - Apart from gcc and ld=2C are there other binar= ies that I should check ? Is
>=3B >=3B there a better ld?
>=3B = It's most probably some stack / overflow problem. gcc=2C ld etc should be f= ine.
>=3B
>=3B >=3B - What are your favourite gcc options? I'm= using CFLAGS=3D"-m68020-60 -Wall"
>=3B As do I :) When not sure what'= s linked in=2C I use Vincent's -Wl=2C-t option=2C too.
>=3B
>=3B= --
>=3B MiKRO / Mystic Bytes
>=3B http://mikro.atari.org
>= =3B
>=3B

= --_dda788c7-3d6a-4a36-8b30-b89519623cec_--