Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 14 Apr 2014 22:34:09 +0200
From:      Ralf Wenk <iz-rpi03@hs-karlsruhe.de>
To:        Andreas Schwarz <freebsd.asc@strcmp.org>
Cc:        freebsd-arm@freebsd.org
Subject:   Re: screen(1) crashes plus weird output for screen -ls
Message-ID:  <E1WZnZx-004VDY-U1@smtp.hs-karlsruhe.de>
In-Reply-To: <443afb9423c.5a6681a6@mail.schwarzes.net>
References:  <freemail.20140405150804.40570.1@xmldata04>  <E1WTAl0-004Ooa-Au@smtp.hs-karlsruhe.de> <E1WYFYa-00CDeU-Gv@smtp.hs-karlsruhe.de> <18ADCE9C-C604-4067-B4D1-32EC07499231@bsdimp.com> <E1WYM5D-008xD0-Oy@smtp.hs-karlsruhe.de> <443afb9423c.5a6681a6@mail.schwarzes.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Andreas Schwarz <freebsd.asc=40strcmp.org> wrote:
> On 10.04.14, Ralf Wenk wrote:
>=20
> > I was not successful in compiling a gcc from the ports on armv6.
> > ONLY_FOR_ARCHS prohibits it. =5B...=5D
>=20
> The legacy standard freebsd gcc is still included in the source tree, y=
ou can use the=20
> WITH_GCC/WITH_GNUCXX options in scr.conf when building and installing w=
orld.

I had some minor trouble to get it going. Adding both options terminates
the build here with

-- gnu/lib/libsupc++__L ---
/root/rpi/head/gnu/lib/libsupc++/../../../contrib/libstdc++/libsupc++/eh_=
arm.cc
:44:1: error: functions that differ only in their return type cannot be=20
overloaded
__cxa_type_match(_Unwind_Exception* ue_header,
=5E

Just setting WITH_GCC got me finally a successful build.
Now back to the problem with screen on arm.


Compiling screen with gcc instead of clang using the standard optimizatio=
n
(-O) results in a correct working screen as =22screen -ls=22 prints

 There is a screen on:
         3164.pts-2.raspberry-pi (Detached)
 1 Socket in /tmp/screens/S-pi.

while a fresh with clang -O compiled screen prints

 There is a screen on:
         3164.pts-2.raspberry-pi (Detached)
 -1073746600 Socket
                   =D0=A0=E1 in =F0=F5=FF=BF8=E8.

Compiling screen with clang -O0 results in the same output as gcc.
I think this supports my idea of suspecting the clang optimizations.

The gcc used is

=24 gcc -v
Using built-in specs.
Target: armv6-undermydesk-freebsd
Configured with: FreeBSD/armv6 system compiler
Thread model: posix
gcc version 4.2.1 20070831 patched =5BFreeBSD=5D
=24


Ralf




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E1WZnZx-004VDY-U1>