Zsh Mailing List Archive
Messages sorted by:
Reverse Date,
Date,
Thread,
Author
compiling zsh on an i686 for an i586 target
- X-seq: zsh-workers 17840
- From: tmk@xxxxxxxxxxxx
- To: zsh-workers@xxxxxxxxxx
- Subject: compiling zsh on an i686 for an i586 target
- Date: Fri, 18 Oct 2002 18:31:34 -0400
- Mailing-list: contact zsh-workers-help@xxxxxxxxxx; run by ezmlm
- Organization: lordzork industries
- Reply-to: tmk@xxxxxxxxxxxx
Greetings,
I have two i586 boxes on my network acting as dns/samba/dhcp/qmail
servers. One of these is a k6-2, the other is a 167 mhz Pentium. I need
to be able to compile zsh on my main i686 box (AMD Athlon) and have it
run on the i586s. So far the resulting zsh binary crashes immediately
upon execution as a result of an "illegal instruction".
All the boxes concerned are running glibc 2.3 under my own home-grown
brand of linux (2.4) with all of the standard GNU tools. I'm using the
latest zsh version, 4.0.6.
Zsh is configured as follows:
./configure --prefix=/usr --bindir=/bin --enable-etcdir=/etc/zsh
--enable-maildir-support --build=i586-pc-linux-gnu
--host=i586-pc-linux-gnu --target=i586-pc-linux-gnu
After configure is finished running, I see among other things:
host operating system : i586-pc-linux-gnu
compiler : gcc
preprocessor flags : -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64
executable compiler flags : -Wall -Wmissing-prototypes -O2
module compiler flags : -Wall -Wmissing-prototypes -O2 -fPIC
executable linker flags : -s -rdynamic
module linker flags : -s -shared
library flags : -ldl -lnsl -lcurses -lm -lc
The gcc version, by the way, is 3.2. I've also tried using 2.95.3 with
the same result. I'm using version 2.13.90.0.4 20020814 of binutils
(which was required to build glibc 2.3).
It seems rather odd to me that this isn't working, because I've
successfully compiled a lot of other i586 binaries on the i686,
including glibc. It's usually enough to just pass the --host= and
--build= options to configure and make sure the cflags are ok.
I did try to make an i586 cross-compiling environment, but gcc wouldn't
build this way. In any case, wouldn't this be unnescessary since the
basic architecture is the same?
If anyone has any ideas, I would very much appreciate hearing them. I
hope I've sent this to the right list; it seemed to me that it had to do
more with the guts of the shell than the driving of it.
t.
--
4.0k -rw-rw-r-- 1 zork zork 17 Apr 4 17:38 .signature
Messages sorted by:
Reverse Date,
Date,
Thread,
Author