On Fri, 24 Sep 1999, Philip wrote:
> I suggested to the
> last person to report it that they could try glibc 2.1.1 instead and I don't
> think I heard any more, so maybe that solved the problem for them.
And from an older E-Mail
>A similar problem has been reported by some people on other platforms, but I
>don't think anybody has worked out the cause (or a cure) yet. The best guess
>so far seems to be that it might be due to a buggy version of awk or some
>other tool. It seems to have occurred suddenly in glibc 2.1.2, so you could
>try going back to 2.1.1 and see if that works any better. Alternatively I
>think you will have to investigate the Makefiles and try to figure out why the
>*.map files are not being correctly built.
Ok! I thought I should give a try to v2.1.1 and resolve the v2.1.2 problem
when I'm advanced enough in my make, bash and gawk studying (I'm _very_ junior
with all of these)
So on my intel sysem with
binutils 2.9.5.0.11, built and installed,
linux 2.2.10 "make dep" for ARM and the resulting includes appropriately copied,
gcc-2.95.1 with 4 patches applied, built and installed,
I tried to ./configure glibc-2.1.1.
With CC=arm-linux-gcc
Configuring results at some point in :
>checking version of gcc... pgcc-2.91.66, ok
>checking for make... (cached) make
>checking version of make... 3.77, ok
>checking for msgfmt... (cached) msgfmt
>checking version of msgfmt... 0.10.35, ok
>checking for makeinfo... no
>configure: warning:
>*** An auxiliary program is missing or too old;
>*** some features will be disabled.
>*** Check the INSTALL file for required versions.
>checking whether the C compiler (gcc ) works... (cached) yes
>checking whether the C compiler (gcc ) is a cross-compiler... (cached) no
>checking whether we are using GNU C... (cached) yes
Which is my native compiler not my 2.95.1 compiler built for ARM.
So of course doing:
# make
results very quickly in:
>make[1]: Entering directory `/usr/src/glibc/glibc-2.1.1/csu'
>gcc ../sysdeps/unix/sysv/linux/arm/sysdep.S -c -I../include -I. ...
>/tmp/ccmAFUZY.s: Assembler messages:
>/tmp/ccmAFUZY.s:1053: Error: no such 386 instruction: `rsb'
>/tmp/ccmAFUZY.s:1289: Error: no such 386 instruction: `ldr'
>/tmp/ccmAFUZY.s:1290: Error: invalid char '[' begining second operand '[r1]'
>/tmp/ccmAFUZY.s:1292: Error: no such 386 instruction: `stmdb'
>/tmp/ccmAFUZY.s:1294: Error: no such 386 instruction: `bl'
>/tmp/ccmAFUZY.s:1295: Error: no such 386 instruction: `ldmia'
>/tmp/ccmAFUZY.s:1296: Error: invalid char '[' begining second operand '[r0]'
>/tmp/ccmAFUZY.s:1300: Error: no such 386 instruction: `mvn'
>/tmp/ccmAFUZY.s:1301: Error: no opcode suffix given and no register operands;
>can't size instruction make[1]: *** [sysdep.o] Error 1
>make[1]: Leaving directory `/usr/src/glibc/glibc-2.1.1/csu'
>make: *** [csu/subdir_lib] Error 2
The very first time I tried this, ./configure told me, something like
>checking version of gcc... gcc-2.95.1, bad!
and eventually failed to complete...
so I tried ./configure with --build=i586-linux instead of --build=i686-pc-linux-gnu
and from that time and since, it chooses the native pgcc.
Is there a quick fix for this?
If not, do I really want to try to solve these issues on glibc-2.1.1, or would
it be preferable that I invest on v2.1.2 ?
--
Fran�ois Desloges
[EMAIL PROTECTED]
unsubscribe: body of `unsubscribe linux-arm' to [EMAIL PROTECTED]
