On Mon, 6 Aug 2001, Ralf Baechle wrote:
> Date: Mon, 6 Aug 2001 11:45:56 +0200
> From: Ralf Baechle <email@example.com>
> To: John Heil <firstname.lastname@example.org>
> Cc: email@example.com, firstname.lastname@example.org,
> Subject: Re: Qube2 gcc 2.7.2 compiler error (fwd)
> On Sun, Aug 05, 2001 at 12:59:23PM +0000, John Heil wrote:
> > Date: Sun, 5 Aug 2001 12:59:23 +0000 ( )
> > From: John Heil <email@example.com>
> > To: firstname.lastname@example.org, email@example.com,
> > firstname.lastname@example.org
> > Subject: Qube2 gcc 2.7.2 compiler error (fwd)
> > On the Qube2, gcc 2.7.2, option -s, to generate MIPS assembler
> > corresponding to the input C code, generates invalid assembler
> > by virtue of generating duplicate labels. The resultant
> > assembler will not assemble, of course, due to the duplicate
> > labels. The code (linux kernel's printk.c) compiles cleanly
> > from C to object code.
> > Q: How do I get valid assembler from gcc on Qube2 ?
> > (My ultimate goal here is to be able to get listings out
> > of gas.)
> gcc 2.7.2 creates a duplicate label for each function label. That is no
> problem as both always have the same value. But I assume you're talking
This is exactly the problem. The fact that the values are the same is
causing the assembler interface to GCC to fail. When gcc -S outputs a
.s assembler file, the GNU as assembler errors out on 'duplicate label'.
Further, using the gcc -Wa,... without using -S, causes the compile to
fail when the assembler is invoked and gives the same errors.
> about a different type of label? Can you send a piece of small piece of
> source code and the assembler code generated from it to demonstrate the
> Anyway, these days gcc 2.7.2 is so old these days it's not even funny. You
> really should upgrade.
I'm happy to upgrade...
What is the recommended level for kernel compiles and where can I find it.
I am required to work on 2.0.34C53_SK using Qube2 for my build platform
so I need compatibility. If I could cross compile on x86 that would be