Re: objdump -k -q -o 0xC0100000 in 2.1.7

Michael L. Galbraith (
Thu, 7 Nov 1996 04:41:07 +0100 (MET)

On 5 Nov 1996, Bradley Ward Allen wrote:

> You're obviously far more knowledgeable than me as well as further
> ahead, but I had tiny kernels myself in 2.1 and wanted to say my
> solution in case it helps (I don't know if it is additional info or
> not). The fixers will undoubtedly get further with your information.
> I had binutils and libc 5.4.10 as well.
Since you mentioned DebiSlackHatSLS, I presume you're refering to me.
Hmmm - you obviously didn't see my _first_ post (laughter).

> Within linux-2.1.5/arch/i386/boot/compressed/Makefile, we have:
> if hash $(ENCAPS) 2> /dev/null; then \
> $(OBJDUMP) $(OBJDUMP_FLAGS) -o $(IMAGE_OFFSET) $(SYSTEM) > $$tmppiggy; \
> else \
> $(OBJCOPY) $(SYSTEM) $$tmppiggy; \
> fi; \
What I did, was to change line 28 in arch/i386/boot/Makefile
The tounge_in_cheek reference to mutation was simply because other folks
didn't have to do this - still wonder about that.

Thanks to everyone who replied. 2.1.7 is up, running well.

Low volume P.S.
I keep my system libs; libc, X-tree ... compiled with the latest/greatest
experimental kernel headers for consistency sake. Yesterday, I recompiled
libc-5.4.10 w. 2.1.7 headers. This broke everything net daemon wise (all).
Recompiling w. 2.0.24 works fine - even w. netscape - zero bus errors since
libc-5.3.12. ???