A bunch of glibc header files were bogusly included.  We should not
depend on having a 32-bit glibc installed, since we don't use it.

Hey guys,

I noticed the appended ipappend options end up in the beginning of the
cmdline argv[] when mboot is entered. mboot then tries to parse the first
argument as the name of the kernel module and promptly fails cuz it can't
find "ip=.." or "bootif=...".

I'm attaching a patch which basically looks for any arguments in the argv[]
array and appends them to the kernel cmdline. The arguments have to be
back-to-back to be considered "appened options", so this should be safe
enough to not take out-of-place arguments and tack onto the kernel cmdline.

Last time i sent in some of the elf section loader patches, i think Peter
had trouble importing time. Hopefully it works now.

thanks,
Ram Yalamanchili

To be absolutely correct, we're supposed to use %* before an indirect
branch target, not just *.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

jmp in x86 can take a memory argument, and since gcc knows everything
that happens all the way up to the jump, allowing it there is safe.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

Fix register constraints of final jump to kernel entry.
When compiled with some GCC versions, mboot.c32 would clobber the kernel
load address and try to jump to 0x2badb002.

Signed-off-by: Tim Deegan <Tim.Deegan@cl.cam.ac.uk>

Load sections from the ELF section headers as well as the program headers,
for kernels which need symbol and debug info but don't ask for them in
the program headers.

Bump the version number to reflect this change.

Signed-off-by: Tim Deegan <Tim.Deegan@cl.cam.ac.uk>

Zero the inputs to BIOS calls, to fix two bugs in mboot.c32's E820 mmap code.

Signed-off-by: Tim Deegan <Tim.Deegan@cl.cam.ac.uk>