On Fri, Sep 02, 2000 at 03:41:33PM +0200, Werner Almesberger wrote:
> Alan Cox wrote:
> > My goal would be to ensure that the bootloader didnt need to be modified.
>
> Yes, I was commenting on Andi's proposal. I think it's very important to
> avoid the need for boot loader modifications - there are simply too many
> of them nowadays.
>
> > As to the tool argument - looking for a 16byte magic key sequence and printing
> > the rest is not hard.
>
> Oh, I agree. What may be quite hard is finding the file. Already with
> LOADLIN, it may get hairy, and who knows what the future will bring ?
embedded system whose memory you initialize over the PCI bus (so your boot
loader runs in host userspace). (I'm doing this for bochs, a friend just
wrote the code for actual ppc hardware).
> Also, if you overwrite (update) your kernel image and/or edit your
> /etc/lilo.conf, you'll get the wrong answer, unless you add some magic
> number (MD5 hash or such) that can be queried for the running kernel.
isn't that what the version string (/proc/version at runtime, start_sys
in the bzImage) is for ?
> But then you're probably already beyond the point of having a
> fully-featured /proc/config.gz, complexity-wise (which, after all,
> would be completely architecture-independent).
Most architectures can boot ELF images -- defining section names for
.config.gz and the version string in the ELF file can be done in an
architecture-independent fashion.
Philipp Rumpf
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Sep 07 2000 - 21:00:13 EST