Re: Can't compile an exact copy of kernel image.
Michael Heldebrant gravada:
> On Thu, 2001-10-04 at 19:36, Ricardo Gazoni wrote:
> > Yes, I did. In fact, step 1 loaded /boot/config-2.2.19pre17 and saved it in
> > /usr/src/kernel-source-2.2.19pre17/.config. To be precise, doing this via make
> > xconfig only adds comment lines to the .config file. I checked it out carefully....
> >
> > Thanks anyway...
> >
> > Osamu Aoki gravada:
> >
> > > Hi,
> > >
> > > Did you copy configuration from /boot/config-....
> > >
> > > Otheise config change makes kernel size different.
> > >
> > > Cheers:-)
> > >
> > > On Thu, Oct 04, 2001 at 06:08:50PM -0300, Ricardo Gazoni wrote:
> > > > I'm a newbie on kernel compiling, and would like to recompile the kernel
> > > > on my
> > > > machine to remove undesired drives. So, I got
> > > > kernel-source-2.2.19pre17.deb,
> > > > kernel-package.deb and gcc272_2.7.2.3-15 and followed the instructions:
> > > >
> > > > 0 - $ make xconfig
> > > > 1 - in xconfig, loaded /boot/config-2.2.19pre17; saved and exited
> > > > 2 - $ make-kpkg clean
> > > > 3 - changed the Makefile, to make EXTRAVERSION = foo17 (to avoid module
> > > > problems)
> > > > 4 - $ fakeroot make-kpkg --revision=foo.1.0 kernel_image
> > > > 5 - (as root) # dpkg -i kernel-image-2.2.19foo17_foo.1.0_i386.deb
> > > >
> > > > Everything worked just fine until now. But the old kernel image
> > > > (vmlinuz-2.2.19pre17) is 1.001.473 bytes long, and the new one
> > > > (vmlinuz-2.2.19foo17) is 1.033.565 bytes long. But the worst thing
> > > > is that the new kernle doesn't boot. What am I doing wrong??
>
> Has the old kernel and the new kernel been compiled with the exact same
> gcc version? If they haven't there's a more than likely chance that the
> kernels will end up differently sized.
>
> About the failed boot? Are you trying to optimzie for a specific
> architecture? Is it complaining about an initrd?
>
> --mike
>
> --
> To UNSUBSCRIBE, email to debian-user-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
The old kernel reports gcc 2.7.2.3 at boot time. So do the new one. The boot failure
happens at the "Partition check" in hda. It reports as follows:
Partition check:
hda: timeout waiting for DMA
hda: irq timeout: status=0x58 { DriveReady SeekComplete DataRequest }
The two last lines repeats twice or three times after a litlle delay, and the boot
process just continues, as if nothing hapened. But everything goes wrong from here on
(can't mount device for rw, and the timeout message periodically blinks on the screen,
even after logon. Didn't go any further).
The funny thing is that I have to turn off the machine (just pressing reset doesn't
work), or the old kernel won't boot also(!).
'Seems my IDE is an "ALI15x3 not 100% native mode (will probe irq's later)". If I compile
a kernel with the specific (experimental) driver things works fine. But the old kernel
also does work fine, and it's not compiled with the experimental driver!!!
What is happening?
Thanks
P.S. Checked the differences between files /boot/System.map-2.2.19foo17 and
/boot/System.map-2.2.19pre17 and they do differ, although I don't know the meaning of the
differences. Gonna try to compile it using the config file withaout the xconfig to see
what happens.
Reply to: