I try to compile a custom kernel based on the lastes hijack kernel.
Compiling the default b13 kernel is not a problem, it works well. But as soon as I apply the three patches rds, voladj and hijack I get the following compile (link) error:

arm-empeg-linux-ld -p -X -T arch/arm/vmlinux.lds arch/arm/kernel/head-armv.o arch/arm/kernel/init_task.o init/main.o init/version.o \
--start-group \
arch/arm/kernel/kernel.o arch/arm/mm/mm.o kernel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o \
fs/filesystems.a \
net/network.a \
drivers/block/block.a drivers/char/char.a drivers/misc/misc.a drivers/net/net.a drivers/net/irda/irda_drivers.a arch/arm/special/special.a \
arch/arm/nwfpe/math-emu.a arch/arm/lib/lib.o arch/arm/lib/lib.a /usr/src/empeg/kernel.20b13hv299/kernel/lib/lib.a /usr/local/armtools/lib/gcc-lib/arm-empeg-linux/2.95.3/libgcc.a \
--end-group \
-o vmlinux
arch/arm/kernel/kernel.o: In function `time_init':
arch/arm/kernel/kernel.o(.text+0x6248): undefined reference to `mktime'
make: *** [vmlinux] Error 1

I found out that the problem is the modified file arch/arm/kernel/time.c
Hijack adds the following lines #if 0 and #endif and comments out the mktime function. I don't know why. If I remove these two "if endif" lines everything is ok, as expected.

Seems that at least on my maschine the "#if 0" get false, where on other maschine this statement return true.

Mark, do you have a explanation for that?

In reply to:

---

I found out that the problem is the modified file arch/arm/kernel/time.c

---

FWIW, I had the same problem and ended up removing the #if. I thought it was something in my kernel changes for ext3 that was needing that mktime, but I guess not. I am using the arm cross compiler from Mark's site.

-Mike

Hi, I've run into this too from time to time. Not sure exactly which combination of tools causes/fixes it, but it's definitely due to conflicting header files in the gcc #includes versus the empeg kernel #includes.

Cheers

I needed to remove that #if 0 as well. The only possiblity I can come up with is that some combination of tools you used was trying to link the kernel against libc.

So did I, I don't know why, and without having it happen I couldn't prove it.

Of course, I should relink my kernel with the correct version so jemplode stops offering to upgrade. It's v299 + ext3, but claims to be v298. Oops.