Started the LFS process in an emulated system, via qemu. Finished making the temporary toolchain with no apparent errors. Next is the chroot phase.

It has occurred to me that I could tar up the tools directory and then untar it to my real (not emulated) system. I would then finish the LFS process (all in a chroot, as per the book) on the actual real partition that the system is intended for.

Would this work? Potential problem would be instruction set differences between original emulated CPU and the real CPU, as reflected somehow in the toolchain, could cause library or compiler snarls.

I'd set the emulator to use KVM, originally to get faster emulation. My understanding is that with KVM, opcodes executed by the emulator are actually somehow passed outside the emulation to be executed directly on the real CPU. This may mean that the emulated system was aware of the exact type of CPU and thus compiled for its capabilities? If so, going outside of emulation for the rest of the process should work? I am not sure.

Thank you.

Never tried it but why not copy tools and sources over and try, You will still have the original stuff where it is. I dont see a problem tho.

Similar case, with different environment.
I am using tools created on vmware to build on real system.
It was OK.

So I tried it. It actually worked. Yay! My LFS boots and runs.

Thanks everyone for the replies.