Re: Linux 6.1-rc3

[Miguel Ojeda]

On Tue, Nov 1, 2022 at 6:58 PM Arnd Bergmann <arnd@xxxxxxxx> wrote:
>
> Linker-generated symbols are usually the cause, but those tend to
> be fixed with the extra pass, and Guenter said that doing an extra
> pass does not avoid the issue here.

I have been trying to reproduce the issue on my side, trying to match
Guenter's environment, but so far I have not been successful, even
when running the toolchain binaries he sent me (thanks a lot!).

What I could reproduce easily, though, was one from the kernel test
robot, powerpc as well, in 5.15.y [1] It requires 4 passes (i.e.
KALLSYMS_EXTRA_PASS is not enough), but it is not an oscillating one.

[1] https://lore.kernel.org/lkml/202209241418.35P6OMjS-lkp@xxxxxxxxx/

> What I've done in the past to debug this was to change
> scripts/link-vmlinux.sh to use a larger number of steps,
> mainly to see if it eventually converges, grows indefinitely
> or oscillates.

Does that mean you saw the oscillating one in the past? Guenter thinks
it is a new issue, since he never saw the oscillating case.

If you did see it, then it may be the case that the kallsyms changes
make some edge case more likely -- at least this is what Boqun and I
originally thought.

Cheers,
Miguel