On Thu, Jul 30, 2020 at 11:03:40AM -0700, Nick Desaulniers wrote:
From: Geert Uytterhoeven geert@linux-m68k.org
commit 59b6359dd92d18f5dc04b14a4c926fa08ab66f7c upstream.
If CONFIG_DEBUG_LOCK_ALLOC=y, the kernel log is spammed with a few hundred identical messages:
unwind: Unknown symbol address c0800300 unwind: Index not found c0800300c0800300 is the return address from the last subroutine call (to __memzero()) in __mmap_switched(). Apparently having this address in the link register confuses the unwinder.
To fix this, reset the link register to zero before jumping to start_kernel().
Fixes: 9520b1a1b5f7a348 ("ARM: head-common.S: speed up startup code") Suggested-by: Ard Biesheuvel ard.biesheuvel@linaro.org Signed-off-by: Geert Uytterhoeven geert+renesas@glider.be Acked-by: Nicolas Pitre nico@linaro.org Signed-off-by: Russell King rmk+kernel@armlinux.org.uk Signed-off-by: Nick Desaulniers ndesaulniers@google.com
Looks like this first landed in v4.15-rc1. Without this, we can't tell during an unwind initiated from start_kernel() when to stop unwinding, which for the clang specific implementation of the arm frame pointer unwinder leads to dereferencing a garbage value, triggering an exception which has no fixup, triggering a panic, triggering an unwind, triggering an infinite loop that prevents booting. I have more patches to send upstream to make the unwinder more resilient, but it's ambiguous as to when to stop unwinding without this patch.
Note, the "Fixes:" tag points at something in 4.15, not 4.14, so are you _SURE_ this is needed in 4.14.y?
thanks,
greg k-h