On Wed, Oct 13, 2021 at 09:37:01PM +0200, Arnd Bergmann wrote:

On Wed, Sep 29, 2021 at 6:10 PM Peter Zijlstra [off-list ref] wrote:

quoted

Make sure to prod idle CPUs so they call klp_update_patch_state().

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 kernel/livepatch/transition.c |    5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

--- a/kernel/livepatch/transition.c
+++ b/kernel/livepatch/transition.c

@@ -413,8 +413,11 @@ void klp_try_complete_transition(void)
        for_each_possible_cpu(cpu) {
                task = idle_task(cpu);
                if (cpu_online(cpu)) {
-                       if (!klp_try_switch_task(task))
+                       if (!klp_try_switch_task(task)) {
                                complete = false;
+                               /* Make idle task go through the main loop. */
+                               wake_up_if_idle(cpu);
+                       }

This caused a build regression on non-SMP kernels:

:-(

x86_64-linux-ld: kernel/livepatch/transition.o: in function
`klp_try_complete_transition':
transition.c:(.text+0x106e): undefined reference to `wake_up_if_idle'

Maybe add a IS_ENABLED(CONFIG_SMP) check to one of the if() conditions?

I'll just add a stub for that function I think. Let me rebase the thing
before I push more crap ontop..