Re: [PATCH] livepatch: Remove klp_arch_set_pc() and asm/livepatch.h
From: Peter Zijlstra <peterz@infradead.org>
Date: 2022-03-28 10:24:41
Also in:
linux-s390, live-patching, lkml
From: Peter Zijlstra <peterz@infradead.org>
Date: 2022-03-28 10:24:41
Also in:
linux-s390, live-patching, lkml
On Mon, Mar 28, 2022 at 08:26:48AM +0200, Christophe Leroy wrote:
All three versions of klp_arch_set_pc() do exactly the same: they call ftrace_instruction_pointer_set(). Call ftrace_instruction_pointer_set() directly and remove klp_arch_set_pc(). As klp_arch_set_pc() was the only thing remaining in asm/livepatch.h on x86 and s390, remove asm/livepatch.h livepatch.h remains on powerpc but its content is exclusively used by powerpc specific code.
The only remaining thing seems to be a klp_init_thread_info() for ppc64, but yeah.. Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
diff --git a/kernel/livepatch/patch.c b/kernel/livepatch/patch.c index c172bf92b576..4c4f5a776d80 100644 --- a/kernel/livepatch/patch.c +++ b/kernel/livepatch/patch.c@@ -118,7 +118,7 @@ static void notrace klp_ftrace_handler(unsigned long ip, if (func->nop) goto unlock; - klp_arch_set_pc(fregs, (unsigned long)func->new_func); + ftrace_instruction_pointer_set(fregs, (unsigned long)func->new_func); unlock: ftrace_test_recursion_unlock(bit);
So ptrace has instruction_pointer_set(), I'm not sure why we have yet another variant in the ftrace_ namespace. Perhaps also delete that?