Re: [PATCHv2] x86/alternatives: Disable KASAN in apply_alternatives()

[Kirill A. Shutemov]

On Wed, Oct 11, 2023 at 09:46:16AM +0200, Peter Zijlstra wrote:
> On Wed, Oct 11, 2023 at 09:58:49AM +0300, Kirill A. Shutemov wrote:
> > Fei has reported that KASAN triggers during apply_alternatives() on
> > 5-level paging machine:
> > 
> 
> Urgh @ KASAN splat, can't we summarize that?

What about this?

	BUG: KASAN: out-of-bounds in rcu_is_watching
	Read of size 4 at addr ff110003ee6419a0 by task swapper/0/0
	...
	__asan_load4
	rcu_is_watching
	? text_poke_early
	trace_hardirqs_on
	? __asan_load4
	text_poke_early
	apply_alternatives
	...

Is it enough details or I overdid summarization?

> > diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
> > index 517ee01503be..b4cc4d7c0825 100644
> > --- a/arch/x86/kernel/alternative.c
> > +++ b/arch/x86/kernel/alternative.c
> > @@ -403,6 +403,17 @@ void __init_or_module noinline apply_alternatives(struct alt_instr *start,
> >  	u8 insn_buff[MAX_PATCH_LEN];
> >  
> >  	DPRINTK(ALT, "alt table %px, -> %px", start, end);
> > +
> > +	/*
> > +	 * In the case CONFIG_X86_5LEVEL=y, KASAN_SHADOW_START is defined using
> > +	 * cpu_feature_enabled(X86_FEATURE_LA57) and is therefore patched here.
> > +	 * During the process, KASAN becomes confused and triggers
> 
> 	because of partial LA57 convertion ..
> 
> > +	 * a false-positive out-of-bound report.
> > +	 *
> > +	 * Disable KASAN until the patching is complete.
> > +	 */
> > +	kasan_disable_current();
> > +
> >  	/*

	/*
	 * In the case CONFIG_X86_5LEVEL=y, KASAN_SHADOW_START is defined using
	 * cpu_feature_enabled(X86_FEATURE_LA57) and is therefore patched here.
	 * During the process, KASAN becomes confused seeing partial LA57
	 * conversion and triggers a false-positive out-of-bound report.
	 *
	 * Disable KASAN until the patching is complete.
	 */

Looks good?

If yes, I will submit v3 with your Ack.

-- 
  Kiryl Shutsemau / Kirill A. Shutemov