Messages in this thread | | | Date | Wed, 11 Oct 2023 11:37:25 +0200 | From | Peter Zijlstra <> | Subject | Re: [PATCHv2] x86/alternatives: Disable KASAN in apply_alternatives() |
| |
On Wed, Oct 11, 2023 at 10:11:46AM +0200, Ingo Molnar wrote: > > * Peter Zijlstra <peterz@infradead.org> wrote: > > > > 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 .. > > Not all LA57 related sites are patched yet at this point, and KASAN sees > a weird & broken mixture of LA48 and LA57 runtime semantics, right? > > Ie. as far as KASAN is concerned, the LA48 -> LA57 behavioral switchover > must be atomic, but during the kernel code patching process it isn't.
Yep, half-way through the patching it observes inconsistencies and goes WTF :-)
| |