powerpc: Fix hard_irq_disable() with sanitizer

As reported by Zhouyi Zhou, WRITE_ONCE() is not atomic
as expected when KASAN or KCSAN are compiled in.

Fix it by re-implementing it using inline assembly.

Fixes: 077fc62b2b66 ("powerpc/irq: remove inline assembly in hard_irq_disable macro")
Reported-by: Zhouyi Zhou <zhouzhouyi@gmail.com>
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/a8298991b3df049a54ee8e558838e34265812014.1661272586.git.christophe.leroy@csgroup.eu

diff --git a/arch/powerpc/include/asm/hw_irq.h b/arch/powerpc/include/asm/hw_irq.h
index 26ede09..3c8cb48 100644 (file)
--- a/arch/powerpc/include/asm/hw_irq.h
+++ b/arch/powerpc/include/asm/hw_irq.h
@@ -282,7 +282,8 @@ static inline bool pmi_irq_pending(void)
        flags = irq_soft_mask_set_return(IRQS_ALL_DISABLED);            \
        local_paca->irq_happened |= PACA_IRQ_HARD_DIS;                  \
        if (!arch_irqs_disabled_flags(flags)) {                         \
-               WRITE_ONCE(local_paca->saved_r1, current_stack_pointer);\
+               asm volatile("std%X0 %1,%0" : "=m" (local_paca->saved_r1) \
+                                           : "r" (current_stack_pointer)); \
                trace_hardirqs_off();                                   \
        }                                                               \
 } while(0)