[PATCH v2 25/34] compiler: Let data_race() imply disabled capability analysis

[Marco Elver <elver@google.com>]

Many patterns that involve data-racy accesses often deliberately ignore
normal synchronization rules to avoid taking a lock.

If we have a lock-guarded variable on which we do a lock-less data-racy
access, rather than having to write capability_unsafe(data_race(..)),
simply make the data_race(..) macro imply capability-unsafety. The
data_race() macro already denotes the intent that something subtly
unsafe is about to happen, so it should be clear enough as-is.

Signed-off-by: Marco Elver <elver@google.com>
---
v2:
* New patch.
---
 include/linux/compiler.h       | 2 ++
 lib/test_capability-analysis.c | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 155385754824..c837464369df 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -186,7 +186,9 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
 #define data_race(expr)							\
 ({									\
 	__kcsan_disable_current();					\
+	disable_capability_analysis();					\
 	__auto_type __v = (expr);					\
+	enable_capability_analysis();					\
 	__kcsan_enable_current();					\
 	__v;								\
 })
diff --git a/lib/test_capability-analysis.c b/lib/test_capability-analysis.c
index 853fdc53840f..13e7732c38a2 100644
--- a/lib/test_capability-analysis.c
+++ b/lib/test_capability-analysis.c
@@ -92,6 +92,8 @@ static void __used test_raw_spinlock_trylock_extra(struct test_raw_spinlock_data
 {
 	unsigned long flags;
 
+	data_race(d->counter++); /* no warning */
+
 	if (raw_spin_trylock_irq(&d->lock)) {
 		d->counter++;
 		raw_spin_unlock_irq(&d->lock);
-- 
2.48.1.711.g2feabab25a-goog