kcov: simplify interrupt check
authorDmitry Vyukov <dvyukov@google.com>
Mon, 8 May 2017 22:56:48 +0000 (15:56 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 9 May 2017 00:15:12 +0000 (17:15 -0700)
in_interrupt() semantics are confusing and wrong for most users as it
also returns true when bh is disabled.  Thus we open coded a proper
check for interrupts in __sanitizer_cov_trace_pc() with a lengthy
explanatory comment.

Use the new in_task() predicate instead.

Link: http://lkml.kernel.org/r/20170321091026.139655-1-dvyukov@google.com
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: James Morse <james.morse@arm.com>
Cc: Alexander Popov <alex.popov@linux.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/kcov.c

index 85e5546..cd77199 100644 (file)
@@ -60,15 +60,8 @@ void notrace __sanitizer_cov_trace_pc(void)
        /*
         * We are interested in code coverage as a function of a syscall inputs,
         * so we ignore code executed in interrupts.
-        * The checks for whether we are in an interrupt are open-coded, because
-        * 1. We can't use in_interrupt() here, since it also returns true
-        *    when we are inside local_bh_disable() section.
-        * 2. We don't want to use (in_irq() | in_serving_softirq() | in_nmi()),
-        *    since that leads to slower generated code (three separate tests,
-        *    one for each of the flags).
         */
-       if (!t || (preempt_count() & (HARDIRQ_MASK | SOFTIRQ_OFFSET
-                                                       | NMI_MASK)))
+       if (!t || !in_task())
                return;
        mode = READ_ONCE(t->kcov_mode);
        if (mode == KCOV_MODE_TRACE) {