s390/preempt: disable __preempt_count_add() optimization for PROFILE_…

…ALL_BRANCHES

[ Upstream commit 63678ee ]

gcc 12 does not (always) optimize away code that should only be generated
if parameters are constant and within in a certain range. This depends on
various obscure kernel config options, however in particular
PROFILE_ALL_BRANCHES can trigger this compile error:

In function ‘__atomic_add_const’,
    inlined from ‘__preempt_count_add.part.0’ at ./arch/s390/include/asm/preempt.h:50:3:
./arch/s390/include/asm/atomic_ops.h:80:9: error: impossible constraint in ‘asm’
   80 |         asm volatile(                                                   \
      |         ^~~

Workaround this by simply disabling the optimization for
PROFILE_ALL_BRANCHES, since the kernel will be so slow, that this
optimization won't matter at all.

Reported-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arch/s390/include/asm/preempt.h

@@ -46,10 +46,17 @@ static inline bool test_preempt_need_resched(void)
 
 static inline void __preempt_count_add(int val)
 {
-	if (__builtin_constant_p(val) && (val >= -128) && (val <= 127))
-		__atomic_add_const(val, &S390_lowcore.preempt_count);
-	else
-		__atomic_add(val, &S390_lowcore.preempt_count);
+	/*
+	 * With some obscure config options and CONFIG_PROFILE_ALL_BRANCHES
+	 * enabled, gcc 12 fails to handle __builtin_constant_p().
+	 */
+	if (!IS_ENABLED(CONFIG_PROFILE_ALL_BRANCHES)) {
+		if (__builtin_constant_p(val) && (val >= -128) && (val <= 127)) {
+			__atomic_add_const(val, &S390_lowcore.preempt_count);
+			return;
+		}
+	}
+	__atomic_add(val, &S390_lowcore.preempt_count);
 }
 
 static inline void __preempt_count_sub(int val)