xtensa: SMP: fix secondary CPU initialization

- add missing memory barriers to the secondary CPU synchronization spin
  loops; add comment to the matching memory barrier in the boot_secondary
  and __cpu_die functions;
- use READ_ONCE/WRITE_ONCE to access cpu_start_id/cpu_start_ccount
  instead of reading/writing them directly;
- re-initialize cpu_running every time before starting secondary CPU to
  flush possible previous CPU startup results.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>

arch/xtensa/kernel/head.S

@@ -276,12 +276,13 @@ should_never_return:
 
 	movi	a2, cpu_start_ccount
 1:
+	memw
 	l32i	a3, a2, 0
 	beqi	a3, 0, 1b
 	movi	a3, 0
 	s32i	a3, a2, 0
-	memw
 1:
+	memw
 	l32i	a3, a2, 0
 	beqi	a3, 0, 1b
 	wsr	a3, ccount
@@ -317,11 +318,13 @@ ENTRY(cpu_restart)
 	rsr	a0, prid
 	neg	a2, a0
 	movi	a3, cpu_start_id
+	memw
 	s32i	a2, a3, 0
 #if XCHAL_DCACHE_IS_WRITEBACK
 	dhwbi	a3, 0
 #endif
 1:
+	memw
 	l32i	a2, a3, 0
 	dhi	a3, 0
 	bne	a2, a0, 1b