Merge tag 'xtensa-20190307' of git://github.com/jcmvbkbc/linux-xtensa

Pull xtensa updates from Max Filippov:

 - use generic spinlock/rwlock implementations

 - clean up IPI processing

 - document boot parameters passing to the kernel

 - fix get_wchan

 - various cleanups in time.c, process.c, traps.c and thread_info.h

* tag 'xtensa-20190307' of git://github.com/jcmvbkbc/linux-xtensa:
  xtensa: simplify trap_init
  xtensa: drop unused definitions
  xtensa: fix get_wchan
  xtensa: use generic spinlock/rwlock implementation
  xtensa: provide xchg for sizes 1 and 2
  xtensa: clean up arch/xtensa/kernel/time.c
  xtensa: SMP: rework IPI processing
  xtensa: document boot parameter passing

10 files changed, 94 insertions, 257 deletions

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 963986a48c62..bacf87ee7975 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -5,6 +5,8 @@ config XTENSA
 	select ARCH_HAS_SYNC_DMA_FOR_CPU
 	select ARCH_HAS_SYNC_DMA_FOR_DEVICE
 	select ARCH_NO_COHERENT_DMA_MMAP if !MMU
+	select ARCH_USE_QUEUED_RWLOCKS
+	select ARCH_USE_QUEUED_SPINLOCKS
 	select ARCH_WANT_FRAME_POINTERS
 	select ARCH_WANT_IPC_PARSE_VERSION
 	select BUILDTIME_EXTABLE_SORT
diff --git a/arch/xtensa/include/asm/Kbuild b/arch/xtensa/include/asm/Kbuild
index 809f39ce08c0..d939e13e8d84 100644
--- a/arch/xtensa/include/asm/Kbuild
+++ b/arch/xtensa/include/asm/Kbuild
@@ -23,6 +23,8 @@ generic-y += mm-arch-hooks.h
 generic-y += param.h
 generic-y += percpu.h
 generic-y += preempt.h
+generic-y += qrwlock.h
+generic-y += qspinlock.h
 generic-y += rwsem.h
 generic-y += sections.h
 generic-y += socket.h
diff --git a/arch/xtensa/include/asm/cmpxchg.h b/arch/xtensa/include/asm/cmpxchg.h
index 201e9009efd8..22a10c715c1f 100644
--- a/arch/xtensa/include/asm/cmpxchg.h
+++ b/arch/xtensa/include/asm/cmpxchg.h
@@ -13,6 +13,7 @@
 
 #ifndef __ASSEMBLY__
 
+#include <linux/bits.h>
 #include <linux/stringify.h>
 
 /*
@@ -138,6 +139,28 @@ static inline unsigned long xchg_u32(volatile int * m, unsigned long val)
 #define xchg(ptr,x) \
 	((__typeof__(*(ptr)))__xchg((unsigned long)(x),(ptr),sizeof(*(ptr))))
 
+static inline u32 xchg_small(volatile void *ptr, u32 x, int size)
+{
+	int off = (unsigned long)ptr % sizeof(u32);
+	volatile u32 *p = ptr - off;
+#ifdef __BIG_ENDIAN
+	int bitoff = (sizeof(u32) - size - off) * BITS_PER_BYTE;
+#else
+	int bitoff = off * BITS_PER_BYTE;
+#endif
+	u32 bitmask = ((0x1 << size * BITS_PER_BYTE) - 1) << bitoff;
+	u32 oldv, newv;
+	u32 ret;
+
+	do {
+		oldv = READ_ONCE(*p);
+		ret = (oldv & bitmask) >> bitoff;
+		newv = (oldv & ~bitmask) | (x << bitoff);
+	} while (__cmpxchg_u32(p, oldv, newv) != oldv);
+
+	return ret;
+}
+
 /*
  * This only works if the compiler isn't horribly bad at optimizing.
  * gcc-2.5.8 reportedly can't handle this, but I define that one to
@@ -150,11 +173,16 @@ static __inline__ unsigned long
 __xchg(unsigned long x, volatile void * ptr, int size)
 {
 	switch (size) {
-		case 4:
-			return xchg_u32(ptr, x);
+	case 1:
+		return xchg_small(ptr, x, 1);
+	case 2:
+		return xchg_small(ptr, x, 2);
+	case 4:
+		return xchg_u32(ptr, x);
+	default:
+		__xchg_called_with_bad_pointer();
+		return x;
 	}
-	__xchg_called_with_bad_pointer();
-	return x;
 }
 
 #endif /* __ASSEMBLY__ */
diff --git a/arch/xtensa/include/asm/spinlock.h b/arch/xtensa/include/asm/spinlock.h
index c6e1290dcbb7..584b0de6f2ca 100644
--- a/arch/xtensa/include/asm/spinlock.h
+++ b/arch/xtensa/include/asm/spinlock.h
@@ -12,188 +12,9 @@
 #define _XTENSA_SPINLOCK_H
 
 #include <asm/barrier.h>
-#include <asm/processor.h>
+#include <asm/qrwlock.h>
+#include <asm/qspinlock.h>
 
-/*
- * spinlock
- *
- * There is at most one owner of a spinlock.  There are not different
- * types of spinlock owners like there are for rwlocks (see below).
- *
- * When trying to obtain a spinlock, the function "spins" forever, or busy-
- * waits, until the lock is obtained.  When spinning, presumably some other
- * owner will soon give up the spinlock making it available to others.  Use
- * the trylock functions to avoid spinning forever.
- *
- * possible values:
- *
- *    0         nobody owns the spinlock
- *    1         somebody owns the spinlock
- */
-
-#define arch_spin_is_locked(x) ((x)->slock != 0)
-
-static inline void arch_spin_lock(arch_spinlock_t *lock)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       wsr     %0, scompare1\n"
-			"1:     movi    %0, 1\n"
-			"       s32c1i  %0, %1, 0\n"
-			"       bnez    %0, 1b\n"
-			: "=&a" (tmp)
-			: "a" (&lock->slock)
-			: "memory");
-}
-
-/* Returns 1 if the lock is obtained, 0 otherwise. */
-
-static inline int arch_spin_trylock(arch_spinlock_t *lock)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       wsr     %0, scompare1\n"
-			"       movi    %0, 1\n"
-			"       s32c1i  %0, %1, 0\n"
-			: "=&a" (tmp)
-			: "a" (&lock->slock)
-			: "memory");
-
-	return tmp == 0 ? 1 : 0;
-}
-
-static inline void arch_spin_unlock(arch_spinlock_t *lock)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       s32ri   %0, %1, 0\n"
-			: "=&a" (tmp)
-			: "a" (&lock->slock)
-			: "memory");
-}
-
-/*
- * rwlock
- *
- * Read-write locks are really a more flexible spinlock.  They allow
- * multiple readers but only one writer.  Write ownership is exclusive
- * (i.e., all other readers and writers are blocked from ownership while
- * there is a write owner).  These rwlocks are unfair to writers.  Writers
- * can be starved for an indefinite time by readers.
- *
- * possible values:
- *
- *   0          nobody owns the rwlock
- *  >0          one or more readers own the rwlock
- *                (the positive value is the actual number of readers)
- *  0x80000000  one writer owns the rwlock, no other writers, no readers
- */
-
-static inline void arch_write_lock(arch_rwlock_t *rw)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       wsr     %0, scompare1\n"
-			"1:     movi    %0, 1\n"
-			"       slli    %0, %0, 31\n"
-			"       s32c1i  %0, %1, 0\n"
-			"       bnez    %0, 1b\n"
-			: "=&a" (tmp)
-			: "a" (&rw->lock)
-			: "memory");
-}
-
-/* Returns 1 if the lock is obtained, 0 otherwise. */
-
-static inline int arch_write_trylock(arch_rwlock_t *rw)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       wsr     %0, scompare1\n"
-			"       movi    %0, 1\n"
-			"       slli    %0, %0, 31\n"
-			"       s32c1i  %0, %1, 0\n"
-			: "=&a" (tmp)
-			: "a" (&rw->lock)
-			: "memory");
-
-	return tmp == 0 ? 1 : 0;
-}
-
-static inline void arch_write_unlock(arch_rwlock_t *rw)
-{
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       movi    %0, 0\n"
-			"       s32ri   %0, %1, 0\n"
-			: "=&a" (tmp)
-			: "a" (&rw->lock)
-			: "memory");
-}
-
-static inline void arch_read_lock(arch_rwlock_t *rw)
-{
-	unsigned long tmp;
-	unsigned long result;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %2, 0\n"
-			"       bltz    %1, 1b\n"
-			"       wsr     %1, scompare1\n"
-			"       addi    %0, %1, 1\n"
-			"       s32c1i  %0, %2, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (result), "=&a" (tmp)
-			: "a" (&rw->lock)
-			: "memory");
-}
-
-/* Returns 1 if the lock is obtained, 0 otherwise. */
-
-static inline int arch_read_trylock(arch_rwlock_t *rw)
-{
-	unsigned long result;
-	unsigned long tmp;
-
-	__asm__ __volatile__(
-			"       l32i    %1, %2, 0\n"
-			"       addi    %0, %1, 1\n"
-			"       bltz    %0, 1f\n"
-			"       wsr     %1, scompare1\n"
-			"       s32c1i  %0, %2, 0\n"
-			"       sub     %0, %0, %1\n"
-			"1:\n"
-			: "=&a" (result), "=&a" (tmp)
-			: "a" (&rw->lock)
-			: "memory");
-
-	return result == 0;
-}
-
-static inline void arch_read_unlock(arch_rwlock_t *rw)
-{
-	unsigned long tmp1, tmp2;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %2, 0\n"
-			"       addi    %0, %1, -1\n"
-			"       wsr     %1, scompare1\n"
-			"       s32c1i  %0, %2, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp1), "=&a" (tmp2)
-			: "a" (&rw->lock)
-			: "memory");
-}
+#define smp_mb__after_spinlock()	smp_mb()
 
 #endif	/* _XTENSA_SPINLOCK_H */
diff --git a/arch/xtensa/include/asm/spinlock_types.h b/arch/xtensa/include/asm/spinlock_types.h
index bb1fe6c1816e..64c9389254f1 100644
--- a/arch/xtensa/include/asm/spinlock_types.h
+++ b/arch/xtensa/include/asm/spinlock_types.h
@@ -2,20 +2,11 @@
 #ifndef __ASM_SPINLOCK_TYPES_H
 #define __ASM_SPINLOCK_TYPES_H
 
-#ifndef __LINUX_SPINLOCK_TYPES_H
+#if !defined(__LINUX_SPINLOCK_TYPES_H) && !defined(__ASM_SPINLOCK_H)
 # error "please don't include this file directly"
 #endif
 
-typedef struct {
-	volatile unsigned int slock;
-} arch_spinlock_t;
-
-#define __ARCH_SPIN_LOCK_UNLOCKED	{ 0 }
-
-typedef struct {
-	volatile unsigned int lock;
-} arch_rwlock_t;
-
-#define __ARCH_RW_LOCK_UNLOCKED		{ 0 }
+#include <asm-generic/qspinlock_types.h>
+#include <asm-generic/qrwlock_types.h>
 
 #endif
diff --git a/arch/xtensa/include/asm/thread_info.h b/arch/xtensa/include/asm/thread_info.h
index f333f10a7650..f092cc3f4e66 100644
--- a/arch/xtensa/include/asm/thread_info.h
+++ b/arch/xtensa/include/asm/thread_info.h
@@ -121,15 +121,6 @@ static inline struct thread_info *current_thread_info(void)
 #define _TIF_WORK_MASK		(_TIF_SYSCALL_TRACE | _TIF_SINGLESTEP | \
 				 _TIF_SYSCALL_TRACEPOINT)
 
-/*
- * Thread-synchronous status.
- *
- * This is different from the flags in that nobody else
- * ever touches our thread-synchronous status, so we don't
- * have to worry about atomic accesses.
- */
-#define TS_USEDFPU		0x0001	/* FPU was used by this task this quantum (SMP) */
-
 #define THREAD_SIZE KERNEL_STACK_SIZE
 #define THREAD_SIZE_ORDER (KERNEL_STACK_SHIFT - PAGE_SHIFT)
 
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index 74969a437a37..db278a9e80c7 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -52,8 +52,6 @@
 extern void ret_from_fork(void);
 extern void ret_from_kernel_thread(void);
 
-struct task_struct *current_set[NR_CPUS] = {&init_task, };
-
 void (*pm_power_off)(void) = NULL;
 EXPORT_SYMBOL(pm_power_off);
 
@@ -321,8 +319,8 @@ unsigned long get_wchan(struct task_struct *p)
 
 		/* Stack layout: sp-4: ra, sp-3: sp' */
 
-		pc = MAKE_PC_FROM_RA(*(unsigned long*)sp - 4, sp);
-		sp = *(unsigned long *)sp - 3;
+		pc = MAKE_PC_FROM_RA(SPILL_SLOT(sp, 0), sp);
+		sp = SPILL_SLOT(sp, 1);
 	} while (count++ < 16);
 	return 0;
 }
diff --git a/arch/xtensa/kernel/smp.c b/arch/xtensa/kernel/smp.c
index be1f280c322c..3699d6d3e479 100644
--- a/arch/xtensa/kernel/smp.c
+++ b/arch/xtensa/kernel/smp.c
@@ -372,8 +372,7 @@ static void send_ipi_message(const struct cpumask *callmask,
 	unsigned long mask = 0;
 
 	for_each_cpu(index, callmask)
-		if (index != smp_processor_id())
-			mask |= 1 << index;
+		mask |= 1 << index;
 
 	set_er(mask, MIPISET(msg_id));
 }
@@ -412,22 +411,31 @@ irqreturn_t ipi_interrupt(int irq, void *dev_id)
 {
 	unsigned int cpu = smp_processor_id();
 	struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
-	unsigned int msg;
-	unsigned i;
 
-	msg = get_er(MIPICAUSE(cpu));
-	for (i = 0; i < IPI_MAX; i++)
-		if (msg & (1 << i)) {
-			set_er(1 << i, MIPICAUSE(cpu));
-			++ipi->ipi_count[i];
+	for (;;) {
+		unsigned int msg;
+
+		msg = get_er(MIPICAUSE(cpu));
+		set_er(msg, MIPICAUSE(cpu));
+
+		if (!msg)
+			break;
+
+		if (msg & (1 << IPI_CALL_FUNC)) {
+			++ipi->ipi_count[IPI_CALL_FUNC];
+			generic_smp_call_function_interrupt();
 		}
 
-	if (msg & (1 << IPI_RESCHEDULE))
-		scheduler_ipi();
-	if (msg & (1 << IPI_CALL_FUNC))
-		generic_smp_call_function_interrupt();
-	if (msg & (1 << IPI_CPU_STOP))
-		ipi_cpu_stop(cpu);
+		if (msg & (1 << IPI_RESCHEDULE)) {
+			++ipi->ipi_count[IPI_RESCHEDULE];
+			scheduler_ipi();
+		}
+
+		if (msg & (1 << IPI_CPU_STOP)) {
+			++ipi->ipi_count[IPI_CPU_STOP];
+			ipi_cpu_stop(cpu);
+		}
+	}
 
 	return IRQ_HANDLED;
 }
diff --git a/arch/xtensa/kernel/time.c b/arch/xtensa/kernel/time.c
index 378186b5eb40..69db8c93c1f9 100644
--- a/arch/xtensa/kernel/time.c
+++ b/arch/xtensa/kernel/time.c
@@ -52,14 +52,11 @@ static struct clocksource ccount_clocksource = {
 	.flags = CLOCK_SOURCE_IS_CONTINUOUS,
 };
 
-static int ccount_timer_set_next_event(unsigned long delta,
-		struct clock_event_device *dev);
 struct ccount_timer {
 	struct clock_event_device evt;
 	int irq_enabled;
 	char name[24];
 };
-static DEFINE_PER_CPU(struct ccount_timer, ccount_timer);
 
 static int ccount_timer_set_next_event(unsigned long delta,
 		struct clock_event_device *dev)
@@ -107,7 +104,30 @@ static int ccount_timer_set_oneshot(struct clock_event_device *evt)
 	return 0;
 }
 
-static irqreturn_t timer_interrupt(int irq, void *dev_id);
+static DEFINE_PER_CPU(struct ccount_timer, ccount_timer) = {
+	.evt = {
+		.features = CLOCK_EVT_FEAT_ONESHOT,
+		.rating = 300,
+		.set_next_event = ccount_timer_set_next_event,
+		.set_state_shutdown = ccount_timer_shutdown,
+		.set_state_oneshot = ccount_timer_set_oneshot,
+		.tick_resume = ccount_timer_set_oneshot,
+	},
+};
+
+static irqreturn_t timer_interrupt(int irq, void *dev_id)
+{
+	struct clock_event_device *evt = &this_cpu_ptr(&ccount_timer)->evt;
+
+	set_linux_timer(get_linux_timer());
+	evt->event_handler(evt);
+
+	/* Allow platform to do something useful (Wdog). */
+	platform_heartbeat();
+
+	return IRQ_HANDLED;
+}
+
 static struct irqaction timer_irqaction = {
 	.handler =	timer_interrupt,
 	.flags =	IRQF_TIMER,
@@ -120,14 +140,8 @@ void local_timer_setup(unsigned cpu)
 	struct clock_event_device *clockevent = &timer->evt;
 
 	timer->irq_enabled = 1;
-	clockevent->name = timer->name;
 	snprintf(timer->name, sizeof(timer->name), "ccount_clockevent_%u", cpu);
-	clockevent->features = CLOCK_EVT_FEAT_ONESHOT;
-	clockevent->rating = 300;
-	clockevent->set_next_event = ccount_timer_set_next_event;
-	clockevent->set_state_shutdown = ccount_timer_shutdown;
-	clockevent->set_state_oneshot = ccount_timer_set_oneshot;
-	clockevent->tick_resume = ccount_timer_set_oneshot;
+	clockevent->name = timer->name;
 	clockevent->cpumask = cpumask_of(cpu);
 	clockevent->irq = irq_create_mapping(NULL, LINUX_TIMER_INT);
 	if (WARN(!clockevent->irq, "error: can't map timer irq"))
@@ -190,23 +204,6 @@ void __init time_init(void)
 	timer_probe();
 }
 
-/*
- * The timer interrupt is called HZ times per second.
- */
-
-irqreturn_t timer_interrupt(int irq, void *dev_id)
-{
-	struct clock_event_device *evt = &this_cpu_ptr(&ccount_timer)->evt;
-
-	set_linux_timer(get_linux_timer());
-	evt->event_handler(evt);
-
-	/* Allow platform to do something useful (Wdog). */
-	platform_heartbeat();
-
-	return IRQ_HANDLED;
-}
-
 #ifndef CONFIG_GENERIC_CALIBRATE_DELAY
 void calibrate_delay(void)
 {
diff --git a/arch/xtensa/kernel/traps.c b/arch/xtensa/kernel/traps.c
index e6fa55aa1ccb..454d53096bc9 100644
--- a/arch/xtensa/kernel/traps.c
+++ b/arch/xtensa/kernel/traps.c
@@ -420,16 +420,15 @@ void __init trap_init(void)
 	/* Setup specific handlers. */
 
 	for(i = 0; dispatch_init_table[i].cause >= 0; i++) {
-
 		int fast = dispatch_init_table[i].fast;
 		int cause = dispatch_init_table[i].cause;
 		void *handler = dispatch_init_table[i].handler;
 
 		if (fast == 0)
 			set_handler(default_handler, cause, handler);
-		if (fast && fast & USER)
+		if ((fast & USER) != 0)
 			set_handler(fast_user_handler, cause, handler);
-		if (fast && fast & KRNL)
+		if ((fast & KRNL) != 0)
 			set_handler(fast_kernel_handler, cause, handler);
 	}