Merge tag 'xtensa-20220523' of https://github.com/jcmvbkbc/linux-xtensa

Pull xtensa architecture updates from Max Filippov:

 - support coprocessors on SMP

 - support KCSAN

 - support handling protection faults in noMMU configurations

 - support using coprocessors in the kernel mode

 - support hibernation

 - enable context tracking

 - enable HAVE_VIRT_CPU_ACCOUNTING_GEN

 - support division by 0 exception on cores without HW division option

 - clean up locking in the ISS network driver

 - clean up kernel entry assemly code

 - various minor fixes

* tag 'xtensa-20220523' of https://github.com/jcmvbkbc/linux-xtensa: (36 commits)
  xtensa: Return true/false (not 1/0) from bool function
  xtensa: improve call0 ABI probing
  xtensa: support artificial division by 0 exception
  xtensa: add trap handler for division by zero
  xtensa/simdisk: fix proc_read_simdisk()
  xtensa: no need to initialise statics to 0
  xtensa: clean up labels in the kernel entry assembly
  xtensa: don't leave invalid TLB entry in fast_store_prohibited
  xtensa: fix declaration of _SecondaryResetVector_text_*
  irqchip: irq-xtensa-mx: fix initial IRQ affinity
  xtensa: enable ARCH_HAS_DEBUG_VM_PGTABLE
  xtensa: add hibernation support
  xtensa: support coprocessors on SMP
  xtensa: get rid of stack frame in coprocessor_flush
  xtensa: merge SAVE_CP_REGS_TAB and LOAD_CP_REGS_TAB
  xtensa: add xtensa_xsr macro
  xtensa: handle coprocessor exceptions in kernel mode
  xtensa: use callx0 opcode in fast_coprocessor
  xtensa: clean up excsave1 initialization
  xtensa: clean up declarations in coprocessor.h
  ...

29 files changed, 879 insertions, 481 deletions

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index bd113bc6e192..0b0f0172cced 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -4,6 +4,7 @@ config XTENSA
 	select ARCH_32BIT_OFF_T
 	select ARCH_HAS_BINFMT_FLAT if !MMU
 	select ARCH_HAS_CURRENT_STACK_POINTER
+	select ARCH_HAS_DEBUG_VM_PGTABLE
 	select ARCH_HAS_DMA_PREP_COHERENT if MMU
 	select ARCH_HAS_SYNC_DMA_FOR_CPU if MMU
 	select ARCH_HAS_SYNC_DMA_FOR_DEVICE if MMU
@@ -29,8 +30,10 @@ config XTENSA
 	select HAVE_ARCH_AUDITSYSCALL
 	select HAVE_ARCH_JUMP_LABEL if !XIP_KERNEL
 	select HAVE_ARCH_KASAN if MMU && !XIP_KERNEL
+	select HAVE_ARCH_KCSAN
 	select HAVE_ARCH_SECCOMP_FILTER
 	select HAVE_ARCH_TRACEHOOK
+	select HAVE_CONTEXT_TRACKING
 	select HAVE_DEBUG_KMEMLEAK
 	select HAVE_DMA_CONTIGUOUS
 	select HAVE_EXIT_THREAD
@@ -42,6 +45,7 @@ config XTENSA
 	select HAVE_PERF_EVENTS
 	select HAVE_STACKPROTECTOR
 	select HAVE_SYSCALL_TRACEPOINTS
+	select HAVE_VIRT_CPU_ACCOUNTING_GEN
 	select IRQ_DOMAIN
 	select MODULES_USE_ELF_RELA
 	select PERF_USE_VMALLOC
@@ -79,6 +83,7 @@ config STACKTRACE_SUPPORT
 
 config MMU
 	def_bool n
+	select PFAULT
 
 config HAVE_XTENSA_GPIO32
 	def_bool n
@@ -178,6 +183,16 @@ config XTENSA_FAKE_NMI
 
 	  If unsure, say N.
 
+config PFAULT
+	bool "Handle protection faults" if EXPERT && !MMU
+	default y
+	help
+	  Handle protection faults. MMU configurations must enable it.
+	  noMMU configurations may disable it if used memory map never
+	  generates protection faults or faults are always fatal.
+
+	  If unsure, say Y.
+
 config XTENSA_UNALIGNED_USER
 	bool "Unaligned memory access in user space"
 	help
@@ -773,6 +788,9 @@ endmenu
 
 menu "Power management options"
 
+config ARCH_HIBERNATION_POSSIBLE
+	def_bool y
+
 source "kernel/power/Kconfig"
 
 endmenu
diff --git a/arch/xtensa/boot/lib/Makefile b/arch/xtensa/boot/lib/Makefile
index e3d717c7bfa1..162d10af36f3 100644
--- a/arch/xtensa/boot/lib/Makefile
+++ b/arch/xtensa/boot/lib/Makefile
@@ -16,6 +16,7 @@ CFLAGS_REMOVE_inffast.o = -pg
 endif
 
 KASAN_SANITIZE := n
+KCSAN_SANITIZE := n
 
 CFLAGS_REMOVE_inflate.o += -fstack-protector -fstack-protector-strong
 CFLAGS_REMOVE_zmem.o += -fstack-protector -fstack-protector-strong
diff --git a/arch/xtensa/include/asm/barrier.h b/arch/xtensa/include/asm/barrier.h
index d6f8d4ddc2bc..898ea397e9bc 100644
--- a/arch/xtensa/include/asm/barrier.h
+++ b/arch/xtensa/include/asm/barrier.h
@@ -11,9 +11,15 @@
 
 #include <asm/core.h>
 
-#define mb()  ({ __asm__ __volatile__("memw" : : : "memory"); })
-#define rmb() barrier()
-#define wmb() mb()
+#define __mb()  ({ __asm__ __volatile__("memw" : : : "memory"); })
+#define __rmb() barrier()
+#define __wmb() __mb()
+
+#ifdef CONFIG_SMP
+#define __smp_mb() __mb()
+#define __smp_rmb() __rmb()
+#define __smp_wmb() __wmb()
+#endif
 
 #if XCHAL_HAVE_S32C1I
 #define __smp_mb__before_atomic()		barrier()
diff --git a/arch/xtensa/include/asm/bitops.h b/arch/xtensa/include/asm/bitops.h
index cd225896c40f..e02ec5833389 100644
--- a/arch/xtensa/include/asm/bitops.h
+++ b/arch/xtensa/include/asm/bitops.h
@@ -99,7 +99,7 @@ static inline unsigned long __fls(unsigned long word)
 #if XCHAL_HAVE_EXCLUSIVE
 
 #define BIT_OP(op, insn, inv)						\
-static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
+static inline void arch_##op##_bit(unsigned int bit, volatile unsigned long *p)\
 {									\
 	unsigned long tmp;						\
 	unsigned long mask = 1UL << (bit & 31);				\
@@ -119,7 +119,7 @@ static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
 
 #define TEST_AND_BIT_OP(op, insn, inv)					\
 static inline int							\
-test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
+arch_test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
 {									\
 	unsigned long tmp, value;					\
 	unsigned long mask = 1UL << (bit & 31);				\
@@ -142,7 +142,7 @@ test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
 #elif XCHAL_HAVE_S32C1I
 
 #define BIT_OP(op, insn, inv)						\
-static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
+static inline void arch_##op##_bit(unsigned int bit, volatile unsigned long *p)\
 {									\
 	unsigned long tmp, value;					\
 	unsigned long mask = 1UL << (bit & 31);				\
@@ -163,7 +163,7 @@ static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
 
 #define TEST_AND_BIT_OP(op, insn, inv)					\
 static inline int							\
-test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
+arch_test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
 {									\
 	unsigned long tmp, value;					\
 	unsigned long mask = 1UL << (bit & 31);				\
@@ -205,6 +205,8 @@ BIT_OPS(change, "xor", )
 #undef BIT_OP
 #undef TEST_AND_BIT_OP
 
+#include <asm-generic/bitops/instrumented-atomic.h>
+
 #include <asm-generic/bitops/le.h>
 
 #include <asm-generic/bitops/ext2-atomic-setbit.h>
diff --git a/arch/xtensa/include/asm/coprocessor.h b/arch/xtensa/include/asm/coprocessor.h
index 0fbe2a740b8d..3b1a0d5d2169 100644
--- a/arch/xtensa/include/asm/coprocessor.h
+++ b/arch/xtensa/include/asm/coprocessor.h
@@ -142,11 +142,12 @@ typedef struct { XCHAL_CP6_SA_LIST(2) } xtregs_cp6_t
 typedef struct { XCHAL_CP7_SA_LIST(2) } xtregs_cp7_t
 	__attribute__ ((aligned (XCHAL_CP7_SA_ALIGN)));
 
-extern struct thread_info* coprocessor_owner[XCHAL_CP_MAX];
-extern void coprocessor_flush(struct thread_info*, int);
-
-extern void coprocessor_release_all(struct thread_info*);
-extern void coprocessor_flush_all(struct thread_info*);
+struct thread_info;
+void coprocessor_flush(struct thread_info *ti, int cp_index);
+void coprocessor_release_all(struct thread_info *ti);
+void coprocessor_flush_all(struct thread_info *ti);
+void coprocessor_flush_release_all(struct thread_info *ti);
+void local_coprocessors_flush_release_all(void);
 
 #endif	/* XTENSA_HAVE_COPROCESSORS */
 
diff --git a/arch/xtensa/include/asm/processor.h b/arch/xtensa/include/asm/processor.h
index 4489a27d527a..76bc63127c66 100644
--- a/arch/xtensa/include/asm/processor.h
+++ b/arch/xtensa/include/asm/processor.h
@@ -246,6 +246,13 @@ extern unsigned long __get_wchan(struct task_struct *p);
 	 v; \
 	 })
 
+#define xtensa_xsr(x, sr) \
+	({ \
+	 unsigned int __v__ = (unsigned int)(x); \
+	 __asm__ __volatile__ ("xsr %0, " __stringify(sr) : "+a"(__v__)); \
+	 __v__; \
+	 })
+
 #if XCHAL_HAVE_EXTERN_REGS
 
 static inline void set_er(unsigned long value, unsigned long addr)
diff --git a/arch/xtensa/include/asm/sections.h b/arch/xtensa/include/asm/sections.h
index a8c42d08e281..3bc6b9afa993 100644
--- a/arch/xtensa/include/asm/sections.h
+++ b/arch/xtensa/include/asm/sections.h
@@ -29,7 +29,7 @@ extern char _Level5InterruptVector_text_end[];
 extern char _Level6InterruptVector_text_start[];
 extern char _Level6InterruptVector_text_end[];
 #endif
-#ifdef CONFIG_SMP
+#ifdef CONFIG_SECONDARY_RESET_VECTOR
 extern char _SecondaryResetVector_text_start[];
 extern char _SecondaryResetVector_text_end[];
 #endif
diff --git a/arch/xtensa/include/asm/thread_info.h b/arch/xtensa/include/asm/thread_info.h
index f6fcbba1d02f..326db1c1d5d8 100644
--- a/arch/xtensa/include/asm/thread_info.h
+++ b/arch/xtensa/include/asm/thread_info.h
@@ -52,12 +52,21 @@ struct thread_info {
 	__u32			cpu;		/* current CPU */
 	__s32			preempt_count;	/* 0 => preemptable,< 0 => BUG*/
 
-	unsigned long		cpenable;
 #if XCHAL_HAVE_EXCLUSIVE
 	/* result of the most recent exclusive store */
 	unsigned long		atomctl8;
 #endif
+#ifdef CONFIG_USER_ABI_CALL0_PROBE
+	/* Address where PS.WOE was enabled by the ABI probing code */
+	unsigned long		ps_woe_fix_addr;
+#endif
 
+	/*
+	 * If i-th bit is set then coprocessor state is loaded into the
+	 * coprocessor i on CPU cp_owner_cpu.
+	 */
+	unsigned long		cpenable;
+	u32			cp_owner_cpu;
 	/* Allocate storage for extra user states and coprocessor states. */
 #if XTENSA_HAVE_COPROCESSORS
 	xtregs_coprocessor_t	xtregs_cp;
diff --git a/arch/xtensa/include/asm/traps.h b/arch/xtensa/include/asm/traps.h
index 6fa47cd8e02d..6f74ccc0c7ea 100644
--- a/arch/xtensa/include/asm/traps.h
+++ b/arch/xtensa/include/asm/traps.h
@@ -12,6 +12,8 @@
 
 #include <asm/ptrace.h>
 
+typedef void xtensa_exception_handler(struct pt_regs *regs);
+
 /*
  * Per-CPU exception handling data structure.
  * EXCSAVE1 points to it.
@@ -25,31 +27,47 @@ struct exc_table {
 	void *fixup;
 	/* For passing a parameter to fixup */
 	void *fixup_param;
+#if XTENSA_HAVE_COPROCESSORS
+	/* Pointers to owner struct thread_info */
+	struct thread_info *coprocessor_owner[XCHAL_CP_MAX];
+#endif
 	/* Fast user exception handlers */
 	void *fast_user_handler[EXCCAUSE_N];
 	/* Fast kernel exception handlers */
 	void *fast_kernel_handler[EXCCAUSE_N];
 	/* Default C-Handlers */
-	void *default_handler[EXCCAUSE_N];
+	xtensa_exception_handler *default_handler[EXCCAUSE_N];
 };
 
-/*
- * handler must be either of the following:
- *  void (*)(struct pt_regs *regs);
- *  void (*)(struct pt_regs *regs, unsigned long exccause);
- */
-extern void * __init trap_set_handler(int cause, void *handler);
-extern void do_unhandled(struct pt_regs *regs, unsigned long exccause);
-void fast_second_level_miss(void);
+DECLARE_PER_CPU(struct exc_table, exc_table);
+
+xtensa_exception_handler *
+__init trap_set_handler(int cause, xtensa_exception_handler *handler);
+
+asmlinkage void fast_illegal_instruction_user(void);
+asmlinkage void fast_syscall_user(void);
+asmlinkage void fast_alloca(void);
+asmlinkage void fast_unaligned(void);
+asmlinkage void fast_second_level_miss(void);
+asmlinkage void fast_store_prohibited(void);
+asmlinkage void fast_coprocessor(void);
+
+asmlinkage void kernel_exception(void);
+asmlinkage void user_exception(void);
+asmlinkage void system_call(struct pt_regs *regs);
+
+void do_IRQ(int hwirq, struct pt_regs *regs);
+void do_page_fault(struct pt_regs *regs);
+void do_unhandled(struct pt_regs *regs);
 
 /* Initialize minimal exc_table structure sufficient for basic paging */
 static inline void __init early_trap_init(void)
 {
-	static struct exc_table exc_table __initdata = {
+	static struct exc_table init_exc_table __initdata = {
 		.fast_kernel_handler[EXCCAUSE_DTLB_MISS] =
 			fast_second_level_miss,
 	};
-	__asm__ __volatile__("wsr  %0, excsave1\n" : : "a" (&exc_table));
+	xtensa_set_sr(&init_exc_table, excsave1);
 }
 
 void secondary_trap_init(void);
diff --git a/arch/xtensa/kernel/Makefile b/arch/xtensa/kernel/Makefile
index 5fd6cd15e0fb..897c1c741058 100644
--- a/arch/xtensa/kernel/Makefile
+++ b/arch/xtensa/kernel/Makefile
@@ -19,6 +19,7 @@ obj-$(CONFIG_XTENSA_VARIANT_HAVE_PERF_EVENTS) += perf_event.o
 obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o
 obj-$(CONFIG_S32C1I_SELFTEST) += s32c1i_selftest.o
 obj-$(CONFIG_JUMP_LABEL) += jump_label.o
+obj-$(CONFIG_HIBERNATION) += hibernate.o
 
 # In the Xtensa architecture, assembly generates literals which must always
 # precede the L32R instruction with a relative offset less than 256 kB.
diff --git a/arch/xtensa/kernel/asm-offsets.c b/arch/xtensa/kernel/asm-offsets.c
index 37278e2785fb..da38de20ae59 100644
--- a/arch/xtensa/kernel/asm-offsets.c
+++ b/arch/xtensa/kernel/asm-offsets.c
@@ -21,6 +21,7 @@
 #include <linux/ptrace.h>
 #include <linux/mm.h>
 #include <linux/kbuild.h>
+#include <linux/suspend.h>
 
 #include <asm/ptrace.h>
 #include <asm/traps.h>
@@ -87,14 +88,19 @@ int main(void)
 	OFFSET(TI_STSTUS, thread_info, status);
 	OFFSET(TI_CPU, thread_info, cpu);
 	OFFSET(TI_PRE_COUNT, thread_info, preempt_count);
+#ifdef CONFIG_USER_ABI_CALL0_PROBE
+	OFFSET(TI_PS_WOE_FIX_ADDR, thread_info, ps_woe_fix_addr);
+#endif
 
 	/* struct thread_info (offset from start_struct) */
 	DEFINE(THREAD_RA, offsetof (struct task_struct, thread.ra));
 	DEFINE(THREAD_SP, offsetof (struct task_struct, thread.sp));
-	DEFINE(THREAD_CPENABLE, offsetof (struct thread_info, cpenable));
 #if XCHAL_HAVE_EXCLUSIVE
 	DEFINE(THREAD_ATOMCTL8, offsetof (struct thread_info, atomctl8));
 #endif
+	DEFINE(THREAD_CPENABLE, offsetof(struct thread_info, cpenable));
+	DEFINE(THREAD_CPU, offsetof(struct thread_info, cpu));
+	DEFINE(THREAD_CP_OWNER_CPU, offsetof(struct thread_info, cp_owner_cpu));
 #if XTENSA_HAVE_COPROCESSORS
 	DEFINE(THREAD_XTREGS_CP0, offsetof(struct thread_info, xtregs_cp.cp0));
 	DEFINE(THREAD_XTREGS_CP1, offsetof(struct thread_info, xtregs_cp.cp1));
@@ -137,11 +143,22 @@ int main(void)
 	DEFINE(EXC_TABLE_DOUBLE_SAVE, offsetof(struct exc_table, double_save));
 	DEFINE(EXC_TABLE_FIXUP, offsetof(struct exc_table, fixup));
 	DEFINE(EXC_TABLE_PARAM, offsetof(struct exc_table, fixup_param));
+#if XTENSA_HAVE_COPROCESSORS
+	DEFINE(EXC_TABLE_COPROCESSOR_OWNER,
+	       offsetof(struct exc_table, coprocessor_owner));
+#endif
 	DEFINE(EXC_TABLE_FAST_USER,
 	       offsetof(struct exc_table, fast_user_handler));
 	DEFINE(EXC_TABLE_FAST_KERNEL,
 	       offsetof(struct exc_table, fast_kernel_handler));
 	DEFINE(EXC_TABLE_DEFAULT, offsetof(struct exc_table, default_handler));
 
+#ifdef CONFIG_HIBERNATION
+	DEFINE(PBE_ADDRESS, offsetof(struct pbe, address));
+	DEFINE(PBE_ORIG_ADDRESS, offsetof(struct pbe, orig_address));
+	DEFINE(PBE_NEXT, offsetof(struct pbe, next));
+	DEFINE(PBE_SIZE, sizeof(struct pbe));
+#endif
+
 	return 0;
 }
diff --git a/arch/xtensa/kernel/coprocessor.S b/arch/xtensa/kernel/coprocessor.S
index c7b9f12896f2..ef33e76e07d8 100644
--- a/arch/xtensa/kernel/coprocessor.S
+++ b/arch/xtensa/kernel/coprocessor.S
@@ -19,6 +19,26 @@
 #include <asm/current.h>
 #include <asm/regs.h>
 
+/*
+ * Rules for coprocessor state manipulation on SMP:
+ *
+ * - a task may have live coprocessors only on one CPU.
+ *
+ * - whether coprocessor context of task T is live on some CPU is
+ *   denoted by T's thread_info->cpenable.
+ *
+ * - non-zero thread_info->cpenable means that thread_info->cp_owner_cpu
+ *   is valid in the T's thread_info. Zero thread_info->cpenable means that
+ *   coprocessor context is valid in the T's thread_info.
+ *
+ * - if a coprocessor context of task T is live on CPU X, only CPU X changes
+ *   T's thread_info->cpenable, cp_owner_cpu and coprocessor save area.
+ *   This is done by making sure that for the task T with live coprocessor
+ *   on CPU X cpenable SR is 0 when T runs on any other CPU Y.
+ *   When fast_coprocessor exception is taken on CPU Y it goes to the
+ *   C-level do_coprocessor that uses IPI to make CPU X flush T's coprocessors.
+ */
+
 #if XTENSA_HAVE_COPROCESSORS
 
 /*
@@ -30,34 +50,30 @@
 		.align 4;						\
 	.Lsave_cp_regs_cp##x:						\
 		xchal_cp##x##_store a2 a3 a4 a5 a6;			\
-		jx	a0;						\
+		ret;							\
 	.endif
 
-#define SAVE_CP_REGS_TAB(x)						\
-	.if XTENSA_HAVE_COPROCESSOR(x);					\
-		.long .Lsave_cp_regs_cp##x;				\
-	.else;								\
-		.long 0;						\
-	.endif;								\
-	.long THREAD_XTREGS_CP##x
-
-
 #define LOAD_CP_REGS(x)							\
 	.if XTENSA_HAVE_COPROCESSOR(x);					\
 		.align 4;						\
 	.Lload_cp_regs_cp##x:						\
 		xchal_cp##x##_load a2 a3 a4 a5 a6;			\
-		jx	a0;						\
+		ret;							\
 	.endif
 
-#define LOAD_CP_REGS_TAB(x)						\
+#define CP_REGS_TAB(x)							\
 	.if XTENSA_HAVE_COPROCESSOR(x);					\
+		.long .Lsave_cp_regs_cp##x;				\
 		.long .Lload_cp_regs_cp##x;				\
 	.else;								\
-		.long 0;						\
+		.long 0, 0;						\
 	.endif;								\
 	.long THREAD_XTREGS_CP##x
 
+#define CP_REGS_TAB_SAVE 0
+#define CP_REGS_TAB_LOAD 4
+#define CP_REGS_TAB_OFFSET 8
+
 	__XTENSA_HANDLER
 
 	SAVE_CP_REGS(0)
@@ -79,25 +95,15 @@
 	LOAD_CP_REGS(7)
 
 	.align 4
-.Lsave_cp_regs_jump_table:
-	SAVE_CP_REGS_TAB(0)
-	SAVE_CP_REGS_TAB(1)
-	SAVE_CP_REGS_TAB(2)
-	SAVE_CP_REGS_TAB(3)
-	SAVE_CP_REGS_TAB(4)
-	SAVE_CP_REGS_TAB(5)
-	SAVE_CP_REGS_TAB(6)
-	SAVE_CP_REGS_TAB(7)
-
-.Lload_cp_regs_jump_table:
-	LOAD_CP_REGS_TAB(0)
-	LOAD_CP_REGS_TAB(1)
-	LOAD_CP_REGS_TAB(2)
-	LOAD_CP_REGS_TAB(3)
-	LOAD_CP_REGS_TAB(4)
-	LOAD_CP_REGS_TAB(5)
-	LOAD_CP_REGS_TAB(6)
-	LOAD_CP_REGS_TAB(7)
+.Lcp_regs_jump_table:
+	CP_REGS_TAB(0)
+	CP_REGS_TAB(1)
+	CP_REGS_TAB(2)
+	CP_REGS_TAB(3)
+	CP_REGS_TAB(4)
+	CP_REGS_TAB(5)
+	CP_REGS_TAB(6)
+	CP_REGS_TAB(7)
 
 /*
  * Entry condition:
@@ -115,9 +121,37 @@
 
 ENTRY(fast_coprocessor)
 
+	s32i	a3, a2, PT_AREG3
+
+#ifdef CONFIG_SMP
+	/*
+	 * Check if any coprocessor context is live on another CPU
+	 * and if so go through the C-level coprocessor exception handler
+	 * to flush it to memory.
+	 */
+	GET_THREAD_INFO (a0, a2)
+	l32i	a3, a0, THREAD_CPENABLE
+	beqz	a3, .Lload_local
+
+	/*
+	 * Pairs with smp_wmb in local_coprocessor_release_all
+	 * and with both memws below.
+	 */
+	memw
+	l32i	a3, a0, THREAD_CPU
+	l32i	a0, a0, THREAD_CP_OWNER_CPU
+	beq	a0, a3, .Lload_local
+
+	rsr	a0, ps
+	l32i	a3, a2, PT_AREG3
+	bbci.l	a0, PS_UM_BIT, 1f
+	call0	user_exception
+1:	call0	kernel_exception
+#endif
+
 	/* Save remaining registers a1-a3 and SAR */
 
-	s32i	a3, a2, PT_AREG3
+.Lload_local:
 	rsr	a3, sar
 	s32i	a1, a2, PT_AREG1
 	s32i	a3, a2, PT_SAR
@@ -125,13 +159,15 @@ ENTRY(fast_coprocessor)
 	rsr	a2, depc
 	s32i	a2, a1, PT_AREG2
 
-	/*
-	 * The hal macros require up to 4 temporary registers. We use a3..a6.
-	 */
+	/* The hal macros require up to 4 temporary registers. We use a3..a6. */
 
 	s32i	a4, a1, PT_AREG4
 	s32i	a5, a1, PT_AREG5
 	s32i	a6, a1, PT_AREG6
+	s32i	a7, a1, PT_AREG7
+	s32i	a8, a1, PT_AREG8
+	s32i	a9, a1, PT_AREG9
+	s32i	a10, a1, PT_AREG10
 
 	/* Find coprocessor number. Subtract first CP EXCCAUSE from EXCCAUSE */
 
@@ -148,58 +184,74 @@ ENTRY(fast_coprocessor)
 	wsr	a0, cpenable
 	rsync
 
-	/* Retrieve previous owner. (a3 still holds CP number) */
+	/* Get coprocessor save/load table entry (a7). */
 
-	movi	a0, coprocessor_owner	# list of owners
-	addx4	a0, a3, a0		# entry for CP
-	l32i	a4, a0, 0
+	movi	a7, .Lcp_regs_jump_table
+	addx8	a7, a3, a7
+	addx4	a7, a3, a7
 
-	beqz	a4, 1f			# skip 'save' if no previous owner
+	/* Retrieve previous owner (a8). */
 
-	/* Disable coprocessor for previous owner. (a2 = 1 << CP number) */
+	rsr	a0, excsave1		# exc_table
+	addx4	a0, a3, a0		# entry for CP
+	l32i	a8, a0, EXC_TABLE_COPROCESSOR_OWNER
+
+	/* Set new owner (a9). */
 
-	l32i	a5, a4, THREAD_CPENABLE
-	xor	a5, a5, a2		# (1 << cp-id) still in a2
-	s32i	a5, a4, THREAD_CPENABLE
+	GET_THREAD_INFO (a9, a1)
+	l32i	a4, a9, THREAD_CPU
+	s32i	a9, a0, EXC_TABLE_COPROCESSOR_OWNER
+	s32i	a4, a9, THREAD_CP_OWNER_CPU
 
 	/*
-	 * Get context save area and 'call' save routine. 
-	 * (a4 still holds previous owner (thread_info), a3 CP number)
+	 * Enable coprocessor for the new owner. (a2 = 1 << CP number)
+	 * This can be done before loading context into the coprocessor.
 	 */
+	l32i	a4, a9, THREAD_CPENABLE
+	or	a4, a4, a2
 
-	movi	a5, .Lsave_cp_regs_jump_table
-	movi	a0, 2f			# a0: 'return' address
-	addx8	a3, a3, a5		# a3: coprocessor number
-	l32i	a2, a3, 4		# a2: xtregs offset
-	l32i	a3, a3, 0		# a3: jump address
-	add	a2, a2, a4
-	jx	a3
+	/*
+	 * Make sure THREAD_CP_OWNER_CPU is in memory before updating
+	 * THREAD_CPENABLE
+	 */
+	memw				# (2)
+	s32i	a4, a9, THREAD_CPENABLE
 
-	/* Note that only a0 and a1 were preserved. */
+	beqz	a8, 1f			# skip 'save' if no previous owner
 
-2:	rsr	a3, exccause
-	addi	a3, a3, -EXCCAUSE_COPROCESSOR0_DISABLED
-	movi	a0, coprocessor_owner
-	addx4	a0, a3, a0
+	/* Disable coprocessor for previous owner. (a2 = 1 << CP number) */
 
-	/* Set new 'owner' (a0 points to the CP owner, a3 contains the CP nr) */
+	l32i	a10, a8, THREAD_CPENABLE
+	xor	a10, a10, a2
 
-1:	GET_THREAD_INFO (a4, a1)
-	s32i	a4, a0, 0
+	/* Get context save area and call save routine. */
 
-	/* Get context save area and 'call' load routine. */
+	l32i	a2, a7, CP_REGS_TAB_OFFSET
+	l32i	a3, a7, CP_REGS_TAB_SAVE
+	add	a2, a2, a8
+	callx0	a3
 
-	movi	a5, .Lload_cp_regs_jump_table
-	movi	a0, 1f
-	addx8	a3, a3, a5
-	l32i	a2, a3, 4		# a2: xtregs offset
-	l32i	a3, a3, 0		# a3: jump address
-	add	a2, a2, a4
-	jx	a3
+	/*
+	 * Make sure coprocessor context and THREAD_CP_OWNER_CPU are in memory
+	 * before updating THREAD_CPENABLE
+	 */
+	memw				# (3)
+	s32i	a10, a8, THREAD_CPENABLE
+1:
+	/* Get context save area and call load routine. */
+
+	l32i	a2, a7, CP_REGS_TAB_OFFSET
+	l32i	a3, a7, CP_REGS_TAB_LOAD
+	add	a2, a2, a9
+	callx0	a3
 
 	/* Restore all registers and return from exception handler. */
 
-1:	l32i	a6, a1, PT_AREG6
+	l32i	a10, a1, PT_AREG10
+	l32i	a9, a1, PT_AREG9
+	l32i	a8, a1, PT_AREG8
+	l32i	a7, a1, PT_AREG7
+	l32i	a6, a1, PT_AREG6
 	l32i	a5, a1, PT_AREG5
 	l32i	a4, a1, PT_AREG4
 
@@ -230,29 +282,21 @@ ENDPROC(fast_coprocessor)
 
 ENTRY(coprocessor_flush)
 
-	/* reserve 4 bytes on stack to save a0 */
-	abi_entry(4)
-
-	s32i	a0, a1, 0
-	movi	a0, .Lsave_cp_regs_jump_table
-	addx8	a3, a3, a0
-	l32i	a4, a3, 4
-	l32i	a3, a3, 0
-	add	a2, a2, a4
-	beqz	a3, 1f
-	callx0	a3
-1:	l32i	a0, a1, 0
-
-	abi_ret(4)
+	abi_entry_default
+
+	movi	a4, .Lcp_regs_jump_table
+	addx8	a4, a3, a4
+	addx4	a3, a3, a4
+	l32i	a4, a3, CP_REGS_TAB_SAVE
+	beqz	a4, 1f
+	l32i	a3, a3, CP_REGS_TAB_OFFSET
+	add	a2, a2, a3
+	mov	a7, a0
+	callx0	a4
+	mov	a0, a7
+1:
+	abi_ret_default
 
 ENDPROC(coprocessor_flush)
 
-	.data
-
-ENTRY(coprocessor_owner)
-
-	.fill XCHAL_CP_MAX, 4, 0
-
-END(coprocessor_owner)
-
 #endif /* XTENSA_HAVE_COPROCESSORS */
diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index 6b6eff658795..e3eae648ba2e 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -28,15 +28,6 @@
 #include <asm/tlbflush.h>
 #include <variant/tie-asm.h>
 
-/* Unimplemented features. */
-
-#undef KERNEL_STACK_OVERFLOW_CHECK
-
-/* Not well tested.
- *
- * - fast_coprocessor
- */
-
 /*
  * Macro to find first bit set in WINDOWBASE from the left + 1
  *
@@ -178,28 +169,26 @@ _user_exception:
 
 	/* Save only live registers. */
 
-UABI_W	_bbsi.l	a2, 1, 1f
+UABI_W	_bbsi.l	a2, 1, .Lsave_window_registers
 	s32i	a4, a1, PT_AREG4
 	s32i	a5, a1, PT_AREG5
 	s32i	a6, a1, PT_AREG6
 	s32i	a7, a1, PT_AREG7
-UABI_W	_bbsi.l	a2, 2, 1f
+UABI_W	_bbsi.l	a2, 2, .Lsave_window_registers
 	s32i	a8, a1, PT_AREG8
 	s32i	a9, a1, PT_AREG9
 	s32i	a10, a1, PT_AREG10
 	s32i	a11, a1, PT_AREG11
-UABI_W	_bbsi.l	a2, 3, 1f
+UABI_W	_bbsi.l	a2, 3, .Lsave_window_registers
 	s32i	a12, a1, PT_AREG12
 	s32i	a13, a1, PT_AREG13
 	s32i	a14, a1, PT_AREG14
 	s32i	a15, a1, PT_AREG15
 
 #if defined(USER_SUPPORT_WINDOWED)
-	_bnei	a2, 1, 1f		# only one valid frame?
+	/* If only one valid frame skip saving regs. */
 
-	/* Only one valid frame, skip saving regs. */
-
-	j	2f
+	beqi	a2, 1, common_exception
 
 	/* Save the remaining registers.
 	 * We have to save all registers up to the first '1' from
@@ -208,8 +197,8 @@ UABI_W	_bbsi.l	a2, 3, 1f
 	 * All register frames starting from the top field to the marked '1'
 	 * must be saved.
 	 */
-
-1:	addi	a3, a2, -1		# eliminate '1' in bit 0: yyyyxxww0
+.Lsave_window_registers:
+	addi	a3, a2, -1		# eliminate '1' in bit 0: yyyyxxww0
 	neg	a3, a3			# yyyyxxww0 -> YYYYXXWW1+1
 	and	a3, a3, a2		# max. only one bit is set
 
@@ -250,7 +239,7 @@ UABI_W	_bbsi.l	a2, 3, 1f
 
 	/* We are back to the original stack pointer (a1) */
 #endif
-2:	/* Now, jump to the common exception handler. */
+	/* Now, jump to the common exception handler. */
 
 	j	common_exception
 
@@ -350,15 +339,6 @@ KABI_W	_bbsi.l	a2, 3, 1f
 	l32i	a0, a1, PT_AREG0	# restore saved a0
 	wsr	a0, depc
 
-#ifdef KERNEL_STACK_OVERFLOW_CHECK
-
-	/*  Stack overflow check, for debugging  */
-	extui	a2, a1, TASK_SIZE_BITS,XX
-	movi	a3, SIZE??
-	_bge	a2, a3, out_of_stack_panic
-
-#endif
-
 /*
  * This is the common exception handler.
  * We get here from the user exception handler or simply by falling through
@@ -442,7 +422,6 @@ KABI_W	or	a3, a3, a0
 	moveqz	a3, a0, a2		# a3 = LOCKLEVEL iff interrupt
 KABI_W	movi	a2, PS_WOE_MASK
 KABI_W	or	a3, a3, a2
-	rsr	a2, exccause
 #endif
 
 	/* restore return address (or 0 if return to userspace) */
@@ -469,42 +448,56 @@ KABI_W	or	a3, a3, a2
 
 	save_xtregs_opt a1 a3 a4 a5 a6 a7 PT_XTREGS_OPT
 	
+#ifdef CONFIG_TRACE_IRQFLAGS
+	rsr		abi_tmp0, ps
+	extui		abi_tmp0, abi_tmp0, PS_INTLEVEL_SHIFT, PS_INTLEVEL_WIDTH
+	beqz		abi_tmp0, 1f
+	abi_call	trace_hardirqs_off
+1:
+#endif
+#ifdef CONFIG_CONTEXT_TRACKING
+	l32i		abi_tmp0, a1, PT_PS
+	bbci.l		abi_tmp0, PS_UM_BIT, 1f
+	abi_call	context_tracking_user_exit
+1:
+#endif
+
 	/* Go to second-level dispatcher. Set up parameters to pass to the
 	 * exception handler and call the exception handler.
 	 */
 
-	rsr	a4, excsave1
-	addx4	a4, a2, a4
-	l32i	a4, a4, EXC_TABLE_DEFAULT		# load handler
-	mov	abi_arg1, a2			# pass EXCCAUSE
-	mov	abi_arg0, a1			# pass stack frame
+	l32i		abi_arg1, a1, PT_EXCCAUSE	# pass EXCCAUSE
+	rsr		abi_tmp0, excsave1
+	addx4		abi_tmp0, abi_arg1, abi_tmp0
+	l32i		abi_tmp0, abi_tmp0, EXC_TABLE_DEFAULT	# load handler
+	mov		abi_arg0, a1			# pass stack frame
 
 	/* Call the second-level handler */
 
-	abi_callx	a4
+	abi_callx	abi_tmp0
 
 	/* Jump here for exception exit */
 	.global common_exception_return
 common_exception_return:
 
 #if XTENSA_FAKE_NMI
-	l32i	abi_tmp0, a1, PT_EXCCAUSE
-	movi	abi_tmp1, EXCCAUSE_MAPPED_NMI
-	l32i	abi_saved1, a1, PT_PS
-	beq	abi_tmp0, abi_tmp1, .Lrestore_state
+	l32i		abi_tmp0, a1, PT_EXCCAUSE
+	movi		abi_tmp1, EXCCAUSE_MAPPED_NMI
+	l32i		abi_saved1, a1, PT_PS
+	beq		abi_tmp0, abi_tmp1, .Lrestore_state
 #endif
 .Ltif_loop:
-	irq_save a2, a3
+	irq_save	abi_tmp0, abi_tmp1
 #ifdef CONFIG_TRACE_IRQFLAGS
 	abi_call	trace_hardirqs_off
 #endif
 
 	/* Jump if we are returning from kernel exceptions. */
 
-	l32i	abi_saved1, a1, PT_PS
-	GET_THREAD_INFO(a2, a1)
-	l32i	a4, a2, TI_FLAGS
-	_bbci.l	abi_saved1, PS_UM_BIT, .Lexit_tif_loop_kernel
+	l32i		abi_saved1, a1, PT_PS
+	GET_THREAD_INFO(abi_tmp0, a1)
+	l32i		abi_saved0, abi_tmp0, TI_FLAGS
+	_bbci.l		abi_saved1, PS_UM_BIT, .Lexit_tif_loop_kernel
 
 	/* Specific to a user exception exit:
 	 * We need to check some flags for signal handling and rescheduling,
@@ -513,75 +506,80 @@ common_exception_return:
 	 * Note that we don't disable interrupts here. 
 	 */
 
-	_bbsi.l	a4, TIF_NEED_RESCHED, .Lresched
-	movi	a2, _TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NOTIFY_SIGNAL
-	bnone	a4, a2, .Lexit_tif_loop_user
+	_bbsi.l		abi_saved0, TIF_NEED_RESCHED, .Lresched
+	movi		abi_tmp0, _TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NOTIFY_SIGNAL
+	bnone		abi_saved0, abi_tmp0, .Lexit_tif_loop_user
 
-	l32i	a4, a1, PT_DEPC
-	bgeui	a4, VALID_DOUBLE_EXCEPTION_ADDRESS, .Lrestore_state
+	l32i		abi_tmp0, a1, PT_DEPC
+	bgeui		abi_tmp0, VALID_DOUBLE_EXCEPTION_ADDRESS, .Lrestore_state
 
 	/* Call do_signal() */
 
 #ifdef CONFIG_TRACE_IRQFLAGS
 	abi_call	trace_hardirqs_on
 #endif
-	rsil	a2, 0
-	mov	abi_arg0, a1
+	rsil		abi_tmp0, 0
+	mov		abi_arg0, a1
 	abi_call	do_notify_resume	# int do_notify_resume(struct pt_regs*)
-	j	.Ltif_loop
+	j		.Ltif_loop
 
 .Lresched:
 #ifdef CONFIG_TRACE_IRQFLAGS
 	abi_call	trace_hardirqs_on
 #endif
-	rsil	a2, 0
+	rsil		abi_tmp0, 0
 	abi_call	schedule	# void schedule (void)
-	j	.Ltif_loop
+	j		.Ltif_loop
 
 .Lexit_tif_loop_kernel:
 #ifdef CONFIG_PREEMPTION
-	_bbci.l	a4, TIF_NEED_RESCHED, .Lrestore_state
+	_bbci.l		abi_saved0, TIF_NEED_RESCHED, .Lrestore_state
 
 	/* Check current_thread_info->preempt_count */
 
-	l32i	a4, a2, TI_PRE_COUNT
-	bnez	a4, .Lrestore_state
+	l32i		abi_tmp1, abi_tmp0, TI_PRE_COUNT
+	bnez		abi_tmp1, .Lrestore_state
 	abi_call	preempt_schedule_irq
 #endif
-	j	.Lrestore_state
+	j		.Lrestore_state
 
 .Lexit_tif_loop_user:
+#ifdef CONFIG_CONTEXT_TRACKING
+	abi_call	context_tracking_user_enter
+#endif
 #ifdef CONFIG_HAVE_HW_BREAKPOINT
-	_bbci.l	a4, TIF_DB_DISABLED, 1f
+	_bbci.l		abi_saved0, TIF_DB_DISABLED, 1f
 	abi_call	restore_dbreak
 1:
 #endif
 #ifdef CONFIG_DEBUG_TLB_SANITY
-	l32i	a4, a1, PT_DEPC
-	bgeui	a4, VALID_DOUBLE_EXCEPTION_ADDRESS, .Lrestore_state
+	l32i		abi_tmp0, a1, PT_DEPC
+	bgeui		abi_tmp0, VALID_DOUBLE_EXCEPTION_ADDRESS, .Lrestore_state
 	abi_call	check_tlb_sanity
 #endif
 
 .Lrestore_state:
 #ifdef CONFIG_TRACE_IRQFLAGS
-	extui	a4, abi_saved1, PS_INTLEVEL_SHIFT, PS_INTLEVEL_WIDTH
-	bgei	a4, LOCKLEVEL, 1f
+	extui		abi_tmp0, abi_saved1, PS_INTLEVEL_SHIFT, PS_INTLEVEL_WIDTH
+	bgei		abi_tmp0, LOCKLEVEL, 1f
 	abi_call	trace_hardirqs_on
 1:
 #endif
-	/* Restore optional registers. */
+	/*
+	 * Restore optional registers.
+	 * abi_arg* are used as temporary registers here.
+	 */
 
-	load_xtregs_opt a1 a2 a4 a5 a6 a7 PT_XTREGS_OPT
+	load_xtregs_opt a1 abi_tmp0 abi_arg0 abi_arg1 abi_arg2 abi_arg3 PT_XTREGS_OPT
 
 	/* Restore SCOMPARE1 */
 
 #if XCHAL_HAVE_S32C1I
-	l32i    a2, a1, PT_SCOMPARE1
-	wsr     a2, scompare1
+	l32i		abi_tmp0, a1, PT_SCOMPARE1
+	wsr		abi_tmp0, scompare1
 #endif
-	wsr	abi_saved1, ps		/* disable interrupts */
-
-	_bbci.l	abi_saved1, PS_UM_BIT, kernel_exception_exit
+	wsr		abi_saved1, ps		/* disable interrupts */
+	_bbci.l		abi_saved1, PS_UM_BIT, kernel_exception_exit
 
 user_exception_exit:
 
@@ -795,7 +793,7 @@ ENDPROC(kernel_exception)
 ENTRY(debug_exception)
 
 	rsr	a0, SREG_EPS + XCHAL_DEBUGLEVEL
-	bbsi.l	a0, PS_EXCM_BIT, 1f	# exception mode
+	bbsi.l	a0, PS_EXCM_BIT, .Ldebug_exception_in_exception	# exception mode
 
 	/* Set EPC1 and EXCCAUSE */
 
@@ -814,10 +812,10 @@ ENTRY(debug_exception)
 
 	/* Switch to kernel/user stack, restore jump vector, and save a0 */
 
-	bbsi.l	a2, PS_UM_BIT, 2f	# jump if user mode
-
+	bbsi.l	a2, PS_UM_BIT, .Ldebug_exception_user	# jump if user mode
 	addi	a2, a1, -16 - PT_KERNEL_SIZE	# assume kernel stack
-3:
+
+.Ldebug_exception_continue:
 	l32i	a0, a3, DT_DEBUG_SAVE
 	s32i	a1, a2, PT_AREG1
 	s32i	a0, a2, PT_AREG0
@@ -845,10 +843,12 @@ ENTRY(debug_exception)
 	bbsi.l	a2, PS_UM_BIT, _user_exception
 	j	_kernel_exception
 
-2:	rsr	a2, excsave1
+.Ldebug_exception_user:
+	rsr	a2, excsave1
 	l32i	a2, a2, EXC_TABLE_KSTK	# load kernel stack pointer
-	j	3b
+	j	.Ldebug_exception_continue
 
+.Ldebug_exception_in_exception:
 #ifdef CONFIG_HAVE_HW_BREAKPOINT
 	/* Debug exception while in exception mode. This may happen when
 	 * window overflow/underflow handler or fast exception handler hits
@@ -856,8 +856,8 @@ ENTRY(debug_exception)
 	 * breakpoints, single-step faulting instruction and restore data
 	 * breakpoints.
 	 */
-1:
-	bbci.l	a0, PS_UM_BIT, 1b	# jump if kernel mode
+
+	bbci.l	a0, PS_UM_BIT, .Ldebug_exception_in_exception	# jump if kernel mode
 
 	rsr	a0, debugcause
 	bbsi.l	a0, DEBUGCAUSE_DBREAK_BIT, .Ldebug_save_dbreak
@@ -901,7 +901,7 @@ ENTRY(debug_exception)
 	rfi	XCHAL_DEBUGLEVEL
 #else
 	/* Debug exception while in exception mode. Should not happen. */
-1:	j	1b	// FIXME!!
+	j	.Ldebug_exception_in_exception	// FIXME!!
 #endif
 
 ENDPROC(debug_exception)
@@ -1056,6 +1056,11 @@ ENTRY(fast_illegal_instruction_user)
 	movi	a3, PS_WOE_MASK
 	or	a0, a0, a3
 	wsr	a0, ps
+#ifdef CONFIG_USER_ABI_CALL0_PROBE
+	GET_THREAD_INFO(a3, a2)
+	rsr	a0, epc1
+	s32i	a0, a3, TI_PS_WOE_FIX_ADDR
+#endif
 	l32i	a3, a2, PT_AREG3
 	l32i	a0, a2, PT_AREG0
 	rsr	a2, depc
@@ -1630,12 +1635,13 @@ ENTRY(fast_second_level_miss)
 
 	GET_CURRENT(a1,a2)
 	l32i	a0, a1, TASK_MM		# tsk->mm
-	beqz	a0, 9f
+	beqz	a0, .Lfast_second_level_miss_no_mm
 
-8:	rsr	a3, excvaddr		# fault address
+.Lfast_second_level_miss_continue:
+	rsr	a3, excvaddr		# fault address
 	_PGD_OFFSET(a0, a3, a1)
 	l32i	a0, a0, 0		# read pmdval
-	beqz	a0, 2f
+	beqz	a0, .Lfast_second_level_miss_no_pmd
 
 	/* Read ptevaddr and convert to top of page-table page.
 	 *
@@ -1678,12 +1684,13 @@ ENTRY(fast_second_level_miss)
 	addi	a3, a3, DTLB_WAY_PGD
 	add	a1, a1, a3		# ... + way_number
 
-3:	wdtlb	a0, a1
+.Lfast_second_level_miss_wdtlb:
+	wdtlb	a0, a1
 	dsync
 
 	/* Exit critical section. */
-
-4:	rsr	a3, excsave1
+.Lfast_second_level_miss_skip_wdtlb:
+	rsr	a3, excsave1
 	movi	a0, 0
 	s32i	a0, a3, EXC_TABLE_FIXUP
 
@@ -1707,19 +1714,21 @@ ENTRY(fast_second_level_miss)
 	esync
 	rfde
 
-9:	l32i	a0, a1, TASK_ACTIVE_MM	# unlikely case mm == 0
-	bnez	a0, 8b
+.Lfast_second_level_miss_no_mm:
+	l32i	a0, a1, TASK_ACTIVE_MM	# unlikely case mm == 0
+	bnez	a0, .Lfast_second_level_miss_continue
 
 	/* Even more unlikely case active_mm == 0.
 	 * We can get here with NMI in the middle of context_switch that
 	 * touches vmalloc area.
 	 */
 	movi	a0, init_mm
-	j	8b
+	j	.Lfast_second_level_miss_continue
 
+.Lfast_second_level_miss_no_pmd:
 #if (DCACHE_WAY_SIZE > PAGE_SIZE)
 
-2:	/* Special case for cache aliasing.
+	/* Special case for cache aliasing.
 	 * We (should) only get here if a clear_user_page, copy_user_page
 	 * or the aliased cache flush functions got preemptively interrupted 
 	 * by another task. Re-establish temporary mapping to the 
@@ -1729,24 +1738,24 @@ ENTRY(fast_second_level_miss)
 	/* We shouldn't be in a double exception */
 
 	l32i	a0, a2, PT_DEPC
-	bgeui	a0, VALID_DOUBLE_EXCEPTION_ADDRESS, 2f
+	bgeui	a0, VALID_DOUBLE_EXCEPTION_ADDRESS, .Lfast_second_level_miss_slow
 
 	/* Make sure the exception originated in the special functions */
 
 	movi	a0, __tlbtemp_mapping_start
 	rsr	a3, epc1
-	bltu	a3, a0, 2f
+	bltu	a3, a0, .Lfast_second_level_miss_slow
 	movi	a0, __tlbtemp_mapping_end
-	bgeu	a3, a0, 2f
+	bgeu	a3, a0, .Lfast_second_level_miss_slow
 
 	/* Check if excvaddr was in one of the TLBTEMP_BASE areas. */
 
 	movi	a3, TLBTEMP_BASE_1
 	rsr	a0, excvaddr
-	bltu	a0, a3, 2f
+	bltu	a0, a3, .Lfast_second_level_miss_slow
 
 	addi	a1, a0, -TLBTEMP_SIZE
-	bgeu	a1, a3, 2f
+	bgeu	a1, a3, .Lfast_second_level_miss_slow
 
 	/* Check if we have to restore an ITLB mapping. */
 
@@ -1772,19 +1781,19 @@ ENTRY(fast_second_level_miss)
 
 	mov	a0, a6
 	movnez	a0, a7, a3
-	j	3b
+	j	.Lfast_second_level_miss_wdtlb
 
 	/* ITLB entry. We only use dst in a6. */
 
 1:	witlb	a6, a1
 	isync
-	j	4b
+	j	.Lfast_second_level_miss_skip_wdtlb
 
 
 #endif	// DCACHE_WAY_SIZE > PAGE_SIZE
 
-
-2:	/* Invalid PGD, default exception handling */
+	/* Invalid PGD, default exception handling */
+.Lfast_second_level_miss_slow:
 
 	rsr	a1, depc
 	s32i	a1, a2, PT_AREG2
@@ -1824,12 +1833,13 @@ ENTRY(fast_store_prohibited)
 
 	GET_CURRENT(a1,a2)
 	l32i	a0, a1, TASK_MM		# tsk->mm
-	beqz	a0, 9f
+	beqz	a0, .Lfast_store_no_mm
 
-8:	rsr	a1, excvaddr		# fault address
+.Lfast_store_continue:
+	rsr	a1, excvaddr		# fault address
 	_PGD_OFFSET(a0, a1, a3)
 	l32i	a0, a0, 0
-	beqz	a0, 2f
+	beqz	a0, .Lfast_store_slow
 
 	/*
 	 * Note that we test _PAGE_WRITABLE_BIT only if PTE is present
@@ -1839,8 +1849,8 @@ ENTRY(fast_store_prohibited)
 	_PTE_OFFSET(a0, a1, a3)
 	l32i	a3, a0, 0		# read pteval
 	movi	a1, _PAGE_CA_INVALID
-	ball	a3, a1, 2f
-	bbci.l	a3, _PAGE_WRITABLE_BIT, 2f
+	ball	a3, a1, .Lfast_store_slow
+	bbci.l	a3, _PAGE_WRITABLE_BIT, .Lfast_store_slow
 
 	movi	a1, _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_HW_WRITE
 	or	a3, a3, a1
@@ -1868,7 +1878,6 @@ ENTRY(fast_store_prohibited)
 	l32i	a2, a2, PT_DEPC
 
 	bgeui	a2, VALID_DOUBLE_EXCEPTION_ADDRESS, 1f
-
 	rsr	a2, depc
 	rfe
 
@@ -1878,11 +1887,17 @@ ENTRY(fast_store_prohibited)
 	esync
 	rfde
 
-9:	l32i	a0, a1, TASK_ACTIVE_MM	# unlikely case mm == 0
-	j	8b
-
-2:	/* If there was a problem, handle fault in C */
+.Lfast_store_no_mm:
+	l32i	a0, a1, TASK_ACTIVE_MM	# unlikely case mm == 0
+	j	.Lfast_store_continue
 
+	/* If there was a problem, handle fault in C */
+.Lfast_store_slow:
+	rsr	a1, excvaddr
+	pdtlb	a0, a1
+	bbci.l	a0, DTLB_HIT_BIT, 1f
+	idtlb	a0
+1:
 	rsr	a3, depc	# still holds a2
 	s32i	a3, a2, PT_AREG2
 	mov	a1, a2
@@ -2071,8 +2086,16 @@ ENTRY(_switch_to)
 
 #if (XTENSA_HAVE_COPROCESSORS || XTENSA_HAVE_IO_PORTS)
 	l32i	a3, a5, THREAD_CPENABLE
-	xsr	a3, cpenable
-	s32i	a3, a4, THREAD_CPENABLE
+#ifdef CONFIG_SMP
+	beqz	a3, 1f
+	memw			# pairs with memw (2) in fast_coprocessor
+	l32i	a6, a5, THREAD_CP_OWNER_CPU
+	l32i	a7, a5, THREAD_CPU
+	beq	a6, a7, 1f	# load 0 into CPENABLE if current CPU is not the owner
+	movi	a3, 0
+1:
+#endif
+	wsr	a3, cpenable
 #endif
 
 #if XCHAL_HAVE_EXCLUSIVE
@@ -2147,3 +2170,95 @@ ENTRY(ret_from_kernel_thread)
 	j		common_exception_return
 
 ENDPROC(ret_from_kernel_thread)
+
+#ifdef CONFIG_HIBERNATION
+
+	.bss
+	.align	4
+.Lsaved_regs:
+#if defined(__XTENSA_WINDOWED_ABI__)
+	.fill	2, 4
+#elif defined(__XTENSA_CALL0_ABI__)
+	.fill	6, 4
+#else
+#error Unsupported Xtensa ABI
+#endif
+	.align	XCHAL_NCP_SA_ALIGN
+.Lsaved_user_regs:
+	.fill	XTREGS_USER_SIZE, 1
+
+	.previous
+
+ENTRY(swsusp_arch_suspend)
+
+	abi_entry_default
+
+	movi		a2, .Lsaved_regs
+	movi		a3, .Lsaved_user_regs
+	s32i		a0, a2, 0
+	s32i		a1, a2, 4
+	save_xtregs_user a3 a4 a5 a6 a7 a8 0
+#if defined(__XTENSA_WINDOWED_ABI__)
+	spill_registers_kernel
+#elif defined(__XTENSA_CALL0_ABI__)
+	s32i		a12, a2, 8
+	s32i		a13, a2, 12
+	s32i		a14, a2, 16
+	s32i		a15, a2, 20
+#else
+#error Unsupported Xtensa ABI
+#endif
+	abi_call	swsusp_save
+	mov		a2, abi_rv
+	abi_ret_default
+
+ENDPROC(swsusp_arch_suspend)
+
+ENTRY(swsusp_arch_resume)
+
+	abi_entry_default
+
+#if defined(__XTENSA_WINDOWED_ABI__)
+	spill_registers_kernel
+#endif
+
+	movi		a2, restore_pblist
+	l32i		a2, a2, 0
+
+.Lcopy_pbe:
+	l32i		a3, a2, PBE_ADDRESS
+	l32i		a4, a2, PBE_ORIG_ADDRESS
+
+	__loopi		a3, a9, PAGE_SIZE, 16
+	l32i		a5, a3, 0
+	l32i		a6, a3, 4
+	l32i		a7, a3, 8
+	l32i		a8, a3, 12
+	addi		a3, a3, 16
+	s32i		a5, a4, 0
+	s32i		a6, a4, 4
+	s32i		a7, a4, 8
+	s32i		a8, a4, 12
+	addi		a4, a4, 16
+	__endl		a3, a9
+
+	l32i		a2, a2, PBE_NEXT
+	bnez		a2, .Lcopy_pbe
+
+	movi		a2, .Lsaved_regs
+	movi		a3, .Lsaved_user_regs
+	l32i		a0, a2, 0
+	l32i		a1, a2, 4
+	load_xtregs_user a3 a4 a5 a6 a7 a8 0
+#if defined(__XTENSA_CALL0_ABI__)
+	l32i		a12, a2, 8
+	l32i		a13, a2, 12
+	l32i		a14, a2, 16
+	l32i		a15, a2, 20
+#endif
+	movi		a2, 0
+	abi_ret_default
+
+ENDPROC(swsusp_arch_resume)
+
+#endif
diff --git a/arch/xtensa/kernel/hibernate.c b/arch/xtensa/kernel/hibernate.c
new file mode 100644
index 000000000000..06984327d6e2
--- /dev/null
+++ b/arch/xtensa/kernel/hibernate.c
@@ -0,0 +1,25 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/mm.h>
+#include <linux/suspend.h>
+#include <asm/coprocessor.h>
+
+int pfn_is_nosave(unsigned long pfn)
+{
+	unsigned long nosave_begin_pfn = PFN_DOWN(__pa(&__nosave_begin));
+	unsigned long nosave_end_pfn = PFN_UP(__pa(&__nosave_end));
+
+	return	(pfn >= nosave_begin_pfn) && (pfn < nosave_end_pfn);
+}
+
+void notrace save_processor_state(void)
+{
+	WARN_ON(num_online_cpus() != 1);
+#if XTENSA_HAVE_COPROCESSORS
+	local_coprocessors_flush_release_all();
+#endif
+}
+
+void notrace restore_processor_state(void)
+{
+}
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index e8bfbca5f001..7e38292dd07a 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -47,6 +47,7 @@
 #include <asm/asm-offsets.h>
 #include <asm/regs.h>
 #include <asm/hw_breakpoint.h>
+#include <asm/traps.h>
 
 extern void ret_from_fork(void);
 extern void ret_from_kernel_thread(void);
@@ -63,52 +64,114 @@ EXPORT_SYMBOL(__stack_chk_guard);
 
 #if XTENSA_HAVE_COPROCESSORS
 
-void coprocessor_release_all(struct thread_info *ti)
+void local_coprocessors_flush_release_all(void)
 {
-	unsigned long cpenable;
-	int i;
+	struct thread_info **coprocessor_owner;
+	struct thread_info *unique_owner[XCHAL_CP_MAX];
+	int n = 0;
+	int i, j;
 
-	/* Make sure we don't switch tasks during this operation. */
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+	xtensa_set_sr(XCHAL_CP_MASK, cpenable);
 
-	preempt_disable();
+	for (i = 0; i < XCHAL_CP_MAX; i++) {
+		struct thread_info *ti = coprocessor_owner[i];
 
-	/* Walk through all cp owners and release it for the requested one. */
+		if (ti) {
+			coprocessor_flush(ti, i);
 
-	cpenable = ti->cpenable;
+			for (j = 0; j < n; j++)
+				if (unique_owner[j] == ti)
+					break;
+			if (j == n)
+				unique_owner[n++] = ti;
 
-	for (i = 0; i < XCHAL_CP_MAX; i++) {
-		if (coprocessor_owner[i] == ti) {
-			coprocessor_owner[i] = 0;
-			cpenable &= ~(1 << i);
+			coprocessor_owner[i] = NULL;
 		}
 	}
+	for (i = 0; i < n; i++) {
+		/* pairs with memw (1) in fast_coprocessor and memw in switch_to */
+		smp_wmb();
+		unique_owner[i]->cpenable = 0;
+	}
+	xtensa_set_sr(0, cpenable);
+}
 
-	ti->cpenable = cpenable;
+static void local_coprocessor_release_all(void *info)
+{
+	struct thread_info *ti = info;
+	struct thread_info **coprocessor_owner;
+	int i;
+
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+
+	/* Walk through all cp owners and release it for the requested one. */
+
+	for (i = 0; i < XCHAL_CP_MAX; i++) {
+		if (coprocessor_owner[i] == ti)
+			coprocessor_owner[i] = NULL;
+	}
+	/* pairs with memw (1) in fast_coprocessor and memw in switch_to */
+	smp_wmb();
+	ti->cpenable = 0;
 	if (ti == current_thread_info())
 		xtensa_set_sr(0, cpenable);
+}
 
-	preempt_enable();
+void coprocessor_release_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_release_all,
+					 ti, true);
+	}
 }
 
-void coprocessor_flush_all(struct thread_info *ti)
+static void local_coprocessor_flush_all(void *info)
 {
-	unsigned long cpenable, old_cpenable;
+	struct thread_info *ti = info;
+	struct thread_info **coprocessor_owner;
+	unsigned long old_cpenable;
 	int i;
 
-	preempt_disable();
-
-	old_cpenable = xtensa_get_sr(cpenable);
-	cpenable = ti->cpenable;
-	xtensa_set_sr(cpenable, cpenable);
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+	old_cpenable = xtensa_xsr(ti->cpenable, cpenable);
 
 	for (i = 0; i < XCHAL_CP_MAX; i++) {
-		if ((cpenable & 1) != 0 && coprocessor_owner[i] == ti)
+		if (coprocessor_owner[i] == ti)
 			coprocessor_flush(ti, i);
-		cpenable >>= 1;
 	}
 	xtensa_set_sr(old_cpenable, cpenable);
+}
 
-	preempt_enable();
+void coprocessor_flush_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_flush_all,
+					 ti, true);
+	}
+}
+
+static void local_coprocessor_flush_release_all(void *info)
+{
+	local_coprocessor_flush_all(info);
+	local_coprocessor_release_all(info);
+}
+
+void coprocessor_flush_release_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_flush_release_all,
+					 ti, true);
+	}
 }
 
 #endif
@@ -140,8 +203,7 @@ void flush_thread(void)
 {
 #if XTENSA_HAVE_COPROCESSORS
 	struct thread_info *ti = current_thread_info();
-	coprocessor_flush_all(ti);
-	coprocessor_release_all(ti);
+	coprocessor_flush_release_all(ti);
 #endif
 	flush_ptrace_hw_breakpoint(current);
 }
diff --git a/arch/xtensa/kernel/ptrace.c b/arch/xtensa/kernel/ptrace.c
index 323c678a691f..22cdaa6729d3 100644
--- a/arch/xtensa/kernel/ptrace.c
+++ b/arch/xtensa/kernel/ptrace.c
@@ -171,8 +171,7 @@ static int tie_set(struct task_struct *target,
 
 #if XTENSA_HAVE_COPROCESSORS
 	/* Flush all coprocessors before we overwrite them. */
-	coprocessor_flush_all(ti);
-	coprocessor_release_all(ti);
+	coprocessor_flush_release_all(ti);
 	ti->xtregs_cp.cp0 = newregs->cp0;
 	ti->xtregs_cp.cp1 = newregs->cp1;
 	ti->xtregs_cp.cp2 = newregs->cp2;
diff --git a/arch/xtensa/kernel/s32c1i_selftest.c b/arch/xtensa/kernel/s32c1i_selftest.c
index 07e56e3a9a8b..8362388c8719 100644
--- a/arch/xtensa/kernel/s32c1i_selftest.c
+++ b/arch/xtensa/kernel/s32c1i_selftest.c
@@ -40,14 +40,13 @@ static inline int probed_compare_swap(int *v, int cmp, int set)
 
 /* Handle probed exception */
 
-static void __init do_probed_exception(struct pt_regs *regs,
-				       unsigned long exccause)
+static void __init do_probed_exception(struct pt_regs *regs)
 {
 	if (regs->pc == rcw_probe_pc) {	/* exception on s32c1i ? */
 		regs->pc += 3;		/* skip the s32c1i instruction */
-		rcw_exc = exccause;
+		rcw_exc = regs->exccause;
 	} else {
-		do_unhandled(regs, exccause);
+		do_unhandled(regs);
 	}
 }
 
diff --git a/arch/xtensa/kernel/signal.c b/arch/xtensa/kernel/signal.c
index 6f68649e86ba..c9ffd42db873 100644
--- a/arch/xtensa/kernel/signal.c
+++ b/arch/xtensa/kernel/signal.c
@@ -162,8 +162,7 @@ setup_sigcontext(struct rt_sigframe __user *frame, struct pt_regs *regs)
 		return err;
 
 #if XTENSA_HAVE_COPROCESSORS
-	coprocessor_flush_all(ti);
-	coprocessor_release_all(ti);
+	coprocessor_flush_release_all(ti);
 	err |= __copy_to_user(&frame->xtregs.cp, &ti->xtregs_cp,
 			      sizeof (frame->xtregs.cp));
 #endif
diff --git a/arch/xtensa/kernel/smp.c b/arch/xtensa/kernel/smp.c
index 1254da07ead1..4dc109dd6214 100644
--- a/arch/xtensa/kernel/smp.c
+++ b/arch/xtensa/kernel/smp.c
@@ -30,6 +30,7 @@
 #include <linux/thread_info.h>
 
 #include <asm/cacheflush.h>
+#include <asm/coprocessor.h>
 #include <asm/kdebug.h>
 #include <asm/mmu_context.h>
 #include <asm/mxregs.h>
@@ -272,6 +273,12 @@ int __cpu_disable(void)
 	 */
 	set_cpu_online(cpu, false);
 
+#if XTENSA_HAVE_COPROCESSORS
+	/*
+	 * Flush coprocessor contexts that are active on the current CPU.
+	 */
+	local_coprocessors_flush_release_all();
+#endif
 	/*
 	 * OK - migrate IRQs away from this CPU
 	 */
diff --git a/arch/xtensa/kernel/traps.c b/arch/xtensa/kernel/traps.c
index 9345007d474d..0c25e035ff10 100644
--- a/arch/xtensa/kernel/traps.c
+++ b/arch/xtensa/kernel/traps.c
@@ -48,25 +48,20 @@
  * Machine specific interrupt handlers
  */
 
-extern void kernel_exception(void);
-extern void user_exception(void);
-
-extern void fast_illegal_instruction_user(void);
-extern void fast_syscall_user(void);
-extern void fast_alloca(void);
-extern void fast_unaligned(void);
-extern void fast_second_level_miss(void);
-extern void fast_store_prohibited(void);
-extern void fast_coprocessor(void);
-
-extern void do_illegal_instruction (struct pt_regs*);
-extern void do_interrupt (struct pt_regs*);
-extern void do_nmi(struct pt_regs *);
-extern void do_unaligned_user (struct pt_regs*);
-extern void do_multihit (struct pt_regs*, unsigned long);
-extern void do_page_fault (struct pt_regs*, unsigned long);
-extern void do_debug (struct pt_regs*);
-extern void system_call (struct pt_regs*);
+static void do_illegal_instruction(struct pt_regs *regs);
+static void do_div0(struct pt_regs *regs);
+static void do_interrupt(struct pt_regs *regs);
+#if XTENSA_FAKE_NMI
+static void do_nmi(struct pt_regs *regs);
+#endif
+#if XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION
+static void do_unaligned_user(struct pt_regs *regs);
+#endif
+static void do_multihit(struct pt_regs *regs);
+#if XTENSA_HAVE_COPROCESSORS
+static void do_coprocessor(struct pt_regs *regs);
+#endif
+static void do_debug(struct pt_regs *regs);
 
 /*
  * The vector table must be preceded by a save area (which
@@ -78,7 +73,8 @@ extern void system_call (struct pt_regs*);
 #define USER		0x02
 
 #define COPROCESSOR(x)							\
-{ EXCCAUSE_COPROCESSOR ## x ## _DISABLED, USER, fast_coprocessor }
+{ EXCCAUSE_COPROCESSOR ## x ## _DISABLED, USER|KRNL, fast_coprocessor },\
+{ EXCCAUSE_COPROCESSOR ## x ## _DISABLED, 0, do_coprocessor }
 
 typedef struct {
 	int cause;
@@ -100,7 +96,7 @@ static dispatch_init_table_t __initdata dispatch_init_table[] = {
 #ifdef SUPPORT_WINDOWED
 { EXCCAUSE_ALLOCA,		USER|KRNL, fast_alloca },
 #endif
-/* EXCCAUSE_INTEGER_DIVIDE_BY_ZERO unhandled */
+{ EXCCAUSE_INTEGER_DIVIDE_BY_ZERO, 0,	   do_div0 },
 /* EXCCAUSE_PRIVILEGED unhandled */
 #if XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION
 #ifdef CONFIG_XTENSA_UNALIGNED_USER
@@ -110,21 +106,21 @@ static dispatch_init_table_t __initdata dispatch_init_table[] = {
 { EXCCAUSE_UNALIGNED,		KRNL,	   fast_unaligned },
 #endif
 #ifdef CONFIG_MMU
-{ EXCCAUSE_ITLB_MISS,		0,	   do_page_fault },
-{ EXCCAUSE_ITLB_MISS,		USER|KRNL, fast_second_level_miss},
+{ EXCCAUSE_ITLB_MISS,			0,	   do_page_fault },
+{ EXCCAUSE_ITLB_MISS,			USER|KRNL, fast_second_level_miss},
+{ EXCCAUSE_DTLB_MISS,			USER|KRNL, fast_second_level_miss},
+{ EXCCAUSE_DTLB_MISS,			0,	   do_page_fault },
+{ EXCCAUSE_STORE_CACHE_ATTRIBUTE,	USER|KRNL, fast_store_prohibited },
+#endif /* CONFIG_MMU */
+#ifdef CONFIG_PFAULT
 { EXCCAUSE_ITLB_MULTIHIT,		0,	   do_multihit },
-{ EXCCAUSE_ITLB_PRIVILEGE,	0,	   do_page_fault },
-/* EXCCAUSE_SIZE_RESTRICTION unhandled */
+{ EXCCAUSE_ITLB_PRIVILEGE,		0,	   do_page_fault },
 { EXCCAUSE_FETCH_CACHE_ATTRIBUTE,	0,	   do_page_fault },
-{ EXCCAUSE_DTLB_MISS,		USER|KRNL, fast_second_level_miss},
-{ EXCCAUSE_DTLB_MISS,		0,	   do_page_fault },
 { EXCCAUSE_DTLB_MULTIHIT,		0,	   do_multihit },
-{ EXCCAUSE_DTLB_PRIVILEGE,	0,	   do_page_fault },
-/* EXCCAUSE_DTLB_SIZE_RESTRICTION unhandled */
-{ EXCCAUSE_STORE_CACHE_ATTRIBUTE,	USER|KRNL, fast_store_prohibited },
+{ EXCCAUSE_DTLB_PRIVILEGE,		0,	   do_page_fault },
 { EXCCAUSE_STORE_CACHE_ATTRIBUTE,	0,	   do_page_fault },
 { EXCCAUSE_LOAD_CACHE_ATTRIBUTE,	0,	   do_page_fault },
-#endif /* CONFIG_MMU */
+#endif
 /* XCCHAL_EXCCAUSE_FLOATING_POINT unhandled */
 #if XTENSA_HAVE_COPROCESSOR(0)
 COPROCESSOR(0),
@@ -179,7 +175,7 @@ __die_if_kernel(const char *str, struct pt_regs *regs, long err)
  * Unhandled Exceptions. Kill user task or panic if in kernel space.
  */
 
-void do_unhandled(struct pt_regs *regs, unsigned long exccause)
+void do_unhandled(struct pt_regs *regs)
 {
 	__die_if_kernel("Caught unhandled exception - should not happen",
 			regs, SIGKILL);
@@ -189,7 +185,7 @@ void do_unhandled(struct pt_regs *regs, unsigned long exccause)
 			    "(pid = %d, pc = %#010lx) - should not happen\n"
 			    "\tEXCCAUSE is %ld\n",
 			    current->comm, task_pid_nr(current), regs->pc,
-			    exccause);
+			    regs->exccause);
 	force_sig(SIGILL);
 }
 
@@ -197,7 +193,7 @@ void do_unhandled(struct pt_regs *regs, unsigned long exccause)
  * Multi-hit exception. This if fatal!
  */
 
-void do_multihit(struct pt_regs *regs, unsigned long exccause)
+static void do_multihit(struct pt_regs *regs)
 {
 	die("Caught multihit exception", regs, SIGKILL);
 }
@@ -206,8 +202,6 @@ void do_multihit(struct pt_regs *regs, unsigned long exccause)
  * IRQ handler.
  */
 
-extern void do_IRQ(int, struct pt_regs *);
-
 #if XTENSA_FAKE_NMI
 
 #define IS_POW2(v) (((v) & ((v) - 1)) == 0)
@@ -240,14 +234,10 @@ irqreturn_t xtensa_pmu_irq_handler(int irq, void *dev_id);
 
 DEFINE_PER_CPU(unsigned long, nmi_count);
 
-void do_nmi(struct pt_regs *regs)
+static void do_nmi(struct pt_regs *regs)
 {
-	struct pt_regs *old_regs;
-
-	if ((regs->ps & PS_INTLEVEL_MASK) < LOCKLEVEL)
-		trace_hardirqs_off();
+	struct pt_regs *old_regs = set_irq_regs(regs);
 
-	old_regs = set_irq_regs(regs);
 	nmi_enter();
 	++*this_cpu_ptr(&nmi_count);
 	check_valid_nmi();
@@ -257,7 +247,7 @@ void do_nmi(struct pt_regs *regs)
 }
 #endif
 
-void do_interrupt(struct pt_regs *regs)
+static void do_interrupt(struct pt_regs *regs)
 {
 	static const unsigned int_level_mask[] = {
 		0,
@@ -269,12 +259,9 @@ void do_interrupt(struct pt_regs *regs)
 		XCHAL_INTLEVEL6_MASK,
 		XCHAL_INTLEVEL7_MASK,
 	};
-	struct pt_regs *old_regs;
+	struct pt_regs *old_regs = set_irq_regs(regs);
 	unsigned unhandled = ~0u;
 
-	trace_hardirqs_off();
-
-	old_regs = set_irq_regs(regs);
 	irq_enter();
 
 	for (;;) {
@@ -306,13 +293,47 @@ void do_interrupt(struct pt_regs *regs)
 	set_irq_regs(old_regs);
 }
 
+static bool check_div0(struct pt_regs *regs)
+{
+	static const u8 pattern[] = {'D', 'I', 'V', '0'};
+	const u8 *p;
+	u8 buf[5];
+
+	if (user_mode(regs)) {
+		if (copy_from_user(buf, (void __user *)regs->pc + 2, 5))
+			return false;
+		p = buf;
+	} else {
+		p = (const u8 *)regs->pc + 2;
+	}
+
+	return memcmp(p, pattern, sizeof(pattern)) == 0 ||
+		memcmp(p + 1, pattern, sizeof(pattern)) == 0;
+}
+
 /*
  * Illegal instruction. Fatal if in kernel space.
  */
 
-void
-do_illegal_instruction(struct pt_regs *regs)
+static void do_illegal_instruction(struct pt_regs *regs)
 {
+#ifdef CONFIG_USER_ABI_CALL0_PROBE
+	/*
+	 * When call0 application encounters an illegal instruction fast
+	 * exception handler will attempt to set PS.WOE and retry failing
+	 * instruction.
+	 * If we get here we know that that instruction is also illegal
+	 * with PS.WOE set, so it's not related to the windowed option
+	 * hence PS.WOE may be cleared.
+	 */
+	if (regs->pc == current_thread_info()->ps_woe_fix_addr)
+		regs->ps &= ~PS_WOE_MASK;
+#endif
+	if (check_div0(regs)) {
+		do_div0(regs);
+		return;
+	}
+
 	__die_if_kernel("Illegal instruction in kernel", regs, SIGKILL);
 
 	/* If in user mode, send SIGILL signal to current process. */
@@ -322,6 +343,11 @@ do_illegal_instruction(struct pt_regs *regs)
 	force_sig(SIGILL);
 }
 
+static void do_div0(struct pt_regs *regs)
+{
+	__die_if_kernel("Unhandled division by 0 in kernel", regs, SIGKILL);
+	force_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)regs->pc);
+}
 
 /*
  * Handle unaligned memory accesses from user space. Kill task.
@@ -331,8 +357,7 @@ do_illegal_instruction(struct pt_regs *regs)
  */
 
 #if XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION
-void
-do_unaligned_user (struct pt_regs *regs)
+static void do_unaligned_user(struct pt_regs *regs)
 {
 	__die_if_kernel("Unhandled unaligned exception in kernel",
 			regs, SIGKILL);
@@ -347,14 +372,20 @@ do_unaligned_user (struct pt_regs *regs)
 }
 #endif
 
+#if XTENSA_HAVE_COPROCESSORS
+static void do_coprocessor(struct pt_regs *regs)
+{
+	coprocessor_flush_release_all(current_thread_info());
+}
+#endif
+
 /* Handle debug events.
  * When CONFIG_HAVE_HW_BREAKPOINT is on this handler is called with
  * preemption disabled to avoid rescheduling and keep mapping of hardware
  * breakpoint structures to debug registers intact, so that
  * DEBUGCAUSE.DBNUM could be used in case of data breakpoint hit.
  */
-void
-do_debug(struct pt_regs *regs)
+static void do_debug(struct pt_regs *regs)
 {
 #ifdef CONFIG_HAVE_HW_BREAKPOINT
 	int ret = check_hw_breakpoint(regs);
@@ -381,7 +412,8 @@ do_debug(struct pt_regs *regs)
 
 /* Set exception C handler - for temporary use when probing exceptions */
 
-void * __init trap_set_handler(int cause, void *handler)
+xtensa_exception_handler *
+__init trap_set_handler(int cause, xtensa_exception_handler *handler)
 {
 	void *previous = per_cpu(exc_table, 0).default_handler[cause];
 
@@ -392,8 +424,7 @@ void * __init trap_set_handler(int cause, void *handler)
 
 static void trap_init_excsave(void)
 {
-	unsigned long excsave1 = (unsigned long)this_cpu_ptr(&exc_table);
-	__asm__ __volatile__("wsr  %0, excsave1\n" : : "a" (excsave1));
+	xtensa_set_sr(this_cpu_ptr(&exc_table), excsave1);
 }
 
 static void trap_init_debug(void)
diff --git a/arch/xtensa/lib/Makefile b/arch/xtensa/lib/Makefile
index 5848c133f7ea..d4e9c397e3fd 100644
--- a/arch/xtensa/lib/Makefile
+++ b/arch/xtensa/lib/Makefile
@@ -8,3 +8,5 @@ lib-y	+= memcopy.o memset.o checksum.o \
 	   divsi3.o udivsi3.o modsi3.o umodsi3.o mulsi3.o \
 	   usercopy.o strncpy_user.o strnlen_user.o
 lib-$(CONFIG_PCI) += pci-auto.o
+lib-$(CONFIG_KCSAN) += kcsan-stubs.o
+KCSAN_SANITIZE_kcsan-stubs.o := n
diff --git a/arch/xtensa/lib/kcsan-stubs.c b/arch/xtensa/lib/kcsan-stubs.c
new file mode 100644
index 000000000000..2b08faa62b86
--- /dev/null
+++ b/arch/xtensa/lib/kcsan-stubs.c
@@ -0,0 +1,54 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/bug.h>
+#include <linux/types.h>
+
+void __atomic_store_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_load_8(const volatile void *p, int i)
+{
+	BUG();
+}
+
+u64 __atomic_exchange_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+bool __atomic_compare_exchange_8(volatile void *p1, void *p2, u64 v, bool b, int i1, int i2)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_add_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_sub_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_and_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_or_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_xor_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
+
+u64 __atomic_fetch_nand_8(volatile void *p, u64 v, int i)
+{
+	BUG();
+}
diff --git a/arch/xtensa/lib/memcopy.S b/arch/xtensa/lib/memcopy.S
index 582d817979ed..b20d206bcb71 100644
--- a/arch/xtensa/lib/memcopy.S
+++ b/arch/xtensa/lib/memcopy.S
@@ -402,13 +402,13 @@ WEAK(memmove)
 	 */
 	# copy 16 bytes per iteration for word-aligned dst and word-aligned src
 #if XCHAL_HAVE_LOOPS
-	loopnez	a7, .backLoop1done
+	loopnez	a7, .LbackLoop1done
 #else /* !XCHAL_HAVE_LOOPS */
-	beqz	a7, .backLoop1done
+	beqz	a7, .LbackLoop1done
 	slli	a8, a7, 4
 	sub	a8, a3, a8	# a8 = start of first 16B source chunk
 #endif /* !XCHAL_HAVE_LOOPS */
-.backLoop1:
+.LbackLoop1:
 	addi	a3, a3, -16
 	l32i	a7, a3, 12
 	l32i	a6, a3,  8
@@ -420,9 +420,9 @@ WEAK(memmove)
 	s32i	a7, a5,  4
 	s32i	a6, a5,  0
 #if !XCHAL_HAVE_LOOPS
-	bne	a3, a8, .backLoop1  # continue loop if a3:src != a8:src_start
+	bne	a3, a8, .LbackLoop1  # continue loop if a3:src != a8:src_start
 #endif /* !XCHAL_HAVE_LOOPS */
-.backLoop1done:
+.LbackLoop1done:
 	bbci.l	a4, 3, .Lback2
 	# copy 8 bytes
 	addi	a3, a3, -8
@@ -479,13 +479,13 @@ WEAK(memmove)
 #endif
 	l32i	a6, a3, 0	# load first word
 #if XCHAL_HAVE_LOOPS
-	loopnez	a7, .backLoop2done
+	loopnez	a7, .LbackLoop2done
 #else /* !XCHAL_HAVE_LOOPS */
-	beqz	a7, .backLoop2done
+	beqz	a7, .LbackLoop2done
 	slli	a10, a7, 4
 	sub	a10, a3, a10	# a10 = start of first 16B source chunk
 #endif /* !XCHAL_HAVE_LOOPS */
-.backLoop2:
+.LbackLoop2:
 	addi	a3, a3, -16
 	l32i	a7, a3, 12
 	l32i	a8, a3,  8
@@ -501,9 +501,9 @@ WEAK(memmove)
 	__src_b	a9, a6, a9
 	s32i	a9, a5,  0
 #if !XCHAL_HAVE_LOOPS
-	bne	a3, a10, .backLoop2 # continue loop if a3:src != a10:src_start
+	bne	a3, a10, .LbackLoop2 # continue loop if a3:src != a10:src_start
 #endif /* !XCHAL_HAVE_LOOPS */
-.backLoop2done:
+.LbackLoop2done:
 	bbci.l	a4, 3, .Lback12
 	# copy 8 bytes
 	addi	a3, a3, -8
diff --git a/arch/xtensa/mm/Makefile b/arch/xtensa/mm/Makefile
index f7fb08ae768f..44153a335951 100644
--- a/arch/xtensa/mm/Makefile
+++ b/arch/xtensa/mm/Makefile
@@ -4,7 +4,8 @@
 #
 
 obj-y			:= init.o misc.o
-obj-$(CONFIG_MMU)	+= cache.o fault.o ioremap.o mmu.o tlb.o
+obj-$(CONFIG_PFAULT)	+= fault.o
+obj-$(CONFIG_MMU)	+= cache.o ioremap.o mmu.o tlb.o
 obj-$(CONFIG_HIGHMEM)	+= highmem.o
 obj-$(CONFIG_KASAN)	+= kasan_init.o
 
diff --git a/arch/xtensa/mm/fault.c b/arch/xtensa/mm/fault.c
index 06d0973a0d74..16f0a5ff5799 100644
--- a/arch/xtensa/mm/fault.c
+++ b/arch/xtensa/mm/fault.c
@@ -21,9 +21,61 @@
 #include <asm/cacheflush.h>
 #include <asm/hardirq.h>
 
-DEFINE_PER_CPU(unsigned long, asid_cache) = ASID_USER_FIRST;
 void bad_page_fault(struct pt_regs*, unsigned long, int);
 
+static void vmalloc_fault(struct pt_regs *regs, unsigned int address)
+{
+#ifdef CONFIG_MMU
+	/* Synchronize this task's top level page-table
+	 * with the 'reference' page table.
+	 */
+	struct mm_struct *act_mm = current->active_mm;
+	int index = pgd_index(address);
+	pgd_t *pgd, *pgd_k;
+	p4d_t *p4d, *p4d_k;
+	pud_t *pud, *pud_k;
+	pmd_t *pmd, *pmd_k;
+	pte_t *pte_k;
+
+	if (act_mm == NULL)
+		goto bad_page_fault;
+
+	pgd = act_mm->pgd + index;
+	pgd_k = init_mm.pgd + index;
+
+	if (!pgd_present(*pgd_k))
+		goto bad_page_fault;
+
+	pgd_val(*pgd) = pgd_val(*pgd_k);
+
+	p4d = p4d_offset(pgd, address);
+	p4d_k = p4d_offset(pgd_k, address);
+	if (!p4d_present(*p4d) || !p4d_present(*p4d_k))
+		goto bad_page_fault;
+
+	pud = pud_offset(p4d, address);
+	pud_k = pud_offset(p4d_k, address);
+	if (!pud_present(*pud) || !pud_present(*pud_k))
+		goto bad_page_fault;
+
+	pmd = pmd_offset(pud, address);
+	pmd_k = pmd_offset(pud_k, address);
+	if (!pmd_present(*pmd) || !pmd_present(*pmd_k))
+		goto bad_page_fault;
+
+	pmd_val(*pmd) = pmd_val(*pmd_k);
+	pte_k = pte_offset_kernel(pmd_k, address);
+
+	if (!pte_present(*pte_k))
+		goto bad_page_fault;
+	return;
+
+bad_page_fault:
+	bad_page_fault(regs, address, SIGKILL);
+#else
+	WARN_ONCE(1, "%s in noMMU configuration\n", __func__);
+#endif
+}
 /*
  * This routine handles page faults.  It determines the address,
  * and the problem, and then passes it off to one of the appropriate
@@ -49,8 +101,10 @@ void do_page_fault(struct pt_regs *regs)
 	/* We fault-in kernel-space virtual memory on-demand. The
 	 * 'reference' page table is init_mm.pgd.
 	 */
-	if (address >= TASK_SIZE && !user_mode(regs))
-		goto vmalloc_fault;
+	if (address >= TASK_SIZE && !user_mode(regs)) {
+		vmalloc_fault(regs, address);
+		return;
+	}
 
 	/* If we're in an interrupt or have no user
 	 * context, we must not take the fault..
@@ -114,7 +168,7 @@ good_area:
 
 	if (fault_signal_pending(fault, regs)) {
 		if (!user_mode(regs))
-			goto bad_page_fault;
+			bad_page_fault(regs, address, SIGKILL);
 		return;
 	}
 
@@ -181,56 +235,6 @@ do_sigbus:
 	if (!user_mode(regs))
 		bad_page_fault(regs, address, SIGBUS);
 	return;
-
-vmalloc_fault:
-	{
-		/* Synchronize this task's top level page-table
-		 * with the 'reference' page table.
-		 */
-		struct mm_struct *act_mm = current->active_mm;
-		int index = pgd_index(address);
-		pgd_t *pgd, *pgd_k;
-		p4d_t *p4d, *p4d_k;
-		pud_t *pud, *pud_k;
-		pmd_t *pmd, *pmd_k;
-		pte_t *pte_k;
-
-		if (act_mm == NULL)
-			goto bad_page_fault;
-
-		pgd = act_mm->pgd + index;
-		pgd_k = init_mm.pgd + index;
-
-		if (!pgd_present(*pgd_k))
-			goto bad_page_fault;
-
-		pgd_val(*pgd) = pgd_val(*pgd_k);
-
-		p4d = p4d_offset(pgd, address);
-		p4d_k = p4d_offset(pgd_k, address);
-		if (!p4d_present(*p4d) || !p4d_present(*p4d_k))
-			goto bad_page_fault;
-
-		pud = pud_offset(p4d, address);
-		pud_k = pud_offset(p4d_k, address);
-		if (!pud_present(*pud) || !pud_present(*pud_k))
-			goto bad_page_fault;
-
-		pmd = pmd_offset(pud, address);
-		pmd_k = pmd_offset(pud_k, address);
-		if (!pmd_present(*pmd) || !pmd_present(*pmd_k))
-			goto bad_page_fault;
-
-		pmd_val(*pmd) = pmd_val(*pmd_k);
-		pte_k = pte_offset_kernel(pmd_k, address);
-
-		if (!pte_present(*pte_k))
-			goto bad_page_fault;
-		return;
-	}
-bad_page_fault:
-	bad_page_fault(regs, address, SIGKILL);
-	return;
 }
 
 
diff --git a/arch/xtensa/mm/mmu.c b/arch/xtensa/mm/mmu.c
index 38acda4f04e8..92e158c69c10 100644
--- a/arch/xtensa/mm/mmu.c
+++ b/arch/xtensa/mm/mmu.c
@@ -18,6 +18,8 @@
 #include <asm/initialize_mmu.h>
 #include <asm/io.h>
 
+DEFINE_PER_CPU(unsigned long, asid_cache) = ASID_USER_FIRST;
+
 #if defined(CONFIG_HIGHMEM)
 static void * __init init_pmd(unsigned long vaddr, unsigned long n_pages)
 {
diff --git a/arch/xtensa/platforms/iss/network.c b/arch/xtensa/platforms/iss/network.c
index be3aaaad8bee..fd84d4891758 100644
--- a/arch/xtensa/platforms/iss/network.c
+++ b/arch/xtensa/platforms/iss/network.c
@@ -38,9 +38,6 @@
 #define ISS_NET_TIMER_VALUE (HZ / 10)
 
 
-static DEFINE_SPINLOCK(opened_lock);
-static LIST_HEAD(opened);
-
 static DEFINE_SPINLOCK(devices_lock);
 static LIST_HEAD(devices);
 
@@ -59,17 +56,27 @@ struct tuntap_info {
 /* ------------------------------------------------------------------------- */
 
 
+struct iss_net_private;
+
+struct iss_net_ops {
+	int (*open)(struct iss_net_private *lp);
+	void (*close)(struct iss_net_private *lp);
+	int (*read)(struct iss_net_private *lp, struct sk_buff **skb);
+	int (*write)(struct iss_net_private *lp, struct sk_buff **skb);
+	unsigned short (*protocol)(struct sk_buff *skb);
+	int (*poll)(struct iss_net_private *lp);
+};
+
 /* This structure contains out private information for the driver. */
 
 struct iss_net_private {
 	struct list_head device_list;
-	struct list_head opened_list;
 
 	spinlock_t lock;
 	struct net_device *dev;
 	struct platform_device pdev;
 	struct timer_list tl;
-	struct net_device_stats stats;
+	struct rtnl_link_stats64 stats;
 
 	struct timer_list timer;
 	unsigned int timer_val;
@@ -82,12 +89,7 @@ struct iss_net_private {
 			struct tuntap_info tuntap;
 		} info;
 
-		int (*open)(struct iss_net_private *lp);
-		void (*close)(struct iss_net_private *lp);
-		int (*read)(struct iss_net_private *lp, struct sk_buff **skb);
-		int (*write)(struct iss_net_private *lp, struct sk_buff **skb);
-		unsigned short (*protocol)(struct sk_buff *skb);
-		int (*poll)(struct iss_net_private *lp);
+		const struct iss_net_ops *net_ops;
 	} tp;
 
 };
@@ -215,6 +217,15 @@ static int tuntap_poll(struct iss_net_private *lp)
 	return simc_poll(lp->tp.info.tuntap.fd);
 }
 
+static const struct iss_net_ops tuntap_ops = {
+	.open		= tuntap_open,
+	.close		= tuntap_close,
+	.read		= tuntap_read,
+	.write		= tuntap_write,
+	.protocol	= tuntap_protocol,
+	.poll		= tuntap_poll,
+};
+
 /*
  * ethX=tuntap,[mac address],device name
  */
@@ -257,13 +268,7 @@ static int tuntap_probe(struct iss_net_private *lp, int index, char *init)
 	lp->mtu = TRANSPORT_TUNTAP_MTU;
 
 	lp->tp.info.tuntap.fd = -1;
-
-	lp->tp.open = tuntap_open;
-	lp->tp.close = tuntap_close;
-	lp->tp.read = tuntap_read;
-	lp->tp.write = tuntap_write;
-	lp->tp.protocol = tuntap_protocol;
-	lp->tp.poll = tuntap_poll;
+	lp->tp.net_ops = &tuntap_ops;
 
 	return 1;
 }
@@ -278,14 +283,16 @@ static int iss_net_rx(struct net_device *dev)
 
 	/* Check if there is any new data. */
 
-	if (lp->tp.poll(lp) == 0)
+	if (lp->tp.net_ops->poll(lp) == 0)
 		return 0;
 
 	/* Try to allocate memory, if it fails, try again next round. */
 
 	skb = dev_alloc_skb(dev->mtu + 2 + ETH_HEADER_OTHER);
 	if (skb == NULL) {
+		spin_lock_bh(&lp->lock);
 		lp->stats.rx_dropped++;
+		spin_unlock_bh(&lp->lock);
 		return 0;
 	}
 
@@ -295,15 +302,17 @@ static int iss_net_rx(struct net_device *dev)
 
 	skb->dev = dev;
 	skb_reset_mac_header(skb);
-	pkt_len = lp->tp.read(lp, &skb);
+	pkt_len = lp->tp.net_ops->read(lp, &skb);
 	skb_put(skb, pkt_len);
 
 	if (pkt_len > 0) {
 		skb_trim(skb, pkt_len);
-		skb->protocol = lp->tp.protocol(skb);
+		skb->protocol = lp->tp.net_ops->protocol(skb);
 
+		spin_lock_bh(&lp->lock);
 		lp->stats.rx_bytes += skb->len;
 		lp->stats.rx_packets++;
+		spin_unlock_bh(&lp->lock);
 		netif_rx(skb);
 		return pkt_len;
 	}
@@ -311,38 +320,24 @@ static int iss_net_rx(struct net_device *dev)
 	return pkt_len;
 }
 
-static int iss_net_poll(void)
+static int iss_net_poll(struct iss_net_private *lp)
 {
-	struct list_head *ele;
 	int err, ret = 0;
 
-	spin_lock(&opened_lock);
-
-	list_for_each(ele, &opened) {
-		struct iss_net_private *lp;
-
-		lp = list_entry(ele, struct iss_net_private, opened_list);
-
-		if (!netif_running(lp->dev))
-			break;
-
-		spin_lock(&lp->lock);
-
-		while ((err = iss_net_rx(lp->dev)) > 0)
-			ret++;
+	if (!netif_running(lp->dev))
+		return 0;
 
-		spin_unlock(&lp->lock);
+	while ((err = iss_net_rx(lp->dev)) > 0)
+		ret++;
 
-		if (err < 0) {
-			pr_err("Device '%s' read returned %d, shutting it down\n",
-			       lp->dev->name, err);
-			dev_close(lp->dev);
-		} else {
-			/* FIXME reactivate_fd(lp->fd, ISS_ETH_IRQ); */
-		}
+	if (err < 0) {
+		pr_err("Device '%s' read returned %d, shutting it down\n",
+		       lp->dev->name, err);
+		dev_close(lp->dev);
+	} else {
+		/* FIXME reactivate_fd(lp->fd, ISS_ETH_IRQ); */
 	}
 
-	spin_unlock(&opened_lock);
 	return ret;
 }
 
@@ -351,10 +346,8 @@ static void iss_net_timer(struct timer_list *t)
 {
 	struct iss_net_private *lp = from_timer(lp, t, timer);
 
-	iss_net_poll();
-	spin_lock(&lp->lock);
+	iss_net_poll(lp);
 	mod_timer(&lp->timer, jiffies + lp->timer_val);
-	spin_unlock(&lp->lock);
 }
 
 
@@ -363,11 +356,9 @@ static int iss_net_open(struct net_device *dev)
 	struct iss_net_private *lp = netdev_priv(dev);
 	int err;
 
-	spin_lock_bh(&lp->lock);
-
-	err = lp->tp.open(lp);
+	err = lp->tp.net_ops->open(lp);
 	if (err < 0)
-		goto out;
+		return err;
 
 	netif_start_queue(dev);
 
@@ -378,36 +369,21 @@ static int iss_net_open(struct net_device *dev)
 	while ((err = iss_net_rx(dev)) > 0)
 		;
 
-	spin_unlock_bh(&lp->lock);
-	spin_lock_bh(&opened_lock);
-	list_add(&lp->opened_list, &opened);
-	spin_unlock_bh(&opened_lock);
-	spin_lock_bh(&lp->lock);
-
 	timer_setup(&lp->timer, iss_net_timer, 0);
 	lp->timer_val = ISS_NET_TIMER_VALUE;
 	mod_timer(&lp->timer, jiffies + lp->timer_val);
 
-out:
-	spin_unlock_bh(&lp->lock);
 	return err;
 }
 
 static int iss_net_close(struct net_device *dev)
 {
 	struct iss_net_private *lp = netdev_priv(dev);
-	netif_stop_queue(dev);
-	spin_lock_bh(&lp->lock);
-
-	spin_lock(&opened_lock);
-	list_del(&opened);
-	spin_unlock(&opened_lock);
 
+	netif_stop_queue(dev);
 	del_timer_sync(&lp->timer);
+	lp->tp.net_ops->close(lp);
 
-	lp->tp.close(lp);
-
-	spin_unlock_bh(&lp->lock);
 	return 0;
 }
 
@@ -417,13 +393,14 @@ static int iss_net_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	int len;
 
 	netif_stop_queue(dev);
-	spin_lock_bh(&lp->lock);
 
-	len = lp->tp.write(lp, &skb);
+	len = lp->tp.net_ops->write(lp, &skb);
 
 	if (len == skb->len) {
+		spin_lock_bh(&lp->lock);
 		lp->stats.tx_packets++;
 		lp->stats.tx_bytes += skb->len;
+		spin_unlock_bh(&lp->lock);
 		netif_trans_update(dev);
 		netif_start_queue(dev);
 
@@ -432,24 +409,29 @@ static int iss_net_start_xmit(struct sk_buff *skb, struct net_device *dev)
 
 	} else if (len == 0) {
 		netif_start_queue(dev);
+		spin_lock_bh(&lp->lock);
 		lp->stats.tx_dropped++;
+		spin_unlock_bh(&lp->lock);
 
 	} else {
 		netif_start_queue(dev);
 		pr_err("%s: %s failed(%d)\n", dev->name, __func__, len);
 	}
 
-	spin_unlock_bh(&lp->lock);
 
 	dev_kfree_skb(skb);
 	return NETDEV_TX_OK;
 }
 
 
-static struct net_device_stats *iss_net_get_stats(struct net_device *dev)
+static void iss_net_get_stats64(struct net_device *dev,
+				struct rtnl_link_stats64 *stats)
 {
 	struct iss_net_private *lp = netdev_priv(dev);
-	return &lp->stats;
+
+	spin_lock_bh(&lp->lock);
+	*stats = lp->stats;
+	spin_unlock_bh(&lp->lock);
 }
 
 static void iss_net_set_multicast_list(struct net_device *dev)
@@ -460,19 +442,6 @@ static void iss_net_tx_timeout(struct net_device *dev, unsigned int txqueue)
 {
 }
 
-static int iss_net_set_mac(struct net_device *dev, void *addr)
-{
-	struct iss_net_private *lp = netdev_priv(dev);
-	struct sockaddr *hwaddr = addr;
-
-	if (!is_valid_ether_addr(hwaddr->sa_data))
-		return -EADDRNOTAVAIL;
-	spin_lock_bh(&lp->lock);
-	eth_hw_addr_set(dev, hwaddr->sa_data);
-	spin_unlock_bh(&lp->lock);
-	return 0;
-}
-
 static int iss_net_change_mtu(struct net_device *dev, int new_mtu)
 {
 	return -EINVAL;
@@ -494,11 +463,11 @@ static int driver_registered;
 static const struct net_device_ops iss_netdev_ops = {
 	.ndo_open		= iss_net_open,
 	.ndo_stop		= iss_net_close,
-	.ndo_get_stats		= iss_net_get_stats,
+	.ndo_get_stats64	= iss_net_get_stats64,
 	.ndo_start_xmit		= iss_net_start_xmit,
 	.ndo_validate_addr	= eth_validate_addr,
 	.ndo_change_mtu		= iss_net_change_mtu,
-	.ndo_set_mac_address	= iss_net_set_mac,
+	.ndo_set_mac_address	= eth_mac_addr,
 	.ndo_tx_timeout		= iss_net_tx_timeout,
 	.ndo_set_rx_mode	= iss_net_set_multicast_list,
 };
@@ -520,7 +489,6 @@ static int iss_net_configure(int index, char *init)
 	lp = netdev_priv(dev);
 	*lp = (struct iss_net_private) {
 		.device_list		= LIST_HEAD_INIT(lp->device_list),
-		.opened_list		= LIST_HEAD_INIT(lp->opened_list),
 		.dev			= dev,
 		.index			= index,
 	};
diff --git a/arch/xtensa/platforms/iss/simdisk.c b/arch/xtensa/platforms/iss/simdisk.c
index 0f0e0724397f..4255b92fa3eb 100644
--- a/arch/xtensa/platforms/iss/simdisk.c
+++ b/arch/xtensa/platforms/iss/simdisk.c
@@ -211,12 +211,18 @@ static ssize_t proc_read_simdisk(struct file *file, char __user *buf,
 	struct simdisk *dev = pde_data(file_inode(file));
 	const char *s = dev->filename;
 	if (s) {
-		ssize_t n = simple_read_from_buffer(buf, size, ppos,
-							s, strlen(s));
-		if (n < 0)
-			return n;
-		buf += n;
-		size -= n;
+		ssize_t len = strlen(s);
+		char *temp = kmalloc(len + 2, GFP_KERNEL);
+
+		if (!temp)
+			return -ENOMEM;
+
+		len = scnprintf(temp, len + 2, "%s\n", s);
+		len = simple_read_from_buffer(buf, size, ppos,
+					      temp, len);
+
+		kfree(temp);
+		return len;
 	}
 	return simple_read_from_buffer(buf, size, ppos, "\n", 1);
 }
diff --git a/arch/xtensa/platforms/xt2000/setup.c b/arch/xtensa/platforms/xt2000/setup.c
index 145d129be76f..0dc22c371614 100644
--- a/arch/xtensa/platforms/xt2000/setup.c
+++ b/arch/xtensa/platforms/xt2000/setup.c
@@ -78,7 +78,7 @@ void __init platform_init(bp_tag_t *first)
 
 void platform_heartbeat(void)
 {
-	static int i=0, t = 0;
+	static int i, t;
 
 	if (--t < 0)
 	{