--- a/arch/arm/atomic.h
+++ b/arch/arm/atomic.h
@@ -22,37 +22,150 @@ static inline int a_ctz_64(uint64_t x)
 	return a_ctz_l(y);
 }
 
-#if ((__ARM_ARCH_6__ || __ARM_ARCH_6K__ || __ARM_ARCH_6ZK__) && !__thumb__) \
- || __ARM_ARCH_7A__ || __ARM_ARCH_7R__ || __ARM_ARCH >= 7
-
 #if __ARM_ARCH_7A__ || __ARM_ARCH_7R__ ||  __ARM_ARCH >= 7
-#define MEM_BARRIER "dmb ish"
-#else
-#define MEM_BARRIER "mcr p15,0,r0,c7,c10,5"
-#endif
 
-static inline int __k_cas(int t, int s, volatile int *p)
+static inline void a_barrier()
 {
-	int ret;
-	__asm__(
-		"	" MEM_BARRIER "\n"
+	__asm__ __volatile__("dmb ish");
+}
+
+static inline int a_cas(volatile int *p, int t, int s)
+{
+	int old;
+	__asm__ __volatile__(
+		"	dmb ish\n"
 		"1:	ldrex %0,%3\n"
-		"	subs %0,%0,%1\n"
-#ifdef __thumb__
-		"	itt eq\n"
-#endif
-		"	strexeq %0,%2,%3\n"
-		"	teqeq %0,#1\n"
-		"	beq 1b\n"
-		"	" MEM_BARRIER "\n"
-		: "=&r"(ret)
+		"	cmp %0,%1\n"
+		"	bne 1f\n"
+		"	strex %0,%2,%3\n"
+		"	cmp %0, #0\n"
+		"	bne 1b\n"
+		"	mov %0, %1\n"
+		"1:	dmb ish\n"
+		: "=&r"(old)
 		: "r"(t), "r"(s), "Q"(*p)
 		: "memory", "cc" );
-	return ret;
+	return old;
+}
+
+static inline int a_swap(volatile int *x, int v)
+{
+	int old, tmp;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%3\n"
+		"	strex %1,%2,%3\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(old), "=&r"(tmp)
+		: "r"(v), "Q"(*x)
+		: "memory", "cc" );
+	return old;
+}
+
+static inline int a_fetch_add(volatile int *x, int v)
+{
+	int old, tmp;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%3\n"
+		"	add %0,%0,%2\n"
+		"	strex %1,%0,%3\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(old), "=&r"(tmp)
+		: "r"(v), "Q"(*x)
+		: "memory", "cc" );
+	return old-v;
+}
+
+static inline void a_inc(volatile int *x)
+{
+	int tmp, tmp2;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%2\n"
+		"	add %0,%0,#1\n"
+		"	strex %1,%0,%2\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(tmp), "=&r"(tmp2)
+		: "Q"(*x)
+		: "memory", "cc" );
+}
+
+static inline void a_dec(volatile int *x)
+{
+	int tmp, tmp2;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%2\n"
+		"	sub %0,%0,#1\n"
+		"	strex %1,%0,%2\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(tmp), "=&r"(tmp2)
+		: "Q"(*x)
+		: "memory", "cc" );
+}
+
+static inline void a_and(volatile int *x, int v)
+{
+	int tmp, tmp2;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%3\n"
+		"	and %0,%0,%2\n"
+		"	strex %1,%0,%3\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(tmp), "=&r"(tmp2)
+		: "r"(v), "Q"(*x)
+		: "memory", "cc" );
+}
+
+static inline void a_or(volatile int *x, int v)
+{
+	int tmp, tmp2;
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"1:	ldrex %0,%3\n"
+		"	orr %0,%0,%2\n"
+		"	strex %1,%0,%3\n"
+		"	cmp %1, #0\n"
+		"	bne 1b\n"
+		"	dmb ish\n"
+		: "=&r"(tmp), "=&r"(tmp2)
+		: "r"(v), "Q"(*x)
+		: "memory", "cc" );
+}
+
+static inline void a_store(volatile int *p, int x)
+{
+	__asm__ __volatile__(
+		"	dmb ish\n"
+		"	str %1,%0\n"
+		"	dmb ish\n"
+		: "=m"(*p)
+		: "r"(x)
+		: "memory", "cc" );
 }
+
 #else
-#define __k_cas ((int (*)(int, int, volatile int *))0xffff0fc0)
-#endif
+
+int __a_cas(int, int, volatile int *) __attribute__((__visibility__("hidden")));
+#define __k_cas __a_cas
+
+static inline void a_barrier()
+{
+	__asm__ __volatile__("bl __a_barrier"
+		: : : "memory", "cc", "ip", "lr" );
+}
 
 static inline int a_cas(volatile int *p, int t, int s)
 {
@@ -65,11 +178,6 @@ static inline int a_cas(volatile int *p,
 	}
 }
 
-static inline void *a_cas_p(volatile void *p, void *t, void *s)
-{
-	return (void *)a_cas(p, (int)t, (int)s);
-}
-
 static inline int a_swap(volatile int *x, int v)
 {
 	int old;
@@ -98,19 +206,9 @@ static inline void a_dec(volatile int *x
 
 static inline void a_store(volatile int *p, int x)
 {
-	while (__k_cas(*p, x, p));
-}
-
-#define a_spin a_barrier
-
-static inline void a_barrier()
-{
-	__k_cas(0, 0, &(int){0});
-}
-
-static inline void a_crash()
-{
-	*(volatile char *)0=0;
+	a_barrier();
+	*p = x;
+	a_barrier();
 }
 
 static inline void a_and(volatile int *p, int v)
@@ -127,6 +225,20 @@ static inline void a_or(volatile int *p,
 	while (__k_cas(old, old|v, p));
 }
 
+#endif
+
+static inline void *a_cas_p(volatile void *p, void *t, void *s)
+{
+	return (void *)a_cas(p, (int)t, (int)s);
+}
+
+#define a_spin a_barrier
+
+static inline void a_crash()
+{
+	*(volatile char *)0=0;
+}
+
 static inline void a_or_l(volatile void *p, long v)
 {
 	a_or(p, v);
--- a/arch/arm/bits/alltypes.h.in
+++ b/arch/arm/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF unsigned wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/arm/bits/syscall.h
+++ b/arch/arm/bits/syscall.h
@@ -339,7 +339,13 @@
 #define __NR_seccomp	383
 #define __NR_getrandom	384
 #define __NR_memfd_create	385
+#define __NR_bpf	386
 
+#define __ARM_NR_breakpoint	0x0f0001
+#define __ARM_NR_cacheflush	0x0f0002
+#define __ARM_NR_usr26		0x0f0003
+#define __ARM_NR_usr32		0x0f0004
+#define __ARM_NR_set_tls	0x0f0005
 
 /* Repeated with SYS_ prefix */
 
@@ -684,3 +690,4 @@
 #define SYS_seccomp	383
 #define SYS_getrandom	384
 #define SYS_memfd_create	385
+#define SYS_bpf	386
--- a/arch/arm/pthread_arch.h
+++ b/arch/arm/pthread_arch.h
@@ -10,9 +10,17 @@ static inline __attribute__((const)) pth
 
 #else
 
-typedef char *(*__ptr_func_t)(void) __attribute__((const));
-#define __pthread_self() \
-	((pthread_t)(((__ptr_func_t)0xffff0fe0)()+8-sizeof(struct pthread)))
+static inline __attribute__((const)) pthread_t __pthread_self()
+{
+#ifdef __clang__
+	char *p;
+	__asm__( "bl __a_gettp\n\tmov %0,r0" : "=r"(p) : : "cc", "r0", "lr" );
+#else
+	register char *p __asm__("r0");
+	__asm__( "bl __a_gettp" : "=r"(p) : : "cc", "lr" );
+#endif
+	return (void *)(p+8-sizeof(struct pthread));
+}
 
 #endif
 
--- /dev/null
+++ b/arch/arm/src/__set_thread_area.c
@@ -0,0 +1,49 @@
+#include <stdint.h>
+#include <elf.h>
+#include "pthread_impl.h"
+#include "libc.h"
+
+#define HWCAP_TLS (1 << 15)
+
+extern const unsigned char __attribute__((__visibility__("hidden")))
+	__a_barrier_dummy[], __a_barrier_oldkuser[],
+	__a_barrier_v6[], __a_barrier_v7[],
+	__a_cas_dummy[], __a_cas_v6[], __a_cas_v7[],
+	__a_gettp_dummy[];
+
+#define __a_barrier_kuser 0xffff0fa0
+#define __a_cas_kuser 0xffff0fc0
+#define __a_gettp_kuser 0xffff0fe0
+
+extern uintptr_t __attribute__((__visibility__("hidden")))
+	__a_barrier_ptr, __a_cas_ptr, __a_gettp_ptr;
+
+#define SET(op,ver) (__a_##op##_ptr = \
+	(uintptr_t)__a_##op##_##ver - (uintptr_t)__a_##op##_dummy)
+
+int __set_thread_area(void *p)
+{
+#if !__ARM_ARCH_7A__ && !__ARM_ARCH_7R__ && __ARM_ARCH < 7
+	if (__hwcap & HWCAP_TLS) {
+		size_t *aux;
+		SET(cas, v7);
+		SET(barrier, v7);
+		for (aux=libc.auxv; *aux; aux+=2) {
+			if (*aux != AT_PLATFORM) continue;
+			const char *s = (void *)aux[1];
+			if (s[0]!='v' || s[1]!='6' || s[2]-'0'<10u) break;
+			SET(cas, v6);
+			SET(barrier, v6);
+			break;
+		}
+	} else {
+		int ver = *(int *)0xffff0ffc;
+		SET(gettp, kuser);
+		SET(cas, kuser);
+		SET(barrier, kuser);
+		if (ver < 2) a_crash();
+		if (ver < 3) SET(barrier, oldkuser);
+	}
+#endif
+	return __syscall(0xf0005, p);
+}
--- /dev/null
+++ b/arch/arm/src/arm/atomics.s
@@ -0,0 +1,116 @@
+.text
+
+.global __a_barrier
+.hidden __a_barrier
+.type __a_barrier,%function
+__a_barrier:
+	ldr ip,1f
+	ldr ip,[pc,ip]
+	add pc,pc,ip
+1:	.word __a_barrier_ptr-1b
+.global __a_barrier_dummy
+.hidden __a_barrier_dummy
+__a_barrier_dummy:
+	tst lr,#1
+	moveq pc,lr
+	bx lr
+.global __a_barrier_oldkuser
+.hidden __a_barrier_oldkuser
+__a_barrier_oldkuser:
+	push {r0,r1,r2,r3,ip,lr}
+	mov r1,r0
+	mov r2,sp
+	ldr ip,=0xffff0fc0
+	mov lr,pc
+	mov pc,ip
+	pop {r0,r1,r2,r3,ip,lr}
+	tst lr,#1
+	moveq pc,lr
+	bx lr
+.global __a_barrier_v6
+.hidden __a_barrier_v6
+__a_barrier_v6:
+	mcr p15,0,r0,c7,c10,5
+	bx lr
+.global __a_barrier_v7
+.hidden __a_barrier_v7
+__a_barrier_v7:
+	.word 0xf57ff05b        /* dmb ish */
+	bx lr
+
+.global __a_cas
+.hidden __a_cas
+.type __a_cas,%function
+__a_cas:
+	ldr ip,1f
+	ldr ip,[pc,ip]
+	add pc,pc,ip
+1:	.word __a_cas_ptr-1b
+.global __a_cas_dummy
+.hidden __a_cas_dummy
+__a_cas_dummy:
+	mov r3,r0
+	ldr r0,[r2]
+	subs r0,r3,r0
+	streq r1,[r2]
+	tst lr,#1
+	moveq pc,lr
+	bx lr
+.global __a_cas_v6
+.hidden __a_cas_v6
+__a_cas_v6:
+	mov r3,r0
+	mcr p15,0,r0,c7,c10,5
+1:	.word 0xe1920f9f        /* ldrex r0,[r2] */
+	subs r0,r3,r0
+	.word 0x01820f91        /* strexeq r0,r1,[r2] */
+	teqeq r0,#1
+	beq 1b
+	mcr p15,0,r0,c7,c10,5
+	bx lr
+.global __a_cas_v7
+.hidden __a_cas_v7
+__a_cas_v7:
+	mov r3,r0
+	.word 0xf57ff05b        /* dmb ish */
+1:	.word 0xe1920f9f        /* ldrex r0,[r2] */
+	subs r0,r3,r0
+	.word 0x01820f91        /* strexeq r0,r1,[r2] */
+	teqeq r0,#1
+	beq 1b
+	.word 0xf57ff05b        /* dmb ish */
+	bx lr
+
+.global __aeabi_read_tp
+.type __aeabi_read_tp,%function
+__aeabi_read_tp:
+
+.global __a_gettp
+.hidden __a_gettp
+.type __a_gettp,%function
+__a_gettp:
+	ldr r0,1f
+	ldr r0,[pc,r0]
+	add pc,pc,r0
+1:	.word __a_gettp_ptr-1b
+.global __a_gettp_dummy
+.hidden __a_gettp_dummy
+__a_gettp_dummy:
+	mrc p15,0,r0,c13,c0,3
+	bx lr
+
+.data
+.global __a_barrier_ptr
+.hidden __a_barrier_ptr
+__a_barrier_ptr:
+	.word 0
+
+.global __a_cas_ptr
+.hidden __a_cas_ptr
+__a_cas_ptr:
+	.word 0
+
+.global __a_gettp_ptr
+.hidden __a_gettp_ptr
+__a_gettp_ptr:
+	.word 0
--- a/arch/arm/syscall_arch.h
+++ b/arch/arm/syscall_arch.h
@@ -5,8 +5,6 @@
 
 long (__syscall)(long, ...);
 
-#ifndef __clang__
-
 #define __asm_syscall(...) do { \
 	__asm__ __volatile__ ( "svc 0" \
 	: "=r"(r0) : __VA_ARGS__ : "memory"); \
@@ -54,41 +52,25 @@ static inline long __syscall4(long n, lo
 	__asm_syscall("r"(r7), "0"(r0), "r"(r1), "r"(r2), "r"(r3));
 }
 
-#else
-
-static inline long __syscall0(long n)
-{
-	return (__syscall)(n);
-}
-
-static inline long __syscall1(long n, long a)
-{
-	return (__syscall)(n, a);
-}
-
-static inline long __syscall2(long n, long a, long b)
-{
-	return (__syscall)(n, a, b);
-}
-
-static inline long __syscall3(long n, long a, long b, long c)
-{
-	return (__syscall)(n, a, b, c);
-}
-
-static inline long __syscall4(long n, long a, long b, long c, long d)
-{
-	return (__syscall)(n, a, b, c, d);
-}
-
-#endif
-
 static inline long __syscall5(long n, long a, long b, long c, long d, long e)
 {
-	return (__syscall)(n, a, b, c, d, e);
+	register long r7 __asm__("r7") = n;
+	register long r0 __asm__("r0") = a;
+	register long r1 __asm__("r1") = b;
+	register long r2 __asm__("r2") = c;
+	register long r3 __asm__("r3") = d;
+	register long r4 __asm__("r4") = e;
+	__asm_syscall("r"(r7), "0"(r0), "r"(r1), "r"(r2), "r"(r3), "r"(r4));
 }
 
 static inline long __syscall6(long n, long a, long b, long c, long d, long e, long f)
 {
-	return (__syscall)(n, a, b, c, d, e, f);
+	register long r7 __asm__("r7") = n;
+	register long r0 __asm__("r0") = a;
+	register long r1 __asm__("r1") = b;
+	register long r2 __asm__("r2") = c;
+	register long r3 __asm__("r3") = d;
+	register long r4 __asm__("r4") = e;
+	register long r5 __asm__("r5") = f;
+	__asm_syscall("r"(r7), "0"(r0), "r"(r1), "r"(r2), "r"(r3), "r"(r4), "r"(r5));
 }
--- a/arch/i386/bits/alltypes.h.in
+++ b/arch/i386/bits/alltypes.h.in
@@ -17,7 +17,6 @@ TYPEDEF __WCHAR_TYPE__ wchar_t;
 TYPEDEF long wchar_t;
 #endif
 #endif
-TYPEDEF unsigned wint_t;
 
 #if defined(__FLT_EVAL_METHOD__) && __FLT_EVAL_METHOD__ == 0
 TYPEDEF float float_t;
--- a/arch/i386/bits/syscall.h
+++ b/arch/i386/bits/syscall.h
@@ -355,6 +355,7 @@
 #define __NR_seccomp		354
 #define __NR_getrandom		355
 #define __NR_memfd_create	356
+#define __NR_bpf		357
 
 
 /* Repeated with SYS_ prefix */
@@ -716,3 +717,4 @@
 #define SYS_seccomp		354
 #define SYS_getrandom		355
 #define SYS_memfd_create	356
+#define SYS_bpf			357
--- a/arch/microblaze/bits/alltypes.h.in
+++ b/arch/microblaze/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF int wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/microblaze/bits/syscall.h
+++ b/arch/microblaze/bits/syscall.h
@@ -381,6 +381,7 @@
 #define __NR_seccomp 384
 #define __NR_getrandom 385
 #define __NR_memfd_create 386
+#define __NR_bpf 387
 
 /* Repeated with SYS_ prefix */
 
@@ -768,3 +769,4 @@
 #define SYS_seccomp 384
 #define SYS_getrandom 385
 #define SYS_memfd_create 386
+#define SYS_bpf 387
--- a/arch/microblaze/syscall_arch.h
+++ b/arch/microblaze/syscall_arch.h
@@ -100,39 +100,7 @@ static inline long __syscall6(long n, lo
 
 #else
 
-static inline long __syscall0(long n)
-{
-	return (__syscall)(n);
-}
-
-static inline long __syscall1(long n, long a)
-{
-	return (__syscall)(n, a);
-}
-
-static inline long __syscall2(long n, long a, long b)
-{
-	return (__syscall)(n, a, b);
-}
-
-static inline long __syscall3(long n, long a, long b, long c)
-{
-	return (__syscall)(n, a, b, c);
-}
-
-static inline long __syscall4(long n, long a, long b, long c, long d)
-{
-	return (__syscall)(n, a, b, c, d);
-}
-
-static inline long __syscall5(long n, long a, long b, long c, long d, long e)
-{
-	return (__syscall)(n, a, b, c, d, e);
-}
-
-static inline long __syscall6(long n, long a, long b, long c, long d, long e, long f)
-{
-	return (__syscall)(n, a, b, c, d, e, f);
-}
+#undef SYSCALL_NO_INLINE
+#define SYSCALL_NO_INLINE
 
 #endif
--- a/arch/mips/bits/alltypes.h.in
+++ b/arch/mips/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF int wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/mips/bits/syscall.h
+++ b/arch/mips/bits/syscall.h
@@ -352,6 +352,7 @@
 #define __NR_seccomp                 4352
 #define __NR_getrandom               4353
 #define __NR_memfd_create            4354
+#define __NR_bpf                     4355
 
 
 /* Repeated with SYS_ prefix */
@@ -709,3 +710,4 @@
 #define SYS_seccomp                 4352
 #define SYS_getrandom               4353
 #define SYS_memfd_create            4354
+#define SYS_bpf                     4355
--- a/arch/or1k/bits/alltypes.h.in
+++ b/arch/or1k/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF unsigned wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/or1k/bits/syscall.h
+++ b/arch/or1k/bits/syscall.h
@@ -263,6 +263,7 @@
 #define __NR_seccomp 277
 #define __NR_getrandom 278
 #define __NR_memfd_create 279
+#define __NR_bpf 280
 
 #define SYS_io_setup __NR_io_setup
 #define SYS_io_destroy __NR_io_destroy
@@ -529,3 +530,4 @@
 #define SYS_seccomp __NR_seccomp
 #define SYS_getrandom __NR_getrandom
 #define SYS_memfd_create __NR_memfd_create
+#define SYS_bpf __NR_bpf
--- a/arch/or1k/syscall_arch.h
+++ b/arch/or1k/syscall_arch.h
@@ -1,7 +1,7 @@
 #define __SYSCALL_LL_E(x) \
 ((union { long long ll; long l[2]; }){ .ll = x }).l[0], \
 ((union { long long ll; long l[2]; }){ .ll = x }).l[1]
-#define __SYSCALL_LL_O(x) 0, __SYSCALL_LL_E((x))
+#define __SYSCALL_LL_O(x) __SYSCALL_LL_E((x))
 
 #define SYSCALL_MMAP2_UNIT 8192ULL
 
@@ -118,39 +118,7 @@ static inline long __syscall6(long n, lo
 
 #else
 
-static inline long __syscall0(long n)
-{
-	return (__syscall)(n);
-}
-
-static inline long __syscall1(long n, long a)
-{
-	return (__syscall)(n, a);
-}
-
-static inline long __syscall2(long n, long a, long b)
-{
-	return (__syscall)(n, a, b);
-}
-
-static inline long __syscall3(long n, long a, long b, long c)
-{
-	return (__syscall)(n, a, b, c);
-}
-
-static inline long __syscall4(long n, long a, long b, long c, long d)
-{
-	return (__syscall)(n, a, b, c, d);
-}
-
-static inline long __syscall5(long n, long a, long b, long c, long d, long e)
-{
-	return (__syscall)(n, a, b, c, d, e);
-}
-
-static inline long __syscall6(long n, long a, long b, long c, long d, long e, long f)
-{
-	return (__syscall)(n, a, b, c, d, e, f);
-}
+#undef SYSCALL_NO_INLINE
+#define SYSCALL_NO_INLINE
 
 #endif
--- a/arch/powerpc/bits/alltypes.h.in
+++ b/arch/powerpc/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF long wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/powerpc/bits/syscall.h
+++ b/arch/powerpc/bits/syscall.h
@@ -374,6 +374,7 @@
 #define __NR_seccomp               358
 #define __NR_getrandom             359
 #define __NR_memfd_create          360
+#define __NR_bpf                   361
 
 /*
  * repeated with SYS prefix
@@ -754,3 +755,4 @@
 #define SYS_seccomp               358
 #define SYS_getrandom             359
 #define SYS_memfd_create          360
+#define SYS_bpf                   361
--- a/arch/powerpc/syscall_arch.h
+++ b/arch/powerpc/syscall_arch.h
@@ -3,39 +3,5 @@
 ((union { long long ll; long l[2]; }){ .ll = x }).l[1]
 #define __SYSCALL_LL_O(x) 0, __SYSCALL_LL_E((x))
 
-long (__syscall)(long, ...);
-
-static inline long __syscall0(long n)
-{
-	return (__syscall)(n, 0, 0, 0, 0, 0, 0);
-}
-
-static inline long __syscall1(long n, long a)
-{
-	return (__syscall)(n, a, 0, 0, 0, 0, 0);
-}
-
-static inline long __syscall2(long n, long a, long b)
-{
-	return (__syscall)(n, a, b, 0, 0, 0, 0);
-}
-
-static inline long __syscall3(long n, long a, long b, long c)
-{
-	return (__syscall)(n, a, b, c, 0, 0, 0);
-}
-
-static inline long __syscall4(long n, long a, long b, long c, long d)
-{
-	return (__syscall)(n, a, b, c, d, 0, 0);
-}
-
-static inline long __syscall5(long n, long a, long b, long c, long d, long e)
-{
-	return (__syscall)(n, a, b, c, d, e, 0);
-}
-
-static inline long __syscall6(long n, long a, long b, long c, long d, long e, long f)
-{
-	return (__syscall)(n, a, b, c, d, e, f);
-}
+#undef SYSCALL_NO_INLINE
+#define SYSCALL_NO_INLINE
--- a/arch/sh/bits/alltypes.h.in
+++ b/arch/sh/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF long wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 TYPEDEF float float_t;
 TYPEDEF double double_t;
--- a/arch/x32/bits/alltypes.h.in
+++ b/arch/x32/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF long wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 #if defined(__FLT_EVAL_METHOD__) && __FLT_EVAL_METHOD__ == 2
 TYPEDEF long double float_t;
--- a/arch/x32/bits/syscall.h
+++ b/arch/x32/bits/syscall.h
@@ -277,6 +277,8 @@
 #define __NR_seccomp (__X32_SYSCALL_BIT + 317)
 #define __NR_getrandom (__X32_SYSCALL_BIT + 318)
 #define __NR_memfd_create (__X32_SYSCALL_BIT + 319)
+#define __NR_kexec_file_load (__X32_SYSCALL_BIT + 320)
+#define __NR_bpf (__X32_SYSCALL_BIT + 321)
 
 #define __NR_rt_sigaction (__X32_SYSCALL_BIT + 512)
 #define __NR_rt_sigreturn (__X32_SYSCALL_BIT + 513)
@@ -604,6 +606,8 @@
 #define SYS_seccomp __NR_seccomp
 #define SYS_getrandom __NR_getrandom
 #define SYS_memfd_create __NR_memfd_create
+#define SYS_kexec_file_load __NR_kexec_file_load
+#define SYS_bpf __NR_bpf
 
 #define SYS_rt_sigaction __NR_rt_sigaction
 #define SYS_rt_sigreturn __NR_rt_sigreturn
--- a/arch/x86_64/bits/alltypes.h.in
+++ b/arch/x86_64/bits/alltypes.h.in
@@ -8,7 +8,6 @@ TYPEDEF __builtin_va_list __isoc_va_list
 #ifndef __cplusplus
 TYPEDEF int wchar_t;
 #endif
-TYPEDEF unsigned wint_t;
 
 #if defined(__FLT_EVAL_METHOD__) && __FLT_EVAL_METHOD__ == 2
 TYPEDEF long double float_t;
--- a/arch/x86_64/bits/syscall.h
+++ b/arch/x86_64/bits/syscall.h
@@ -318,6 +318,8 @@
 #define __NR_seccomp				317
 #define __NR_getrandom				318
 #define __NR_memfd_create			319
+#define __NR_kexec_file_load			320
+#define __NR_bpf				321
 
 
 #undef __NR_fstatat
@@ -654,6 +656,8 @@
 #define SYS_seccomp				317
 #define SYS_getrandom				318
 #define SYS_memfd_create			319
+#define SYS_kexec_file_load			320
+#define SYS_bpf					321
 
 #undef SYS_fstatat
 #undef SYS_pread
--- a/include/alltypes.h.in
+++ b/include/alltypes.h.in
@@ -28,6 +28,7 @@ TYPEDEF _Int64 blkcnt_t;
 TYPEDEF unsigned _Int64 fsblkcnt_t;
 TYPEDEF unsigned _Int64 fsfilcnt_t;
 
+TYPEDEF unsigned wint_t;
 TYPEDEF unsigned long wctype_t;
 
 TYPEDEF void * timer_t;
--- a/include/arpa/nameser.h
+++ b/include/arpa/nameser.h
@@ -1,6 +1,11 @@
 #ifndef _ARPA_NAMESER_H
 #define _ARPA_NAMESER_H
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stddef.h>
 #include <stdint.h>
 
 #define __NAMESER	19991006
@@ -48,6 +53,8 @@ extern const struct _ns_flagdata _ns_fla
 #define ns_msg_end(handle) ((handle)._eom + 0)
 #define ns_msg_size(handle) ((handle)._eom - (handle)._msg)
 #define ns_msg_count(handle, section) ((handle)._counts[section] + 0)
+#define ns_msg_getflag(handle, flag) \
+	(((handle)._flags & _ns_flagdata[flag].mask) >> _ns_flagdata[flag].shift)
 
 typedef	struct __ns_rr {
 	char		name[NS_MAXDNAME];
@@ -296,43 +303,20 @@ typedef enum __ns_cert_types {
 #define NS_OPT_DNSSEC_OK        0x8000U
 #define NS_OPT_NSID		3
 
-#define NS_GET16(s, cp) do { \
-	register const unsigned char *t_cp = (const unsigned char *)(cp); \
-	(s) = ((uint16_t)t_cp[0] << 8) \
-	    | ((uint16_t)t_cp[1]) \
-	    ; \
-	(cp) += NS_INT16SZ; \
-} while (0)
-
-#define NS_GET32(l, cp) do { \
-	register const unsigned char *t_cp = (const unsigned char *)(cp); \
-	(l) = ((uint32_t)t_cp[0] << 24) \
-	    | ((uint32_t)t_cp[1] << 16) \
-	    | ((uint32_t)t_cp[2] << 8) \
-	    | ((uint32_t)t_cp[3]) \
-	    ; \
-	(cp) += NS_INT32SZ; \
-} while (0)
-
-#define NS_PUT16(s, cp) do { \
-	register uint16_t t_s = (uint16_t)(s); \
-	register unsigned char *t_cp = (unsigned char *)(cp); \
-	*t_cp++ = t_s >> 8; \
-	*t_cp   = t_s; \
-	(cp) += NS_INT16SZ; \
-} while (0)
-
-#define NS_PUT32(l, cp) do { \
-	register uint32_t t_l = (uint32_t)(l); \
-	register unsigned char *t_cp = (unsigned char *)(cp); \
-	*t_cp++ = t_l >> 24; \
-	*t_cp++ = t_l >> 16; \
-	*t_cp++ = t_l >> 8; \
-	*t_cp   = t_l; \
-	(cp) += NS_INT32SZ; \
-} while (0)
-
-
+#define NS_GET16(s, cp) (void)((s) = ns_get16(((cp)+=2)-2))
+#define NS_GET32(l, cp) (void)((l) = ns_get32(((cp)+=4)-4))
+#define NS_PUT16(s, cp) ns_put16((s), ((cp)+=2)-2)
+#define NS_PUT32(l, cp) ns_put32((l), ((cp)+=4)-4)
+
+unsigned ns_get16(const unsigned char *);
+unsigned long ns_get32(const unsigned char *);
+void ns_put16(unsigned, unsigned char *);
+void ns_put32(unsigned long, unsigned char *);
+
+int ns_initparse(const unsigned char *, int, ns_msg *);
+int ns_parserr(ns_msg *, ns_sect, int, ns_rr *);
+int ns_skiprr(const unsigned char *, const unsigned char *, ns_sect, int);
+int ns_name_uncompress(const unsigned char *, const unsigned char *, const unsigned char *, char *, size_t);
 
 
 #define	__BIND		19950621
@@ -464,4 +448,8 @@ typedef struct {
 #define	PUTSHORT		NS_PUT16
 #define	PUTLONG			NS_PUT32
 
+#ifdef __cplusplus
+}
+#endif
+
 #endif
--- a/include/complex.h
+++ b/include/complex.h
@@ -7,9 +7,9 @@ extern "C" {
 
 #define complex _Complex
 #ifdef __GNUC__
-#define _Complex_I (__extension__ 1.0fi)
+#define _Complex_I (__extension__ (0.0f+1.0fi))
 #else
-#define _Complex_I 1.0fi
+#define _Complex_I (0.0f+1.0fi)
 #endif
 #define I _Complex_I
 
@@ -101,8 +101,9 @@ double creal(double complex);
 float crealf(float complex);
 long double creall(long double complex);
 
+#ifndef __cplusplus
 #define __CIMAG(x, t) \
-	((union { _Complex t __z; t __xy[2]; }){(_Complex t)(x)}.__xy[1])
+	(+(union { _Complex t __z; t __xy[2]; }){(_Complex t)(x)}.__xy[1])
 
 #define creal(x) ((double)(x))
 #define crealf(x) ((float)(x))
@@ -111,13 +112,20 @@ long double creall(long double complex);
 #define cimag(x) __CIMAG(x, double)
 #define cimagf(x) __CIMAG(x, float)
 #define cimagl(x) __CIMAG(x, long double)
+#endif
 
-#define __CMPLX(x, y, t) \
-	((union { _Complex t __z; t __xy[2]; }){.__xy = {(x),(y)}}.__z)
-
+#if __STDC_VERSION__ >= 201112L
+#if defined(_Imaginary_I)
+#define __CMPLX(x, y, t) ((t)(x) + _Imaginary_I*(t)(y)))
+#elif defined(__clang__)
+#define __CMPLX(x, y, t) (+(_Complex t){ (t)(x), (t)(y) })
+#else
+#define __CMPLX(x, y, t) (__builtin_complex((t)(x), (t)(y)))
+#endif
 #define CMPLX(x, y) __CMPLX(x, y, double)
 #define CMPLXF(x, y) __CMPLX(x, y, float)
 #define CMPLXL(x, y) __CMPLX(x, y, long double)
+#endif
 
 #ifdef __cplusplus
 }
--- a/include/pthread.h
+++ b/include/pthread.h
@@ -84,7 +84,9 @@ __attribute__((const))
 pthread_t pthread_self(void);
 
 int pthread_equal(pthread_t, pthread_t);
+#ifndef __cplusplus
 #define pthread_equal(x,y) ((x)==(y))
+#endif
 
 int pthread_setcancelstate(int, int *);
 int pthread_setcanceltype(int, int *);
--- a/include/stdint.h
+++ b/include/stdint.h
@@ -47,8 +47,8 @@ typedef uint64_t uint_least64_t;
 
 #define UINT8_MAX  (0xff)
 #define UINT16_MAX (0xffff)
-#define UINT32_MAX (0xffffffff)
-#define UINT64_MAX (0xffffffffffffffff)
+#define UINT32_MAX (0xffffffffu)
+#define UINT64_MAX (0xffffffffffffffffu)
 
 #define INT_FAST8_MIN   INT8_MIN
 #define INT_FAST64_MIN  INT64_MIN
@@ -78,7 +78,7 @@ typedef uint64_t uint_least64_t;
 #define INTMAX_MAX  INT64_MAX
 #define UINTMAX_MAX UINT64_MAX
 
-#define WINT_MIN 0
+#define WINT_MIN 0U
 #define WINT_MAX UINT32_MAX
 
 #if L'\0'-1 > 0
--- a/include/sys/prctl.h
+++ b/include/sys/prctl.h
@@ -5,6 +5,8 @@
 extern "C" {
 #endif
 
+#include <stdint.h>
+
 #define PR_SET_PDEATHSIG  1
 #define PR_GET_PDEATHSIG  2
 #define PR_GET_DUMPABLE   3
@@ -80,6 +82,25 @@ extern "C" {
 #define PR_SET_MM_ENV_END              11
 #define PR_SET_MM_AUXV                 12
 #define PR_SET_MM_EXE_FILE             13
+#define PR_SET_MM_MAP                  14
+#define PR_SET_MM_MAP_SIZE             15
+
+struct prctl_mm_map {
+	uint64_t start_code;
+	uint64_t end_code;
+	uint64_t start_data;
+	uint64_t end_data;
+	uint64_t start_brk;
+	uint64_t brk;
+	uint64_t start_stack;
+	uint64_t arg_start;
+	uint64_t arg_end;
+	uint64_t env_start;
+	uint64_t env_end;
+	uint64_t *auxv;
+	uint32_t auxv_size;
+	uint32_t exe_fd;
+};
 
 #define PR_SET_PTRACER 0x59616d61
 #define PR_SET_PTRACER_ANY (-1UL)
--- a/include/threads.h
+++ b/include/threads.h
@@ -51,7 +51,9 @@ void thrd_yield(void);
 
 thrd_t thrd_current(void);
 int thrd_equal(thrd_t, thrd_t);
+#ifndef __cplusplus
 #define thrd_equal(A, B) ((A) == (B))
+#endif
 
 void call_once(once_flag *, void (*)(void));
 
--- a/include/utmp.h
+++ b/include/utmp.h
@@ -35,6 +35,8 @@ void         setutent(void);
 
 void updwtmp(const char *, const struct utmp *);
 
+int login_tty(int);
+
 #define _PATH_UTMP "/dev/null/utmp"
 #define _PATH_WTMP "/dev/null/wtmp"
 
--- a/src/fcntl/open.c
+++ b/src/fcntl/open.c
@@ -7,7 +7,7 @@ int open(const char *filename, int flags
 {
 	mode_t mode = 0;
 
-	if (flags & O_CREAT) {
+	if ((flags & O_CREAT) || (flags & O_TMPFILE) == O_TMPFILE) {
 		va_list ap;
 		va_start(ap, flags);
 		mode = va_arg(ap, mode_t);
--- a/src/fcntl/openat.c
+++ b/src/fcntl/openat.c
@@ -5,11 +5,15 @@
 
 int openat(int fd, const char *filename, int flags, ...)
 {
-	mode_t mode;
-	va_list ap;
-	va_start(ap, flags);
-	mode = va_arg(ap, mode_t);
-	va_end(ap);
+	mode_t mode = 0;
+
+	if ((flags & O_CREAT) || (flags & O_TMPFILE) == O_TMPFILE) {
+		va_list ap;
+		va_start(ap, flags);
+		mode = va_arg(ap, mode_t);
+		va_end(ap);
+	}
+
 	return syscall_cp(SYS_openat, fd, filename, flags|O_LARGEFILE, mode);
 }
 
--- a/src/internal/libm.h
+++ b/src/internal/libm.h
@@ -128,6 +128,18 @@ do {                                    
   (d) = __u.f;                                    \
 } while (0)
 
+#undef __CMPLX
+#undef CMPLX
+#undef CMPLXF
+#undef CMPLXL
+
+#define __CMPLX(x, y, t) \
+	((union { _Complex t __z; t __xy[2]; }){.__xy = {(x),(y)}}.__z)
+
+#define CMPLX(x, y) __CMPLX(x, y, double)
+#define CMPLXF(x, y) __CMPLX(x, y, float)
+#define CMPLXL(x, y) __CMPLX(x, y, long double)
+
 /* fdlibm kernel functions */
 
 int    __rem_pio2_large(double*,double*,int,int,int);
--- a/src/internal/syscall.h
+++ b/src/internal/syscall.h
@@ -24,12 +24,22 @@ long __syscall_ret(unsigned long), __sys
 	__syscall_cp(syscall_arg_t, syscall_arg_t, syscall_arg_t, syscall_arg_t,
 	             syscall_arg_t, syscall_arg_t, syscall_arg_t);
 
+#ifdef SYSCALL_NO_INLINE
+#define __syscall0(n) (__syscall)(n)
+#define __syscall1(n,a) (__syscall)(n,__scc(a))
+#define __syscall2(n,a,b) (__syscall)(n,__scc(a),__scc(b))
+#define __syscall3(n,a,b,c) (__syscall)(n,__scc(a),__scc(b),__scc(c))
+#define __syscall4(n,a,b,c,d) (__syscall)(n,__scc(a),__scc(b),__scc(c),__scc(d))
+#define __syscall5(n,a,b,c,d,e) (__syscall)(n,__scc(a),__scc(b),__scc(c),__scc(d),__scc(e))
+#define __syscall6(n,a,b,c,d,e,f) (__syscall)(n,__scc(a),__scc(b),__scc(c),__scc(d),__scc(e),__scc(f))
+#else
 #define __syscall1(n,a) __syscall1(n,__scc(a))
 #define __syscall2(n,a,b) __syscall2(n,__scc(a),__scc(b))
 #define __syscall3(n,a,b,c) __syscall3(n,__scc(a),__scc(b),__scc(c))
 #define __syscall4(n,a,b,c,d) __syscall4(n,__scc(a),__scc(b),__scc(c),__scc(d))
 #define __syscall5(n,a,b,c,d,e) __syscall5(n,__scc(a),__scc(b),__scc(c),__scc(d),__scc(e))
 #define __syscall6(n,a,b,c,d,e,f) __syscall6(n,__scc(a),__scc(b),__scc(c),__scc(d),__scc(e),__scc(f))
+#endif
 #define __syscall7(n,a,b,c,d,e,f,g) (__syscall)(n,__scc(a),__scc(b),__scc(c),__scc(d),__scc(e),__scc(f),__scc(g))
 
 #define __SYSCALL_NARGS_X(a,b,c,d,e,f,g,h,n,...) n
--- a/src/ldso/dynlink.c
+++ b/src/ldso/dynlink.c
@@ -645,6 +645,8 @@ static void decode_dyn(struct dso *p)
 		p->hashtab = (void *)(p->base + dyn[DT_HASH]);
 	if (dyn[0]&(1<<DT_RPATH))
 		p->rpath_orig = (void *)(p->strings + dyn[DT_RPATH]);
+	if (dyn[0]&(1<<DT_RUNPATH))
+		p->rpath_orig = (void *)(p->strings + dyn[DT_RUNPATH]);
 	if (search_vec(p->dynv, dyn, DT_GNU_HASH))
 		p->ghashtab = (void *)(p->base + *dyn);
 	if (search_vec(p->dynv, dyn, DT_VERSYM))
@@ -1126,6 +1128,7 @@ void *__dynlink(int argc, char **argv)
 		libc.secure = 1;
 	}
 	libc.page_size = aux[AT_PAGESZ];
+	libc.auxv = auxv;
 
 	/* If the dynamic linker was invoked as a program itself, AT_BASE
 	 * will not be set. In that case, we assume the base address is
--- a/src/math/__rem_pio2.c
+++ b/src/math/__rem_pio2.c
@@ -19,6 +19,12 @@
 
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+
 /*
  * invpio2:  53 bits of 2/pi
  * pio2_1:   first  33 bit of pi/2
@@ -29,6 +35,7 @@
  * pio2_3t:  pi/2 - (pio2_1+pio2_2+pio2_3)
  */
 static const double
+toint   = 1.5/EPS,
 invpio2 = 6.36619772367581382433e-01, /* 0x3FE45F30, 0x6DC9C883 */
 pio2_1  = 1.57079632673412561417e+00, /* 0x3FF921FB, 0x54400000 */
 pio2_1t = 6.07710050650619224932e-11, /* 0x3DD0B461, 0x1A626331 */
@@ -41,8 +48,8 @@ pio2_3t = 8.47842766036889956997e-32; /*
 int __rem_pio2(double x, double *y)
 {
 	union {double f; uint64_t i;} u = {x};
-	double_t z,w,t,r;
-	double tx[3],ty[2],fn;
+	double_t z,w,t,r,fn;
+	double tx[3],ty[2];
 	uint32_t ix;
 	int sign, n, ex, ey, i;
 
@@ -111,8 +118,7 @@ int __rem_pio2(double x, double *y)
 	if (ix < 0x413921fb) {  /* |x| ~< 2^20*(pi/2), medium size */
 medium:
 		/* rint(x/(pi/2)), Assume round-to-nearest. */
-		fn = x*invpio2 + 0x1.8p52;
-		fn = fn - 0x1.8p52;
+		fn = x*invpio2 + toint - toint;
 		n = (int32_t)fn;
 		r = x - fn*pio2_1;
 		w = fn*pio2_1t;  /* 1st round, good to 85 bits */
--- a/src/math/__rem_pio2f.c
+++ b/src/math/__rem_pio2f.c
@@ -22,12 +22,19 @@
 
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+
 /*
  * invpio2:  53 bits of 2/pi
  * pio2_1:   first 25 bits of pi/2
  * pio2_1t:  pi/2 - pio2_1
  */
 static const double
+toint   = 1.5/EPS,
 invpio2 = 6.36619772367581382433e-01, /* 0x3FE45F30, 0x6DC9C883 */
 pio2_1  = 1.57079631090164184570e+00, /* 0x3FF921FB, 0x50000000 */
 pio2_1t = 1.58932547735281966916e-08; /* 0x3E5110b4, 0x611A6263 */
@@ -35,7 +42,8 @@ pio2_1t = 1.58932547735281966916e-08; /*
 int __rem_pio2f(float x, double *y)
 {
 	union {float f; uint32_t i;} u = {x};
-	double tx[1],ty[1],fn;
+	double tx[1],ty[1];
+	double_t fn;
 	uint32_t ix;
 	int n, sign, e0;
 
@@ -43,8 +51,7 @@ int __rem_pio2f(float x, double *y)
 	/* 25+53 bit pi is good enough for medium size */
 	if (ix < 0x4dc90fdb) {  /* |x| ~< 2^28*(pi/2), medium size */
 		/* Use a specialized rint() to get fn.  Assume round-to-nearest. */
-		fn = x*invpio2 + 0x1.8p52;
-		fn = fn - 0x1.8p52;
+		fn = x*invpio2 + toint - toint;
 		n  = (int32_t)fn;
 		*y = x - fn*pio2_1 - fn*pio2_1t;
 		return n;
--- a/src/math/__rem_pio2l.c
+++ b/src/math/__rem_pio2l.c
@@ -20,10 +20,11 @@
  * use __rem_pio2_large() for large x
  */
 
+static const long double toint = 1.5/LDBL_EPSILON;
+
 #if LDBL_MANT_DIG == 64
 /* u ~< 0x1p25*pi/2 */
 #define SMALL(u) (((u.i.se & 0x7fffU)<<16 | u.i.m>>48) < ((0x3fff + 25)<<16 | 0x921f>>1 | 0x8000))
-#define TOINT 0x1.8p63
 #define QUOBITS(x) ((uint32_t)(int32_t)x & 0x7fffffff)
 #define ROUND1 22
 #define ROUND2 61
@@ -50,7 +51,6 @@ pio2_3t = -2.75299651904407171810e-37L; 
 #elif LDBL_MANT_DIG == 113
 /* u ~< 0x1p45*pi/2 */
 #define SMALL(u) (((u.i.se & 0x7fffU)<<16 | u.i.top) < ((0x3fff + 45)<<16 | 0x921f))
-#define TOINT 0x1.8p112
 #define QUOBITS(x) ((uint32_t)(int64_t)x & 0x7fffffff)
 #define ROUND1 51
 #define ROUND2 119
@@ -77,7 +77,7 @@ int __rem_pio2l(long double x, long doub
 	ex = u.i.se & 0x7fff;
 	if (SMALL(u)) {
 		/* rint(x/(pi/2)), Assume round-to-nearest. */
-		fn = x*invpio2 + TOINT - TOINT;
+		fn = x*invpio2 + toint - toint;
 		n = QUOBITS(fn);
 		r = x-fn*pio2_1;
 		w = fn*pio2_1t;  /* 1st round good to 102/180 bits (ld80/ld128) */
--- a/src/math/ceil.c
+++ b/src/math/ceil.c
@@ -1,5 +1,12 @@
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const double_t toint = 1/EPS;
+
 double ceil(double x)
 {
 	union {double f; uint64_t i;} u = {x};
@@ -10,9 +17,9 @@ double ceil(double x)
 		return x;
 	/* y = int(x) - x, where int(x) is an integer neighbor of x */
 	if (u.i >> 63)
-		y = (double)(x - 0x1p52) + 0x1p52 - x;
+		y = x - toint + toint - x;
 	else
-		y = (double)(x + 0x1p52) - 0x1p52 - x;
+		y = x + toint - toint - x;
 	/* special case because of non-nearest rounding modes */
 	if (e <= 0x3ff-1) {
 		FORCE_EVAL(y);
--- a/src/math/ceill.c
+++ b/src/math/ceill.c
@@ -6,11 +6,9 @@ long double ceill(long double x)
 	return ceil(x);
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double ceill(long double x)
 {
 	union ldshape u = {x};
@@ -21,9 +19,9 @@ long double ceill(long double x)
 		return x;
 	/* y = int(x) - x, where int(x) is an integer neighbor of x */
 	if (u.i.se >> 15)
-		y = x - TOINT + TOINT - x;
+		y = x - toint + toint - x;
 	else
-		y = x + TOINT - TOINT - x;
+		y = x + toint - toint - x;
 	/* special case because of non-nearest rounding modes */
 	if (e <= 0x3fff-1) {
 		FORCE_EVAL(y);
--- a/src/math/floor.c
+++ b/src/math/floor.c
@@ -1,5 +1,12 @@
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const double_t toint = 1/EPS;
+
 double floor(double x)
 {
 	union {double f; uint64_t i;} u = {x};
@@ -10,9 +17,9 @@ double floor(double x)
 		return x;
 	/* y = int(x) - x, where int(x) is an integer neighbor of x */
 	if (u.i >> 63)
-		y = (double)(x - 0x1p52) + 0x1p52 - x;
+		y = x - toint + toint - x;
 	else
-		y = (double)(x + 0x1p52) - 0x1p52 - x;
+		y = x + toint - toint - x;
 	/* special case because of non-nearest rounding modes */
 	if (e <= 0x3ff-1) {
 		FORCE_EVAL(y);
--- a/src/math/floorl.c
+++ b/src/math/floorl.c
@@ -6,11 +6,9 @@ long double floorl(long double x)
 	return floor(x);
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double floorl(long double x)
 {
 	union ldshape u = {x};
@@ -21,9 +19,9 @@ long double floorl(long double x)
 		return x;
 	/* y = int(x) - x, where int(x) is an integer neighbor of x */
 	if (u.i.se >> 15)
-		y = x - TOINT + TOINT - x;
+		y = x - toint + toint - x;
 	else
-		y = x + TOINT - TOINT - x;
+		y = x + toint - toint - x;
 	/* special case because of non-nearest rounding modes */
 	if (e <= 0x3fff-1) {
 		FORCE_EVAL(y);
--- a/src/math/i386/fmod.s
+++ b/src/math/i386/fmod.s
@@ -4,7 +4,7 @@ fmod:
 	fldl 12(%esp)
 	fldl 4(%esp)
 1:	fprem
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/fmodf.s
+++ b/src/math/i386/fmodf.s
@@ -4,7 +4,7 @@ fmodf:
 	flds 8(%esp)
 	flds 4(%esp)
 1:	fprem
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/fmodl.s
+++ b/src/math/i386/fmodl.s
@@ -4,7 +4,7 @@ fmodl:
 	fldt 16(%esp)
 	fldt 4(%esp)
 1:	fprem
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/remainder.s
+++ b/src/math/i386/remainder.s
@@ -7,7 +7,7 @@ drem:
 	fldl 12(%esp)
 	fldl 4(%esp)
 1:	fprem1
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/remainderf.s
+++ b/src/math/i386/remainderf.s
@@ -7,7 +7,7 @@ dremf:
 	flds 8(%esp)
 	flds 4(%esp)
 1:	fprem1
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/remainderl.s
+++ b/src/math/i386/remainderl.s
@@ -4,7 +4,7 @@ remainderl:
 	fldt 16(%esp)
 	fldt 4(%esp)
 1:	fprem1
-	fstsw %ax
+	fnstsw %ax
 	sahf
 	jp 1b
 	fstp %st(1)
--- a/src/math/i386/sqrt.s
+++ b/src/math/i386/sqrt.s
@@ -2,7 +2,7 @@
 .type sqrt,@function
 sqrt:	fldl 4(%esp)
 	fsqrt
-	fstsw %ax
+	fnstsw %ax
 	sub $12,%esp
 	fld %st(0)
 	fstpt (%esp)
--- a/src/math/modfl.c
+++ b/src/math/modfl.c
@@ -11,11 +11,9 @@ long double modfl(long double x, long do
 	return r;
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double modfl(long double x, long double *iptr)
 {
 	union ldshape u = {x};
@@ -40,7 +38,7 @@ long double modfl(long double x, long do
 
 	/* raises spurious inexact */
 	absx = s ? -x : x;
-	y = absx + TOINT - TOINT - absx;
+	y = absx + toint - toint - absx;
 	if (y == 0) {
 		*iptr = x;
 		return s ? -0.0 : 0.0;
--- a/src/math/rint.c
+++ b/src/math/rint.c
@@ -1,6 +1,14 @@
+#include <float.h>
 #include <math.h>
 #include <stdint.h>
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const double_t toint = 1/EPS;
+
 double rint(double x)
 {
 	union {double f; uint64_t i;} u = {x};
@@ -11,9 +19,9 @@ double rint(double x)
 	if (e >= 0x3ff+52)
 		return x;
 	if (s)
-		y = (double)(x - 0x1p52) + 0x1p52;
+		y = x - toint + toint;
 	else
-		y = (double)(x + 0x1p52) - 0x1p52;
+		y = x + toint - toint;
 	if (y == 0)
 		return s ? -0.0 : 0;
 	return y;
--- a/src/math/rintf.c
+++ b/src/math/rintf.c
@@ -1,6 +1,16 @@
+#include <float.h>
 #include <math.h>
 #include <stdint.h>
 
+#if FLT_EVAL_METHOD==0
+#define EPS FLT_EPSILON
+#elif FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const float_t toint = 1/EPS;
+
 float rintf(float x)
 {
 	union {float f; uint32_t i;} u = {x};
@@ -11,9 +21,9 @@ float rintf(float x)
 	if (e >= 0x7f+23)
 		return x;
 	if (s)
-		y = (float)(x - 0x1p23f) + 0x1p23f;
+		y = x - toint + toint;
 	else
-		y = (float)(x + 0x1p23f) - 0x1p23f;
+		y = x + toint - toint;
 	if (y == 0)
 		return s ? -0.0f : 0.0f;
 	return y;
--- a/src/math/rintl.c
+++ b/src/math/rintl.c
@@ -6,11 +6,9 @@ long double rintl(long double x)
 	return rint(x);
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double rintl(long double x)
 {
 	union ldshape u = {x};
@@ -21,9 +19,9 @@ long double rintl(long double x)
 	if (e >= 0x3fff+LDBL_MANT_DIG-1)
 		return x;
 	if (s)
-		y = x - TOINT + TOINT;
+		y = x - toint + toint;
 	else
-		y = x + TOINT - TOINT;
+		y = x + toint - toint;
 	if (y == 0)
 		return 0*x;
 	return y;
--- a/src/math/round.c
+++ b/src/math/round.c
@@ -1,5 +1,12 @@
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const double_t toint = 1/EPS;
+
 double round(double x)
 {
 	union {double f; uint64_t i;} u = {x};
@@ -12,10 +19,10 @@ double round(double x)
 		x = -x;
 	if (e < 0x3ff-1) {
 		/* raise inexact if x!=0 */
-		FORCE_EVAL(x + 0x1p52);
+		FORCE_EVAL(x + toint);
 		return 0*u.f;
 	}
-	y = (double)(x + 0x1p52) - 0x1p52 - x;
+	y = x + toint - toint - x;
 	if (y > 0.5)
 		y = y + x - 1;
 	else if (y <= -0.5)
--- a/src/math/roundf.c
+++ b/src/math/roundf.c
@@ -1,5 +1,14 @@
 #include "libm.h"
 
+#if FLT_EVAL_METHOD==0
+#define EPS FLT_EPSILON
+#elif FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+static const float_t toint = 1/EPS;
+
 float roundf(float x)
 {
 	union {float f; uint32_t i;} u = {x};
@@ -11,10 +20,10 @@ float roundf(float x)
 	if (u.i >> 31)
 		x = -x;
 	if (e < 0x7f-1) {
-		FORCE_EVAL(x + 0x1p23f);
+		FORCE_EVAL(x + toint);
 		return 0*u.f;
 	}
-	y = (float)(x + 0x1p23f) - 0x1p23f - x;
+	y = x + toint - toint - x;
 	if (y > 0.5f)
 		y = y + x - 1;
 	else if (y <= -0.5f)
--- a/src/math/roundl.c
+++ b/src/math/roundl.c
@@ -6,11 +6,9 @@ long double roundl(long double x)
 	return round(x);
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double roundl(long double x)
 {
 	union ldshape u = {x};
@@ -22,10 +20,10 @@ long double roundl(long double x)
 	if (u.i.se >> 15)
 		x = -x;
 	if (e < 0x3fff-1) {
-		FORCE_EVAL(x + TOINT);
+		FORCE_EVAL(x + toint);
 		return 0*u.f;
 	}
-	y = x + TOINT - TOINT - x;
+	y = x + toint - toint - x;
 	if (y > 0.5)
 		y = y + x - 1;
 	else if (y <= -0.5)
--- a/src/math/truncl.c
+++ b/src/math/truncl.c
@@ -6,11 +6,9 @@ long double truncl(long double x)
 	return trunc(x);
 }
 #elif (LDBL_MANT_DIG == 64 || LDBL_MANT_DIG == 113) && LDBL_MAX_EXP == 16384
-#if LDBL_MANT_DIG == 64
-#define TOINT 0x1p63
-#elif LDBL_MANT_DIG == 113
-#define TOINT 0x1p112
-#endif
+
+static const long double toint = 1/LDBL_EPSILON;
+
 long double truncl(long double x)
 {
 	union ldshape u = {x};
@@ -27,7 +25,7 @@ long double truncl(long double x)
 	/* y = int(|x|) - |x|, where int(|x|) is an integer neighbor of |x| */
 	if (s)
 		x = -x;
-	y = x + TOINT - TOINT - x;
+	y = x + toint - toint - x;
 	if (y > 0)
 		y -= 1;
 	x += y;
--- a/src/math/x32/exp2l.s
+++ b/src/math/x32/exp2l.s
@@ -6,9 +6,7 @@ expm1l:
 	fmulp
 	movl $0xc2820000,-4(%esp)
 	flds -4(%esp)
-	fucomp %st(1)
-	fnstsw %ax
-	sahf
+	fucomip %st(1)
 	fld1
 	jb 1f
 		# x*log2e <= -65, return -1 without underflow
@@ -17,11 +15,8 @@ expm1l:
 	ret
 1:	fld %st(1)
 	fabs
-	fucom %st(1)
-	fnstsw %ax
+	fucomip %st(1)
 	fstp %st(0)
-	fstp %st(0)
-	sahf
 	ja 1f
 	f2xm1
 	ret
@@ -53,9 +48,7 @@ exp2l:
 	fld %st(1)
 	fsub %st(1)
 	faddp
-	fucomp %st(1)
-	fnstsw
-	sahf
+	fucomip %st(1)
 	je 2f             # x - 0x1p63 + 0x1p63 == x
 	movl $1,(%esp)
 	flds (%esp)       # 0x1p-149
--- a/src/math/x32/fmodl.s
+++ b/src/math/x32/fmodl.s
@@ -4,8 +4,8 @@ fmodl:
 	fldt 24(%esp)
 	fldt 8(%esp)
 1:	fprem
-	fstsw %ax
-	sahf
-	jp 1b
+	fnstsw %ax
+	testb $4,%ah
+	jnz 1b
 	fstp %st(1)
 	ret
--- a/src/math/x32/remainderl.s
+++ b/src/math/x32/remainderl.s
@@ -4,8 +4,8 @@ remainderl:
 	fldt 24(%esp)
 	fldt 8(%esp)
 1:	fprem1
-	fstsw %ax
-	sahf
-	jp 1b
+	fnstsw %ax
+	testb $4,%ah
+	jnz 1b
 	fstp %st(1)
 	ret
--- a/src/math/x86_64/exp2l.s
+++ b/src/math/x86_64/exp2l.s
@@ -6,9 +6,7 @@ expm1l:
 	fmulp
 	movl $0xc2820000,-4(%rsp)
 	flds -4(%rsp)
-	fucomp %st(1)
-	fnstsw %ax
-	sahf
+	fucomip %st(1)
 	fld1
 	jb 1f
 		# x*log2e <= -65, return -1 without underflow
@@ -17,11 +15,8 @@ expm1l:
 	ret
 1:	fld %st(1)
 	fabs
-	fucom %st(1)
-	fnstsw %ax
+	fucomip %st(1)
 	fstp %st(0)
-	fstp %st(0)
-	sahf
 	ja 1f
 	f2xm1
 	ret
@@ -53,9 +48,7 @@ exp2l:
 	fld %st(1)
 	fsub %st(1)
 	faddp
-	fucomp %st(1)
-	fnstsw
-	sahf
+	fucomip %st(1)
 	je 2f             # x - 0x1p63 + 0x1p63 == x
 	movl $1,(%rsp)
 	flds (%rsp)       # 0x1p-149
--- a/src/math/x86_64/fmodl.s
+++ b/src/math/x86_64/fmodl.s
@@ -4,8 +4,8 @@ fmodl:
 	fldt 24(%rsp)
 	fldt 8(%rsp)
 1:	fprem
-	fstsw %ax
-	sahf
-	jp 1b
+	fnstsw %ax
+	testb $4,%ah
+	jnz 1b
 	fstp %st(1)
 	ret
--- a/src/math/x86_64/remainderl.s
+++ b/src/math/x86_64/remainderl.s
@@ -4,8 +4,8 @@ remainderl:
 	fldt 24(%rsp)
 	fldt 8(%rsp)
 1:	fprem1
-	fstsw %ax
-	sahf
-	jp 1b
+	fnstsw %ax
+	testb $4,%ah
+	jnz 1b
 	fstp %st(1)
 	ret
--- a/src/misc/forkpty.c
+++ b/src/misc/forkpty.c
@@ -1,38 +1,57 @@
 #include <pty.h>
+#include <utmp.h>
 #include <unistd.h>
-#include <sys/ioctl.h>
+#include <errno.h>
 #include <fcntl.h>
+#include <sys/wait.h>
+#include <pthread.h>
 
-int forkpty(int *m, char *name, const struct termios *tio, const struct winsize *ws)
+int forkpty(int *pm, char *name, const struct termios *tio, const struct winsize *ws)
 {
-	int s, t, i, istmp[3]={0};
-	pid_t pid;
+	int m, s, ec=0, p[2], cs;
+	pid_t pid=-1;
+	sigset_t set, oldset;
 
-	if (openpty(m, &s, name, tio, ws) < 0) return -1;
+	if (openpty(&m, &s, name, tio, ws) < 0) return -1;
 
-	/* Ensure before forking that we don't exceed fd limit */
-	for (i=0; i<3; i++) {
-		if (fcntl(i, F_GETFL) < 0) {
-			t = fcntl(s, F_DUPFD, i);
-			if (t<0) break;
-			else if (t!=i) close(t);
-			else istmp[i] = 1;
-		}
+	sigfillset(&set);
+	pthread_sigmask(SIG_BLOCK, &set, &oldset);
+	pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cs);
+
+	if (pipe2(p, O_CLOEXEC)) {
+		close(s);
+		goto out;
 	}
-	pid = i==3 ? fork() : -1;
+
+	pid = fork();
 	if (!pid) {
-		close(*m);
-		setsid();
-		ioctl(s, TIOCSCTTY, (char *)0);
-		dup2(s, 0);
-		dup2(s, 1);
-		dup2(s, 2);
-		if (s>2) close(s);
+		close(m);
+		close(p[0]);
+		if (login_tty(s)) {
+			write(p[1], &errno, sizeof errno);
+			_exit(127);
+		}
+		close(p[1]);
+		pthread_setcancelstate(cs, 0);
+		pthread_sigmask(SIG_SETMASK, &oldset, 0);
 		return 0;
 	}
-	for (i=0; i<3; i++)
-		if (istmp[i]) close(i);
 	close(s);
-	if (pid < 0) close(*m);
+	close(p[1]);
+	if (read(p[0], &ec, sizeof ec) > 0) {
+		int status;
+		waitpid(pid, &status, 0);
+		pid = -1;
+		errno = ec;
+	}
+	close(p[0]);
+
+out:
+	if (pid > 0) *pm = m;
+	else close(m);
+
+	pthread_setcancelstate(cs, 0);
+	pthread_sigmask(SIG_SETMASK, &oldset, 0);
+
 	return pid;
 }
--- a/src/misc/getopt.c
+++ b/src/misc/getopt.c
@@ -4,6 +4,7 @@
 #include <limits.h>
 #include <stdlib.h>
 #include "libc.h"
+#include "locale_impl.h"
 
 char *optarg;
 int optind=1, opterr=1, optopt, __optpos, __optreset=0;
@@ -11,6 +12,18 @@ int optind=1, opterr=1, optopt, __optpos
 #define optpos __optpos
 weak_alias(__optreset, optreset);
 
+void __getopt_msg(const char *a, const char *b, const char *c, size_t l)
+{
+	FILE *f = stderr;
+	b = __lctrans_cur(b);
+	flockfile(f);
+	fwrite(a, strlen(a), 1, f)
+	&& fwrite(b, strlen(b), 1, f)
+	&& fwrite(c, l, 1, f)
+	&& putc('\n', f);
+	funlockfile(f);
+}
+
 int getopt(int argc, char * const argv[], const char *optstring)
 {
 	int i;
@@ -24,8 +37,20 @@ int getopt(int argc, char * const argv[]
 		optind = 1;
 	}
 
-	if (optind >= argc || !argv[optind] || argv[optind][0] != '-' || !argv[optind][1])
+	if (optind >= argc || !argv[optind])
+		return -1;
+
+	if (argv[optind][0] != '-') {
+		if (optstring[0] == '-') {
+			optarg = argv[optind++];
+			return 1;
+		}
+		return -1;
+	}
+
+	if (!argv[optind][1])
 		return -1;
+
 	if (argv[optind][1] == '-' && !argv[optind][2])
 		return optind++, -1;
 
@@ -43,30 +68,31 @@ int getopt(int argc, char * const argv[]
 		optpos = 0;
 	}
 
-	for (i=0; (l = mbtowc(&d, optstring+i, MB_LEN_MAX)) && d!=c; i+=l>0?l:1);
+	if (optstring[0] == '-' || optstring[0] == '+')
+		optstring++;
+
+	i = 0;
+	d = 0;
+	do {
+		l = mbtowc(&d, optstring+i, MB_LEN_MAX);
+		if (l>0) i+=l; else i++;
+	} while (l && d != c);
 
 	if (d != c) {
-		if (optstring[0] != ':' && opterr) {
-			write(2, argv[0], strlen(argv[0]));
-			write(2, ": illegal option: ", 18);
-			write(2, optchar, k);
-			write(2, "\n", 1);
-		}
+		if (optstring[0] != ':' && opterr)
+			__getopt_msg(argv[0], ": unrecognized option: ", optchar, k);
 		return '?';
 	}
-	if (optstring[i+1] == ':') {
-		if (optind >= argc) {
+	if (optstring[i] == ':') {
+		if (optstring[i+1] == ':') optarg = 0;
+		else if (optind >= argc) {
 			if (optstring[0] == ':') return ':';
-			if (opterr) {
-				write(2, argv[0], strlen(argv[0]));
-				write(2, ": option requires an argument: ", 31);
-				write(2, optchar, k);
-				write(2, "\n", 1);
-			}
+			if (opterr) __getopt_msg(argv[0],
+				": option requires an argument: ",
+				optchar, k);
 			return '?';
 		}
-		if (optstring[i+2] == ':') optarg = 0;
-		if (optstring[i+2] != ':' || optpos) {
+		if (optstring[i+1] != ':' || optpos) {
 			optarg = argv[optind++] + optpos;
 			optpos = 0;
 		}
--- a/src/misc/getopt_long.c
+++ b/src/misc/getopt_long.c
@@ -2,37 +2,106 @@
 #include <stddef.h>
 #include <getopt.h>
 #include <stdio.h>
+#include <string.h>
 
 extern int __optpos, __optreset;
 
+static void permute(char *const *argv, int dest, int src)
+{
+	char **av = (char **)argv;
+	char *tmp = av[src];
+	int i;
+	for (i=src; i>dest; i--)
+		av[i] = av[i-1];
+	av[dest] = tmp;
+}
+
+void __getopt_msg(const char *, const char *, const char *, size_t);
+
+static int __getopt_long_core(int argc, char *const *argv, const char *optstring, const struct option *longopts, int *idx, int longonly);
+
 static int __getopt_long(int argc, char *const *argv, const char *optstring, const struct option *longopts, int *idx, int longonly)
 {
+	int ret, skipped, resumed;
+	const char *optstring2 = optstring + 1;
 	if (!optind || __optreset) {
 		__optreset = 0;
 		__optpos = 0;
 		optind = 1;
 	}
-	if (optind >= argc || !argv[optind] || argv[optind][0] != '-') return -1;
-	if ((longonly && argv[optind][1]) ||
-		(argv[optind][1] == '-' && argv[optind][2]))
-	{
+	if (optind >= argc || !argv[optind]) return -1;
+	skipped = optind;
+	if (optstring[0] != '+' && optstring[0] != '-') {
 		int i;
-		for (i=0; longopts[i].name; i++) {
+		for (i=optind; ; i++) {
+			if (i >= argc || !argv[i]) return -1;
+			if (argv[i][0] == '-' && argv[i][1]) break;
+		}
+		optind = i;
+		optstring2 = optstring;
+	}
+	resumed = optind;
+	ret = __getopt_long_core(argc, argv, optstring2, longopts, idx, longonly);
+	if (resumed > skipped) {
+		int i, cnt = optind-resumed;
+		for (i=0; i<cnt; i++)
+			permute(argv, skipped, optind-1);
+		optind = skipped + cnt;
+	}
+	return ret;
+}
+
+static int __getopt_long_core(int argc, char *const *argv, const char *optstring, const struct option *longopts, int *idx, int longonly)
+{
+
+	if (longopts && argv[optind][0] == '-' &&
+		((longonly && argv[optind][1]) ||
+		 (argv[optind][1] == '-' && argv[optind][2])))
+	{
+		int i, cnt, match;
+		char *opt;
+		for (cnt=i=0; longopts[i].name; i++) {
 			const char *name = longopts[i].name;
-			char *opt = argv[optind]+1;
+			opt = argv[optind]+1;
 			if (*opt == '-') opt++;
 			for (; *name && *name == *opt; name++, opt++);
-			if (*name || (*opt && *opt != '=')) continue;
+			if (*opt && *opt != '=') continue;
+			match = i;
+			if (!*name) {
+				cnt = 1;
+				break;
+			}
+			cnt++;
+		}
+		if (cnt==1) {
+			i = match;
+			optind++;
+			optopt = longopts[i].val;
 			if (*opt == '=') {
-				if (!longopts[i].has_arg) continue;
+				if (!longopts[i].has_arg) {
+					if (optstring[0] == ':' || !opterr)
+						return '?';
+					__getopt_msg(argv[0],
+						": option does not take an argument: ",
+						longopts[i].name,
+						strlen(longopts[i].name));
+					return '?';
+				}
 				optarg = opt+1;
 			} else {
 				if (longopts[i].has_arg == required_argument) {
-					if (!(optarg = argv[++optind]))
-						return ':';
+					if (!(optarg = argv[optind])) {
+						if (optstring[0] == ':' || !opterr)
+							return ':';
+						__getopt_msg(argv[0],
+							": option requires an argument: ",
+							longopts[i].name,
+							strlen(longopts[i].name));
+						return '?';
+					}
+					optind++;
 				} else optarg = NULL;
 			}
-			optind++;
 			if (idx) *idx = i;
 			if (longopts[i].flag) {
 				*longopts[i].flag = longopts[i].val;
@@ -41,6 +110,12 @@ static int __getopt_long(int argc, char 
 			return longopts[i].val;
 		}
 		if (argv[optind][1] == '-') {
+			if (optstring[0] != ':' && opterr)
+				__getopt_msg(argv[0], cnt ?
+					": option is ambiguous: " :
+					": unrecognized option: ",
+					argv[optind]+2,
+					strlen(argv[optind]+2));
 			optind++;
 			return '?';
 		}
--- /dev/null
+++ b/src/misc/login_tty.c
@@ -0,0 +1,14 @@
+#include <utmp.h>
+#include <sys/ioctl.h>
+#include <unistd.h>
+
+int login_tty(int fd)
+{
+	setsid();
+	if (ioctl(fd, TIOCSCTTY, (char *)0)) return -1;
+	dup2(fd, 0);
+	dup2(fd, 1);
+	dup2(fd, 2);
+	if (fd>2) close(fd);
+	return 0;
+}
--- a/src/misc/openpty.c
+++ b/src/misc/openpty.c
@@ -3,31 +3,38 @@
 #include <unistd.h>
 #include <pty.h>
 #include <stdio.h>
+#include <pthread.h>
 
 /* Nonstandard, but vastly superior to the standard functions */
 
-int openpty(int *m, int *s, char *name, const struct termios *tio, const struct winsize *ws)
+int openpty(int *pm, int *ps, char *name, const struct termios *tio, const struct winsize *ws)
 {
-	int n=0;
+	int m, s, n=0, cs;
 	char buf[20];
 
-	*m = open("/dev/ptmx", O_RDWR|O_NOCTTY);
-	if (*m < 0) return -1;
+	m = open("/dev/ptmx", O_RDWR|O_NOCTTY);
+	if (m < 0) return -1;
 
-	if (ioctl(*m, TIOCSPTLCK, &n) || ioctl (*m, TIOCGPTN, &n)) {
-		close(*m);
-		return -1;
-	}
+	pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cs);
+
+	if (ioctl(m, TIOCSPTLCK, &n) || ioctl (m, TIOCGPTN, &n))
+		goto fail;
 
 	if (!name) name = buf;
 	snprintf(name, sizeof buf, "/dev/pts/%d", n);
-	if ((*s = open(name, O_RDWR|O_NOCTTY)) < 0) {
-		close(*m);
-		return -1;
-	}
+	if ((s = open(name, O_RDWR|O_NOCTTY)) < 0)
+		goto fail;
+
+	if (tio) tcsetattr(s, TCSANOW, tio);
+	if (ws) ioctl(s, TIOCSWINSZ, ws);
 
-	if (tio) tcsetattr(*s, TCSANOW, tio);
-	if (ws) ioctl(*s, TIOCSWINSZ, ws);
+	*pm = m;
+	*ps = s;
 
+	pthread_setcancelstate(cs, 0);
 	return 0;
+fail:
+	close(m);
+	pthread_setcancelstate(cs, 0);
+	return -1;
 }
--- a/src/misc/syslog.c
+++ b/src/misc/syslog.c
@@ -46,8 +46,12 @@ void closelog(void)
 
 static void __openlog()
 {
-	log_fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0);
-	if (log_fd >= 0) connect(log_fd, (void *)&log_addr, sizeof log_addr);
+	int fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0);
+	if (fd < 0) return;
+	if (connect(fd, (void *)&log_addr, sizeof log_addr) < 0)
+		close(fd);
+	else
+		log_fd = fd;
 }
 
 void openlog(const char *ident, int opt, int facility)
--- a/src/multibyte/c16rtomb.c
+++ b/src/multibyte/c16rtomb.c
@@ -4,6 +4,8 @@
 
 size_t c16rtomb(char *restrict s, char16_t c16, mbstate_t *restrict ps)
 {
+	static unsigned internal_state;
+	if (!ps) ps = (void *)&internal_state;
 	unsigned *x = (unsigned *)ps;
 	wchar_t wc;
 
--- a/src/multibyte/mbrtoc16.c
+++ b/src/multibyte/mbrtoc16.c
@@ -3,6 +3,8 @@
 
 size_t mbrtoc16(char16_t *restrict pc16, const char *restrict s, size_t n, mbstate_t *restrict ps)
 {
+	static unsigned internal_state;
+	if (!ps) ps = (void *)&internal_state;
 	unsigned *pending = (unsigned *)ps;
 
 	if (!s) return mbrtoc16(0, "", 1, ps);
--- a/src/multibyte/mbrtoc32.c
+++ b/src/multibyte/mbrtoc32.c
@@ -3,6 +3,8 @@
 
 size_t mbrtoc32(char32_t *restrict pc32, const char *restrict s, size_t n, mbstate_t *restrict ps)
 {
+	static unsigned internal_state;
+	if (!ps) ps = (void *)&internal_state;
 	if (!s) return mbrtoc32(0, "", 1, ps);
 	wchar_t wc;
 	size_t ret = mbrtowc(&wc, s, n, ps);
--- a/src/multibyte/wcsnrtombs.c
+++ b/src/multibyte/wcsnrtombs.c
@@ -40,7 +40,7 @@ size_t wcsnrtombs(char *restrict dst, co
 		ws++; wn--;
 		/* safe - this loop runs fewer than sizeof(buf) times */
 		s+=l; n-=l;
-		cnt++;
+		cnt += l;
 	}
 	if (dst) *wcs = ws;
 	return cnt;
--- /dev/null
+++ b/src/network/ns_parse.c
@@ -0,0 +1,171 @@
+#define _BSD_SOURCE
+#include <errno.h>
+#include <stddef.h>
+#include <resolv.h>
+#include <arpa/nameser.h>
+
+const struct _ns_flagdata _ns_flagdata[16] = {
+	{ 0x8000, 15 },
+	{ 0x7800, 11 },
+	{ 0x0400, 10 },
+	{ 0x0200, 9 },
+	{ 0x0100, 8 },
+	{ 0x0080, 7 },
+	{ 0x0040, 6 },
+	{ 0x0020, 5 },
+	{ 0x0010, 4 },
+	{ 0x000f, 0 },
+	{ 0x0000, 0 },
+	{ 0x0000, 0 },
+	{ 0x0000, 0 },
+	{ 0x0000, 0 },
+	{ 0x0000, 0 },
+	{ 0x0000, 0 },
+};
+
+unsigned ns_get16(const unsigned char *cp)
+{
+	return cp[0]<<8 | cp[1];
+}
+
+unsigned long ns_get32(const unsigned char *cp)
+{
+	return (unsigned)cp[0]<<24 | cp[1]<<16 | cp[2]<<8 | cp[3];
+}
+
+void ns_put16(unsigned s, unsigned char *cp)
+{
+	*cp++ = s>>8;
+	*cp++ = s;
+}
+
+void ns_put32(unsigned long l, unsigned char *cp)
+{
+	*cp++ = l>>24;
+	*cp++ = l>>16;
+	*cp++ = l>>8;
+	*cp++ = l;
+}
+
+int ns_initparse(const unsigned char *msg, int msglen, ns_msg *handle)
+{
+	int i, r;
+
+	handle->_msg = msg;
+	handle->_eom = msg + msglen;
+	if (msglen < (2 + ns_s_max) * NS_INT16SZ) goto bad;
+	NS_GET16(handle->_id, msg);
+	NS_GET16(handle->_flags, msg);
+	for (i = 0; i < ns_s_max; i++) NS_GET16(handle->_counts[i], msg);
+	for (i = 0; i < ns_s_max; i++) {
+		if (handle->_counts[i]) {
+			handle->_sections[i] = msg;
+			r = ns_skiprr(msg, handle->_eom, i, handle->_counts[i]);
+			if (r < 0) return -1;
+			msg += r;
+		} else {
+			handle->_sections[i] = NULL;
+		}
+	}
+	if (msg != handle->_eom) goto bad;
+	handle->_sect = ns_s_max;
+	handle->_rrnum = -1;
+	handle->_msg_ptr = NULL;
+	return 0;
+bad:
+	errno = EMSGSIZE;
+	return -1;
+}
+
+int ns_skiprr(const unsigned char *ptr, const unsigned char *eom, ns_sect section, int count)
+{
+	const unsigned char *p = ptr;
+	int r;
+
+	while (count--) {
+		r = dn_skipname(p, eom);
+		if (r < 0) goto bad;
+		if (r + 2 * NS_INT16SZ > eom - p) goto bad;
+		p += r + 2 * NS_INT16SZ;
+		if (section != ns_s_qd) {
+			if (NS_INT32SZ + NS_INT16SZ > eom - p) goto bad;
+			p += NS_INT32SZ;
+			NS_GET16(r, p);
+			if (r > eom - p) goto bad;
+			p += r;
+		}
+	}
+	return ptr - p;
+bad:
+	errno = EMSGSIZE;
+	return -1;
+}
+
+int ns_parserr(ns_msg *handle, ns_sect section, int rrnum, ns_rr *rr)
+{
+	int r;
+
+	if (section < 0 || section >= ns_s_max) goto bad;
+	if (section != handle->_sect) {
+		handle->_sect = section;
+		handle->_rrnum = 0;
+		handle->_msg_ptr = handle->_sections[section];
+	}
+	if (rrnum == -1) rrnum = handle->_rrnum;
+	if (rrnum < 0 || rrnum >= handle->_counts[section]) goto bad;
+	if (rrnum < handle->_rrnum) {
+		handle->_rrnum = 0;
+		handle->_msg_ptr = handle->_sections[section];
+	}
+	if (rrnum > handle->_rrnum) {
+		r = ns_skiprr(handle->_msg_ptr, handle->_eom, section, rrnum - handle->_rrnum);
+		if (r < 0) return -1;
+		handle->_msg_ptr += r;
+		handle->_rrnum = rrnum;
+	}
+	r = ns_name_uncompress(handle->_msg, handle->_eom, handle->_msg_ptr, rr->name, NS_MAXDNAME);
+	if (r < 0) return -1;
+	handle->_msg_ptr += r;
+	if (2 * NS_INT16SZ > handle->_eom - handle->_msg_ptr) goto size;
+	NS_GET16(rr->type, handle->_msg_ptr);
+	NS_GET16(rr->rr_class, handle->_msg_ptr);
+	if (section != ns_s_qd) {
+		if (NS_INT32SZ + NS_INT16SZ > handle->_eom - handle->_msg_ptr) goto size;
+		NS_GET32(rr->ttl, handle->_msg_ptr);
+		NS_GET16(rr->rdlength, handle->_msg_ptr);
+		if (rr->rdlength > handle->_eom - handle->_msg_ptr) goto size;
+		rr->rdata = handle->_msg_ptr;
+		handle->_msg_ptr += rr->rdlength;
+	} else {
+		rr->ttl = 0;
+		rr->rdlength = 0;
+		rr->rdata = NULL;
+	}
+	handle->_rrnum++;
+	if (handle->_rrnum > handle->_counts[section]) {
+		handle->_sect = section + 1;
+		if (handle->_sect == ns_s_max) {
+			handle->_rrnum = -1;
+			handle->_msg_ptr = NULL;
+		} else {
+			handle->_rrnum = 0;
+		}
+	}
+	return 0;
+bad:
+	errno = ENODEV;
+	return -1;
+size:
+	errno = EMSGSIZE;
+	return -1;
+}
+
+int ns_name_uncompress(const unsigned char *msg, const unsigned char *eom,
+                       const unsigned char *src, char *dst, size_t dstsiz)
+{
+	int r;
+	r = dn_expand(msg, eom, src, dst, dstsiz);
+	if (r < 0) errno = EMSGSIZE;
+	return r;
+}
+
--- a/src/process/posix_spawn.c
+++ b/src/process/posix_spawn.c
@@ -102,8 +102,7 @@ static int child(void *args_vp)
 			}
 			switch(op->cmd) {
 			case FDOP_CLOSE:
-				if ((ret=__syscall(SYS_close, op->fd)))
-					goto fail;
+				__syscall(SYS_close, op->fd);
 				break;
 			case FDOP_DUP2:
 				if ((ret=__sys_dup2(op->srcfd, op->fd))<0)
@@ -137,7 +136,7 @@ static int child(void *args_vp)
 fail:
 	/* Since sizeof errno < PIPE_BUF, the write is atomic. */
 	ret = -ret;
-	if (ret) while (write(p, &ret, sizeof ret) < 0);
+	if (ret) while (__syscall(SYS_write, p, &ret, sizeof ret) < 0);
 	_exit(127);
 }
 
--- a/src/regex/fnmatch.c
+++ b/src/regex/fnmatch.c
@@ -97,7 +97,13 @@ escaped:
 	return pat[0];
 }
 
-static int match_bracket(const char *p, int k)
+static int casefold(int k)
+{
+	int c = towupper(k);
+	return c == k ? towlower(k) : c;
+}
+
+static int match_bracket(const char *p, int k, int kfold)
 {
 	wchar_t wc;
 	int inv = 0;
@@ -119,7 +125,10 @@ static int match_bracket(const char *p, 
 			wchar_t wc2;
 			int l = mbtowc(&wc2, p+1, 4);
 			if (l < 0) return 0;
-			if (wc<=wc2 && (unsigned)k-wc <= wc2-wc) return !inv;
+			if (wc <= wc2)
+				if ((unsigned)k-wc <= wc2-wc ||
+				    (unsigned)kfold-wc <= wc2-wc)
+					return !inv;
 			p += l-1;
 			continue;
 		}
@@ -132,7 +141,9 @@ static int match_bracket(const char *p, 
 				char buf[16];
 				memcpy(buf, p0, p-1-p0);
 				buf[p-1-p0] = 0;
-				if (iswctype(k, wctype(buf))) return !inv;
+				if (iswctype(k, wctype(buf)) ||
+				    iswctype(kfold, wctype(buf)))
+					return !inv;
 			}
 			continue;
 		}
@@ -143,7 +154,7 @@ static int match_bracket(const char *p, 
 			if (l < 0) return 0;
 			p += l-1;
 		}
-		if (wc==k) return !inv;
+		if (wc==k || wc==kfold) return !inv;
 	}
 	return inv;
 }
@@ -153,7 +164,7 @@ static int fnmatch_internal(const char *
 	const char *p, *ptail, *endpat;
 	const char *s, *stail, *endstr;
 	size_t pinc, sinc, tailcnt=0;
-	int c, k;
+	int c, k, kfold;
 
 	if (flags & FNM_PERIOD) {
 		if (*str == '.' && *pat != '.')
@@ -173,10 +184,11 @@ static int fnmatch_internal(const char *
 				return (c==END) ? 0 : FNM_NOMATCH;
 			str += sinc;
 			n -= sinc;
+			kfold = flags & FNM_CASEFOLD ? casefold(k) : k;
 			if (c == BRACKET) {
-				if (!match_bracket(pat, k))
+				if (!match_bracket(pat, k, kfold))
 					return FNM_NOMATCH;
-			} else if (c != QUESTION && k != c) {
+			} else if (c != QUESTION && k != c && kfold != c) {
 				return FNM_NOMATCH;
 			}
 			pat+=pinc;
@@ -233,10 +245,11 @@ static int fnmatch_internal(const char *
 			break;
 		}
 		s += sinc;
+		kfold = flags & FNM_CASEFOLD ? casefold(k) : k;
 		if (c == BRACKET) {
-			if (!match_bracket(p-pinc, k))
+			if (!match_bracket(p-pinc, k, kfold))
 				return FNM_NOMATCH;
-		} else if (c != QUESTION && k != c) {
+		} else if (c != QUESTION && k != c && kfold != c) {
 			return FNM_NOMATCH;
 		}
 	}
@@ -261,10 +274,11 @@ static int fnmatch_internal(const char *
 			k = str_next(s, endstr-s, &sinc);
 			if (!k)
 				return FNM_NOMATCH;
+			kfold = flags & FNM_CASEFOLD ? casefold(k) : k;
 			if (c == BRACKET) {
-				if (!match_bracket(p-pinc, k))
+				if (!match_bracket(p-pinc, k, kfold))
 					break;
-			} else if (c != QUESTION && k != c) {
+			} else if (c != QUESTION && k != c && kfold != c) {
 				break;
 			}
 			s += sinc;
--- a/src/sched/affinity.c
+++ b/src/sched/affinity.c
@@ -1,5 +1,6 @@
 #define _GNU_SOURCE
 #include <sched.h>
+#include <string.h>
 #include "pthread_impl.h"
 #include "syscall.h"
 
@@ -10,17 +11,23 @@ int sched_setaffinity(pid_t tid, size_t 
 
 int pthread_setaffinity_np(pthread_t td, size_t size, const cpu_set_t *set)
 {
-	return syscall(SYS_sched_setaffinity, td->tid, size, set);
+	return -__syscall(SYS_sched_setaffinity, td->tid, size, set);
 }
 
-int sched_getaffinity(pid_t tid, size_t size, cpu_set_t *set)
+static int do_getaffinity(pid_t tid, size_t size, cpu_set_t *set)
 {
 	long ret = __syscall(SYS_sched_getaffinity, tid, size, set);
-	if (ret > 0) ret = 0;
-	return __syscall_ret(ret);
+	if (ret < 0) return ret;
+	if (ret < size) memset((char *)set+ret, 0, size-ret);
+	return 0;
+}
+
+int sched_getaffinity(pid_t tid, size_t size, cpu_set_t *set)
+{
+	return __syscall_ret(do_getaffinity(tid, size, set));
 }
 
 int pthread_getaffinity_np(pthread_t td, size_t size, cpu_set_t *set)
 {
-	return sched_getaffinity(td->tid, size, set);
+	return -do_getaffinity(td->tid, size, set);
 }
--- a/src/setjmp/arm/longjmp.s
+++ b/src/setjmp/arm/longjmp.s
@@ -20,7 +20,7 @@ longjmp:
 	ldc p2, cr4, [ip], #48
 2:	tst r1,#0x40
 	beq 2f
-	ldc p11, cr8, [ip], #64
+	.word 0xecbc8b10 /* vldmia ip!, {d8-d15} */
 2:	tst r1,#0x200
 	beq 3f
 	ldcl p1, cr10, [ip], #8
--- a/src/setjmp/arm/setjmp.s
+++ b/src/setjmp/arm/setjmp.s
@@ -22,7 +22,7 @@ setjmp:
 	stc p2, cr4, [ip], #48
 2:	tst r1,#0x40
 	beq 2f
-	stc p11, cr8, [ip], #64
+	.word 0xecac8b10 /* vstmia ip!, {d8-d15} */
 2:	tst r1,#0x200
 	beq 3f
 	stcl p1, cr10, [ip], #8
--- a/src/signal/raise.c
+++ b/src/signal/raise.c
@@ -5,12 +5,11 @@
 
 int raise(int sig)
 {
-	int pid, tid, ret;
+	int tid, ret;
 	sigset_t set;
 	__block_app_sigs(&set);
 	tid = __syscall(SYS_gettid);
-	pid = __syscall(SYS_getpid);
-	ret = syscall(SYS_tgkill, pid, tid, sig);
+	ret = syscall(SYS_tkill, tid, sig);
 	__restore_sigs(&set);
 	return ret;
 }
--- a/src/stdio/vfprintf.c
+++ b/src/stdio/vfprintf.c
@@ -158,7 +158,7 @@ static void pop_arg(union arg *arg, int 
 
 static void out(FILE *f, const char *s, size_t l)
 {
-	__fwritex((void *)s, l, f);
+	if (!(f->flags & F_ERR)) __fwritex((void *)s, l, f);
 }
 
 static void pad(FILE *f, char c, int w, int l, int fl)
@@ -225,7 +225,7 @@ static int fmt_fp(FILE *f, long double y
 
 	if (!isfinite(y)) {
 		char *s = (t&32)?"inf":"INF";
-		if (y!=y) s=(t&32)?"nan":"NAN", pl=0;
+		if (y!=y) s=(t&32)?"nan":"NAN";
 		pad(f, ' ', w, 3+pl, fl&~ZERO_PAD);
 		out(f, prefix, pl);
 		out(f, s, 3);
@@ -570,7 +570,7 @@ static int printf_core(FILE *f, const ch
 			if (0) {
 		case 'o':
 			a = fmt_o(arg.i, z);
-			if ((fl&ALT_FORM) && arg.i) prefix+=5, pl=1;
+			if ((fl&ALT_FORM) && p<z-a+1) p=z-a+1;
 			} if (0) {
 		case 'd': case 'i':
 			pl=1;
@@ -656,6 +656,7 @@ int vfprintf(FILE *restrict f, const cha
 	int nl_type[NL_ARGMAX+1] = {0};
 	union arg nl_arg[NL_ARGMAX+1];
 	unsigned char internal_buf[80], *saved_buf = 0;
+	int olderr;
 	int ret;
 
 	/* the copy allows passing va_list* even if va_list is an array */
@@ -666,6 +667,8 @@ int vfprintf(FILE *restrict f, const cha
 	}
 
 	FLOCK(f);
+	olderr = f->flags & F_ERR;
+	if (f->mode < 1) f->flags &= ~F_ERR;
 	if (!f->buf_size) {
 		saved_buf = f->buf;
 		f->wpos = f->wbase = f->buf = internal_buf;
@@ -680,6 +683,8 @@ int vfprintf(FILE *restrict f, const cha
 		f->buf_size = 0;
 		f->wpos = f->wbase = f->wend = 0;
 	}
+	if (f->flags & F_ERR) ret = -1;
+	f->flags |= olderr;
 	FUNLOCK(f);
 	va_end(ap2);
 	return ret;
--- a/src/stdio/vfwprintf.c
+++ b/src/stdio/vfwprintf.c
@@ -149,7 +149,7 @@ static void pop_arg(union arg *arg, int 
 
 static void out(FILE *f, const wchar_t *s, size_t l)
 {
-	while (l--) fputwc(*s++, f);
+	while (l-- && !(f->flags & F_ERR)) fputwc(*s++, f);
 }
 
 static int getint(wchar_t **s) {
@@ -345,6 +345,7 @@ int vfwprintf(FILE *restrict f, const wc
 	va_list ap2;
 	int nl_type[NL_ARGMAX] = {0};
 	union arg nl_arg[NL_ARGMAX];
+	int olderr;
 	int ret;
 
 	/* the copy allows passing va_list* even if va_list is an array */
@@ -356,7 +357,11 @@ int vfwprintf(FILE *restrict f, const wc
 
 	FLOCK(f);
 	f->mode |= f->mode+1;
+	olderr = f->flags & F_ERR;
+	f->flags &= ~F_ERR;
 	ret = wprintf_core(f, fmt, &ap2, nl_arg, nl_type);
+	if (f->flags & F_ERR) ret = -1;
+	f->flags |= olderr;
 	FUNLOCK(f);
 	va_end(ap2);
 	return ret;
--- a/src/string/armel/memcpy.s
+++ b/src/string/armel/memcpy.s
@@ -49,113 +49,113 @@ memcpy:
 	 * ARM ABI. Since we have to save R0, we might as well save R4
 	 * which we can use for better pipelining of the reads below
 	 */
-        .fnstart
-        .save       {r0, r4, lr}
-        stmfd       sp!, {r0, r4, lr}
-        /* Making room for r5-r11 which will be spilled later */
-        .pad        #28
-        sub         sp, sp, #28
-
-        /* it simplifies things to take care of len<4 early */
-        cmp	r2, #4
-        blo	copy_last_3_and_return
-
-        /* compute the offset to align the source
-         * offset = (4-(src&3))&3 = -src & 3
-         */
-        rsb	r3, r1, #0
-        ands	r3, r3, #3
-        beq	src_aligned
-
-        /* align source to 32 bits. We need to insert 2 instructions between
-         * a ldr[b|h] and str[b|h] because byte and half-word instructions
-         * stall 2 cycles.
-         */
-        movs	r12, r3, lsl #31
-        sub	r2, r2, r3		/* we know that r3 <= r2 because r2 >= 4 */
-        ldrmib	r3, [r1], #1
-        ldrcsb	r4, [r1], #1
-        ldrcsb	r12,[r1], #1
-        strmib	r3, [r0], #1
-        strcsb	r4, [r0], #1
-        strcsb	r12,[r0], #1
+	.fnstart
+	.save       {r0, r4, lr}
+	stmfd       sp!, {r0, r4, lr}
+	/* Making room for r5-r11 which will be spilled later */
+	.pad        #28
+	sub         sp, sp, #28
+
+	/* it simplifies things to take care of len<4 early */
+	cmp     r2, #4
+	blo     copy_last_3_and_return
+
+	/* compute the offset to align the source
+	 * offset = (4-(src&3))&3 = -src & 3
+	 */
+	rsb     r3, r1, #0
+	ands    r3, r3, #3
+	beq     src_aligned
+
+	/* align source to 32 bits. We need to insert 2 instructions between
+	 * a ldr[b|h] and str[b|h] because byte and half-word instructions
+	 * stall 2 cycles.
+	 */
+	movs    r12, r3, lsl #31
+	sub     r2, r2, r3              /* we know that r3 <= r2 because r2 >= 4 */
+	.word 0x44d13001 /* ldrbmi r3, [r1], #1 */
+	.word 0x24d14001 /* ldrbcs r4, [r1], #1 */
+	.word 0x24d1c001 /* ldrbcs r12,[r1], #1 */
+	.word 0x44c03001 /* strbmi r3, [r0], #1 */
+	.word 0x24c04001 /* strbcs r4, [r0], #1 */
+	.word 0x24c0c001 /* strbcs r12,[r0], #1 */
 
 src_aligned:
 
 	/* see if src and dst are aligned together (congruent) */
-	eor	r12, r0, r1
-        tst	r12, #3
-        bne	non_congruent
-
-        /* Use post-incriment mode for stm to spill r5-r11 to reserved stack
-         * frame. Don't update sp.
-         */
-        stmea	sp, {r5-r11}
-
-        /* align the destination to a cache-line */
-        rsb	r3, r0, #0
-        ands	r3, r3, #0x1C
-        beq    	congruent_aligned32
-        cmp    	r3, r2
-        andhi	r3, r2, #0x1C
-
-        /* conditionnaly copies 0 to 7 words (length in r3) */
-        movs	r12, r3, lsl #28
-        ldmcsia	r1!, {r4, r5, r6, r7}	/* 16 bytes */
-        ldmmiia	r1!, {r8, r9}			/*  8 bytes */
-        stmcsia	r0!, {r4, r5, r6, r7}
-        stmmiia	r0!, {r8, r9}
-        tst    	r3, #0x4
-        ldrne	r10,[r1], #4			/*  4 bytes */
-        strne	r10,[r0], #4
-        sub    	r2, r2, r3
+	eor     r12, r0, r1
+	tst     r12, #3
+	bne     non_congruent
+
+	/* Use post-incriment mode for stm to spill r5-r11 to reserved stack
+	 * frame. Don't update sp.
+	 */
+	stmea   sp, {r5-r11}
+
+	/* align the destination to a cache-line */
+	rsb     r3, r0, #0
+	ands    r3, r3, #0x1C
+	beq     congruent_aligned32
+	cmp     r3, r2
+	andhi   r3, r2, #0x1C
+
+	/* conditionnaly copies 0 to 7 words (length in r3) */
+	movs    r12, r3, lsl #28
+	ldmcs   r1!, {r4, r5, r6, r7}           /* 16 bytes */
+	ldmmi   r1!, {r8, r9}                   /*  8 bytes */
+	stmcs   r0!, {r4, r5, r6, r7}
+	stmmi   r0!, {r8, r9}
+	tst     r3, #0x4
+	ldrne   r10,[r1], #4                    /*  4 bytes */
+	strne   r10,[r0], #4
+	sub     r2, r2, r3
 
 congruent_aligned32:
 	/*
- 	 * here source is aligned to 32 bytes.
- 	 */
+	 * here source is aligned to 32 bytes.
+	 */
 
 cached_aligned32:
-        subs   	r2, r2, #32
-        blo    	less_than_32_left
+	subs    r2, r2, #32
+	blo     less_than_32_left
 
-        /*
-         * We preload a cache-line up to 64 bytes ahead. On the 926, this will
-         * stall only until the requested world is fetched, but the linefill
-         * continues in the the background.
-         * While the linefill is going, we write our previous cache-line
-         * into the write-buffer (which should have some free space).
-         * When the linefill is done, the writebuffer will
-         * start dumping its content into memory
-         *
-         * While all this is going, we then load a full cache line into
-         * 8 registers, this cache line should be in the cache by now
-         * (or partly in the cache).
-         *
-         * This code should work well regardless of the source/dest alignment.
-         *
-         */
-
-        /* Align the preload register to a cache-line because the cpu does
-         * "critical word first" (the first word requested is loaded first).
-         */
-        @ bic    	r12, r1, #0x1F
-        @ add    	r12, r12, #64
-
-1:      ldmia  	r1!, { r4-r11 }
-        subs   	r2, r2, #32
-
-        /* 
-         * NOTE: if r12 is more than 64 ahead of r1, the following ldrhi
-         * for ARM9 preload will not be safely guarded by the preceding subs.
-         * When it is safely guarded the only possibility to have SIGSEGV here
-         * is because the caller overstates the length.
-         */
-        @ ldrhi  	r3, [r12], #32      /* cheap ARM9 preload */
-        stmia  	r0!, { r4-r11 }
-	bhs    	1b
+	/*
+	 * We preload a cache-line up to 64 bytes ahead. On the 926, this will
+	 * stall only until the requested world is fetched, but the linefill
+	 * continues in the the background.
+	 * While the linefill is going, we write our previous cache-line
+	 * into the write-buffer (which should have some free space).
+	 * When the linefill is done, the writebuffer will
+	 * start dumping its content into memory
+	 *
+	 * While all this is going, we then load a full cache line into
+	 * 8 registers, this cache line should be in the cache by now
+	 * (or partly in the cache).
+	 *
+	 * This code should work well regardless of the source/dest alignment.
+	 *
+	 */
 
-        add	r2, r2, #32
+	/* Align the preload register to a cache-line because the cpu does
+	 * "critical word first" (the first word requested is loaded first).
+	 */
+	@ bic           r12, r1, #0x1F
+	@ add           r12, r12, #64
+
+1:      ldmia   r1!, { r4-r11 }
+	subs    r2, r2, #32
+
+	/* 
+	 * NOTE: if r12 is more than 64 ahead of r1, the following ldrhi
+	 * for ARM9 preload will not be safely guarded by the preceding subs.
+	 * When it is safely guarded the only possibility to have SIGSEGV here
+	 * is because the caller overstates the length.
+	 */
+	@ ldrhi         r3, [r12], #32      /* cheap ARM9 preload */
+	stmia   r0!, { r4-r11 }
+	bhs     1b
+
+	add     r2, r2, #32
 
 less_than_32_left:
 	/*
@@ -166,30 +166,30 @@ less_than_32_left:
 	 * be a common case (if not executed the code below takes
 	 * about 16 cycles)
 	 */
-	tst	r2, #0x1F
-        beq	1f
+	tst     r2, #0x1F
+	beq     1f
 
-        /* conditionnaly copies 0 to 31 bytes */
-        movs	r12, r2, lsl #28
-        ldmcsia	r1!, {r4, r5, r6, r7}	/* 16 bytes */
-        ldmmiia	r1!, {r8, r9}			/*  8 bytes */
-        stmcsia	r0!, {r4, r5, r6, r7}
-        stmmiia	r0!, {r8, r9}
-        movs	r12, r2, lsl #30
-        ldrcs	r3, [r1], #4			/*  4 bytes */
-        ldrmih	r4, [r1], #2			/*  2 bytes */
-        strcs	r3, [r0], #4
-        strmih	r4, [r0], #2
-        tst    	r2, #0x1
-        ldrneb	r3, [r1]				/*  last byte  */
-        strneb	r3, [r0]
-
-        /* we're done! restore everything and return */
-1:	ldmfd	sp!, {r5-r11}
-        ldmfd	sp!, {r0, r4, lr}
-        tst	lr, #1
-        moveq	pc, lr
-        bx	lr
+	/* conditionnaly copies 0 to 31 bytes */
+	movs    r12, r2, lsl #28
+	ldmcs   r1!, {r4, r5, r6, r7}           /* 16 bytes */
+	ldmmi   r1!, {r8, r9}                   /*  8 bytes */
+	stmcs   r0!, {r4, r5, r6, r7}
+	stmmi   r0!, {r8, r9}
+	movs    r12, r2, lsl #30
+	ldrcs   r3, [r1], #4                    /*  4 bytes */
+	.word 0x40d140b2 /* ldrhmi r4, [r1], #2 */ /*  2 bytes */
+	strcs   r3, [r0], #4
+	.word 0x40c040b2 /* strhmi r4, [r0], #2 */
+	tst     r2, #0x1
+	.word 0x15d13000 /* ldrbne r3, [r1] */  /*  last byte  */
+	.word 0x15c03000 /* strbne r3, [r0] */
+
+	/* we're done! restore everything and return */
+1:      ldmfd   sp!, {r5-r11}
+	ldmfd   sp!, {r0, r4, lr}
+	tst     lr, #1
+	moveq   pc, lr
+	bx      lr
 
 	/********************************************************************/
 
@@ -202,180 +202,180 @@ non_congruent:
 	 * (the number of bytes written is always smaller, because we have
 	 * partial words in the shift queue)
 	 */
-	cmp	r2, #4
-	blo	copy_last_3_and_return
+	cmp     r2, #4
+	blo     copy_last_3_and_return
 
-        /* Use post-incriment mode for stm to spill r5-r11 to reserved stack
-         * frame. Don't update sp.
-         */
-        stmea	sp, {r5-r11}
-
-        /* compute shifts needed to align src to dest */
-        rsb	r5, r0, #0
-        and	r5, r5, #3			/* r5 = # bytes in partial words */
-        mov	r12, r5, lsl #3		/* r12 = right */
-        rsb	lr, r12, #32		/* lr = left  */
-
-        /* read the first word */
-        ldr	r3, [r1], #4
-        sub	r2, r2, #4
-
-        /* write a partial word (0 to 3 bytes), such that destination
-         * becomes aligned to 32 bits (r5 = nb of words to copy for alignment)
-         */
-        movs	r5, r5, lsl #31
-        strmib	r3, [r0], #1
-        movmi	r3, r3, lsr #8
-        strcsb	r3, [r0], #1
-        movcs	r3, r3, lsr #8
-        strcsb	r3, [r0], #1
-        movcs	r3, r3, lsr #8
-
-        cmp	r2, #4
-        blo	partial_word_tail
-
-        /* Align destination to 32 bytes (cache line boundary) */
-1:	tst	r0, #0x1c
-        beq	2f
-        ldr	r5, [r1], #4
-        sub    	r2, r2, #4
-        orr	r4, r3, r5,		lsl lr
-        mov	r3, r5,			lsr r12
-        str	r4, [r0], #4
-        cmp    	r2, #4
-        bhs	1b
-        blo	partial_word_tail
+	/* Use post-incriment mode for stm to spill r5-r11 to reserved stack
+	 * frame. Don't update sp.
+	 */
+	stmea   sp, {r5-r11}
+
+	/* compute shifts needed to align src to dest */
+	rsb     r5, r0, #0
+	and     r5, r5, #3                      /* r5 = # bytes in partial words */
+	mov     r12, r5, lsl #3         /* r12 = right */
+	rsb     lr, r12, #32            /* lr = left  */
+
+	/* read the first word */
+	ldr     r3, [r1], #4
+	sub     r2, r2, #4
+
+	/* write a partial word (0 to 3 bytes), such that destination
+	 * becomes aligned to 32 bits (r5 = nb of words to copy for alignment)
+	 */
+	movs    r5, r5, lsl #31
+	.word 0x44c03001 /* strbmi r3, [r0], #1 */
+	movmi   r3, r3, lsr #8
+	.word 0x24c03001 /* strbcs r3, [r0], #1 */
+	movcs   r3, r3, lsr #8
+	.word 0x24c03001 /* strbcs r3, [r0], #1 */
+	movcs   r3, r3, lsr #8
+
+	cmp     r2, #4
+	blo     partial_word_tail
+
+	/* Align destination to 32 bytes (cache line boundary) */
+1:      tst     r0, #0x1c
+	beq     2f
+	ldr     r5, [r1], #4
+	sub     r2, r2, #4
+	orr     r4, r3, r5,             lsl lr
+	mov     r3, r5,                 lsr r12
+	str     r4, [r0], #4
+	cmp     r2, #4
+	bhs     1b
+	blo     partial_word_tail
 
 	/* copy 32 bytes at a time */
-2:	subs	r2, r2, #32
-	blo	less_than_thirtytwo
+2:      subs    r2, r2, #32
+	blo     less_than_thirtytwo
+
+	/* Use immediate mode for the shifts, because there is an extra cycle
+	 * for register shifts, which could account for up to 50% of
+	 * performance hit.
+	 */
 
-        /* Use immediate mode for the shifts, because there is an extra cycle
-         * for register shifts, which could account for up to 50% of
-         * performance hit.
-         */
-
-        cmp	r12, #24
-        beq	loop24
-        cmp	r12, #8
-        beq	loop8
+	cmp     r12, #24
+	beq     loop24
+	cmp     r12, #8
+	beq     loop8
 
 loop16:
-        ldr    	r12, [r1], #4
-1:      mov    	r4, r12
-	ldmia	r1!, {   r5,r6,r7,  r8,r9,r10,r11}
-        subs   	r2, r2, #32
-        ldrhs  	r12, [r1], #4
-	orr	r3, r3, r4, lsl #16
-        mov	r4, r4, lsr #16
-        orr	r4, r4, r5, lsl #16
-        mov	r5, r5, lsr #16
-        orr	r5, r5, r6, lsl #16
-        mov	r6, r6, lsr #16
-        orr	r6, r6, r7, lsl #16
-        mov	r7, r7, lsr #16
-        orr	r7, r7, r8, lsl #16
-        mov	r8, r8, lsr #16
-        orr	r8, r8, r9, lsl #16
-        mov	r9, r9, lsr #16
-        orr	r9, r9, r10, lsl #16
-        mov	r10, r10,		lsr #16
-        orr	r10, r10, r11, lsl #16
-        stmia	r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
-        mov	r3, r11, lsr #16
-        bhs	1b
-        b	less_than_thirtytwo
+	ldr     r12, [r1], #4
+1:      mov     r4, r12
+	ldmia   r1!, {   r5,r6,r7,  r8,r9,r10,r11}
+	subs    r2, r2, #32
+	ldrhs   r12, [r1], #4
+	orr     r3, r3, r4, lsl #16
+	mov     r4, r4, lsr #16
+	orr     r4, r4, r5, lsl #16
+	mov     r5, r5, lsr #16
+	orr     r5, r5, r6, lsl #16
+	mov     r6, r6, lsr #16
+	orr     r6, r6, r7, lsl #16
+	mov     r7, r7, lsr #16
+	orr     r7, r7, r8, lsl #16
+	mov     r8, r8, lsr #16
+	orr     r8, r8, r9, lsl #16
+	mov     r9, r9, lsr #16
+	orr     r9, r9, r10, lsl #16
+	mov     r10, r10,               lsr #16
+	orr     r10, r10, r11, lsl #16
+	stmia   r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
+	mov     r3, r11, lsr #16
+	bhs     1b
+	b       less_than_thirtytwo
 
 loop8:
-        ldr    	r12, [r1], #4
-1:      mov    	r4, r12
-	ldmia	r1!, {   r5,r6,r7,  r8,r9,r10,r11}
-	subs	r2, r2, #32
-        ldrhs  	r12, [r1], #4
-        orr	r3, r3, r4, lsl #24
-        mov	r4, r4, lsr #8
-        orr	r4, r4, r5, lsl #24
-        mov	r5, r5, lsr #8
-        orr	r5, r5, r6, lsl #24
-        mov	r6, r6,	 lsr #8
-        orr	r6, r6, r7, lsl #24
-        mov	r7, r7,	 lsr #8
-        orr	r7, r7, r8,		lsl #24
-        mov	r8, r8,	 lsr #8
-        orr	r8, r8, r9,		lsl #24
-        mov	r9, r9,	 lsr #8
-        orr	r9, r9, r10,	lsl #24
-        mov	r10, r10, lsr #8
-        orr	r10, r10, r11,	lsl #24
-        stmia	r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
-        mov	r3, r11, lsr #8
-        bhs	1b
-        b	less_than_thirtytwo
+	ldr     r12, [r1], #4
+1:      mov     r4, r12
+	ldmia   r1!, {   r5,r6,r7,  r8,r9,r10,r11}
+	subs    r2, r2, #32
+	ldrhs   r12, [r1], #4
+	orr     r3, r3, r4, lsl #24
+	mov     r4, r4, lsr #8
+	orr     r4, r4, r5, lsl #24
+	mov     r5, r5, lsr #8
+	orr     r5, r5, r6, lsl #24
+	mov     r6, r6,  lsr #8
+	orr     r6, r6, r7, lsl #24
+	mov     r7, r7,  lsr #8
+	orr     r7, r7, r8,             lsl #24
+	mov     r8, r8,  lsr #8
+	orr     r8, r8, r9,             lsl #24
+	mov     r9, r9,  lsr #8
+	orr     r9, r9, r10,    lsl #24
+	mov     r10, r10, lsr #8
+	orr     r10, r10, r11,  lsl #24
+	stmia   r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
+	mov     r3, r11, lsr #8
+	bhs     1b
+	b       less_than_thirtytwo
 
 loop24:
-        ldr    	r12, [r1], #4
-1:      mov    	r4, r12
-	ldmia	r1!, {   r5,r6,r7,  r8,r9,r10,r11}
-        subs	r2, r2, #32
-        ldrhs  	r12, [r1], #4
-        orr	r3, r3, r4, lsl #8
-        mov	r4, r4, lsr #24
-        orr	r4, r4, r5, lsl #8
-        mov	r5, r5, lsr #24
-        orr	r5, r5, r6, lsl #8
-        mov	r6, r6, lsr #24
-        orr	r6, r6, r7, lsl #8
-        mov	r7, r7, lsr #24
-        orr	r7, r7, r8, lsl #8
-        mov	r8, r8, lsr #24
-        orr	r8, r8, r9, lsl #8
-        mov	r9, r9, lsr #24
-        orr	r9, r9, r10, lsl #8
-        mov	r10, r10, lsr #24
-        orr	r10, r10, r11, lsl #8
-        stmia	r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
-        mov	r3, r11, lsr #24
-        bhs	1b
+	ldr     r12, [r1], #4
+1:      mov     r4, r12
+	ldmia   r1!, {   r5,r6,r7,  r8,r9,r10,r11}
+	subs    r2, r2, #32
+	ldrhs   r12, [r1], #4
+	orr     r3, r3, r4, lsl #8
+	mov     r4, r4, lsr #24
+	orr     r4, r4, r5, lsl #8
+	mov     r5, r5, lsr #24
+	orr     r5, r5, r6, lsl #8
+	mov     r6, r6, lsr #24
+	orr     r6, r6, r7, lsl #8
+	mov     r7, r7, lsr #24
+	orr     r7, r7, r8, lsl #8
+	mov     r8, r8, lsr #24
+	orr     r8, r8, r9, lsl #8
+	mov     r9, r9, lsr #24
+	orr     r9, r9, r10, lsl #8
+	mov     r10, r10, lsr #24
+	orr     r10, r10, r11, lsl #8
+	stmia   r0!, {r3,r4,r5,r6, r7,r8,r9,r10}
+	mov     r3, r11, lsr #24
+	bhs     1b
 
 less_than_thirtytwo:
 	/* copy the last 0 to 31 bytes of the source */
-	rsb	r12, lr, #32		/* we corrupted r12, recompute it  */
-        add	r2, r2, #32
-        cmp	r2, #4
-        blo	partial_word_tail
-
-1:	ldr	r5, [r1], #4
-        sub    	r2, r2, #4
-        orr	r4, r3, r5,		lsl lr
-        mov	r3,	r5,			lsr r12
-        str	r4, [r0], #4
-        cmp    	r2, #4
-        bhs	1b
+	rsb     r12, lr, #32            /* we corrupted r12, recompute it  */
+	add     r2, r2, #32
+	cmp     r2, #4
+	blo     partial_word_tail
+
+1:      ldr     r5, [r1], #4
+	sub     r2, r2, #4
+	orr     r4, r3, r5,             lsl lr
+	mov     r3,     r5,                     lsr r12
+	str     r4, [r0], #4
+	cmp     r2, #4
+	bhs     1b
 
 partial_word_tail:
 	/* we have a partial word in the input buffer */
-	movs	r5, lr, lsl #(31-3)
-	strmib	r3, [r0], #1
-        movmi	r3, r3, lsr #8
-        strcsb	r3, [r0], #1
-        movcs	r3, r3, lsr #8
-        strcsb	r3, [r0], #1
+	movs    r5, lr, lsl #(31-3)
+	.word 0x44c03001 /* strbmi r3, [r0], #1 */
+	movmi   r3, r3, lsr #8
+	.word 0x24c03001 /* strbcs r3, [r0], #1 */
+	movcs   r3, r3, lsr #8
+	.word 0x24c03001 /* strbcs r3, [r0], #1 */
 
-        /* Refill spilled registers from the stack. Don't update sp. */
-        ldmfd	sp, {r5-r11}
+	/* Refill spilled registers from the stack. Don't update sp. */
+	ldmfd   sp, {r5-r11}
 
 copy_last_3_and_return:
-	movs	r2, r2, lsl #31	/* copy remaining 0, 1, 2 or 3 bytes */
-        ldrmib	r2, [r1], #1
-        ldrcsb	r3, [r1], #1
-        ldrcsb	r12,[r1]
-        strmib	r2, [r0], #1
-        strcsb	r3, [r0], #1
-        strcsb	r12,[r0]
-
-        /* we're done! restore sp and spilled registers and return */
-        add    	sp,  sp, #28
-        ldmfd	sp!, {r0, r4, lr}
-        tst	lr, #1
-        moveq	pc, lr
-        bx	lr
+	movs    r2, r2, lsl #31 /* copy remaining 0, 1, 2 or 3 bytes */
+	.word 0x44d12001 /* ldrbmi r2, [r1], #1 */
+	.word 0x24d13001 /* ldrbcs r3, [r1], #1 */
+	.word 0x25d1c000 /* ldrbcs r12,[r1] */
+	.word 0x44c02001 /* strbmi r2, [r0], #1 */
+	.word 0x24c03001 /* strbcs r3, [r0], #1 */
+	.word 0x25c0c000 /* strbcs r12,[r0] */
+
+	/* we're done! restore sp and spilled registers and return */
+	add     sp,  sp, #28
+	ldmfd   sp!, {r0, r4, lr}
+	tst     lr, #1
+	moveq   pc, lr
+	bx      lr
--- a/src/thread/arm/__set_thread_area.s
+++ b/src/thread/arm/__set_thread_area.s
@@ -1,12 +1 @@
-.text
-.global __set_thread_area
-.type   __set_thread_area,%function
-__set_thread_area:
-	mov r1,r7
-	mov r7,#0x0f0000
-	add r7,r7,#5
-	svc 0
-	mov r7,r1
-	tst lr,#1
-	moveq pc,lr
-	bx lr
+/* Replaced by C code in arch/arm/src */
--- a/src/thread/arm/tls.s
+++ /dev/null
@@ -1,4 +0,0 @@
-.global __aeabi_read_tp
-.type __aeabi_read_tp,%function
-__aeabi_read_tp:
-	ldr pc,=0xffff0fe0
--- a/src/thread/pthread_once.c
+++ b/src/thread/pthread_once.c
@@ -8,15 +8,8 @@ static void undo(void *control)
 		__wake(control, -1, 1);
 }
 
-int __pthread_once(pthread_once_t *control, void (*init)(void))
+int __pthread_once_full(pthread_once_t *control, void (*init)(void))
 {
-	/* Return immediately if init finished before, but ensure that
-	 * effects of the init routine are visible to the caller. */
-	if (*control == 2) {
-		a_barrier();
-		return 0;
-	}
-
 	/* Try to enter initializing state. Four possibilities:
 	 *  0 - we're the first or the other cancelled; run init
 	 *  1 - another thread is running init; wait
@@ -43,4 +36,15 @@ int __pthread_once(pthread_once_t *contr
 	}
 }
 
+int __pthread_once(pthread_once_t *control, void (*init)(void))
+{
+	/* Return immediately if init finished before, but ensure that
+	 * effects of the init routine are visible to the caller. */
+	if (*control == 2) {
+		a_barrier();
+		return 0;
+	}
+	return __pthread_once_full(control, init);
+}
+
 weak_alias(__pthread_once, pthread_once);