14 files changed, 243 insertions, 81 deletions

diff --git a/src/internal/aio_impl.h b/src/internal/aio_impl.h
new file mode 100644
index 00000000..a8657665
--- /dev/null
+++ b/src/internal/aio_impl.h
@@ -0,0 +1,9 @@
+#ifndef AIO_IMPL_H
+#define AIO_IMPL_H
+
+extern hidden volatile int __aio_fut;
+
+extern hidden int __aio_close(int);
+extern hidden void __aio_atfork(int);
+
+#endif
diff --git a/src/internal/atomic.h b/src/internal/atomic.h
index f938879b..8f71c8cd 100644
--- a/src/internal/atomic.h
+++ b/src/internal/atomic.h
@@ -194,7 +194,7 @@ static inline void a_store(volatile int *p, int v)
 
 #ifndef a_barrier
 #define a_barrier a_barrier
-static void a_barrier()
+static inline void a_barrier()
 {
 	volatile int tmp = 0;
 	a_cas(&tmp, 0, 0);
@@ -315,4 +315,19 @@ static inline int a_clz_64(uint64_t x)
 }
 #endif
 
+#ifndef a_clz_32
+#define a_clz_32 a_clz_32
+static inline int a_clz_32(uint32_t x)
+{
+	x >>= 1;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x++;
+	return 31-a_ctz_32(x);
+}
+#endif
+
 #endif
diff --git a/src/internal/dynlink.h b/src/internal/dynlink.h
index ffd06b04..40c743e2 100644
--- a/src/internal/dynlink.h
+++ b/src/internal/dynlink.h
@@ -73,6 +73,10 @@ struct fdpic_dummy_loadmap {
 #define DL_NOMMU_SUPPORT 0
 #endif
 
+#ifndef TLSDESC_BACKWARDS
+#define TLSDESC_BACKWARDS 0
+#endif
+
 #if !DL_FDPIC
 #define IS_RELATIVE(x,s) ( \
 	(R_TYPE(x) == REL_RELATIVE) || \
@@ -92,11 +96,14 @@ struct fdpic_dummy_loadmap {
 #define DT_DEBUG_INDIRECT 0
 #endif
 
+#ifndef DT_DEBUG_INDIRECT_REL
+#define DT_DEBUG_INDIRECT_REL 0
+#endif
+
 #define AUX_CNT 32
-#define DYN_CNT 32
+#define DYN_CNT 37
 
 typedef void (*stage2_func)(unsigned char *, size_t *);
-typedef void (*stage3_func)(size_t *);
 
 hidden void *__dlsym(void *restrict, const char *restrict, void *restrict);
 
@@ -106,4 +113,9 @@ hidden void __dl_vseterr(const char *, va_list);
 
 hidden ptrdiff_t __tlsdesc_static(), __tlsdesc_dynamic();
 
+hidden extern int __malloc_replaced;
+hidden extern int __aligned_alloc_replaced;
+hidden void __malloc_donate(char *, char *);
+hidden int __malloc_allzerop(void *);
+
 #endif
diff --git a/src/internal/emulate_wait4.c b/src/internal/emulate_wait4.c
new file mode 100644
index 00000000..f6303412
--- /dev/null
+++ b/src/internal/emulate_wait4.c
@@ -0,0 +1,55 @@
+#include <sys/wait.h>
+#include "syscall.h"
+
+#ifndef SYS_wait4
+hidden long __emulate_wait4(int pid, int *status, int options, void *kru, int cp)
+{
+	idtype_t t;
+	int r;
+	siginfo_t info;
+
+	info.si_pid = 0;
+	if (pid < -1) {
+		t = P_PGID;
+		pid = -pid;
+	} else if (pid == -1) {
+		t = P_ALL;
+	} else if (pid == 0) {
+		t = P_PGID;
+	} else {
+		t = P_PID;
+	}
+
+	if (cp) r = __syscall_cp(SYS_waitid, t, pid, &info, options|WEXITED, kru);
+	else r = __syscall(SYS_waitid, t, pid, &info, options|WEXITED, kru);
+
+	if (r<0) return r;
+
+	if (info.si_pid && status) {
+		int sw=0;
+		switch (info.si_code) {
+		case CLD_CONTINUED:
+			sw = 0xffff;
+			break;
+		case CLD_DUMPED:
+			sw = info.si_status&0x7f | 0x80;
+			break;
+		case CLD_EXITED:
+			sw = (info.si_status&0xff) << 8;
+			break;
+		case CLD_KILLED:
+			sw = info.si_status&0x7f;
+			break;
+		case CLD_STOPPED:
+		case CLD_TRAPPED:
+			/* see ptrace(2); the high bits of si_status can contain */
+			/* PTRACE_EVENT_ values which must be preserved */
+			sw = (info.si_status << 8) + 0x7f;
+			break;
+		}
+		*status = sw;
+	}
+
+	return info.si_pid;
+}
+#endif
diff --git a/src/internal/fork_impl.h b/src/internal/fork_impl.h
new file mode 100644
index 00000000..f995fce2
--- /dev/null
+++ b/src/internal/fork_impl.h
@@ -0,0 +1,21 @@
+#include <features.h>
+
+extern hidden volatile int *const __at_quick_exit_lockptr;
+extern hidden volatile int *const __atexit_lockptr;
+extern hidden volatile int *const __gettext_lockptr;
+extern hidden volatile int *const __locale_lockptr;
+extern hidden volatile int *const __random_lockptr;
+extern hidden volatile int *const __sem_open_lockptr;
+extern hidden volatile int *const __stdio_ofl_lockptr;
+extern hidden volatile int *const __syslog_lockptr;
+extern hidden volatile int *const __timezone_lockptr;
+
+extern hidden volatile int *const __bump_lockptr;
+
+extern hidden volatile int *const __vmlock_lockptr;
+
+hidden void __malloc_atfork(int);
+hidden void __ldso_atfork(int);
+hidden void __pthread_key_atfork(int);
+
+hidden void __post_Fork(int);
diff --git a/src/internal/ksigaction.h b/src/internal/ksigaction.h
index 8ebd5938..ef333f33 100644
--- a/src/internal/ksigaction.h
+++ b/src/internal/ksigaction.h
@@ -6,8 +6,13 @@
 struct k_sigaction {
 	void (*handler)(int);
 	unsigned long flags;
+#ifdef SA_RESTORER
 	void (*restorer)(void);
+#endif
 	unsigned mask[2];
+#ifndef SA_RESTORER
+	void *unused;
+#endif
 };
 
 hidden void __restore(), __restore_rt();
diff --git a/src/internal/libc.h b/src/internal/libc.h
index ac97dc7e..619bba86 100644
--- a/src/internal/libc.h
+++ b/src/internal/libc.h
@@ -18,10 +18,11 @@ struct tls_module {
 };
 
 struct __libc {
-	int can_do_threads;
-	int threaded;
-	int secure;
-	volatile int threads_minus_1;
+	char can_do_threads;
+	char threaded;
+	char secure;
+	volatile signed char need_locks;
+	int threads_minus_1;
 	size_t *auxv;
 	struct tls_module *tls_head;
 	size_t tls_size, tls_align, tls_cnt;
diff --git a/src/internal/libm.h b/src/internal/libm.h
index b5bd26b8..72ad17d8 100644
--- a/src/internal/libm.h
+++ b/src/internal/libm.h
@@ -236,13 +236,13 @@ hidden int    __rem_pio2(double,double*);
 hidden double __sin(double,double,int);
 hidden double __cos(double,double);
 hidden double __tan(double,double,int);
-hidden double __expo2(double);
+hidden double __expo2(double,double);
 
 hidden int    __rem_pio2f(float,double*);
 hidden float  __sindf(double);
 hidden float  __cosdf(double);
 hidden float  __tandf(double,int);
-hidden float  __expo2f(float);
+hidden float  __expo2f(float,float);
 
 hidden int __rem_pio2l(long double, long double *);
 hidden long double __sinl(long double, long double, int);
@@ -267,5 +267,8 @@ hidden double __math_uflow(uint32_t);
 hidden double __math_oflow(uint32_t);
 hidden double __math_divzero(uint32_t);
 hidden double __math_invalid(double);
+#if LDBL_MANT_DIG != DBL_MANT_DIG
+hidden long double __math_invalidl(long double);
+#endif
 
 #endif
diff --git a/src/internal/locale_impl.h b/src/internal/locale_impl.h
index 741a71c4..4431a92e 100644
--- a/src/internal/locale_impl.h
+++ b/src/internal/locale_impl.h
@@ -15,6 +15,8 @@ struct __locale_map {
 	const struct __locale_map *next;
 };
 
+extern hidden volatile int __locale_lock[1];
+
 extern hidden const struct __locale_map __c_dot_utf8;
 extern hidden const struct __locale_struct __c_locale;
 extern hidden const struct __locale_struct __c_dot_utf8_locale;
diff --git a/src/internal/malloc_impl.h b/src/internal/malloc_impl.h
deleted file mode 100644
index 59785a7f..00000000
--- a/src/internal/malloc_impl.h
+++ /dev/null
@@ -1,46 +0,0 @@
-#ifndef MALLOC_IMPL_H
-#define MALLOC_IMPL_H
-
-#include <sys/mman.h>
-
-hidden void *__expand_heap(size_t *);
-
-hidden void __malloc_donate(char *, char *);
-
-hidden void *__memalign(size_t, size_t);
-
-struct chunk {
-	size_t psize, csize;
-	struct chunk *next, *prev;
-};
-
-struct bin {
-	volatile int lock[2];
-	struct chunk *head;
-	struct chunk *tail;
-};
-
-#define SIZE_ALIGN (4*sizeof(size_t))
-#define SIZE_MASK (-SIZE_ALIGN)
-#define OVERHEAD (2*sizeof(size_t))
-#define MMAP_THRESHOLD (0x1c00*SIZE_ALIGN)
-#define DONTCARE 16
-#define RECLAIM 163840
-
-#define CHUNK_SIZE(c) ((c)->csize & -2)
-#define CHUNK_PSIZE(c) ((c)->psize & -2)
-#define PREV_CHUNK(c) ((struct chunk *)((char *)(c) - CHUNK_PSIZE(c)))
-#define NEXT_CHUNK(c) ((struct chunk *)((char *)(c) + CHUNK_SIZE(c)))
-#define MEM_TO_CHUNK(p) (struct chunk *)((char *)(p) - OVERHEAD)
-#define CHUNK_TO_MEM(c) (void *)((char *)(c) + OVERHEAD)
-#define BIN_TO_CHUNK(i) (MEM_TO_CHUNK(&mal.bins[i].head))
-
-#define C_INUSE  ((size_t)1)
-
-#define IS_MMAPPED(c) !((c)->csize & (C_INUSE))
-
-hidden void __bin_chunk(struct chunk *);
-
-hidden extern int __malloc_replaced;
-
-#endif
diff --git a/src/internal/pthread_impl.h b/src/internal/pthread_impl.h
index 5742dfc5..de2b9d8b 100644
--- a/src/internal/pthread_impl.h
+++ b/src/internal/pthread_impl.h
@@ -11,16 +11,25 @@
 #include "atomic.h"
 #include "futex.h"
 
+#include "pthread_arch.h"
+
 #define pthread __pthread
 
 struct pthread {
 	/* Part 1 -- these fields may be external or
 	 * internal (accessed via asm) ABI. Do not change. */
 	struct pthread *self;
+#ifndef TLS_ABOVE_TP
 	uintptr_t *dtv;
+#endif
 	struct pthread *prev, *next; /* non-ABI */
 	uintptr_t sysinfo;
-	uintptr_t canary, canary2;
+#ifndef TLS_ABOVE_TP
+#ifdef CANARY_PAD
+	uintptr_t canary_pad;
+#endif
+	uintptr_t canary;
+#endif
 
 	/* Part 2 -- implementation details, non-ABI. */
 	int tid;
@@ -43,6 +52,7 @@ struct pthread {
 		long off;
 		volatile void *volatile pending;
 	} robust_list;
+	int h_errno_val;
 	volatile int timer_id;
 	locale_t locale;
 	volatile int killlock[1];
@@ -51,21 +61,19 @@ struct pthread {
 
 	/* Part 3 -- the positions of these fields relative to
 	 * the end of the structure is external and internal ABI. */
-	uintptr_t canary_at_end;
-	uintptr_t *dtv_copy;
+#ifdef TLS_ABOVE_TP
+	uintptr_t canary;
+	uintptr_t *dtv;
+#endif
 };
 
 enum {
-	DT_EXITING = 0,
+	DT_EXITED = 0,
+	DT_EXITING,
 	DT_JOINABLE,
 	DT_DETACHED,
 };
 
-struct __timer {
-	int timerid;
-	pthread_t thread;
-};
-
 #define __SU (sizeof(size_t)/sizeof(int))
 
 #define _a_stacksize __u.__s[0]
@@ -98,16 +106,22 @@ struct __timer {
 #define _b_waiters2 __u.__vi[4]
 #define _b_inst __u.__p[3]
 
-#include "pthread_arch.h"
-
-#ifndef CANARY
-#define CANARY canary
+#ifndef TP_OFFSET
+#define TP_OFFSET 0
 #endif
 
 #ifndef DTP_OFFSET
 #define DTP_OFFSET 0
 #endif
 
+#ifdef TLS_ABOVE_TP
+#define TP_ADJ(p) ((char *)(p) + sizeof(struct pthread) + TP_OFFSET)
+#define __pthread_self() ((pthread_t)(__get_tp() - sizeof(struct __pthread) - TP_OFFSET))
+#else
+#define TP_ADJ(p) (p)
+#define __pthread_self() ((pthread_t)__get_tp())
+#endif
+
 #ifndef tls_mod_off_t
 #define tls_mod_off_t size_t
 #endif
@@ -141,7 +155,6 @@ hidden int __pthread_key_delete_impl(pthread_key_t);
 
 extern hidden volatile size_t __pthread_tsd_size;
 extern hidden void *__pthread_tsd_main[];
-extern hidden volatile int __aio_fut;
 extern hidden volatile int __eintr_valid_flag;
 
 hidden int __clone(int (*)(void *), void *, int, void *, ...);
@@ -176,6 +189,8 @@ hidden void __tl_sync(pthread_t);
 
 extern hidden volatile int __thread_list_lock;
 
+extern hidden volatile int __abort_lock[1];
+
 extern hidden unsigned __default_stacksize;
 extern hidden unsigned __default_guardsize;
 
diff --git a/src/internal/shgetc.c b/src/internal/shgetc.c
index a4a9c633..7455d2f0 100644
--- a/src/internal/shgetc.c
+++ b/src/internal/shgetc.c
@@ -32,6 +32,6 @@ int __shgetc(FILE *f)
 	else
 		f->shend = f->rend;
 	f->shcnt = f->buf - f->rpos + cnt;
-	if (f->rpos[-1] != c) f->rpos[-1] = c;
+	if (f->rpos <= f->buf) f->rpos[-1] = c;
 	return c;
 }
diff --git a/src/internal/stdio_impl.h b/src/internal/stdio_impl.h
index d7398f59..0b2438d6 100644
--- a/src/internal/stdio_impl.h
+++ b/src/internal/stdio_impl.h
@@ -60,8 +60,6 @@ hidden size_t __stdout_write(FILE *, const unsigned char *, size_t);
 hidden off_t __stdio_seek(FILE *, off_t, int);
 hidden int __stdio_close(FILE *);
 
-hidden size_t __string_read(FILE *, unsigned char *, size_t);
-
 hidden int __toread(FILE *);
 hidden int __towrite(FILE *);
 
diff --git a/src/internal/syscall.h b/src/internal/syscall.h
index 9f2784db..33d981f9 100644
--- a/src/internal/syscall.h
+++ b/src/internal/syscall.h
@@ -2,6 +2,7 @@
 #define _INTERNAL_SYSCALL_H
 
 #include <features.h>
+#include <errno.h>
 #include <sys/syscall.h>
 #include "syscall_arch.h"
 
@@ -57,15 +58,22 @@ hidden long __syscall_ret(unsigned long),
 #define __syscall_cp(...) __SYSCALL_DISP(__syscall_cp,__VA_ARGS__)
 #define syscall_cp(...) __syscall_ret(__syscall_cp(__VA_ARGS__))
 
-#ifndef SYSCALL_USE_SOCKETCALL
-#define __socketcall(nm,a,b,c,d,e,f) __syscall(SYS_##nm, a, b, c, d, e, f)
-#define __socketcall_cp(nm,a,b,c,d,e,f) __syscall_cp(SYS_##nm, a, b, c, d, e, f)
-#else
-#define __socketcall(nm,a,b,c,d,e,f) __syscall(SYS_socketcall, __SC_##nm, \
-    ((long [6]){ (long)a, (long)b, (long)c, (long)d, (long)e, (long)f }))
-#define __socketcall_cp(nm,a,b,c,d,e,f) __syscall_cp(SYS_socketcall, __SC_##nm, \
-    ((long [6]){ (long)a, (long)b, (long)c, (long)d, (long)e, (long)f }))
-#endif
+static inline long __alt_socketcall(int sys, int sock, int cp, syscall_arg_t a, syscall_arg_t b, syscall_arg_t c, syscall_arg_t d, syscall_arg_t e, syscall_arg_t f)
+{
+	long r;
+	if (cp) r = __syscall_cp(sys, a, b, c, d, e, f);
+	else r = __syscall(sys, a, b, c, d, e, f);
+	if (r != -ENOSYS) return r;
+#ifdef SYS_socketcall
+	if (cp) r = __syscall_cp(SYS_socketcall, sock, ((long[6]){a, b, c, d, e, f}));
+	else r = __syscall(SYS_socketcall, sock, ((long[6]){a, b, c, d, e, f}));
+#endif
+	return r;
+}
+#define __socketcall(nm, a, b, c, d, e, f) __alt_socketcall(SYS_##nm, __SC_##nm, 0, \
+	__scc(a), __scc(b), __scc(c), __scc(d), __scc(e), __scc(f))
+#define __socketcall_cp(nm, a, b, c, d, e, f) __alt_socketcall(SYS_##nm, __SC_##nm, 1, \
+	__scc(a), __scc(b), __scc(c), __scc(d), __scc(e), __scc(f))
 
 /* fixup legacy 16-bit junk */
 
@@ -193,6 +201,45 @@ hidden long __syscall_ret(unsigned long),
 #define SYS_sendfile SYS_sendfile64
 #endif
 
+#ifdef SYS_timer_settime32
+#define SYS_timer_settime SYS_timer_settime32
+#endif
+
+#ifdef SYS_timer_gettime32
+#define SYS_timer_gettime SYS_timer_gettime32
+#endif
+
+#ifdef SYS_timerfd_settime32
+#define SYS_timerfd_settime SYS_timerfd_settime32
+#endif
+
+#ifdef SYS_timerfd_gettime32
+#define SYS_timerfd_gettime SYS_timerfd_gettime32
+#endif
+
+#ifdef SYS_clock_settime32
+#define SYS_clock_settime SYS_clock_settime32
+#endif
+
+#ifdef SYS_clock_gettime32
+#define SYS_clock_gettime SYS_clock_gettime32
+#endif
+
+#ifdef SYS_clock_getres_time32
+#define SYS_clock_getres SYS_clock_getres_time32
+#endif
+
+#ifdef SYS_clock_nanosleep_time32
+#define SYS_clock_nanosleep SYS_clock_nanosleep_time32
+#endif
+
+#ifdef SYS_gettimeofday_time32
+#define SYS_gettimeofday SYS_gettimeofday_time32
+#endif
+
+#ifdef SYS_settimeofday_time32
+#define SYS_settimeofday SYS_settimeofday_time32
+#endif
 
 /* Ensure that the plain syscall names are defined even for "time64-only"
  * archs. These facilitate callers passing null time arguments, and make
@@ -299,6 +346,12 @@ hidden long __syscall_ret(unsigned long),
 #define __SC_recvmmsg    19
 #define __SC_sendmmsg    20
 
+/* This is valid only because all socket syscalls are made via
+ * socketcall, which always fills unused argument slots with zeros. */
+#ifndef SYS_accept
+#define SYS_accept SYS_accept4
+#endif
+
 #ifndef SO_RCVTIMEO_OLD
 #define SO_RCVTIMEO_OLD  20
 #endif
@@ -306,6 +359,13 @@ hidden long __syscall_ret(unsigned long),
 #define SO_SNDTIMEO_OLD  21
 #endif
 
+#define SO_TIMESTAMP_OLD    29
+#define SO_TIMESTAMPNS_OLD  35
+#define SO_TIMESTAMPING_OLD 37
+#define SCM_TIMESTAMP_OLD    SO_TIMESTAMP_OLD
+#define SCM_TIMESTAMPNS_OLD  SO_TIMESTAMPNS_OLD
+#define SCM_TIMESTAMPING_OLD SO_TIMESTAMPING_OLD
+
 #ifndef SIOCGSTAMP_OLD
 #define SIOCGSTAMP_OLD 0x8906
 #endif
@@ -331,6 +391,18 @@ hidden long __syscall_ret(unsigned long),
 #define __sys_open_cp(...) __SYSCALL_DISP(__sys_open_cp,,__VA_ARGS__)
 #define sys_open_cp(...) __syscall_ret(__sys_open_cp(__VA_ARGS__))
 
+#ifdef SYS_wait4
+#define __sys_wait4(a,b,c,d) __syscall(SYS_wait4,a,b,c,d)
+#define __sys_wait4_cp(a,b,c,d) __syscall_cp(SYS_wait4,a,b,c,d)
+#else
+hidden long __emulate_wait4(int, int *, int, void *, int);
+#define __sys_wait4(a,b,c,d) __emulate_wait4(a,b,c,d,0)
+#define __sys_wait4_cp(a,b,c,d) __emulate_wait4(a,b,c,d,1)
+#endif
+
+#define sys_wait4(a,b,c,d) __syscall_ret(__sys_wait4(a,b,c,d))
+#define sys_wait4_cp(a,b,c,d) __syscall_ret(__sys_wait4_cp(a,b,c,d))
+
 hidden void __procfdname(char __buf[static 15+3*sizeof(int)], unsigned);
 
 hidden void *__vdsosym(const char *, const char *);