16 files changed, 259 insertions, 80 deletions

diff --git a/src/linux/cache.c b/src/linux/cache.c
index 0eb051c2..e76f7812 100644
--- a/src/linux/cache.c
+++ b/src/linux/cache.c
@@ -21,7 +21,7 @@ weak_alias(__cachectl, cachectl);
 #ifdef SYS_riscv_flush_icache
 
 #define VDSO_FLUSH_ICACHE_SYM "__vdso_flush_icache"
-#define VDSO_FLUSH_ICACHE_VER "LINUX_4.5"
+#define VDSO_FLUSH_ICACHE_VER "LINUX_4.15"
 
 static void *volatile vdso_func;
 
@@ -45,6 +45,7 @@ int __riscv_flush_icache(void *start, void *end, unsigned long int flags)
 		if (!r) return r;
 		if (r != -ENOSYS) return __syscall_ret(r);
 	}
+	return syscall(SYS_riscv_flush_icache, start, end, flags);
 }
 weak_alias(__riscv_flush_icache, riscv_flush_icache);
 #endif
diff --git a/src/linux/clock_adjtime.c b/src/linux/clock_adjtime.c
index 2f531397..d4d03d24 100644
--- a/src/linux/clock_adjtime.c
+++ b/src/linux/clock_adjtime.c
@@ -38,9 +38,60 @@ int clock_adjtime (clockid_t clock_id, struct timex *utx)
 {
 	int r = -ENOSYS;
 #ifdef SYS_clock_adjtime64
-	if (SYS_clock_adjtime == SYS_clock_adjtime64 ||
-	    (utx->modes & ADJ_SETOFFSET) && !IS32BIT(utx->time.tv_sec)) {
-		struct ktimex64 ktx = {
+	struct ktimex64 ktx = {
+		.modes = utx->modes,
+		.offset = utx->offset,
+		.freq = utx->freq,
+		.maxerror = utx->maxerror,
+		.esterror = utx->esterror,
+		.status = utx->status,
+		.constant = utx->constant,
+		.precision = utx->precision,
+		.tolerance = utx->tolerance,
+		.time_sec = utx->time.tv_sec,
+		.time_usec = utx->time.tv_usec,
+		.tick = utx->tick,
+		.ppsfreq = utx->ppsfreq,
+		.jitter = utx->jitter,
+		.shift = utx->shift,
+		.stabil = utx->stabil,
+		.jitcnt = utx->jitcnt,
+		.calcnt = utx->calcnt,
+		.errcnt = utx->errcnt,
+		.stbcnt = utx->stbcnt,
+		.tai = utx->tai,
+	};
+	r = __syscall(SYS_clock_adjtime64, clock_id, &ktx);
+	if (r>=0) {
+		utx->modes = ktx.modes;
+		utx->offset = ktx.offset;
+		utx->freq = ktx.freq;
+		utx->maxerror = ktx.maxerror;
+		utx->esterror = ktx.esterror;
+		utx->status = ktx.status;
+		utx->constant = ktx.constant;
+		utx->precision = ktx.precision;
+		utx->tolerance = ktx.tolerance;
+		utx->time.tv_sec = ktx.time_sec;
+		utx->time.tv_usec = ktx.time_usec;
+		utx->tick = ktx.tick;
+		utx->ppsfreq = ktx.ppsfreq;
+		utx->jitter = ktx.jitter;
+		utx->shift = ktx.shift;
+		utx->stabil = ktx.stabil;
+		utx->jitcnt = ktx.jitcnt;
+		utx->calcnt = ktx.calcnt;
+		utx->errcnt = ktx.errcnt;
+		utx->stbcnt = ktx.stbcnt;
+		utx->tai = ktx.tai;
+	}
+	if (SYS_clock_adjtime == SYS_clock_adjtime64 || r!=-ENOSYS)
+		return __syscall_ret(r);
+	if ((utx->modes & ADJ_SETOFFSET) && !IS32BIT(utx->time.tv_sec))
+		return __syscall_ret(-ENOTSUP);
+#endif
+	if (sizeof(time_t) > sizeof(long)) {
+		struct ktimex ktx = {
 			.modes = utx->modes,
 			.offset = utx->offset,
 			.freq = utx->freq,
@@ -63,6 +114,10 @@ int clock_adjtime (clockid_t clock_id, struct timex *utx)
 			.stbcnt = utx->stbcnt,
 			.tai = utx->tai,
 		};
+#ifdef SYS_adjtimex
+		if (clock_id==CLOCK_REALTIME) r = __syscall(SYS_adjtimex, &ktx);
+		else
+#endif
 		r = __syscall(SYS_clock_adjtime, clock_id, &ktx);
 		if (r>=0) {
 			utx->modes = ktx.modes;
@@ -87,29 +142,6 @@ int clock_adjtime (clockid_t clock_id, struct timex *utx)
 			utx->stbcnt = ktx.stbcnt;
 			utx->tai = ktx.tai;
 		}
-	}
-	if (SYS_clock_adjtime == SYS_clock_adjtime64 || r!=-ENOSYS)
-		return __syscall_ret(r);
-	if ((utx->modes & ADJ_SETOFFSET) && !IS32BIT(utx->time.tv_sec))
-		return __syscall_ret(-ENOTSUP);
-#endif
-	if (sizeof(time_t) > sizeof(long)) {
-		union {
-			struct timex utx;
-			struct ktimex ktx;
-		} u = { *utx };
-		u.ktx.time_sec = utx->time.tv_sec;
-		u.ktx.time_usec = utx->time.tv_usec;
-#ifdef SYS_adjtimex
-		if (clock_id==CLOCK_REALTIME) r = __syscall(SYS_adjtimex, &u);
-		else
-#endif
-		r = __syscall(SYS_clock_adjtime, clock_id, &u);
-		if (r>=0) {
-			*utx = u.utx;
-			utx->time.tv_sec = u.ktx.time_sec;
-			utx->time.tv_usec = u.ktx.time_usec;
-		}
 		return __syscall_ret(r);
 	}
 #ifdef SYS_adjtimex
diff --git a/src/linux/clone.c b/src/linux/clone.c
index 8c1af7d3..257c1cec 100644
--- a/src/linux/clone.c
+++ b/src/linux/clone.c
@@ -4,18 +4,62 @@
 #include <sched.h>
 #include "pthread_impl.h"
 #include "syscall.h"
+#include "lock.h"
+#include "fork_impl.h"
+
+struct clone_start_args {
+	int (*func)(void *);
+	void *arg;
+	sigset_t sigmask;
+};
+
+static int clone_start(void *arg)
+{
+	struct clone_start_args *csa = arg;
+	__post_Fork(0);
+	__restore_sigs(&csa->sigmask);
+	return csa->func(csa->arg);
+}
 
 int clone(int (*func)(void *), void *stack, int flags, void *arg, ...)
 {
+	struct clone_start_args csa;
 	va_list ap;
-	pid_t *ptid, *ctid;
-	void  *tls;
+	pid_t *ptid = 0, *ctid = 0;
+	void  *tls = 0;
+
+	/* Flags that produce an invalid thread/TLS state are disallowed. */
+	int badflags = CLONE_THREAD | CLONE_SETTLS | CLONE_CHILD_CLEARTID;
+
+	if ((flags & badflags) || !stack)
+		return __syscall_ret(-EINVAL);
 
 	va_start(ap, arg);
-	ptid = va_arg(ap, pid_t *);
-	tls  = va_arg(ap, void *);
-	ctid = va_arg(ap, pid_t *);
+	if (flags & (CLONE_PIDFD | CLONE_PARENT_SETTID | CLONE_CHILD_SETTID))
+	 	ptid = va_arg(ap, pid_t *);
+	if (flags & CLONE_CHILD_SETTID) {
+		tls = va_arg(ap, void *);
+		ctid = va_arg(ap, pid_t *);
+	}
 	va_end(ap);
 
-	return __syscall_ret(__clone(func, stack, flags, arg, ptid, tls, ctid));
+	/* If CLONE_VM is used, it's impossible to give the child a consistent
+	 * thread structure. In this case, the best we can do is assume the
+	 * caller is content with an extremely restrictive execution context
+	 * like the one vfork() would provide. */
+	if (flags & CLONE_VM) return __syscall_ret(
+		__clone(func, stack, flags, arg, ptid, tls, ctid));
+
+	__block_all_sigs(&csa.sigmask);
+	LOCK(__abort_lock);
+
+	/* Setup the a wrapper start function for the child process to do
+	 * mimic _Fork in producing a consistent execution state. */
+	csa.func = func;
+	csa.arg = arg;
+	int ret = __clone(clone_start, stack, flags, &csa, ptid, tls, ctid);
+
+	__post_Fork(ret);
+	__restore_sigs(&csa.sigmask);
+	return __syscall_ret(ret);
 }
diff --git a/src/linux/epoll.c b/src/linux/epoll.c
index deff5b10..e56e8f4c 100644
--- a/src/linux/epoll.c
+++ b/src/linux/epoll.c
@@ -5,6 +5,7 @@
 
 int epoll_create(int size)
 {
+	if (size<=0) return __syscall_ret(-EINVAL);
 	return epoll_create1(0);
 }
 
@@ -24,9 +25,9 @@ int epoll_ctl(int fd, int op, int fd2, struct epoll_event *ev)
 
 int epoll_pwait(int fd, struct epoll_event *ev, int cnt, int to, const sigset_t *sigs)
 {
-	int r = __syscall(SYS_epoll_pwait, fd, ev, cnt, to, sigs, _NSIG/8);
+	int r = __syscall_cp(SYS_epoll_pwait, fd, ev, cnt, to, sigs, _NSIG/8);
 #ifdef SYS_epoll_wait
-	if (r==-ENOSYS && !sigs) r = __syscall(SYS_epoll_wait, fd, ev, cnt, to);
+	if (r==-ENOSYS && !sigs) r = __syscall_cp(SYS_epoll_wait, fd, ev, cnt, to);
 #endif
 	return __syscall_ret(r);
 }
diff --git a/src/linux/fallocate.c b/src/linux/fallocate.c
index 7d68bc8f..9146350e 100644
--- a/src/linux/fallocate.c
+++ b/src/linux/fallocate.c
@@ -7,6 +7,3 @@ int fallocate(int fd, int mode, off_t base, off_t len)
 	return syscall(SYS_fallocate, fd, mode, __SYSCALL_LL_E(base),
 		__SYSCALL_LL_E(len));
 }
-
-#undef fallocate64
-weak_alias(fallocate, fallocate64);
diff --git a/src/linux/getdents.c b/src/linux/getdents.c
index 796c1e5c..97f76e14 100644
--- a/src/linux/getdents.c
+++ b/src/linux/getdents.c
@@ -8,5 +8,3 @@ int getdents(int fd, struct dirent *buf, size_t len)
 	if (len>INT_MAX) len = INT_MAX;
 	return syscall(SYS_getdents, fd, buf, len);
 }
-
-weak_alias(getdents, getdents64);
diff --git a/src/linux/gettid.c b/src/linux/gettid.c
new file mode 100644
index 00000000..70767137
--- /dev/null
+++ b/src/linux/gettid.c
@@ -0,0 +1,8 @@
+#define _GNU_SOURCE
+#include <unistd.h>
+#include "pthread_impl.h"
+
+pid_t gettid(void)
+{
+	return __pthread_self()->tid;
+}
diff --git a/src/linux/membarrier.c b/src/linux/membarrier.c
index 9ebe906e..f64fe7e1 100644
--- a/src/linux/membarrier.c
+++ b/src/linux/membarrier.c
@@ -9,13 +9,8 @@ static void dummy_0(void)
 {
 }
 
-static void dummy_1(pthread_t t)
-{
-}
-
 weak_alias(dummy_0, __tl_lock);
 weak_alias(dummy_0, __tl_unlock);
-weak_alias(dummy_1, __tl_sync);
 
 static sem_t barrier_sem;
 
@@ -40,7 +35,7 @@ int __membarrier(int cmd, int flags)
 		__tl_lock();
 		sem_init(&barrier_sem, 0, 0);
 		struct sigaction sa = {
-			.sa_flags = SA_RESTART,
+			.sa_flags = SA_RESTART | SA_ONSTACK,
 			.sa_handler = bcast_barrier
 		};
 		memset(&sa.sa_mask, -1, sizeof sa.sa_mask);
diff --git a/src/linux/ppoll.c b/src/linux/ppoll.c
deleted file mode 100644
index e614600a..00000000
--- a/src/linux/ppoll.c
+++ /dev/null
@@ -1,26 +0,0 @@
-#define _GNU_SOURCE
-#include <poll.h>
-#include <signal.h>
-#include <errno.h>
-#include "syscall.h"
-
-#define IS32BIT(x) !((x)+0x80000000ULL>>32)
-#define CLAMP(x) (int)(IS32BIT(x) ? (x) : 0x7fffffffU+((0ULL+(x))>>63))
-
-int ppoll(struct pollfd *fds, nfds_t n, const struct timespec *to, const sigset_t *mask)
-{
-	time_t s = to ? to->tv_sec : 0;
-	long ns = to ? to->tv_nsec : 0;
-#ifdef SYS_ppoll_time64
-	int r = -ENOSYS;
-	if (SYS_ppoll == SYS_ppoll_time64 || !IS32BIT(s))
-		r = __syscall_cp(SYS_ppoll_time64, fds, n,
-			to ? ((long long[]){s, ns}) : 0,
-			mask, _NSIG/8);
-	if (SYS_ppoll == SYS_ppoll_time64 || r != -ENOSYS)
-		return __syscall_ret(r);
-	s = CLAMP(s);
-#endif
-	return syscall_cp(SYS_ppoll, fds, n,
-		to ? ((long[]){s, ns}) : 0, mask, _NSIG/8);
-}
diff --git a/src/linux/preadv2.c b/src/linux/preadv2.c
new file mode 100644
index 00000000..5e7ab70f
--- /dev/null
+++ b/src/linux/preadv2.c
@@ -0,0 +1,17 @@
+#define _GNU_SOURCE
+#include <sys/uio.h>
+#include <unistd.h>
+#include "syscall.h"
+
+ssize_t preadv2(int fd, const struct iovec *iov, int count, off_t ofs, int flags)
+{
+#ifdef SYS_preadv
+	if (!flags) {
+		if (ofs==-1) return readv(fd, iov, count);
+		return syscall_cp(SYS_preadv, fd, iov, count,
+			(long)(ofs), (long)(ofs>>32));
+	}
+#endif
+	return syscall_cp(SYS_preadv2, fd, iov, count,
+		(long)(ofs), (long)(ofs>>32), flags);
+}
diff --git a/src/linux/prlimit.c b/src/linux/prlimit.c
index 3df9ffba..fcf45aab 100644
--- a/src/linux/prlimit.c
+++ b/src/linux/prlimit.c
@@ -21,6 +21,3 @@ int prlimit(pid_t pid, int resource, const struct rlimit *new_limit, struct rlim
 	}
 	return r;
 }
-
-#undef prlimit64
-weak_alias(prlimit, prlimit64);
diff --git a/src/linux/pwritev2.c b/src/linux/pwritev2.c
new file mode 100644
index 00000000..ece90d7c
--- /dev/null
+++ b/src/linux/pwritev2.c
@@ -0,0 +1,17 @@
+#define _GNU_SOURCE
+#include <sys/uio.h>
+#include <unistd.h>
+#include "syscall.h"
+
+ssize_t pwritev2(int fd, const struct iovec *iov, int count, off_t ofs, int flags)
+{
+#ifdef SYS_pwritev
+	if (!flags) {
+		if (ofs==-1) return writev(fd, iov, count);
+		return syscall_cp(SYS_pwritev, fd, iov, count,
+			(long)(ofs), (long)(ofs>>32));
+	}
+#endif
+	return syscall_cp(SYS_pwritev2, fd, iov, count,
+		(long)(ofs), (long)(ofs>>32), flags);
+}
diff --git a/src/linux/sendfile.c b/src/linux/sendfile.c
index 9afe6dd6..fc1577d3 100644
--- a/src/linux/sendfile.c
+++ b/src/linux/sendfile.c
@@ -5,5 +5,3 @@ ssize_t sendfile(int out_fd, int in_fd, off_t *ofs, size_t count)
 {
 	return syscall(SYS_sendfile, out_fd, in_fd, ofs, count);
 }
-
-weak_alias(sendfile, sendfile64);
diff --git a/src/linux/setgroups.c b/src/linux/setgroups.c
index 1248fdbf..47142f14 100644
--- a/src/linux/setgroups.c
+++ b/src/linux/setgroups.c
@@ -1,8 +1,36 @@
 #define _GNU_SOURCE
 #include <unistd.h>
+#include <signal.h>
 #include "syscall.h"
+#include "libc.h"
+
+struct ctx {
+	size_t count;
+	const gid_t *list;
+	int ret;
+};
+
+static void do_setgroups(void *p)
+{
+	struct ctx *c = p;
+	if (c->ret<0) return;
+	int ret = __syscall(SYS_setgroups, c->count, c->list);
+	if (ret && !c->ret) {
+		/* If one thread fails to set groups after another has already
+		 * succeeded, forcibly killing the process is the only safe
+		 * thing to do. State is inconsistent and dangerous. Use
+		 * SIGKILL because it is uncatchable. */
+		__block_all_sigs(0);
+		__syscall(SYS_kill, __syscall(SYS_getpid), SIGKILL);
+	}
+	c->ret = ret;
+}
 
 int setgroups(size_t count, const gid_t list[])
 {
-	return syscall(SYS_setgroups, count, list);
+	/* ret is initially nonzero so that failure of the first thread does not
+	 * trigger the safety kill above. */
+	struct ctx c = { .count = count, .list = list, .ret = 1 };
+	__synccall(do_setgroups, &c);
+	return __syscall_ret(c.ret);
 }
diff --git a/src/linux/statx.c b/src/linux/statx.c
new file mode 100644
index 00000000..4616bff4
--- /dev/null
+++ b/src/linux/statx.c
@@ -0,0 +1,42 @@
+#define _GNU_SOURCE
+#include <sys/stat.h>
+#include <string.h>
+#include <syscall.h>
+#include <sys/sysmacros.h>
+#include <errno.h>
+
+int statx(int dirfd, const char *restrict path, int flags, unsigned mask, struct statx *restrict stx)
+{
+	int ret = __syscall(SYS_statx, dirfd, path, flags, mask, stx);
+
+#ifndef SYS_fstatat
+	return __syscall_ret(ret);
+#endif
+
+	if (ret != -ENOSYS) return __syscall_ret(ret);
+
+	struct stat st;
+	ret = fstatat(dirfd, path, &st, flags);
+	if (ret) return ret;
+
+	stx->stx_dev_major = major(st.st_dev);
+	stx->stx_dev_minor = minor(st.st_dev);
+	stx->stx_ino = st.st_ino;
+	stx->stx_mode = st.st_mode;
+	stx->stx_nlink = st.st_nlink;
+	stx->stx_uid = st.st_uid;
+	stx->stx_gid = st.st_gid;
+	stx->stx_size = st.st_size;
+	stx->stx_blksize = st.st_blksize;
+	stx->stx_blocks = st.st_blocks;
+	stx->stx_atime.tv_sec = st.st_atim.tv_sec;
+	stx->stx_atime.tv_nsec = st.st_atim.tv_nsec;
+	stx->stx_mtime.tv_sec = st.st_mtim.tv_sec;
+	stx->stx_mtime.tv_nsec = st.st_mtim.tv_nsec;
+	stx->stx_ctime.tv_sec = st.st_ctim.tv_sec;
+	stx->stx_ctime.tv_nsec = st.st_ctim.tv_nsec;
+	stx->stx_btime = (struct statx_timestamp){.tv_sec=0, .tv_nsec=0};
+	stx->stx_mask = STATX_BASIC_STATS;
+
+	return 0;
+}
diff --git a/src/linux/wait4.c b/src/linux/wait4.c
index 97f12cc5..fb08c0d0 100644
--- a/src/linux/wait4.c
+++ b/src/linux/wait4.c
@@ -1,9 +1,39 @@
 #define _GNU_SOURCE
 #include <sys/wait.h>
 #include <sys/resource.h>
+#include <string.h>
+#include <errno.h>
 #include "syscall.h"
 
-pid_t wait4(pid_t pid, int *status, int options, struct rusage *usage)
+pid_t wait4(pid_t pid, int *status, int options, struct rusage *ru)
 {
-	return syscall(SYS_wait4, pid, status, options, usage);
+	int r;
+#ifdef SYS_wait4_time64
+	if (ru) {
+		long long kru64[18];
+		r = __syscall(SYS_wait4_time64, pid, status, options, kru64);
+		if (r > 0) {
+			ru->ru_utime = (struct timeval)
+				{ .tv_sec = kru64[0], .tv_usec = kru64[1] };
+			ru->ru_stime = (struct timeval)
+				{ .tv_sec = kru64[2], .tv_usec = kru64[3] };
+			char *slots = (char *)&ru->ru_maxrss;
+			for (int i=0; i<14; i++)
+				*(long *)(slots + i*sizeof(long)) = kru64[4+i];
+		}
+		if (SYS_wait4_time64 == SYS_wait4 || r != -ENOSYS)
+			return __syscall_ret(r);
+	}
+#endif
+	char *dest = ru ? (char *)&ru->ru_maxrss - 4*sizeof(long) : 0;
+	r = __sys_wait4(pid, status, options, dest);
+	if (r>0 && ru && sizeof(time_t) > sizeof(long)) {
+		long kru[4];
+		memcpy(kru, dest, 4*sizeof(long));
+		ru->ru_utime = (struct timeval)
+			{ .tv_sec = kru[0], .tv_usec = kru[1] };
+		ru->ru_stime = (struct timeval)
+			{ .tv_sec = kru[2], .tv_usec = kru[3] };
+	}
+	return __syscall_ret(r);
 }