Add abstime kqueue(2) timers and expand struct kevent members.

This change implements NOTE_ABSTIME flag for EVFILT_TIMER, which
specifies that the data field contains absolute time to fire the
event.

To make this useful, data member of the struct kevent must be extended
to 64bit.  Using the opportunity, I also added ext members.  This
changes struct kevent almost to Apple struct kevent64, except I did
not changed type of ident and udata, the later would cause serious API
incompatibilities.

The type of ident was kept uintptr_t since EVFILT_AIO returns a
pointer in this field, and e.g. CHERI is sensitive to the type
(discussed with brooks, jhb).

Unlike Apple kevent64, symbol versioning allows us to claim ABI
compatibility and still name the new syscall kevent(2).  Compat shims
are provided for both host native and compat32.

Requested by:	bapt
Reviewed by:	bapt, brooks, ngie (previous version)
Sponsored by:	The FreeBSD Foundation
Differential revision:	https://reviews.freebsd.org/D11025

contrib/netbsd-tests/kernel/kqueue/t_proc1.c

@@ -139,11 +139,7 @@ ATF_TC_BODY(proc1, tc)
 			printf(" NOTE_FORK");
 		}
 		if (event[0].fflags & NOTE_CHILD)
-#ifdef __FreeBSD__
-			printf(" NOTE_CHILD, parent = %" PRIdPTR, event[0].data);
-#else
 			printf(" NOTE_CHILD, parent = %" PRId64, event[0].data);
-#endif
 
 		printf("\n");
 	}

contrib/netbsd-tests/kernel/kqueue/t_sig.c

@@ -127,11 +127,7 @@ ATF_TC_BODY(sig, tc)
 		if (n == 0)
 			continue;
 
-#ifdef __FreeBSD__
-		(void)printf("sig: kevent flags: 0x%x, data: %" PRIdPTR " (# "
-#else
 		(void)printf("sig: kevent flags: 0x%x, data: %" PRId64 " (# "
-#endif
 		    "times signal posted)\n", event[0].flags, event[0].data);
 	}
 

lib/libc/include/compat.h

@@ -65,6 +65,8 @@ __sym_compat(statfs, freebsd11_statfs, FBSD_1.0);
 __sym_compat(mknod, freebsd11_mknod, FBSD_1.0);
 __sym_compat(mknodat, freebsd11_mknodat, FBSD_1.1);
 
+__sym_compat(kevent, freebsd11_kevent, FBSD_1.0);
+
 #undef __sym_compat
 
 #define	__weak_reference(sym,alias)	\

lib/libc/sys/Symbol.map

@@ -121,7 +121,6 @@ FBSD_1.0 {
 	jail;
 	jail_attach;
 	kenv;
-	kevent;
 	kill;
 	kldfind;
 	kldfirstmod;
@@ -393,6 +392,7 @@ FBSD_1.5 {
 	getdents;
 	getdirentries;
 	getfsstat;
+	kevent;
 	lstat;
 	mknod;
 	mknodat;

lib/libc/sys/kqueue.2

@@ -24,7 +24,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd April 18, 2017
+.Dd June 17, 2017
 .Dt KQUEUE 2
 .Os
 .Sh NAME
@@ -148,12 +148,13 @@ The
 structure is defined as:
 .Bd -literal
 struct kevent {
-	uintptr_t ident;	/* identifier for this event */
+	uintptr_t  ident;	/* identifier for this event */
 	short	  filter;	/* filter for event */
 	u_short	  flags;	/* action flags for kqueue */
 	u_int	  fflags;	/* filter flag value */
-	intptr_t  data;		/* filter data value */
+	int64_t   data;		/* filter data value */
 	void	  *udata;	/* opaque user data identifier */
+	uint64_t  ext[4];	/* extentions */
 };
 .Ed
 .Pp
@@ -177,6 +178,20 @@ Filter-specific flags.
 Filter-specific data value.
 .It Fa udata
 Opaque user-defined value passed through the kernel unchanged.
+.It Fa ext
+Extended data passed to and from kernel.
+The
+.Fa ext[0]
+and
+.Fa ext[1]
+members use is defined by the filter.
+If the filter does not use them, the members are copied unchanged.
+The
+.Fa ext[2]
+and
+.Fa ext[3]
+members are always passed throught the kernel as-is,
+making additional context available to application.
 .El
 .Pp
 The
@@ -515,16 +530,26 @@ Establishes an arbitrary timer identified by
 .Va ident .
 When adding a timer,
 .Va data
-specifies the timeout period.
+specifies the moment to fire the timer (for
+.Dv NOTE_ABSTIME )
+or the timeout period.
 The timer will be periodic unless
 .Dv EV_ONESHOT
+or
+.Dv NOTE_ABSTIME
 is specified.
 On return,
 .Va data
 contains the number of times the timeout has expired since the last call to
 .Fn kevent .
-This filter automatically sets the EV_CLEAR flag internally.
-.Bl -tag -width "Dv NOTE_USECONDS"
+For non-monotonic timers, this filter automatically sets the
+.Dv EV_CLEAR
+flag internally.
+.Pp
+The filter accepts the following flags in the
+.Va fflags
+argument:
+.Bl -tag -width "Dv NOTE_MSECONDS"
 .It Dv NOTE_SECONDS
 .Va data
 is in seconds.
@@ -537,6 +562,8 @@ is in microseconds.
 .It Dv NOTE_NSECONDS
 .Va data
 is in nanoseconds.
+.It Dv NOTE_ABSTIME
+The specified expiration time is absolute.
 .El
 .Pp
 If

sys/compat/freebsd32/freebsd32.h

@@ -137,12 +137,13 @@ struct statfs32 {
 };
 
 struct kevent32 {
-	u_int32_t	ident;		/* identifier for this event */
+	uint32_t	ident;		/* identifier for this event */
 	short		filter;		/* filter for event */
 	u_short		flags;
 	u_int		fflags;
-	int32_t		data;
-	u_int32_t	udata;		/* opaque user data identifier */
+	int32_t		data1, data2;
+	uint32_t	udata;		/* opaque user data identifier */
+	uint32_t	ext64[8];
 };
 
 struct iovec32 {

sys/compat/freebsd32/freebsd32_misc.c

@@ -119,7 +119,7 @@ CTASSERT(sizeof(struct statfs32) == 256);
 CTASSERT(sizeof(struct rusage32) == 72);
 #endif
 CTASSERT(sizeof(struct sigaltstack32) == 12);
-CTASSERT(sizeof(struct kevent32) == 20);
+CTASSERT(sizeof(struct kevent32) == 56);
 CTASSERT(sizeof(struct iovec32) == 8);
 CTASSERT(sizeof(struct msghdr32) == 28);
 #ifdef __amd64__
@@ -622,7 +622,8 @@ freebsd32_kevent_copyout(void *arg, struct kevent *kevp, int count)
 {
 	struct freebsd32_kevent_args *uap;
 	struct kevent32	ks32[KQ_NEVENTS];
-	int i, error = 0;
+	uint64_t e;
+	int i, j, error;
 
 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
 	uap = (struct freebsd32_kevent_args *)arg;
@@ -632,8 +633,24 @@ freebsd32_kevent_copyout(void *arg, struct kevent *kevp, int count)
 		CP(kevp[i], ks32[i], filter);
 		CP(kevp[i], ks32[i], flags);
 		CP(kevp[i], ks32[i], fflags);
-		CP(kevp[i], ks32[i], data);
+#if BYTE_ORDER == LITTLE_ENDIAN
+		ks32[i].data1 = kevp[i].data;
+		ks32[i].data2 = kevp[i].data >> 32;
+#else
+		ks32[i].data1 = kevp[i].data >> 32;
+		ks32[i].data2 = kevp[i].data;
+#endif
 		PTROUT_CP(kevp[i], ks32[i], udata);
+		for (j = 0; j < nitems(kevp->ext); j++) {
+			e = kevp[i].ext[j];
+#if BYTE_ORDER == LITTLE_ENDIAN
+			ks32[i].ext64[2 * j] = e;
+			ks32[i].ext64[2 * j + 1] = e >> 32;
+#else
+			ks32[i].ext64[2 * j] = e >> 32;
+			ks32[i].ext64[2 * j + 1] = e;
+#endif
+		}
 	}
 	error = copyout(ks32, uap->eventlist, count * sizeof *ks32);
 	if (error == 0)
@@ -649,7 +666,8 @@ freebsd32_kevent_copyin(void *arg, struct kevent *kevp, int count)
 {
 	struct freebsd32_kevent_args *uap;
 	struct kevent32	ks32[KQ_NEVENTS];
-	int i, error = 0;
+	uint64_t e;
+	int i, j, error;
 
 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
 	uap = (struct freebsd32_kevent_args *)arg;
@@ -664,8 +682,20 @@ freebsd32_kevent_copyin(void *arg, struct kevent *kevp, int count)
 		CP(ks32[i], kevp[i], filter);
 		CP(ks32[i], kevp[i], flags);
 		CP(ks32[i], kevp[i], fflags);
-		CP(ks32[i], kevp[i], data);
+		kevp[i].data = PAIR32TO64(uint64_t, ks32[i].data);
 		PTRIN_CP(ks32[i], kevp[i], udata);
+		for (j = 0; j < nitems(kevp->ext); j++) {
+#if BYTE_ORDER == LITTLE_ENDIAN
+			e = ks32[i].ext64[2 * j + 1];
+			e <<= 32;
+			e += ks32[i].ext64[2 * j];
+#else
+			e = ks32[i].ext64[2 * j];
+			e <<= 32;
+			e += ks32[i].ext64[2 * j + 1];
+#endif
+			kevp[i].ext[j] = e;
+		}
 	}
 done:
 	return (error);
@@ -683,6 +713,98 @@ freebsd32_kevent(struct thread *td, struct freebsd32_kevent_args *uap)
 	};
 	int error;
 
+	if (uap->timeout) {
+		error = copyin(uap->timeout, &ts32, sizeof(ts32));
+		if (error)
+			return (error);
+		CP(ts32, ts, tv_sec);
+		CP(ts32, ts, tv_nsec);
+		tsp = &ts;
+	} else
+		tsp = NULL;
+	error = kern_kevent(td, uap->fd, uap->nchanges, uap->nevents,
+	    &k_ops, tsp);
+	return (error);
+}
+
+#ifdef COMPAT_FREEBSD11
+struct kevent32_freebsd11 {
+	u_int32_t	ident;		/* identifier for this event */
+	short		filter;		/* filter for event */
+	u_short		flags;
+	u_int		fflags;
+	int32_t		data;
+	u_int32_t	udata;		/* opaque user data identifier */
+};
+
+static int
+freebsd32_kevent11_copyout(void *arg, struct kevent *kevp, int count)
+{
+	struct freebsd11_freebsd32_kevent_args *uap;
+	struct kevent32_freebsd11 ks32[KQ_NEVENTS];
+	int i, error;
+
+	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
+	uap = (struct freebsd11_freebsd32_kevent_args *)arg;
+
+	for (i = 0; i < count; i++) {
+		CP(kevp[i], ks32[i], ident);
+		CP(kevp[i], ks32[i], filter);
+		CP(kevp[i], ks32[i], flags);
+		CP(kevp[i], ks32[i], fflags);
+		CP(kevp[i], ks32[i], data);
+		PTROUT_CP(kevp[i], ks32[i], udata);
+	}
+	error = copyout(ks32, uap->eventlist, count * sizeof *ks32);
+	if (error == 0)
+		uap->eventlist += count;
+	return (error);
+}
+
+/*
+ * Copy 'count' items from the list pointed to by uap->changelist.
+ */
+static int
+freebsd32_kevent11_copyin(void *arg, struct kevent *kevp, int count)
+{
+	struct freebsd11_freebsd32_kevent_args *uap;
+	struct kevent32_freebsd11 ks32[KQ_NEVENTS];
+	int i, j, error;
+
+	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
+	uap = (struct freebsd11_freebsd32_kevent_args *)arg;
+
+	error = copyin(uap->changelist, ks32, count * sizeof *ks32);
+	if (error)
+		goto done;
+	uap->changelist += count;
+
+	for (i = 0; i < count; i++) {
+		CP(ks32[i], kevp[i], ident);
+		CP(ks32[i], kevp[i], filter);
+		CP(ks32[i], kevp[i], flags);
+		CP(ks32[i], kevp[i], fflags);
+		CP(ks32[i], kevp[i], data);
+		PTRIN_CP(ks32[i], kevp[i], udata);
+		for (j = 0; j < nitems(kevp->ext); j++)
+			kevp[i].ext[j] = 0;
+	}
+done:
+	return (error);
+}
+
+int
+freebsd11_freebsd32_kevent(struct thread *td,
+    struct freebsd11_freebsd32_kevent_args *uap)
+{
+	struct timespec32 ts32;
+	struct timespec ts, *tsp;
+	struct kevent_copyops k_ops = {
+		.arg = uap,
+		.k_copyout = freebsd32_kevent11_copyout,
+		.k_copyin = freebsd32_kevent11_copyin,
+	};
+	int error;
 
 	if (uap->timeout) {
 		error = copyin(uap->timeout, &ts32, sizeof(ts32));
@@ -697,6 +819,7 @@ freebsd32_kevent(struct thread *td, struct freebsd32_kevent_args *uap)
 	    &k_ops, tsp);
 	return (error);
 }
+#endif
 
 int
 freebsd32_gettimeofday(struct thread *td,

sys/compat/freebsd32/syscalls.master

@@ -667,10 +667,12 @@
 361	AUE_GETRESGID	NOPROTO	{ int getresgid(gid_t *rgid, gid_t *egid, \
 				    gid_t *sgid); }
 362	AUE_KQUEUE	NOPROTO	{ int kqueue(void); }
-363	AUE_KEVENT	STD	{ int freebsd32_kevent(int fd, \
-				    const struct kevent32 *changelist, \
+363	AUE_KEVENT	COMPAT11 { int freebsd32_kevent(int fd, \
+				    const struct kevent32_freebsd11 * \
+				    changelist, \
 				    int nchanges, \
-				    struct kevent32 *eventlist, int nevents, \
+				    struct kevent32_freebsd11 *eventlist, \
+				    int nevents, \
 				    const struct timespec32 *timeout); }
 364	AUE_NULL	UNIMPL	__cap_get_proc
 365	AUE_NULL	UNIMPL	__cap_set_proc
@@ -1111,3 +1113,9 @@
 				    struct statfs32 *buf); }
 559	AUE_MKNODAT	NOPROTO	{ int mknodat(int fd, char *path, mode_t mode, \
 				    dev_t dev); }
+560	AUE_KEVENT	STD	{ int freebsd32_kevent(int fd, \
+				    const struct kevent32 *changelist, \
+				    int nchanges, \
+				    struct kevent32 *eventlist, \
+				    int nevents, \
+				    const struct timespec32 *timeout); }