MFhead@r344270

UPDATING

@@ -38,6 +38,12 @@ NOTE TO PEOPLE WHO THINK THAT FreeBSD 13.x IS SLOW:
 	modules on kernels not having 'device iflib', the iflib.ko module
 	is loaded automatically.
 
+20190125:
+	The IEEE80211_AMPDU_AGE and AH_SUPPORT_AR5416 kernel configuration
+	options no longer exist since r343219 and r343427 respectively;
+	nothing uses them, so they should be just removed from custom
+	kernel config files.
+
 20181230:
 	r342635 changes the way efibootmgr(8) works by requiring users to add
 	the -b (bootnum) parameter for commands where the bootnum was previously
@@ -231,7 +237,7 @@ NOTE TO PEOPLE WHO THINK THAT FreeBSD 13.x IS SLOW:
 20180719:
 	ARM64 now have efifb support, if you want to have serial console
 	on your arm64 board when an screen is connected and the bootloader
-	setup a frambuffer for us to use, just add :
+	setup a framebuffer for us to use, just add :
 	boot_serial=YES
 	boot_multicons=YES
 	in /boot/loader.conf

contrib/libarchive/libarchive/archive_read_disk_posix.c

@@ -860,6 +860,7 @@ next_entry(struct archive_read_disk *a, struct tree *t,
 	struct archive_string delayed_str;
 
 	delayed = ARCHIVE_OK;
+	delayed_errno = 0;
 	archive_string_init(&delayed_str);
 
 	st = NULL;

contrib/libc++/include/type_traits

@@ -733,12 +733,6 @@ _LIBCPP_INLINE_VAR _LIBCPP_CONSTEXPR bool is_integral_v
 // is_floating_point
 
 template <class _Tp> struct __libcpp_is_floating_point              : public false_type {};
-#ifdef __clang__
-template <>          struct __libcpp_is_floating_point<__fp16>      : public true_type {};
-#endif
-#ifdef __FLT16_MANT_DIG__
-template <>          struct __libcpp_is_floating_point<_Float16>    : public true_type {};
-#endif
 template <>          struct __libcpp_is_floating_point<float>       : public true_type {};
 template <>          struct __libcpp_is_floating_point<double>      : public true_type {};
 template <>          struct __libcpp_is_floating_point<long double> : public true_type {};

contrib/llvm/lib/MC/ELFObjectWriter.cpp

@@ -1258,14 +1258,20 @@ void ELFObjectWriter::executePostLayoutBinding(MCAssembler &Asm,
     if (!Symbol.isUndefined() && !Rest.startswith("@@@"))
       continue;
 
-    // FIXME: produce a better error message.
+    // FIXME: Get source locations for these errors or diagnose them earlier.
     if (Symbol.isUndefined() && Rest.startswith("@@") &&
-        !Rest.startswith("@@@"))
+        !Rest.startswith("@@@")) {
-      report_fatal_error("A @@ version cannot be undefined");
+      Asm.getContext().reportError(SMLoc(), "versioned symbol " + AliasName +
+                                                " must be defined");
+      continue;
+    }
 
-    if (Renames.count(&Symbol) && Renames[&Symbol] != Alias)
+    if (Renames.count(&Symbol) && Renames[&Symbol] != Alias) {
-      report_fatal_error(llvm::Twine("Multiple symbol versions defined for ") +
+      Asm.getContext().reportError(
-                         Symbol.getName());
+          SMLoc(), llvm::Twine("multiple symbol versions defined for ") +
+                       Symbol.getName());
+      continue;
+    }
 
     Renames.insert(std::make_pair(&Symbol, Alias));
   }

etc/mtree/BSD.root.dist

@@ -26,6 +26,8 @@
         ..
         modules
         ..
+        uboot
+        ..
         zfs
         ..
     ..

kerberos5/tools/asn1_compile/Makefile

@@ -6,6 +6,7 @@ LIBROKEN_A=	${.OBJDIR:H:H}/lib/libroken/libroken.a
 LIBADD=	vers
 LDADD=	${LIBROKEN_A}
 DPADD=	${LIBROKEN_A}
+MK_PIE:=	no
 
 SRCS=	\
 	asn1parse.y \

kerberos5/tools/slc/Makefile

@@ -6,6 +6,7 @@ LIBADD=	vers
 LDADD=  ${LIBROKEN_A}
 DPADD=  ${LIBROKEN_A}
 MAN=
+MK_PIE:=	no
 
 SRCS=	roken.h \
 	slc-gram.y \

lib/clang/Makefile.inc

@@ -2,6 +2,8 @@
 
 .include <bsd.compiler.mk>
 
+MK_PIE:=	no	# Explicit libXXX.a references
+
 .if ${COMPILER_TYPE} == "clang"
 DEBUG_FILES_CFLAGS= -gline-tables-only
 .else

lib/clang/libllvmminimal/Makefile

@@ -18,6 +18,7 @@ SRCS+=		Support/Errno.cpp
 SRCS+=		Support/Error.cpp
 SRCS+=		Support/ErrorHandling.cpp
 SRCS+=		Support/FoldingSet.cpp
+SRCS+=		Support/FormatVariadic.cpp
 SRCS+=		Support/FormattedStream.cpp
 SRCS+=		Support/Hashing.cpp
 SRCS+=		Support/Host.cpp

lib/libbe/be.c

@@ -45,6 +45,11 @@ __FBSDID("$FreeBSD$");
 #include "be.h"
 #include "be_impl.h"
 
+struct be_destroy_data {
+	libbe_handle_t		*lbh;
+	char			*snapname;
+};
+
 #if SOON
 static int be_create_child_noent(libbe_handle_t *lbh, const char *active,
     const char *child_path);
@@ -186,12 +191,38 @@ be_nicenum(uint64_t num, char *buf, size_t buflen)
 static int
 be_destroy_cb(zfs_handle_t *zfs_hdl, void *data)
 {
+	char path[BE_MAXPATHLEN];
+	struct be_destroy_data *bdd;
+	zfs_handle_t *snap;
 	int err;
 
-	if ((err = zfs_iter_children(zfs_hdl, be_destroy_cb, data)) != 0)
+	bdd = (struct be_destroy_data *)data;
-		return (err);
+	if (bdd->snapname == NULL) {
-	if ((err = zfs_destroy(zfs_hdl, false)) != 0)
+		err = zfs_iter_children(zfs_hdl, be_destroy_cb, data);
+		if (err != 0)
+			return (err);
+		return (zfs_destroy(zfs_hdl, false));
+	}
+	/* If we're dealing with snapshots instead, delete that one alone */
+	err = zfs_iter_filesystems(zfs_hdl, be_destroy_cb, data);
+	if (err != 0)
 		return (err);
+	/*
+	 * This part is intentionally glossing over any potential errors,
+	 * because there's a lot less potential for errors when we're cleaning
+	 * up snapshots rather than a full deep BE.  The primary error case
+	 * here being if the snapshot doesn't exist in the first place, which
+	 * the caller will likely deem insignificant as long as it doesn't
+	 * exist after the call.  Thus, such a missing snapshot shouldn't jam
+	 * up the destruction.
+	 */
+	snprintf(path, sizeof(path), "%s@%s", zfs_get_name(zfs_hdl),
+	    bdd->snapname);
+	if (!zfs_dataset_exists(bdd->lbh->lzh, path, ZFS_TYPE_SNAPSHOT))
+		return (0);
+	snap = zfs_open(bdd->lbh->lzh, path, ZFS_TYPE_SNAPSHOT);
+	if (snap != NULL)
+		zfs_destroy(snap, false);
 	return (0);
 }
 
@@ -199,22 +230,26 @@ be_destroy_cb(zfs_handle_t *zfs_hdl, void *data)
  * Destroy the boot environment or snapshot specified by the name
  * parameter. Options are or'd together with the possible values:
  * BE_DESTROY_FORCE : forces operation on mounted datasets
+ * BE_DESTROY_ORIGIN: destroy the origin snapshot as well
  */
 int
 be_destroy(libbe_handle_t *lbh, const char *name, int options)
 {
+	struct be_destroy_data bdd;
 	char origin[BE_MAXPATHLEN], path[BE_MAXPATHLEN];
 	zfs_handle_t *fs;
-	char *p;
+	char *snapdelim;
 	int err, force, mounted;
+	size_t rootlen;
 
-	p = path;
+	bdd.lbh = lbh;
+	bdd.snapname = NULL;
 	force = options & BE_DESTROY_FORCE;
 	*origin = '\0';
 
 	be_root_concat(lbh, name, path);
 
-	if (strchr(name, '@') == NULL) {
+	if ((snapdelim = strchr(path, '@')) == NULL) {
 		if (!zfs_dataset_exists(lbh->lzh, path, ZFS_TYPE_FILESYSTEM))
 			return (set_error(lbh, BE_ERR_NOENT));
 
@@ -222,9 +257,10 @@ be_destroy(libbe_handle_t *lbh, const char *name, int options)
 		    strcmp(path, lbh->bootfs) == 0)
 			return (set_error(lbh, BE_ERR_DESTROYACT));
 
-		fs = zfs_open(lbh->lzh, p, ZFS_TYPE_FILESYSTEM);
+		fs = zfs_open(lbh->lzh, path, ZFS_TYPE_FILESYSTEM);
 		if (fs == NULL)
 			return (set_error(lbh, BE_ERR_ZFSOPEN));
+
 		if ((options & BE_DESTROY_ORIGIN) != 0 &&
 		    zfs_prop_get(fs, ZFS_PROP_ORIGIN, origin, sizeof(origin),
 		    NULL, NULL, 0, 1) != 0)
@@ -233,41 +269,57 @@ be_destroy(libbe_handle_t *lbh, const char *name, int options)
 		if (!zfs_dataset_exists(lbh->lzh, path, ZFS_TYPE_SNAPSHOT))
 			return (set_error(lbh, BE_ERR_NOENT));
 
-		fs = zfs_open(lbh->lzh, p, ZFS_TYPE_SNAPSHOT);
+		bdd.snapname = strdup(snapdelim + 1);
-		if (fs == NULL)
+		if (bdd.snapname == NULL)
+			return (set_error(lbh, BE_ERR_NOMEM));
+		*snapdelim = '\0';
+		fs = zfs_open(lbh->lzh, path, ZFS_TYPE_DATASET);
+		if (fs == NULL) {
+			free(bdd.snapname);
 			return (set_error(lbh, BE_ERR_ZFSOPEN));
+		}
 	}
 
 	/* Check if mounted, unmount if force is specified */
 	if ((mounted = zfs_is_mounted(fs, NULL)) != 0) {
-		if (force)
+		if (force) {
 			zfs_unmount(fs, NULL, 0);
-		else
+		} else {
+			free(bdd.snapname);
 			return (set_error(lbh, BE_ERR_DESTROYMNT));
+		}
 	}
 
-	if ((err = be_destroy_cb(fs, NULL)) != 0) {
+	err = be_destroy_cb(fs, &bdd);
+	zfs_close(fs);
+	free(bdd.snapname);
+	if (err != 0) {
 		/* Children are still present or the mount is referenced */
 		if (err == EBUSY)
 			return (set_error(lbh, BE_ERR_DESTROYMNT));
 		return (set_error(lbh, BE_ERR_UNKNOWN));
 	}
 
-	if (*origin != '\0') {
+	if ((options & BE_DESTROY_ORIGIN) == 0)
-		fs = zfs_open(lbh->lzh, origin, ZFS_TYPE_SNAPSHOT);
+		return (0);
-		if (fs == NULL)
-			return (set_error(lbh, BE_ERR_ZFSOPEN));
-		err = zfs_destroy(fs, false);
-		if (err == EBUSY)
-			return (set_error(lbh, BE_ERR_DESTROYMNT));
-		else if (err != 0)
-			return (set_error(lbh, BE_ERR_UNKNOWN));
-	}
 
-	return (0);
+	/* The origin can't possibly be shorter than the BE root */
+	rootlen = strlen(lbh->root);
+	if (*origin == '\0' || strlen(origin) <= rootlen + 1)
+		return (set_error(lbh, BE_ERR_INVORIGIN));
+
+	/*
+	 * We'll be chopping off the BE root and running this back through
+	 * be_destroy, so that we properly handle the origin snapshot whether
+	 * it be that of a deep BE or not.
+	 */
+	if (strncmp(origin, lbh->root, rootlen) != 0 || origin[rootlen] != '/')
+		return (0);
+
+	return (be_destroy(lbh, origin + rootlen + 1,
+	    options & ~BE_DESTROY_ORIGIN));
 }
 
-
 int
 be_snapshot(libbe_handle_t *lbh, const char *source, const char *snap_name,
     bool recursive, char *result)

lib/libbe/be.h

@@ -59,6 +59,7 @@ typedef enum be_error {
 	BE_ERR_NOPOOL,		/* operation not supported on this pool */
 	BE_ERR_NOMEM,		/* insufficient memory */
 	BE_ERR_UNKNOWN,         /* unknown error */
+	BE_ERR_INVORIGIN,       /* invalid origin */
 } be_error_t;
 
 

lib/libbe/be_error.c

@@ -105,6 +105,9 @@ libbe_error_description(libbe_handle_t *lbh)
 	case BE_ERR_UNKNOWN:
 		return ("unknown error");
 
+	case BE_ERR_INVORIGIN:
+		return ("invalid origin");
+
 	default:
 		assert(lbh->error == BE_ERR_SUCCESS);
 		return ("no error");

lib/libbe/libbe.3

@@ -28,7 +28,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd February 11, 2019
+.Dd February 12, 2019
 .Dt LIBBE 3
 .Os
 .Sh NAME
@@ -489,6 +489,8 @@ BE_ERR_NOPOOL
 BE_ERR_NOMEM
 .It
 BE_ERR_UNKNOWN
+.It
+BE_ERR_INVORIGIN
 .El
 .Sh SEE ALSO
 .Xr bectl 8

lib/libc/Makefile

@@ -122,6 +122,7 @@ NOASM=
 .endif
 .if ${LIBC_ARCH} == "i386" || ${LIBC_ARCH} == "amd64"
 .include "${LIBC_SRCTOP}/x86/sys/Makefile.inc"
+.include "${LIBC_SRCTOP}/x86/gen/Makefile.inc"
 .endif
 .if ${MK_NIS} != "no"
 CFLAGS+= -DYP

lib/libc/amd64/gen/Makefile.inc

@@ -2,7 +2,7 @@
 # $FreeBSD$
 
 SRCS+=	_setjmp.S _set_tp.c rfork_thread.S setjmp.S sigsetjmp.S \
-	fabs.S getcontextx.c \
+	fabs.S \
 	infinity.c ldexp.c makecontext.c signalcontext.c \
 	flt_rounds.c fpgetmask.c fpsetmask.c fpgetprec.c fpsetprec.c \
 	fpgetround.c fpsetround.c fpgetsticky.c

lib/libc/gen/readpassphrase.3

@@ -178,4 +178,6 @@ extension and should not be used if portability is desired.
 The
 .Fn readpassphrase
 function first appeared in
+.Fx 4.6
+and
 .Ox 2.9 .

lib/libc/i386/gen/Makefile.inc

@@ -2,5 +2,5 @@
 # $FreeBSD$
 
 SRCS+=	_ctx_start.S _setjmp.S _set_tp.c fabs.S \
-	flt_rounds.c getcontextx.c infinity.c ldexp.c makecontext.c \
+	flt_rounds.c infinity.c ldexp.c makecontext.c \
 	rfork_thread.S setjmp.S signalcontext.c sigsetjmp.S

lib/libc/i386/gen/getcontextx.c

@@ -1,145 +0,0 @@
-/*-
- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
- *
- * Copyright (c) 2011 Konstantin Belousov <kib@FreeBSD.org>
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- *
- * 1. Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
- * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
- * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
-#include <sys/types.h>
-#include <sys/ucontext.h>
-#include <errno.h>
-#include <stdarg.h>
-#include <stdlib.h>
-#include <machine/npx.h>
-#include <machine/specialreg.h>
-#include <machine/sysarch.h>
-
-static int xstate_sz = -1;
-
-int
-__getcontextx_size(void)
-{
-	u_int p[4];
-	int cpuid_supported;
-
-	if (xstate_sz == -1) {
-		__asm __volatile(
-		    "	pushfl\n"
-		    "	popl	%%eax\n"
-		    "	movl    %%eax,%%ecx\n"
-		    "	xorl    $0x200000,%%eax\n"
-		    "	pushl	%%eax\n"
-		    "	popfl\n"
-		    "	pushfl\n"
-		    "	popl    %%eax\n"
-		    "	xorl    %%eax,%%ecx\n"
-		    "	je	1f\n"
-		    "	movl	$1,%0\n"
-		    "	jmp	2f\n"
-		    "1:	movl	$0,%0\n"
-		    "2:\n"
-		    : "=r" (cpuid_supported) : : "eax", "ecx");
-		if (cpuid_supported) {
-			__asm __volatile(
-			    "	pushl	%%ebx\n"
-			    "	cpuid\n"
-			    "	movl	%%ebx,%1\n"
-			    "	popl	%%ebx\n"
-			    : "=a" (p[0]), "=r" (p[1]), "=c" (p[2]), "=d" (p[3])
-			    :  "0" (0x1));
-			if ((p[2] & CPUID2_OSXSAVE) != 0) {
-				__asm __volatile(
-				    "	pushl	%%ebx\n"
-				    "	cpuid\n"
-				    "	movl	%%ebx,%1\n"
-				    "	popl	%%ebx\n"
-				    : "=a" (p[0]), "=r" (p[1]), "=c" (p[2]),
-					"=d" (p[3])
-				    :  "0" (0xd), "2" (0x0));
-				xstate_sz = p[1] - sizeof(struct savexmm);
-			} else
-				xstate_sz = 0;
-		} else
-			xstate_sz = 0;
-	}
-
-	return (sizeof(ucontext_t) + xstate_sz);
-}
-
-int
-__fillcontextx2(char *ctx)
-{
-	struct i386_get_xfpustate xfpu;
-	ucontext_t *ucp;
-
-	ucp = (ucontext_t *)ctx;
-	if (xstate_sz != 0) {
-		xfpu.addr = (char *)(ucp + 1);
-		xfpu.len = xstate_sz;
-		if (sysarch(I386_GET_XFPUSTATE, &xfpu) == -1)
-			return (-1);
-		ucp->uc_mcontext.mc_xfpustate = (__register_t)xfpu.addr;
-		ucp->uc_mcontext.mc_xfpustate_len = xstate_sz;
-		ucp->uc_mcontext.mc_flags |= _MC_HASFPXSTATE;
-	} else {
-		ucp->uc_mcontext.mc_xfpustate = 0;
-		ucp->uc_mcontext.mc_xfpustate_len = 0;
-	}
-	return (0);
-}
-
-int
-__fillcontextx(char *ctx)
-{
-	ucontext_t *ucp;
-
-	ucp = (ucontext_t *)ctx;
-	if (getcontext(ucp) == -1)
-		return (-1);
-	__fillcontextx2(ctx);
-	return (0);
-}
-
-__weak_reference(__getcontextx, getcontextx);
-
-ucontext_t *
-__getcontextx(void)
-{
-	char *ctx;
-	int error;
-
-	ctx = malloc(__getcontextx_size());
-	if (ctx == NULL)
-		return (NULL);
-	if (__fillcontextx(ctx) == -1) {
-		error = errno;
-		free(ctx);
-		errno = error;
-		return (NULL);
-	}
-	return ((ucontext_t *)ctx);
-}

lib/libc/sys/sendfile.2

@@ -25,7 +25,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd January 25, 2019
+.Dd February 15, 2019
 .Dt SENDFILE 2
 .Os
 .Sh NAME
@@ -224,6 +224,19 @@ implementation of
 .Fn sendfile
 is "zero-copy", meaning that it has been optimized so that copying of the file data is avoided.
 .Sh TUNING
+.Ss physical paging buffers
+.Fn sendfile
+uses vnode pager to read file pages into memory.
+The pager uses a pool of physical buffers to run its I/O operations.
+When system runs out of pbufs, sendfile will block and report state
+.Dq Li zonelimit .
+Size of the pool can be tuned with
+.Va vm.vnode_pbufs
+.Xr loader.conf 5
+tunable and can be checked with
+.Xr sysctl 8
+OID of the same name at runtime.
+.Ss sendfile(2) buffers
 On some architectures, this system call internally uses a special
 .Fn sendfile
 buffer
@@ -279,9 +292,11 @@ buffers usage respectively.
 These values may also be viewed through
 .Nm netstat Fl m .
 .Pp
-If a value of zero is reported for
+If
-.Va kern.ipc.nsfbufs ,
+.Xr sysctl 8
-your architecture does not need to use
+OID
+.Va kern.ipc.nsfbufs
+doesn't exist, your architecture does not need to use
 .Fn sendfile
 buffers because their task can be efficiently performed
 by the generic virtual memory structures.
@@ -363,11 +378,13 @@ does not support
 The socket peer has closed the connection.
 .El
 .Sh SEE ALSO
+.Xr loader.conf 5 ,
 .Xr netstat 1 ,
 .Xr open 2 ,
 .Xr send 2 ,
 .Xr socket 2 ,
 .Xr writev 2 ,
+.Xr sysctl 8 ,
 .Xr tuning 7
 .Rs
 .%A K. Elmeleegy

lib/libc/x86/gen/Makefile.inc

@@ -0,0 +1,6 @@
+# $FreeBSD$
+
+.PATH:	${LIBC_SRCTOP}/x86/gen
+
+SRCS+= \
+	getcontextx.c

lib/libc/x86/gen/getcontextx.c

@@ -35,51 +35,78 @@ __FBSDID("$FreeBSD$");
 #include <stdarg.h>
 #include <stdlib.h>
 #include <machine/cpufunc.h>
-#include <machine/fpu.h>
 #include <machine/specialreg.h>
 #include <machine/sysarch.h>
+#include <x86/ifunc.h>
+#include <x86/fpu.h>
 
-static int xstate_sz = -1;
+#if defined __i386__
+#define	X86_GET_XFPUSTATE	I386_GET_XFPUSTATE
+typedef struct savexmm savex86_t ;
+typedef struct i386_get_xfpustate x86_get_xfpustate_t;
+#elif defined __amd64__
+#define	X86_GET_XFPUSTATE	AMD64_GET_XFPUSTATE
+typedef struct savefpu savex86_t;
+typedef struct amd64_get_xfpustate x86_get_xfpustate_t;
+#else
+#error "Wrong arch"
+#endif
 
-int
+static int xstate_sz = 0;
-__getcontextx_size(void)
+
+static int
+__getcontextx_size_xfpu(void)
 {
-	u_int p[4];
-
-	if (xstate_sz == -1) {
-		do_cpuid(1, p);
-		if ((p[2] & CPUID2_OSXSAVE) != 0) {
-			cpuid_count(0xd, 0x0, p);
-			xstate_sz = p[1] - sizeof(struct savefpu);
-		} else
-			xstate_sz = 0;
-	}
 
 	return (sizeof(ucontext_t) + xstate_sz);
 }
 
-int
+DEFINE_UIFUNC(, int, __getcontextx_size, (void), static)
-__fillcontextx2(char *ctx)
 {
-	struct amd64_get_xfpustate xfpu;
+	u_int p[4];
+
+	if ((cpu_feature2 & CPUID2_OSXSAVE) != 0) {
+		cpuid_count(0xd, 0x0, p);
+		xstate_sz = p[1] - sizeof(savex86_t);
+	}
+	return (__getcontextx_size_xfpu);
+}
+
+static int
+__fillcontextx2_xfpu(char *ctx)
+{
+	x86_get_xfpustate_t xfpu;
 	ucontext_t *ucp;
 
 	ucp = (ucontext_t *)ctx;
-	if (xstate_sz != 0) {
+	xfpu.addr = (char *)(ucp + 1);
-		xfpu.addr = (char *)(ucp + 1);
+	xfpu.len = xstate_sz;
-		xfpu.len = xstate_sz;
+	if (sysarch(X86_GET_XFPUSTATE, &xfpu) == -1)
-		if (sysarch(AMD64_GET_XFPUSTATE, &xfpu) == -1)
+		return (-1);
-			return (-1);
+	ucp->uc_mcontext.mc_xfpustate = (__register_t)xfpu.addr;
-		ucp->uc_mcontext.mc_xfpustate = (__register_t)xfpu.addr;
+	ucp->uc_mcontext.mc_xfpustate_len = xstate_sz;
-		ucp->uc_mcontext.mc_xfpustate_len = xstate_sz;
+	ucp->uc_mcontext.mc_flags |= _MC_HASFPXSTATE;
-		ucp->uc_mcontext.mc_flags |= _MC_HASFPXSTATE;
-	} else {
-		ucp->uc_mcontext.mc_xfpustate = 0;
-		ucp->uc_mcontext.mc_xfpustate_len = 0;
-	}
 	return (0);
 }
 
+static int
+__fillcontextx2_noxfpu(char *ctx)
+{
+	ucontext_t *ucp;
+
+	ucp = (ucontext_t *)ctx;
+	ucp->uc_mcontext.mc_xfpustate = 0;
+	ucp->uc_mcontext.mc_xfpustate_len = 0;
+	return (0);
+}
+
+DEFINE_UIFUNC(, int, __fillcontextx2, (char *), static)
+{
+
+	return ((cpu_feature2 & CPUID2_OSXSAVE) != 0 ? __fillcontextx2_xfpu : 
+	    __fillcontextx2_noxfpu);
+}
+
 int
 __fillcontextx(char *ctx)
 {

lib/libc/x86/sys/__vdso_gettc.c

@@ -53,31 +53,6 @@ __FBSDID("$FreeBSD$");
 #include <x86/ifunc.h>
 #include "libc_private.h"
 
-static void
-cpuidp(u_int leaf, u_int p[4])
-{
-
-	__asm __volatile(
-#if defined(__i386__)
-	    "	pushl	%%ebx\n"
-#endif
-	    "	cpuid\n"
-#if defined(__i386__)
-	    "	movl	%%ebx,%1\n"
-	    "	popl	%%ebx"
-#endif
-	    : "=a" (p[0]),
-#if defined(__i386__)
-	    "=r" (p[1]),
-#elif defined(__amd64__)
-	    "=b" (p[1]),
-#else
-#error "Arch"
-#endif
-	    "=c" (p[2]), "=d" (p[3])
-	    :  "0" (leaf));
-}
-
 static void
 rdtsc_mb_lfence(void)
 {
@@ -100,12 +75,12 @@ rdtsc_mb_none(void)
 DEFINE_UIFUNC(static, void, rdtsc_mb, (void), static)
 {
 	u_int p[4];
-	/* Not a typo, string matches our cpuidp() registers use. */
+	/* Not a typo, string matches our do_cpuid() registers use. */
 	static const char intel_id[] = "GenuntelineI";
 
 	if ((cpu_feature & CPUID_SSE2) == 0)
 		return (rdtsc_mb_none);
-	cpuidp(0, p);
+	do_cpuid(0, p);
 	return (memcmp(p + 1, intel_id, sizeof(intel_id) - 1) == 0 ?
 	    rdtsc_mb_lfence : rdtsc_mb_mfence);
 }

lib/libmemstat/memstat_uma.c

@@ -213,6 +213,15 @@ memstat_sysctl_uma(struct memory_type_list *list, int flags)
 			mtp->mt_numfrees += upsp->ups_frees;
 		}
 
+		/*
+		 * Values for uth_allocs and uth_frees frees are snap.
+		 * It may happen that kernel reports that number of frees
+		 * is greater than number of allocs. See counter(9) for
+		 * details.
+		 */
+		if (mtp->mt_numallocs < mtp->mt_numfrees)
+			mtp->mt_numallocs = mtp->mt_numfrees;
+
 		mtp->mt_size = uthp->uth_size;
 		mtp->mt_rsize = uthp->uth_rsize;
 		mtp->mt_memalloced = mtp->mt_numallocs * uthp->uth_size;

lib/libthr/arch/powerpc/include/pthread_md.h

@@ -72,14 +72,15 @@ _tcb_set(struct tcb *tcb)
 static __inline struct tcb *
 _tcb_get(void)
 {
-	register uint8_t *_tp;
+        register struct tcb *tcb;
+
 #ifdef __powerpc64__
-	__asm __volatile("mr %0,13" : "=r"(_tp));
+	__asm __volatile("addi %0,13,%1" : "=r"(tcb) : "i"(-TP_OFFSET));
 #else
-	__asm __volatile("mr %0,2" : "=r"(_tp));
+	__asm __volatile("addi %0,2,%1" : "=r"(tcb) : "i"(-TP_OFFSET));
 #endif
 
-	return ((struct tcb *)(_tp - TP_OFFSET));
+	return (tcb);
 }
 
 static __inline struct pthread *

libexec/rc/rc.d/nfsd

@@ -13,6 +13,7 @@ name="nfsd"
 desc="Remote NFS server"
 rcvar="nfs_server_enable"
 command="/usr/sbin/${name}"
+nfs_server_vhost=""
 
 load_rc_config $name
 start_precmd="nfsd_precmd"
@@ -20,6 +21,7 @@ sig_stop="USR1"
 
 nfsd_precmd()
 {
+	local	_vhost
 	rc_flags="${nfs_server_flags}"
 
 	# Load the modules now, so that the vfs.nfsd sysctl
@@ -46,6 +48,9 @@ nfsd_precmd()
 
 	force_depend rpcbind || return 1
 	force_depend mountd || return 1
+	if [ -n "${nfs_server_vhost}" ]; then
+		command_args="-V \"${nfs_server_vhost}\""
+	fi
 }
 
 run_rc_command "$1"

libexec/rtld-elf/Makefile

@@ -7,6 +7,7 @@
 .include <src.opts.mk>
 PACKAGE=	clibs
 MK_BIND_NOW=	no
+MK_PIE=		no # Always position independent using local rules
 MK_SSP=		no
 
 CONFS=		libmap.conf

sbin/mdmfs/mdmfs.c

@@ -196,6 +196,7 @@ main(int argc, char **argv)
 				usage();
 			mdtype = MD_MALLOC;
 			have_mdtype = true;
+			argappend(&mdconfig_arg, "-o reserve");
 			break;
 		case 'm':
 			argappend(&newfs_arg, "-m %s", optarg);

sbin/nvmecontrol/firmware.c

@@ -177,7 +177,7 @@ static void
 firmware(const struct nvme_function *nf, int argc, char *argv[])
 {
 	int				fd = -1, slot = 0;
-	int				a_flag, s_flag, f_flag;
+	int				a_flag, f_flag;
 	int				activate_action, reboot_required;
 	int				opt;
 	char				*p, *image = NULL;
@@ -188,7 +188,7 @@ firmware(const struct nvme_function *nf, int argc, char *argv[])
 	uint8_t				fw_slot1_ro, fw_num_slots;
 	struct nvme_controller_data	cdata;
 
-	a_flag = s_flag = f_flag = false;
+	a_flag = f_flag = false;
 
 	while ((opt = getopt(argc, argv, "af:s:")) != -1) {
 		switch (opt) {
@@ -214,7 +214,6 @@ firmware(const struct nvme_function *nf, int argc, char *argv[])
 				    "7.\n", optarg);
 				usage(nf);
 			}
-			s_flag = true;
 			break;
 		case 'f':
 			image = optarg;

share/man/man5/src.conf.5

@@ -1,6 +1,6 @@
 .\" DO NOT EDIT-- this file is @generated by tools/build/options/makeman.
 .\" $FreeBSD$
-.Dd January 31, 2019
+.Dd February 15, 2019
 .Dt SRC.CONF 5
 .Os
 .Sh NAME
@@ -406,7 +406,8 @@ Set to build the Clang C/C++ compiler during the bootstrap phase of the build.
 This is a default setting on
 amd64/amd64, arm/arm, arm/armv6, arm/armv7, arm64/aarch64 and i386/i386.
 .It Va WITH_CLANG_EXTRAS
-Set to build additional clang and llvm tools, such as bugpoint.
+Set to build additional clang and llvm tools, such as bugpoint and
+clang-format.
 .It Va WITHOUT_CLANG_FULL
 Set to avoid building the ARCMigrate, Rewriter and StaticAnalyzer components of
 the Clang C/C++ compiler.
@@ -1542,6 +1543,9 @@ When set, it enforces these options:
 .It
 .Va WITHOUT_AUTHPF
 .El
+.It Va WITH_PIE
+Build dynamically linked binaries as
+Position-Independent Executable (PIE).
 .It Va WITHOUT_PKGBOOTSTRAP
 Set to not build
 .Xr pkg 7

share/mk/bsd.lib.mk

@@ -91,13 +91,16 @@ CTFFLAGS+= -g
 # prefer .s to a .c, add .po, remove stuff not used in the BSD libraries
 # .pico used for PIC object files
 # .nossppico used for NOSSP PIC object files
-.SUFFIXES: .out .o .bc .ll .po .pico .nossppico .S .asm .s .c .cc .cpp .cxx .C .f .y .l .ln
+# .pieo used for PIE object files
+.SUFFIXES: .out .o .bc .ll .po .pico .nossppico .pieo .S .asm .s .c .cc .cpp .cxx .C .f .y .l .ln
 
 .if !defined(PICFLAG)
 .if ${MACHINE_CPUARCH} == "sparc64"
 PICFLAG=-fPIC
+PIEFLAG=-fPIE
 .else
 PICFLAG=-fpic
+PIEFLAG=-fpie
 .endif
 .endif
 
@@ -115,6 +118,10 @@ PO_FLAG=-pg
 	${CC} ${PICFLAG} -DPIC ${SHARED_CFLAGS:C/^-fstack-protector.*$//} ${CFLAGS:C/^-fstack-protector.*$//} -c ${.IMPSRC} -o ${.TARGET}
 	${CTFCONVERT_CMD}
 
+.c.pieo:
+	${CC} ${PIEFLAG} -DPIC ${SHARED_CFLAGS} ${CFLAGS} -c ${.IMPSRC} -o ${.TARGET}
+	${CTFCONVERT_CMD}
+
 .cc.po .C.po .cpp.po .cxx.po:
 	${CXX} ${PO_FLAG} ${STATIC_CXXFLAGS} ${PO_CXXFLAGS} -c ${.IMPSRC} -o ${.TARGET}
 
@@ -124,6 +131,9 @@ PO_FLAG=-pg
 .cc.nossppico .C.nossppico .cpp.nossppico .cxx.nossppico:
 	${CXX} ${PICFLAG} -DPIC ${SHARED_CXXFLAGS:C/^-fstack-protector.*$//} ${CXXFLAGS:C/^-fstack-protector.*$//} -c ${.IMPSRC} -o ${.TARGET}
 
+.cc.pieo .C.pieo .cpp.pieo .cxx.pieo:
+	${CXX} ${PIEFLAG} ${SHARED_CXXFLAGS} ${CXXFLAGS} -c ${.IMPSRC} -o ${.TARGET}
+
 .f.po:
 	${FC} -pg ${FFLAGS} -o ${.TARGET} -c ${.IMPSRC}
 	${CTFCONVERT_CMD}
@@ -136,7 +146,7 @@ PO_FLAG=-pg
 	${FC} ${PICFLAG} -DPIC ${FFLAGS:C/^-fstack-protector.*$//} -o ${.TARGET} -c ${.IMPSRC}
 	${CTFCONVERT_CMD}
 
-.s.po .s.pico .s.nossppico:
+.s.po .s.pico .s.nossppico .s.pieo:
 	${AS} ${AFLAGS} -o ${.TARGET} ${.IMPSRC}
 	${CTFCONVERT_CMD}
 
@@ -155,6 +165,11 @@ PO_FLAG=-pg
 	    ${CFLAGS:C/^-fstack-protector.*$//} ${ACFLAGS} -c ${.IMPSRC} -o ${.TARGET}
 	${CTFCONVERT_CMD}
 
+.asm.pieo:
+	${CC:N${CCACHE_BIN}} -x assembler-with-cpp ${PIEFLAG} -DPIC \
+	    ${CFLAGS} ${ACFLAGS} -c ${.IMPSRC} -o ${.TARGET}
+	${CTFCONVERT_CMD}
+
 .S.po:
 	${CC:N${CCACHE_BIN}} -DPROF ${PO_CFLAGS} ${ACFLAGS} -c ${.IMPSRC} \
 	    -o ${.TARGET}
@@ -170,6 +185,11 @@ PO_FLAG=-pg
 	    -c ${.IMPSRC} -o ${.TARGET}
 	${CTFCONVERT_CMD}
 
+.S.pieo:
+	${CC:N${CCACHE_BIN}} ${PIEFLAG} -DPIC ${CFLAGS} ${ACFLAGS} \
+	    -c ${.IMPSRC} -o ${.TARGET}
+	${CTFCONVERT_CMD}
+
 _LIBDIR:=${LIBDIR}
 _SHLIBDIR:=${SHLIBDIR}
 
@@ -334,6 +354,20 @@ lib${LIB_PRIVATE}${LIB}_nossp_pic.a: ${NOSSPSOBJS}
 
 .endif # !defined(INTERNALLIB)
 
+.if defined(INTERNALLIB) && ${MK_PIE} != "no"
+PIEOBJS+=	${OBJS:.o=.pieo}
+DEPENDOBJS+=	${PIEOBJS}
+CLEANFILES+=	${PIEOBJS}
+
+_LIBS+=		lib${LIB_PRIVATE}${LIB}_pie.a
+
+lib${LIB_PRIVATE}${LIB}_pie.a: ${PIEOBJS}
+	@${ECHO} building pie ${LIB} library
+	@rm -f ${.TARGET}
+	${AR} ${ARFLAGS} ${.TARGET} ${PIEOBJS} ${ARADD}
+	${RANLIB} ${RANLIBFLAGS} ${.TARGET}
+.endif
+
 .if defined(_SKIP_BUILD)
 all:
 .else

share/mk/bsd.opts.mk

@@ -73,6 +73,7 @@ __DEFAULT_NO_OPTIONS = \
     CCACHE_BUILD \
     CTF \
     INSTALL_AS_USER \
+    PIE \
     RETPOLINE \
     STALE_STAGED
 

share/mk/bsd.prog.mk

@@ -38,11 +38,16 @@ MK_DEBUG_FILES=	no
 .if ${MK_BIND_NOW} != "no"
 LDFLAGS+= -Wl,-znow
 .endif
+.if ${MK_PIE} != "no" && (!defined(NO_SHARED) || ${NO_SHARED:tl} == "no")
+CFLAGS+= -fPIE
+CXXFLAGS+= -fPIE
+LDFLAGS+= -pie
+.endif
 .if ${MK_RETPOLINE} != "no"
 CFLAGS+= -mretpoline
 CXXFLAGS+= -mretpoline
 # retpolineplt is broken with static linking (PR 233336)
-.if !defined(NO_SHARED) || ${NO_SHARED} == "no" || ${NO_SHARED} == "NO"
+.if !defined(NO_SHARED) || ${NO_SHARED:tl} == "no"
 LDFLAGS+= -Wl,-zretpolineplt
 .endif
 .endif
@@ -68,7 +73,7 @@ TAGS+=		package=${PACKAGE:Uruntime}
 TAG_ARGS=	-T ${TAGS:[*]:S/ /,/g}
 .endif
 
-.if defined(NO_SHARED) && (${NO_SHARED} != "no" && ${NO_SHARED} != "NO")
+.if defined(NO_SHARED) && ${NO_SHARED:tl} != "no"
 LDFLAGS+= -static
 .endif
 

share/mk/src.libnames.mk

@@ -385,6 +385,10 @@ LDADD_gtest_main= -lprivategtest_main
 LIB${_l:tu}?=	${LIBDESTDIR}${LIBDIR_BASE}/libprivate${_l}.a
 .endfor
 
+.if ${MK_PIE} != "no"
+PIE_SUFFIX=	_pie
+.endif
+
 .for _l in ${_LIBRARIES}
 .if ${_INTERNALLIBS:M${_l}} || !defined(SYSROOT)
 LDADD_${_l}_L+=		-L${LIB${_l:tu}DIR}
@@ -392,12 +396,14 @@ LDADD_${_l}_L+=		-L${LIB${_l:tu}DIR}
 DPADD_${_l}?=	${LIB${_l:tu}}
 .if ${_PRIVATELIBS:M${_l}}
 LDADD_${_l}?=	-lprivate${_l}
+.elif ${_INTERNALLIBS:M${_l}}
+LDADD_${_l}?=	${LDADD_${_l}_L} -l${_l:S/${PIE_SUFFIX}//}${PIE_SUFFIX}
 .else
 LDADD_${_l}?=	${LDADD_${_l}_L} -l${_l}
 .endif
 # Add in all dependencies for static linkage.
 .if defined(_DP_${_l}) && (${_INTERNALLIBS:M${_l}} || \
-    (defined(NO_SHARED) && (${NO_SHARED} != "no" && ${NO_SHARED} != "NO")))
+    (defined(NO_SHARED) && ${NO_SHARED:tl} != "no"))
 .for _d in ${_DP_${_l}}
 DPADD_${_l}+=	${DPADD_${_d}}
 LDADD_${_l}+=	${LDADD_${_d}}
@@ -444,69 +450,69 @@ LDADD+=		${LDADD_${_l}}
 
 # INTERNALLIB definitions.
 LIBELFTCDIR=	${OBJTOP}/lib/libelftc
-LIBELFTC?=	${LIBELFTCDIR}/libelftc.a
+LIBELFTC?=	${LIBELFTCDIR}/libelftc${PIE_SUFFIX}.a
 
 LIBPEDIR=	${OBJTOP}/lib/libpe
-LIBPE?=		${LIBPEDIR}/libpe.a
+LIBPE?=		${LIBPEDIR}/libpe${PIE_SUFFIX}.a
 
 LIBOPENBSDDIR=	${OBJTOP}/lib/libopenbsd
-LIBOPENBSD?=	${LIBOPENBSDDIR}/libopenbsd.a
+LIBOPENBSD?=	${LIBOPENBSDDIR}/libopenbsd${PIE_SUFFIX}.a
 
 LIBSMDIR=	${OBJTOP}/lib/libsm
-LIBSM?=		${LIBSMDIR}/libsm.a
+LIBSM?=		${LIBSMDIR}/libsm${PIE_SUFFIX}.a
 
 LIBSMDBDIR=	${OBJTOP}/lib/libsmdb
-LIBSMDB?=	${LIBSMDBDIR}/libsmdb.a
+LIBSMDB?=	${LIBSMDBDIR}/libsmdb${PIE_SUFFIX}.a
 
 LIBSMUTILDIR=	${OBJTOP}/lib/libsmutil
-LIBSMUTIL?=	${LIBSMUTILDIR}/libsmutil.a
+LIBSMUTIL?=	${LIBSMUTILDIR}/libsmutil${PIE_SUFFIX}.a
 
 LIBNETBSDDIR?=	${OBJTOP}/lib/libnetbsd
-LIBNETBSD?=	${LIBNETBSDDIR}/libnetbsd.a
+LIBNETBSD?=	${LIBNETBSDDIR}/libnetbsd${PIE_SUFFIX}.a
 
 LIBVERSDIR?=	${OBJTOP}/kerberos5/lib/libvers
-LIBVERS?=	${LIBVERSDIR}/libvers.a
+LIBVERS?=	${LIBVERSDIR}/libvers${PIE_SUFFIX}.a
 
 LIBSLDIR=	${OBJTOP}/kerberos5/lib/libsl
-LIBSL?=		${LIBSLDIR}/libsl.a
+LIBSL?=		${LIBSLDIR}/libsl${PIE_SUFFIX}.a
 
 LIBIPFDIR=	${OBJTOP}/sbin/ipf/libipf
-LIBIPF?=	${LIBIPFDIR}/libipf.a
+LIBIPF?=	${LIBIPFDIR}/libipf${PIE_SUFFIX}.a
 
 LIBTELNETDIR=	${OBJTOP}/lib/libtelnet
-LIBTELNET?=	${LIBTELNETDIR}/libtelnet.a
+LIBTELNET?=	${LIBTELNETDIR}/libtelnet${PIE_SUFFIX}.a
 
 LIBCRONDIR=	${OBJTOP}/usr.sbin/cron/lib
-LIBCRON?=	${LIBCRONDIR}/libcron.a
+LIBCRON?=	${LIBCRONDIR}/libcron${PIE_SUFFIX}.a
 
 LIBNTPDIR=	${OBJTOP}/usr.sbin/ntp/libntp
-LIBNTP?=	${LIBNTPDIR}/libntp.a
+LIBNTP?=	${LIBNTPDIR}/libntp${PIE_SUFFIX}.a
 
 LIBNTPEVENTDIR=	${OBJTOP}/usr.sbin/ntp/libntpevent
-LIBNTPEVENT?=	${LIBNTPEVENTDIR}/libntpevent.a
+LIBNTPEVENT?=	${LIBNTPEVENTDIR}/libntpevent${PIE_SUFFIX}.a
 
 LIBOPTSDIR=	${OBJTOP}/usr.sbin/ntp/libopts
-LIBOPTS?=	${LIBOPTSDIR}/libopts.a
+LIBOPTS?=	${LIBOPTSDIR}/libopts${PIE_SUFFIX}.a
 
 LIBPARSEDIR=	${OBJTOP}/usr.sbin/ntp/libparse
-LIBPARSE?=	${LIBPARSEDIR}/libparse.a
+LIBPARSE?=	${LIBPARSEDIR}/libparse${PIE_SUFFIX}.a
 
 LIBLPRDIR=	${OBJTOP}/usr.sbin/lpr/common_source
-LIBLPR?=	${LIBLPRDIR}/liblpr.a
+LIBLPR?=	${LIBLPRDIR}/liblpr${PIE_SUFFIX}.a
 
 LIBFIFOLOGDIR=	${OBJTOP}/usr.sbin/fifolog/lib
-LIBFIFOLOG?=	${LIBFIFOLOGDIR}/libfifolog.a
+LIBFIFOLOG?=	${LIBFIFOLOGDIR}/libfifolog${PIE_SUFFIX}.a
 
 LIBBSNMPTOOLSDIR=	${OBJTOP}/usr.sbin/bsnmpd/tools/libbsnmptools
-LIBBSNMPTOOLS?=	${LIBBSNMPTOOLSDIR}/libbsnmptools.a
+LIBBSNMPTOOLS?=	${LIBBSNMPTOOLSDIR}/libbsnmptools${PIE_SUFFIX}.a
 
 LIBAMUDIR=	${OBJTOP}/usr.sbin/amd/libamu
-LIBAMU?=	${LIBAMUDIR}/libamu.a
+LIBAMU?=	${LIBAMUDIR}/libamu${PIE_SUFFIX}.a
 
-LIBBE?=		${LIBBEDIR}/libbe.a
+LIBBE?=		${LIBBEDIR}/libbe${PIE_SUFFIX}.a
 
 LIBPMCSTATDIR=	${OBJTOP}/lib/libpmcstat
-LIBPMCSTAT?=	${LIBPMCSTATDIR}/libpmcstat.a
+LIBPMCSTAT?=	${LIBPMCSTATDIR}/libpmcstat${PIE_SUFFIX}.a
 
 LIBC_NOSSP_PICDIR=	${OBJTOP}/lib/libc
 LIBC_NOSSP_PIC?=	${LIBC_NOSSP_PICDIR}/libc_nossp_pic.a

stand/common/dev_net.c

@@ -122,13 +122,15 @@ net_open(struct open_file *f, ...)
 {
 	struct iodesc *d;
 	va_list args;
-	char *devname;		/* Device part of file name (or NULL). */
+	struct devdesc *dev;
+	const char *devname;	/* Device part of file name (or NULL). */
 	int error = 0;
 
 	va_start(args, f);
-	devname = va_arg(args, char*);
+	dev = va_arg(args, struct devdesc *);
 	va_end(args);
 
+	devname = dev->d_dev->dv_name;
 	/* Before opening another interface, close the previous one first. */
 	if (netdev_sock >= 0 && strcmp(devname, netdev_name) != 0)
 		net_cleanup();
@@ -137,7 +139,7 @@ net_open(struct open_file *f, ...)
 	if (netdev_opens == 0) {
 		/* Find network interface. */
 		if (netdev_sock < 0) {
-			netdev_sock = netif_open(devname);
+			netdev_sock = netif_open(dev);
 			if (netdev_sock < 0) {
 				printf("net_open: netif_open() failed\n");
 				return (ENXIO);

stand/common/disk.c

@@ -75,7 +75,7 @@ display_size(uint64_t size, u_int sectorsize)
 		size /= 1024;
 		unit = 'M';
 	}
-	sprintf(buf, "%ld%cB", (long)size, unit);
+	sprintf(buf, "%4ld%cB", (long)size, unit);
 	return (buf);
 }
 
@@ -102,7 +102,6 @@ ptblread(void *d, void *buf, size_t blocks, uint64_t offset)
 	    blocks * od->sectorsize, (char *)buf, NULL));
 }
 
-#define	PWIDTH	35
 static int
 ptable_print(void *arg, const char *pname, const struct ptable_entry *part)
 {
@@ -112,16 +111,16 @@ ptable_print(void *arg, const char *pname, const struct ptable_entry *part)
 	struct ptable *table;
 	char line[80];
 	int res;
+	u_int sectsize;
+	uint64_t partsize;
 
 	pa = (struct print_args *)arg;
 	od = (struct open_disk *)pa->dev->dd.d_opendata;
-	sprintf(line, "  %s%s: %s", pa->prefix, pname,
+	sectsize = od->sectorsize;
-	    parttype2str(part->type));
+	partsize = part->end - part->start + 1;
-	if (pa->verbose)
+	sprintf(line, "  %s%s: %s\t%s\n", pa->prefix, pname,
-		sprintf(line, "%-*s%s", PWIDTH, line,
+	    parttype2str(part->type),
-		    display_size(part->end - part->start + 1,
+	    pa->verbose ? display_size(partsize, sectsize) : "");
-		    od->sectorsize));
-	strcat(line, "\n");
 	if (pager_output(line))
 		return 1;
 	res = 0;
@@ -131,10 +130,15 @@ ptable_print(void *arg, const char *pname, const struct ptable_entry *part)
 		dev.dd.d_unit = pa->dev->dd.d_unit;
 		dev.d_slice = part->index;
 		dev.d_partition = -1;
-		if (disk_open(&dev, part->end - part->start + 1,
+		if (disk_open(&dev, partsize, sectsize) == 0) {
-		    od->sectorsize) == 0) {
+			/*
-			table = ptable_open(&dev, part->end - part->start + 1,
+			 * disk_open() for partition -1 on a bsd slice assumes
-			    od->sectorsize, ptblread);
+			 * you want the first bsd partition.  Reset things so
+			 * that we're looking at the start of the raw slice.
+			 */
+			dev.d_partition = -1;
+			dev.d_offset = part->start;
+			table = ptable_open(&dev, partsize, sectsize, ptblread);
 			if (table != NULL) {
 				sprintf(line, "  %s%s", pa->prefix, pname);
 				bsd.dev = pa->dev;
@@ -149,7 +153,6 @@ ptable_print(void *arg, const char *pname, const struct ptable_entry *part)
 
 	return (res);
 }
-#undef PWIDTH
 
 int
 disk_print(struct disk_devdesc *dev, char *prefix, int verbose)

stand/common/part.c

@@ -788,6 +788,9 @@ ptable_close(struct ptable *table)
 {
 	struct pentry *entry;
 
+	if (table == NULL)
+		return;
+
 	while (!STAILQ_EMPTY(&table->entries)) {
 		entry = STAILQ_FIRST(&table->entries);
 		STAILQ_REMOVE_HEAD(&table->entries, entry);

stand/i386/Makefile.inc

@@ -7,6 +7,7 @@
 LOADER_ADDRESS?=0x200000
 LDFLAGS+=	-nostdlib
 LDFLAGS.lld+=	-Wl,--no-rosegment
+MK_PIE:=	no
 
 # BTX components
 BTXDIR=		${BOOTOBJ}/i386/btx

stand/i386/zfsboot/zfsboot.c

@@ -545,32 +545,19 @@ probe_drive(struct zfsdsk *zdsk)
     char *sec;
     unsigned i;
 
-    /*
-     * If we find a vdev on the whole disk, stop here.
-     */
-    if (vdev_probe(vdev_read2, zdsk, NULL) == 0)
-	return;
-
 #ifdef LOADER_GELI_SUPPORT
     /*
-     * Taste the disk, if it is GELI encrypted, decrypt it and check to see if
+     * Taste the disk, if it is GELI encrypted, decrypt it then dig out the
-     * it is a usable vdev then. Otherwise dig
+     * partition table and probe each slice/partition in turn for a vdev or
-     * out the partition table and probe each slice/partition
+     * GELI encrypted vdev.
-     * in turn for a vdev or GELI encrypted vdev.
      */
     elba = drvsize_ext(zdsk);
     if (elba > 0) {
 	elba--;
     }
     zdsk->gdev = geli_taste(vdev_read, zdsk, elba, "disk%u:0:");
-    if (zdsk->gdev != NULL) {
+    if ((zdsk->gdev != NULL) && (geli_havekey(zdsk->gdev) == 0))
-	if (geli_havekey(zdsk->gdev) == 0 || 
+	    geli_passphrase(zdsk->gdev, gelipw);
-	    geli_passphrase(zdsk->gdev, gelipw) == 0) {
-	    if (vdev_probe(vdev_read2, zdsk, NULL) == 0) {
-		return;
-	    }
-	}
-    }
 #endif /* LOADER_GELI_SUPPORT */
 
     sec = dmadat->secbuf;

stand/libsa/cd9660.c

@@ -241,6 +241,10 @@ dirmatch(struct open_file *f, const char *path, struct iso_directory_record *dp,
 		icase = 1;
 	} else
 		icase = 0;
+
+	if (strlen(path) != len)
+		return (0);
+
 	for (i = len; --i >= 0; path++, cp++) {
 		if (!*path || *path == '/')
 			break;

stand/libsa/zfs/zfs.c

@@ -363,51 +363,100 @@ static int
 vdev_read(vdev_t *vdev, void *priv, off_t offset, void *buf, size_t bytes)
 {
 	int fd, ret;
-	size_t res, size, remainder, rb_size, blksz;
+	size_t res, head, tail, total_size, full_sec_size;
-	unsigned secsz;
+	unsigned secsz, do_tail_read;
-	off_t off;
+	off_t start_sec;
-	char *bouncebuf, *rb_buf;
+	char *outbuf, *bouncebuf;
 
 	fd = (uintptr_t) priv;
+	outbuf = (char *) buf;
 	bouncebuf = NULL;
 
 	ret = ioctl(fd, DIOCGSECTORSIZE, &secsz);
 	if (ret != 0)
 		return (ret);
 
-	off = offset / secsz;
+	/*
-	remainder = offset % secsz;
+	 * Handling reads of arbitrary offset and size - multi-sector case
-	if (lseek(fd, off * secsz, SEEK_SET) == -1)
+	 * and single-sector case.
-		return (errno);
+	 *
+	 *                        Multi-sector Case
+	 *                (do_tail_read = true if tail > 0)
+	 *
+	 *   |<----------------------total_size--------------------->|
+	 *   |                                                       |
+	 *   |<--head-->|<--------------bytes------------>|<--tail-->|
+	 *   |          |                                 |          |
+	 *   |          |       |<~full_sec_size~>|       |          |
+	 *   +------------------+                 +------------------+
+	 *   |          |0101010|     .  .  .     |0101011|          |
+	 *   +------------------+                 +------------------+
+	 *         start_sec                         start_sec + n
+	 *
+	 *
+	 *                      Single-sector Case
+	 *                    (do_tail_read = false)
+	 *
+	 *              |<------total_size = secsz----->|
+	 *              |                               |
+	 *              |<-head->|<---bytes--->|<-tail->|
+	 *              +-------------------------------+
+	 *              |        |0101010101010|        |
+	 *              +-------------------------------+
+	 *                          start_sec
+	 */
+	start_sec = offset / secsz;
+	head = offset % secsz;
+	total_size = roundup2(head + bytes, secsz);
+	tail = total_size - (head + bytes);
+	do_tail_read = ((tail > 0) && (head + bytes > secsz));
+	full_sec_size = total_size;
+	if (head > 0)
+		full_sec_size -= secsz;
+	if (do_tail_read)
+		full_sec_size -= secsz;
 
-	rb_buf = buf;
+	/* Return of partial sector data requires a bounce buffer. */
-	rb_size = bytes;
+	if ((head > 0) || do_tail_read) {
-	size = roundup2(bytes + remainder, secsz);
-	blksz = size;
-	if (remainder != 0 || size != bytes) {
 		bouncebuf = zfs_alloc(secsz);
 		if (bouncebuf == NULL) {
 			printf("vdev_read: out of memory\n");
 			return (ENOMEM);
 		}
-		rb_buf = bouncebuf;
-		blksz = rb_size - remainder;
 	}
 
-	while (bytes > 0) {
+	if (lseek(fd, start_sec * secsz, SEEK_SET) == -1)
-		res = read(fd, rb_buf, rb_size);
+		return (errno);
-		if (res != rb_size) {
+
+	/* Partial data return from first sector */
+	if (head > 0) {
+		res = read(fd, bouncebuf, secsz);
+		if (res != secsz) {
 			ret = EIO;
 			goto error;
 		}
-		if (bytes < blksz)
+		memcpy(outbuf, bouncebuf + head, min(secsz - head, bytes));
-			blksz = bytes;
+		outbuf += min(secsz - head, bytes);
-		if (bouncebuf != NULL)
+	}
-			memcpy(buf, rb_buf + remainder, blksz);
+
-		buf = (void *)((uintptr_t)buf + blksz);
+	/* Full data return from read sectors */
-		bytes -= blksz;
+	if (full_sec_size > 0) {
-		remainder = 0;
+		res = read(fd, outbuf, full_sec_size);
-		blksz = rb_size;
+		if (res != full_sec_size) {
+			ret = EIO;
+			goto error;
+		}
+		outbuf += full_sec_size;
+	}
+
+	/* Partial data return from last sector */
+	if (do_tail_read) {
+		res = read(fd, bouncebuf, secsz);
+		if (res != secsz) {
+			ret = EIO;
+			goto error;
+		}
+		memcpy(outbuf, bouncebuf, secsz - tail);
 	}
 
 	ret = 0;

stand/lua/password.lua

@@ -38,6 +38,7 @@ local INCORRECT_PASSWORD = "loader: incorrect password"
 -- Asterisks as a password mask
 local show_password_mask = false
 local twiddle_chars = {"/", "-", "\\", "|"}
+local screen_setup = false
 
 -- Module exports
 function password.read(prompt_length)
@@ -80,8 +81,6 @@ function password.read(prompt_length)
 end
 
 function password.check()
-	screen.clear()
-	screen.defcursor()
 	-- pwd is optionally supplied if we want to check it
 	local function doPrompt(prompt, pwd)
 		local attempts = 1
@@ -90,6 +89,12 @@ function password.check()
 			printc("\r" .. string.rep(" ", #INCORRECT_PASSWORD))
 		end
 
+		if not screen_setup then
+			screen.clear()
+			screen.defcursor()
+			screen_setup = true
+		end
+
 		while true do
 			if attempts > 1 then
 				clear_incorrect_text_prompt()

stand/powerpc/uboot/Makefile

@@ -11,6 +11,7 @@ LOADER_BZIP2_SUPPORT?=	no
 
 .include <bsd.init.mk>
 
+BINDIR= 	/boot/uboot
 PROG=		ubldr
 NEWVERSWHAT=	"U-Boot loader" ${MACHINE_ARCH}
 INSTALLFLAGS=	-b

stand/uboot/common/main.c

@@ -156,7 +156,7 @@ get_device_type(const char *devstr, int *devtype)
 		printf("Unknown device type '%s'\n", devstr);
 	}
 
-	*devtype = -1;
+	*devtype = DEV_TYP_NONE;
 	return (NULL);
 }
 
@@ -182,6 +182,14 @@ device_typename(int type)
  * The returned values for slice and partition are interpreted by
  * disk_open().
  *
+ * The device string can be a standard loader(8) disk specifier:
+ *
+ * disk<unit>s<slice>              disk0s1
+ * disk<unit>s<slice><partition>   disk1s2a
+ * disk<unit>p<partition>          disk0p4
+ *
+ * or one of the following formats:
+ *
  * Valid device strings:                     For device types:
  *
  * <type_name>                               DEV_TYP_STOR, DEV_TYP_NET
@@ -198,11 +206,12 @@ device_typename(int type)
 static void
 get_load_device(int *type, int *unit, int *slice, int *partition)
 {
+	struct disk_devdesc dev;
 	char *devstr;
 	const char *p;
 	char *endp;
 
-	*type = -1;
+	*type = DEV_TYP_NONE;
 	*unit = -1;
 	*slice = 0;
 	*partition = -1;
@@ -216,18 +225,31 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 
 	p = get_device_type(devstr, type);
 
+	/*
+	 * If type is DEV_TYP_STOR we have a disk-like device.  If we can parse
+	 * the remainder of the string as a standard unit+slice+partition (e.g.,
+	 * 0s2a or 1p12), return those results.  Otherwise we'll fall through to
+	 * the code that parses the legacy format.
+	 */
+	if ((*type & DEV_TYP_STOR) && disk_parsedev(&dev, p, NULL) == 0) {
+		*unit = dev.dd.d_unit;
+		*slice = dev.d_slice;
+		*partition = dev.d_partition;
+		return;
+	}
+
 	/* Ignore optional spaces after the device name. */
 	while (*p == ' ')
 		p++;
 
 	/* Unknown device name, or a known name without unit number.  */
-	if ((*type == -1) || (*p == '\0')) {
+	if ((*type == DEV_TYP_NONE) || (*p == '\0')) {
 		return;
 	}
 
 	/* Malformed unit number. */
 	if (!isdigit(*p)) {
-		*type = -1;
+		*type = DEV_TYP_NONE;
 		return;
 	}
 
@@ -242,7 +264,7 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 
 	/* Device string is malformed beyond unit number. */
 	if (*p != ':') {
-		*type = -1;
+		*type = DEV_TYP_NONE;
 		*unit = -1;
 		return;
 	}
@@ -255,7 +277,7 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 
 	/* Only DEV_TYP_STOR devices can have a slice specification. */
 	if (!(*type & DEV_TYP_STOR)) {
-		*type = -1;
+		*type = DEV_TYP_NONE;
 		*unit = -1;
 		return;
 	}
@@ -264,7 +286,7 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 
 	/* Malformed slice number. */
 	if (p == endp) {
-		*type = -1;
+		*type = DEV_TYP_NONE;
 		*unit = -1;
 		*slice = 0;
 		return;
@@ -278,7 +300,7 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 
 	/* Device string is malformed beyond slice number. */
 	if (*p != '.') {
-		*type = -1;
+		*type = DEV_TYP_NONE;
 		*unit = -1;
 		*slice = 0;
 		return;
@@ -298,7 +320,7 @@ get_load_device(int *type, int *unit, int *slice, int *partition)
 		return;
 
 	/* Junk beyond partition number. */
-	*type = -1;
+	*type = DEV_TYP_NONE;
 	*unit = -1;
 	*slice = 0;
 	*partition = -1;
@@ -310,13 +332,13 @@ print_disk_probe_info()
 	char slice[32];
 	char partition[32];
 
-	if (currdev.d_disk.slice > 0)
+	if (currdev.d_disk.d_slice > 0)
-		sprintf(slice, "%d", currdev.d_disk.slice);
+		sprintf(slice, "%d", currdev.d_disk.d_slice);
 	else
 		strcpy(slice, "<auto>");
 
-	if (currdev.d_disk.partition >= 0)
+	if (currdev.d_disk.d_partition >= 0)
-		sprintf(partition, "%d", currdev.d_disk.partition);
+		sprintf(partition, "%d", currdev.d_disk.d_partition);
 	else
 		strcpy(partition, "<auto>");
 
@@ -332,8 +354,8 @@ probe_disks(int devidx, int load_type, int load_unit, int load_slice,
 	int open_result, unit;
 	struct open_file f;
 
-	currdev.d_disk.slice = load_slice;
+	currdev.d_disk.d_slice = load_slice;
-	currdev.d_disk.partition = load_partition;
+	currdev.d_disk.d_partition = load_partition;
 
 	f.f_devdata = &currdev;
 	open_result = -1;
@@ -467,14 +489,14 @@ main(int argc, char **argv)
 		currdev.dd.d_dev = devsw[i];
 		currdev.dd.d_unit = 0;
 
-		if ((load_type == -1 || (load_type & DEV_TYP_STOR)) &&
+		if ((load_type == DEV_TYP_NONE || (load_type & DEV_TYP_STOR)) &&
 		    strcmp(devsw[i]->dv_name, "disk") == 0) {
 			if (probe_disks(i, load_type, load_unit, load_slice, 
 			    load_partition) == 0)
 				break;
 		}
 
-		if ((load_type == -1 || (load_type & DEV_TYP_NET)) &&
+		if ((load_type == DEV_TYP_NONE || (load_type & DEV_TYP_NET)) &&
 		    strcmp(devsw[i]->dv_name, "net") == 0)
 			break;
 	}

stand/uboot/lib/libuboot.h

@@ -27,18 +27,14 @@
  * $FreeBSD$
  */
 
-struct uboot_devdesc {
+#include <disk.h>
-	struct devdesc		dd;	/* Must be first. */
-	union {
-		struct {
-			int	slice;
-			int	partition;
-			off_t	offset;
-		} disk;
-	} d_kind;
-};
 
-#define d_disk d_kind.disk
+struct uboot_devdesc {
+	union {
+		struct devdesc      dd;
+		struct disk_devdesc d_disk;
+	};
+};
 
 /*
  * Default network packet alignment in memory.  On arm arches packets must be

sys/amd64/amd64/pmap.c

@@ -353,10 +353,6 @@ pt_entry_t pg_nx;
 
 static SYSCTL_NODE(_vm, OID_AUTO, pmap, CTLFLAG_RD, 0, "VM/pmap parameters");
 
-static int pat_works = 1;
-SYSCTL_INT(_vm_pmap, OID_AUTO, pat_works, CTLFLAG_RD, &pat_works, 1,
-    "Is page attribute table fully functional?");
-
 static int pg_ps_enabled = 1;
 SYSCTL_INT(_vm_pmap, OID_AUTO, pg_ps_enabled, CTLFLAG_RDTUN | CTLFLAG_NOFETCH,
     &pg_ps_enabled, 0, "Are large page mappings enabled?");
@@ -1222,7 +1218,6 @@ pmap_bootstrap(vm_paddr_t *firstaddr)
 void
 pmap_init_pat(void)
 {
-	int pat_table[PAT_INDEX_SIZE];
 	uint64_t pat_msr;
 	u_long cr0, cr4;
 	int i;
@@ -1233,45 +1228,32 @@ pmap_init_pat(void)
 
 	/* Set default PAT index table. */
 	for (i = 0; i < PAT_INDEX_SIZE; i++)
-		pat_table[i] = -1;
+		pat_index[i] = -1;
-	pat_table[PAT_WRITE_BACK] = 0;
+	pat_index[PAT_WRITE_BACK] = 0;
-	pat_table[PAT_WRITE_THROUGH] = 1;
+	pat_index[PAT_WRITE_THROUGH] = 1;
-	pat_table[PAT_UNCACHEABLE] = 3;
+	pat_index[PAT_UNCACHEABLE] = 3;
-	pat_table[PAT_WRITE_COMBINING] = 3;
+	pat_index[PAT_WRITE_COMBINING] = 6;
-	pat_table[PAT_WRITE_PROTECTED] = 3;
+	pat_index[PAT_WRITE_PROTECTED] = 5;
-	pat_table[PAT_UNCACHED] = 3;
+	pat_index[PAT_UNCACHED] = 2;
 
-	/* Initialize default PAT entries. */
+	/*
+	 * Initialize default PAT entries.
+	 * Leave the indices 0-3 at the default of WB, WT, UC-, and UC.
+	 * Program 5 and 6 as WP and WC.
+	 *
+	 * Leave 4 and 7 as WB and UC.  Note that a recursive page table
+	 * mapping for a 2M page uses a PAT value with the bit 3 set due
+	 * to its overload with PG_PS.
+	 */
 	pat_msr = PAT_VALUE(0, PAT_WRITE_BACK) |
 	    PAT_VALUE(1, PAT_WRITE_THROUGH) |
 	    PAT_VALUE(2, PAT_UNCACHED) |
 	    PAT_VALUE(3, PAT_UNCACHEABLE) |
 	    PAT_VALUE(4, PAT_WRITE_BACK) |
-	    PAT_VALUE(5, PAT_WRITE_THROUGH) |
+	    PAT_VALUE(5, PAT_WRITE_PROTECTED) |
-	    PAT_VALUE(6, PAT_UNCACHED) |
+	    PAT_VALUE(6, PAT_WRITE_COMBINING) |
 	    PAT_VALUE(7, PAT_UNCACHEABLE);
 
-	if (pat_works) {
-		/*
-		 * Leave the indices 0-3 at the default of WB, WT, UC-, and UC.
-		 * Program 5 and 6 as WP and WC.
-		 * Leave 4 and 7 as WB and UC.
-		 */
-		pat_msr &= ~(PAT_MASK(5) | PAT_MASK(6));
-		pat_msr |= PAT_VALUE(5, PAT_WRITE_PROTECTED) |
-		    PAT_VALUE(6, PAT_WRITE_COMBINING);
-		pat_table[PAT_UNCACHED] = 2;
-		pat_table[PAT_WRITE_PROTECTED] = 5;
-		pat_table[PAT_WRITE_COMBINING] = 6;
-	} else {
-		/*
-		 * Just replace PAT Index 2 with WC instead of UC-.
-		 */
-		pat_msr &= ~PAT_MASK(2);
-		pat_msr |= PAT_VALUE(2, PAT_WRITE_COMBINING);
-		pat_table[PAT_WRITE_COMBINING] = 2;
-	}
-
 	/* Disable PGE. */
 	cr4 = rcr4();
 	load_cr4(cr4 & ~CR4_PGE);
@@ -1286,8 +1268,6 @@ pmap_init_pat(void)
 
 	/* Update PAT and index table. */
 	wrmsr(MSR_PAT, pat_msr);
-	for (i = 0; i < PAT_INDEX_SIZE; i++)
-		pat_index[i] = pat_table[i];
 
 	/* Flush caches and TLBs again. */
 	wbinvd();

sys/amd64/sgx/sgx_linux.c

@@ -70,30 +70,26 @@ sgx_linux_ioctl(struct thread *td, struct linux_ioctl_args *args)
 	cmd = args->cmd;
 
 	args->cmd &= ~(LINUX_IOC_IN | LINUX_IOC_OUT);
-	if (cmd & LINUX_IOC_IN)
+	if ((cmd & LINUX_IOC_IN) != 0)
 		args->cmd |= IOC_IN;
-	if (cmd & LINUX_IOC_OUT)
+	if ((cmd & LINUX_IOC_OUT) != 0)
 		args->cmd |= IOC_OUT;
 
 	len = IOCPARM_LEN(cmd);
 	if (len > SGX_IOCTL_MAX_DATA_LEN) {
-		printf("%s: Can't copy data: cmd len is too big %d\n",
+		error = EINVAL;
-		    __func__, len);
+		goto out;
-		return (EINVAL);
 	}
 
-	if (cmd & LINUX_IOC_IN) {
+	if ((cmd & LINUX_IOC_IN) != 0) {
 		error = copyin((void *)args->arg, data, len);
-		if (error) {
+		if (error != 0)
-			printf("%s: Can't copy data, error %d\n",
+			goto out;
-			    __func__, error);
-			return (EINVAL);
-		}
 	}
 
-	error = (fo_ioctl(fp, args->cmd, (caddr_t)data, td->td_ucred, td));
+	error = fo_ioctl(fp, args->cmd, (caddr_t)data, td->td_ucred, td);
+out:
 	fdrop(fp, td);
-
 	return (error);
 }
 

sys/arm/allwinner/axp81x.c

@@ -120,6 +120,10 @@ MALLOC_DEFINE(M_AXP8XX_REG, "AXP8xx regulator", "AXP8xx power regulator");
 #define	 AXP_VOLTCTL_MASK	0x7f
 #define	AXP_POWERBAT		0x32
 #define	 AXP_POWERBAT_SHUTDOWN	(1 << 7)
+#define	AXP_CHARGERCTL1		0x33
+#define	 AXP_CHARGERCTL1_MIN	0
+#define	 AXP_CHARGERCTL1_MAX	13
+#define	 AXP_CHARGERCTL1_CMASK	0xf
 #define	AXP_IRQEN1		0x40
 #define	 AXP_IRQEN1_ACIN_HI	(1 << 6)
 #define	 AXP_IRQEN1_ACIN_LO	(1 << 5)
@@ -614,13 +618,13 @@ static const struct axp8xx_sensors axp8xx_common_sensors[] = {
 		.id = AXP_SENSOR_BATT_CHARGE_CURRENT,
 		.name = "batchargecurrent",
 		.format = "I",
-		.desc = "Battery Charging Current",
+		.desc = "Average Battery Charging Current",
 	},
 	{
 		.id = AXP_SENSOR_BATT_DISCHARGE_CURRENT,
 		.name = "batdischargecurrent",
 		.format = "I",
-		.desc = "Battery Discharging Current",
+		.desc = "Average Battery Discharging Current",
 	},
 	{
 		.id = AXP_SENSOR_BATT_CAPACITY_PERCENT,
@@ -889,6 +893,33 @@ axp8xx_shutdown(void *devp, int howto)
 	axp8xx_write(dev, AXP_POWERBAT, AXP_POWERBAT_SHUTDOWN);
 }
 
+static int
+axp8xx_sysctl_chargecurrent(SYSCTL_HANDLER_ARGS)
+{
+	device_t dev = arg1;
+	uint8_t data;
+	int val, error;
+
+	error = axp8xx_read(dev, AXP_CHARGERCTL1, &data, 1);
+	if (error != 0)
+		return (error);
+
+	if (bootverbose)
+		device_printf(dev, "Raw CHARGECTL1 val: 0x%0x\n", data);
+	val = (data & AXP_CHARGERCTL1_CMASK);
+	error = sysctl_handle_int(oidp, &val, 0, req);
+	if (error || !req->newptr) /* error || read request */
+		return (error);
+
+	if ((val < AXP_CHARGERCTL1_MIN) || (val > AXP_CHARGERCTL1_MAX))
+		return (EINVAL);
+
+	val |= (data & (AXP_CHARGERCTL1_CMASK << 4));
+	axp8xx_write(dev, AXP_CHARGERCTL1, val);
+
+	return (0);
+}
+
 static int
 axp8xx_sysctl(SYSCTL_HANDLER_ARGS)
 {
@@ -1482,6 +1513,16 @@ axp8xx_attach(device_t dev)
 		    sc->sensors[i].format,
 		    sc->sensors[i].desc);
 	}
+	SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev),
+	    SYSCTL_CHILDREN(device_get_sysctl_tree(dev)),
+	    OID_AUTO, "batchargecurrentstep",
+	    CTLTYPE_INT | CTLFLAG_RW,
+	    dev, 0, axp8xx_sysctl_chargecurrent,
+	    "I", "Battery Charging Current Step, "
+	    "0: 200mA, 1: 400mA, 2: 600mA, 3: 800mA, "
+	    "4: 1000mA, 5: 1200mA, 6: 1400mA, 7: 1600mA, "
+	    "8: 1800mA, 9: 2000mA, 10: 2200mA, 11: 2400mA, "
+	    "12: 2600mA, 13: 2800mA");
 
 	/* Get thresholds */
 	if (axp8xx_read(dev, AXP_BAT_CAP_WARN, &val, 1) == 0) {

sys/arm/arm/elf_machdep.c

@@ -84,7 +84,7 @@ struct sysentvec elf32_freebsd_sysvec = {
 #if __ARM_ARCH >= 6
 			  SV_ASLR | SV_SHP | SV_TIMEKEEP |
 #endif
-			  SV_ABI_FREEBSD | SV_ILP32,
+			  SV_ABI_FREEBSD | SV_ILP32 | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,

sys/arm/freescale/imx/imx6_snvs.c

@@ -73,6 +73,7 @@ struct snvs_softc {
 };
 
 static struct ofw_compat_data compat_data[] = {
+	{"fsl,sec-v4.0-mon-rtc-lp", true},
 	{"fsl,sec-v4.0-mon", true},
 	{NULL,               false}
 };

sys/arm64/arm64/elf_machdep.c

@@ -79,7 +79,8 @@ static struct sysentvec elf64_freebsd_sysvec = {
 	.sv_setregs	= exec_setregs,
 	.sv_fixlimit	= NULL,
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_SHP | SV_TIMEKEEP | SV_ABI_FREEBSD | SV_LP64,
+	.sv_flags	= SV_SHP | SV_TIMEKEEP | SV_ABI_FREEBSD | SV_LP64 |
+	    SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,

sys/conf/NOTES

@@ -1759,14 +1759,6 @@ hint.ata.1.at="isa"
 hint.ata.1.port="0x170"
 hint.ata.1.irq="15"
 
-#
-# The following options are valid on the ATA driver:
-#
-# ATA_REQUEST_TIMEOUT:	the number of seconds to wait for an ATA request
-#			before timing out.
-
-#options 	ATA_REQUEST_TIMEOUT=10
-
 #
 # Standard floppy disk controllers and floppy tapes, supports
 # the Y-E DATA External FDD (PC Card)

sys/conf/files

@@ -4847,6 +4847,8 @@ crypto/libsodium/randombytes.c	optional crypto \
 	compile-with "${NORMAL_C} -I$S/contrib/libsodium/src/libsodium/include -I$S/crypto/libsodium"
 crypto/libsodium/utils.c	optional crypto \
 	compile-with "${NORMAL_C} -I$S/contrib/libsodium/src/libsodium/include -I$S/crypto/libsodium"
+opencrypto/cbc_mac.c		optional crypto
+opencrypto/xform_cbc_mac.c	optional crypto
 rpc/auth_none.c			optional krpc | nfslockd | nfscl | nfsd
 rpc/auth_unix.c			optional krpc | nfslockd | nfscl | nfsd
 rpc/authunix_prot.c		optional krpc | nfslockd | nfscl | nfsd

sys/conf/ldscript.riscv

@@ -87,6 +87,10 @@ SECTIONS
   .sdata     : { *(.sdata) }
   _edata  =  .;
   PROVIDE (edata = .);
+  /* Ensure __bss_start is associated with the next section in case orphan
+     sections are placed directly after .sdata, as has been seen to happen with
+     LLD.  */
+  . = .;
   __bss_start = .;
   .sbss      : { *(.sbss) *(.scommon) }
   .bss       :

sys/contrib/libnv/nvpair.c

@@ -229,16 +229,8 @@ nvpair_remove_nvlist_array(nvpair_t *nvp)
 	nvlarray = __DECONST(nvlist_t **,
 	    nvpair_get_nvlist_array(nvp, &count));
 	for (i = 0; i < count; i++) {
-		nvlist_t *nvl;
+		nvlist_set_array_next(nvlarray[i], NULL);
-		nvpair_t *nnvp;
+		nvlist_set_parent(nvlarray[i], NULL);
-
-		nvl = nvlarray[i];
-		nnvp = nvlist_get_array_next_nvpair(nvl);
-		if (nnvp != NULL) {
-			nvpair_free_structure(nnvp);
-		}
-		nvlist_set_array_next(nvl, NULL);
-		nvlist_set_parent(nvl, NULL);
 	}
 }
 

sys/dev/ata/ata-all.h

@@ -203,10 +203,6 @@
 #define ATA_OP_FINISHED                 1
 #define ATA_MAX_28BIT_LBA               268435455UL
 
-#ifndef	ATA_REQUEST_TIMEOUT
-#define	ATA_REQUEST_TIMEOUT		10
-#endif
-
 /* structure used for composite atomic operations */
 #define MAX_COMPOSITES          32              /* u_int32_t bits */
 struct ata_composite {

sys/dev/ena/ena.c

@@ -1046,10 +1046,6 @@ ena_refill_rx_bufs(struct ena_ring *rx_ring, uint32_t num)
 		    "RX buffer - next to use: %d", next_to_use);
 
 		req_id = rx_ring->free_rx_ids[next_to_use];
-		rc = validate_rx_req_id(rx_ring, req_id);
-		if (unlikely(rc != 0))
-			break;
-
 		rx_info = &rx_ring->rx_buffer_info[req_id];
 
 		rc = ena_alloc_rx_mbuf(adapter, rx_ring, rx_info);
@@ -1472,21 +1468,24 @@ ena_rx_mbuf(struct ena_ring *rx_ring, struct ena_com_rx_buf_info *ena_bufs,
 	struct ena_rx_buffer *rx_info;
 	struct ena_adapter *adapter;
 	unsigned int descs = ena_rx_ctx->descs;
+	int rc;
 	uint16_t ntc, len, req_id, buf = 0;
 
 	ntc = *next_to_clean;
 	adapter = rx_ring->adapter;
-	rx_info = &rx_ring->rx_buffer_info[ntc];
 
+	len = ena_bufs[buf].len;
+	req_id = ena_bufs[buf].req_id;
+	rc = validate_rx_req_id(rx_ring, req_id);
+	if (unlikely(rc != 0))
+		return (NULL);
+
+	rx_info = &rx_ring->rx_buffer_info[req_id];
 	if (unlikely(rx_info->mbuf == NULL)) {
 		device_printf(adapter->pdev, "NULL mbuf in rx_info");
 		return (NULL);
 	}
 
-	len = ena_bufs[buf].len;
-	req_id = ena_bufs[buf].req_id;
-	rx_info = &rx_ring->rx_buffer_info[req_id];
-
 	ena_trace(ENA_DBG | ENA_RXPTH, "rx_info %p, mbuf %p, paddr %jx",
 	    rx_info, rx_info->mbuf, (uintmax_t)rx_info->ena_buf.paddr);
 
@@ -1517,6 +1516,16 @@ ena_rx_mbuf(struct ena_ring *rx_ring, struct ena_com_rx_buf_info *ena_bufs,
 		++buf;
 		len = ena_bufs[buf].len;
 		req_id = ena_bufs[buf].req_id;
+		rc = validate_rx_req_id(rx_ring, req_id);
+		if (unlikely(rc != 0)) {
+			/*
+			 * If the req_id is invalid, then the device will be
+			 * reset. In that case we must free all mbufs that
+			 * were already gathered.
+			 */
+			m_freem(mbuf);
+			return (NULL);
+		}
 		rx_info = &rx_ring->rx_buffer_info[req_id];
 
 		if (unlikely(rx_info->mbuf == NULL)) {

sys/dev/ena/ena.h

@@ -41,7 +41,7 @@
 
 #define DRV_MODULE_VER_MAJOR	0
 #define DRV_MODULE_VER_MINOR	8
-#define DRV_MODULE_VER_SUBMINOR 2
+#define DRV_MODULE_VER_SUBMINOR 3
 
 #define DRV_MODULE_NAME		"ena"
 

sys/dev/ixl/if_ixl.c

@@ -932,7 +932,7 @@ ixl_if_msix_intr_assign(if_ctx_t ctx, int msix)
 		return (err);
 	}
 	/* Create soft IRQ for handling VFLRs */
-	iflib_softirq_alloc_generic(ctx, &pf->iov_irq, IFLIB_INTR_IOV, pf, 0, "iov");
+	iflib_softirq_alloc_generic(ctx, NULL, IFLIB_INTR_IOV, pf, 0, "iov");
 
 	/* Now set up the stations */
 	for (i = 0, vector = 1; i < vsi->shared->isc_nrxqsets; i++, vector++, rx_que++) {

sys/dev/ixl/ixl_pf.h

@@ -138,7 +138,6 @@ struct ixl_pf {
 	struct ixl_vf		*vfs;
 	int			num_vfs;
 	uint16_t		veb_seid;
-	struct if_irq		iov_irq;
 };
 
 /*

sys/dev/netmap/netmap_freebsd.c

@@ -105,6 +105,7 @@ int nm_os_selinfo_init(NM_SELINFO_T *si, const char *name) {
 	snprintf(si->mtxname, sizeof(si->mtxname), "nmkl%s", name);
 	mtx_init(&si->m, si->mtxname, NULL, MTX_DEF);
 	knlist_init_mtx(&si->si.si_note, &si->m);
+	si->kqueue_users = 0;
 
 	return (0);
 }
@@ -1351,7 +1352,9 @@ void
 nm_os_selwakeup(struct nm_selinfo *si)
 {
 	selwakeuppri(&si->si, PI_NET);
-	taskqueue_enqueue(si->ntfytq, &si->ntfytask);
+	if (si->kqueue_users > 0) {
+		taskqueue_enqueue(si->ntfytq, &si->ntfytask);
+	}
 }
 
 void
@@ -1364,20 +1367,28 @@ static void
 netmap_knrdetach(struct knote *kn)
 {
 	struct netmap_priv_d *priv = (struct netmap_priv_d *)kn->kn_hook;
-	struct selinfo *si = &priv->np_si[NR_RX]->si;
+	struct nm_selinfo *si = priv->np_si[NR_RX];
 
-	nm_prinf("remove selinfo %p", si);
+	knlist_remove(&si->si.si_note, kn, /*islocked=*/0);
-	knlist_remove(&si->si_note, kn, /*islocked=*/0);
+	NMG_LOCK();
+	KASSERT(si->kqueue_users > 0, ("kqueue_user underflow on %s",
+	    si->mtxname));
+	si->kqueue_users--;
+	nm_prinf("kqueue users for %s: %d", si->mtxname, si->kqueue_users);
+	NMG_UNLOCK();
 }
 
 static void
 netmap_knwdetach(struct knote *kn)
 {
 	struct netmap_priv_d *priv = (struct netmap_priv_d *)kn->kn_hook;
-	struct selinfo *si = &priv->np_si[NR_TX]->si;
+	struct nm_selinfo *si = priv->np_si[NR_TX];
 
-	nm_prinf("remove selinfo %p", si);
+	knlist_remove(&si->si.si_note, kn, /*islocked=*/0);
-	knlist_remove(&si->si_note, kn, /*islocked=*/0);
+	NMG_LOCK();
+	si->kqueue_users--;
+	nm_prinf("kqueue users for %s: %d", si->mtxname, si->kqueue_users);
+	NMG_UNLOCK();
 }
 
 /*
@@ -1465,6 +1476,10 @@ netmap_kqfilter(struct cdev *dev, struct knote *kn)
 	kn->kn_fop = (ev == EVFILT_WRITE) ?
 		&netmap_wfiltops : &netmap_rfiltops;
 	kn->kn_hook = priv;
+	NMG_LOCK();
+	si->kqueue_users++;
+	nm_prinf("kqueue users for %s: %d", si->mtxname, si->kqueue_users);
+	NMG_UNLOCK();
 	knlist_add(&si->si.si_note, kn, /*islocked=*/0);
 
 	return 0;

sys/dev/netmap/netmap_kern.h

@@ -132,11 +132,14 @@ struct netmap_adapter *netmap_getna(if_t ifp);
 #define MBUF_QUEUED(m)		1
 
 struct nm_selinfo {
+	/* Support for select(2) and poll(2). */
 	struct selinfo si;
+	/* Support for kqueue(9). See comments in netmap_freebsd.c */
 	struct taskqueue *ntfytq;
 	struct task ntfytask;
 	struct mtx m;
 	char mtxname[32];
+	int kqueue_users;
 };
 
 

sys/fs/fuse/fuse_internal.c

@@ -373,7 +373,6 @@ fuse_internal_readdir_processdata(struct uio *uio,
 
 /* remove */
 
-#define INVALIDATE_CACHED_VATTRS_UPON_UNLINK 1
 int
 fuse_internal_remove(struct vnode *dvp,
     struct vnode *vp,
@@ -381,15 +380,11 @@ fuse_internal_remove(struct vnode *dvp,
     enum fuse_opcode op)
 {
 	struct fuse_dispatcher fdi;
+	struct fuse_vnode_data *fvdat;
+	int err;
 
-	struct vattr *vap = VTOVA(vp);
+	err = 0;
-
+	fvdat = VTOFUD(vp);
-#if INVALIDATE_CACHED_VATTRS_UPON_UNLINK
-	int need_invalidate = 0;
-	uint64_t target_nlink = 0;
-
-#endif
-	int err = 0;
 
 	debug_printf("dvp=%p, cnp=%p, op=%d\n", vp, cnp, op);
 
@@ -399,13 +394,6 @@ fuse_internal_remove(struct vnode *dvp,
 	memcpy(fdi.indata, cnp->cn_nameptr, cnp->cn_namelen);
 	((char *)fdi.indata)[cnp->cn_namelen] = '\0';
 
-#if INVALIDATE_CACHED_VATTRS_UPON_UNLINK
-	if (vap->va_nlink > 1) {
-		need_invalidate = 1;
-		target_nlink = vap->va_nlink;
-	}
-#endif
-
 	err = fdisp_wait_answ(&fdi);
 	fdisp_destroy(&fdi);
 	return err;
@@ -483,13 +471,13 @@ fuse_internal_newentry_core(struct vnode *dvp,
 	if ((err = fuse_internal_checkentry(feo, vtyp))) {
 		return err;
 	}
-	err = fuse_vnode_get(mp, feo->nodeid, dvp, vpp, cnp, vtyp);
+	err = fuse_vnode_get(mp, feo, feo->nodeid, dvp, vpp, cnp, vtyp);
 	if (err) {
 		fuse_internal_forget_send(mp, cnp->cn_thread, cnp->cn_cred,
 		    feo->nodeid, 1);
 		return err;
 	}
-	cache_attrs(*vpp, feo);
+	cache_attrs(*vpp, feo, NULL);
 
 	return err;
 }
@@ -563,6 +551,7 @@ fuse_internal_vnode_disappear(struct vnode *vp)
 
 	ASSERT_VOP_ELOCKED(vp, "fuse_internal_vnode_disappear");
 	fvdat->flag |= FN_REVOKED;
+	fvdat->valid_attr_cache = false;
 	cache_purge(vp);
 }
 

sys/fs/fuse/fuse_internal.h

@@ -200,15 +200,47 @@ fuse_internal_access(struct vnode *vp,
 
 /* attributes */
 
+/*
+ * Cache FUSE attributes 'fat', with nominal expiration
+ * 'attr_valid'.'attr_valid_nsec', in attr cache associated with vnode 'vp'.
+ * Optionally, if argument 'vap' is not NULL, store a copy of the converted
+ * attributes there as well.
+ *
+ * If the nominal attribute cache TTL is zero, do not cache on the 'vp' (but do
+ * return the result to the caller).
+ */
 static __inline
 void
-fuse_internal_attr_fat2vat(struct mount *mp,
+fuse_internal_attr_fat2vat(struct vnode *vp,
                            struct fuse_attr *fat,
+                           uint64_t attr_valid,
+                           uint32_t attr_valid_nsec,
                            struct vattr *vap)
 {
+    struct mount *mp;
+    struct fuse_vnode_data *fvdat;
+    struct vattr *vp_cache_at;
+
+    mp = vnode_mount(vp);
+    fvdat = VTOFUD(vp);
+
     DEBUGX(FUSE_DEBUG_INTERNAL,
         "node #%ju, mode 0%o\n", (uintmax_t)fat->ino, fat->mode);
 
+    /* Honor explicit do-not-cache requests from user filesystems. */
+    if (attr_valid == 0 && attr_valid_nsec == 0)
+        fvdat->valid_attr_cache = false;
+    else
+        fvdat->valid_attr_cache = true;
+
+    vp_cache_at = VTOVA(vp);
+
+    if (vap == NULL && vp_cache_at == NULL)
+        return;
+
+    if (vap == NULL)
+        vap = vp_cache_at;
+
     vattr_null(vap);
 
     vap->va_fsid = mp->mnt_stat.f_fsid.val[0];
@@ -227,21 +259,17 @@ fuse_internal_attr_fat2vat(struct mount *mp,
     vap->va_ctime.tv_nsec = fat->ctimensec;
     vap->va_blocksize = PAGE_SIZE;
     vap->va_type = IFTOVT(fat->mode);
-
-#if (S_BLKSIZE == 512)
-    /* Optimize this case */
-    vap->va_bytes = fat->blocks << 9;
-#else
     vap->va_bytes = fat->blocks * S_BLKSIZE;
-#endif
-
     vap->va_flags = 0;
+
+    if (vap != vp_cache_at && vp_cache_at != NULL)
+        memcpy(vp_cache_at, vap, sizeof(*vap));
 }
 
 
-#define	cache_attrs(vp, fuse_out)					\
+#define	cache_attrs(vp, fuse_out, vap_out)				\
-	fuse_internal_attr_fat2vat(vnode_mount(vp), &(fuse_out)->attr,	\
+        fuse_internal_attr_fat2vat((vp), &(fuse_out)->attr,		\
-	    VTOVA(vp));
+            (fuse_out)->attr_valid, (fuse_out)->attr_valid_nsec, (vap_out))
 
 /* fsync */
 

sys/fs/fuse/fuse_io.c

@@ -155,7 +155,13 @@ fuse_io_dispatch(struct vnode *vp, struct uio *uio, int ioflag,
 		}
 		break;
 	case UIO_WRITE:
-		if (directio) {
+		/*
+		 * Kludge: simulate write-through caching via write-around
+		 * caching.  Same effect, as far as never caching dirty data,
+		 * but slightly pessimal in that newly written data is not
+		 * cached.
+		 */
+		if (directio || fuse_data_cache_mode == FUSE_CACHE_WT) {
 			FS_DEBUG("direct write of vnode %ju via file handle %ju\n",
 			    (uintmax_t)VTOILLU(vp), (uintmax_t)fufh->fh_id);
 			err = fuse_write_directbackend(vp, uio, cred, fufh, ioflag);
@@ -362,8 +368,11 @@ fuse_write_directbackend(struct vnode *vp, struct uio *uio,
 		}
 		uio->uio_resid += diff;
 		uio->uio_offset -= diff;
-		if (uio->uio_offset > fvdat->filesize)
+		if (uio->uio_offset > fvdat->filesize &&
+		    fuse_data_cache_mode != FUSE_CACHE_UC) {
 			fuse_vnode_setsize(vp, cred, uio->uio_offset);
+			fvdat->flag &= ~FN_SIZECHANGE;
+		}
 	}
 
 	fdisp_destroy(&fdi);
@@ -655,6 +664,7 @@ fuse_io_strategy(struct vnode *vp, struct buf *bp)
 		uiop->uio_offset = ((off_t)bp->b_blkno) * biosize;
 		error = fuse_read_directbackend(vp, uiop, cred, fufh);
 
+		/* XXXCEM: Potentially invalid access to cached_attrs here */
 		if ((!error && uiop->uio_resid) ||
 		    (fsess_opt_brokenio(vnode_mount(vp)) && error == EIO &&
 		    uiop->uio_offset < fvdat->filesize && fvdat->filesize > 0 &&

sys/fs/fuse/fuse_ipc.h

@@ -214,7 +214,13 @@ struct fuse_data {
 #define FSESS_NO_MMAP             0x0800 /* disable mmap */
 #define FSESS_BROKENIO            0x1000 /* fix broken io */
 
-extern int fuse_data_cache_enable;
+enum fuse_data_cache_mode {
+	FUSE_CACHE_UC,
+	FUSE_CACHE_WT,
+	FUSE_CACHE_WB,
+};
+
+extern int fuse_data_cache_mode;
 extern int fuse_data_cache_invalidate;
 extern int fuse_mmap_enable;
 extern int fuse_sync_resize;
@@ -248,7 +254,7 @@ fsess_opt_datacache(struct mount *mp)
 {
     struct fuse_data *data = fuse_get_mpdata(mp);
 
-    return (fuse_data_cache_enable ||
+    return (fuse_data_cache_mode != FUSE_CACHE_UC &&
         (data->dataflags & FSESS_NO_DATACACHE) == 0);
 }
 
@@ -257,7 +263,7 @@ fsess_opt_mmap(struct mount *mp)
 {
     struct fuse_data *data = fuse_get_mpdata(mp);
 
-    if (!(fuse_mmap_enable && fuse_data_cache_enable))
+    if (!fuse_mmap_enable || fuse_data_cache_mode == FUSE_CACHE_UC)
         return 0;
     return ((data->dataflags & (FSESS_NO_DATACACHE | FSESS_NO_MMAP)) == 0);
 }

sys/fs/fuse/fuse_node.c

@@ -94,16 +94,19 @@ __FBSDID("$FreeBSD$");
 
 MALLOC_DEFINE(M_FUSEVN, "fuse_vnode", "fuse vnode private data");
 
+static int sysctl_fuse_cache_mode(SYSCTL_HANDLER_ARGS);
+
 static int fuse_node_count = 0;
 
 SYSCTL_INT(_vfs_fuse, OID_AUTO, node_count, CTLFLAG_RD,
     &fuse_node_count, 0, "Count of FUSE vnodes");
 
-int	fuse_data_cache_enable = 1;
+int	fuse_data_cache_mode = FUSE_CACHE_WT;
 
-SYSCTL_INT(_vfs_fuse, OID_AUTO, data_cache_enable, CTLFLAG_RW,
+SYSCTL_PROC(_vfs_fuse, OID_AUTO, data_cache_mode, CTLTYPE_INT|CTLFLAG_RW,
-    &fuse_data_cache_enable, 0,
+    &fuse_data_cache_mode, 0, sysctl_fuse_cache_mode, "I",
-    "enable caching of FUSE file data (including dirty data)");
+    "Zero: disable caching of FUSE file data; One: write-through caching "
+    "(default); Two: write-back caching (generally unsafe)");
 
 int	fuse_data_cache_invalidate = 0;
 
@@ -116,7 +119,7 @@ int	fuse_mmap_enable = 1;
 
 SYSCTL_INT(_vfs_fuse, OID_AUTO, mmap_enable, CTLFLAG_RW,
     &fuse_mmap_enable, 0,
-    "If non-zero, and data_cache_enable is also non-zero, enable mmap(2) of "
+    "If non-zero, and data_cache_mode is also non-zero, enable mmap(2) of "
     "FUSE files");
 
 int	fuse_refresh_size = 0;
@@ -140,6 +143,28 @@ SYSCTL_INT(_vfs_fuse, OID_AUTO, fix_broken_io, CTLFLAG_RW,
     "If non-zero, print a diagnostic warning if a userspace filesystem returns"
     " EIO on reads of recently extended portions of files");
 
+static int
+sysctl_fuse_cache_mode(SYSCTL_HANDLER_ARGS)
+{
+	int val, error;
+
+	val = *(int *)arg1;
+	error = sysctl_handle_int(oidp, &val, 0, req);
+	if (error || !req->newptr)
+		return (error);
+
+	switch (val) {
+	case FUSE_CACHE_UC:
+	case FUSE_CACHE_WT:
+	case FUSE_CACHE_WB:
+		*(int *)arg1 = val;
+		break;
+	default:
+		return (EDOM);
+	}
+	return (0);
+}
+
 static void
 fuse_vnode_init(struct vnode *vp, struct fuse_vnode_data *fvdat,
     uint64_t nodeid, enum vtype vtyp)
@@ -147,6 +172,7 @@ fuse_vnode_init(struct vnode *vp, struct fuse_vnode_data *fvdat,
 	int i;
 
 	fvdat->nid = nodeid;
+	vattr_null(&fvdat->cached_attrs);
 	if (nodeid == FUSE_ROOT_ID) {
 		vp->v_vflag |= VV_ROOT;
 	}
@@ -240,6 +266,7 @@ fuse_vnode_alloc(struct mount *mp,
 
 int
 fuse_vnode_get(struct mount *mp,
+    struct fuse_entry_out *feo,
     uint64_t nodeid,
     struct vnode *dvp,
     struct vnode **vpp,
@@ -260,7 +287,9 @@ fuse_vnode_get(struct mount *mp,
 		MPASS(!(cnp->cn_namelen == 1 && cnp->cn_nameptr[0] == '.'));
 		fuse_vnode_setparent(*vpp, dvp);
 	}
-	if (dvp != NULL && cnp != NULL && (cnp->cn_flags & MAKEENTRY) != 0) {
+	if (dvp != NULL && cnp != NULL && (cnp->cn_flags & MAKEENTRY) != 0 &&
+	    feo != NULL &&
+	    (feo->entry_valid != 0 || feo->entry_valid_nsec != 0)) {
 		ASSERT_VOP_LOCKED(*vpp, "fuse_vnode_get");
 		ASSERT_VOP_LOCKED(dvp, "fuse_vnode_get");
 		cache_enter(dvp, *vpp, cnp);
@@ -371,6 +400,7 @@ fuse_vnode_refreshsize(struct vnode *vp, struct ucred *cred)
 	struct vattr va;
 
 	if ((fvdat->flag & FN_SIZECHANGE) != 0 ||
+	    fuse_data_cache_mode == FUSE_CACHE_UC ||
 	    (fuse_refresh_size == 0 && fvdat->filesize != 0))
 		return;
 

sys/fs/fuse/fuse_node.h

@@ -86,6 +86,7 @@ struct fuse_vnode_data {
     uint32_t   flag;
 
     /** meta **/
+    bool              valid_attr_cache;
     struct vattr      cached_attrs;
     off_t             filesize;
     uint64_t          nlookup;
@@ -95,7 +96,9 @@ struct fuse_vnode_data {
 #define VTOFUD(vp) \
     ((struct fuse_vnode_data *)((vp)->v_data))
 #define VTOI(vp)    (VTOFUD(vp)->nid)
-#define VTOVA(vp)   (&(VTOFUD(vp)->cached_attrs))
+#define VTOVA(vp) \
+    (VTOFUD(vp)->valid_attr_cache ? \
+    &(VTOFUD(vp)->cached_attrs) : NULL)
 #define VTOILLU(vp) ((uint64_t)(VTOFUD(vp) ? VTOI(vp) : 0))
 
 #define FUSE_NULL_ID 0
@@ -114,6 +117,7 @@ fuse_vnode_setparent(struct vnode *vp, struct vnode *dvp)
 void fuse_vnode_destroy(struct vnode *vp);
 
 int fuse_vnode_get(struct mount         *mp,
+                   struct fuse_entry_out *feo,
                    uint64_t              nodeid,
                    struct vnode         *dvp,
                    struct vnode        **vpp,

sys/fs/fuse/fuse_vfsops.c

@@ -444,7 +444,8 @@ fuse_vfsop_root(struct mount *mp, int lkflags, struct vnode **vpp)
 		if (err == 0)
 			*vpp = data->vroot;
 	} else {
-		err = fuse_vnode_get(mp, FUSE_ROOT_ID, NULL, vpp, NULL, VDIR);
+		err = fuse_vnode_get(mp, NULL, FUSE_ROOT_ID, NULL, vpp, NULL,
+		    VDIR);
 		if (err == 0) {
 			FUSE_LOCK();
 			MPASS(data->vroot == NULL || data->vroot == *vpp);

sys/fs/fuse/fuse_vnops.c

@@ -384,7 +384,7 @@ fuse_vnop_create(struct vop_create_args *ap)
 	if ((err = fuse_internal_checkentry(feo, VREG))) {
 		goto out;
 	}
-	err = fuse_vnode_get(mp, feo->nodeid, dvp, vpp, cnp, VREG);
+	err = fuse_vnode_get(mp, feo, feo->nodeid, dvp, vpp, cnp, VREG);
 	if (err) {
 		struct fuse_release_in *fri;
 		uint64_t nodeid = feo->nodeid;
@@ -518,10 +518,8 @@ fuse_vnop_getattr(struct vop_getattr_args *ap)
 		}
 		goto out;
 	}
-	cache_attrs(vp, (struct fuse_attr_out *)fdi.answ);
+
-	if (vap != VTOVA(vp)) {
+	cache_attrs(vp, (struct fuse_attr_out *)fdi.answ, vap);
-		memcpy(vap, VTOVA(vp), sizeof(*vap));
-	}
 	if (vap->va_type != vnode_vtype(vp)) {
 		fuse_internal_vnode_disappear(vp);
 		err = ENOENT;
@@ -540,6 +538,7 @@ fuse_vnop_getattr(struct vop_getattr_args *ap)
 
 		if (fvdat->filesize != new_filesize) {
 			fuse_vnode_setsize(vp, cred, new_filesize);
+			fvdat->flag &= ~FN_SIZECHANGE;
 		}
 	}
 	debug_printf("fuse_getattr e: returning 0\n");
@@ -628,9 +627,15 @@ fuse_vnop_link(struct vop_link_args *ap)
 	if (vnode_mount(tdvp) != vnode_mount(vp)) {
 		return EXDEV;
 	}
-	if (vap->va_nlink >= FUSE_LINK_MAX) {
+
+	/*
+	 * This is a seatbelt check to protect naive userspace filesystems from
+	 * themselves and the limitations of the FUSE IPC protocol.  If a
+	 * filesystem does not allow attribute caching, assume it is capable of
+	 * validating that nlink does not overflow.
+	 */
+	if (vap != NULL && vap->va_nlink >= FUSE_LINK_MAX)
 		return EMLINK;
-	}
 	fli.oldnodeid = VTOI(vp);
 
 	fdisp_init(&fdi, 0);
@@ -853,8 +858,8 @@ fuse_vnop_lookup(struct vop_lookup_args *ap)
 				vref(dvp);
 				*vpp = dvp;
 			} else {
-				err = fuse_vnode_get(dvp->v_mount, nid, dvp,
+				err = fuse_vnode_get(dvp->v_mount, feo, nid,
-				    &vp, cnp, IFTOVT(fattr->mode));
+				    dvp, &vp, cnp, IFTOVT(fattr->mode));
 				if (err)
 					goto out;
 				*vpp = vp;
@@ -889,12 +894,8 @@ fuse_vnop_lookup(struct vop_lookup_args *ap)
 				err = EISDIR;
 				goto out;
 			}
-			err = fuse_vnode_get(vnode_mount(dvp),
+			err = fuse_vnode_get(vnode_mount(dvp), feo, nid, dvp,
-			    nid,
+			    &vp, cnp, IFTOVT(fattr->mode));
-			    dvp,
-			    &vp,
-			    cnp,
-			    IFTOVT(fattr->mode));
 			if (err) {
 				goto out;
 			}
@@ -932,12 +933,8 @@ fuse_vnop_lookup(struct vop_lookup_args *ap)
 				}
 			}
 			VOP_UNLOCK(dvp, 0);
-			err = fuse_vnode_get(vnode_mount(dvp),
+			err = fuse_vnode_get(vnode_mount(dvp), feo, nid, NULL,
-			    nid,
+			    &vp, cnp, IFTOVT(fattr->mode));
-			    NULL,
-			    &vp,
-			    cnp,
-			    IFTOVT(fattr->mode));
 			vfs_unbusy(mp);
 			vn_lock(dvp, ltype | LK_RETRY);
 			if ((dvp->v_iflag & VI_DOOMED) != 0) {
@@ -952,23 +949,54 @@ fuse_vnop_lookup(struct vop_lookup_args *ap)
 			vref(dvp);
 			*vpp = dvp;
 		} else {
-			err = fuse_vnode_get(vnode_mount(dvp),
+			struct fuse_vnode_data *fvdat;
-			    nid,
+
-			    dvp,
+			err = fuse_vnode_get(vnode_mount(dvp), feo, nid, dvp,
-			    &vp,
+			    &vp, cnp, IFTOVT(fattr->mode));
-			    cnp,
-			    IFTOVT(fattr->mode));
 			if (err) {
 				goto out;
 			}
 			fuse_vnode_setparent(vp, dvp);
+
+			/*
+			 * In the case where we are looking up a FUSE node
+			 * represented by an existing cached vnode, and the
+			 * true size reported by FUSE_LOOKUP doesn't match
+			 * the vnode's cached size, fix the vnode cache to
+			 * match the real object size.
+			 *
+			 * This can occur via FUSE distributed filesystems,
+			 * irregular files, etc.
+			 */
+			fvdat = VTOFUD(vp);
+			if (vnode_isreg(vp) &&
+			    fattr->size != fvdat->filesize) {
+				/*
+				 * The FN_SIZECHANGE flag reflects a dirty
+				 * append.  If userspace lets us know our cache
+				 * is invalid, that write was lost.  (Dirty
+				 * writes that do not cause append are also
+				 * lost, but we don't detect them here.)
+				 *
+				 * XXX: Maybe disable WB caching on this mount.
+				 */
+				if (fvdat->flag & FN_SIZECHANGE)
+					printf("%s: WB cache incoherent on "
+					    "%s!\n", __func__,
+					    vnode_mount(vp)->mnt_stat.f_mntonname);
+
+				(void)fuse_vnode_setsize(vp, cred, fattr->size);
+				fvdat->flag &= ~FN_SIZECHANGE;
+			}
 			*vpp = vp;
 		}
 
 		if (op == FUSE_GETATTR) {
-			cache_attrs(*vpp, (struct fuse_attr_out *)fdi.answ);
+			cache_attrs(*vpp, (struct fuse_attr_out *)fdi.answ,
+			    NULL);
 		} else {
-			cache_attrs(*vpp, (struct fuse_entry_out *)fdi.answ);
+			cache_attrs(*vpp, (struct fuse_entry_out *)fdi.answ,
+			    NULL);
 		}
 
 		/* Insert name into cache if appropriate. */
@@ -1643,9 +1671,9 @@ fuse_vnop_setattr(struct vop_setattr_args *ap)
 			err = EAGAIN;
 		}
 	}
-	if (!err && !sizechanged) {
+	if (err == 0)
-		cache_attrs(vp, (struct fuse_attr_out *)fdi.answ);
+		cache_attrs(vp, (struct fuse_attr_out *)fdi.answ, NULL);
-	}
+
 out:
 	fdisp_destroy(&fdi);
 	if (!err && sizechanged) {

sys/i386/include/cpufunc.h

@@ -108,21 +108,47 @@ disable_intr(void)
 	__asm __volatile("cli" : : : "memory");
 }
 
+#ifdef _KERNEL
 static __inline void
 do_cpuid(u_int ax, u_int *p)
 {
 	__asm __volatile("cpuid"
-			 : "=a" (p[0]), "=b" (p[1]), "=c" (p[2]), "=d" (p[3])
+	    : "=a" (p[0]), "=b" (p[1]), "=c" (p[2]), "=d" (p[3])
-			 :  "0" (ax));
+	    :  "0" (ax));
 }
 
 static __inline void
 cpuid_count(u_int ax, u_int cx, u_int *p)
 {
 	__asm __volatile("cpuid"
-			 : "=a" (p[0]), "=b" (p[1]), "=c" (p[2]), "=d" (p[3])
+	    : "=a" (p[0]), "=b" (p[1]), "=c" (p[2]), "=d" (p[3])
-			 :  "0" (ax), "c" (cx));
+	    :  "0" (ax), "c" (cx));
 }
+#else
+static __inline void
+do_cpuid(u_int ax, u_int *p)
+{
+	__asm __volatile(
+	    "pushl\t%%ebx\n\t"
+	    "cpuid\n\t"
+	    "movl\t%%ebx,%1\n\t"
+	    "popl\t%%ebx"
+	    : "=a" (p[0]), "=DS" (p[1]), "=c" (p[2]), "=d" (p[3])
+	    :  "0" (ax));
+}
+
+static __inline void
+cpuid_count(u_int ax, u_int cx, u_int *p)
+{
+	__asm __volatile(
+	    "pushl\t%%ebx\n\t"
+	    "cpuid\n\t"
+	    "movl\t%%ebx,%1\n\t"
+	    "popl\t%%ebx"
+	    : "=a" (p[0]), "=DS" (p[1]), "=c" (p[2]), "=d" (p[3])
+	    :  "0" (ax), "c" (cx));
+}
+#endif
 
 static __inline void
 enable_intr(void)

sys/kern/kern_resource.c

@@ -863,13 +863,88 @@ rufetchtd(struct thread *td, struct rusage *ru)
 	calcru1(p, &td->td_rux, &ru->ru_utime, &ru->ru_stime);
 }
 
+/* XXX: the MI version is too slow to use: */
+#ifndef __HAVE_INLINE_FLSLL
+#define	flsll(x)	(fls((x) >> 32) != 0 ? fls((x) >> 32) + 32 : fls(x))
+#endif
+
 static uint64_t
 mul64_by_fraction(uint64_t a, uint64_t b, uint64_t c)
 {
+	uint64_t acc, bh, bl;
+	int i, s, sa, sb;
+
 	/*
-	 * Compute floor(a * (b / c)) without overflowing, (b / c) <= 1.0.
+	 * Calculate (a * b) / c accurately enough without overflowing.  c
+	 * must be nonzero, and its top bit must be 0.  a or b must be
+	 * <= c, and the implementation is tuned for b <= c.
+	 *
+	 * The comments about times are for use in calcru1() with units of
+	 * microseconds for 'a' and stathz ticks at 128 Hz for b and c.
+	 *
+	 * Let n be the number of top zero bits in c.  Each iteration
+	 * either returns, or reduces b by right shifting it by at least n.
+	 * The number of iterations is at most 1 + 64 / n, and the error is
+	 * at most the number of iterations.
+	 *
+	 * It is very unusual to need even 2 iterations.  Previous
+	 * implementations overflowed essentially by returning early in the
+	 * first iteration, with n = 38 giving overflow at 105+ hours and
+	 * n = 32 giving overlow at at 388+ days despite a more careful
+	 * calculation.  388 days is a reasonable uptime, and the calculation
+	 * needs to work for the uptime times the number of CPUs since 'a'
+	 * is per-process.
 	 */
-	return ((a / c) * b + (a % c) * (b / c) + (a % c) * (b % c) / c);
+	if (a >= (uint64_t)1 << 63)
+		return (0);		/* Unsupported arg -- can't happen. */
+	acc = 0;
+	for (i = 0; i < 128; i++) {
+		sa = flsll(a);
+		sb = flsll(b);
+		if (sa + sb <= 64)
+			/* Up to 105 hours on first iteration. */
+			return (acc + (a * b) / c);
+		if (a >= c) {
+			/*
+			 * This reduction is based on a = q * c + r, with the
+			 * remainder r < c.  'a' may be large to start, and
+			 * moving bits from b into 'a' at the end of the loop
+			 * sets the top bit of 'a', so the reduction makes
+			 * significant progress.
+			 */
+			acc += (a / c) * b;
+			a %= c;
+			sa = flsll(a);
+			if (sa + sb <= 64)
+				/* Up to 388 days on first iteration. */
+				return (acc + (a * b) / c);
+		}
+
+		/*
+		 * This step writes a * b as a * ((bh << s) + bl) =
+		 * a * (bh << s) + a * bl = (a << s) * bh + a * bl.  The 2
+		 * additive terms are handled separately.  Splitting in
+		 * this way is linear except for rounding errors.
+		 *
+		 * s = 64 - sa is the maximum such that a << s fits in 64
+		 * bits.  Since a < c and c has at least 1 zero top bit,
+		 * sa < 64 and s > 0.  Thus this step makes progress by
+		 * reducing b (it increases 'a', but taking remainders on
+		 * the next iteration completes the reduction).
+		 *
+		 * Finally, the choice for s is just what is needed to keep
+		 * a * bl from overflowing, so we don't need complications
+		 * like a recursive call mul64_by_fraction(a, bl, c) to
+		 * handle the second additive term.
+		 */
+		s = 64 - sa;
+		bh = b >> s;
+		bl = b - (bh << s);
+		acc += (a * bl) / c;
+		a <<= s;
+		b = bh;
+	}
+	return (0);		/* Algorithm failure -- can't happen. */
 }
 
 static void
@@ -896,15 +971,23 @@ calcru1(struct proc *p, struct rusage_ext *ruxp, struct timeval *up,
 		tu = ruxp->rux_tu;
 	}
 
+	/* Subdivide tu.  Avoid overflow in the multiplications. */
+	if (__predict_true(tu <= ((uint64_t)1 << 38) && tt <= (1 << 26))) {
+		/* Up to 76 hours when stathz is 128. */
+		uu = (tu * ut) / tt;
+		su = (tu * st) / tt;
+	} else {
+		uu = mul64_by_fraction(tu, ut, tt);
+		su = mul64_by_fraction(tu, ut, st);
+	}
+
 	if (tu >= ruxp->rux_tu) {
 		/*
 		 * The normal case, time increased.
 		 * Enforce monotonicity of bucketed numbers.
 		 */
-		uu = mul64_by_fraction(tu, ut, tt);
 		if (uu < ruxp->rux_uu)
 			uu = ruxp->rux_uu;
-		su = mul64_by_fraction(tu, st, tt);
 		if (su < ruxp->rux_su)
 			su = ruxp->rux_su;
 	} else if (tu + 3 > ruxp->rux_tu || 101 * tu > 100 * ruxp->rux_tu) {
@@ -933,8 +1016,6 @@ calcru1(struct proc *p, struct rusage_ext *ruxp, struct timeval *up,
 		    "to %ju usec for pid %d (%s)\n",
 		    (uintmax_t)ruxp->rux_tu, (uintmax_t)tu,
 		    p->p_pid, p->p_comm);
-		uu = mul64_by_fraction(tu, ut, tt);
-		su = mul64_by_fraction(tu, st, tt);
 	}
 
 	ruxp->rux_uu = uu;

sys/kern/sys_pipe.c

@@ -379,9 +379,7 @@ void
 pipe_dtor(struct pipe *dpipe)
 {
 	struct pipe *peer;
-	ino_t ino;
 
-	ino = dpipe->pipe_ino;
 	peer = (dpipe->pipe_state & PIPE_NAMED) != 0 ? dpipe->pipe_peer : NULL;
 	funsetown(&dpipe->pipe_sigio);
 	pipeclose(dpipe);

sys/mips/mips/elf_machdep.c

@@ -76,7 +76,7 @@ struct sysentvec elf64_freebsd_sysvec = {
 	.sv_setregs	= exec_setregs,
 	.sv_fixlimit	= NULL,
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_ABI_FREEBSD | SV_LP64,
+	.sv_flags	= SV_ABI_FREEBSD | SV_LP64 | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,
@@ -131,7 +131,7 @@ struct sysentvec elf32_freebsd_sysvec = {
 	.sv_setregs	= exec_setregs,
 	.sv_fixlimit	= NULL,
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_ABI_FREEBSD | SV_ILP32,
+	.sv_flags	= SV_ABI_FREEBSD | SV_ILP32 | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,

sys/modules/crypto/Makefile

@@ -68,5 +68,7 @@ CFLAGS.utils.c			+= -I${LIBSODIUM_INC} -I${LIBSODIUM_COMPAT}
 
 SRCS	+= opt_param.h cryptodev_if.h bus_if.h device_if.h
 SRCS	+= opt_ddb.h
+SRCS	+= cbc_mac.c
+SRCS	+= xform_cbc_mac.c
 
 .include <bsd.kmod.mk>

sys/net/if_lagg.c

@@ -133,6 +133,7 @@ static int	lagg_ioctl(struct ifnet *, u_long, caddr_t);
 static int	lagg_snd_tag_alloc(struct ifnet *,
 		    union if_snd_tag_alloc_params *,
 		    struct m_snd_tag **);
+static void	lagg_snd_tag_free(struct m_snd_tag *);
 #endif
 static int	lagg_setmulti(struct lagg_port *);
 static int	lagg_clrmulti(struct lagg_port *);
@@ -514,6 +515,7 @@ lagg_clone_create(struct if_clone *ifc, int unit, caddr_t params)
 	ifp->if_flags = IFF_SIMPLEX | IFF_BROADCAST | IFF_MULTICAST;
 #ifdef RATELIMIT
 	ifp->if_snd_tag_alloc = lagg_snd_tag_alloc;
+	ifp->if_snd_tag_free = lagg_snd_tag_free;
 #endif
 	ifp->if_capenable = ifp->if_capabilities = IFCAP_HWSTATS;
 
@@ -1568,6 +1570,13 @@ lagg_snd_tag_alloc(struct ifnet *ifp,
 	/* forward allocation request */
 	return (ifp->if_snd_tag_alloc(ifp, params, ppmt));
 }
+
+static void
+lagg_snd_tag_free(struct m_snd_tag *tag)
+{
+	tag->ifp->if_snd_tag_free(tag);
+}
+
 #endif
 
 static int

sys/net/if_vlan.c

@@ -267,6 +267,7 @@ static	int vlan_ioctl(struct ifnet *ifp, u_long cmd, caddr_t addr);
 #ifdef RATELIMIT
 static	int vlan_snd_tag_alloc(struct ifnet *,
     union if_snd_tag_alloc_params *, struct m_snd_tag **);
+static void vlan_snd_tag_free(struct m_snd_tag *);
 #endif
 static	void vlan_qflush(struct ifnet *ifp);
 static	int vlan_setflag(struct ifnet *ifp, int flag, int status,
@@ -1047,6 +1048,7 @@ vlan_clone_create(struct if_clone *ifc, char *name, size_t len, caddr_t params)
 	ifp->if_ioctl = vlan_ioctl;
 #ifdef RATELIMIT
 	ifp->if_snd_tag_alloc = vlan_snd_tag_alloc;
+	ifp->if_snd_tag_free = vlan_snd_tag_free;
 #endif
 	ifp->if_flags = VLAN_IFFLAGS;
 	ether_ifattach(ifp, eaddr);
@@ -1934,4 +1936,10 @@ vlan_snd_tag_alloc(struct ifnet *ifp,
 	/* forward allocation request */
 	return (ifp->if_snd_tag_alloc(ifp, params, ppmt));
 }
+
+static void
+vlan_snd_tag_free(struct m_snd_tag *tag)
+{
+	tag->ifp->if_snd_tag_free(tag);
+}
 #endif

sys/net/iflib.c

@@ -1468,12 +1468,17 @@ iflib_fast_intr(void *arg)
 {
 	iflib_filter_info_t info = arg;
 	struct grouptask *gtask = info->ifi_task;
+	int result;
+
 	if (!iflib_started)
-		return (FILTER_HANDLED);
+		return (FILTER_STRAY);
 
 	DBG_COUNTER_INC(fast_intrs);
-	if (info->ifi_filter != NULL && info->ifi_filter(info->ifi_filter_arg) == FILTER_HANDLED)
+	if (info->ifi_filter != NULL) {
-		return (FILTER_HANDLED);
+		result = info->ifi_filter(info->ifi_filter_arg);
+		if ((result & FILTER_SCHEDULE_THREAD) == 0)
+			return (result);
+	}
 
 	GROUPTASK_ENQUEUE(gtask);
 	return (FILTER_HANDLED);
@@ -1488,15 +1493,18 @@ iflib_fast_intr_rxtx(void *arg)
 	iflib_rxq_t rxq = (iflib_rxq_t)info->ifi_ctx;
 	iflib_txq_t txq;
 	void *sc;
-	int i, cidx;
+	int i, cidx, result;
 	qidx_t txqid;
 
 	if (!iflib_started)
-		return (FILTER_HANDLED);
+		return (FILTER_STRAY);
 
 	DBG_COUNTER_INC(fast_intrs);
-	if (info->ifi_filter != NULL && info->ifi_filter(info->ifi_filter_arg) == FILTER_HANDLED)
+	if (info->ifi_filter != NULL) {
-		return (FILTER_HANDLED);
+		result = info->ifi_filter(info->ifi_filter_arg);
+		if ((result & FILTER_SCHEDULE_THREAD) == 0)
+			return (result);
+	}
 
 	ctx = rxq->ifr_ctx;
 	sc = ctx->ifc_softc;
@@ -1531,13 +1539,17 @@ iflib_fast_intr_ctx(void *arg)
 {
 	iflib_filter_info_t info = arg;
 	struct grouptask *gtask = info->ifi_task;
+	int result;
 
 	if (!iflib_started)
-		return (FILTER_HANDLED);
+		return (FILTER_STRAY);
 
 	DBG_COUNTER_INC(fast_intrs);
-	if (info->ifi_filter != NULL && info->ifi_filter(info->ifi_filter_arg) == FILTER_HANDLED)
+	if (info->ifi_filter != NULL) {
-		return (FILTER_HANDLED);
+		result = info->ifi_filter(info->ifi_filter_arg);
+		if ((result & FILTER_SCHEDULE_THREAD) == 0)
+			return (result);
+	}
 
 	GROUPTASK_ENQUEUE(gtask);
 	return (FILTER_HANDLED);

sys/netinet/in_pcb.c

@@ -1565,6 +1565,7 @@ in_pcbfree_deferred(epoch_context_t ctx)
 	inp = __containerof(ctx, struct inpcb, inp_epoch_ctx);
 
 	INP_WLOCK(inp);
+	CURVNET_SET(inp->inp_vnet);
 #ifdef INET
 	struct ip_moptions *imo = inp->inp_moptions;
 	inp->inp_moptions = NULL;
@@ -1597,6 +1598,7 @@ in_pcbfree_deferred(epoch_context_t ctx)
 #ifdef INET
 	inp_freemoptions(imo);
 #endif	
+	CURVNET_RESTORE();
 }
 
 /*

sys/netinet/tcp_timewait.c

@@ -302,7 +302,7 @@ tcp_twstart(struct tcpcb *tp)
 	if (SEQ_GT(tp->rcv_adv, tp->rcv_nxt) &&
 	    recwin < (tp->rcv_adv - tp->rcv_nxt))
 		recwin = (tp->rcv_adv - tp->rcv_nxt);
-	tw->last_win = htons((u_short)(recwin >> tp->rcv_scale));
+	tw->last_win = (u_short)(recwin >> tp->rcv_scale);
 
 	/*
 	 * Set t_recent if timestamps are used on the connection.

sys/opencrypto/cbc_mac.c

@@ -0,0 +1,252 @@
+/*
+ * Copyright (c) 2018-2019 iXsystems Inc.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <sys/types.h>
+#include <sys/systm.h>
+#include <sys/param.h>
+#include <sys/endian.h>
+#include <opencrypto/cbc_mac.h>
+#include <opencrypto/xform_auth.h>
+
+/*
+ * Given two CCM_CBC_BLOCK_LEN blocks, xor
+ * them into dst, and then encrypt dst.
+ */
+static void
+xor_and_encrypt(struct aes_cbc_mac_ctx *ctx,
+		const uint8_t *src, uint8_t *dst)
+{
+	const uint64_t *b1;
+	uint64_t *b2;
+	uint64_t temp_block[CCM_CBC_BLOCK_LEN/sizeof(uint64_t)];
+
+	b1 = (const uint64_t*)src;
+	b2 = (uint64_t*)dst;
+
+	for (size_t count = 0;
+	     count < CCM_CBC_BLOCK_LEN/sizeof(uint64_t);
+	     count++) {
+		temp_block[count] = b1[count] ^ b2[count];
+	}
+	rijndaelEncrypt(ctx->keysched, ctx->rounds, (void*)temp_block, dst);
+}
+
+void
+AES_CBC_MAC_Init(struct aes_cbc_mac_ctx *ctx)
+{
+	bzero(ctx, sizeof(*ctx));
+}
+
+void
+AES_CBC_MAC_Setkey(struct aes_cbc_mac_ctx *ctx, const uint8_t *key, uint16_t klen)
+{
+	ctx->rounds = rijndaelKeySetupEnc(ctx->keysched, key, klen * 8);
+}
+
+/*
+ * This is called to set the nonce, aka IV.
+ * Before this call, the authDataLength and cryptDataLength fields
+ * MUST have been set.  Sadly, there's no way to return an error.
+ *
+ * The CBC-MAC algorithm requires that the first block contain the
+ * nonce, as well as information about the sizes and lengths involved.
+ */
+void
+AES_CBC_MAC_Reinit(struct aes_cbc_mac_ctx *ctx, const uint8_t *nonce, uint16_t nonceLen)
+{
+	uint8_t b0[CCM_CBC_BLOCK_LEN];
+	uint8_t *bp = b0, flags = 0;
+	uint8_t L = 0;
+	uint64_t dataLength = ctx->cryptDataLength;
+	
+	KASSERT(ctx->authDataLength != 0 || ctx->cryptDataLength != 0,
+	    ("Auth Data and Data lengths cannot both be 0"));
+
+	KASSERT(nonceLen >= 7 && nonceLen <= 13,
+	    ("nonceLen must be between 7 and 13 bytes"));
+
+	ctx->nonce = nonce;
+	ctx->nonceLength = nonceLen;
+	
+	ctx->authDataCount = 0;
+	ctx->blockIndex = 0;
+	explicit_bzero(ctx->staging_block, sizeof(ctx->staging_block));
+	
+	/*
+	 * Need to determine the L field value.  This is the number of
+	 * bytes needed to specify the length of the message; the length
+	 * is whatever is left in the 16 bytes after specifying flags and
+	 * the nonce.
+	 */
+	L = 15 - nonceLen;
+	
+	flags = ((ctx->authDataLength > 0) << 6) +
+	    (((AES_CBC_MAC_HASH_LEN - 2) / 2) << 3) +
+	    L - 1;
+	/*
+	 * Now we need to set up the first block, which has flags, nonce,
+	 * and the message length.
+	 */
+	b0[0] = flags;
+	bcopy(nonce, b0 + 1, nonceLen);
+	bp = b0 + 1 + nonceLen;
+
+	/* Need to copy L' [aka L-1] bytes of cryptDataLength */
+	for (uint8_t *dst = b0 + sizeof(b0) - 1; dst >= bp; dst--) {
+		*dst = dataLength;
+		dataLength >>= 8;
+	}
+	/* Now need to encrypt b0 */
+	rijndaelEncrypt(ctx->keysched, ctx->rounds, b0, ctx->block);
+	/* If there is auth data, we need to set up the staging block */
+	if (ctx->authDataLength) {
+		if (ctx->authDataLength < ((1<<16) - (1<<8))) {
+			uint16_t sizeVal = htobe16(ctx->authDataLength);
+			bcopy(&sizeVal, ctx->staging_block, sizeof(sizeVal));
+			ctx->blockIndex = sizeof(sizeVal);
+		} else if (ctx->authDataLength < (1ULL<<32)) {
+			uint32_t sizeVal = htobe32(ctx->authDataLength);
+			ctx->staging_block[0] = 0xff;
+			ctx->staging_block[1] = 0xfe;
+			bcopy(&sizeVal, ctx->staging_block+2, sizeof(sizeVal));
+			ctx->blockIndex = 2 + sizeof(sizeVal);
+		} else {
+			uint64_t sizeVal = htobe64(ctx->authDataLength);
+			ctx->staging_block[0] = 0xff;
+			ctx->staging_block[1] = 0xff;
+			bcopy(&sizeVal, ctx->staging_block+2, sizeof(sizeVal));
+			ctx->blockIndex = 2 + sizeof(sizeVal);
+		}
+	}
+}
+
+int
+AES_CBC_MAC_Update(struct aes_cbc_mac_ctx *ctx, const uint8_t *data,
+    uint16_t length)
+{
+	size_t copy_amt;
+	
+	/*
+	 * This will be called in one of two phases:
+	 * (1)  Applying authentication data, or
+	 * (2)  Applying the payload data.
+	 *
+	 * Because CBC-MAC puts the authentication data size before the
+	 * data, subsequent calls won't be block-size-aligned.  Which
+	 * complicates things a fair bit.
+	 *
+	 * The payload data doesn't have that problem.
+	 */
+				
+	if (ctx->authDataCount < ctx->authDataLength) {
+		/*
+		 * We need to process data as authentication data.
+		 * Since we may be out of sync, we may also need
+		 * to pad out the staging block.
+		 */
+		const uint8_t *ptr = data;
+		while (length > 0) {
+
+			copy_amt = MIN(length,
+			    sizeof(ctx->staging_block) - ctx->blockIndex);
+
+			bcopy(ptr, ctx->staging_block + ctx->blockIndex,
+			    copy_amt);
+			ptr += copy_amt;
+			length -= copy_amt;
+			ctx->authDataCount += copy_amt;
+			ctx->blockIndex += copy_amt;
+			ctx->blockIndex %= sizeof(ctx->staging_block);
+			if (ctx->authDataCount == ctx->authDataLength)
+				length = 0;
+			if (ctx->blockIndex == 0 ||
+			    ctx->authDataCount >= ctx->authDataLength) {
+				/*
+				 * We're done with this block, so we
+				 * xor staging_block with block, and then
+				 * encrypt it.
+				 */
+				xor_and_encrypt(ctx, ctx->staging_block, ctx->block);
+				bzero(ctx->staging_block, sizeof(ctx->staging_block));
+				ctx->blockIndex = 0;
+			}
+		}
+		return (0);
+	}
+	/*
+	 * If we're here, then we're encoding payload data.
+	 * This is marginally easier, except that _Update can
+	 * be called with non-aligned update lengths. As a result,
+	 * we still need to use the staging block.
+	 */
+	KASSERT((length + ctx->cryptDataCount) <= ctx->cryptDataLength,
+	    ("More encryption data than allowed"));
+
+	while (length) {
+		uint8_t *ptr;
+		
+		copy_amt = MIN(sizeof(ctx->staging_block) - ctx->blockIndex,
+		    length);
+		ptr = ctx->staging_block + ctx->blockIndex;
+		bcopy(data, ptr, copy_amt);
+		data += copy_amt;
+		ctx->blockIndex += copy_amt;
+		ctx->cryptDataCount += copy_amt;
+		length -= copy_amt;
+		if (ctx->blockIndex == sizeof(ctx->staging_block)) {
+			/* We've got a full block */
+			xor_and_encrypt(ctx, ctx->staging_block, ctx->block);
+			ctx->blockIndex = 0;
+			bzero(ctx->staging_block, sizeof(ctx->staging_block));
+		}
+	}
+	return (0);
+}
+
+void
+AES_CBC_MAC_Final(uint8_t *buf, struct aes_cbc_mac_ctx *ctx)
+{
+	uint8_t s0[CCM_CBC_BLOCK_LEN];
+	
+	/*
+	 * We first need to check to see if we've got any data
+	 * left over to encrypt.
+	 */
+	if (ctx->blockIndex != 0) {
+		xor_and_encrypt(ctx, ctx->staging_block, ctx->block);
+		ctx->cryptDataCount += ctx->blockIndex;
+		ctx->blockIndex = 0;
+		explicit_bzero(ctx->staging_block, sizeof(ctx->staging_block));
+	}
+	bzero(s0, sizeof(s0));
+	s0[0] = (15 - ctx->nonceLength) - 1;
+	bcopy(ctx->nonce, s0 + 1, ctx->nonceLength);
+	rijndaelEncrypt(ctx->keysched, ctx->rounds, s0, s0);
+	for (size_t indx = 0; indx < AES_CBC_MAC_HASH_LEN; indx++)
+		buf[indx] = ctx->block[indx] ^ s0[indx];
+	explicit_bzero(s0, sizeof(s0));
+}

sys/opencrypto/cbc_mac.h

@@ -0,0 +1,67 @@
+/*
+ * Copyright (c) 2014 The FreeBSD Foundation
+ * Copyright (c) 2018, iXsystems Inc.
+ * All rights reserved.
+ *
+ * This software was developed by Sean Eric Fagan, with lots of references
+ * to existing AES-CCM (gmac) code.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1.  Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ * 2.  Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *	$FreeBSD$
+ *
+ */
+
+#ifndef _CBC_CCM_H
+# define _CBC_CCM_H
+
+# include <sys/types.h>
+# include <crypto/rijndael/rijndael.h>
+
+# define CCM_CBC_BLOCK_LEN	16	/* 128 bits */
+# define CCM_CBC_MAX_DIGEST_LEN	16
+# define CCM_CBC_MIN_DIGEST_LEN	4
+
+/*
+ * This is the authentication context structure;
+ * the encryption one is similar.
+ */
+struct aes_cbc_mac_ctx {
+	uint64_t	authDataLength, authDataCount;
+	uint64_t	cryptDataLength, cryptDataCount;
+	int		blockIndex;
+	uint8_t		staging_block[CCM_CBC_BLOCK_LEN];
+	uint8_t		block[CCM_CBC_BLOCK_LEN];
+	const uint8_t	*nonce;
+	int		nonceLength;	/* This one is in bytes, not bits! */
+	/* AES state data */
+	int		rounds;
+	uint32_t	keysched[4*(RIJNDAEL_MAXNR+1)];
+};
+
+void AES_CBC_MAC_Init(struct aes_cbc_mac_ctx *);
+void AES_CBC_MAC_Setkey(struct aes_cbc_mac_ctx *, const uint8_t *, uint16_t);
+void AES_CBC_MAC_Reinit(struct aes_cbc_mac_ctx *, const uint8_t *, uint16_t);
+int AES_CBC_MAC_Update(struct aes_cbc_mac_ctx *, const uint8_t *, uint16_t);
+void AES_CBC_MAC_Final(uint8_t *, struct aes_cbc_mac_ctx *);
+
+#endif /* _CBC_CCM_H */

sys/opencrypto/cryptodev.c

@@ -444,6 +444,9 @@ cryptof_ioctl(
 		case CRYPTO_CHACHA20:
 			txform = &enc_xform_chacha20;
 			break;
+		case CRYPTO_AES_CCM_16:
+			txform = &enc_xform_ccm;
+			break;
 
 		default:
 			CRYPTDEB("invalid cipher");
@@ -488,6 +491,25 @@ cryptof_ioctl(
 			thash = &auth_hash_nist_gmac_aes_256;
 			break;
 
+		case CRYPTO_AES_CCM_CBC_MAC:
+			switch (sop->keylen) {
+			case 16:
+				thash = &auth_hash_ccm_cbc_mac_128;
+				break;
+			case 24:
+				thash = &auth_hash_ccm_cbc_mac_192;
+				break;
+			case 32:
+				thash = &auth_hash_ccm_cbc_mac_256;
+				break;
+			default:
+				CRYPTDEB("Invalid CBC MAC key size %d",
+				    sop->keylen);
+				SDT_PROBE1(opencrypto, dev, ioctl,
+				    error, __LINE__);
+				return (EINVAL);
+			}
+			break;
 #ifdef notdef
 		case CRYPTO_MD5:
 			thash = &auth_hash_md5;
@@ -1003,12 +1025,13 @@ cryptodev_aead(
 	}
 
 	/*
-	 * For GCM, crd_len covers only the AAD.  For other ciphers
+	 * For GCM/CCM, crd_len covers only the AAD.  For other ciphers
 	 * chained with an HMAC, crd_len covers both the AAD and the
 	 * cipher text.
 	 */
 	crda->crd_skip = 0;
-	if (cse->cipher == CRYPTO_AES_NIST_GCM_16)
+	if (cse->cipher == CRYPTO_AES_NIST_GCM_16 ||
+	    cse->cipher == CRYPTO_AES_CCM_16)
 		crda->crd_len = caead->aadlen;
 	else
 		crda->crd_len = caead->aadlen + caead->len;

sys/opencrypto/cryptodev.h

@@ -86,6 +86,7 @@
 #define	SHA1_KPDK_HASH_LEN	20
 #define	AES_GMAC_HASH_LEN	16
 #define	POLY1305_HASH_LEN	16
+#define	AES_CBC_MAC_HASH_LEN	16
 /* Maximum hash algorithm result length */
 #define	HASH_MAX_LEN		SHA2_512_HASH_LEN /* Keep this updated */
 
@@ -107,6 +108,9 @@
 #define	AES_128_GMAC_KEY_LEN		16
 #define	AES_192_GMAC_KEY_LEN		24
 #define	AES_256_GMAC_KEY_LEN		32
+#define	AES_128_CBC_MAC_KEY_LEN		16
+#define	AES_192_CBC_MAC_KEY_LEN		24
+#define	AES_256_CBC_MAC_KEY_LEN		32
 
 #define	POLY1305_KEY_LEN		32
 
@@ -129,6 +133,7 @@
 
 #define	ARC4_IV_LEN		1
 #define	AES_GCM_IV_LEN		12
+#define	AES_CCM_IV_LEN		12
 #define	AES_XTS_IV_LEN		8
 #define	AES_XTS_ALPHA		0x87	/* GF(2^128) generator polynomial */
 
@@ -199,7 +204,9 @@
 #define	CRYPTO_SHA2_384		36
 #define	CRYPTO_SHA2_512		37
 #define	CRYPTO_POLY1305		38
-#define	CRYPTO_ALGORITHM_MAX	38 /* Keep updated - see below */
+#define	CRYPTO_AES_CCM_CBC_MAC	39	/* auth side */
+#define	CRYPTO_AES_CCM_16	40	/* cipher side */
+#define	CRYPTO_ALGORITHM_MAX	40	/* Keep updated - see below */
 
 #define	CRYPTO_ALGO_VALID(x)	((x) >= CRYPTO_ALGORITHM_MIN && \
 				 (x) <= CRYPTO_ALGORITHM_MAX)

sys/opencrypto/cryptosoft.c

@@ -62,6 +62,9 @@ __FBSDID("$FreeBSD$");
 #include <sys/bus.h>
 #include "cryptodev_if.h"
 
+_Static_assert(AES_CCM_IV_LEN == AES_GCM_IV_LEN,
+    "AES_GCM_IV_LEN must currently be the same as AES_CCM_IV_LEN");
+
 static	int32_t swcr_id;
 
 u_int8_t hmac_ipad_buffer[HMAC_MAX_BLOCK_LEN];
@@ -506,6 +509,7 @@ swcr_authenc(struct cryptop *crp)
 	caddr_t buf = (caddr_t)crp->crp_buf;
 	uint32_t *blkp;
 	int aadlen, blksz, i, ivlen, len, iskip, oskip, r;
+	int isccm = 0;
 
 	ivlen = blksz = iskip = oskip = 0;
 
@@ -520,13 +524,18 @@ swcr_authenc(struct cryptop *crp)
 
 		sw = &ses->swcr_algorithms[i];
 		switch (sw->sw_alg) {
+		case CRYPTO_AES_CCM_16:
 		case CRYPTO_AES_NIST_GCM_16:
 		case CRYPTO_AES_NIST_GMAC:
 			swe = sw;
 			crde = crd;
 			exf = swe->sw_exf;
-			ivlen = 12;
+			/* AES_CCM_IV_LEN and AES_GCM_IV_LEN are both 12 */
+			ivlen = AES_CCM_IV_LEN;
 			break;
+		case CRYPTO_AES_CCM_CBC_MAC:
+			isccm = 1;
+			/* FALLTHROUGH */
 		case CRYPTO_AES_128_NIST_GMAC:
 		case CRYPTO_AES_192_NIST_GMAC:
 		case CRYPTO_AES_256_NIST_GMAC:
@@ -544,8 +553,26 @@ swcr_authenc(struct cryptop *crp)
 	}
 	if (crde == NULL || crda == NULL)
 		return (EINVAL);
+	/*
+	 * We need to make sure that the auth algorithm matches the
+	 * encr algorithm.  Specifically, for AES-GCM must go with
+	 * AES NIST GMAC, and AES-CCM must go with CBC-MAC.
+	 */
+	if (crde->crd_alg == CRYPTO_AES_NIST_GCM_16) {
+		switch (crda->crd_alg) {
+		case CRYPTO_AES_128_NIST_GMAC:
+		case CRYPTO_AES_192_NIST_GMAC:
+		case CRYPTO_AES_256_NIST_GMAC:
+			break;	/* Good! */
+		default:
+			return (EINVAL);	/* Not good! */
+		}
+	} else if (crde->crd_alg == CRYPTO_AES_CCM_16 &&
+	    crda->crd_alg != CRYPTO_AES_CCM_CBC_MAC)
+		return (EINVAL);
 
-	if (crde->crd_alg == CRYPTO_AES_NIST_GCM_16 &&
+	if ((crde->crd_alg == CRYPTO_AES_NIST_GCM_16 ||
+	     crde->crd_alg == CRYPTO_AES_CCM_16) &&
 	    (crde->crd_flags & CRD_F_IV_EXPLICIT) == 0)
 		return (EINVAL);
 
@@ -576,6 +603,15 @@ swcr_authenc(struct cryptop *crp)
 		}
 	}
 
+	if (swa->sw_alg == CRYPTO_AES_CCM_CBC_MAC) {
+		/*
+		 * AES CCM-CBC needs to know the length of
+		 * both the auth data, and payload data, before
+		 * doing the auth computation.
+		 */
+		ctx.aes_cbc_mac_ctx.authDataLength = crda->crd_len;
+		ctx.aes_cbc_mac_ctx.cryptDataLength = crde->crd_len;
+	}
 	/* Supply MAC with IV */
 	if (axf->Reinit)
 		axf->Reinit(&ctx, iv, ivlen);
@@ -610,16 +646,30 @@ swcr_authenc(struct cryptop *crp)
 			bzero(blk, blksz);
 		crypto_copydata(crp->crp_flags, buf, crde->crd_skip + i, len,
 		    blk);
+		/*
+		 * One of the problems with CCM+CBC is that the authentication
+		 * is done on the unecncrypted data.  As a result, we have
+		 * to do the authentication update at different times,
+		 * depending on whether it's CCM or not.
+		 */
 		if (crde->crd_flags & CRD_F_ENCRYPT) {
+			if (isccm)
+				axf->Update(&ctx, blk, len);
 			if (exf->encrypt_multi != NULL)
 				exf->encrypt_multi(swe->sw_kschedule, blk,
 				    len);
 			else
 				exf->encrypt(swe->sw_kschedule, blk);
-			axf->Update(&ctx, blk, len);
+			if (!isccm)
+				axf->Update(&ctx, blk, len);
 			crypto_copyback(crp->crp_flags, buf,
 			    crde->crd_skip + i, len, blk);
 		} else {
+			if (isccm) {
+				KASSERT(exf->encrypt_multi == NULL,
+				    ("assume CCM is single-block only"));
+				exf->decrypt(swe->sw_kschedule, blk);
+			}
 			axf->Update(&ctx, blk, len);
 		}
 	}
@@ -650,6 +700,11 @@ swcr_authenc(struct cryptop *crp)
 		r = timingsafe_bcmp(aalg, uaalg, axf->hashsize);
 		if (r == 0) {
 			/* tag matches, decrypt data */
+			if (isccm) {
+				KASSERT(exf->reinit != NULL,
+				    ("AES-CCM reinit function must be set"));
+				exf->reinit(swe->sw_kschedule, iv);
+			}
 			for (i = 0; i < crde->crd_len; i += blksz) {
 				len = MIN(crde->crd_len - i, blksz);
 				if (len < blksz)
@@ -799,6 +854,9 @@ swcr_newsession(device_t dev, crypto_session_t cses, struct cryptoini *cri)
 		case CRYPTO_AES_NIST_GCM_16:
 			txf = &enc_xform_aes_nist_gcm;
 			goto enccommon;
+		case CRYPTO_AES_CCM_16:
+			txf = &enc_xform_ccm;
+			goto enccommon;
 		case CRYPTO_AES_NIST_GMAC:
 			txf = &enc_xform_aes_nist_gmac;
 			swd->sw_exf = txf;
@@ -943,6 +1001,22 @@ swcr_newsession(device_t dev, crypto_session_t cses, struct cryptoini *cri)
 			swd->sw_axf = axf;
 			break;
 
+		case CRYPTO_AES_CCM_CBC_MAC:
+			switch (cri->cri_klen) {
+			case 128:
+				axf = &auth_hash_ccm_cbc_mac_128;
+				break;
+			case 192:
+				axf = &auth_hash_ccm_cbc_mac_192;
+				break;
+			case 256:
+				axf = &auth_hash_ccm_cbc_mac_256;
+				break;
+			default:
+				swcr_freesession(dev, cses);
+				return EINVAL;
+			}
+			goto auth4common;
 		case CRYPTO_AES_128_NIST_GMAC:
 			axf = &auth_hash_nist_gmac_aes_128;
 			goto auth4common;
@@ -1042,6 +1116,7 @@ swcr_freesession(device_t dev, crypto_session_t cses)
 		case CRYPTO_CAMELLIA_CBC:
 		case CRYPTO_NULL_CBC:
 		case CRYPTO_CHACHA20:
+		case CRYPTO_AES_CCM_16:
 			txf = swd->sw_exf;
 
 			if (swd->sw_kschedule)
@@ -1056,6 +1131,7 @@ swcr_freesession(device_t dev, crypto_session_t cses)
 		case CRYPTO_SHA2_512_HMAC:
 		case CRYPTO_RIPEMD160_HMAC:
 		case CRYPTO_NULL_HMAC:
+		case CRYPTO_AES_CCM_CBC_MAC:
 			axf = swd->sw_axf;
 
 			if (swd->sw_ictx) {
@@ -1201,6 +1277,8 @@ swcr_process(device_t dev, struct cryptop *crp, int hint)
 		case CRYPTO_AES_128_NIST_GMAC:
 		case CRYPTO_AES_192_NIST_GMAC:
 		case CRYPTO_AES_256_NIST_GMAC:
+		case CRYPTO_AES_CCM_16:
+		case CRYPTO_AES_CCM_CBC_MAC:
 			crp->crp_etype = swcr_authenc(crp);
 			goto done;
 
@@ -1291,6 +1369,8 @@ swcr_attach(device_t dev)
 	REGISTER(CRYPTO_BLAKE2B);
 	REGISTER(CRYPTO_BLAKE2S);
 	REGISTER(CRYPTO_CHACHA20);
+	REGISTER(CRYPTO_AES_CCM_16);
+	REGISTER(CRYPTO_AES_CCM_CBC_MAC);
 	REGISTER(CRYPTO_POLY1305);
 #undef REGISTER
 

sys/opencrypto/xform_aes_icm.c

@@ -57,6 +57,7 @@ static	void aes_icm_crypt(caddr_t, u_int8_t *);
 static	void aes_icm_zerokey(u_int8_t **);
 static	void aes_icm_reinit(caddr_t, u_int8_t *);
 static	void aes_gcm_reinit(caddr_t, u_int8_t *);
+static	void aes_ccm_reinit(caddr_t, u_int8_t *);
 
 /* Encryption instances */
 struct enc_xform enc_xform_aes_icm = {
@@ -79,6 +80,18 @@ struct enc_xform enc_xform_aes_nist_gcm = {
 	aes_gcm_reinit,
 };
 
+struct enc_xform enc_xform_ccm = {
+	.type = CRYPTO_AES_CCM_16,
+	.name = "AES-CCM",
+	.blocksize = AES_ICM_BLOCK_LEN, .ivsize = AES_CCM_IV_LEN,
+	.minkey = AES_MIN_KEY, .maxkey = AES_MAX_KEY,
+	.encrypt = aes_icm_crypt,
+	.decrypt = aes_icm_crypt,
+	.setkey = aes_icm_setkey,
+	.zerokey = aes_icm_zerokey,
+	.reinit = aes_ccm_reinit,
+};
+
 /*
  * Encryption wrapper routines.
  */
@@ -104,6 +117,21 @@ aes_gcm_reinit(caddr_t key, u_int8_t *iv)
 	ctx->ac_block[AESICM_BLOCKSIZE - 1] = 2;
 }
 
+static void
+aes_ccm_reinit(caddr_t key, u_int8_t *iv)
+{
+	struct aes_icm_ctx *ctx;
+
+	ctx = (struct aes_icm_ctx*)key;
+
+	/* CCM has flags, then the IV, then the counter, which starts at 1 */
+	bzero(ctx->ac_block, sizeof(ctx->ac_block));
+	/* 3 bytes for length field; this gives a nonce of 12 bytes */
+	ctx->ac_block[0] = (15 - AES_CCM_IV_LEN) - 1;
+	bcopy(iv, ctx->ac_block+1, AES_CCM_IV_LEN);
+	ctx->ac_block[AESICM_BLOCKSIZE - 1] = 1;
+}
+
 static void
 aes_icm_crypt(caddr_t key, u_int8_t *data)
 {

sys/opencrypto/xform_auth.h

@@ -42,6 +42,7 @@
 #include <crypto/sha2/sha512.h>
 #include <opencrypto/rmd160.h>
 #include <opencrypto/gmac.h>
+#include <opencrypto/cbc_mac.h>
 
 #include <opencrypto/cryptodev.h>
 #include <opencrypto/xform_userland.h>
@@ -85,6 +86,9 @@ extern struct auth_hash auth_hash_nist_gmac_aes_256;
 extern struct auth_hash auth_hash_blake2b;
 extern struct auth_hash auth_hash_blake2s;
 extern struct auth_hash auth_hash_poly1305;
+extern struct auth_hash auth_hash_ccm_cbc_mac_128;
+extern struct auth_hash auth_hash_ccm_cbc_mac_192;
+extern struct auth_hash auth_hash_ccm_cbc_mac_256;
 
 union authctx {
 	MD5_CTX md5ctx;
@@ -95,6 +99,7 @@ union authctx {
 	SHA384_CTX sha384ctx;
 	SHA512_CTX sha512ctx;
 	struct aes_gmac_ctx aes_gmac_ctx;
+	struct aes_cbc_mac_ctx aes_cbc_mac_ctx;
 };
 
 #endif /* _CRYPTO_XFORM_AUTH_H_ */

sys/opencrypto/xform_cbc_mac.c

@@ -0,0 +1,55 @@
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <opencrypto/cbc_mac.h>
+#include <opencrypto/xform_auth.h>
+
+/* Authentication instances */
+struct auth_hash auth_hash_ccm_cbc_mac_128 = {
+	.type = CRYPTO_AES_CCM_CBC_MAC,
+	.name = "CBC-CCM-AES-128",
+	.keysize = AES_128_CBC_MAC_KEY_LEN,
+	.hashsize = AES_CBC_MAC_HASH_LEN,
+	.ctxsize = sizeof(struct aes_cbc_mac_ctx),
+	.blocksize = CCM_CBC_BLOCK_LEN,
+	.Init = (void (*)(void *)) AES_CBC_MAC_Init,
+	.Setkey =
+	    (void (*)(void *, const u_int8_t *, u_int16_t))AES_CBC_MAC_Setkey,
+	.Reinit =
+	    (void (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Reinit,
+	.Update =
+	    (int  (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Update,
+	.Final = (void (*)(u_int8_t *, void *)) AES_CBC_MAC_Final,
+};
+struct auth_hash auth_hash_ccm_cbc_mac_192 = {
+	.type = CRYPTO_AES_CCM_CBC_MAC,
+	.name = "CBC-CCM-AES-192",
+	.keysize = AES_192_CBC_MAC_KEY_LEN,
+	.hashsize = AES_CBC_MAC_HASH_LEN,
+	.ctxsize = sizeof(struct aes_cbc_mac_ctx),
+	.blocksize = CCM_CBC_BLOCK_LEN,
+	.Init = (void (*)(void *)) AES_CBC_MAC_Init,
+	.Setkey =
+	    (void (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Setkey,
+	.Reinit =
+	    (void (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Reinit,
+	.Update =
+	    (int  (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Update,
+	.Final = (void (*)(u_int8_t *, void *)) AES_CBC_MAC_Final,
+};
+struct auth_hash auth_hash_ccm_cbc_mac_256 = {
+	.type = CRYPTO_AES_CCM_CBC_MAC,
+	.name = "CBC-CCM-AES-256",
+	.keysize = AES_256_CBC_MAC_KEY_LEN,
+	.hashsize = AES_CBC_MAC_HASH_LEN,
+	.ctxsize = sizeof(struct aes_cbc_mac_ctx),
+	.blocksize = CCM_CBC_BLOCK_LEN,
+	.Init = (void (*)(void *)) AES_CBC_MAC_Init,
+	.Setkey =
+	    (void (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Setkey,
+	.Reinit =
+	    (void (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Reinit,
+	.Update =
+	    (int  (*)(void *, const u_int8_t *, u_int16_t)) AES_CBC_MAC_Update,
+	.Final = (void (*)(u_int8_t *, void *)) AES_CBC_MAC_Final,
+};

sys/opencrypto/xform_enc.h

@@ -84,6 +84,7 @@ extern struct enc_xform enc_xform_aes_xts;
 extern struct enc_xform enc_xform_arc4;
 extern struct enc_xform enc_xform_camellia;
 extern struct enc_xform enc_xform_chacha20;
+extern struct enc_xform enc_xform_ccm;
 
 struct aes_icm_ctx {
 	u_int32_t	ac_ek[4*(RIJNDAEL_MAXNR + 1)];

sys/powerpc/booke/pmap.c

@@ -2973,14 +2973,19 @@ mmu_booke_zero_page_area(mmu_t mmu, vm_page_t m, int off, int size)
 
 	/* XXX KASSERT off and size are within a single page? */
 
-	mtx_lock(&zero_page_mutex);
+	if (hw_direct_map) {
-	va = zero_page_va;
+		va = PHYS_TO_DMAP(VM_PAGE_TO_PHYS(m));
+		bzero((caddr_t)va + off, size);
+	} else {
+		mtx_lock(&zero_page_mutex);
+		va = zero_page_va;
 
-	mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m));
+		mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m));
-	bzero((caddr_t)va + off, size);
+		bzero((caddr_t)va + off, size);
-	mmu_booke_kremove(mmu, va);
+		mmu_booke_kremove(mmu, va);
 
-	mtx_unlock(&zero_page_mutex);
+		mtx_unlock(&zero_page_mutex);
+	}
 }
 
 /*
@@ -2991,15 +2996,23 @@ mmu_booke_zero_page(mmu_t mmu, vm_page_t m)
 {
 	vm_offset_t off, va;
 
-	mtx_lock(&zero_page_mutex);
+	if (hw_direct_map) {
-	va = zero_page_va;
+		va = PHYS_TO_DMAP(VM_PAGE_TO_PHYS(m));
+	} else {
+		va = zero_page_va;
+		mtx_lock(&zero_page_mutex);
+
+		mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m));
+	}
 
-	mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m));
 	for (off = 0; off < PAGE_SIZE; off += cacheline_size)
 		__asm __volatile("dcbz 0,%0" :: "r"(va + off));
-	mmu_booke_kremove(mmu, va);
 
-	mtx_unlock(&zero_page_mutex);
+	if (!hw_direct_map) {
+		mmu_booke_kremove(mmu, va);
+
+		mtx_unlock(&zero_page_mutex);
+	}
 }
 
 /*
@@ -3015,13 +3028,20 @@ mmu_booke_copy_page(mmu_t mmu, vm_page_t sm, vm_page_t dm)
 	sva = copy_page_src_va;
 	dva = copy_page_dst_va;
 
-	mtx_lock(&copy_page_mutex);
+	if (hw_direct_map) {
-	mmu_booke_kenter(mmu, sva, VM_PAGE_TO_PHYS(sm));
+		sva = PHYS_TO_DMAP(VM_PAGE_TO_PHYS(sm));
-	mmu_booke_kenter(mmu, dva, VM_PAGE_TO_PHYS(dm));
+		dva = PHYS_TO_DMAP(VM_PAGE_TO_PHYS(dm));
+	} else {
+		mtx_lock(&copy_page_mutex);
+		mmu_booke_kenter(mmu, sva, VM_PAGE_TO_PHYS(sm));
+		mmu_booke_kenter(mmu, dva, VM_PAGE_TO_PHYS(dm));
+	}
 	memcpy((caddr_t)dva, (caddr_t)sva, PAGE_SIZE);
-	mmu_booke_kremove(mmu, dva);
+	if (!hw_direct_map) {
-	mmu_booke_kremove(mmu, sva);
+		mmu_booke_kremove(mmu, dva);
-	mtx_unlock(&copy_page_mutex);
+		mmu_booke_kremove(mmu, sva);
+		mtx_unlock(&copy_page_mutex);
+	}
 }
 
 static inline void
@@ -3032,26 +3052,34 @@ mmu_booke_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset,
 	vm_offset_t a_pg_offset, b_pg_offset;
 	int cnt;
 
-	mtx_lock(&copy_page_mutex);
+	if (hw_direct_map) {
-	while (xfersize > 0) {
+		a_cp = (caddr_t)((uintptr_t)PHYS_TO_DMAP(VM_PAGE_TO_PHYS(*ma)) +
-		a_pg_offset = a_offset & PAGE_MASK;
+		    a_offset);
-		cnt = min(xfersize, PAGE_SIZE - a_pg_offset);
+		b_cp = (caddr_t)((uintptr_t)PHYS_TO_DMAP(VM_PAGE_TO_PHYS(*mb)) +
-		mmu_booke_kenter(mmu, copy_page_src_va,
+		    b_offset);
-		    VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT]));
+		bcopy(a_cp, b_cp, xfersize);
-		a_cp = (char *)copy_page_src_va + a_pg_offset;
+	} else {
-		b_pg_offset = b_offset & PAGE_MASK;
+		mtx_lock(&copy_page_mutex);
-		cnt = min(cnt, PAGE_SIZE - b_pg_offset);
+		while (xfersize > 0) {
-		mmu_booke_kenter(mmu, copy_page_dst_va,
+			a_pg_offset = a_offset & PAGE_MASK;
-		    VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT]));
+			cnt = min(xfersize, PAGE_SIZE - a_pg_offset);
-		b_cp = (char *)copy_page_dst_va + b_pg_offset;
+			mmu_booke_kenter(mmu, copy_page_src_va,
-		bcopy(a_cp, b_cp, cnt);
+			    VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT]));
-		mmu_booke_kremove(mmu, copy_page_dst_va);
+			a_cp = (char *)copy_page_src_va + a_pg_offset;
-		mmu_booke_kremove(mmu, copy_page_src_va);
+			b_pg_offset = b_offset & PAGE_MASK;
-		a_offset += cnt;
+			cnt = min(cnt, PAGE_SIZE - b_pg_offset);
-		b_offset += cnt;
+			mmu_booke_kenter(mmu, copy_page_dst_va,
-		xfersize -= cnt;
+			    VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT]));
+			b_cp = (char *)copy_page_dst_va + b_pg_offset;
+			bcopy(a_cp, b_cp, cnt);
+			mmu_booke_kremove(mmu, copy_page_dst_va);
+			mmu_booke_kremove(mmu, copy_page_src_va);
+			a_offset += cnt;
+			b_offset += cnt;
+			xfersize -= cnt;
+		}
+		mtx_unlock(&copy_page_mutex);
 	}
-	mtx_unlock(&copy_page_mutex);
 }
 
 static vm_offset_t
@@ -3064,6 +3092,9 @@ mmu_booke_quick_enter_page(mmu_t mmu, vm_page_t m)
 
 	paddr = VM_PAGE_TO_PHYS(m);
 
+	if (hw_direct_map)
+		return (PHYS_TO_DMAP(paddr));
+
 	flags = PTE_SR | PTE_SW | PTE_SX | PTE_WIRED | PTE_VALID;
 	flags |= tlb_calc_wimg(paddr, pmap_page_get_memattr(m)) << PTE_MAS2_SHIFT;
 	flags |= PTE_PS_4KB;
@@ -3097,6 +3128,9 @@ mmu_booke_quick_remove_page(mmu_t mmu, vm_offset_t addr)
 {
 	pte_t *pte;
 
+	if (hw_direct_map)
+		return;
+
 	pte = pte_find(mmu, kernel_pmap, addr);
 
 	KASSERT(PCPU_GET(qmap_addr) == addr,
@@ -3880,29 +3914,23 @@ tlb1_write_entry_int(void *arg)
 	mas0 = MAS0_TLBSEL(1) | MAS0_ESEL(args->idx);
 
 	mtspr(SPR_MAS0, mas0);
-	__asm __volatile("isync");
 	mtspr(SPR_MAS1, args->e->mas1);
-	__asm __volatile("isync");
 	mtspr(SPR_MAS2, args->e->mas2);
-	__asm __volatile("isync");
 	mtspr(SPR_MAS3, args->e->mas3);
-	__asm __volatile("isync");
 	switch ((mfpvr() >> 16) & 0xFFFF) {
 	case FSL_E500mc:
 	case FSL_E5500:
 	case FSL_E6500:
 		mtspr(SPR_MAS8, 0);
-		__asm __volatile("isync");
 		/* FALLTHROUGH */
 	case FSL_E500v2:
 		mtspr(SPR_MAS7, args->e->mas7);
-		__asm __volatile("isync");
 		break;
 	default:
 		break;
 	}
 
-	__asm __volatile("tlbwe; isync; msync");
+	__asm __volatile("isync; tlbwe; isync; msync");
 
 }
 
@@ -4325,12 +4353,26 @@ tid_flush(tlbtid_t tid)
 	msr = mfmsr();
 	__asm __volatile("wrteei 0");
 
+	/*
+	 * Newer (e500mc and later) have tlbilx, which doesn't broadcast, so use
+	 * it for PID invalidation.
+	 */
+	switch ((mfpvr() >> 16) & 0xffff) {
+	case FSL_E500mc:
+	case FSL_E5500:
+	case FSL_E6500:
+		mtspr(SPR_MAS6, tid << MAS6_SPID0_SHIFT);
+		/* tlbilxpid */
+		__asm __volatile("isync; .long 0x7c000024; isync; msync");
+		mtmsr(msr);
+		return;
+	}
+
 	for (way = 0; way < TLB0_WAYS; way++)
 		for (entry = 0; entry < TLB0_ENTRIES_PER_WAY; entry++) {
 
 			mas0 = MAS0_TLBSEL(0) | MAS0_ESEL(way);
 			mtspr(SPR_MAS0, mas0);
-			__asm __volatile("isync");
 
 			mas2 = entry << MAS2_TLB0_ENTRY_IDX_SHIFT;
 			mtspr(SPR_MAS2, mas2);
@@ -4407,7 +4449,6 @@ DB_SHOW_COMMAND(tlb0, tlb0_print_tlbentries)
 
 			mas0 = MAS0_TLBSEL(0) | MAS0_ESEL(way);
 			mtspr(SPR_MAS0, mas0);
-			__asm __volatile("isync");
 
 			mas2 = entryidx << MAS2_TLB0_ENTRY_IDX_SHIFT;
 			mtspr(SPR_MAS2, mas2);

sys/powerpc/powerpc/elf32_machdep.c

@@ -115,7 +115,7 @@ struct sysentvec elf32_freebsd_sysvec = {
 	.sv_fixlimit	= NULL,
 #endif
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_ABI_FREEBSD | SV_ILP32 | SV_SHP,
+	.sv_flags	= SV_ABI_FREEBSD | SV_ILP32 | SV_SHP | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_shared_page_base = FREEBSD32_SHAREDPAGE,

sys/powerpc/powerpc/elf64_machdep.c

@@ -79,7 +79,7 @@ struct sysentvec elf64_freebsd_sysvec_v1 = {
 	.sv_setregs	= exec_setregs_funcdesc,
 	.sv_fixlimit	= NULL,
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_ABI_FREEBSD | SV_LP64 | SV_SHP,
+	.sv_flags	= SV_ABI_FREEBSD | SV_LP64 | SV_SHP | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,

sys/powerpc/powerpc/exec_machdep.c

@@ -474,6 +474,10 @@ set_mcontext(struct thread *td, mcontext_t *mcp)
 	else
 		tf->fixreg[2] = tls;
 
+	/* Disable FPU */
+	tf->srr1 &= ~PSL_FP;
+	pcb->pcb_flags &= ~PCB_FPU;
+
 	if (mcp->mc_flags & _MC_FP_VALID) {
 		/* enable_fpu() will happen lazily on a fault */
 		pcb->pcb_flags |= PCB_FPREGS;

sys/riscv/include/param.h

@@ -82,7 +82,7 @@
 #define	PAGE_SIZE	(1 << PAGE_SHIFT)	/* Page size */
 #define	PAGE_MASK	(PAGE_SIZE - 1)
 
-#define	MAXPAGESIZES	1		/* maximum number of supported page sizes */
+#define	MAXPAGESIZES	3	/* maximum number of supported page sizes */
 
 #ifndef KSTACK_PAGES
 #define	KSTACK_PAGES	4	/* pages of kernel stack (with pcb) */

sys/riscv/include/pcb.h

@@ -55,7 +55,6 @@ struct pcb {
 #define	PCB_FP_STARTED	0x1
 #define	PCB_FP_USERMASK	0x1
 	uint64_t	pcb_sepc;	/* Supervisor exception pc */
-	vm_offset_t	pcb_l1addr;	/* L1 page tables base address */
 	vm_offset_t	pcb_onfault;	/* Copyinout fault handler */
 };
 

sys/riscv/include/pcpu.h

@@ -45,6 +45,7 @@
 #define	ALT_STACK_SIZE	128
 
 #define	PCPU_MD_FIELDS							\
+	struct pmap *pc_curpmap;	/* Currently active pmap */	\
 	uint32_t pc_pending_ipis;	/* IPIs pending to this CPU */	\
 	char __pad[61]
 

sys/riscv/include/pmap.h

@@ -41,9 +41,12 @@
 #ifndef LOCORE
 
 #include <sys/queue.h>
+#include <sys/_cpuset.h>
 #include <sys/_lock.h>
 #include <sys/_mutex.h>
 
+#include <vm/_vm_radix.h>
+
 #ifdef _KERNEL
 
 #define	vtophys(va)	pmap_kextract((vm_offset_t)(va))
@@ -78,8 +81,11 @@ struct pmap {
 	struct mtx		pm_mtx;
 	struct pmap_statistics	pm_stats;	/* pmap statictics */
 	pd_entry_t		*pm_l1;
+	u_long			pm_satp;	/* value for SATP register */
+	cpuset_t		pm_active;	/* active on cpus */
 	TAILQ_HEAD(,pv_chunk)	pm_pvchunk;	/* list of mappings in pmap */
 	LIST_ENTRY(pmap)	pm_list;	/* List of all pmaps */
+	struct vm_radix		pm_root;
 };
 
 typedef struct pv_entry {
@@ -134,11 +140,16 @@ extern vm_offset_t virtual_end;
 #define	L1_MAPPABLE_P(va, pa, size)					\
 	((((va) | (pa)) & L1_OFFSET) == 0 && (size) >= L1_SIZE)
 
+struct thread;
+
+void	pmap_activate_boot(pmap_t);
+void	pmap_activate_sw(struct thread *);
 void	pmap_bootstrap(vm_offset_t, vm_paddr_t, vm_size_t);
 void	pmap_kenter_device(vm_offset_t, vm_size_t, vm_paddr_t);
 vm_paddr_t pmap_kextract(vm_offset_t va);
 void	pmap_kremove(vm_offset_t);
 void	pmap_kremove_device(vm_offset_t, vm_size_t);
+bool	pmap_ps_enabled(pmap_t);
 
 void	*pmap_mapdev(vm_offset_t, vm_size_t);
 void	*pmap_mapbios(vm_paddr_t, vm_size_t);

sys/riscv/include/pte.h

@@ -62,7 +62,8 @@ typedef	uint64_t	pn_t;			/* page number */
 #define	L3_SIZE 	(1 << L3_SHIFT)
 #define	L3_OFFSET 	(L3_SIZE - 1)
 
-#define	Ln_ENTRIES	(1 << 9)
+#define	Ln_ENTRIES_SHIFT 9
+#define	Ln_ENTRIES	(1 << Ln_ENTRIES_SHIFT)
 #define	Ln_ADDR_MASK	(Ln_ENTRIES - 1)
 
 /* Bits 9:8 are reserved for software */
@@ -79,6 +80,8 @@ typedef	uint64_t	pn_t;			/* page number */
 #define	PTE_RWX		(PTE_R | PTE_W | PTE_X)
 #define	PTE_RX		(PTE_R | PTE_X)
 #define	PTE_KERN	(PTE_V | PTE_R | PTE_W | PTE_A | PTE_D)
+#define	PTE_PROMOTE	(PTE_V | PTE_RWX | PTE_D | PTE_A | PTE_G | PTE_U | \
+			 PTE_SW_MANAGED | PTE_SW_WIRED)
 
 #define	PTE_PPN0_S	10
 #define	PTE_PPN1_S	19

sys/riscv/include/vmparam.h

@@ -99,10 +99,10 @@
 #define	VM_NFREEORDER		12
 
 /*
- * Disable superpage reservations.
+ * Enable superpage reservations: 1 level.
  */
 #ifndef	VM_NRESERVLEVEL
-#define	VM_NRESERVLEVEL		0
+#define	VM_NRESERVLEVEL		1
 #endif
 
 /*

sys/riscv/riscv/elf_machdep.c

@@ -82,7 +82,7 @@ struct sysentvec elf64_freebsd_sysvec = {
 	.sv_setregs	= exec_setregs,
 	.sv_fixlimit	= NULL,
 	.sv_maxssiz	= NULL,
-	.sv_flags	= SV_ABI_FREEBSD | SV_LP64 | SV_SHP,
+	.sv_flags	= SV_ABI_FREEBSD | SV_LP64 | SV_SHP | SV_ASLR,
 	.sv_set_syscall_retval = cpu_set_syscall_retval,
 	.sv_fetch_syscall_args = cpu_fetch_syscall_args,
 	.sv_syscallnames = syscallnames,

sys/riscv/riscv/genassym.c

@@ -63,7 +63,6 @@ ASSYM(TDF_ASTPENDING, TDF_ASTPENDING);
 ASSYM(TDF_NEEDRESCHED, TDF_NEEDRESCHED);
 
 ASSYM(PCB_ONFAULT, offsetof(struct pcb, pcb_onfault));
-ASSYM(PCB_L1ADDR, offsetof(struct pcb, pcb_l1addr));
 ASSYM(PCB_SIZE, sizeof(struct pcb));
 ASSYM(PCB_RA, offsetof(struct pcb, pcb_ra));
 ASSYM(PCB_SP, offsetof(struct pcb, pcb_sp));