Merge fcbc54cd31 into f2233ac33a

sysctl(8) prints a newline after the description, no need for this extra
newline.

MFC after:	1 week

ObsoleteFiles.inc

@@ -51,6 +51,13 @@
 #   xargs -n1 | sort | uniq -d;
 # done
 
+# 20241124: library and tests of OpenBSD dc
+OLD_FILES+=usr/share/misc/bc.library
+OLD_FILES+=usr/tests/usr.bin/dc/Kyuafile
+OLD_FILES+=usr/tests/usr.bin/dc/bcode
+OLD_FILES+=usr/tests/usr.bin/dc/inout
+OLD_DIRS+=usr/tests/usr.bin/dc
+
 # 20241119: rewrite mv tests
 OLD_FILES+=usr/tests/bin/mv/legacy_test
 

RELNOTES

@@ -10,6 +10,11 @@ newline.  Entries should be separated by a newline.
 
 Changes to this file should not be MFCed.
 
+b2f7c53430c3:
+	Kernel TLS is now enabled by default in kernels including KTLS
+	support.  KTLS is included in GENERIC kernels for aarch64,
+	amd64, powerpc64, and powerpc64le.
+
 f57efe95cc25:
 	New mididump(1) utility which dumps MIDI 1.0 events in real time.
 

UPDATING

@@ -27,6 +27,11 @@ NOTE TO PEOPLE WHO THINK THAT FreeBSD 15.x IS SLOW:
 	world, or to merely disable the most expensive debugging functionality
 	at runtime, run "ln -s 'abort:false,junk:false' /etc/malloc.conf".)
 
+20241124:
+	The OpenBSD derived bc and dc implementations and the WITHOUT_GH_BC
+	option that allowed building them instead of the advanced version
+	imported more than 4 years ago have been removed.
+
 20241025:
 	The support for the rc_fast_and_loose variable has been removed from
 	rc.subr(8). Users setting rc_fast_and_loose on their systems are

etc/mtree/BSD.tests.dist

@@ -505,6 +505,8 @@
         ..
         route
         ..
+        savecore
+        ..
         sysctl
         ..
     ..
@@ -1089,8 +1091,6 @@
         ..
         cut
         ..
-        dc
-        ..
         diff
         ..
         diff3

lib/libc/gen/getentropy.c

@@ -31,6 +31,7 @@
 
 #include <errno.h>
 #include <signal.h>
+#include <unistd.h>
 #include <ssp/ssp.h>
 
 #include "libc_private.h"

lib/libsys/fhreadlink.2

@@ -60,7 +60,7 @@ code in the global variable
 .Va errno .
 .Sh ERRORS
 The
-.Fn readlink
+.Fn fhreadlink
 system call
 will fail if:
 .Bl -tag -width Er
@@ -87,7 +87,8 @@ is no longer valid
 .El
 .Sh SEE ALSO
 .Xr fhlink 2 ,
-.Xr fhstat 2
+.Xr fhstat 2 ,
+.Xr readlink 2
 .Sh HISTORY
 The
 .Fn fhreadlink

lib/libsys/getsockopt.2

@@ -25,7 +25,7 @@
 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 .\" SUCH DAMAGE.
 .\"
-.Dd July 8, 2024
+.Dd November 25, 2024
 .Dt GETSOCKOPT 2
 .Os
 .Sh NAME
@@ -568,9 +568,14 @@ struct so_splice {
 .Pp
 Data received on
 .Fa s
-will automatically be transmitted from the socket specified in
+will automatically be transmitted via the socket specified in
 .Fa sp_fd
 without any intervention by userspace.
+That is, the data will be transmitted via
+.Fa sp_fd
+as if userspace had called
+.Xr send 2
+directly.
 Splicing is a one-way operation; a given pair of sockets may be
 spliced in one or both directions.
 Currently only connected

lib/libsys/readlink.2

@@ -138,6 +138,7 @@ is neither
 nor a file descriptor associated with a directory.
 .El
 .Sh SEE ALSO
+.Xr fhreadlink 2 ,
 .Xr lstat 2 ,
 .Xr stat 2 ,
 .Xr symlink 2 ,

libexec/rc/rc.conf

@@ -263,8 +263,8 @@ icmp_log_redirect="NO"		# Set to YES to log ICMP REDIRECT packets
 network_interfaces="auto"	# List of network interfaces (or "auto").
 cloned_interfaces=""		# List of cloned network interfaces to create.
 #cloned_interfaces="gif0 gif1 gif2 gif3" # Pre-cloning GENERIC config.
-#ifconfig_lo0="inet 127.0.0.1"	# default loopback device configuration.
-#ifconfig_lo0_alias0="inet 127.0.0.254 netmask 0xffffffff" # Sample alias entry.
+#ifconfig_lo0="inet 127.0.0.1/8"	# default loopback device configuration.
+#ifconfig_lo0_alias0="inet 127.0.0.254/32" # Sample alias entry.
 #ifconfig_em0_ipv6="inet6 2001:db8:1::1 prefixlen 64" # Sample IPv6 addr entry
 #ifconfig_em0_alias0="inet6 2001:db8:2::1 prefixlen 64" # Sample IPv6 alias
 #ifconfig_em0_name="net0"	# Change interface name from em0 to net0.

libexec/rc/rc.d/var_run

@@ -27,7 +27,7 @@ _var_run_load() {
 
 _var_run_save() {
 	if [ ! -d $(dirname ${var_run_mtree}) ]; then
-		mkdir -p ${var_run_mtree}
+		mkdir -p $(dirname ${var_run_mtree})
 	fi
 	mtree -dcbj -p /var/run > ${var_run_mtree}
 }

sbin/ffsinfo/ffsinfo.8

@@ -36,7 +36,7 @@
 .\"
 .\" $TSHeader: src/sbin/ffsinfo/ffsinfo.8,v 1.3 2000/12/12 19:30:55 tomsoft Exp $
 .\"
-.Dd September 8, 2000
+.Dd November 19, 2024
 .Dt FFSINFO 8
 .Os
 .Sh NAME
@@ -120,6 +120,7 @@ to
 .Pa /var/tmp/ffsinfo
 with all available information.
 .Sh SEE ALSO
+.Xr ffs 4 ,
 .Xr dumpfs 8 ,
 .Xr fsck 8 ,
 .Xr gpart 8 ,

sbin/fsck_msdosfs/fsck_msdosfs.8

@@ -28,7 +28,7 @@
 .Os
 .Sh NAME
 .Nm fsck_msdosfs
-.Nd DOS/Windows (FAT) file system consistency checker
+.Nd File Allocation Table (FAT) file system consistency checker
 .Sh SYNOPSIS
 .Nm
 .Fl p
@@ -121,9 +121,8 @@ to assume
 as the answer to all operator questions.
 .El
 .Sh SEE ALSO
-.Xr fsck 8 ,
-.Xr fsck_ffs 8 ,
-.Xr mount_msdosfs 8
+.Xr msdosfs 4 ,
+.Xr fsck 8
 .Sh HISTORY
 The
 .Nm

sbin/mount/mount.8

@@ -25,7 +25,7 @@
 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 .\" SUCH DAMAGE.
 .\"
-.Dd January 24, 2024
+.Dd November 19, 2024
 .Dt MOUNT 8
 .Os
 .Sh NAME
@@ -568,6 +568,7 @@ support for a particular file system might be provided either on a static
 .Xr cd9660 4 ,
 .Xr devfs 4 ,
 .Xr ext2fs 4 ,
+.Xr ffs 4 ,
 .Xr mac 4 ,
 .Xr procfs 4 ,
 .Xr tarfs 4 ,

sbin/mount_msdosfs/mount_msdosfs.8

@@ -33,7 +33,7 @@
 .Os
 .Sh NAME
 .Nm mount_msdosfs
-.Nd mount an MS-DOS file system
+.Nd mount a File Allocation Table (FAT) file system
 .Sh SYNOPSIS
 .Nm
 .Op Fl 9ls

sbin/newfs/newfs.8

@@ -25,7 +25,7 @@
 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 .\" SUCH DAMAGE.
 .\"
-.Dd May 18, 2024
+.Dd November 19, 2024
 .Dt NEWFS 8
 .Os
 .Sh NAME
@@ -350,6 +350,7 @@ than the historical defaults
 This large fragment size may lead to much wasted space
 on file systems that contain many small files.
 .Sh SEE ALSO
+.Xr ffs 4 ,
 .Xr geom 4 ,
 .Xr disktab 5 ,
 .Xr fs 5 ,

sbin/newfs_msdos/newfs_msdos.8

@@ -28,7 +28,7 @@
 .Os
 .Sh NAME
 .Nm newfs_msdos
-.Nd construct a new MS-DOS (FAT) file system
+.Nd construct a new File Allocation Table (FAT) file system
 .Sh SYNOPSIS
 .Nm
 .Op Fl N
@@ -252,6 +252,7 @@ Create a 30MB image file, with the FAT partition starting
 newfs_msdos -C 30M -@63s ./somefile
 .Ed
 .Sh SEE ALSO
+.Xr msdosfs 4 ,
 .Xr gpart 8 ,
 .Xr newfs 8
 .Sh HISTORY

sbin/pfctl/pfctl.8

@@ -24,7 +24,7 @@
 .\" (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 .\" THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 .\"
-.Dd July 23, 2024
+.Dd November 20, 2024
 .Dt PFCTL 8
 .Os
 .Sh NAME
@@ -506,8 +506,8 @@ Automatically create a nonexisting table.
 Show the content (addresses) of a table.
 .It Fl T Cm test
 Test if the given addresses match a table.
-.It Fl T Cm zero
-Clear all the statistics of a table.
+.It Fl T Cm zero Op Ar address ...
+Clear all the statistics of a table, or only for specified addresses.
 .It Fl T Cm load
 Load only the table definitions from
 .Xr pf.conf 5 .

sbin/pfctl/pfctl.h

@@ -60,6 +60,7 @@ int	 pfr_del_tables(struct pfr_table *, int, int *, int);
 int	 pfr_get_tables(struct pfr_table *, struct pfr_table *, int *, int);
 int	 pfr_get_tstats(struct pfr_table *, struct pfr_tstats *, int *, int);
 int	 pfr_clr_tstats(struct pfr_table *, int, int *, int);
+int	 pfr_clr_astats(struct pfr_table *, struct pfr_addr *, int, int *, int);
 int	 pfr_clr_addrs(struct pfr_table *, int *, int);
 int	 pfr_add_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int);
 int	 pfr_del_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int);

sbin/pfctl/pfctl_radix.c

@@ -286,6 +286,29 @@ pfr_get_astats(struct pfr_table *tbl, struct pfr_astats *addr, int *size,
 	return (0);
 }
 
+int
+pfr_clr_astats(struct pfr_table *tbl, struct pfr_addr *addr, int size,
+    int *nzero, int flags)
+{
+	struct pfioc_table io;
+
+	if (size < 0 || (size && !tbl) || addr == NULL) {
+		errno = EINVAL;
+		return (-1);
+	}
+	bzero(&io, sizeof io);
+	io.pfrio_flags = flags;
+	io.pfrio_table = *tbl;
+	io.pfrio_buffer = addr;
+	io.pfrio_esize = sizeof(*addr);
+	io.pfrio_size = size;
+	if (ioctl(dev, DIOCRCLRASTATS, &io) == -1)
+		return (-1);
+	if (nzero)
+		*nzero = io.pfrio_nzero;
+	return (0);
+}
+
 int
 pfr_clr_tstats(struct pfr_table *tbl, int size, int *nzero, int flags)
 {

sbin/pfctl/pfctl_table.c

@@ -344,9 +344,22 @@ pfctl_table(int argc, char *argv[], char *tname, const char *command,
 		}
 		if (nmatch < b.pfrb_size)
 			rv = 2;
+	} else if (!strcmp(command, "zero") && (argc || file != NULL)) {
+		b.pfrb_type = PFRB_ADDRS;
+		if (load_addr(&b, argc, argv, file, 0))
+			goto _error;
+		if (opts & PF_OPT_VERBOSE)
+			flags |= PFR_FLAG_FEEDBACK;
+		RVTEST(pfr_clr_astats(&table, b.pfrb_caddr, b.pfrb_size,
+		    &nzero, flags));
+		xprintf(opts, "%d/%d addresses cleared", nzero, b.pfrb_size);
+		if (opts & PF_OPT_VERBOSE)
+			PFRB_FOREACH(a, &b)
+				if (opts & PF_OPT_VERBOSE2 ||
+				    a->pfra_fback != PFR_FB_NONE)
+					print_addrx(a, NULL,
+					    opts & PF_OPT_USEDNS);
 	} else if (!strcmp(command, "zero")) {
-		if (argc || file != NULL)
-			usage();
 		flags |= PFR_FLAG_ADDRSTOO;
 		RVTEST(pfr_clr_tstats(&table, 1, &nzero, flags));
 		xprintf(opts, "%d table/stats cleared", nzero);

sbin/savecore/Makefile

@@ -18,4 +18,7 @@ CFLAGS+=	-DWITH_CASPER
 LIBADD+=	casper cap_fileargs cap_syslog
 .endif
 
+HAS_TESTS=
+SUBDIR.${MK_TESTS}+= tests
+
 .include <bsd.prog.mk>

sbin/savecore/tests/Makefile

@@ -0,0 +1,3 @@
+ATF_TESTS_SH=	livedump_test
+
+.include <bsd.test.mk>

sbin/savecore/tests/livedump_test.sh

@@ -0,0 +1,54 @@
+#
+# SPDX-License-Identifier: BSD-2-Clause
+#
+# Copyright (c) 2024 Mark Johnston <markj@FreeBSD.org>
+#
+
+atf_test_case livedump_kldstat
+livedump_kldstat_head()
+{
+	atf_set "descr" "Test livedump integrity"
+	atf_set "require.progs" kgdb
+	atf_set "require.user" root
+}
+livedump_kldstat_body()
+{
+	atf_check savecore -L .
+
+	kernel=$(sysctl -n kern.bootfile)
+
+	if ! [ -f /usr/lib/debug/${kernel}.debug ]; then
+		atf_skip "No debug symbols for the running kernel"
+	fi
+
+	# Implement kldstat using gdb script.
+	cat >./kldstat.gdb <<'__EOF__'
+printf "Id Refs Address                Size Name\n"
+set $_lf = linker_files.tqh_first
+while ($_lf)
+    printf "%2d %4d %p %8x %s\n", $_lf->id, $_lf->refs, $_lf->address, $_lf->size, $_lf->filename
+    set $_lf = $_lf->link.tqe_next
+end
+__EOF__
+
+	# Ignore stderr since kgdb prints some warnings about inaccessible
+	# source files.
+	#
+	# Use a script to source the main gdb script, otherwise kgdb prints
+	# a bunch of line noise that is annoying to filter out.
+	echo "source ./kldstat.gdb" > ./script.gdb
+	atf_check -o save:out -e ignore \
+	    kgdb -q ${kernel} ./livecore.0 < ./script.gdb
+
+	# Get rid of gunk printed by kgdb.
+	sed -i '' -n -e 's/^(kgdb) //' -e '/^Id Refs /,$p' out
+
+	# The output of kgdb should match the output of kldstat.
+	atf_check -o save:kldstat kldstat
+	atf_check diff kldstat out
+}
+
+atf_init_test_cases()
+{
+	atf_add_test_case livedump_kldstat
+}

share/man/man4/ccr.4

@@ -22,7 +22,7 @@
 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 .\" SUCH DAMAGE.
 .\"
-.Dd June 3, 2019
+.Dd November 25, 2024
 .Dt CCR 4
 .Os
 .Sh NAME
@@ -52,7 +52,10 @@ The driver accelerates AES-CBC, AES-CCM, AES-CTR, AES-GCM, AES-XTS,
 SHA1, SHA2-224, SHA2-256, SHA2-384, SHA2-512,
 SHA1-HMAC, SHA2-224-HMAC, SHA2-256-HMAC, SHA2-384-HMAC, and SHA2-512-HMAC
 operations for
-.Xr crypto 4
+.Xr crypto 9
+consumers such as
+.Xr ktls 4 ,
+.Xr geli 4 ,
 and
 .Xr ipsec 4 .
 The driver also supports chaining one of AES-CBC, AES-CTR, or AES-XTS with
@@ -97,7 +100,11 @@ email all the specific information related to the issue to
 .Sh SEE ALSO
 .Xr crypto 4 ,
 .Xr cxgbe 4 ,
-.Xr ipsec 4
+.Xr geli 4 ,
+.Xr ipsec 4 ,
+.Xr ktls 4 ,
+.Xr crypto 7 ,
+.Xr crypto 9
 .Sh HISTORY
 The
 .Nm

share/man/man4/msdosfs.4

@@ -6,7 +6,7 @@
 .Os
 .Sh NAME
 .Nm msdosfs
-.Nd MS-DOS file system
+.Nd File Allocation Table (FAT) file system
 .Sh SYNOPSIS
 .Cd "options MSDOSFS"
 .Sh DESCRIPTION
@@ -65,8 +65,10 @@ may also be used to extract this information.
 .Sh SEE ALSO
 .Xr mount 2 ,
 .Xr unmount 2 ,
+.Xr fsck_msdosfs 8 ,
 .Xr mount 8 ,
 .Xr mount_msdosfs 8 ,
+.Xr newfs_msdos 8 ,
 .Xr umount 8
 .Sh AUTHORS
 This manual page was written by

share/man/man4/vt.4

@@ -50,6 +50,7 @@ In
 .Cd kern.vt.color.<colornum>.rgb="<colorspec>"
 .Cd kern.vt.fb.default_mode="<X>x<Y>"
 .Cd kern.vt.fb.modes.<connector>="<X>x<Y>"
+.Cd kern.vt.slow_down=<delay>"
 .Cd screen.font="<X>x<Y>"
 .Pp
 In
@@ -266,6 +267,16 @@ It will contain a list of connectors and their associated tunables.
 This is currently only supported by the
 .Cm vt_fb
 backend when it is paired with a KMS video driver.
+.It Va kern.vt.slow_down
+When debugging the kernel on modern laptops, the screen is often
+the only available console, and relevant information will scroll
+out of view before it can be captured by eye or camera.
+.Pp
+Setting
+.Va kern.vt.slow_down
+to a non-zero number will make console output synchronous (ie:
+not dependent on timers and interrupts) and slow it down in proportion
+to the number.
 .It Va screen.font
 Set this value to the base name of the desired font file located in
 .Pa /boot/fonts .

share/man/man5/src.conf.5

@@ -1,5 +1,5 @@
 .\" DO NOT EDIT-- this file is @generated by tools/build/options/makeman.
-.Dd October 30, 2024
+.Dd November 22, 2024
 .Dt SRC.CONF 5
 .Os
 .Sh NAME
@@ -712,12 +712,6 @@ and
 .Xr ftpd 8 .
 .It Va WITHOUT_GAMES
 Do not build games.
-.It Va WITHOUT_GH_BC
-Install the traditional FreeBSD
-.Xr bc 1
-and
-.Xr dc 1
-programs instead of the enhanced versions.
 .It Va WITHOUT_GNU_DIFF
 Do not build GNU
 .Xr diff3 1 ;
@@ -1510,11 +1504,6 @@ and related programs.
 Do not build
 .Xr ppp 8
 and related programs.
-.It Va WITH_PROFILE
-Build profiled libraries for use with
-.Xr gprof 8 .
-This option is deprecated and may not be present in a future version of
-.Fx .
 .It Va WITHOUT_PTHREADS_ASSERTIONS
 Disable debugging assertions in pthreads library.
 .It Va WITHOUT_QUOTAS

share/mk/src.opts.mk

@@ -101,7 +101,6 @@ __DEFAULT_YES_OPTIONS = \
     FREEBSD_UPDATE \
     FTP \
     GAMES \
-    GH_BC \
     GNU_DIFF \
     GOOGLETEST \
     GPIO \

sys/amd64/amd64/pmap.c

@@ -4082,7 +4082,19 @@ pmap_qremove(vm_offset_t sva, int count)
 
 	va = sva;
 	while (count-- > 0) {
+		/*
+		 * pmap_enter() calls within the kernel virtual
+		 * address space happen on virtual addresses from
+		 * subarenas that import superpage-sized and -aligned
+		 * address ranges.  So, the virtual address that we
+		 * allocate to use with pmap_qenter() can't be close
+		 * enough to one of those pmap_enter() calls for it to
+		 * be caught up in a promotion.
+		 */
 		KASSERT(va >= VM_MIN_KERNEL_ADDRESS, ("usermode va %lx", va));
+		KASSERT((*vtopde(va) & X86_PG_PS) == 0,
+		    ("pmap_qremove on promoted va %#lx", va));
+
 		pmap_kremove(va);
 		va += PAGE_SIZE;
 	}
@@ -10506,8 +10518,7 @@ pmap_map_io_transient(vm_page_t page[], vm_offset_t vaddr[], int count,
 {
 	vm_paddr_t paddr;
 	bool needs_mapping;
-	pt_entry_t *pte;
-	int cache_bits, error __unused, i;
+	int error __unused, i;
 
 	/*
 	 * Allocate any KVA space that we need, this is done in a separate
@@ -10552,11 +10563,8 @@ pmap_map_io_transient(vm_page_t page[], vm_offset_t vaddr[], int count,
 				 */
 				pmap_qenter(vaddr[i], &page[i], 1);
 			} else {
-				pte = vtopte(vaddr[i]);
-				cache_bits = pmap_cache_bits(kernel_pmap,
-				    page[i]->md.pat_mode, false);
-				pte_store(pte, paddr | X86_PG_RW | X86_PG_V |
-				    cache_bits);
+				pmap_kenter_attr(vaddr[i], paddr,
+				    page[i]->md.pat_mode);
 				pmap_invlpg(kernel_pmap, vaddr[i]);
 			}
 		}

sys/amd64/amd64/trap.c

@@ -598,6 +598,18 @@ trap(struct trapframe *frame)
 			 * FALLTHROUGH (TRCTRAP kernel mode, kernel address)
 			 */
 		case T_BPTFLT:
+			/*
+			 * Most likely, EFI RT hitting INT3.  This
+			 * check prevents kdb from handling
+			 * breakpoints set on the BIOS text, if such
+			 * option is ever needed.
+			 */
+			if ((td->td_pflags & TDP_EFIRT) != 0 &&
+			    curpcb->pcb_onfault != NULL) {
+				frame->tf_rip = (long)curpcb->pcb_onfault;
+				return;
+			}
+
 			/*
 			 * If KDB is enabled, let it handle the debugger trap.
 			 * Otherwise, debugger traps "can't happen".

sys/arm/arm/exec_machdep.c

@@ -101,14 +101,19 @@ get_vfpcontext(struct thread *td, mcontext_vfp_t *vfp)
 	    P_SHOULDSTOP(td->td_proc));
 
 	pcb = td->td_pcb;
-	if ((pcb->pcb_fpflags & PCB_FP_STARTED) != 0 && td == curthread) {
+	if (td == curthread) {
 		critical_enter();
 		vfp_store(&pcb->pcb_vfpstate, false);
 		critical_exit();
 	}
 	KASSERT(pcb->pcb_vfpsaved == &pcb->pcb_vfpstate,
 		("Called get_vfpcontext while the kernel is using the VFP"));
-	memcpy(vfp, &pcb->pcb_vfpstate, sizeof(*vfp));
+
+	memset(vfp, 0, sizeof(*vfp));
+	memcpy(vfp->mcv_reg, pcb->pcb_vfpstate.reg,
+	    sizeof(vfp->mcv_reg));
+	vfp->mcv_fpscr = pcb->pcb_vfpstate.fpscr;
+
 }
 
 /*
@@ -127,7 +132,10 @@ set_vfpcontext(struct thread *td, mcontext_vfp_t *vfp)
 	}
 	KASSERT(pcb->pcb_vfpsaved == &pcb->pcb_vfpstate,
 		("Called set_vfpcontext while the kernel is using the VFP"));
-	memcpy(&pcb->pcb_vfpstate, vfp, sizeof(*vfp));
+	memcpy(pcb->pcb_vfpstate.reg, vfp->mcv_reg,
+	    sizeof(pcb->pcb_vfpstate.reg));
+	pcb->pcb_vfpstate.fpscr = vfp->mcv_fpscr;
+
 }
 #endif
 
@@ -163,8 +171,6 @@ get_mcontext(struct thread *td, mcontext_t *mcp, int clear_ret)
 {
 	struct trapframe *tf = td->td_frame;
 	__greg_t *gr = mcp->__gregs;
-	mcontext_vfp_t	mcontext_vfp;
-	int rv;
 
 	if (clear_ret & GET_MC_CLEAR_RET) {
 		gr[_REG_R0] = 0;
@@ -189,19 +195,9 @@ get_mcontext(struct thread *td, mcontext_t *mcp, int clear_ret)
 	gr[_REG_LR]   = tf->tf_usr_lr;
 	gr[_REG_PC]   = tf->tf_pc;
 
-#ifdef VFP
-	if (mcp->mc_vfp_size != sizeof(mcontext_vfp_t))
-		return (EINVAL);
-	get_vfpcontext(td, &mcontext_vfp);
-#else
-	bzero(&mcontext_vfp, sizeof(mcontext_vfp));
-#endif
-
-	if (mcp->mc_vfp_ptr != NULL) {
-		rv = copyout(&mcontext_vfp, mcp->mc_vfp_ptr,  sizeof(mcontext_vfp));
-		if (rv != 0)
-			return (rv);
-	}
+	mcp->mc_vfp_size = 0;
+	mcp->mc_vfp_ptr = NULL;
+	memset(&mcp->mc_spare, 0, sizeof(mcp->mc_spare));
 
 	return (0);
 }
@@ -315,6 +311,16 @@ sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask)
 	/* Populate the siginfo frame. */
 	bzero(&frame, sizeof(frame));
 	get_mcontext(td, &frame.sf_uc.uc_mcontext, 0);
+
+#ifdef VFP
+	get_vfpcontext(td, &frame.sf_vfp);
+	frame.sf_uc.uc_mcontext.mc_vfp_size = sizeof(fp->sf_vfp);
+	frame.sf_uc.uc_mcontext.mc_vfp_ptr = &fp->sf_vfp;
+#else
+	frame.sf_uc.uc_mcontext.mc_vfp_size = 0;
+	frame.sf_uc.uc_mcontext.mc_vfp_ptr = NULL;
+#endif
+
 	frame.sf_si = ksi->ksi_info;
 	frame.sf_uc.uc_sigmask = *mask;
 	frame.sf_uc.uc_stack = td->td_sigstk;

sys/arm64/arm64/gicv3_its.c

@@ -444,7 +444,7 @@ gicv3_its_table_page_size(struct gicv3_its_softc *sc, int table)
 	reg = gic_its_read_8(sc, GITS_BASER(table));
 
 	while (1) {
-		reg &= GITS_BASER_PSZ_MASK;
+		reg &= ~GITS_BASER_PSZ_MASK;
 		switch (page_size) {
 		case PAGE_SIZE_4K:	/* 4KB */
 			reg |= GITS_BASER_PSZ_4K << GITS_BASER_PSZ_SHIFT;

sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c

@@ -741,7 +741,7 @@ dtrace_canstore_statvar(uint64_t addr, size_t sz, size_t *remain,
 		return (0);
 
 	maxglobalsize = dtrace_statvar_maxsize + sizeof (uint64_t);
-	maxlocalsize = maxglobalsize * NCPU;
+	maxlocalsize = maxglobalsize * (mp_maxid + 1);
 
 	for (i = 0; i < nsvars; i++) {
 		dtrace_statvar_t *svar = svars[i];
@@ -1744,7 +1744,7 @@ dtrace_dynvar_clean(dtrace_dstate_t *dstate)
 	dtrace_dynvar_t **rinsep;
 	int i, j, work = 0;
 
-	for (i = 0; i < NCPU; i++) {
+	CPU_FOREACH(i) {
 		dcpu = &dstate->dtds_percpu[i];
 		rinsep = &dcpu->dtdsc_rinsing;
 
@@ -1782,7 +1782,7 @@ dtrace_dynvar_clean(dtrace_dstate_t *dstate)
 			 * rinsing list -- and then we borrow this CPU to
 			 * rinse our dirty list.
 			 */
-			for (j = 0; j < NCPU; j++) {
+			CPU_FOREACH(j) {
 				dtrace_dstate_percpu_t *rinser;
 
 				rinser = &dstate->dtds_percpu[j];
@@ -1800,7 +1800,7 @@ dtrace_dynvar_clean(dtrace_dstate_t *dstate)
 				break;
 			}
 
-			if (j == NCPU) {
+			if (j > mp_maxid) {
 				/*
 				 * We were unable to find another CPU that
 				 * could accept this dirty list -- we are
@@ -1841,7 +1841,7 @@ dtrace_dynvar_clean(dtrace_dstate_t *dstate)
 
 	dtrace_sync();
 
-	for (i = 0; i < NCPU; i++) {
+	CPU_FOREACH(i) {
 		dcpu = &dstate->dtds_percpu[i];
 
 		if (dcpu->dtdsc_rinsing == NULL)
@@ -2190,7 +2190,7 @@ retry:
 				case DTRACE_DSTATE_CLEAN: {
 					void *sp = &dstate->dtds_state;
 
-					if (++cpu >= NCPU)
+					if (++cpu > mp_maxid)
 						cpu = 0;
 
 					if (dcpu->dtdsc_dirty != NULL &&
@@ -6504,7 +6504,7 @@ dtrace_dif_emulate(dtrace_difo_t *difo, dtrace_mstate_t *mstate,
 				size_t lim;
 
 				sz += sizeof (uint64_t);
-				ASSERT(svar->dtsv_size == NCPU * sz);
+				ASSERT(svar->dtsv_size == (mp_maxid + 1) * sz);
 				a += curcpu * sz;
 
 				if (*(uint8_t *)a == UINT8_MAX) {
@@ -6521,7 +6521,8 @@ dtrace_dif_emulate(dtrace_difo_t *difo, dtrace_mstate_t *mstate,
 				break;
 			}
 
-			ASSERT(svar->dtsv_size == NCPU * sizeof (uint64_t));
+			ASSERT(svar->dtsv_size ==
+			    (mp_maxid + 1) * sizeof (uint64_t));
 			tmp = (uint64_t *)(uintptr_t)svar->dtsv_data;
 			regs[rd] = tmp[curcpu];
 			break;
@@ -6544,7 +6545,7 @@ dtrace_dif_emulate(dtrace_difo_t *difo, dtrace_mstate_t *mstate,
 				size_t lim;
 
 				sz += sizeof (uint64_t);
-				ASSERT(svar->dtsv_size == NCPU * sz);
+				ASSERT(svar->dtsv_size == (mp_maxid + 1) * sz);
 				a += curcpu * sz;
 
 				if (regs[rd] == 0) {
@@ -6565,7 +6566,8 @@ dtrace_dif_emulate(dtrace_difo_t *difo, dtrace_mstate_t *mstate,
 				break;
 			}
 
-			ASSERT(svar->dtsv_size == NCPU * sizeof (uint64_t));
+			ASSERT(svar->dtsv_size ==
+			    (mp_maxid + 1) * sizeof (uint64_t));
 			tmp = (uint64_t *)(uintptr_t)svar->dtsv_data;
 			tmp[curcpu] = regs[rd];
 			break;
@@ -10678,10 +10680,11 @@ dtrace_difo_init(dtrace_difo_t *dp, dtrace_vstate_t *vstate)
 			svarp = &vstate->dtvs_locals;
 
 			if (v->dtdv_type.dtdt_flags & DIF_TF_BYREF)
-				dsize = NCPU * (v->dtdv_type.dtdt_size +
+				dsize = (mp_maxid + 1) *
+				    (v->dtdv_type.dtdt_size +
 				    sizeof (uint64_t));
 			else
-				dsize = NCPU * sizeof (uint64_t);
+				dsize = (mp_maxid + 1) * sizeof (uint64_t);
 
 			break;
 
@@ -12603,7 +12606,7 @@ dtrace_buffer_consumed(dtrace_buffer_t *bufs, hrtime_t when)
 {
 	int i;
 
-	for (i = 0; i < NCPU; i++) {
+	CPU_FOREACH(i) {
 		dtrace_buffer_t *buf = &bufs[i];
 
 		if (buf->dtb_size == 0)
@@ -12627,7 +12630,7 @@ dtrace_buffer_free(dtrace_buffer_t *bufs)
 {
 	int i;
 
-	for (i = 0; i < NCPU; i++) {
+	CPU_FOREACH(i) {
 		dtrace_buffer_t *buf = &bufs[i];
 
 		if (buf->dtb_tomax == NULL) {
@@ -14379,7 +14382,8 @@ dtrace_dstate_init(dtrace_dstate_t *dstate, size_t size)
 	dstate->dtds_size = size;
 	dstate->dtds_base = base;
 	dstate->dtds_percpu = kmem_cache_alloc(dtrace_state_cache, KM_SLEEP);
-	bzero(dstate->dtds_percpu, NCPU * sizeof (dtrace_dstate_percpu_t));
+	bzero(dstate->dtds_percpu,
+	    (mp_maxid + 1) * sizeof (dtrace_dstate_percpu_t));
 
 	hashsize = size / (dstate->dtds_chunksize + sizeof (dtrace_dynhash_t));
 
@@ -14413,14 +14417,10 @@ dtrace_dstate_init(dtrace_dstate_t *dstate, size_t size)
 	VERIFY((uintptr_t)start < limit);
 	VERIFY((uintptr_t)start >= (uintptr_t)base);
 
-	maxper = (limit - (uintptr_t)start) / NCPU;
+	maxper = (limit - (uintptr_t)start) / (mp_maxid + 1);
 	maxper = (maxper / dstate->dtds_chunksize) * dstate->dtds_chunksize;
 
-#ifndef illumos
 	CPU_FOREACH(i) {
-#else
-	for (i = 0; i < NCPU; i++) {
-#endif
 		dstate->dtds_percpu[i].dtdsc_free = dvar = start;
 
 		/*
@@ -14430,7 +14430,7 @@ dtrace_dstate_init(dtrace_dstate_t *dstate, size_t size)
 		 * whatever is left over.  In either case, we set the limit to
 		 * be the limit of the dynamic variable space.
 		 */
-		if (maxper == 0 || i == NCPU - 1) {
+		if (maxper == 0 || i == mp_maxid) {
 			limit = (uintptr_t)base + size;
 			start = NULL;
 		} else {
@@ -14603,7 +14603,7 @@ dtrace_state_create(struct cdev *dev, struct ucred *cred __unused)
 	char c[30];
 	dtrace_state_t *state;
 	dtrace_optval_t *opt;
-	int bufsize = NCPU * sizeof (dtrace_buffer_t), i;
+	int bufsize = (mp_maxid + 1) * sizeof (dtrace_buffer_t), i;
 	int cpu_it;
 
 	ASSERT(MUTEX_HELD(&dtrace_lock));
@@ -14651,12 +14651,6 @@ dtrace_state_create(struct cdev *dev, struct ucred *cred __unused)
 	state->dts_dev = dev;
 #endif
 
-	/*
-	 * We allocate NCPU buffers.  On the one hand, this can be quite
-	 * a bit of memory per instance (nearly 36K on a Starcat).  On the
-	 * other hand, it saves an additional memory reference in the probe
-	 * path.
-	 */
 	state->dts_buffer = kmem_zalloc(bufsize, KM_SLEEP);
 	state->dts_aggbuffer = kmem_zalloc(bufsize, KM_SLEEP);
 
@@ -14666,12 +14660,12 @@ dtrace_state_create(struct cdev *dev, struct ucred *cred __unused)
          * assumed to be seeded at this point (if from Fortuna seed file).
 	 */
 	arc4random_buf(&state->dts_rstate[0], 2 * sizeof(uint64_t));
-	for (cpu_it = 1; cpu_it < NCPU; cpu_it++) {
+	for (cpu_it = 1; cpu_it <= mp_maxid; cpu_it++) {
 		/*
 		 * Each CPU is assigned a 2^64 period, non-overlapping
 		 * subsequence.
 		 */
-		dtrace_xoroshiro128_plus_jump(state->dts_rstate[cpu_it-1],
+		dtrace_xoroshiro128_plus_jump(state->dts_rstate[cpu_it - 1],
 		    state->dts_rstate[cpu_it]); 
 	}
 
@@ -14970,7 +14964,7 @@ dtrace_state_go(dtrace_state_t *state, processorid_t *cpu)
 	cyc_handler_t hdlr;
 	cyc_time_t when;
 #endif
-	int rval = 0, i, bufsize = NCPU * sizeof (dtrace_buffer_t);
+	int rval = 0, i, bufsize = (mp_maxid + 1) * sizeof (dtrace_buffer_t);
 	dtrace_icookie_t cookie;
 
 	mutex_enter(&cpu_lock);
@@ -15224,10 +15218,10 @@ dtrace_state_go(dtrace_state_t *state, processorid_t *cpu)
 	 * We enable anonymous tracing before APs are started, so we must
 	 * activate buffers using the current CPU.
 	 */
-	if (state == dtrace_anon.dta_state)
-		for (int i = 0; i < NCPU; i++)
+	if (state == dtrace_anon.dta_state) {
+		CPU_FOREACH(i)
 			dtrace_buffer_activate_cpu(state, i);
-	else
+	} else
 		dtrace_xcall(DTRACE_CPUALL,
 		    (dtrace_xcall_t)dtrace_buffer_activate, state);
 #else
@@ -15410,7 +15404,7 @@ dtrace_state_destroy(dtrace_state_t *state)
 #ifdef illumos
 	minor_t minor = getminor(state->dts_dev);
 #endif
-	int i, bufsize = NCPU * sizeof (dtrace_buffer_t);
+	int i, bufsize = (mp_maxid + 1) * sizeof (dtrace_buffer_t);
 	dtrace_speculation_t *spec = state->dts_speculations;
 	int nspec = state->dts_nspeculations;
 	uint32_t match;
@@ -15724,7 +15718,7 @@ dtrace_helper_trace(dtrace_helper_action_t *helper,
 		if ((svar = vstate->dtvs_locals[i]) == NULL)
 			continue;
 
-		ASSERT(svar->dtsv_size >= NCPU * sizeof (uint64_t));
+		ASSERT(svar->dtsv_size >= (mp_maxid + 1) * sizeof (uint64_t));
 		ent->dtht_locals[i] =
 		    ((uint64_t *)(uintptr_t)svar->dtsv_data)[curcpu];
 	}

sys/cddl/contrib/opensolaris/uts/common/sys/dtrace_impl.h

@@ -51,11 +51,7 @@ extern "C" {
 #include <sys/file.h>
 
 #ifndef illumos
-#ifdef __sparcv9
-typedef uint32_t		pc_t;
-#else
 typedef uintptr_t		pc_t;
-#endif
 typedef	u_long			greg_t;
 #endif
 

sys/cddl/dev/dtrace/amd64/dtrace_asm.S

@@ -58,6 +58,8 @@
 
 	ENTRY(dtrace_invop_start)
 
+	KMSAN_ENTER
+
 	/*
 	 * #BP traps with %rip set to the next address. We need to decrement
 	 * the value to indicate the address of the int3 (0xcc) instruction
@@ -77,6 +79,12 @@
 	call	dtrace_invop
 	addq	$16, %rsp
 
+#ifdef KMSAN
+	movq	%rax, %r12
+	KMSAN_LEAVE
+	movq	%r12, %rax
+#endif
+
 	cmpl	$DTRACE_INVOP_PUSHL_EBP, %eax
 	je	bp_push
 	cmpl	$DTRACE_INVOP_CALL, %eax

sys/cddl/dev/dtrace/amd64/dtrace_isa.c

@@ -29,6 +29,7 @@
 #include <sys/systm.h>
 #include <sys/dtrace_impl.h>
 #include <sys/kernel.h>
+#include <sys/msan.h>
 #include <sys/stack.h>
 #include <sys/pcpu.h>
 
@@ -73,6 +74,8 @@ dtrace_getpcstack(pc_t *pcstack, int pcstack_limit, int aframes,
 	frame = (struct amd64_frame *)rbp;
 	td = curthread;
 	while (depth < pcstack_limit) {
+		kmsan_mark(frame, sizeof(*frame), KMSAN_STATE_INITED);
+
 		if (!kstack_contains(curthread, (vm_offset_t)frame,
 		    sizeof(*frame)))
 			break;
@@ -99,6 +102,7 @@ dtrace_getpcstack(pc_t *pcstack, int pcstack_limit, int aframes,
 	for (; depth < pcstack_limit; depth++) {
 		pcstack[depth] = 0;
 	}
+	kmsan_check(pcstack, pcstack_limit * sizeof(*pcstack), "dtrace");
 }
 
 static int
@@ -399,8 +403,10 @@ dtrace_getarg(int arg, int aframes)
 		goto load;
 	}
 
-	for (i = 1; i <= aframes; i++)
+	for (i = 1; i <= aframes; i++) {
+		kmsan_mark(fp, sizeof(*fp), KMSAN_STATE_INITED);
 		fp = fp->f_frame;
+	}
 
 	/*
 	 * We know that we did not come through a trap to get into
@@ -430,6 +436,8 @@ load:
 	val = stack[arg];
 	DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT);
 
+	kmsan_mark(&val, sizeof(val), KMSAN_STATE_INITED);
+
 	return (val);
 }
 
@@ -444,10 +452,13 @@ dtrace_getstackdepth(int aframes)
 	rbp = dtrace_getfp();
 	frame = (struct amd64_frame *)rbp;
 	depth++;
-	for(;;) {
+	for (;;) {
+		kmsan_mark(frame, sizeof(*frame), KMSAN_STATE_INITED);
+
 		if (!kstack_contains(curthread, (vm_offset_t)frame,
 		    sizeof(*frame)))
 			break;
+
 		depth++;
 		if (frame->f_frame <= frame)
 			break;
@@ -574,76 +585,100 @@ void
 dtrace_copyin(uintptr_t uaddr, uintptr_t kaddr, size_t size,
     volatile uint16_t *flags)
 {
-	if (dtrace_copycheck(uaddr, kaddr, size))
+	if (dtrace_copycheck(uaddr, kaddr, size)) {
 		dtrace_copy(uaddr, kaddr, size);
+		kmsan_mark((void *)kaddr, size, KMSAN_STATE_INITED);
+	}
 }
 
 void
 dtrace_copyout(uintptr_t kaddr, uintptr_t uaddr, size_t size,
     volatile uint16_t *flags)
 {
-	if (dtrace_copycheck(uaddr, kaddr, size))
+	if (dtrace_copycheck(uaddr, kaddr, size)) {
+		kmsan_check((void *)kaddr, size, "dtrace_copyout");
 		dtrace_copy(kaddr, uaddr, size);
+	}
 }
 
 void
 dtrace_copyinstr(uintptr_t uaddr, uintptr_t kaddr, size_t size,
     volatile uint16_t *flags)
 {
-	if (dtrace_copycheck(uaddr, kaddr, size))
+	if (dtrace_copycheck(uaddr, kaddr, size)) {
 		dtrace_copystr(uaddr, kaddr, size, flags);
+		kmsan_mark((void *)kaddr, size, KMSAN_STATE_INITED);
+	}
 }
 
 void
 dtrace_copyoutstr(uintptr_t kaddr, uintptr_t uaddr, size_t size,
     volatile uint16_t *flags)
 {
-	if (dtrace_copycheck(uaddr, kaddr, size))
+	if (dtrace_copycheck(uaddr, kaddr, size)) {
+		kmsan_check((void *)kaddr, size, "dtrace_copyoutstr");
 		dtrace_copystr(kaddr, uaddr, size, flags);
+	}
 }
 
 uint8_t
 dtrace_fuword8(void *uaddr)
 {
+	uint8_t val;
+
 	if ((uintptr_t)uaddr > VM_MAXUSER_ADDRESS) {
 		DTRACE_CPUFLAG_SET(CPU_DTRACE_BADADDR);
 		cpu_core[curcpu].cpuc_dtrace_illval = (uintptr_t)uaddr;
 		return (0);
 	}
-	return (dtrace_fuword8_nocheck(uaddr));
+	val = dtrace_fuword8_nocheck(uaddr);
+	kmsan_mark(&val, sizeof(val), KMSAN_STATE_INITED);
+	return (val);
 }
 
 uint16_t
 dtrace_fuword16(void *uaddr)
 {
+	uint16_t val;
+
 	if ((uintptr_t)uaddr > VM_MAXUSER_ADDRESS) {
 		DTRACE_CPUFLAG_SET(CPU_DTRACE_BADADDR);
 		cpu_core[curcpu].cpuc_dtrace_illval = (uintptr_t)uaddr;
 		return (0);
 	}
-	return (dtrace_fuword16_nocheck(uaddr));
+	val = dtrace_fuword16_nocheck(uaddr);
+	kmsan_mark(&val, sizeof(val), KMSAN_STATE_INITED);
+	return (val);
 }
 
 uint32_t
 dtrace_fuword32(void *uaddr)
 {
+	uint32_t val;
+
 	if ((uintptr_t)uaddr > VM_MAXUSER_ADDRESS) {
 		DTRACE_CPUFLAG_SET(CPU_DTRACE_BADADDR);
 		cpu_core[curcpu].cpuc_dtrace_illval = (uintptr_t)uaddr;
 		return (0);
 	}
-	return (dtrace_fuword32_nocheck(uaddr));
+	val = dtrace_fuword32_nocheck(uaddr);
+	kmsan_mark(&val, sizeof(val), KMSAN_STATE_INITED);
+	return (val);
 }
 
 uint64_t
 dtrace_fuword64(void *uaddr)
 {
+	uint64_t val;
+
 	if ((uintptr_t)uaddr > VM_MAXUSER_ADDRESS) {
 		DTRACE_CPUFLAG_SET(CPU_DTRACE_BADADDR);
 		cpu_core[curcpu].cpuc_dtrace_illval = (uintptr_t)uaddr;
 		return (0);
 	}
-	return (dtrace_fuword64_nocheck(uaddr));
+	val = dtrace_fuword64_nocheck(uaddr);
+	kmsan_mark(&val, sizeof(val), KMSAN_STATE_INITED);
+	return (val);
 }
 
 /*

sys/cddl/dev/dtrace/amd64/dtrace_subr.c

@@ -33,6 +33,7 @@
 #include <sys/systm.h>
 #include <sys/kernel.h>
 #include <sys/malloc.h>
+#include <sys/msan.h>
 #include <sys/proc.h>
 #include <sys/smp.h>
 #include <sys/dtrace_impl.h>
@@ -68,6 +69,8 @@ dtrace_invop(uintptr_t addr, struct trapframe *frame, void **scratch)
 	dtrace_invop_hdlr_t *hdlr;
 	int rval;
 
+	kmsan_mark(frame, sizeof(*frame), KMSAN_STATE_INITED);
+
 	td = curthread;
 	td->t_dtrace_trapframe = frame;
 	rval = 0;

sys/cddl/dev/dtrace/dtrace_ioctl.c

@@ -229,7 +229,7 @@ dtrace_ioctl(struct cdev *dev, u_long cmd, caddr_t addr,
 		    "DTRACEIOC_AGGSNAP":"DTRACEIOC_BUFSNAP",
 		    curcpu, desc.dtbd_cpu);
 
-		if (desc.dtbd_cpu >= MAXCPU || CPU_ABSENT(desc.dtbd_cpu))
+		if (desc.dtbd_cpu > mp_maxid || CPU_ABSENT(desc.dtbd_cpu))
 			return (ENOENT);
 
 		mutex_enter(&dtrace_lock);

sys/cddl/dev/dtrace/dtrace_load.c

@@ -100,8 +100,8 @@ dtrace_load(void *dummy)
 	mutex_enter(&dtrace_lock);
 
 	dtrace_state_cache = kmem_cache_create("dtrace_state_cache",
-	    sizeof (dtrace_dstate_percpu_t) * NCPU, DTRACE_STATE_ALIGN,
-	    NULL, NULL, NULL, NULL, NULL, 0);
+	    sizeof (dtrace_dstate_percpu_t) * (mp_maxid + 1),
+	    DTRACE_STATE_ALIGN, NULL, NULL, NULL, NULL, NULL, 0);
 
 	ASSERT(MUTEX_HELD(&cpu_lock));
 	dtrace_bymod = dtrace_hash_create(offsetof(dtrace_probe_t, dtpr_mod),

sys/cddl/dev/dtrace/x86/dis_tables.c

@@ -3720,8 +3720,8 @@ dtrace_disx86(dis86_t *x, uint_t cpu_mode)
 	x->d86_len = 0;
 	x->d86_rmindex = -1;
 	x->d86_error = 0;
-#ifdef DIS_TEXT
 	x->d86_numopnds = 0;
+#ifdef DIS_TEXT
 	x->d86_seg_prefix = NULL;
 	x->d86_mnem[0] = 0;
 	for (i = 0; i < 4; ++i) {
@@ -4860,6 +4860,7 @@ mm_shift:
 		dtrace_rex_adjust(rex_prefix, mode, NULL, &r_m);
 		dtrace_get_operand(x, mode, r_m, LONG_OPND, vbit);
 		dtrace_get_operand(x, REG_ONLY, reg, SEG_OPND, 1 - vbit);
+		wbit = SEG_OPND;
 		break;
 
 	/*
@@ -5787,6 +5788,7 @@ xmmprm:
 #endif
 			dtrace_rex_adjust(rex_prefix, mode, &reg, &r_m);
 			dtrace_get_operand(x, mode, r_m, BYTE_OPND, 0);
+			wbit = LONG_OPND;
 		}
 		break;
 
@@ -6442,7 +6444,7 @@ done:
 		else
 			x->d86_memsize = dp->it_size;
 
-	} else if (wbit == 0) {
+	} else if (wbit == BYTE_OPND) {
 		x->d86_memsize = 1;
 
 	} else if (wbit == LONG_OPND) {

sys/cddl/dev/fbt/fbt.c

@@ -136,6 +136,13 @@ fbt_excluded(const char *name)
 	    strcmp(name, "owner_sx") == 0)
 		return (1);
 
+	/*
+	 * The KMSAN runtime can't be instrumented safely.
+	 */
+	if (strncmp(name, "__msan", 6) == 0 ||
+	    strncmp(name, "kmsan_", 6) == 0)
+		return (1);
+
 	/*
 	 * Stack unwinders may be called from probe context on some
 	 * platforms.

sys/cddl/dev/kinst/kinst.c

@@ -132,6 +132,13 @@ kinst_excluded(const char *name)
 	    strcmp(name, "owner_sx") == 0)
 		return (true);
 
+	/*
+	 * The KMSAN runtime can't be instrumented safely.
+	 */
+	if (strncmp(name, "__msan", 6) == 0 ||
+	    strncmp(name, "kmsan_", 6) == 0)
+		return (1);
+
 	/*
 	 * When DTrace is built into the kernel we need to exclude the kinst
 	 * functions from instrumentation.

sys/conf/files.riscv

@@ -36,6 +36,7 @@ riscv/riscv/bus_machdep.c	standard
 riscv/riscv/bus_space_asm.S	standard
 riscv/riscv/busdma_bounce.c	standard
 riscv/riscv/busdma_machdep.c	standard
+riscv/riscv/cache.c		standard
 riscv/riscv/clock.c		standard
 riscv/riscv/copyinout.S		standard
 riscv/riscv/cpufunc_asm.S	standard
@@ -83,5 +84,7 @@ riscv/vmm/vmm_riscv.c				optional	vmm
 riscv/vmm/vmm_sbi.c				optional	vmm
 riscv/vmm/vmm_switch.S				optional	vmm
 
+riscv/thead/thead.c		standard
+
 # Zstd
 contrib/zstd/lib/freebsd/zstd_kfreebsd.c		optional zstdio compile-with ${ZSTD_C}

sys/contrib/openzfs/.mailmap

@@ -70,6 +70,7 @@ Rob Norris <robn@despairlabs.com>
 Rob Norris <rob.norris@klarasystems.com>
 Sam Lunt <samuel.j.lunt@gmail.com>
 Sanjeev Bagewadi <sanjeev.bagewadi@gmail.com>
+Sebastian Wuerl <s.wuerl@mailbox.org>
 Stoiko Ivanov <github@nomore.at>
 Tamas TEVESZ <ice@extreme.hu>
 WHR <msl0000023508@gmail.com>
@@ -78,6 +79,7 @@ Youzhong Yang <youzhong@gmail.com>
 
 # Signed-off-by: overriding Author:
 Ryan <errornointernet@envs.net> <error.nointernet@gmail.com>
+Sietse <sietse@wizdom.nu> <uglymotha@wizdom.nu>
 Qiuhao Chen <chenqiuhao1997@gmail.com> <haohao0924@126.com>
 Yuxin Wang <yuxinwang9999@gmail.com> <Bi11gates9999@gmail.com>
 Zhenlei Huang <zlei@FreeBSD.org> <zlei.huang@gmail.com>

sys/contrib/openzfs/AUTHORS

@@ -423,6 +423,7 @@ CONTRIBUTORS:
     Mathieu Velten <matmaul@gmail.com>
     Matt Fiddaman <github@m.fiddaman.uk>
     Matthew Ahrens <matt@delphix.com>
+    Matthew Heller <matthew.f.heller@gmail.com>
     Matthew Thode <mthode@mthode.org>
     Matthias Blankertz <matthias@blankertz.org>
     Matt Johnston <matt@fugro-fsi.com.au>
@@ -562,6 +563,7 @@ CONTRIBUTORS:
     Scot W. Stevenson <scot.stevenson@gmail.com>
     Sean Eric Fagan <sef@ixsystems.com>
     Sebastian Gottschall <s.gottschall@dd-wrt.com>
+    Sebastian Wuerl <s.wuerl@mailbox.org>
     Sebastien Roy <seb@delphix.com>
     Sen Haerens <sen@senhaerens.be>
     Serapheim Dimitropoulos <serapheim@delphix.com>
@@ -574,6 +576,7 @@ CONTRIBUTORS:
     Shawn Bayern <sbayern@law.fsu.edu>
     Shengqi Chen <harry-chen@outlook.com>
     Shen Yan <shenyanxxxy@qq.com>
+    Sietse <sietse@wizdom.nu>
     Simon Guest <simon.guest@tesujimath.org>
     Simon Klinkert <simon.klinkert@gmail.com>
     Sowrabha Gopal <sowrabha.gopal@delphix.com>
@@ -629,6 +632,7 @@ CONTRIBUTORS:
     Trevor Bautista <trevrb@trevrb.net>
     Trey Dockendorf <treydock@gmail.com>
     Troels Nørgaard <tnn@tradeshift.com>
+    tstabrawa <tstabrawa@users.noreply.github.com>
     Tulsi Jain <tulsi.jain@delphix.com>
     Turbo Fredriksson <turbo@bayour.com>
     Tyler J. Stachecki <stachecki.tyler@gmail.com>

sys/contrib/openzfs/META

@@ -6,5 +6,5 @@ Release:       1
 Release-Tags:  relext
 License:       CDDL
 Author:        OpenZFS
-Linux-Maximum: 6.11
+Linux-Maximum: 6.12
 Linux-Minimum: 4.18

sys/contrib/openzfs/cmd/arc_summary

@@ -662,10 +662,7 @@ def section_arc(kstats_dict):
     print()
 
     print('ARC hash breakdown:')
-    prt_i1('Elements max:', f_hits(arc_stats['hash_elements_max']))
-    prt_i2('Elements current:',
-           f_perc(arc_stats['hash_elements'], arc_stats['hash_elements_max']),
-           f_hits(arc_stats['hash_elements']))
+    prt_i1('Elements:', f_hits(arc_stats['hash_elements']))
     prt_i1('Collisions:', f_hits(arc_stats['hash_collisions']))
 
     prt_i1('Chain max:', f_hits(arc_stats['hash_chain_max']))

sys/contrib/openzfs/cmd/zdb/zdb.c

@@ -2119,9 +2119,6 @@ dump_brt(spa_t *spa)
 		return;
 	}
 
-	brt_t *brt = spa->spa_brt;
-	VERIFY(brt);
-
 	char count[32], used[32], saved[32];
 	zdb_nicebytes(brt_get_used(spa), used, sizeof (used));
 	zdb_nicebytes(brt_get_saved(spa), saved, sizeof (saved));
@@ -2132,11 +2129,8 @@ dump_brt(spa_t *spa)
 	if (dump_opt['T'] < 2)
 		return;
 
-	for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
-		brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
-		if (brtvd == NULL)
-			continue;
-
+	for (uint64_t vdevid = 0; vdevid < spa->spa_brt_nvdevs; vdevid++) {
+		brt_vdev_t *brtvd = spa->spa_brt_vdevs[vdevid];
 		if (!brtvd->bv_initiated) {
 			printf("BRT: vdev %" PRIu64 ": empty\n", vdevid);
 			continue;
@@ -2160,20 +2154,21 @@ dump_brt(spa_t *spa)
 	if (!do_histo)
 		printf("\n%-16s %-10s\n", "DVA", "REFCNT");
 
-	for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
-		brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
-		if (brtvd == NULL || !brtvd->bv_initiated)
+	for (uint64_t vdevid = 0; vdevid < spa->spa_brt_nvdevs; vdevid++) {
+		brt_vdev_t *brtvd = spa->spa_brt_vdevs[vdevid];
+		if (!brtvd->bv_initiated)
 			continue;
 
 		uint64_t counts[64] = {};
 
 		zap_cursor_t zc;
 		zap_attribute_t *za = zap_attribute_alloc();
-		for (zap_cursor_init(&zc, brt->brt_mos, brtvd->bv_mos_entries);
+		for (zap_cursor_init(&zc, spa->spa_meta_objset,
+		    brtvd->bv_mos_entries);
 		    zap_cursor_retrieve(&zc, za) == 0;
 		    zap_cursor_advance(&zc)) {
 			uint64_t refcnt;
-			VERIFY0(zap_lookup_uint64(brt->brt_mos,
+			VERIFY0(zap_lookup_uint64(spa->spa_meta_objset,
 			    brtvd->bv_mos_entries,
 			    (const uint64_t *)za->za_name, 1,
 			    za->za_integer_length, za->za_num_integers,
@@ -8227,14 +8222,11 @@ dump_mos_leaks(spa_t *spa)
 		}
 	}
 
-	if (spa->spa_brt != NULL) {
-		brt_t *brt = spa->spa_brt;
-		for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
-			brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
-			if (brtvd != NULL && brtvd->bv_initiated) {
-				mos_obj_refd(brtvd->bv_mos_brtvdev);
-				mos_obj_refd(brtvd->bv_mos_entries);
-			}
+	for (uint64_t vdevid = 0; vdevid < spa->spa_brt_nvdevs; vdevid++) {
+		brt_vdev_t *brtvd = spa->spa_brt_vdevs[vdevid];
+		if (brtvd->bv_initiated) {
+			mos_obj_refd(brtvd->bv_mos_brtvdev);
+			mos_obj_refd(brtvd->bv_mos_entries);
 		}
 	}
 

sys/contrib/openzfs/cmd/zed/agents/zfs_retire.c

@@ -445,8 +445,8 @@ zfs_retire_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl,
 		 * its a loopback event from spa_async_remove(). Just
 		 * ignore it.
 		 */
-		if (vs->vs_state == VDEV_STATE_REMOVED &&
-		    state == VDEV_STATE_REMOVED)
+		if ((vs->vs_state == VDEV_STATE_REMOVED && state ==
+		    VDEV_STATE_REMOVED) || vs->vs_state == VDEV_STATE_OFFLINE)
 			return;
 
 		/* Remove the vdev since device is unplugged */

sys/contrib/openzfs/include/os/freebsd/spl/sys/debug.h

@@ -201,7 +201,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    "failed (%lld " #OP " %lld) " STR "\n",		\
 		    (long long)(_verify3_left),				\
 		    (long long)(_verify3_right),			\
-		    __VA_ARGS);						\
+		    __VA_ARGS__);					\
 	} while (0)
 
 #define	VERIFY3UF(LEFT, OP, RIGHT, STR, ...)	do {			\
@@ -213,7 +213,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    "failed (%llu " #OP " %llu) " STR "\n",		\
 		    (unsigned long long)(_verify3_left),		\
 		    (unsigned long long)(_verify3_right),		\
-		    __VA_ARGS);						\
+		    __VA_ARGS__);					\
 	} while (0)
 
 #define	VERIFY3PF(LEFT, OP, RIGHT, STR, ...)	do {			\

sys/contrib/openzfs/include/os/freebsd/spl/sys/vnode.h

@@ -98,11 +98,9 @@ vn_flush_cached_data(vnode_t *vp, boolean_t sync)
 {
 	if (vm_object_mightbedirty(vp->v_object)) {
 		int flags = sync ? OBJPC_SYNC : 0;
-		vn_lock(vp, LK_SHARED | LK_RETRY);
 		zfs_vmobject_wlock(vp->v_object);
 		vm_object_page_clean(vp->v_object, 0, 0, flags);
 		zfs_vmobject_wunlock(vp->v_object);
-		VOP_UNLOCK(vp);
 	}
 }
 #endif

sys/contrib/openzfs/include/os/linux/spl/sys/debug.h

@@ -205,7 +205,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    "failed (%lld " #OP " %lld) " STR "\n",		\
 		    (long long)(_verify3_left),				\
 		    (long long)(_verify3_right),			\
-		    __VA_ARGS);						\
+		    __VA_ARGS__);					\
 	} while (0)
 
 #define	VERIFY3UF(LEFT, OP, RIGHT, STR, ...)	do {			\
@@ -217,7 +217,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    "failed (%llu " #OP " %llu) " STR "\n",		\
 		    (unsigned long long)(_verify3_left),		\
 		    (unsigned long long)(_verify3_right),		\
-		    __VA_ARGS);						\
+		    __VA_ARGS__);					\
 	} while (0)
 
 #define	VERIFY3PF(LEFT, OP, RIGHT, STR, ...)	do {			\

sys/contrib/openzfs/include/sys/arc.h

@@ -347,6 +347,7 @@ void l2arc_fini(void);
 void l2arc_start(void);
 void l2arc_stop(void);
 void l2arc_spa_rebuild_start(spa_t *spa);
+void l2arc_spa_rebuild_stop(spa_t *spa);
 
 #ifndef _KERNEL
 extern boolean_t arc_watch;

sys/contrib/openzfs/include/sys/arc_impl.h

@@ -942,6 +942,7 @@ typedef struct arc_sums {
 	wmsum_t arcstat_evict_l2_eligible_mru;
 	wmsum_t arcstat_evict_l2_ineligible;
 	wmsum_t arcstat_evict_l2_skip;
+	wmsum_t arcstat_hash_elements;
 	wmsum_t arcstat_hash_collisions;
 	wmsum_t arcstat_hash_chains;
 	aggsum_t arcstat_size;

sys/contrib/openzfs/include/sys/brt_impl.h

@@ -86,28 +86,38 @@ typedef struct brt_vdev_phys {
 	uint64_t	bvp_savedspace;
 } brt_vdev_phys_t;
 
-typedef struct brt_vdev {
+struct brt_vdev {
+	/*
+	 * Pending changes from open contexts.
+	 */
+	kmutex_t	bv_pending_lock;
+	avl_tree_t	bv_pending_tree[TXG_SIZE];
+	/*
+	 * Protects bv_mos_*.
+	 */
+	krwlock_t	bv_mos_entries_lock ____cacheline_aligned;
+	/*
+	 * Protects all the fields starting from bv_initiated.
+	 */
+	krwlock_t	bv_lock ____cacheline_aligned;
 	/*
 	 * VDEV id.
 	 */
-	uint64_t	bv_vdevid;
-	/*
-	 * Is the structure initiated?
-	 * (bv_entcount and bv_bitmap are allocated?)
-	 */
-	boolean_t	bv_initiated;
+	uint64_t	bv_vdevid ____cacheline_aligned;
 	/*
 	 * Object number in the MOS for the entcount array and brt_vdev_phys.
 	 */
 	uint64_t	bv_mos_brtvdev;
 	/*
-	 * Object number in the MOS for the entries table.
+	 * Object number in the MOS and dnode for the entries table.
 	 */
 	uint64_t	bv_mos_entries;
+	dnode_t		*bv_mos_entries_dnode;
 	/*
-	 * Entries to sync.
+	 * Is the structure initiated?
+	 * (bv_entcount and bv_bitmap are allocated?)
 	 */
-	avl_tree_t	bv_tree;
+	boolean_t	bv_initiated;
 	/*
 	 * Does the bv_entcount[] array needs byte swapping?
 	 */
@@ -120,6 +130,26 @@ typedef struct brt_vdev {
 	 * This is the array with BRT entry count per BRT_RANGESIZE.
 	 */
 	uint16_t	*bv_entcount;
+	/*
+	 * bv_entcount[] potentially can be a bit too big to sychronize it all
+	 * when we just changed few entcounts. The fields below allow us to
+	 * track updates to bv_entcount[] array since the last sync.
+	 * A single bit in the bv_bitmap represents as many entcounts as can
+	 * fit into a single BRT_BLOCKSIZE.
+	 * For example we have 65536 entcounts in the bv_entcount array
+	 * (so the whole array is 128kB). We updated bv_entcount[2] and
+	 * bv_entcount[5]. In that case only first bit in the bv_bitmap will
+	 * be set and we will write only first BRT_BLOCKSIZE out of 128kB.
+	 */
+	ulong_t		*bv_bitmap;
+	/*
+	 * bv_entcount[] needs updating on disk.
+	 */
+	boolean_t	bv_entcount_dirty;
+	/*
+	 * brt_vdev_phys needs updating on disk.
+	 */
+	boolean_t	bv_meta_dirty;
 	/*
 	 * Sum of all bv_entcount[]s.
 	 */
@@ -133,65 +163,27 @@ typedef struct brt_vdev {
 	 */
 	uint64_t	bv_savedspace;
 	/*
-	 * brt_vdev_phys needs updating on disk.
+	 * Entries to sync.
 	 */
-	boolean_t	bv_meta_dirty;
-	/*
-	 * bv_entcount[] needs updating on disk.
-	 */
-	boolean_t	bv_entcount_dirty;
-	/*
-	 * bv_entcount[] potentially can be a bit too big to sychronize it all
-	 * when we just changed few entcounts. The fields below allow us to
-	 * track updates to bv_entcount[] array since the last sync.
-	 * A single bit in the bv_bitmap represents as many entcounts as can
-	 * fit into a single BRT_BLOCKSIZE.
-	 * For example we have 65536 entcounts in the bv_entcount array
-	 * (so the whole array is 128kB). We updated bv_entcount[2] and
-	 * bv_entcount[5]. In that case only first bit in the bv_bitmap will
-	 * be set and we will write only first BRT_BLOCKSIZE out of 128kB.
-	 */
-	ulong_t		*bv_bitmap;
-	uint64_t	bv_nblocks;
-} brt_vdev_t;
+	avl_tree_t	bv_tree;
+};
 
-/*
- * In-core brt
- */
-typedef struct brt {
-	krwlock_t	brt_lock;
-	spa_t		*brt_spa;
-#define	brt_mos		brt_spa->spa_meta_objset
-	uint64_t	brt_rangesize;
-	uint64_t	brt_usedspace;
-	uint64_t	brt_savedspace;
-	avl_tree_t	brt_pending_tree[TXG_SIZE];
-	kmutex_t	brt_pending_lock[TXG_SIZE];
-	/* Sum of all entries across all bv_trees. */
-	uint64_t	brt_nentries;
-	brt_vdev_t	*brt_vdevs;
-	uint64_t	brt_nvdevs;
-} brt_t;
-
-/* Size of bre_offset / sizeof (uint64_t). */
+/* Size of offset / sizeof (uint64_t). */
 #define	BRT_KEY_WORDS	(1)
 
+#define	BRE_OFFSET(bre)	(DVA_GET_OFFSET(&(bre)->bre_bp.blk_dva[0]))
+
 /*
  * In-core brt entry.
- * On-disk we use bre_offset as the key and bre_refcount as the value.
+ * On-disk we use ZAP with offset as the key and count as the value.
  */
 typedef struct brt_entry {
-	uint64_t	bre_offset;
-	uint64_t	bre_refcount;
 	avl_node_t	bre_node;
+	blkptr_t	bre_bp;
+	uint64_t	bre_count;
+	uint64_t	bre_pcount;
 } brt_entry_t;
 
-typedef struct brt_pending_entry {
-	blkptr_t	bpe_bp;
-	int		bpe_count;
-	avl_node_t	bpe_node;
-} brt_pending_entry_t;
-
 #ifdef	__cplusplus
 }
 #endif

sys/contrib/openzfs/include/sys/spa.h

@@ -53,6 +53,7 @@ extern "C" {
 /*
  * Forward references that lots of things need.
  */
+typedef struct brt_vdev brt_vdev_t;
 typedef struct spa spa_t;
 typedef struct vdev vdev_t;
 typedef struct metaslab metaslab_t;

sys/contrib/openzfs/include/sys/spa_impl.h

@@ -412,8 +412,12 @@ struct spa {
 	uint64_t	spa_dedup_dspace;	/* Cache get_dedup_dspace() */
 	uint64_t	spa_dedup_checksum;	/* default dedup checksum */
 	uint64_t	spa_dspace;		/* dspace in normal class */
+	uint64_t	spa_rdspace;		/* raw (non-dedup) --//-- */
 	boolean_t	spa_active_ddt_prune;	/* ddt prune process active */
-	struct brt	*spa_brt;		/* in-core BRT */
+	brt_vdev_t	**spa_brt_vdevs;	/* array of per-vdev BRTs */
+	uint64_t	spa_brt_nvdevs;		/* number of vdevs in BRT */
+	uint64_t	spa_brt_rangesize;	/* pool's BRT range size */
+	krwlock_t	spa_brt_lock;		/* Protects brt_vdevs/nvdevs */
 	kmutex_t	spa_vdev_top_lock;	/* dueling offline/remove */
 	kmutex_t	spa_proc_lock;		/* protects spa_proc* */
 	kcondvar_t	spa_proc_cv;		/* spa_proc_state transitions */

sys/contrib/openzfs/include/sys/zap.h

@@ -223,11 +223,15 @@ int zap_lookup_norm(objset_t *ds, uint64_t zapobj, const char *name,
     boolean_t *normalization_conflictp);
 int zap_lookup_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
     int key_numints, uint64_t integer_size, uint64_t num_integers, void *buf);
+int zap_lookup_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints, uint64_t integer_size, uint64_t num_integers, void *buf);
 int zap_contains(objset_t *ds, uint64_t zapobj, const char *name);
 int zap_prefetch(objset_t *os, uint64_t zapobj, const char *name);
 int zap_prefetch_object(objset_t *os, uint64_t zapobj);
 int zap_prefetch_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
     int key_numints);
+int zap_prefetch_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints);
 
 int zap_lookup_by_dnode(dnode_t *dn, const char *name,
     uint64_t integer_size, uint64_t num_integers, void *buf);
@@ -236,9 +240,6 @@ int zap_lookup_norm_by_dnode(dnode_t *dn, const char *name,
     matchtype_t mt, char *realname, int rn_len,
     boolean_t *ncp);
 
-int zap_count_write_by_dnode(dnode_t *dn, const char *name,
-    int add, zfs_refcount_t *towrite, zfs_refcount_t *tooverwrite);
-
 /*
  * Create an attribute with the given name and value.
  *

sys/contrib/openzfs/man/man8/zpool-remove.8

@@ -109,7 +109,7 @@ Stops and cancels an in-progress removal of a top-level vdev.
 .El
 .
 .Sh EXAMPLES
-.\" These are, respectively, examples 14 from zpool.8
+.\" These are, respectively, examples 15 from zpool.8
 .\" Make sure to update them bidirectionally
 .Ss Example 1 : No Removing a Mirrored top-level (Log or Data) Device
 The following commands remove the mirrored log device
@@ -142,9 +142,43 @@ The command to remove the mirrored log
 .Ar mirror-2 No is :
 .Dl # Nm zpool Cm remove Ar tank mirror-2
 .Pp
+At this point, the log device no longer exists
+(both sides of the mirror have been removed):
+.Bd -literal -compact -offset Ds
+  pool: tank
+ state: ONLINE
+  scan: none requested
+config:
+
+        NAME        STATE     READ WRITE CKSUM
+        tank        ONLINE       0     0     0
+          mirror-0  ONLINE       0     0     0
+            sda     ONLINE       0     0     0
+            sdb     ONLINE       0     0     0
+          mirror-1  ONLINE       0     0     0
+            sdc     ONLINE       0     0     0
+            sdd     ONLINE       0     0     0
+.Ed
+.Pp
 The command to remove the mirrored data
 .Ar mirror-1 No is :
 .Dl # Nm zpool Cm remove Ar tank mirror-1
+.Pp
+After
+.Ar mirror-1 No has been evacuated, the pool remains redundant, but
+the total amount of space is reduced:
+.Bd -literal -compact -offset Ds
+  pool: tank
+ state: ONLINE
+  scan: none requested
+config:
+
+        NAME        STATE     READ WRITE CKSUM
+        tank        ONLINE       0     0     0
+          mirror-0  ONLINE       0     0     0
+            sda     ONLINE       0     0     0
+            sdb     ONLINE       0     0     0
+.Ed
 .
 .Sh SEE ALSO
 .Xr zpool-add 8 ,

sys/contrib/openzfs/man/man8/zpool.8

@@ -405,9 +405,43 @@ The command to remove the mirrored log
 .Ar mirror-2 No is :
 .Dl # Nm zpool Cm remove Ar tank mirror-2
 .Pp
+At this point, the log device no longer exists
+(both sides of the mirror have been removed):
+.Bd -literal -compact -offset Ds
+  pool: tank
+ state: ONLINE
+  scan: none requested
+config:
+
+        NAME        STATE     READ WRITE CKSUM
+        tank        ONLINE       0     0     0
+          mirror-0  ONLINE       0     0     0
+            sda     ONLINE       0     0     0
+            sdb     ONLINE       0     0     0
+          mirror-1  ONLINE       0     0     0
+            sdc     ONLINE       0     0     0
+            sdd     ONLINE       0     0     0
+.Ed
+.Pp
 The command to remove the mirrored data
 .Ar mirror-1 No is :
 .Dl # Nm zpool Cm remove Ar tank mirror-1
+.Pp
+After
+.Ar mirror-1 No has been evacuated, the pool remains redundant, but
+the total amount of space is reduced:
+.Bd -literal -compact -offset Ds
+  pool: tank
+ state: ONLINE
+  scan: none requested
+config:
+
+        NAME        STATE     READ WRITE CKSUM
+        tank        ONLINE       0     0     0
+          mirror-0  ONLINE       0     0     0
+            sda     ONLINE       0     0     0
+            sdb     ONLINE       0     0     0
+.Ed
 .
 .Ss Example 16 : No Displaying expanded space on a device
 The following command displays the detailed information for the pool

sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c

@@ -291,8 +291,12 @@ zfs_ioctl(vnode_t *vp, ulong_t com, intptr_t data, int flag, cred_t *cred,
 	case F_SEEK_HOLE:
 	{
 		off = *(offset_t *)data;
+		error = vn_lock(vp, LK_SHARED);
+		if (error)
+			return (error);
 		/* offset parameter is in/out */
 		error = zfs_holey(VTOZ(vp), com, &off);
+		VOP_UNLOCK(vp);
 		if (error)
 			return (error);
 		*(offset_t *)data = off;
@@ -452,8 +456,10 @@ mappedread_sf(znode_t *zp, int nbytes, zfs_uio_t *uio)
 				if (!vm_page_wired(pp) && pp->valid == 0 &&
 				    vm_page_busy_tryupgrade(pp))
 					vm_page_free(pp);
-				else
+				else {
+					vm_page_deactivate_noreuse(pp);
 					vm_page_sunbusy(pp);
+				}
 				zfs_vmobject_wunlock(obj);
 			}
 		} else {
@@ -3928,6 +3934,7 @@ zfs_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind,
 	if (zfs_enter_verify_zp(zfsvfs, zp, FTAG) != 0)
 		return (zfs_vm_pagerret_error);
 
+	object = ma[0]->object;
 	start = IDX_TO_OFF(ma[0]->pindex);
 	end = IDX_TO_OFF(ma[count - 1]->pindex + 1);
 
@@ -3936,33 +3943,47 @@ zfs_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind,
 	 * Note that we need to handle the case of the block size growing.
 	 */
 	for (;;) {
+		uint64_t len;
+
 		blksz = zp->z_blksz;
+		len = roundup(end, blksz) - rounddown(start, blksz);
+
 		lr = zfs_rangelock_tryenter(&zp->z_rangelock,
-		    rounddown(start, blksz),
-		    roundup(end, blksz) - rounddown(start, blksz), RL_READER);
+		    rounddown(start, blksz), len, RL_READER);
 		if (lr == NULL) {
-			if (rahead != NULL) {
-				*rahead = 0;
-				rahead = NULL;
+			/*
+			 * Avoid a deadlock with update_pages().  We need to
+			 * hold the range lock when copying from the DMU, so
+			 * give up the busy lock to allow update_pages() to
+			 * proceed.  We might need to allocate new pages, which
+			 * isn't quite right since this allocation isn't subject
+			 * to the page fault handler's OOM logic, but this is
+			 * the best we can do for now.
+			 */
+			for (int i = 0; i < count; i++) {
+				ASSERT(vm_page_none_valid(ma[i]));
+				vm_page_xunbusy(ma[i]);
 			}
-			if (rbehind != NULL) {
-				*rbehind = 0;
-				rbehind = NULL;
-			}
-			break;
+
+			lr = zfs_rangelock_enter(&zp->z_rangelock,
+			    rounddown(start, blksz), len, RL_READER);
+
+			zfs_vmobject_wlock(object);
+			(void) vm_page_grab_pages(object, OFF_TO_IDX(start),
+			    VM_ALLOC_NORMAL | VM_ALLOC_WAITOK | VM_ALLOC_ZERO,
+			    ma, count);
+			zfs_vmobject_wunlock(object);
 		}
 		if (blksz == zp->z_blksz)
 			break;
 		zfs_rangelock_exit(lr);
 	}
 
-	object = ma[0]->object;
 	zfs_vmobject_wlock(object);
 	obj_size = object->un_pager.vnp.vnp_size;
 	zfs_vmobject_wunlock(object);
 	if (IDX_TO_OFF(ma[count - 1]->pindex) >= obj_size) {
-		if (lr != NULL)
-			zfs_rangelock_exit(lr);
+		zfs_rangelock_exit(lr);
 		zfs_exit(zfsvfs, FTAG);
 		return (zfs_vm_pagerret_bad);
 	}
@@ -3987,11 +4008,33 @@ zfs_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind,
 	 * ZFS will panic if we request DMU to read beyond the end of the last
 	 * allocated block.
 	 */
-	error = dmu_read_pages(zfsvfs->z_os, zp->z_id, ma, count, &pgsin_b,
-	    &pgsin_a, MIN(end, obj_size) - (end - PAGE_SIZE));
+	for (int i = 0; i < count; i++) {
+		int dummypgsin, count1, j, last_size;
 
-	if (lr != NULL)
-		zfs_rangelock_exit(lr);
+		if (vm_page_any_valid(ma[i])) {
+			ASSERT(vm_page_all_valid(ma[i]));
+			continue;
+		}
+		for (j = i + 1; j < count; j++) {
+			if (vm_page_any_valid(ma[j])) {
+				ASSERT(vm_page_all_valid(ma[j]));
+				break;
+			}
+		}
+		count1 = j - i;
+		dummypgsin = 0;
+		last_size = j == count ?
+		    MIN(end, obj_size) - (end - PAGE_SIZE) : PAGE_SIZE;
+		error = dmu_read_pages(zfsvfs->z_os, zp->z_id, &ma[i], count1,
+		    i == 0 ? &pgsin_b : &dummypgsin,
+		    j == count ? &pgsin_a : &dummypgsin,
+		    last_size);
+		if (error != 0)
+			break;
+		i += count1 - 1;
+	}
+
+	zfs_rangelock_exit(lr);
 	ZFS_ACCESSTIME_STAMP(zfsvfs, zp);
 
 	dataset_kstats_update_read_kstats(&zfsvfs->z_kstat, count*PAGE_SIZE);
@@ -6159,7 +6202,7 @@ zfs_freebsd_copy_file_range(struct vop_copy_file_range_args *ap)
 	} else {
 #if (__FreeBSD_version >= 1302506 && __FreeBSD_version < 1400000) || \
 	__FreeBSD_version >= 1400086
-		vn_lock_pair(invp, false, LK_EXCLUSIVE, outvp, false,
+		vn_lock_pair(invp, false, LK_SHARED, outvp, false,
 		    LK_EXCLUSIVE);
 #else
 		vn_lock_pair(invp, false, outvp, false);

sys/contrib/openzfs/module/os/linux/zfs/zpl_super.c

@@ -375,7 +375,18 @@ zpl_prune_sb(uint64_t nr_to_scan, void *arg)
 	struct super_block *sb = (struct super_block *)arg;
 	int objects = 0;
 
-	(void) -zfs_prune(sb, nr_to_scan, &objects);
+	/*
+	 * deactivate_locked_super calls shrinker_free and only then
+	 * sops->kill_sb cb, resulting in UAF on umount when trying to reach
+	 * for the shrinker functions in zpl_prune_sb of in-umount dataset.
+	 * Increment if s_active is not zero, but don't prune if it is -
+	 * umount could be underway.
+	 */
+	if (atomic_inc_not_zero(&sb->s_active)) {
+		(void) -zfs_prune(sb, nr_to_scan, &objects);
+		atomic_dec(&sb->s_active);
+	}
+
 }
 
 const struct super_operations zpl_super_operations = {

sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c

@@ -1176,7 +1176,7 @@ zvol_queue_limits_init(zvol_queue_limits_t *limits, zvol_state_t *zv,
 		limits->zql_max_segment_size = UINT_MAX;
 	}
 
-	limits->zql_io_opt = zv->zv_volblocksize;
+	limits->zql_io_opt = DMU_MAX_ACCESS / 2;
 
 	limits->zql_physical_block_size = zv->zv_volblocksize;
 	limits->zql_max_discard_sectors =

sys/contrib/openzfs/module/zfs/arc.c

@@ -1074,12 +1074,9 @@ buf_hash_insert(arc_buf_hdr_t *hdr, kmutex_t **lockp)
 		ARCSTAT_BUMP(arcstat_hash_collisions);
 		if (i == 1)
 			ARCSTAT_BUMP(arcstat_hash_chains);
-
 		ARCSTAT_MAX(arcstat_hash_chain_max, i);
 	}
-	uint64_t he = atomic_inc_64_nv(
-	    &arc_stats.arcstat_hash_elements.value.ui64);
-	ARCSTAT_MAX(arcstat_hash_elements_max, he);
+	ARCSTAT_BUMP(arcstat_hash_elements);
 
 	return (NULL);
 }
@@ -1103,8 +1100,7 @@ buf_hash_remove(arc_buf_hdr_t *hdr)
 	arc_hdr_clear_flags(hdr, ARC_FLAG_IN_HASH_TABLE);
 
 	/* collect some hash table performance data */
-	atomic_dec_64(&arc_stats.arcstat_hash_elements.value.ui64);
-
+	ARCSTAT_BUMPDOWN(arcstat_hash_elements);
 	if (buf_hash_table.ht_table[idx] &&
 	    buf_hash_table.ht_table[idx]->b_hash_next == NULL)
 		ARCSTAT_BUMPDOWN(arcstat_hash_chains);
@@ -7008,6 +7004,9 @@ arc_kstat_update(kstat_t *ksp, int rw)
 	    wmsum_value(&arc_sums.arcstat_evict_l2_ineligible);
 	as->arcstat_evict_l2_skip.value.ui64 =
 	    wmsum_value(&arc_sums.arcstat_evict_l2_skip);
+	as->arcstat_hash_elements.value.ui64 =
+	    as->arcstat_hash_elements_max.value.ui64 =
+	    wmsum_value(&arc_sums.arcstat_hash_elements);
 	as->arcstat_hash_collisions.value.ui64 =
 	    wmsum_value(&arc_sums.arcstat_hash_collisions);
 	as->arcstat_hash_chains.value.ui64 =
@@ -7432,6 +7431,7 @@ arc_state_init(void)
 	wmsum_init(&arc_sums.arcstat_evict_l2_eligible_mru, 0);
 	wmsum_init(&arc_sums.arcstat_evict_l2_ineligible, 0);
 	wmsum_init(&arc_sums.arcstat_evict_l2_skip, 0);
+	wmsum_init(&arc_sums.arcstat_hash_elements, 0);
 	wmsum_init(&arc_sums.arcstat_hash_collisions, 0);
 	wmsum_init(&arc_sums.arcstat_hash_chains, 0);
 	aggsum_init(&arc_sums.arcstat_size, 0);
@@ -7590,6 +7590,7 @@ arc_state_fini(void)
 	wmsum_fini(&arc_sums.arcstat_evict_l2_eligible_mru);
 	wmsum_fini(&arc_sums.arcstat_evict_l2_ineligible);
 	wmsum_fini(&arc_sums.arcstat_evict_l2_skip);
+	wmsum_fini(&arc_sums.arcstat_hash_elements);
 	wmsum_fini(&arc_sums.arcstat_hash_collisions);
 	wmsum_fini(&arc_sums.arcstat_hash_chains);
 	aggsum_fini(&arc_sums.arcstat_size);
@@ -9287,6 +9288,14 @@ skip:
 			hdr->b_l2hdr.b_hits = 0;
 			hdr->b_l2hdr.b_arcs_state =
 			    hdr->b_l1hdr.b_state->arcs_state;
+			arc_hdr_set_flags(hdr, ARC_FLAG_HAS_L2HDR |
+			    ARC_FLAG_L2_WRITING);
+
+			(void) zfs_refcount_add_many(&dev->l2ad_alloc,
+			    arc_hdr_size(hdr), hdr);
+			l2arc_hdr_arcstats_increment(hdr);
+			vdev_space_update(dev->l2ad_vdev, asize, 0, 0);
+
 			mutex_enter(&dev->l2ad_mtx);
 			if (pio == NULL) {
 				/*
@@ -9298,12 +9307,6 @@ skip:
 			}
 			list_insert_head(&dev->l2ad_buflist, hdr);
 			mutex_exit(&dev->l2ad_mtx);
-			arc_hdr_set_flags(hdr, ARC_FLAG_HAS_L2HDR |
-			    ARC_FLAG_L2_WRITING);
-
-			(void) zfs_refcount_add_many(&dev->l2ad_alloc,
-			    arc_hdr_size(hdr), hdr);
-			l2arc_hdr_arcstats_increment(hdr);
 
 			boolean_t commit = l2arc_log_blk_insert(dev, hdr);
 			mutex_exit(hash_lock);
@@ -9333,7 +9336,6 @@ skip:
 			write_psize += psize;
 			write_asize += asize;
 			dev->l2ad_hand += asize;
-			vdev_space_update(dev->l2ad_vdev, asize, 0, 0);
 
 			if (commit) {
 				/* l2ad_hand will be adjusted inside. */
@@ -9844,6 +9846,37 @@ l2arc_spa_rebuild_start(spa_t *spa)
 	}
 }
 
+void
+l2arc_spa_rebuild_stop(spa_t *spa)
+{
+	ASSERT(MUTEX_HELD(&spa_namespace_lock) ||
+	    spa->spa_export_thread == curthread);
+
+	for (int i = 0; i < spa->spa_l2cache.sav_count; i++) {
+		l2arc_dev_t *dev =
+		    l2arc_vdev_get(spa->spa_l2cache.sav_vdevs[i]);
+		if (dev == NULL)
+			continue;
+		mutex_enter(&l2arc_rebuild_thr_lock);
+		dev->l2ad_rebuild_cancel = B_TRUE;
+		mutex_exit(&l2arc_rebuild_thr_lock);
+	}
+	for (int i = 0; i < spa->spa_l2cache.sav_count; i++) {
+		l2arc_dev_t *dev =
+		    l2arc_vdev_get(spa->spa_l2cache.sav_vdevs[i]);
+		if (dev == NULL)
+			continue;
+		mutex_enter(&l2arc_rebuild_thr_lock);
+		if (dev->l2ad_rebuild_began == B_TRUE) {
+			while (dev->l2ad_rebuild == B_TRUE) {
+				cv_wait(&l2arc_rebuild_thr_cv,
+				    &l2arc_rebuild_thr_lock);
+			}
+		}
+		mutex_exit(&l2arc_rebuild_thr_lock);
+	}
+}
+
 /*
  * Main entry point for L2ARC rebuilding.
  */
@@ -9852,12 +9885,12 @@ l2arc_dev_rebuild_thread(void *arg)
 {
 	l2arc_dev_t *dev = arg;
 
-	VERIFY(!dev->l2ad_rebuild_cancel);
 	VERIFY(dev->l2ad_rebuild);
 	(void) l2arc_rebuild(dev);
 	mutex_enter(&l2arc_rebuild_thr_lock);
 	dev->l2ad_rebuild_began = B_FALSE;
 	dev->l2ad_rebuild = B_FALSE;
+	cv_signal(&l2arc_rebuild_thr_cv);
 	mutex_exit(&l2arc_rebuild_thr_lock);
 
 	thread_exit();
@@ -10008,8 +10041,6 @@ l2arc_rebuild(l2arc_dev_t *dev)
 		for (;;) {
 			mutex_enter(&l2arc_rebuild_thr_lock);
 			if (dev->l2ad_rebuild_cancel) {
-				dev->l2ad_rebuild = B_FALSE;
-				cv_signal(&l2arc_rebuild_thr_cv);
 				mutex_exit(&l2arc_rebuild_thr_lock);
 				err = SET_ERROR(ECANCELED);
 				goto out;
@@ -10585,6 +10616,8 @@ l2arc_log_blk_commit(l2arc_dev_t *dev, zio_t *pio, l2arc_write_callback_t *cb)
 	(void) zio_nowait(wzio);
 
 	dev->l2ad_hand += asize;
+	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);
+
 	/*
 	 * Include the committed log block's pointer  in the list of pointers
 	 * to log blocks present in the L2ARC device.
@@ -10598,7 +10631,6 @@ l2arc_log_blk_commit(l2arc_dev_t *dev, zio_t *pio, l2arc_write_callback_t *cb)
 	zfs_refcount_add_many(&dev->l2ad_lb_asize, asize, lb_ptr_buf);
 	zfs_refcount_add(&dev->l2ad_lb_count, lb_ptr_buf);
 	mutex_exit(&dev->l2ad_mtx);
-	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);
 
 	/* bump the kstats */
 	ARCSTAT_INCR(arcstat_l2_write_bytes, asize);

sys/contrib/openzfs/module/zfs/dbuf.c

@@ -89,7 +89,6 @@ typedef struct dbuf_stats {
 	kstat_named_t hash_misses;
 	kstat_named_t hash_collisions;
 	kstat_named_t hash_elements;
-	kstat_named_t hash_elements_max;
 	/*
 	 * Number of sublists containing more than one dbuf in the dbuf
 	 * hash table. Keep track of the longest hash chain.
@@ -134,7 +133,6 @@ dbuf_stats_t dbuf_stats = {
 	{ "hash_misses",			KSTAT_DATA_UINT64 },
 	{ "hash_collisions",			KSTAT_DATA_UINT64 },
 	{ "hash_elements",			KSTAT_DATA_UINT64 },
-	{ "hash_elements_max",			KSTAT_DATA_UINT64 },
 	{ "hash_chains",			KSTAT_DATA_UINT64 },
 	{ "hash_chain_max",			KSTAT_DATA_UINT64 },
 	{ "hash_insert_race",			KSTAT_DATA_UINT64 },
@@ -154,6 +152,7 @@ struct {
 	wmsum_t hash_hits;
 	wmsum_t hash_misses;
 	wmsum_t hash_collisions;
+	wmsum_t hash_elements;
 	wmsum_t hash_chains;
 	wmsum_t hash_insert_race;
 	wmsum_t metadata_cache_count;
@@ -432,8 +431,7 @@ dbuf_hash_insert(dmu_buf_impl_t *db)
 	db->db_hash_next = h->hash_table[idx];
 	h->hash_table[idx] = db;
 	mutex_exit(DBUF_HASH_MUTEX(h, idx));
-	uint64_t he = atomic_inc_64_nv(&dbuf_stats.hash_elements.value.ui64);
-	DBUF_STAT_MAX(hash_elements_max, he);
+	DBUF_STAT_BUMP(hash_elements);
 
 	return (NULL);
 }
@@ -506,7 +504,7 @@ dbuf_hash_remove(dmu_buf_impl_t *db)
 	    h->hash_table[idx]->db_hash_next == NULL)
 		DBUF_STAT_BUMPDOWN(hash_chains);
 	mutex_exit(DBUF_HASH_MUTEX(h, idx));
-	atomic_dec_64(&dbuf_stats.hash_elements.value.ui64);
+	DBUF_STAT_BUMPDOWN(hash_elements);
 }
 
 typedef enum {
@@ -903,6 +901,8 @@ dbuf_kstat_update(kstat_t *ksp, int rw)
 	    wmsum_value(&dbuf_sums.hash_misses);
 	ds->hash_collisions.value.ui64 =
 	    wmsum_value(&dbuf_sums.hash_collisions);
+	ds->hash_elements.value.ui64 =
+	    wmsum_value(&dbuf_sums.hash_elements);
 	ds->hash_chains.value.ui64 =
 	    wmsum_value(&dbuf_sums.hash_chains);
 	ds->hash_insert_race.value.ui64 =
@@ -1004,6 +1004,7 @@ dbuf_init(void)
 	wmsum_init(&dbuf_sums.hash_hits, 0);
 	wmsum_init(&dbuf_sums.hash_misses, 0);
 	wmsum_init(&dbuf_sums.hash_collisions, 0);
+	wmsum_init(&dbuf_sums.hash_elements, 0);
 	wmsum_init(&dbuf_sums.hash_chains, 0);
 	wmsum_init(&dbuf_sums.hash_insert_race, 0);
 	wmsum_init(&dbuf_sums.metadata_cache_count, 0);
@@ -1077,6 +1078,7 @@ dbuf_fini(void)
 	wmsum_fini(&dbuf_sums.hash_hits);
 	wmsum_fini(&dbuf_sums.hash_misses);
 	wmsum_fini(&dbuf_sums.hash_collisions);
+	wmsum_fini(&dbuf_sums.hash_elements);
 	wmsum_fini(&dbuf_sums.hash_chains);
 	wmsum_fini(&dbuf_sums.hash_insert_race);
 	wmsum_fini(&dbuf_sums.metadata_cache_count);
@@ -2578,8 +2580,11 @@ dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
 		 * We are freeing a block that we cloned in the same
 		 * transaction group.
 		 */
-		brt_pending_remove(dmu_objset_spa(db->db_objset),
-		    &dr->dt.dl.dr_overridden_by, tx);
+		blkptr_t *bp = &dr->dt.dl.dr_overridden_by;
+		if (!BP_IS_HOLE(bp) && !BP_IS_EMBEDDED(bp)) {
+			brt_pending_remove(dmu_objset_spa(db->db_objset),
+			    bp, tx);
+		}
 	}
 
 	dnode_t *dn = dr->dr_dnode;

sys/contrib/openzfs/module/zfs/dsl_dataset.c

@@ -68,6 +68,7 @@
 #include <sys/zio_compress.h>
 #include <zfs_fletcher.h>
 #include <sys/zio_checksum.h>
+#include <sys/brt.h>
 
 /*
  * The SPA supports block sizes up to 16MB.  However, very large blocks
@@ -289,8 +290,26 @@ dsl_dataset_block_kill(dsl_dataset_t *ds, const blkptr_t *bp, dmu_tx_t *tx,
 	if (BP_GET_LOGICAL_BIRTH(bp) > dsl_dataset_phys(ds)->ds_prev_snap_txg) {
 		int64_t delta;
 
-		dprintf_bp(bp, "freeing ds=%llu", (u_longlong_t)ds->ds_object);
-		dsl_free(tx->tx_pool, tx->tx_txg, bp);
+		/*
+		 * Put blocks that would create IO on the pool's deadlist for
+		 * dsl_process_async_destroys() to find. This is to prevent
+		 * zio_free() from creating a ZIO_TYPE_FREE IO for them, which
+		 * are very heavy and can lead to out-of-memory conditions if
+		 * something tries to free millions of blocks on the same txg.
+		 */
+		boolean_t defer = spa_version(spa) >= SPA_VERSION_DEADLISTS &&
+		    (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) ||
+		    brt_maybe_exists(spa, bp));
+
+		if (defer) {
+			dprintf_bp(bp, "putting on free list: %s", "");
+			bpobj_enqueue(&ds->ds_dir->dd_pool->dp_free_bpobj,
+			    bp, B_FALSE, tx);
+		} else {
+			dprintf_bp(bp, "freeing ds=%llu",
+			    (u_longlong_t)ds->ds_object);
+			dsl_free(tx->tx_pool, tx->tx_txg, bp);
+		}
 
 		mutex_enter(&ds->ds_lock);
 		ASSERT(dsl_dataset_phys(ds)->ds_unique_bytes >= used ||
@@ -298,9 +317,14 @@ dsl_dataset_block_kill(dsl_dataset_t *ds, const blkptr_t *bp, dmu_tx_t *tx,
 		delta = parent_delta(ds, -used);
 		dsl_dataset_phys(ds)->ds_unique_bytes -= used;
 		mutex_exit(&ds->ds_lock);
+
 		dsl_dir_diduse_transfer_space(ds->ds_dir,
 		    delta, -compressed, -uncompressed, -used,
 		    DD_USED_REFRSRV, DD_USED_HEAD, tx);
+
+		if (defer)
+			dsl_dir_diduse_space(tx->tx_pool->dp_free_dir,
+			    DD_USED_HEAD, used, compressed, uncompressed, tx);
 	} else {
 		dprintf_bp(bp, "putting on dead list: %s", "");
 		if (async) {

sys/contrib/openzfs/module/zfs/spa.c

@@ -2081,6 +2081,7 @@ spa_unload(spa_t *spa)
 			vdev_trim_stop_all(root_vdev, VDEV_TRIM_ACTIVE);
 			vdev_autotrim_stop_all(spa);
 			vdev_rebuild_stop_all(spa);
+			l2arc_spa_rebuild_stop(spa);
 		}
 	}
 
@@ -7115,6 +7116,7 @@ spa_export_common(const char *pool, int new_state, nvlist_t **oldconfig,
 		vdev_trim_stop_all(rvd, VDEV_TRIM_ACTIVE);
 		vdev_autotrim_stop_all(spa);
 		vdev_rebuild_stop_all(spa);
+		l2arc_spa_rebuild_stop(spa);
 
 		/*
 		 * We want this to be reflected on every label,

sys/contrib/openzfs/module/zfs/spa_misc.c

@@ -1870,13 +1870,7 @@ spa_get_slop_space(spa_t *spa)
 	if (spa->spa_dedup_dspace == ~0ULL)
 		spa_update_dspace(spa);
 
-	/*
-	 * spa_get_dspace() includes the space only logically "used" by
-	 * deduplicated data, so since it's not useful to reserve more
-	 * space with more deduplicated data, we subtract that out here.
-	 */
-	space =
-	    spa_get_dspace(spa) - spa->spa_dedup_dspace - brt_get_dspace(spa);
+	space = spa->spa_rdspace;
 	slop = MIN(space >> spa_slop_shift, spa_max_slop);
 
 	/*
@@ -1912,8 +1906,7 @@ spa_get_checkpoint_space(spa_t *spa)
 void
 spa_update_dspace(spa_t *spa)
 {
-	spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) +
-	    ddt_get_dedup_dspace(spa) + brt_get_dspace(spa);
+	spa->spa_rdspace = metaslab_class_get_dspace(spa_normal_class(spa));
 	if (spa->spa_nonallocating_dspace > 0) {
 		/*
 		 * Subtract the space provided by all non-allocating vdevs that
@@ -1933,9 +1926,11 @@ spa_update_dspace(spa_t *spa)
 		 * doesn't matter that the data we are moving may be
 		 * allocated twice (on the old device and the new device).
 		 */
-		ASSERT3U(spa->spa_dspace, >=, spa->spa_nonallocating_dspace);
-		spa->spa_dspace -= spa->spa_nonallocating_dspace;
+		ASSERT3U(spa->spa_rdspace, >=, spa->spa_nonallocating_dspace);
+		spa->spa_rdspace -= spa->spa_nonallocating_dspace;
 	}
+	spa->spa_dspace = spa->spa_rdspace + ddt_get_dedup_dspace(spa) +
+	    brt_get_dspace(spa);
 }
 
 /*

sys/contrib/openzfs/module/zfs/zap_leaf.c

@@ -248,20 +248,63 @@ zap_leaf_array_create(zap_leaf_t *l, const char *buf,
 	return (chunk_head);
 }
 
-static void
-zap_leaf_array_free(zap_leaf_t *l, uint16_t *chunkp)
+/*
+ * Non-destructively copy array between leaves.
+ */
+static uint16_t
+zap_leaf_array_copy(zap_leaf_t *l, uint16_t chunk, zap_leaf_t *nl)
 {
-	uint16_t chunk = *chunkp;
-
-	*chunkp = CHAIN_END;
+	uint16_t new_chunk;
+	uint16_t *nchunkp = &new_chunk;
 
 	while (chunk != CHAIN_END) {
-		uint_t nextchunk = ZAP_LEAF_CHUNK(l, chunk).l_array.la_next;
-		ASSERT3U(ZAP_LEAF_CHUNK(l, chunk).l_array.la_type, ==,
-		    ZAP_CHUNK_ARRAY);
-		zap_leaf_chunk_free(l, chunk);
-		chunk = nextchunk;
+		ASSERT3U(chunk, <, ZAP_LEAF_NUMCHUNKS(l));
+		uint16_t nchunk = zap_leaf_chunk_alloc(nl);
+
+		struct zap_leaf_array *la =
+		    &ZAP_LEAF_CHUNK(l, chunk).l_array;
+		struct zap_leaf_array *nla =
+		    &ZAP_LEAF_CHUNK(nl, nchunk).l_array;
+		ASSERT3U(la->la_type, ==, ZAP_CHUNK_ARRAY);
+
+		*nla = *la; /* structure assignment */
+
+		chunk = la->la_next;
+		*nchunkp = nchunk;
+		nchunkp = &nla->la_next;
 	}
+	*nchunkp = CHAIN_END;
+	return (new_chunk);
+}
+
+/*
+ * Free array.  Unlike trivial loop of zap_leaf_chunk_free() this does
+ * not reverse order of chunks in the free list, reducing fragmentation.
+ */
+static void
+zap_leaf_array_free(zap_leaf_t *l, uint16_t chunk)
+{
+	struct zap_leaf_header *hdr = &zap_leaf_phys(l)->l_hdr;
+	uint16_t *tailp = &hdr->lh_freelist;
+	uint16_t oldfree = *tailp;
+
+	while (chunk != CHAIN_END) {
+		ASSERT3U(chunk, <, ZAP_LEAF_NUMCHUNKS(l));
+		zap_leaf_chunk_t *c = &ZAP_LEAF_CHUNK(l, chunk);
+		ASSERT3U(c->l_array.la_type, ==, ZAP_CHUNK_ARRAY);
+
+		*tailp = chunk;
+		chunk = c->l_array.la_next;
+
+		c->l_free.lf_type = ZAP_CHUNK_FREE;
+		memset(c->l_free.lf_pad, 0, sizeof (c->l_free.lf_pad));
+		tailp = &c->l_free.lf_next;
+
+		ASSERT3U(hdr->lh_nfree, <, ZAP_LEAF_NUMCHUNKS(l));
+		hdr->lh_nfree++;
+	}
+
+	*tailp = oldfree;
 }
 
 /* array_len and buf_len are in integers, not bytes */
@@ -515,7 +558,7 @@ zap_entry_update(zap_entry_handle_t *zeh,
 	if ((int)zap_leaf_phys(l)->l_hdr.lh_nfree < delta_chunks)
 		return (SET_ERROR(EAGAIN));
 
-	zap_leaf_array_free(l, &le->le_value_chunk);
+	zap_leaf_array_free(l, le->le_value_chunk);
 	le->le_value_chunk =
 	    zap_leaf_array_create(l, buf, integer_size, num_integers);
 	le->le_value_numints = num_integers;
@@ -534,10 +577,11 @@ zap_entry_remove(zap_entry_handle_t *zeh)
 	struct zap_leaf_entry *le = ZAP_LEAF_ENTRY(l, entry_chunk);
 	ASSERT3U(le->le_type, ==, ZAP_CHUNK_ENTRY);
 
-	zap_leaf_array_free(l, &le->le_name_chunk);
-	zap_leaf_array_free(l, &le->le_value_chunk);
-
 	*zeh->zeh_chunkp = le->le_next;
+
+	/* Free in opposite order to reduce fragmentation. */
+	zap_leaf_array_free(l, le->le_value_chunk);
+	zap_leaf_array_free(l, le->le_name_chunk);
 	zap_leaf_chunk_free(l, entry_chunk);
 
 	zap_leaf_phys(l)->l_hdr.lh_nentries--;
@@ -701,34 +745,6 @@ zap_leaf_rehash_entry(zap_leaf_t *l, struct zap_leaf_entry *le, uint16_t entry)
 	return (chunkp);
 }
 
-static uint16_t
-zap_leaf_transfer_array(zap_leaf_t *l, uint16_t chunk, zap_leaf_t *nl)
-{
-	uint16_t new_chunk;
-	uint16_t *nchunkp = &new_chunk;
-
-	while (chunk != CHAIN_END) {
-		uint16_t nchunk = zap_leaf_chunk_alloc(nl);
-		struct zap_leaf_array *nla =
-		    &ZAP_LEAF_CHUNK(nl, nchunk).l_array;
-		struct zap_leaf_array *la =
-		    &ZAP_LEAF_CHUNK(l, chunk).l_array;
-		uint_t nextchunk = la->la_next;
-
-		ASSERT3U(chunk, <, ZAP_LEAF_NUMCHUNKS(l));
-		ASSERT3U(nchunk, <, ZAP_LEAF_NUMCHUNKS(l));
-
-		*nla = *la; /* structure assignment */
-
-		zap_leaf_chunk_free(l, chunk);
-		chunk = nextchunk;
-		*nchunkp = nchunk;
-		nchunkp = &nla->la_next;
-	}
-	*nchunkp = CHAIN_END;
-	return (new_chunk);
-}
-
 static void
 zap_leaf_transfer_entry(zap_leaf_t *l, uint_t entry, zap_leaf_t *nl)
 {
@@ -741,10 +757,12 @@ zap_leaf_transfer_entry(zap_leaf_t *l, uint_t entry, zap_leaf_t *nl)
 
 	(void) zap_leaf_rehash_entry(nl, nle, chunk);
 
-	nle->le_name_chunk = zap_leaf_transfer_array(l, le->le_name_chunk, nl);
-	nle->le_value_chunk =
-	    zap_leaf_transfer_array(l, le->le_value_chunk, nl);
+	nle->le_name_chunk = zap_leaf_array_copy(l, le->le_name_chunk, nl);
+	nle->le_value_chunk = zap_leaf_array_copy(l, le->le_value_chunk, nl);
 
+	/* Free in opposite order to reduce fragmentation. */
+	zap_leaf_array_free(l, le->le_value_chunk);
+	zap_leaf_array_free(l, le->le_name_chunk);
 	zap_leaf_chunk_free(l, entry);
 
 	zap_leaf_phys(l)->l_hdr.lh_nentries--;

sys/contrib/openzfs/module/zfs/zap_micro.c

@@ -1227,6 +1227,21 @@ zap_lookup_norm_by_dnode(dnode_t *dn, const char *name,
 	return (err);
 }
 
+static int
+zap_prefetch_uint64_impl(zap_t *zap, const uint64_t *key, int key_numints)
+{
+	zap_name_t *zn = zap_name_alloc_uint64(zap, key, key_numints);
+	if (zn == NULL) {
+		zap_unlockdir(zap, FTAG);
+		return (SET_ERROR(ENOTSUP));
+	}
+
+	fzap_prefetch(zn);
+	zap_name_free(zn);
+	zap_unlockdir(zap, FTAG);
+	return (0);
+}
+
 int
 zap_prefetch_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
     int key_numints)
@@ -1237,13 +1252,37 @@ zap_prefetch_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
 	    zap_lockdir(os, zapobj, NULL, RW_READER, TRUE, FALSE, FTAG, &zap);
 	if (err != 0)
 		return (err);
+	err = zap_prefetch_uint64_impl(zap, key, key_numints);
+	/* zap_prefetch_uint64_impl() calls zap_unlockdir() */
+	return (err);
+}
+
+int
+zap_prefetch_uint64_by_dnode(dnode_t *dn, const uint64_t *key, int key_numints)
+{
+	zap_t *zap;
+
+	int err =
+	    zap_lockdir_by_dnode(dn, NULL, RW_READER, TRUE, FALSE, FTAG, &zap);
+	if (err != 0)
+		return (err);
+	err = zap_prefetch_uint64_impl(zap, key, key_numints);
+	/* zap_prefetch_uint64_impl() calls zap_unlockdir() */
+	return (err);
+}
+
+static int
+zap_lookup_uint64_impl(zap_t *zap, const uint64_t *key,
+    int key_numints, uint64_t integer_size, uint64_t num_integers, void *buf)
+{
 	zap_name_t *zn = zap_name_alloc_uint64(zap, key, key_numints);
 	if (zn == NULL) {
 		zap_unlockdir(zap, FTAG);
 		return (SET_ERROR(ENOTSUP));
 	}
 
-	fzap_prefetch(zn);
+	int err = fzap_lookup(zn, integer_size, num_integers, buf,
+	    NULL, 0, NULL);
 	zap_name_free(zn);
 	zap_unlockdir(zap, FTAG);
 	return (err);
@@ -1259,16 +1298,25 @@ zap_lookup_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
 	    zap_lockdir(os, zapobj, NULL, RW_READER, TRUE, FALSE, FTAG, &zap);
 	if (err != 0)
 		return (err);
-	zap_name_t *zn = zap_name_alloc_uint64(zap, key, key_numints);
-	if (zn == NULL) {
-		zap_unlockdir(zap, FTAG);
-		return (SET_ERROR(ENOTSUP));
-	}
+	err = zap_lookup_uint64_impl(zap, key, key_numints, integer_size,
+	    num_integers, buf);
+	/* zap_lookup_uint64_impl() calls zap_unlockdir() */
+	return (err);
+}
 
-	err = fzap_lookup(zn, integer_size, num_integers, buf,
-	    NULL, 0, NULL);
-	zap_name_free(zn);
-	zap_unlockdir(zap, FTAG);
+int
+zap_lookup_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints, uint64_t integer_size, uint64_t num_integers, void *buf)
+{
+	zap_t *zap;
+
+	int err =
+	    zap_lockdir_by_dnode(dn, NULL, RW_READER, TRUE, FALSE, FTAG, &zap);
+	if (err != 0)
+		return (err);
+	err = zap_lookup_uint64_impl(zap, key, key_numints, integer_size,
+	    num_integers, buf);
+	/* zap_lookup_uint64_impl() calls zap_unlockdir() */
 	return (err);
 }
 

sys/contrib/openzfs/module/zfs/zio.c

@@ -2192,31 +2192,20 @@ zio_delay_interrupt(zio_t *zio)
 		} else {
 			taskqid_t tid;
 			hrtime_t diff = zio->io_target_timestamp - now;
-			clock_t expire_at_tick = ddi_get_lbolt() +
-			    NSEC_TO_TICK(diff);
+			int ticks = MAX(1, NSEC_TO_TICK(diff));
+			clock_t expire_at_tick = ddi_get_lbolt() + ticks;
 
 			DTRACE_PROBE3(zio__delay__hit, zio_t *, zio,
 			    hrtime_t, now, hrtime_t, diff);
 
-			if (NSEC_TO_TICK(diff) == 0) {
-				/* Our delay is less than a jiffy - just spin */
-				zfs_sleep_until(zio->io_target_timestamp);
-				zio_interrupt(zio);
-			} else {
+			tid = taskq_dispatch_delay(system_taskq, zio_interrupt,
+			    zio, TQ_NOSLEEP, expire_at_tick);
+			if (tid == TASKQID_INVALID) {
 				/*
-				 * Use taskq_dispatch_delay() in the place of
-				 * OpenZFS's timeout_generic().
+				 * Couldn't allocate a task.  Just finish the
+				 * zio without a delay.
 				 */
-				tid = taskq_dispatch_delay(system_taskq,
-				    zio_interrupt, zio, TQ_NOSLEEP,
-				    expire_at_tick);
-				if (tid == TASKQID_INVALID) {
-					/*
-					 * Couldn't allocate a task.  Just
-					 * finish the zio without a delay.
-					 */
-					zio_interrupt(zio);
-				}
+				zio_interrupt(zio);
 			}
 		}
 		return;

sys/contrib/openzfs/module/zfs/zio_checksum.c

@@ -160,6 +160,12 @@ abd_fletcher_4_byteswap(abd_t *abd, uint64_t size,
 	abd_fletcher_4_impl(abd, size, &acd);
 }
 
+/*
+ * Checksum vectors.
+ *
+ * Note: you cannot change the name string for these functions, as they are
+ * embedded in on-disk data in some places (eg dedup table names).
+ */
 zio_checksum_info_t zio_checksum_table[ZIO_CHECKSUM_FUNCTIONS] = {
 	{{NULL, NULL}, NULL, NULL, 0, "inherit"},
 	{{NULL, NULL}, NULL, NULL, 0, "on"},

sys/contrib/openzfs/module/zfs/zio_compress.c

@@ -44,10 +44,6 @@ static unsigned long zio_decompress_fail_fraction = 0;
 
 /*
  * Compression vectors.
- *
- * NOTE: DO NOT CHANGE THE NAMES OF THESE COMPRESSION FUNCTIONS.
- * THEY ARE USED AS ZAP KEY NAMES BY FAST DEDUP AND THEREFORE
- * PART OF THE ON-DISK FORMAT.
  */
 zio_compress_info_t zio_compress_table[ZIO_COMPRESS_FUNCTIONS] = {
 	{"inherit",	0,	NULL,	NULL, NULL},

sys/contrib/openzfs/rpm/generic/zfs-dkms.spec.in

@@ -32,6 +32,7 @@ Requires(post): gcc, make, perl, diffutils
 %if 0%{?rhel}%{?fedora}%{?mageia}%{?suse_version}%{?openEuler}
 Requires:       kernel-devel >= @ZFS_META_KVER_MIN@, kernel-devel <= @ZFS_META_KVER_MAX@.999
 Requires(post): kernel-devel >= @ZFS_META_KVER_MIN@, kernel-devel <= @ZFS_META_KVER_MAX@.999
+Conflicts:      kernel-devel < @ZFS_META_KVER_MIN@, kernel-devel > @ZFS_META_KVER_MAX@.999
 Obsoletes:      spl-dkms <= %{version}
 %endif
 Provides:       %{module}-kmod = %{version}

sys/contrib/openzfs/tests/zfs-tests/cmd/getversion.c

@@ -19,9 +19,13 @@
  */
 
 #include <sys/ioctl.h>
+#ifdef _KERNEL
+#include <sys/fcntl.h>
+#else
+#include <fcntl.h>
+#endif
 #include <linux/fs.h>
 #include <err.h>
-#include <fcntl.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>

sys/contrib/openzfs/tests/zfs-tests/tests/functional/bclone/bclone_prop_sync.ksh

@@ -41,9 +41,11 @@ log_must zfs set compress=zle $TESTDSTFS
 
 for prop in "${sync_prop_vals[@]}"; do
     log_must zfs set sync=$prop $TESTSRCFS
+    # 15*8=120, which is greater than 113, so we are sure the data won't
+    # be embedded into BP.
     # 32767*8=262136, which is larger than a single default recordsize of
     # 131072.
-    FILESIZE=$(random_int_between 1 32767)
+    FILESIZE=$(random_int_between 15 32767)
     FILESIZE=$((FILESIZE * 8))
     bclone_test random $FILESIZE false $TESTSRCDIR $TESTSRCDIR
 done
@@ -52,9 +54,11 @@ for srcprop in "${sync_prop_vals[@]}"; do
     log_must zfs set sync=$srcprop $TESTSRCFS
     for dstprop in "${sync_prop_vals[@]}"; do
         log_must zfs set sync=$dstprop $TESTDSTFS
+        # 15*8=120, which is greater than 113, so we are sure the data won't
+        # be embedded into BP.
         # 32767*8=262136, which is larger than a single default recordsize of
         # 131072.
-        FILESIZE=$(random_int_between 1 32767)
+        FILESIZE=$(random_int_between 15 32767)
         FILESIZE=$((FILESIZE * 8))
         bclone_test random $FILESIZE false $TESTSRCDIR $TESTDSTDIR
     done

sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zpool_status/zpool_status_008_pos.ksh

@@ -69,15 +69,16 @@ for raid_type in "draid2:3d:6c:1s" "raidz2"; do
 	log_mustnot eval "zpool status -e $TESTPOOL2 | grep ONLINE"
 
 	# Check no ONLINE slow vdevs are show.  Then mark IOs greater than
-	# 160ms slow, delay IOs 320ms to vdev6, check slow IOs.
+	# 750ms slow, delay IOs 1000ms to vdev6, check slow IOs.
 	log_must check_vdev_state $TESTPOOL2 $TESTDIR/vdev6 "ONLINE"
 	log_mustnot eval "zpool status -es $TESTPOOL2 | grep ONLINE"
 
-	log_must set_tunable64 ZIO_SLOW_IO_MS 160
-	log_must zinject -d $TESTDIR/vdev6 -D320:100 $TESTPOOL2
+	log_must set_tunable64 ZIO_SLOW_IO_MS 750
+	log_must zinject -d $TESTDIR/vdev6 -D1000:100 $TESTPOOL2
 	log_must mkfile 1048576 /$TESTPOOL2/testfile
 	sync_pool $TESTPOOL2
 	log_must set_tunable64 ZIO_SLOW_IO_MS $OLD_SLOW_IO
+	log_must zinject -c all
 
 	# Check vdev6 slow IOs are only shown when requested with -s.
 	log_mustnot eval "zpool status -e $TESTPOOL2 | grep $TESTDIR/vdev6 | grep ONLINE"
@@ -95,10 +96,9 @@ for raid_type in "draid2:3d:6c:1s" "raidz2"; do
 	log_mustnot eval "zpool status -es $TESTPOOL2 | grep $TESTDIR/vdev2 | grep ONLINE"
 	log_mustnot eval "zpool status -es $TESTPOOL2 | grep $TESTDIR/vdev3 | grep ONLINE"
 
-	log_must zinject -c all
 	log_must zpool status -es $TESTPOOL2
 
-	zpool destroy $TESTPOOL2
+	log_must zpool destroy $TESTPOOL2
 done
 
 log_pass "Verify zpool status -e shows only unhealthy vdevs"

sys/dev/e1000/e1000_phy.c

@@ -1707,9 +1707,10 @@ s32 e1000_setup_copper_link_generic(struct e1000_hw *hw)
 		 * autonegotiation.
 		 */
 		ret_val = e1000_copper_link_autoneg(hw);
-		if (ret_val)
+		if (ret_val && !hw->mac.forced_speed_duplex)
 			return ret_val;
-	} else {
+	}
+	if (!hw->mac.autoneg || (ret_val && hw->mac.forced_speed_duplex)) {
 		/* PHY will be set to 10H, 10F, 100H or 100F
 		 * depending on user settings.
 		 */

sys/dev/e1000/em_txrx.c

@@ -108,16 +108,19 @@ em_dump_rs(struct e1000_softc *sc)
 			cur = txr->tx_rsq[rs_cidx];
 			status = txr->tx_base[cur].upper.fields.status;
 			if (!(status & E1000_TXD_STAT_DD))
-				printf("qid[%d]->tx_rsq[%d]: %d clear ", qid, rs_cidx, cur);
+				printf("qid[%d]->tx_rsq[%d]: %d clear ",
+				    qid, rs_cidx, cur);
 		} else {
 			rs_cidx = (rs_cidx-1)&(ntxd-1);
 			cur = txr->tx_rsq[rs_cidx];
-			printf("qid[%d]->tx_rsq[rs_cidx-1=%d]: %d  ", qid, rs_cidx, cur);
+			printf("qid[%d]->tx_rsq[rs_cidx-1=%d]: %d  ",
+			    qid, rs_cidx, cur);
 		}
 		printf("cidx_prev=%d rs_pidx=%d ",txr->tx_cidx_processed,
 		    txr->tx_rs_pidx);
 		for (i = 0; i < ntxd; i++) {
-			if (txr->tx_base[i].upper.fields.status & E1000_TXD_STAT_DD)
+			if (txr->tx_base[i].upper.fields.status &
+			    E1000_TXD_STAT_DD)
 				printf("%d set ", i);
 		}
 		printf("\n");
@@ -143,8 +146,8 @@ em_tso_setup(struct e1000_softc *sc, if_pkt_info_t pi, uint32_t *txd_upper,
 
 	hdr_len = pi->ipi_ehdrlen + pi->ipi_ip_hlen + pi->ipi_tcp_hlen;
 	*txd_lower = (E1000_TXD_CMD_DEXT |	/* Extended descr type */
-		      E1000_TXD_DTYP_D |	/* Data descr type */
-		      E1000_TXD_CMD_TSE);	/* Do TSE on this packet */
+	    E1000_TXD_DTYP_D |			/* Data descr type */
+	    E1000_TXD_CMD_TSE);			/* Do TSE on this packet */
 
 	cur = pi->ipi_pidx;
 	TXD = (struct e1000_context_desc *)&txr->tx_base[cur];
@@ -157,7 +160,8 @@ em_tso_setup(struct e1000_softc *sc, if_pkt_info_t pi, uint32_t *txd_upper,
 	switch(pi->ipi_etype) {
 	case ETHERTYPE_IP:
 		/* IP and/or TCP header checksum calculation and insertion. */
-		*txd_upper = (E1000_TXD_POPTS_IXSM | E1000_TXD_POPTS_TXSM) << 8;
+		*txd_upper =
+		    (E1000_TXD_POPTS_IXSM | E1000_TXD_POPTS_TXSM) << 8;
 
 		TXD->lower_setup.ip_fields.ipcse =
 		    htole16(pi->ipi_ehdrlen + pi->ipi_ip_hlen - 1);
@@ -183,7 +187,8 @@ em_tso_setup(struct e1000_softc *sc, if_pkt_info_t pi, uint32_t *txd_upper,
 	TXD->upper_setup.tcp_fields.tucss = pi->ipi_ehdrlen + pi->ipi_ip_hlen;
 	TXD->upper_setup.tcp_fields.tucse = 0;
 	TXD->upper_setup.tcp_fields.tucso =
-	    pi->ipi_ehdrlen + pi->ipi_ip_hlen + offsetof(struct tcphdr, th_sum);
+	    pi->ipi_ehdrlen + pi->ipi_ip_hlen +
+	    offsetof(struct tcphdr, th_sum);
 
 	/*
 	 * Payload size per packet w/o any headers.
@@ -211,8 +216,8 @@ em_tso_setup(struct e1000_softc *sc, if_pkt_info_t pi, uint32_t *txd_upper,
 	if (++cur == scctx->isc_ntxd[0]) {
 		cur = 0;
 	}
-	DPRINTF(iflib_get_dev(sc->ctx), "%s: pidx: %d cur: %d\n", __FUNCTION__,
-	    pi->ipi_pidx, cur);
+	DPRINTF(iflib_get_dev(sc->ctx), "%s: pidx: %d cur: %d\n",
+	    __FUNCTION__, pi->ipi_pidx, cur);
 	return (cur);
 }
 
@@ -277,8 +282,8 @@ em_transmit_checksum_setup(struct e1000_softc *sc, if_pkt_info_t pi,
 	 * ipcse - End offset for header checksum calculation.
 	 * ipcso - Offset of place to put the checksum.
 	 *
-	 * We set ipcsX values regardless of IP version to work around HW issues
-	 * and ipcse must be 0 for IPv6 per "PCIe GbE SDM 2.5" page 61.
+	 * We set ipcsX values regardless of IP version to work around HW
+	 * issues and ipcse must be 0 for IPv6 per "PCIe GbE SDM 2.5" page 61.
 	 * IXSM controls whether it's inserted.
 	 */
 	TXD->lower_setup.ip_fields.ipcss = pi->ipi_ehdrlen;
@@ -296,7 +301,8 @@ em_transmit_checksum_setup(struct e1000_softc *sc, if_pkt_info_t pi,
 	 * tucse - End offset for payload checksum calculation.
 	 * tucso - Offset of place to put the checksum.
 	 */
-	if (csum_flags & (CSUM_TCP | CSUM_UDP | CSUM_IP6_TCP | CSUM_IP6_UDP)) {
+	if (csum_flags & (CSUM_TCP | CSUM_UDP | CSUM_IP6_TCP |
+	    CSUM_IP6_UDP)) {
 		uint8_t tucso;
 
 		*txd_upper |= E1000_TXD_POPTS_TXSM << 8;
@@ -326,7 +332,8 @@ em_transmit_checksum_setup(struct e1000_softc *sc, if_pkt_info_t pi,
 		cur = 0;
 	}
 	DPRINTF(iflib_get_dev(sc->ctx),
-	    "checksum_setup csum_flags=%x txd_upper=%x txd_lower=%x hdr_len=%d cmd=%x\n",
+	    "checksum_setup csum_flags=%x txd_upper=%x txd_lower=%x"
+	    " hdr_len=%d cmd=%x\n",
 	    csum_flags, *txd_upper, *txd_lower, hdr_len, cmd);
 	return (cur);
 }
@@ -372,7 +379,8 @@ em_isc_txd_encap(void *arg, if_pkt_info_t pi)
 		i = em_tso_setup(sc, pi, &txd_upper, &txd_lower);
 		tso_desc = true;
 	} else if (csum_flags & EM_CSUM_OFFLOAD) {
-		i = em_transmit_checksum_setup(sc, pi, &txd_upper, &txd_lower);
+		i = em_transmit_checksum_setup(sc, pi, &txd_upper,
+		    &txd_lower);
 	}
 
 	if (pi->ipi_mflags & M_VLANTAG) {
@@ -414,7 +422,8 @@ em_isc_txd_encap(void *arg, if_pkt_info_t pi)
 			/* Now make the sentinel */
 			ctxd = &txr->tx_base[i];
 			ctxd->buffer_addr = htole64(seg_addr + seg_len);
-			ctxd->lower.data = htole32(cmd | txd_lower | TSO_WORKAROUND);
+			ctxd->lower.data =
+			    htole32(cmd | txd_lower | TSO_WORKAROUND);
 			ctxd->upper.data = htole32(txd_upper);
 			pidx_last = i;
 			if (++i == scctx->isc_ntxd[0])
@@ -429,7 +438,8 @@ em_isc_txd_encap(void *arg, if_pkt_info_t pi)
 			pidx_last = i;
 			if (++i == scctx->isc_ntxd[0])
 				i = 0;
-			DPRINTF(iflib_get_dev(sc->ctx), "pidx_last=%d i=%d ntxd[0]=%d\n",
+			DPRINTF(iflib_get_dev(sc->ctx),
+			    "pidx_last=%d i=%d ntxd[0]=%d\n",
 			    pidx_last, i, scctx->isc_ntxd[0]);
 		}
 	}
@@ -449,7 +459,8 @@ em_isc_txd_encap(void *arg, if_pkt_info_t pi)
 	}
 	ctxd->lower.data |= htole32(E1000_TXD_CMD_EOP | txd_flags);
 	DPRINTF(iflib_get_dev(sc->ctx),
-	    "tx_buffers[%d]->eop = %d ipi_new_pidx=%d\n", first, pidx_last, i);
+	    "tx_buffers[%d]->eop = %d ipi_new_pidx=%d\n",
+	    first, pidx_last, i);
 	pi->ipi_new_pidx = i;
 
 	/* Sent data accounting for AIM */
@@ -508,8 +519,8 @@ em_isc_txd_credits_update(void *arg, uint16_t txqid, bool clear)
 			delta += ntxd;
 		MPASS(delta > 0);
 		DPRINTF(iflib_get_dev(sc->ctx),
-			      "%s: cidx_processed=%u cur=%u clear=%d delta=%d\n",
-			      __FUNCTION__, prev, cur, clear, delta);
+		    "%s: cidx_processed=%u cur=%u clear=%d delta=%d\n",
+		    __FUNCTION__, prev, cur, clear, delta);
 
 		processed += delta;
 		prev  = cur;
@@ -699,7 +710,8 @@ lem_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 
 	if (scctx->isc_capenable & IFCAP_VLAN_HWTAGGING &&
 	    status & E1000_RXD_STAT_VP) {
-		ri->iri_vtag = le16toh(rxd->special & E1000_RXD_SPC_VLAN_MASK);
+		ri->iri_vtag =
+		    le16toh(rxd->special & E1000_RXD_SPC_VLAN_MASK);
 		ri->iri_flags |= M_VLANTAG;
 	}
 
@@ -789,7 +801,8 @@ em_receive_checksum(uint16_t status, uint8_t errors, if_rxd_info_t ri)
 		return;
 
 	/* If there is a layer 3 or 4 error we are done */
-	if (__predict_false(errors & (E1000_RXD_ERR_IPE | E1000_RXD_ERR_TCPE)))
+	if (__predict_false(errors & (E1000_RXD_ERR_IPE |
+	    E1000_RXD_ERR_TCPE)))
 		return;
 
 	/* IP Checksum Good */

sys/dev/e1000/igb_txrx.c

@@ -102,14 +102,15 @@ igb_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
 		break;
 	default:
 		panic("%s: CSUM_TSO but no supported IP version (0x%04x)",
-		      __func__, ntohs(pi->ipi_etype));
+		    __func__, ntohs(pi->ipi_etype));
 		break;
 	}
 
-	TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[pi->ipi_pidx];
+	TXD = (struct e1000_adv_tx_context_desc *)&txr->tx_base[pi->ipi_pidx];
 
 	/* This is used in the transmit desc in encap */
-	paylen = pi->ipi_len - pi->ipi_ehdrlen - pi->ipi_ip_hlen - pi->ipi_tcp_hlen;
+	paylen = pi->ipi_len - pi->ipi_ehdrlen - pi->ipi_ip_hlen -
+	    pi->ipi_tcp_hlen;
 
 	/* VLAN MACLEN IPLEN */
 	if (pi->ipi_mflags & M_VLANTAG) {
@@ -147,8 +148,8 @@ igb_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
  *
  **********************************************************************/
 static int
-igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
-    uint32_t *olinfo_status)
+igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi,
+    uint32_t *cmd_type_len, uint32_t *olinfo_status)
 {
 	struct e1000_adv_tx_context_desc *TXD;
 	struct e1000_softc *sc = txr->sc;
@@ -164,7 +165,7 @@ igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
 	*olinfo_status |= pi->ipi_len << E1000_ADVTXD_PAYLEN_SHIFT;
 
 	/* Now ready a context descriptor */
-	TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[pi->ipi_pidx];
+	TXD = (struct e1000_adv_tx_context_desc *)&txr->tx_base[pi->ipi_pidx];
 
 	/*
 	** In advanced descriptors the vlan tag must
@@ -246,8 +247,8 @@ igb_isc_txd_encap(void *arg, if_pkt_info_t pi)
 
 	pidx_last = olinfo_status = 0;
 	/* Basic descriptor defines */
-	cmd_type_len = (E1000_ADVTXD_DTYP_DATA |
-			E1000_ADVTXD_DCMD_IFCS | E1000_ADVTXD_DCMD_DEXT);
+	cmd_type_len = (E1000_ADVTXD_DTYP_DATA | E1000_ADVTXD_DCMD_IFCS |
+	    E1000_ADVTXD_DCMD_DEXT);
 
 	if (pi->ipi_mflags & M_VLANTAG)
 		cmd_type_len |= E1000_ADVTXD_DCMD_VLE;
@@ -300,9 +301,9 @@ igb_isc_txd_encap(void *arg, if_pkt_info_t pi)
 static void
 igb_isc_txd_flush(void *arg, uint16_t txqid, qidx_t pidx)
 {
-	struct e1000_softc *sc	= arg;
-	struct em_tx_queue *que	= &sc->tx_queues[txqid];
-	struct tx_ring *txr	= &que->txr;
+	struct e1000_softc *sc = arg;
+	struct em_tx_queue *que = &sc->tx_queues[txqid];
+	struct tx_ring *txr = &que->txr;
 
 	E1000_WRITE_REG(&sc->hw, E1000_TDT(txr->me), pidx);
 }
@@ -351,7 +352,8 @@ igb_isc_txd_credits_update(void *arg, uint16_t txqid, bool clear)
 		if (rs_cidx  == txr->tx_rs_pidx)
 			break;
 		cur = txr->tx_rsq[rs_cidx];
-		status = ((union e1000_adv_tx_desc *)&txr->tx_base[cur])->wb.status;
+		status = ((union e1000_adv_tx_desc *)
+		    &txr->tx_base[cur])->wb.status;
 	} while ((status & E1000_TXD_STAT_DD));
 
 	txr->tx_rs_cidx = rs_cidx;
@@ -387,7 +389,8 @@ igb_isc_rxd_refill(void *arg, if_rxd_update_t iru)
 }
 
 static void
-igb_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, qidx_t pidx)
+igb_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused,
+    qidx_t pidx)
 {
 	struct e1000_softc *sc = arg;
 	struct em_rx_queue *que = &sc->rx_queues[rxqid];
@@ -453,7 +456,8 @@ igb_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 		MPASS ((staterr & E1000_RXD_STAT_DD) != 0);
 
 		len = le16toh(rxd->wb.upper.length);
-		ptype = le32toh(rxd->wb.lower.lo_dword.data) &  IGB_PKTTYPE_MASK;
+		ptype =
+		    le32toh(rxd->wb.lower.lo_dword.data) &  IGB_PKTTYPE_MASK;
 
 		ri->iri_len += len;
 		rxr->rx_bytes += ri->iri_len;
@@ -462,7 +466,8 @@ igb_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 		eop = ((staterr & E1000_RXD_STAT_EOP) == E1000_RXD_STAT_EOP);
 
 		/* Make sure bad packets are discarded */
-		if (eop && ((staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) != 0)) {
+		if (eop &&
+		    ((staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) != 0)) {
 			sc->dropped_pkts++;
 			++rxr->rx_discarded;
 			return (EBADMSG);
@@ -524,7 +529,8 @@ igb_rx_checksum(uint32_t staterr, if_rxd_info_t ri, uint32_t ptype)
 		return;
 
 	/* If there is a layer 3 or 4 error we are done */
-	if (__predict_false(errors & (E1000_RXD_ERR_IPE | E1000_RXD_ERR_TCPE)))
+	if (__predict_false(errors &
+	    (E1000_RXD_ERR_IPE | E1000_RXD_ERR_TCPE)))
 		return;
 
 	/* IP Checksum Good */
@@ -535,11 +541,13 @@ igb_rx_checksum(uint32_t staterr, if_rxd_info_t ri, uint32_t ptype)
 	if (__predict_true(status &
 	    (E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS))) {
 		/* SCTP header present */
-		if (__predict_false((ptype & E1000_RXDADV_PKTTYPE_ETQF) == 0 &&
+		if (__predict_false(
+		    (ptype & E1000_RXDADV_PKTTYPE_ETQF) == 0 &&
 		    (ptype & E1000_RXDADV_PKTTYPE_SCTP) != 0)) {
 			ri->iri_csum_flags |= CSUM_SCTP_VALID;
 		} else {
-			ri->iri_csum_flags |= CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
+			ri->iri_csum_flags |=
+			    CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
 			ri->iri_csum_data = htons(0xffff);
 		}
 	}

sys/dev/efidev/efirt.c

@@ -309,6 +309,9 @@ efi_enter(void)
 		fpu_kern_leave(td, NULL);
 		mtx_unlock(&efi_lock);
 		PMAP_UNLOCK(curpmap);
+	} else {
+		MPASS((td->td_pflags & TDP_EFIRT) == 0);
+		td->td_pflags |= TDP_EFIRT;
 	}
 	return (error);
 }
@@ -319,10 +322,13 @@ efi_leave(void)
 	struct thread *td;
 	pmap_t curpmap;
 
+	td = curthread;
+	MPASS((td->td_pflags & TDP_EFIRT) != 0);
+	td->td_pflags &= ~TDP_EFIRT;
+
 	efi_arch_leave();
 
 	curpmap = &curproc->p_vmspace->vm_pmap;
-	td = curthread;
 	fpu_kern_leave(td, NULL);
 	mtx_unlock(&efi_lock);
 	PMAP_UNLOCK(curpmap);

sys/dev/ena/ena_sysctl.c

@@ -158,7 +158,7 @@ SYSCTL_INT(_hw_ena, OID_AUTO, enable_9k_mbufs, CTLFLAG_RDTUN,
 int ena_force_large_llq_header = ENA_LLQ_HEADER_SIZE_POLICY_DEFAULT;
 SYSCTL_INT(_hw_ena, OID_AUTO, force_large_llq_header, CTLFLAG_RDTUN,
     &ena_force_large_llq_header, 0,
-    "Change default LLQ entry size received from the device\n");
+    "Change default LLQ entry size received from the device");
 
 int ena_rss_table_size = ENA_RX_RSS_TABLE_SIZE;
 

sys/dev/hid/hms.c

@@ -122,6 +122,7 @@ struct hms_softc {
 	hid_size_t		isize;
 	uint32_t		drift_cnt;
 	uint32_t		drift_thresh;
+	struct hid_location	wheel_loc;
 #endif
 };
 
@@ -131,6 +132,7 @@ hms_intr(void *context, void *buf, hid_size_t len)
 {
 	struct hidmap *hm = context;
 	struct hms_softc *sc = device_get_softc(hm->dev);
+	int32_t wheel;
 
 	if (len > sc->isize)
 		len = sc->isize;
@@ -140,8 +142,18 @@ hms_intr(void *context, void *buf, hid_size_t len)
 	 * to return last report data in sampling mode even after touch has
 	 * been ended.  That results in cursor drift.  Filter out such a
 	 * reports through comparing with previous one.
+	 *
+	 * Except this results in dropping consecutive mouse wheel events,
+	 * because differently from cursor movement they always move by the
+	 * same amount.  So, don't do it when there's mouse wheel movement.
 	 */
-	if (len == sc->last_irsize && memcmp(buf, sc->last_ir, len) == 0) {
+	if (sc->wheel_loc.size != 0)
+		wheel = hid_get_data(buf, len, &sc->wheel_loc);
+	else
+		wheel = 0;
+
+	if (len == sc->last_irsize && memcmp(buf, sc->last_ir, len) == 0 &&
+	    wheel == 0) {
 		sc->drift_cnt++;
 		if (sc->drift_thresh != 0 && sc->drift_cnt >= sc->drift_thresh)
 			return;
@@ -285,9 +297,25 @@ hms_attach(device_t dev)
 	/* Count number of input usages of variable type mapped to buttons */
 	for (hi = sc->hm.hid_items;
 	     hi < sc->hm.hid_items + sc->hm.nhid_items;
-	     hi++)
+	     hi++) {
 		if (hi->type == HIDMAP_TYPE_VARIABLE && hi->evtype == EV_KEY)
 			nbuttons++;
+#ifdef IICHID_SAMPLING
+		/*
+		 * Make note of which part of the report descriptor is the wheel.
+		 */
+		if (hi->type == HIDMAP_TYPE_VARIABLE &&
+		    hi->evtype == EV_REL && hi->code == REL_WHEEL) {
+			sc->wheel_loc = hi->loc;
+			/*
+			 * Account for the leading Report ID byte
+			 * if it is a multi-report device.
+			 */
+			if (hi->id != 0)
+				sc->wheel_loc.pos += 8;
+		}
+#endif
+	}
 
 	/* announce information about the mouse */
 	device_printf(dev, "%d buttons and [%s%s%s%s%s] coordinates ID=%u\n",

sys/dev/igc/if_igc.c

@@ -49,22 +49,38 @@
 static const pci_vendor_info_t igc_vendor_info_array[] =
 {
 	/* Intel(R) PRO/1000 Network Connection - igc */
-	PVID(0x8086, IGC_DEV_ID_I225_LM, "Intel(R) Ethernet Controller I225-LM"),
-	PVID(0x8086, IGC_DEV_ID_I225_V, "Intel(R) Ethernet Controller I225-V"),
-	PVID(0x8086, IGC_DEV_ID_I225_K, "Intel(R) Ethernet Controller I225-K"),
-	PVID(0x8086, IGC_DEV_ID_I225_I, "Intel(R) Ethernet Controller I225-I"),
-	PVID(0x8086, IGC_DEV_ID_I220_V, "Intel(R) Ethernet Controller I220-V"),
-	PVID(0x8086, IGC_DEV_ID_I225_K2, "Intel(R) Ethernet Controller I225-K(2)"),
-	PVID(0x8086, IGC_DEV_ID_I225_LMVP, "Intel(R) Ethernet Controller I225-LMvP(2)"),
-	PVID(0x8086, IGC_DEV_ID_I226_K, "Intel(R) Ethernet Controller I226-K"),
-	PVID(0x8086, IGC_DEV_ID_I226_LMVP, "Intel(R) Ethernet Controller I226-LMvP"),
-	PVID(0x8086, IGC_DEV_ID_I225_IT, "Intel(R) Ethernet Controller I225-IT(2)"),
-	PVID(0x8086, IGC_DEV_ID_I226_LM, "Intel(R) Ethernet Controller I226-LM"),
-	PVID(0x8086, IGC_DEV_ID_I226_V, "Intel(R) Ethernet Controller I226-V"),
-	PVID(0x8086, IGC_DEV_ID_I226_IT, "Intel(R) Ethernet Controller I226-IT"),
-	PVID(0x8086, IGC_DEV_ID_I221_V, "Intel(R) Ethernet Controller I221-V"),
-	PVID(0x8086, IGC_DEV_ID_I226_BLANK_NVM, "Intel(R) Ethernet Controller I226(blankNVM)"),
-	PVID(0x8086, IGC_DEV_ID_I225_BLANK_NVM, "Intel(R) Ethernet Controller I225(blankNVM)"),
+	PVID(0x8086, IGC_DEV_ID_I225_LM,
+	    "Intel(R) Ethernet Controller I225-LM"),
+	PVID(0x8086, IGC_DEV_ID_I225_V,
+	    "Intel(R) Ethernet Controller I225-V"),
+	PVID(0x8086, IGC_DEV_ID_I225_K,
+	    "Intel(R) Ethernet Controller I225-K"),
+	PVID(0x8086, IGC_DEV_ID_I225_I,
+	    "Intel(R) Ethernet Controller I225-I"),
+	PVID(0x8086, IGC_DEV_ID_I220_V,
+	    "Intel(R) Ethernet Controller I220-V"),
+	PVID(0x8086, IGC_DEV_ID_I225_K2,
+	    "Intel(R) Ethernet Controller I225-K(2)"),
+	PVID(0x8086, IGC_DEV_ID_I225_LMVP,
+	    "Intel(R) Ethernet Controller I225-LMvP(2)"),
+	PVID(0x8086, IGC_DEV_ID_I226_K,
+	    "Intel(R) Ethernet Controller I226-K"),
+	PVID(0x8086, IGC_DEV_ID_I226_LMVP,
+	    "Intel(R) Ethernet Controller I226-LMvP"),
+	PVID(0x8086, IGC_DEV_ID_I225_IT,
+	    "Intel(R) Ethernet Controller I225-IT(2)"),
+	PVID(0x8086, IGC_DEV_ID_I226_LM,
+	    "Intel(R) Ethernet Controller I226-LM"),
+	PVID(0x8086, IGC_DEV_ID_I226_V,
+	    "Intel(R) Ethernet Controller I226-V"),
+	PVID(0x8086, IGC_DEV_ID_I226_IT,
+	    "Intel(R) Ethernet Controller I226-IT"),
+	PVID(0x8086, IGC_DEV_ID_I221_V,
+	    "Intel(R) Ethernet Controller I221-V"),
+	PVID(0x8086, IGC_DEV_ID_I226_BLANK_NVM,
+	    "Intel(R) Ethernet Controller I226(blankNVM)"),
+	PVID(0x8086, IGC_DEV_ID_I225_BLANK_NVM,
+	    "Intel(R) Ethernet Controller I225(blankNVM)"),
 	/* required last entry */
 	PVID_END
 };
@@ -80,8 +96,10 @@ static int	igc_if_shutdown(if_ctx_t);
 static int	igc_if_suspend(if_ctx_t);
 static int	igc_if_resume(if_ctx_t);
 
-static int	igc_if_tx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int, int);
-static int	igc_if_rx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int, int);
+static int	igc_if_tx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int,
+    int);
+static int	igc_if_rx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int,
+    int);
 static void	igc_if_queues_free(if_ctx_t);
 
 static uint64_t	igc_if_get_counter(if_ctx_t, ift_counter);
@@ -229,11 +247,12 @@ SYSCTL_INT(_hw_igc, OID_AUTO, disable_crc_stripping, CTLFLAG_RDTUN,
     &igc_disable_crc_stripping, 0, "Disable CRC Stripping");
 
 static int igc_smart_pwr_down = false;
-SYSCTL_INT(_hw_igc, OID_AUTO, smart_pwr_down, CTLFLAG_RDTUN, &igc_smart_pwr_down,
+SYSCTL_INT(_hw_igc, OID_AUTO, smart_pwr_down, CTLFLAG_RDTUN,
+    &igc_smart_pwr_down,
     0, "Set to true to leave smart power down enabled on newer adapters");
 
 /* Controls whether promiscuous also shows bad packets */
-static int igc_debug_sbp = true;
+static int igc_debug_sbp = false;
 SYSCTL_INT(_hw_igc, OID_AUTO, sbp, CTLFLAG_RDTUN, &igc_debug_sbp, 0,
     "Show bad packets in promiscuous mode");
 
@@ -277,7 +296,8 @@ static struct if_shared_ctx igc_sctx_init = {
 	.isc_vendor_info = igc_vendor_info_array,
 	.isc_driver_version = "1",
 	.isc_driver = &igc_if_driver,
-	.isc_flags = IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP | IFLIB_NEED_ZERO_CSUM,
+	.isc_flags =
+	    IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP | IFLIB_NEED_ZERO_CSUM,
 
 	.isc_nrxd_min = {IGC_MIN_RXD},
 	.isc_ntxd_min = {IGC_MIN_TXD},
@@ -383,15 +403,20 @@ static int igc_get_regs(SYSCTL_HANDLER_ARGS)
 	for (j = 0; j < nrxd; j++) {
 		u32 staterr = le32toh(rxr->rx_base[j].wb.upper.status_error);
 		u32 length =  le32toh(rxr->rx_base[j].wb.upper.length);
-		sbuf_printf(sb, "\tReceive Descriptor Address %d: %08" PRIx64 "  Error:%d  Length:%d\n", j, rxr->rx_base[j].read.buffer_addr, staterr, length);
+		sbuf_printf(sb, "\tReceive Descriptor Address %d: %08"
+		    PRIx64 "  Error:%d  Length:%d\n",
+		    j, rxr->rx_base[j].read.buffer_addr, staterr, length);
 	}
 
 	for (j = 0; j < min(ntxd, 256); j++) {
 		unsigned int *ptr = (unsigned int *)&txr->tx_base[j];
 
-		sbuf_printf(sb, "\tTXD[%03d] [0]: %08x [1]: %08x [2]: %08x [3]: %08x  eop: %d DD=%d\n",
-			    j, ptr[0], ptr[1], ptr[2], ptr[3], buf->eop,
-			    buf->eop != -1 ? txr->tx_base[buf->eop].upper.fields.status & IGC_TXD_STAT_DD : 0);
+		sbuf_printf(sb, "\tTXD[%03d] [0]: %08x [1]: %08x [2]: %08x"
+		    "[3]: %08x  eop: %d DD=%d\n",
+		    j, ptr[0], ptr[1], ptr[2], ptr[3], buf->eop,
+		    buf->eop != -1 ?
+		    txr->tx_base[buf->eop].upper.fields.status &
+		    IGC_TXD_STAT_DD : 0);
 
 	}
 	}
@@ -523,13 +548,16 @@ igc_if_attach_pre(if_ctx_t ctx)
 	igc_identify_hardware(ctx);
 
 	scctx->isc_tx_nsegments = IGC_MAX_SCATTER;
-	scctx->isc_nrxqsets_max = scctx->isc_ntxqsets_max = igc_set_num_queues(ctx);
+	scctx->isc_nrxqsets_max =
+	    scctx->isc_ntxqsets_max = igc_set_num_queues(ctx);
 	if (bootverbose)
 		device_printf(dev, "attach_pre capping queues at %d\n",
 		    scctx->isc_ntxqsets_max);
 
-	scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0] * sizeof(union igc_adv_tx_desc), IGC_DBA_ALIGN);
-	scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] * sizeof(union igc_adv_rx_desc), IGC_DBA_ALIGN);
+	scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0] *
+	    sizeof(union igc_adv_tx_desc), IGC_DBA_ALIGN);
+	scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] *
+	    sizeof(union igc_adv_rx_desc), IGC_DBA_ALIGN);
 	scctx->isc_txd_size[0] = sizeof(union igc_adv_tx_desc);
 	scctx->isc_rxd_size[0] = sizeof(union igc_adv_rx_desc);
 	scctx->isc_txrx = &igc_txrx;
@@ -588,7 +616,8 @@ igc_if_attach_pre(if_ctx_t ctx)
 	sc->mta = malloc(sizeof(u8) * ETHER_ADDR_LEN *
 	    MAX_NUM_MULTICAST_ADDRESSES, M_DEVBUF, M_NOWAIT);
 	if (sc->mta == NULL) {
-		device_printf(dev, "Can not allocate multicast setup array\n");
+		device_printf(dev,
+		    "Can not allocate multicast setup array\n");
 		error = ENOMEM;
 		goto err_late;
 	}
@@ -632,7 +661,7 @@ igc_if_attach_pre(if_ctx_t ctx)
 	/* Copy the permanent MAC address out of the EEPROM */
 	if (igc_read_mac_addr(hw) < 0) {
 		device_printf(dev, "EEPROM read error while reading MAC"
-			      " address\n");
+		    " address\n");
 		error = EIO;
 		goto err_late;
 	}
@@ -772,7 +801,7 @@ igc_if_mtu_set(if_ctx_t ctx, uint32_t mtu)
 	struct igc_softc *sc = iflib_get_softc(ctx);
 	if_softc_ctx_t scctx = iflib_get_softc_ctx(ctx);
 
-	 IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)");
+	IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)");
 
 	 /* 9K Jumbo Frame size */
 	 max_frame_size = 9234;
@@ -817,7 +846,8 @@ igc_if_init(if_ctx_t ctx)
 	igc_reset(ctx);
 	igc_if_update_admin_status(ctx);
 
-	for (i = 0, tx_que = sc->tx_queues; i < sc->tx_num_queues; i++, tx_que++) {
+	for (i = 0, tx_que = sc->tx_queues; i < sc->tx_num_queues;
+	    i++, tx_que++) {
 		struct tx_ring *txr = &tx_que->txr;
 
 		txr->tx_rs_cidx = txr->tx_rs_pidx;
@@ -906,7 +936,7 @@ igc_neweitr(struct igc_softc *sc, struct igc_rx_queue *que,
 			goto igc_set_next_eitr;
 		}
 
-		/* Get the largest values from the associated tx and rx ring */
+		/* Get largest values from the associated tx and rx ring */
 		if (txr->tx_bytes && txr->tx_packets) {
 			bytes = txr->tx_bytes;
 			bytes_packets = txr->tx_bytes/txr->tx_packets;
@@ -914,7 +944,8 @@ igc_neweitr(struct igc_softc *sc, struct igc_rx_queue *que,
 		}
 		if (rxr->rx_bytes && rxr->rx_packets) {
 			bytes = max(bytes, rxr->rx_bytes);
-			bytes_packets = max(bytes_packets, rxr->rx_bytes/rxr->rx_packets);
+			bytes_packets = max(bytes_packets,
+			    rxr->rx_bytes/rxr->rx_packets);
 			packets = max(packets, rxr->rx_packets);
 		}
 
@@ -935,7 +966,8 @@ igc_neweitr(struct igc_softc *sc, struct igc_rx_queue *que,
 				/* Handle TSO */
 				if (bytes_packets > 8000)
 					nextlatency = eitr_latency_bulk;
-				else if ((packets < 10) || (bytes_packets > 1200))
+				else if ((packets < 10) ||
+				    (bytes_packets > 1200))
 					nextlatency = eitr_latency_bulk;
 				else if (packets > 35)
 					nextlatency = eitr_latency_lowest;
@@ -954,7 +986,8 @@ igc_neweitr(struct igc_softc *sc, struct igc_rx_queue *que,
 			break;
 		default:
 			nextlatency = eitr_latency_low;
-			device_printf(sc->dev, "Unexpected neweitr transition %d\n",
+			device_printf(sc->dev,
+			    "Unexpected neweitr transition %d\n",
 			    nextlatency);
 			break;
 		}
@@ -1170,13 +1203,13 @@ igc_if_media_status(if_ctx_t ctx, struct ifmediareq *ifmr)
 		break;
 	case 100:
 		ifmr->ifm_active |= IFM_100_TX;
-                break;
+		break;
 	case 1000:
 		ifmr->ifm_active |= IFM_1000_T;
 		break;
 	case 2500:
-                ifmr->ifm_active |= IFM_2500_T;
-                break;
+		ifmr->ifm_active |= IFM_2500_T;
+		break;
 	}
 
 	if (sc->link_duplex == FULL_DUPLEX)
@@ -1210,9 +1243,9 @@ igc_if_media_change(if_ctx_t ctx)
 	case IFM_AUTO:
 		sc->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT;
 		break;
-        case IFM_2500_T:
-                sc->hw.phy.autoneg_advertised = ADVERTISE_2500_FULL;
-                break;
+	case IFM_2500_T:
+		sc->hw.phy.autoneg_advertised = ADVERTISE_2500_FULL;
+		break;
 	case IFM_1000_T:
 		sc->hw.phy.autoneg_advertised = ADVERTISE_1000_FULL;
 		break;
@@ -1296,7 +1329,7 @@ igc_if_multi_set(if_ctx_t ctx)
 {
 	struct igc_softc *sc = iflib_get_softc(ctx);
 	if_t ifp = iflib_get_ifp(ctx);
-	u8  *mta; /* Multicast array memory */
+	u8 *mta; /* Multicast array memory */
 	u32 reg_rctl = 0;
 	int mcnt = 0;
 
@@ -1315,10 +1348,10 @@ igc_if_multi_set(if_ctx_t ctx)
 		if (igc_debug_sbp)
 			reg_rctl |= IGC_RCTL_SBP;
 	} else if (mcnt >= MAX_NUM_MULTICAST_ADDRESSES ||
-	      if_getflags(ifp) & IFF_ALLMULTI) {
-                reg_rctl |= IGC_RCTL_MPE;
+		    if_getflags(ifp) & IFF_ALLMULTI) {
+		reg_rctl |= IGC_RCTL_MPE;
 		reg_rctl &= ~IGC_RCTL_UPE;
-        } else
+	} else
 		reg_rctl &= ~(IGC_RCTL_UPE | IGC_RCTL_MPE);
 
 	if (mcnt < MAX_NUM_MULTICAST_ADDRESSES)
@@ -1463,7 +1496,8 @@ igc_allocate_pci_resources(if_ctx_t ctx)
 	sc->memory = bus_alloc_resource_any(dev, SYS_RES_MEMORY,
 	    &rid, RF_ACTIVE);
 	if (sc->memory == NULL) {
-		device_printf(dev, "Unable to allocate bus resource: memory\n");
+		device_printf(dev,
+		    "Unable to allocate bus resource: memory\n");
 		return (ENXIO);
 	}
 	sc->osdep.mem_bus_space_tag = rman_get_bustag(sc->memory);
@@ -1494,9 +1528,12 @@ igc_if_msix_intr_assign(if_ctx_t ctx, int msix)
 	for (i = 0; i < sc->rx_num_queues; i++, rx_que++, vector++) {
 		rid = vector + 1;
 		snprintf(buf, sizeof(buf), "rxq%d", i);
-		error = iflib_irq_alloc_generic(ctx, &rx_que->que_irq, rid, IFLIB_INTR_RXTX, igc_msix_que, rx_que, rx_que->me, buf);
+		error = iflib_irq_alloc_generic(ctx, &rx_que->que_irq, rid,
+		    IFLIB_INTR_RXTX, igc_msix_que, rx_que, rx_que->me, buf);
 		if (error) {
-			device_printf(iflib_get_dev(ctx), "Failed to allocate que int %d err: %d", i, error);
+			device_printf(iflib_get_dev(ctx),
+			    "Failed to allocate que int %d err: %d",
+			    i, error);
 			sc->rx_num_queues = i + 1;
 			goto fail;
 		}
@@ -1534,10 +1571,12 @@ igc_if_msix_intr_assign(if_ctx_t ctx, int msix)
 
 	/* Link interrupt */
 	rid = rx_vectors + 1;
-	error = iflib_irq_alloc_generic(ctx, &sc->irq, rid, IFLIB_INTR_ADMIN, igc_msix_link, sc, 0, "aq");
+	error = iflib_irq_alloc_generic(ctx, &sc->irq, rid, IFLIB_INTR_ADMIN,
+	    igc_msix_link, sc, 0, "aq");
 
 	if (error) {
-		device_printf(iflib_get_dev(ctx), "Failed to register admin handler");
+		device_printf(iflib_get_dev(ctx),
+		    "Failed to register admin handler");
 		goto fail;
 	}
 	sc->linkvec = rx_vectors;
@@ -1662,12 +1701,12 @@ igc_setup_msix(if_ctx_t ctx)
 static void
 igc_init_dmac(struct igc_softc *sc, u32 pba)
 {
-	device_t	dev = sc->dev;
+	device_t dev = sc->dev;
 	struct igc_hw *hw = &sc->hw;
-	u32 		dmac, reg = ~IGC_DMACR_DMAC_EN;
-	u16		hwm;
-	u16		max_frame_size;
-	int		status;
+	u32 dmac, reg = ~IGC_DMACR_DMAC_EN;
+	u16 hwm;
+	u16 max_frame_size;
+	int status;
 
 	max_frame_size = sc->shared->isc_max_frame_size;
 
@@ -1777,7 +1816,8 @@ igc_reset(if_ctx_t ctx)
 	 * response (Rx) to Ethernet PAUSE frames.
 	 * - High water mark should allow for at least two frames to be
 	 *   received after sending an XOFF.
-	 * - Low water mark works best when it is very near the high water mark.
+	 * - Low water mark works best when it is very near the high water
+	 *   mark.
 	 *   This allows the receiver to restart by sending XON when it has
 	 *   drained a bit. Here we use an arbitrary value of 1500 which will
 	 *   restart after one full frame is pulled from the buffer. There
@@ -1957,7 +1997,8 @@ igc_setup_interface(if_ctx_t ctx)
 }
 
 static int
-igc_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxqs, int ntxqsets)
+igc_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs,
+    int ntxqs, int ntxqsets)
 {
 	struct igc_softc *sc = iflib_get_softc(ctx);
 	if_softc_ctx_t scctx = sc->shared;
@@ -1972,7 +2013,8 @@ igc_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxq
 	if (!(sc->tx_queues =
 	    (struct igc_tx_queue *) malloc(sizeof(struct igc_tx_queue) *
 	    sc->tx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) {
-		device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n");
+		device_printf(iflib_get_dev(ctx),
+		    "Unable to allocate queue memory\n");
 		return(ENOMEM);
 	}
 
@@ -1984,14 +2026,16 @@ igc_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxq
 		que->me = txr->me =  i;
 
 		/* Allocate report status array */
-		if (!(txr->tx_rsq = (qidx_t *) malloc(sizeof(qidx_t) * scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) {
-			device_printf(iflib_get_dev(ctx), "failed to allocate rs_idxs memory\n");
+		if (!(txr->tx_rsq = (qidx_t *) malloc(sizeof(qidx_t) *
+		    scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) {
+			device_printf(iflib_get_dev(ctx),
+			    "failed to allocate rs_idxs memory\n");
 			error = ENOMEM;
 			goto fail;
 		}
 		for (j = 0; j < scctx->isc_ntxd[0]; j++)
 			txr->tx_rsq[j] = QIDX_INVALID;
-		/* get the virtual and physical address of the hardware queues */
+		/* get virtual and physical address of the hardware queues */
 		txr->tx_base = (struct igc_tx_desc *)vaddrs[i*ntxqs];
 		txr->tx_paddr = paddrs[i*ntxqs];
 	}
@@ -2006,7 +2050,8 @@ fail:
 }
 
 static int
-igc_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxqs, int nrxqsets)
+igc_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs,
+    int nrxqs, int nrxqsets)
 {
 	struct igc_softc *sc = iflib_get_softc(ctx);
 	int error = IGC_SUCCESS;
@@ -2020,7 +2065,8 @@ igc_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxq
 	if (!(sc->rx_queues =
 	    (struct igc_rx_queue *) malloc(sizeof(struct igc_rx_queue) *
 	    sc->rx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) {
-		device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n");
+		device_printf(iflib_get_dev(ctx),
+		    "Unable to allocate queue memory\n");
 		error = ENOMEM;
 		goto fail;
 	}
@@ -2032,7 +2078,7 @@ igc_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxq
 		rxr->que = que;
 		que->me = rxr->me =  i;
 
-		/* get the virtual and physical address of the hardware queues */
+		/* get virtual and physical address of the hardware queues */
 		rxr->rx_base = (union igc_rx_desc_extended *)vaddrs[i*nrxqs];
 		rxr->rx_paddr = paddrs[i*nrxqs];
 	}
@@ -2137,7 +2183,7 @@ igc_initialize_transmit_unit(if_ctx_t ctx)
 	tctl = IGC_READ_REG(&sc->hw, IGC_TCTL);
 	tctl &= ~IGC_TCTL_CT;
 	tctl |= (IGC_TCTL_PSP | IGC_TCTL_RTLC | IGC_TCTL_EN |
-		   (IGC_COLLISION_THRESHOLD << IGC_CT_SHIFT));
+	    (IGC_COLLISION_THRESHOLD << IGC_CT_SHIFT));
 
 	/* This write will effectively turn on the transmit unit. */
 	IGC_WRITE_REG(&sc->hw, IGC_TCTL, tctl);
@@ -2247,11 +2293,9 @@ igc_initialize_receive_unit(if_ctx_t ctx)
 #endif
 
 		IGC_WRITE_REG(hw, IGC_RDLEN(i),
-			      scctx->isc_nrxd[0] * sizeof(struct igc_rx_desc));
-		IGC_WRITE_REG(hw, IGC_RDBAH(i),
-			      (uint32_t)(bus_addr >> 32));
-		IGC_WRITE_REG(hw, IGC_RDBAL(i),
-			      (uint32_t)bus_addr);
+		      scctx->isc_nrxd[0] * sizeof(struct igc_rx_desc));
+		IGC_WRITE_REG(hw, IGC_RDBAH(i), (uint32_t)(bus_addr >> 32));
+		IGC_WRITE_REG(hw, IGC_RDBAL(i), (uint32_t)bus_addr);
 		IGC_WRITE_REG(hw, IGC_SRRCTL(i), srrctl);
 		/* Setup the Head and Tail Descriptor Pointers */
 		IGC_WRITE_REG(hw, IGC_RDH(i), 0);
@@ -2648,17 +2692,17 @@ igc_add_hw_stats(struct igc_softc *sc)
 
 	/* Driver Statistics */
 	SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "dropped",
-			CTLFLAG_RD, &sc->dropped_pkts,
-			"Driver dropped packets");
+	    CTLFLAG_RD, &sc->dropped_pkts,
+	    "Driver dropped packets");
 	SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "link_irq",
-			CTLFLAG_RD, &sc->link_irq,
-			"Link MSI-X IRQ Handled");
+	    CTLFLAG_RD, &sc->link_irq,
+	    "Link MSI-X IRQ Handled");
 	SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_overruns",
-			CTLFLAG_RD, &sc->rx_overruns,
-			"RX overruns");
+	    CTLFLAG_RD, &sc->rx_overruns,
+	    "RX overruns");
 	SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "watchdog_timeouts",
-			CTLFLAG_RD, &sc->watchdog_events,
-			"Watchdog timeouts");
+	    CTLFLAG_RD, &sc->watchdog_events,
+	    "Watchdog timeouts");
 	SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "device_control",
 	    CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT,
 	    sc, IGC_CTRL, igc_sysctl_reg_handler, "IU",
@@ -2668,11 +2712,11 @@ igc_add_hw_stats(struct igc_softc *sc)
 	    sc, IGC_RCTL, igc_sysctl_reg_handler, "IU",
 	    "Receiver Control Register");
 	SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_high_water",
-			CTLFLAG_RD, &sc->hw.fc.high_water, 0,
-			"Flow Control High Watermark");
+	    CTLFLAG_RD, &sc->hw.fc.high_water, 0,
+	    "Flow Control High Watermark");
 	SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_low_water",
-			CTLFLAG_RD, &sc->hw.fc.low_water, 0,
-			"Flow Control Low Watermark");
+	    CTLFLAG_RD, &sc->hw.fc.low_water, 0,
+	    "Flow Control Low Watermark");
 
 	for (int i = 0; i < sc->tx_num_queues; i++, tx_que++) {
 		struct tx_ring *txr = &tx_que->txr;
@@ -2694,8 +2738,8 @@ igc_add_hw_stats(struct igc_softc *sc)
 		    IGC_TDT(txr->me), igc_sysctl_reg_handler, "IU",
 		    "Transmit Descriptor Tail");
 		SYSCTL_ADD_ULONG(ctx, queue_list, OID_AUTO, "tx_irq",
-				CTLFLAG_RD, &txr->tx_irq,
-				"Queue MSI-X Transmit Interrupts");
+		    CTLFLAG_RD, &txr->tx_irq,
+		    "Queue MSI-X Transmit Interrupts");
 	}
 
 	for (int j = 0; j < sc->rx_num_queues; j++, rx_que++) {
@@ -2707,8 +2751,8 @@ igc_add_hw_stats(struct igc_softc *sc)
 
 		SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "interrupt_rate",
 		    CTLTYPE_UINT | CTLFLAG_RD, rx_que,
-			false, igc_sysctl_interrupt_rate_handler, "IU",
-			"Interrupt Rate");
+		    false, igc_sysctl_interrupt_rate_handler, "IU",
+		    "Interrupt Rate");
 		SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "rxd_head",
 		    CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, sc,
 		    IGC_RDH(rxr->me), igc_sysctl_reg_handler, "IU",
@@ -2718,181 +2762,179 @@ igc_add_hw_stats(struct igc_softc *sc)
 		    IGC_RDT(rxr->me), igc_sysctl_reg_handler, "IU",
 		    "Receive Descriptor Tail");
 		SYSCTL_ADD_ULONG(ctx, queue_list, OID_AUTO, "rx_irq",
-				CTLFLAG_RD, &rxr->rx_irq,
-				"Queue MSI-X Receive Interrupts");
+		    CTLFLAG_RD, &rxr->rx_irq,
+		    "Queue MSI-X Receive Interrupts");
 	}
 
 	/* MAC stats get their own sub node */
-
 	stat_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "mac_stats",
 	    CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Statistics");
 	stat_list = SYSCTL_CHILDREN(stat_node);
 
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "excess_coll",
-			CTLFLAG_RD, &stats->ecol,
-			"Excessive collisions");
+	    CTLFLAG_RD, &stats->ecol,
+	    "Excessive collisions");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "single_coll",
-			CTLFLAG_RD, &stats->scc,
-			"Single collisions");
+	    CTLFLAG_RD, &stats->scc,
+	    "Single collisions");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "multiple_coll",
-			CTLFLAG_RD, &stats->mcc,
-			"Multiple collisions");
+	    CTLFLAG_RD, &stats->mcc,
+	    "Multiple collisions");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "late_coll",
-			CTLFLAG_RD, &stats->latecol,
-			"Late collisions");
+	    CTLFLAG_RD, &stats->latecol,
+	    "Late collisions");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "collision_count",
-			CTLFLAG_RD, &stats->colc,
-			"Collision Count");
+	    CTLFLAG_RD, &stats->colc,
+	    "Collision Count");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "symbol_errors",
-			CTLFLAG_RD, &sc->stats.symerrs,
-			"Symbol Errors");
+	    CTLFLAG_RD, &sc->stats.symerrs,
+	    "Symbol Errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "sequence_errors",
-			CTLFLAG_RD, &sc->stats.sec,
-			"Sequence Errors");
+	    CTLFLAG_RD, &sc->stats.sec,
+	    "Sequence Errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "defer_count",
-			CTLFLAG_RD, &sc->stats.dc,
-			"Defer Count");
+	    CTLFLAG_RD, &sc->stats.dc,
+	    "Defer Count");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "missed_packets",
-			CTLFLAG_RD, &sc->stats.mpc,
-			"Missed Packets");
+	    CTLFLAG_RD, &sc->stats.mpc,
+	    "Missed Packets");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_length_errors",
-			CTLFLAG_RD, &sc->stats.rlec,
-			"Receive Length Errors");
+	    CTLFLAG_RD, &sc->stats.rlec,
+	    "Receive Length Errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_no_buff",
-			CTLFLAG_RD, &sc->stats.rnbc,
-			"Receive No Buffers");
+	    CTLFLAG_RD, &sc->stats.rnbc,
+	    "Receive No Buffers");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_undersize",
-			CTLFLAG_RD, &sc->stats.ruc,
-			"Receive Undersize");
+	    CTLFLAG_RD, &sc->stats.ruc,
+	    "Receive Undersize");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_fragmented",
-			CTLFLAG_RD, &sc->stats.rfc,
-			"Fragmented Packets Received ");
+	    CTLFLAG_RD, &sc->stats.rfc,
+	    "Fragmented Packets Received ");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_oversize",
-			CTLFLAG_RD, &sc->stats.roc,
-			"Oversized Packets Received");
+	    CTLFLAG_RD, &sc->stats.roc,
+	    "Oversized Packets Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_jabber",
-			CTLFLAG_RD, &sc->stats.rjc,
-			"Recevied Jabber");
+	    CTLFLAG_RD, &sc->stats.rjc,
+	    "Recevied Jabber");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_errs",
-			CTLFLAG_RD, &sc->stats.rxerrc,
-			"Receive Errors");
+	    CTLFLAG_RD, &sc->stats.rxerrc,
+	    "Receive Errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "crc_errs",
-			CTLFLAG_RD, &sc->stats.crcerrs,
-			"CRC errors");
+	    CTLFLAG_RD, &sc->stats.crcerrs,
+	    "CRC errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "alignment_errs",
-			CTLFLAG_RD, &sc->stats.algnerrc,
-			"Alignment Errors");
+	    CTLFLAG_RD, &sc->stats.algnerrc,
+	    "Alignment Errors");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_recvd",
-			CTLFLAG_RD, &sc->stats.xonrxc,
-			"XON Received");
+	    CTLFLAG_RD, &sc->stats.xonrxc,
+	    "XON Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_txd",
-			CTLFLAG_RD, &sc->stats.xontxc,
-			"XON Transmitted");
+	    CTLFLAG_RD, &sc->stats.xontxc,
+	    "XON Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_recvd",
-			CTLFLAG_RD, &sc->stats.xoffrxc,
-			"XOFF Received");
+	    CTLFLAG_RD, &sc->stats.xoffrxc,
+	    "XOFF Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_txd",
-			CTLFLAG_RD, &sc->stats.xofftxc,
-			"XOFF Transmitted");
+	    CTLFLAG_RD, &sc->stats.xofftxc,
+	    "XOFF Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "unsupported_fc_recvd",
-			CTLFLAG_RD, &sc->stats.fcruc,
-			"Unsupported Flow Control Received");
+	    CTLFLAG_RD, &sc->stats.fcruc,
+	    "Unsupported Flow Control Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_recvd",
-			CTLFLAG_RD, &sc->stats.mgprc,
-			"Management Packets Received");
+	    CTLFLAG_RD, &sc->stats.mgprc,
+	    "Management Packets Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_drop",
-			CTLFLAG_RD, &sc->stats.mgpdc,
-			"Management Packets Dropped");
+	    CTLFLAG_RD, &sc->stats.mgpdc,
+	    "Management Packets Dropped");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_txd",
-			CTLFLAG_RD, &sc->stats.mgptc,
-			"Management Packets Transmitted");
+	    CTLFLAG_RD, &sc->stats.mgptc,
+	    "Management Packets Transmitted");
 
 	/* Packet Reception Stats */
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_recvd",
-			CTLFLAG_RD, &sc->stats.tpr,
-			"Total Packets Received ");
+	    CTLFLAG_RD, &sc->stats.tpr,
+	    "Total Packets Received ");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_recvd",
-			CTLFLAG_RD, &sc->stats.gprc,
-			"Good Packets Received");
+	    CTLFLAG_RD, &sc->stats.gprc,
+	    "Good Packets Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_recvd",
-			CTLFLAG_RD, &sc->stats.bprc,
-			"Broadcast Packets Received");
+	    CTLFLAG_RD, &sc->stats.bprc,
+	    "Broadcast Packets Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_recvd",
-			CTLFLAG_RD, &sc->stats.mprc,
-			"Multicast Packets Received");
+	    CTLFLAG_RD, &sc->stats.mprc,
+	    "Multicast Packets Received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_64",
-			CTLFLAG_RD, &sc->stats.prc64,
-			"64 byte frames received ");
+	    CTLFLAG_RD, &sc->stats.prc64,
+	    "64 byte frames received ");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_65_127",
-			CTLFLAG_RD, &sc->stats.prc127,
-			"65-127 byte frames received");
+	    CTLFLAG_RD, &sc->stats.prc127,
+	    "65-127 byte frames received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_128_255",
-			CTLFLAG_RD, &sc->stats.prc255,
-			"128-255 byte frames received");
+	    CTLFLAG_RD, &sc->stats.prc255,
+	    "128-255 byte frames received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_256_511",
-			CTLFLAG_RD, &sc->stats.prc511,
-			"256-511 byte frames received");
+	    CTLFLAG_RD, &sc->stats.prc511,
+	    "256-511 byte frames received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_512_1023",
-			CTLFLAG_RD, &sc->stats.prc1023,
-			"512-1023 byte frames received");
+	    CTLFLAG_RD, &sc->stats.prc1023,
+	    "512-1023 byte frames received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_1024_1522",
-			CTLFLAG_RD, &sc->stats.prc1522,
-			"1023-1522 byte frames received");
+	    CTLFLAG_RD, &sc->stats.prc1522,
+	    "1023-1522 byte frames received");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_recvd",
-			CTLFLAG_RD, &sc->stats.gorc,
-			"Good Octets Received");
+	    CTLFLAG_RD, &sc->stats.gorc,
+	    "Good Octets Received");
 
 	/* Packet Transmission Stats */
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_txd",
-			CTLFLAG_RD, &sc->stats.gotc,
-			"Good Octets Transmitted");
+	    CTLFLAG_RD, &sc->stats.gotc,
+	    "Good Octets Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_txd",
-			CTLFLAG_RD, &sc->stats.tpt,
-			"Total Packets Transmitted");
+	    CTLFLAG_RD, &sc->stats.tpt,
+	    "Total Packets Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_txd",
-			CTLFLAG_RD, &sc->stats.gptc,
-			"Good Packets Transmitted");
+	    CTLFLAG_RD, &sc->stats.gptc,
+	    "Good Packets Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_txd",
-			CTLFLAG_RD, &sc->stats.bptc,
-			"Broadcast Packets Transmitted");
+	    CTLFLAG_RD, &sc->stats.bptc,
+	    "Broadcast Packets Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_txd",
-			CTLFLAG_RD, &sc->stats.mptc,
-			"Multicast Packets Transmitted");
+	    CTLFLAG_RD, &sc->stats.mptc,
+	    "Multicast Packets Transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_64",
-			CTLFLAG_RD, &sc->stats.ptc64,
-			"64 byte frames transmitted ");
+	    CTLFLAG_RD, &sc->stats.ptc64,
+	    "64 byte frames transmitted ");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_65_127",
-			CTLFLAG_RD, &sc->stats.ptc127,
-			"65-127 byte frames transmitted");
+	    CTLFLAG_RD, &sc->stats.ptc127,
+	    "65-127 byte frames transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_128_255",
-			CTLFLAG_RD, &sc->stats.ptc255,
-			"128-255 byte frames transmitted");
+	    CTLFLAG_RD, &sc->stats.ptc255,
+	    "128-255 byte frames transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_256_511",
-			CTLFLAG_RD, &sc->stats.ptc511,
-			"256-511 byte frames transmitted");
+	    CTLFLAG_RD, &sc->stats.ptc511,
+	    "256-511 byte frames transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_512_1023",
-			CTLFLAG_RD, &sc->stats.ptc1023,
-			"512-1023 byte frames transmitted");
+	    CTLFLAG_RD, &sc->stats.ptc1023,
+	    "512-1023 byte frames transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_1024_1522",
-			CTLFLAG_RD, &sc->stats.ptc1522,
-			"1024-1522 byte frames transmitted");
+	    CTLFLAG_RD, &sc->stats.ptc1522,
+	    "1024-1522 byte frames transmitted");
 	SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tso_txd",
-			CTLFLAG_RD, &sc->stats.tsctc,
-			"TSO Contexts Transmitted");
+	    CTLFLAG_RD, &sc->stats.tsctc,
+	    "TSO Contexts Transmitted");
 
 	/* Interrupt Stats */
-
 	int_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "interrupts",
 	    CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Interrupt Statistics");
 	int_list = SYSCTL_CHILDREN(int_node);
 
 	SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "asserts",
-			CTLFLAG_RD, &sc->stats.iac,
-			"Interrupt Assertion Count");
+	    CTLFLAG_RD, &sc->stats.iac,
+	    "Interrupt Assertion Count");
 
 	SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "rx_desc_min_thresh",
-			CTLFLAG_RD, &sc->stats.rxdmtc,
-			"Rx Desc Min Thresh Count");
+	    CTLFLAG_RD, &sc->stats.rxdmtc,
+	    "Rx Desc Min Thresh Count");
 }
 
 static void
@@ -2913,21 +2955,22 @@ igc_sbuf_fw_version(struct igc_fw_version *fw_ver, struct sbuf *buf)
 
 	if (fw_ver->eep_major || fw_ver->eep_minor || fw_ver->eep_build) {
 		sbuf_printf(buf, "EEPROM V%d.%d-%d", fw_ver->eep_major,
-			    fw_ver->eep_minor, fw_ver->eep_build);
+		    fw_ver->eep_minor, fw_ver->eep_build);
 		space = " ";
 	}
 
-	if (fw_ver->invm_major || fw_ver->invm_minor || fw_ver->invm_img_type) {
+	if (fw_ver->invm_major || fw_ver->invm_minor ||
+	    fw_ver->invm_img_type) {
 		sbuf_printf(buf, "%sNVM V%d.%d imgtype%d",
-			    space, fw_ver->invm_major, fw_ver->invm_minor,
-			    fw_ver->invm_img_type);
+		    space, fw_ver->invm_major, fw_ver->invm_minor,
+		    fw_ver->invm_img_type);
 		space = " ";
 	}
 
 	if (fw_ver->or_valid) {
 		sbuf_printf(buf, "%sOption ROM V%d-b%d-p%d",
-			    space, fw_ver->or_major, fw_ver->or_build,
-			    fw_ver->or_patch);
+		    space, fw_ver->or_major, fw_ver->or_build,
+		    fw_ver->or_patch);
 		space = " ";
 	}
 
@@ -3085,7 +3128,7 @@ igc_set_flowcntl(SYSCTL_HANDLER_ARGS)
 {
 	int error;
 	static int input = 3; /* default is full */
-	struct igc_softc	*sc = (struct igc_softc *) arg1;
+	struct igc_softc *sc = (struct igc_softc *) arg1;
 
 	error = sysctl_handle_int(oidp, &input, 0, req);
 
@@ -3253,14 +3296,14 @@ igc_print_debug_info(struct igc_softc *sc)
 	for (int i = 0; i < sc->tx_num_queues; i++, txr++) {
 		device_printf(dev, "TX Queue %d ------\n", i);
 		device_printf(dev, "hw tdh = %d, hw tdt = %d\n",
-			IGC_READ_REG(&sc->hw, IGC_TDH(i)),
-			IGC_READ_REG(&sc->hw, IGC_TDT(i)));
+		    IGC_READ_REG(&sc->hw, IGC_TDH(i)),
+		    IGC_READ_REG(&sc->hw, IGC_TDT(i)));
 
 	}
 	for (int j=0; j < sc->rx_num_queues; j++, rxr++) {
 		device_printf(dev, "RX Queue %d ------\n", j);
 		device_printf(dev, "hw rdh = %d, hw rdt = %d\n",
-			IGC_READ_REG(&sc->hw, IGC_RDH(j)),
-			IGC_READ_REG(&sc->hw, IGC_RDT(j)));
+		    IGC_READ_REG(&sc->hw, IGC_RDH(j)),
+		    IGC_READ_REG(&sc->hw, IGC_RDT(j)));
 	}
 }

sys/dev/igc/igc_txrx.c

@@ -97,15 +97,19 @@ igc_dump_rs(struct igc_softc *sc)
 			cur = txr->tx_rsq[rs_cidx];
 			status = txr->tx_base[cur].upper.fields.status;
 			if (!(status & IGC_TXD_STAT_DD))
-				printf("qid[%d]->tx_rsq[%d]: %d clear ", qid, rs_cidx, cur);
+				printf("qid[%d]->tx_rsq[%d]: %d clear ",
+				    qid, rs_cidx, cur);
 		} else {
 			rs_cidx = (rs_cidx-1)&(ntxd-1);
 			cur = txr->tx_rsq[rs_cidx];
-			printf("qid[%d]->tx_rsq[rs_cidx-1=%d]: %d  ", qid, rs_cidx, cur);
+			printf("qid[%d]->tx_rsq[rs_cidx-1=%d]: %d  ",
+			    qid, rs_cidx, cur);
 		}
-		printf("cidx_prev=%d rs_pidx=%d ",txr->tx_cidx_processed, txr->tx_rs_pidx);
+		printf("cidx_prev=%d rs_pidx=%d ",txr->tx_cidx_processed,
+		    txr->tx_rs_pidx);
 		for (i = 0; i < ntxd; i++) {
-			if (txr->tx_base[i].upper.fields.status & IGC_TXD_STAT_DD)
+			if (txr->tx_base[i].upper.fields.status &
+			    IGC_TXD_STAT_DD)
 				printf("%d set ", i);
 		}
 		printf("\n");
@@ -138,14 +142,15 @@ igc_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
 		break;
 	default:
 		panic("%s: CSUM_TSO but no supported IP version (0x%04x)",
-		      __func__, ntohs(pi->ipi_etype));
+		    __func__, ntohs(pi->ipi_etype));
 		break;
 	}
 
 	TXD = (struct igc_adv_tx_context_desc *) &txr->tx_base[pi->ipi_pidx];
 
 	/* This is used in the transmit desc in encap */
-	paylen = pi->ipi_len - pi->ipi_ehdrlen - pi->ipi_ip_hlen - pi->ipi_tcp_hlen;
+	paylen = pi->ipi_len - pi->ipi_ehdrlen - pi->ipi_ip_hlen -
+	    pi->ipi_tcp_hlen;
 
 	/* VLAN MACLEN IPLEN */
 	if (pi->ipi_mflags & M_VLANTAG) {
@@ -180,8 +185,8 @@ igc_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
  *
  **********************************************************************/
 static int
-igc_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, uint32_t *cmd_type_len,
-    uint32_t *olinfo_status)
+igc_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi,
+    uint32_t *cmd_type_len, uint32_t *olinfo_status)
 {
 	struct igc_adv_tx_context_desc *TXD;
 	uint32_t vlan_macip_lens, type_tucmd_mlhl;
@@ -275,7 +280,7 @@ igc_isc_txd_encap(void *arg, if_pkt_info_t pi)
 	pidx_last = olinfo_status = 0;
 	/* Basic descriptor defines */
 	cmd_type_len = (IGC_ADVTXD_DTYP_DATA |
-			IGC_ADVTXD_DCMD_IFCS | IGC_ADVTXD_DCMD_DEXT);
+	    IGC_ADVTXD_DCMD_IFCS | IGC_ADVTXD_DCMD_DEXT);
 
 	if (pi->ipi_mflags & M_VLANTAG)
 		cmd_type_len |= IGC_ADVTXD_DCMD_VLE;
@@ -324,9 +329,9 @@ igc_isc_txd_encap(void *arg, if_pkt_info_t pi)
 static void
 igc_isc_txd_flush(void *arg, uint16_t txqid, qidx_t pidx)
 {
-	struct igc_softc *sc	= arg;
-	struct igc_tx_queue *que	= &sc->tx_queues[txqid];
-	struct tx_ring *txr	= &que->txr;
+	struct igc_softc *sc = arg;
+	struct igc_tx_queue *que = &sc->tx_queues[txqid];
+	struct tx_ring *txr = &que->txr;
 
 	IGC_WRITE_REG(&sc->hw, IGC_TDT(txr->me), pidx);
 }
@@ -370,12 +375,13 @@ igc_isc_txd_credits_update(void *arg, uint16_t txqid, bool clear)
 		MPASS(delta > 0);
 
 		processed += delta;
-		prev  = cur;
+		prev = cur;
 		rs_cidx = (rs_cidx + 1) & (ntxd-1);
-		if (rs_cidx  == txr->tx_rs_pidx)
+		if (rs_cidx == txr->tx_rs_pidx)
 			break;
 		cur = txr->tx_rsq[rs_cidx];
-		status = ((union igc_adv_tx_desc *)&txr->tx_base[cur])->wb.status;
+		status =
+		    ((union igc_adv_tx_desc *)&txr->tx_base[cur])->wb.status;
 	} while ((status & IGC_TXD_STAT_DD));
 
 	txr->tx_rs_cidx = rs_cidx;
@@ -411,7 +417,8 @@ igc_isc_rxd_refill(void *arg, if_rxd_update_t iru)
 }
 
 static void
-igc_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, qidx_t pidx)
+igc_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused,
+    qidx_t pidx)
 {
 	struct igc_softc *sc = arg;
 	struct igc_rx_queue *que = &sc->rx_queues[rxqid];
@@ -477,7 +484,8 @@ igc_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 		MPASS ((staterr & IGC_RXD_STAT_DD) != 0);
 
 		len = le16toh(rxd->wb.upper.length);
-		ptype = le32toh(rxd->wb.lower.lo_dword.data) &  IGC_PKTTYPE_MASK;
+		ptype =
+		    le32toh(rxd->wb.lower.lo_dword.data) & IGC_PKTTYPE_MASK;
 
 		ri->iri_len += len;
 		rxr->rx_bytes += ri->iri_len;
@@ -558,7 +566,8 @@ igc_rx_checksum(uint32_t staterr, if_rxd_info_t ri, uint32_t ptype)
 		    (ptype & IGC_RXDADV_PKTTYPE_SCTP) != 0)) {
 			ri->iri_csum_flags |= CSUM_SCTP_VALID;
 		} else {
-			ri->iri_csum_flags |= CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
+			ri->iri_csum_flags |=
+			    CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
 			ri->iri_csum_data = htons(0xffff);
 		}
 	}

sys/dev/ixgbe/if_bypass.c

@@ -1,4 +1,4 @@
-/******************************************************************************
+/*****************************************************************************
 
   Copyright (c) 2001-2017, Intel Corporation
   All rights reserved.
@@ -29,7 +29,7 @@
   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   POSSIBILITY OF SUCH DAMAGE.
 
-******************************************************************************/
+*****************************************************************************/
 
 
 #include "ixgbe.h"
@@ -114,11 +114,11 @@ ixgbe_get_bypass_time(u32 *year, u32 *sec)
 static int
 ixgbe_bp_version(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      version = 0;
-	u32             cmd;
+	int error = 0;
+	static int version = 0;
+	u32 cmd;
 
 	ixgbe_bypass_mutex_enter(sc);
 	cmd = BYPASS_PAGE_CTL2 | BYPASS_WE;
@@ -154,15 +154,14 @@ err:
 static int
 ixgbe_bp_set_state(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      state = 0;
+	int error = 0;
+	static int state = 0;
 
 	/* Get the current state */
 	ixgbe_bypass_mutex_enter(sc);
-	error = hw->mac.ops.bypass_rw(hw,
-	    BYPASS_PAGE_CTL0, &state);
+	error = hw->mac.ops.bypass_rw(hw, BYPASS_PAGE_CTL0, &state);
 	ixgbe_bypass_mutex_clear(sc);
 	if (error != 0)
 		return (error);
@@ -216,10 +215,10 @@ out:
 static int
 ixgbe_bp_timeout(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      timeout = 0;
+	int error = 0;
+	static int timeout = 0;
 
 	/* Get the current value */
 	ixgbe_bypass_mutex_enter(sc);
@@ -259,10 +258,10 @@ ixgbe_bp_timeout(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_main_on(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      main_on = 0;
+	int error = 0;
+	static int main_on = 0;
 
 	ixgbe_bypass_mutex_enter(sc);
 	error = hw->mac.ops.bypass_rw(hw, BYPASS_PAGE_CTL0, &main_on);
@@ -301,10 +300,10 @@ ixgbe_bp_main_on(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_main_off(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      main_off = 0;
+	int error = 0;
+	static int main_off = 0;
 
 	ixgbe_bypass_mutex_enter(sc);
 	error = hw->mac.ops.bypass_rw(hw, BYPASS_PAGE_CTL0, &main_off);
@@ -343,10 +342,10 @@ ixgbe_bp_main_off(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_aux_on(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      aux_on = 0;
+	int error = 0;
+	static int aux_on = 0;
 
 	ixgbe_bypass_mutex_enter(sc);
 	error = hw->mac.ops.bypass_rw(hw, BYPASS_PAGE_CTL0, &aux_on);
@@ -385,10 +384,10 @@ ixgbe_bp_aux_on(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_aux_off(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
-	static int      aux_off = 0;
+	int error = 0;
+	static int aux_off = 0;
 
 	ixgbe_bypass_mutex_enter(sc);
 	error = hw->mac.ops.bypass_rw(hw, BYPASS_PAGE_CTL0, &aux_off);
@@ -432,11 +431,11 @@ ixgbe_bp_aux_off(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_wd_set(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error, tmp;
-	static int      timeout = 0;
-	u32             mask, arg;
+	int error, tmp;
+	static int timeout = 0;
+	u32 mask, arg;
 
 	/* Get the current hardware value */
 	ixgbe_bypass_mutex_enter(sc);
@@ -503,11 +502,11 @@ ixgbe_bp_wd_set(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_wd_reset(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc  *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             sec, year;
-	int             cmd, count = 0, error = 0;
-	int             reset_wd = 0;
+	u32 sec, year;
+	int cmd, count = 0, error = 0;
+	int reset_wd = 0;
 
 	error = sysctl_handle_int(oidp, &reset_wd, 0, req);
 	if ((error) || (req->newptr == NULL))
@@ -549,14 +548,14 @@ ixgbe_bp_wd_reset(SYSCTL_HANDLER_ARGS)
 static int
 ixgbe_bp_log(SYSCTL_HANDLER_ARGS)
 {
-	struct ixgbe_softc             *sc = (struct ixgbe_softc *) arg1;
-	struct ixgbe_hw            *hw = &sc->hw;
-	u32                        cmd, base, head;
-	u32                        log_off, count = 0;
-	static int                 status = 0;
-	u8                         data;
+	struct ixgbe_softc *sc = (struct ixgbe_softc *) arg1;
+	struct ixgbe_hw *hw = &sc->hw;
+	u32 cmd, base, head;
+	u32 log_off, count = 0;
+	static int status = 0;
+	u8 data;
 	struct ixgbe_bypass_eeprom eeprom[BYPASS_MAX_LOGS];
-	int                        i, error = 0;
+	int i, error = 0;
 
 	error = sysctl_handle_int(oidp, &status, 0, req);
 	if ((error) || (req->newptr == NULL))
@@ -639,12 +638,15 @@ ixgbe_bp_log(SYSCTL_HANDLER_ARGS)
 		    BYPASS_LOG_EVENT_SHIFT;
 		u8 action =  eeprom[count].actions & BYPASS_LOG_ACTION_M;
 		u16 day_mon[2][13] = {
-		  {0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334, 365},
-		  {0, 31, 59, 91, 121, 152, 182, 213, 244, 274, 305, 335, 366}
+			{0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304,
+			    334, 365},
+			{0, 31, 59, 91, 121, 152, 182, 213, 244, 274, 305,
+			    335, 366}
 		};
 		char *event_str[] = {"unknown", "main on", "aux on",
 		    "main off", "aux off", "WDT", "user" };
-		char *action_str[] = {"ignore", "normal", "bypass", "isolate",};
+		char *action_str[] =
+		    {"ignore", "normal", "bypass", "isolate",};
 
 		/* verify vaild data  1 - 6 */
 		if (event < BYPASS_EVENT_MAIN_ON || event > BYPASS_EVENT_USR)
@@ -711,11 +713,11 @@ unlock_err:
 void
 ixgbe_bypass_init(struct ixgbe_softc *sc)
 {
-	struct ixgbe_hw        *hw = &sc->hw;
-	device_t               dev = sc->dev;
-	struct sysctl_oid      *bp_node;
+	struct ixgbe_hw *hw = &sc->hw;
+	device_t dev = sc->dev;
+	struct sysctl_oid *bp_node;
 	struct sysctl_oid_list *bp_list;
-	u32                    mask, value, sec, year;
+	u32 mask, value, sec, year;
 
 	if (!(sc->feat_cap & IXGBE_FEATURE_BYPASS))
 		return;
@@ -723,13 +725,13 @@ ixgbe_bypass_init(struct ixgbe_softc *sc)
 	/* First set up time for the hardware */
 	ixgbe_get_bypass_time(&year, &sec);
 
-	mask = BYPASS_CTL1_TIME_M
-	     | BYPASS_CTL1_VALID_M
-	     | BYPASS_CTL1_OFFTRST_M;
+	mask = BYPASS_CTL1_TIME_M |
+	    BYPASS_CTL1_VALID_M |
+	    BYPASS_CTL1_OFFTRST_M;
 
-	value = (sec & BYPASS_CTL1_TIME_M)
-	      | BYPASS_CTL1_VALID
-	      | BYPASS_CTL1_OFFTRST;
+	value = (sec & BYPASS_CTL1_TIME_M) |
+	    BYPASS_CTL1_VALID |
+	    BYPASS_CTL1_OFFTRST;
 
 	ixgbe_bypass_mutex_enter(sc);
 	hw->mac.ops.bypass_set(hw, BYPASS_PAGE_CTL1, mask, value);

sys/dev/ixgbe/if_fdir.c

@@ -1,4 +1,4 @@
-/******************************************************************************
+/*****************************************************************************
 
   Copyright (c) 2001-2017, Intel Corporation
   All rights reserved.
@@ -29,7 +29,7 @@
   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   POSSIBILITY OF SUCH DAMAGE.
 
-******************************************************************************/
+*****************************************************************************/
 
 #include "ixgbe.h"
 
@@ -51,9 +51,9 @@ ixgbe_init_fdir(struct ixgbe_softc *sc)
 void
 ixgbe_reinit_fdir(void *context)
 {
-	if_ctx_t       ctx = context;
+	if_ctx_t ctx = context;
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	if_t           ifp = iflib_get_ifp(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
 
 	if (!(sc->feat_en & IXGBE_FEATURE_FDIR))
 		return;
@@ -79,16 +79,16 @@ ixgbe_reinit_fdir(void *context)
 void
 ixgbe_atr(struct tx_ring *txr, struct mbuf *mp)
 {
-	struct ixgbe_softc             *sc = txr->sc;
-	struct ix_queue            *que;
-	struct ip                  *ip;
-	struct tcphdr              *th;
-	struct udphdr              *uh;
-	struct ether_vlan_header   *eh;
+	struct ixgbe_softc *sc = txr->sc;
+	struct ix_queue *que;
+	struct ip *ip;
+	struct tcphdr *th;
+	struct udphdr *uh;
+	struct ether_vlan_header *eh;
 	union ixgbe_atr_hash_dword input = {.dword = 0};
 	union ixgbe_atr_hash_dword common = {.dword = 0};
-	int                        ehdrlen, ip_hlen;
-	u16                        etype;
+	int ehdrlen, ip_hlen;
+	u16 etype;
 
 	eh = mtod(mp, struct ether_vlan_header *);
 	if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) {

sys/dev/ixgbe/if_ixv.c

@@ -1,4 +1,4 @@
-/******************************************************************************
+/*****************************************************************************
 
   Copyright (c) 2001-2017, Intel Corporation
   All rights reserved.
@@ -29,7 +29,7 @@
   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   POSSIBILITY OF SUCH DAMAGE.
 
-******************************************************************************/
+*****************************************************************************/
 
 
 #include "opt_inet.h"
@@ -58,13 +58,18 @@ static const char ixv_driver_version[] = "2.0.1-k";
  ************************************************************************/
 static const pci_vendor_info_t ixv_vendor_info_array[] =
 {
-	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_82599_VF, "Intel(R) X520 82599 Virtual Function"),
-	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X540_VF, "Intel(R) X540 Virtual Function"),
-	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550_VF, "Intel(R) X550 Virtual Function"),
-	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550EM_X_VF, "Intel(R) X552 Virtual Function"),
-	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550EM_A_VF, "Intel(R) X553 Virtual Function"),
+	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_82599_VF,
+	    "Intel(R) X520 82599 Virtual Function"),
+	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X540_VF,
+	    "Intel(R) X540 Virtual Function"),
+	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550_VF,
+	    "Intel(R) X550 Virtual Function"),
+	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550EM_X_VF,
+	    "Intel(R) X552 Virtual Function"),
+	PVID(IXGBE_INTEL_VENDOR_ID, IXGBE_DEV_ID_X550EM_A_VF,
+	    "Intel(R) X553 Virtual Function"),
 	/* required last entry */
-PVID_END
+	PVID_END
 };
 
 /************************************************************************
@@ -76,8 +81,10 @@ static int      ixv_if_attach_post(if_ctx_t);
 static int      ixv_if_detach(if_ctx_t);
 
 static int      ixv_if_rx_queue_intr_enable(if_ctx_t, uint16_t);
-static int      ixv_if_tx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int, int);
-static int      ixv_if_rx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int, int);
+static int      ixv_if_tx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int,
+    int);
+static int      ixv_if_rx_queues_alloc(if_ctx_t, caddr_t *, uint64_t *, int,
+    int);
 static void     ixv_if_queues_free(if_ctx_t);
 static void     ixv_identify_hardware(if_ctx_t);
 static void     ixv_init_device_features(struct ixgbe_softc *);
@@ -239,17 +246,17 @@ ixv_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs,
     int ntxqs, int ntxqsets)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	if_softc_ctx_t     scctx = sc->shared;
+	if_softc_ctx_t scctx = sc->shared;
 	struct ix_tx_queue *que;
-	int                i, j, error;
+	int i, j, error;
 
 	MPASS(sc->num_tx_queues == ntxqsets);
 	MPASS(ntxqs == 1);
 
 	/* Allocate queue structure memory */
 	sc->tx_queues =
-	    (struct ix_tx_queue *)malloc(sizeof(struct ix_tx_queue) * ntxqsets,
-	                                 M_DEVBUF, M_NOWAIT | M_ZERO);
+	    (struct ix_tx_queue *)malloc(sizeof(struct ix_tx_queue) *
+	    ntxqsets, M_DEVBUF, M_NOWAIT | M_ZERO);
 	if (!sc->tx_queues) {
 		device_printf(iflib_get_dev(ctx),
 		    "Unable to allocate TX ring memory\n");
@@ -263,13 +270,14 @@ ixv_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs,
 		txr->sc =  que->sc = sc;
 
 		/* Allocate report status array */
-		if (!(txr->tx_rsq = (qidx_t *)malloc(sizeof(qidx_t) * scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) {
+		if (!(txr->tx_rsq = (qidx_t *)malloc(sizeof(qidx_t) *
+		    scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) {
 			error = ENOMEM;
 			goto fail;
 		}
 		for (j = 0; j < scctx->isc_ntxd[0]; j++)
 			txr->tx_rsq[j] = QIDX_INVALID;
-		/* get the virtual and physical address of the hardware queues */
+		/* get virtual and physical address of the hardware queues */
 		txr->tail = IXGBE_VFTDT(txr->me);
 		txr->tx_base = (union ixgbe_adv_tx_desc *)vaddrs[i*ntxqs];
 		txr->tx_paddr = paddrs[i*ntxqs];
@@ -299,15 +307,15 @@ ixv_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs,
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ix_rx_queue *que;
-	int                i, error;
+	int i, error;
 
 	MPASS(sc->num_rx_queues == nrxqsets);
 	MPASS(nrxqs == 1);
 
 	/* Allocate queue structure memory */
 	sc->rx_queues =
-	    (struct ix_rx_queue *)malloc(sizeof(struct ix_rx_queue) * nrxqsets,
-	                                 M_DEVBUF, M_NOWAIT | M_ZERO);
+	    (struct ix_rx_queue *)malloc(sizeof(struct ix_rx_queue) *
+	    nrxqsets, M_DEVBUF, M_NOWAIT | M_ZERO);
 	if (!sc->rx_queues) {
 		device_printf(iflib_get_dev(ctx),
 		    "Unable to allocate TX ring memory\n");
@@ -348,7 +356,7 @@ ixv_if_queues_free(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ix_tx_queue *que = sc->tx_queues;
-	int                i;
+	int i;
 
 	if (que == NULL)
 		goto free;
@@ -382,11 +390,11 @@ free:
 static int
 ixv_if_attach_pre(if_ctx_t ctx)
 {
-	struct ixgbe_softc  *sc;
-	device_t        dev;
-	if_softc_ctx_t  scctx;
+	struct ixgbe_softc *sc;
+	device_t dev;
+	if_softc_ctx_t scctx;
 	struct ixgbe_hw *hw;
-	int             error = 0;
+	int error = 0;
 
 	INIT_DEBUGOUT("ixv_attach: begin");
 
@@ -458,7 +466,7 @@ ixv_if_attach_pre(if_ctx_t ctx)
 	/* Check if VF was disabled by PF */
 	error = hw->mac.ops.get_link_state(hw, &sc->link_enabled);
 	if (error) {
-		/* PF is not capable of controlling VF state. Enable the link. */
+		/* PF is not capable of controlling VF state. Enable link. */
 		sc->link_enabled = true;
 	}
 
@@ -522,8 +530,8 @@ static int
 ixv_if_attach_post(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	device_t       dev = iflib_get_dev(ctx);
-	int            error = 0;
+	device_t dev = iflib_get_dev(ctx);
+	int error = 0;
 
 	/* Setup OS specific network interface */
 	error = ixv_setup_interface(ctx);
@@ -568,7 +576,7 @@ ixv_if_mtu_set(if_ctx_t ctx, uint32_t mtu)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	if_t ifp = iflib_get_ifp(ctx);
-	int            error = 0;
+	int error = 0;
 
 	IOCTL_DEBUGOUT("ioctl: SIOCSIFMTU (Set Interface MTU)");
 	if (mtu > IXGBE_MAX_FRAME_SIZE - IXGBE_MTU_HDR) {
@@ -596,9 +604,9 @@ ixv_if_init(if_ctx_t ctx)
 {
 	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
 	if_t ifp = iflib_get_ifp(ctx);
-	device_t        dev = iflib_get_dev(ctx);
+	device_t dev = iflib_get_dev(ctx);
 	struct ixgbe_hw *hw = &sc->hw;
-	int             error = 0;
+	int error = 0;
 
 	INIT_DEBUGOUT("ixv_if_init: begin");
 	hw->adapter_stopped = false;
@@ -670,8 +678,8 @@ static inline void
 ixv_enable_queue(struct ixgbe_softc *sc, u32 vector)
 {
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             queue = 1 << vector;
-	u32             mask;
+	u32 queue = 1 << vector;
+	u32 mask;
 
 	mask = (IXGBE_EIMS_RTX_QUEUE & queue);
 	IXGBE_WRITE_REG(hw, IXGBE_VTEIMS, mask);
@@ -684,8 +692,8 @@ static inline void
 ixv_disable_queue(struct ixgbe_softc *sc, u32 vector)
 {
 	struct ixgbe_hw *hw = &sc->hw;
-	u64             queue = (u64)(1 << vector);
-	u32             mask;
+	u64 queue = (u64)(1 << vector);
+	u32 mask;
 
 	mask = (IXGBE_EIMS_RTX_QUEUE & queue);
 	IXGBE_WRITE_REG(hw, IXGBE_VTEIMC, mask);
@@ -699,7 +707,7 @@ static int
 ixv_msix_que(void *arg)
 {
 	struct ix_rx_queue *que = arg;
-	struct ixgbe_softc     *sc = que->sc;
+	struct ixgbe_softc *sc = que->sc;
 
 	ixv_disable_queue(sc, que->msix);
 	++que->irqs;
@@ -713,9 +721,9 @@ ixv_msix_que(void *arg)
 static int
 ixv_msix_mbx(void *arg)
 {
-	struct ixgbe_softc  *sc = arg;
+	struct ixgbe_softc *sc = arg;
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             reg;
+	u32 reg;
 
 	++sc->link_irq;
 
@@ -811,11 +819,13 @@ static int
 ixv_negotiate_api(struct ixgbe_softc *sc)
 {
 	struct ixgbe_hw *hw = &sc->hw;
-	int             mbx_api[] = { ixgbe_mbox_api_12,
-	                              ixgbe_mbox_api_11,
-	                              ixgbe_mbox_api_10,
-	                              ixgbe_mbox_api_unknown };
-	int             i = 0;
+	int mbx_api[] = {
+		ixgbe_mbox_api_12,
+		ixgbe_mbox_api_11,
+		ixgbe_mbox_api_10,
+		ixgbe_mbox_api_unknown
+	};
+	int i = 0;
 
 	while (mbx_api[i] != ixgbe_mbox_api_unknown) {
 		if (ixgbevf_negotiate_api_version(hw, mbx_api[i]) == 0)
@@ -830,7 +840,8 @@ ixv_negotiate_api(struct ixgbe_softc *sc)
 static u_int
 ixv_if_multi_set_cb(void *cb_arg, struct sockaddr_dl *addr, u_int cnt)
 {
-	bcopy(LLADDR(addr), &((u8 *)cb_arg)[cnt * IXGBE_ETH_LENGTH_OF_ADDRESS],
+	bcopy(LLADDR(addr),
+	    &((u8 *)cb_arg)[cnt * IXGBE_ETH_LENGTH_OF_ADDRESS],
 	    IXGBE_ETH_LENGTH_OF_ADDRESS);
 
 	return (++cnt);
@@ -844,11 +855,11 @@ ixv_if_multi_set_cb(void *cb_arg, struct sockaddr_dl *addr, u_int cnt)
 static void
 ixv_if_multi_set(if_ctx_t ctx)
 {
-	u8       mta[MAX_NUM_MULTICAST_ADDRESSES * IXGBE_ETH_LENGTH_OF_ADDRESS];
-	struct ixgbe_softc     *sc = iflib_get_softc(ctx);
-	u8                 *update_ptr;
-	if_t               ifp = iflib_get_ifp(ctx);
-	int                mcnt = 0;
+	u8 mta[MAX_NUM_MULTICAST_ADDRESSES * IXGBE_ETH_LENGTH_OF_ADDRESS];
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
+	u8 *update_ptr;
+	if_t ifp = iflib_get_ifp(ctx);
+	int mcnt = 0;
 
 	IOCTL_DEBUGOUT("ixv_if_multi_set: begin");
 
@@ -908,8 +919,8 @@ static void
 ixv_if_update_admin_status(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	device_t       dev = iflib_get_dev(ctx);
-	s32            status;
+	device_t dev = iflib_get_dev(ctx);
+	s32 status;
 
 	sc->hw.mac.get_link_status = true;
 
@@ -955,7 +966,7 @@ ixv_if_update_admin_status(if_ctx_t ctx)
 static void
 ixv_if_stop(if_ctx_t ctx)
 {
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ixgbe_hw *hw = &sc->hw;
 
 	INIT_DEBUGOUT("ixv_stop: begin\n");
@@ -981,8 +992,8 @@ ixv_if_stop(if_ctx_t ctx)
 static void
 ixv_identify_hardware(if_ctx_t ctx)
 {
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
-	device_t        dev = iflib_get_dev(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
+	device_t dev = iflib_get_dev(ctx);
 	struct ixgbe_hw *hw = &sc->hw;
 
 	/* Save off the information about this board */
@@ -1023,22 +1034,24 @@ static int
 ixv_if_msix_intr_assign(if_ctx_t ctx, int msix)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	device_t           dev = iflib_get_dev(ctx);
+	device_t dev = iflib_get_dev(ctx);
 	struct ix_rx_queue *rx_que = sc->rx_queues;
 	struct ix_tx_queue *tx_que;
-	int                error, rid, vector = 0;
-	char               buf[16];
+	int error, rid, vector = 0;
+	char buf[16];
 
 	for (int i = 0; i < sc->num_rx_queues; i++, vector++, rx_que++) {
 		rid = vector + 1;
 
 		snprintf(buf, sizeof(buf), "rxq%d", i);
 		error = iflib_irq_alloc_generic(ctx, &rx_que->que_irq, rid,
-		    IFLIB_INTR_RXTX, ixv_msix_que, rx_que, rx_que->rxr.me, buf);
+		    IFLIB_INTR_RXTX, ixv_msix_que, rx_que, rx_que->rxr.me,
+		    buf);
 
 		if (error) {
 			device_printf(iflib_get_dev(ctx),
-			    "Failed to allocate que int %d err: %d", i, error);
+			    "Failed to allocate que int %d err: %d",
+			    i, error);
 			sc->num_rx_queues = i + 1;
 			goto fail;
 		}
@@ -1074,7 +1087,8 @@ ixv_if_msix_intr_assign(if_ctx_t ctx, int msix)
 	if (sc->hw.mac.type == ixgbe_mac_82599_vf) {
 		int msix_ctrl;
 		if (pci_find_cap(dev, PCIY_MSIX, &rid)) {
-			device_printf(dev, "Finding MSIX capability failed\n");
+			device_printf(dev,
+			    "Finding MSIX capability failed\n");
 		} else {
 			rid += PCIR_MSIX_CTRL;
 			msix_ctrl = pci_read_config(dev, rid, 2);
@@ -1101,21 +1115,21 @@ static int
 ixv_allocate_pci_resources(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	device_t       dev = iflib_get_dev(ctx);
-	int            rid;
+	device_t dev = iflib_get_dev(ctx);
+	int rid;
 
 	rid = PCIR_BAR(0);
 	sc->pci_mem = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid,
 	    RF_ACTIVE);
 
 	if (!(sc->pci_mem)) {
-		device_printf(dev, "Unable to allocate bus resource: memory\n");
+		device_printf(dev,
+		    "Unable to allocate bus resource: memory\n");
 		return (ENXIO);
 	}
 
 	sc->osdep.mem_bus_space_tag = rman_get_bustag(sc->pci_mem);
-	sc->osdep.mem_bus_space_handle =
-	    rman_get_bushandle(sc->pci_mem);
+	sc->osdep.mem_bus_space_handle = rman_get_bushandle(sc->pci_mem);
 	sc->hw.hw_addr = (u8 *)&sc->osdep.mem_bus_space_handle;
 
 	return (0);
@@ -1129,7 +1143,7 @@ ixv_free_pci_resources(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ix_rx_queue *que = sc->rx_queues;
-	device_t           dev = iflib_get_dev(ctx);
+	device_t dev = iflib_get_dev(ctx);
 
 	/* Release all MSI-X queue resources */
 	if (sc->intr_type == IFLIB_INTR_MSIX)
@@ -1156,7 +1170,7 @@ ixv_setup_interface(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	if_softc_ctx_t scctx = sc->shared;
-	if_t           ifp = iflib_get_ifp(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
 
 	INIT_DEBUGOUT("ixv_setup_interface: begin");
 
@@ -1178,7 +1192,7 @@ static uint64_t
 ixv_if_get_counter(if_ctx_t ctx, ift_counter cnt)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	if_t           ifp = iflib_get_ifp(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
 
 	switch (cnt) {
 	case IFCOUNTER_IPACKETS:
@@ -1222,16 +1236,16 @@ static void
 ixv_initialize_transmit_units(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	struct ixgbe_hw    *hw = &sc->hw;
-	if_softc_ctx_t     scctx = sc->shared;
+	struct ixgbe_hw *hw = &sc->hw;
+	if_softc_ctx_t scctx = sc->shared;
 	struct ix_tx_queue *que = sc->tx_queues;
-	int                i;
+	int i;
 
 	for (i = 0; i < sc->num_tx_queues; i++, que++) {
 		struct tx_ring *txr = &que->txr;
-		u64            tdba = txr->tx_paddr;
-		u32            txctrl, txdctl;
-		int            j = txr->me;
+		u64 tdba = txr->tx_paddr;
+		u32 txctrl, txdctl;
+		int j = txr->me;
 
 		/* Set WTHRESH to 8, burst writeback */
 		txdctl = IXGBE_READ_REG(hw, IXGBE_VFTXDCTL(j));
@@ -1281,10 +1295,10 @@ static void
 ixv_initialize_rss_mapping(struct ixgbe_softc *sc)
 {
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             reta = 0, mrqc, rss_key[10];
-	int             queue_id;
-	int             i, j;
-	u32             rss_hash_config;
+	u32 reta = 0, mrqc, rss_key[10];
+	int queue_id;
+	int i, j;
+	u32 rss_hash_config;
 
 	if (sc->feat_en & IXGBE_FEATURE_RSS) {
 		/* Fetch the configured RSS key */
@@ -1351,18 +1365,21 @@ ixv_initialize_rss_mapping(struct ixgbe_softc *sc)
 	if (rss_hash_config & RSS_HASHTYPE_RSS_TCP_IPV6)
 		mrqc |= IXGBE_MRQC_RSS_FIELD_IPV6_TCP;
 	if (rss_hash_config & RSS_HASHTYPE_RSS_IPV6_EX)
-		device_printf(sc->dev, "%s: RSS_HASHTYPE_RSS_IPV6_EX defined, but not supported\n",
-		    __func__);
+		device_printf(sc->dev,
+		    "%s: RSS_HASHTYPE_RSS_IPV6_EX defined,"
+		    " but not supported\n", __func__);
 	if (rss_hash_config & RSS_HASHTYPE_RSS_TCP_IPV6_EX)
-		device_printf(sc->dev, "%s: RSS_HASHTYPE_RSS_TCP_IPV6_EX defined, but not supported\n",
-		    __func__);
+		device_printf(sc->dev,
+		    "%s: RSS_HASHTYPE_RSS_TCP_IPV6_EX defined,"
+		    " but not supported\n", __func__);
 	if (rss_hash_config & RSS_HASHTYPE_RSS_UDP_IPV4)
 		mrqc |= IXGBE_MRQC_RSS_FIELD_IPV4_UDP;
 	if (rss_hash_config & RSS_HASHTYPE_RSS_UDP_IPV6)
 		mrqc |= IXGBE_MRQC_RSS_FIELD_IPV6_UDP;
 	if (rss_hash_config & RSS_HASHTYPE_RSS_UDP_IPV6_EX)
-		device_printf(sc->dev, "%s: RSS_HASHTYPE_RSS_UDP_IPV6_EX defined, but not supported\n",
-		    __func__);
+		device_printf(sc->dev,
+		    "%s: RSS_HASHTYPE_RSS_UDP_IPV6_EX defined,"
+		    " but not supported\n", __func__);
 	IXGBE_WRITE_REG(hw, IXGBE_VFMRQC, mrqc);
 } /* ixv_initialize_rss_mapping */
 
@@ -1374,22 +1391,22 @@ static void
 ixv_initialize_receive_units(if_ctx_t ctx)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	if_softc_ctx_t     scctx;
-	struct ixgbe_hw    *hw = &sc->hw;
+	if_softc_ctx_t scctx;
+	struct ixgbe_hw *hw = &sc->hw;
 #ifdef DEV_NETMAP
-	if_t               ifp = iflib_get_ifp(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
 #endif
 	struct ix_rx_queue *que = sc->rx_queues;
-	u32                bufsz, psrtype;
+	u32 bufsz, psrtype;
 
 	bufsz = (sc->rx_mbuf_sz + BSIZEPKT_ROUNDUP) >>
 	    IXGBE_SRRCTL_BSIZEPKT_SHIFT;
 
-	psrtype = IXGBE_PSRTYPE_TCPHDR
-	        | IXGBE_PSRTYPE_UDPHDR
-	        | IXGBE_PSRTYPE_IPV4HDR
-	        | IXGBE_PSRTYPE_IPV6HDR
-	        | IXGBE_PSRTYPE_L2HDR;
+	psrtype = IXGBE_PSRTYPE_TCPHDR |
+	    IXGBE_PSRTYPE_UDPHDR |
+	    IXGBE_PSRTYPE_IPV4HDR |
+	    IXGBE_PSRTYPE_IPV6HDR |
+	    IXGBE_PSRTYPE_L2HDR;
 
 	if (sc->num_rx_queues > 1)
 		psrtype |= 1 << 29;
@@ -1398,15 +1415,18 @@ ixv_initialize_receive_units(if_ctx_t ctx)
 
 	/* Tell PF our max_frame size */
 	if (ixgbevf_rlpml_set_vf(hw, sc->max_frame_size) != 0) {
-		device_printf(sc->dev, "There is a problem with the PF setup.  It is likely the receive unit for this VF will not function correctly.\n");
+		device_printf(sc->dev,
+		    "There is a problem with the PF setup.  It is likely the"
+		    " receive unit for this VF will not function correctly."
+		    "\n");
 	}
 	scctx = sc->shared;
 
 	for (int i = 0; i < sc->num_rx_queues; i++, que++) {
 		struct rx_ring *rxr = &que->rxr;
-		u64            rdba = rxr->rx_paddr;
-		u32            reg, rxdctl;
-		int            j = rxr->me;
+		u64 rdba = rxr->rx_paddr;
+		u32 reg, rxdctl;
+		int j = rxr->me;
 
 		/* Disable the queue */
 		rxdctl = IXGBE_READ_REG(hw, IXGBE_VFRXDCTL(j));
@@ -1497,10 +1517,10 @@ ixv_initialize_receive_units(if_ctx_t ctx)
 static void
 ixv_setup_vlan_support(if_ctx_t ctx)
 {
-	if_t            ifp = iflib_get_ifp(ctx);
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             ctrl, vid, vfta, retry;
+	u32 ctrl, vid, vfta, retry;
 
 	/*
 	 * We get here thru if_init, meaning
@@ -1571,7 +1591,7 @@ static void
 ixv_if_register_vlan(if_ctx_t ctx, u16 vtag)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	u16            index, bit;
+	u16 index, bit;
 
 	index = (vtag >> 5) & 0x7F;
 	bit = vtag & 0x1F;
@@ -1589,7 +1609,7 @@ static void
 ixv_if_unregister_vlan(if_ctx_t ctx, u16 vtag)
 {
 	struct ixgbe_softc *sc = iflib_get_softc(ctx);
-	u16            index, bit;
+	u16 index, bit;
 
 	index = (vtag >> 5) & 0x7F;
 	bit = vtag & 0x1F;
@@ -1603,10 +1623,10 @@ ixv_if_unregister_vlan(if_ctx_t ctx, u16 vtag)
 static void
 ixv_if_enable_intr(if_ctx_t ctx)
 {
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ixgbe_hw *hw = &sc->hw;
 	struct ix_rx_queue *que = sc->rx_queues;
-	u32             mask = (IXGBE_EIMS_ENABLE_MASK & ~IXGBE_EIMS_RTX_QUEUE);
+	u32 mask = (IXGBE_EIMS_ENABLE_MASK & ~IXGBE_EIMS_RTX_QUEUE);
 
 	IXGBE_WRITE_REG(hw, IXGBE_VTEIMS, mask);
 
@@ -1638,7 +1658,7 @@ ixv_if_disable_intr(if_ctx_t ctx)
 static int
 ixv_if_rx_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid)
 {
-	struct ixgbe_softc	*sc = iflib_get_softc(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ix_rx_queue *que = &sc->rx_queues[rxqid];
 
 	ixv_enable_queue(sc, que->rxr.me);
@@ -1658,7 +1678,7 @@ static void
 ixv_set_ivar(struct ixgbe_softc *sc, u8 entry, u8 vector, s8 type)
 {
 	struct ixgbe_hw *hw = &sc->hw;
-	u32             ivar, index;
+	u32 ivar, index;
 
 	vector |= IXGBE_IVAR_ALLOC_VAL;
 
@@ -1808,18 +1828,18 @@ ixv_update_stats(struct ixgbe_softc *sc)
 static void
 ixv_add_stats_sysctls(struct ixgbe_softc *sc)
 {
-	device_t                dev = sc->dev;
-	struct ix_tx_queue      *tx_que = sc->tx_queues;
-	struct ix_rx_queue      *rx_que = sc->rx_queues;
-	struct sysctl_ctx_list  *ctx = device_get_sysctl_ctx(dev);
-	struct sysctl_oid       *tree = device_get_sysctl_tree(dev);
-	struct sysctl_oid_list  *child = SYSCTL_CHILDREN(tree);
+	device_t dev = sc->dev;
+	struct ix_tx_queue *tx_que = sc->tx_queues;
+	struct ix_rx_queue *rx_que = sc->rx_queues;
+	struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev);
+	struct sysctl_oid *tree = device_get_sysctl_tree(dev);
+	struct sysctl_oid_list *child = SYSCTL_CHILDREN(tree);
 	struct ixgbevf_hw_stats *stats = &sc->stats.vf;
-	struct sysctl_oid       *stat_node, *queue_node;
-	struct sysctl_oid_list  *stat_list, *queue_list;
+	struct sysctl_oid *stat_node, *queue_node;
+	struct sysctl_oid_list *stat_list, *queue_list;
 
 #define QUEUE_NAME_LEN 32
-	char                    namebuf[QUEUE_NAME_LEN];
+	char namebuf[QUEUE_NAME_LEN];
 
 	/* Driver Statistics */
 	SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "watchdog_events",
@@ -1922,9 +1942,9 @@ ixv_sysctl_debug(SYSCTL_HANDLER_ARGS)
 static void
 ixv_init_device_features(struct ixgbe_softc *sc)
 {
-	sc->feat_cap = IXGBE_FEATURE_NETMAP
-				    | IXGBE_FEATURE_VF
-				    | IXGBE_FEATURE_LEGACY_TX;
+	sc->feat_cap = IXGBE_FEATURE_NETMAP |
+	    IXGBE_FEATURE_VF |
+	    IXGBE_FEATURE_LEGACY_TX;
 
 	/* A tad short on feature flags for VFs, atm. */
 	switch (sc->hw.mac.type) {

sys/dev/ixgbe/if_sriov.c

@@ -217,7 +217,7 @@ ixgbe_ping_all_vfs(struct ixgbe_softc *sc)
 
 static void
 ixgbe_vf_set_default_vlan(struct ixgbe_softc *sc, struct ixgbe_vf *vf,
-                          uint16_t tag)
+    uint16_t tag)
 {
 	struct ixgbe_hw *hw;
 	uint32_t vmolr, vmvir;
@@ -269,7 +269,6 @@ ixgbe_clear_vfmbmem(struct ixgbe_softc *sc, struct ixgbe_vf *vf)
 static boolean_t
 ixgbe_vf_frame_size_compatible(struct ixgbe_softc *sc, struct ixgbe_vf *vf)
 {
-
 	/*
 	 * Frame size compatibility between PF and VF is only a problem on
 	 * 82599-based cards.  X540 and later support any combination of jumbo
@@ -282,8 +281,8 @@ ixgbe_vf_frame_size_compatible(struct ixgbe_softc *sc, struct ixgbe_vf *vf)
 	case IXGBE_API_VER_1_0:
 	case IXGBE_API_VER_UNKNOWN:
 		/*
-		 * On legacy (1.0 and older) VF versions, we don't support jumbo
-		 * frames on either the PF or the VF.
+		 * On legacy (1.0 and older) VF versions, we don't support
+		 * jumbo frames on either the PF or the VF.
 		 */
 		if (sc->max_frame_size > ETHER_MAX_LEN ||
 		    vf->maximum_frame_size > ETHER_MAX_LEN)
@@ -302,8 +301,8 @@ ixgbe_vf_frame_size_compatible(struct ixgbe_softc *sc, struct ixgbe_vf *vf)
 			return (true);
 
 		/*
-		 * Jumbo frames only work with VFs if the PF is also using jumbo
-		 * frames.
+		 * Jumbo frames only work with VFs if the PF is also using
+		 * jumbo frames.
 		 */
 		if (sc->max_frame_size <= ETHER_MAX_LEN)
 			return (true);
@@ -526,7 +525,7 @@ ixgbe_vf_set_lpe(struct ixgbe_softc *sc, struct ixgbe_vf *vf, uint32_t *msg)
 
 static void
 ixgbe_vf_set_macvlan(struct ixgbe_softc *sc, struct ixgbe_vf *vf,
-                     uint32_t *msg)
+    uint32_t *msg)
 {
 	//XXX implement this
 	ixgbe_send_vf_failure(sc, vf, msg[0]);
@@ -537,7 +536,6 @@ static void
 ixgbe_vf_api_negotiate(struct ixgbe_softc *sc, struct ixgbe_vf *vf,
     uint32_t *msg)
 {
-
 	switch (msg[1]) {
 	case IXGBE_API_VER_1_0:
 	case IXGBE_API_VER_1_1:
@@ -553,7 +551,8 @@ ixgbe_vf_api_negotiate(struct ixgbe_softc *sc, struct ixgbe_vf *vf,
 
 
 static void
-ixgbe_vf_get_queues(struct ixgbe_softc *sc, struct ixgbe_vf *vf, uint32_t *msg)
+ixgbe_vf_get_queues(struct ixgbe_softc *sc, struct ixgbe_vf *vf,
+    uint32_t *msg)
 {
 	struct ixgbe_hw *hw;
 	uint32_t resp[IXGBE_VF_GET_QUEUES_RESP_LEN];
@@ -585,9 +584,9 @@ ixgbe_vf_get_queues(struct ixgbe_softc *sc, struct ixgbe_vf *vf, uint32_t *msg)
 static void
 ixgbe_process_vf_msg(if_ctx_t ctx, struct ixgbe_vf *vf)
 {
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 #ifdef KTR
-	if_t		ifp = iflib_get_ifp(ctx);
+	if_t ifp = iflib_get_ifp(ctx);
 #endif
 	struct ixgbe_hw *hw;
 	uint32_t msg[IXGBE_VFMAILBOX_SIZE];
@@ -639,13 +638,12 @@ ixgbe_process_vf_msg(if_ctx_t ctx, struct ixgbe_vf *vf)
 	}
 } /* ixgbe_process_vf_msg */
 
-
 /* Tasklet for handling VF -> PF mailbox messages */
 void
 ixgbe_handle_mbx(void *context)
 {
-	if_ctx_t        ctx = context;
-	struct ixgbe_softc  *sc = iflib_get_softc(ctx);
+	if_ctx_t ctx = context;
+	struct ixgbe_softc *sc = iflib_get_softc(ctx);
 	struct ixgbe_hw *hw;
 	struct ixgbe_vf *vf;
 	int i;
@@ -656,13 +654,16 @@ ixgbe_handle_mbx(void *context)
 		vf = &sc->vfs[i];
 
 		if (vf->flags & IXGBE_VF_ACTIVE) {
-			if (hw->mbx.ops[vf->pool].check_for_rst(hw, vf->pool) == 0)
+			if (hw->mbx.ops[vf->pool].check_for_rst(hw,
+			    vf->pool) == 0)
 				ixgbe_process_vf_reset(sc, vf);
 
-			if (hw->mbx.ops[vf->pool].check_for_msg(hw, vf->pool) == 0)
+			if (hw->mbx.ops[vf->pool].check_for_msg(hw,
+			    vf->pool) == 0)
 				ixgbe_process_vf_msg(ctx, vf);
 
-			if (hw->mbx.ops[vf->pool].check_for_ack(hw, vf->pool) == 0)
+			if (hw->mbx.ops[vf->pool].check_for_ack(hw,
+			    vf->pool) == 0)
 				ixgbe_process_vf_ack(sc, vf);
 		}
 	}
@@ -799,27 +800,27 @@ ixgbe_initialize_iov(struct ixgbe_softc *sc)
 
 	/* RMW appropriate registers based on IOV mode */
 	/* Read... */
-	mrqc    = IXGBE_READ_REG(hw, IXGBE_MRQC);
+	mrqc = IXGBE_READ_REG(hw, IXGBE_MRQC);
 	gcr_ext = IXGBE_READ_REG(hw, IXGBE_GCR_EXT);
-	gpie    = IXGBE_READ_REG(hw, IXGBE_GPIE);
+	gpie = IXGBE_READ_REG(hw, IXGBE_GPIE);
 	/* Modify... */
-	mrqc    &= ~IXGBE_MRQC_MRQE_MASK;
-	mtqc     =  IXGBE_MTQC_VT_ENA;      /* No initial MTQC read needed */
-	gcr_ext |=  IXGBE_GCR_EXT_MSIX_EN;
+	mrqc &= ~IXGBE_MRQC_MRQE_MASK;
+	mtqc = IXGBE_MTQC_VT_ENA;      /* No initial MTQC read needed */
+	gcr_ext |= IXGBE_GCR_EXT_MSIX_EN;
 	gcr_ext &= ~IXGBE_GCR_EXT_VT_MODE_MASK;
-	gpie    &= ~IXGBE_GPIE_VTMODE_MASK;
+	gpie &= ~IXGBE_GPIE_VTMODE_MASK;
 	switch (sc->iov_mode) {
 	case IXGBE_64_VM:
-		mrqc    |= IXGBE_MRQC_VMDQRSS64EN;
-		mtqc    |= IXGBE_MTQC_64VF;
+		mrqc |= IXGBE_MRQC_VMDQRSS64EN;
+		mtqc |= IXGBE_MTQC_64VF;
 		gcr_ext |= IXGBE_GCR_EXT_VT_MODE_64;
-		gpie    |= IXGBE_GPIE_VTMODE_64;
+		gpie |= IXGBE_GPIE_VTMODE_64;
 		break;
 	case IXGBE_32_VM:
-		mrqc    |= IXGBE_MRQC_VMDQRSS32EN;
-		mtqc    |= IXGBE_MTQC_32VF;
+		mrqc |= IXGBE_MRQC_VMDQRSS32EN;
+		mtqc |= IXGBE_MTQC_32VF;
 		gcr_ext |= IXGBE_GCR_EXT_VT_MODE_32;
-		gpie    |= IXGBE_GPIE_VTMODE_32;
+		gpie |= IXGBE_GPIE_VTMODE_32;
 		break;
 	default:
 		panic("Unexpected SR-IOV mode %d", sc->iov_mode);

sys/dev/ixgbe/ix_txrx.c

@@ -1,4 +1,4 @@
-/******************************************************************************
+/*****************************************************************************
 
   Copyright (c) 2001-2017, Intel Corporation
   All rights reserved.
@@ -29,7 +29,7 @@
   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   POSSIBILITY OF SUCH DAMAGE.
 
-******************************************************************************/
+*****************************************************************************/
 
 #ifndef IXGBE_STANDALONE_BUILD
 #include "opt_inet.h"
@@ -80,7 +80,7 @@ ixgbe_tx_ctx_setup(struct ixgbe_adv_tx_context_desc *TXD, if_pkt_info_t pi)
 {
 	uint32_t vlan_macip_lens, type_tucmd_mlhl;
 	uint32_t olinfo_status, mss_l4len_idx, pktlen, offload;
-	u8  ehdrlen;
+	u8 ehdrlen;
 
 	offload = true;
 	olinfo_status = mss_l4len_idx = vlan_macip_lens = type_tucmd_mlhl = 0;
@@ -105,9 +105,12 @@ ixgbe_tx_ctx_setup(struct ixgbe_adv_tx_context_desc *TXD, if_pkt_info_t pi)
 	/* First check if TSO is to be used */
 	if (pi->ipi_csum_flags & CSUM_TSO) {
 		/* This is used in the transmit desc in encap */
-		pktlen = pi->ipi_len - ehdrlen - pi->ipi_ip_hlen - pi->ipi_tcp_hlen;
-		mss_l4len_idx |= (pi->ipi_tso_segsz << IXGBE_ADVTXD_MSS_SHIFT);
-		mss_l4len_idx |= (pi->ipi_tcp_hlen << IXGBE_ADVTXD_L4LEN_SHIFT);
+		pktlen = pi->ipi_len - ehdrlen - pi->ipi_ip_hlen -
+		    pi->ipi_tcp_hlen;
+		mss_l4len_idx |=
+		    (pi->ipi_tso_segsz << IXGBE_ADVTXD_MSS_SHIFT);
+		mss_l4len_idx |=
+		    (pi->ipi_tcp_hlen << IXGBE_ADVTXD_L4LEN_SHIFT);
 	}
 
 	olinfo_status |= pktlen << IXGBE_ADVTXD_PAYLEN_SHIFT;
@@ -126,7 +129,8 @@ ixgbe_tx_ctx_setup(struct ixgbe_adv_tx_context_desc *TXD, if_pkt_info_t pi)
 
 	switch (pi->ipi_ipproto) {
 	case IPPROTO_TCP:
-		if (pi->ipi_csum_flags & (CSUM_IP_TCP | CSUM_IP6_TCP | CSUM_TSO))
+		if (pi->ipi_csum_flags &
+		    (CSUM_IP_TCP | CSUM_IP6_TCP | CSUM_TSO))
 			type_tucmd_mlhl |= IXGBE_ADVTXD_TUCMD_L4T_TCP;
 		else
 			offload = false;
@@ -168,17 +172,17 @@ ixgbe_tx_ctx_setup(struct ixgbe_adv_tx_context_desc *TXD, if_pkt_info_t pi)
 static int
 ixgbe_isc_txd_encap(void *arg, if_pkt_info_t pi)
 {
-	struct ixgbe_softc               *sc = arg;
-	if_softc_ctx_t                   scctx = sc->shared;
-	struct ix_tx_queue               *que = &sc->tx_queues[pi->ipi_qsidx];
-	struct tx_ring                   *txr = &que->txr;
-	int                              nsegs = pi->ipi_nsegs;
-	bus_dma_segment_t                *segs = pi->ipi_segs;
-	union ixgbe_adv_tx_desc          *txd = NULL;
+	struct ixgbe_softc *sc = arg;
+	if_softc_ctx_t scctx = sc->shared;
+	struct ix_tx_queue *que = &sc->tx_queues[pi->ipi_qsidx];
+	struct tx_ring *txr = &que->txr;
+	int nsegs = pi->ipi_nsegs;
+	bus_dma_segment_t *segs = pi->ipi_segs;
+	union ixgbe_adv_tx_desc *txd = NULL;
 	struct ixgbe_adv_tx_context_desc *TXD;
-	int                              i, j, first, pidx_last;
-	uint32_t                         olinfo_status, cmd, flags;
-	qidx_t                           ntxd;
+	int i, j, first, pidx_last;
+	uint32_t olinfo_status, cmd, flags;
+	qidx_t ntxd;
 
 	cmd =  (IXGBE_ADVTXD_DTYP_DATA |
 		IXGBE_ADVTXD_DCMD_IFCS | IXGBE_ADVTXD_DCMD_DEXT);
@@ -249,9 +253,9 @@ ixgbe_isc_txd_encap(void *arg, if_pkt_info_t pi)
 static void
 ixgbe_isc_txd_flush(void *arg, uint16_t txqid, qidx_t pidx)
 {
-	struct ixgbe_softc     *sc = arg;
+	struct ixgbe_softc *sc = arg;
 	struct ix_tx_queue *que = &sc->tx_queues[txqid];
-	struct tx_ring     *txr = &que->txr;
+	struct tx_ring *txr = &que->txr;
 
 	IXGBE_WRITE_REG(&sc->hw, txr->tail, pidx);
 } /* ixgbe_isc_txd_flush */
@@ -263,14 +267,14 @@ static int
 ixgbe_isc_txd_credits_update(void *arg, uint16_t txqid, bool clear)
 {
 	struct ixgbe_softc *sc = arg;
-	if_softc_ctx_t     scctx = sc->shared;
+	if_softc_ctx_t scctx = sc->shared;
 	struct ix_tx_queue *que = &sc->tx_queues[txqid];
-	struct tx_ring     *txr = &que->txr;
-	qidx_t             processed = 0;
-	int                updated;
-	qidx_t             cur, prev, ntxd, rs_cidx;
-	int32_t            delta;
-	uint8_t            status;
+	struct tx_ring *txr = &que->txr;
+	qidx_t processed = 0;
+	int updated;
+	qidx_t cur, prev, ntxd, rs_cidx;
+	int32_t delta;
+	uint8_t status;
 
 	rs_cidx = txr->tx_rs_cidx;
 	if (rs_cidx == txr->tx_rs_pidx)
@@ -319,9 +323,9 @@ ixgbe_isc_txd_credits_update(void *arg, uint16_t txqid, bool clear)
 static void
 ixgbe_isc_rxd_refill(void *arg, if_rxd_update_t iru)
 {
-	struct ixgbe_softc *sc   = arg;
-	struct ix_rx_queue *que  = &sc->rx_queues[iru->iru_qsidx];
-	struct rx_ring *rxr      = &que->rxr;
+	struct ixgbe_softc *sc = arg;
+	struct ix_rx_queue *que = &sc->rx_queues[iru->iru_qsidx];
+	struct rx_ring *rxr = &que->rxr;
 	uint64_t *paddrs;
 	int i;
 	uint32_t next_pidx, pidx;
@@ -342,11 +346,12 @@ ixgbe_isc_rxd_refill(void *arg, if_rxd_update_t iru)
  * ixgbe_isc_rxd_flush
  ************************************************************************/
 static void
-ixgbe_isc_rxd_flush(void *arg, uint16_t qsidx, uint8_t flidx __unused, qidx_t pidx)
+ixgbe_isc_rxd_flush(void *arg, uint16_t qsidx, uint8_t flidx __unused,
+    qidx_t pidx)
 {
-	struct ixgbe_softc *sc  = arg;
+	struct ixgbe_softc *sc = arg;
 	struct ix_rx_queue *que = &sc->rx_queues[qsidx];
-	struct rx_ring     *rxr = &que->rxr;
+	struct rx_ring *rxr = &que->rxr;
 
 	IXGBE_WRITE_REG(&sc->hw, rxr->tail, pidx);
 } /* ixgbe_isc_rxd_flush */
@@ -357,12 +362,12 @@ ixgbe_isc_rxd_flush(void *arg, uint16_t qsidx, uint8_t flidx __unused, qidx_t pi
 static int
 ixgbe_isc_rxd_available(void *arg, uint16_t qsidx, qidx_t pidx, qidx_t budget)
 {
-	struct ixgbe_softc      *sc = arg;
-	struct ix_rx_queue      *que = &sc->rx_queues[qsidx];
-	struct rx_ring          *rxr = &que->rxr;
+	struct ixgbe_softc *sc = arg;
+	struct ix_rx_queue *que = &sc->rx_queues[qsidx];
+	struct rx_ring *rxr = &que->rxr;
 	union ixgbe_adv_rx_desc *rxd;
-	uint32_t                 staterr;
-	int                      cnt, i, nrxd;
+	uint32_t staterr;
+	int cnt, i, nrxd;
 
 	nrxd = sc->shared->isc_nrxd[0];
 	for (cnt = 0, i = pidx; cnt < nrxd && cnt <= budget;) {
@@ -391,16 +396,16 @@ ixgbe_isc_rxd_available(void *arg, uint16_t qsidx, qidx_t pidx, qidx_t budget)
 static int
 ixgbe_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 {
-	struct ixgbe_softc       *sc = arg;
-	if_softc_ctx_t		 scctx = sc->shared;
-	struct ix_rx_queue       *que = &sc->rx_queues[ri->iri_qsidx];
-	struct rx_ring           *rxr = &que->rxr;
-	union ixgbe_adv_rx_desc  *rxd;
+	struct ixgbe_softc *sc = arg;
+	if_softc_ctx_t scctx = sc->shared;
+	struct ix_rx_queue *que = &sc->rx_queues[ri->iri_qsidx];
+	struct rx_ring *rxr = &que->rxr;
+	union ixgbe_adv_rx_desc *rxd;
 
-	uint16_t                  pkt_info, len, cidx, i;
-	uint32_t                  ptype;
-	uint32_t                  staterr = 0;
-	bool                      eop;
+	uint16_t pkt_info, len, cidx, i;
+	uint32_t ptype;
+	uint32_t staterr = 0;
+	bool eop;
 
 	i = 0;
 	cidx = ri->iri_cidx;
@@ -425,7 +430,8 @@ ixgbe_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri)
 		/* Make sure bad packets are discarded */
 		if (eop && (staterr & IXGBE_RXDADV_ERR_FRAME_ERR_MASK) != 0) {
 			if (sc->feat_en & IXGBE_FEATURE_VF)
-				if_inc_counter(ri->iri_ifp, IFCOUNTER_IERRORS, 1);
+				if_inc_counter(ri->iri_ifp, IFCOUNTER_IERRORS,
+				    1);
 
 			rxr->rx_discarded++;
 			return (EBADMSG);
@@ -478,7 +484,8 @@ ixgbe_rx_checksum(uint32_t staterr, if_rxd_info_t ri, uint32_t ptype)
 	uint8_t errors = (uint8_t)(staterr >> 24);
 
 	/* If there is a layer 3 or 4 error we are done */
-	if (__predict_false(errors & (IXGBE_RXD_ERR_IPE | IXGBE_RXD_ERR_TCPE)))
+	if (__predict_false(errors &
+	    (IXGBE_RXD_ERR_IPE | IXGBE_RXD_ERR_TCPE)))
 		return;
 
 	/* IP Checksum Good */
@@ -492,7 +499,8 @@ ixgbe_rx_checksum(uint32_t staterr, if_rxd_info_t ri, uint32_t ptype)
 		    (ptype & IXGBE_RXDADV_PKTTYPE_SCTP) != 0)) {
 			ri->iri_csum_flags |= CSUM_SCTP_VALID;
 		} else {
-			ri->iri_csum_flags |= CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
+			ri->iri_csum_flags |=
+			    CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
 			ri->iri_csum_data = htons(0xffff);
 		}
 	}

sys/dev/ixgbe/ixgbe.h

@@ -1,4 +1,4 @@
-/******************************************************************************
+/*****************************************************************************
   SPDX-License-Identifier: BSD-3-Clause
 
   Copyright (c) 2001-2017, Intel Corporation
@@ -30,7 +30,7 @@
   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   POSSIBILITY OF SUCH DAMAGE.
 
-******************************************************************************/
+*****************************************************************************/
 
 #ifndef _IXGBE_H_
 #define _IXGBE_H_
@@ -435,8 +435,8 @@ struct ixgbe_softc {
 	struct ixgbe_bp_data	bypass;
 
 	/* Firmware error check */
-	int                     recovery_mode;
-	struct callout          fw_mode_timer;
+	int			recovery_mode;
+	struct callout		fw_mode_timer;
 
 	/* Misc stats maintained by the driver */
 	unsigned long		dropped_pkts;

sys/dev/mlx5/driver.h

@@ -723,7 +723,8 @@ struct mlx5_core_dev {
 	u32			vsc_addr;
 	u32			issi;
 	struct mlx5_special_contexts special_contexts;
-	unsigned int module_status[MLX5_MAX_PORTS];
+	unsigned int module_status;
+	unsigned int module_num;
 	struct mlx5_flow_root_namespace *root_ns;
 	struct mlx5_flow_root_namespace *fdb_root_ns;
 	struct mlx5_flow_root_namespace *esw_egress_root_ns;

sys/dev/mlx5/mlx5_core/mlx5_eq.c

@@ -690,9 +690,9 @@ static const char *mlx5_port_module_event_error_type_to_string(u8 error_type)
 
 unsigned int mlx5_query_module_status(struct mlx5_core_dev *dev, int module_num)
 {
-	if (module_num < 0 || module_num >= MLX5_MAX_PORTS)
-		return 0;		/* undefined */
-	return dev->module_status[module_num];
+	if (module_num != dev->module_num)
+		return 0;		/* module num doesn't equal to what FW reported */
+	return dev->module_status;
 }
 
 static void mlx5_port_module_event(struct mlx5_core_dev *dev,
@@ -740,8 +740,8 @@ static void mlx5_port_module_event(struct mlx5_core_dev *dev,
 		    "Module %u, unknown status %d\n", module_num, module_status);
 	}
 	/* store module status */
-	if (module_num < MLX5_MAX_PORTS)
-		dev->module_status[module_num] = module_status;
+	dev->module_status = module_status;
+	dev->module_num = module_num;
 }
 
 static void mlx5_port_general_notification_event(struct mlx5_core_dev *dev,

sys/dev/mlx5/mlx5_en/en_hw_tls.h

@@ -84,7 +84,7 @@ struct mlx5e_tls {
 	struct workqueue_struct *wq;
 	uma_zone_t zone;
 	uint32_t max_resources;		/* max number of resources */
-	volatile uint32_t num_resources;	/* current number of resources */
+	int zone_max;
 	int init;			/* set when ready */
 	char zname[32];
 };

sys/dev/mlx5/mlx5_en/mlx5_en_hw_tls.c

@@ -81,17 +81,57 @@ static const char *mlx5e_tls_stats_desc[] = {
 
 static void mlx5e_tls_work(struct work_struct *);
 
+/*
+ * Expand the tls tag UMA zone in a sleepable context
+ */
+
+static void
+mlx5e_prealloc_tags(struct mlx5e_priv *priv, int nitems)
+{
+	struct mlx5e_tls_tag **tags;
+	int i;
+
+	tags = malloc(sizeof(tags[0]) * nitems,
+	    M_MLX5E_TLS, M_WAITOK);
+	for (i = 0; i < nitems; i++)
+		tags[i] = uma_zalloc(priv->tls.zone, M_WAITOK);
+	__compiler_membar();
+	for (i = 0; i < nitems; i++)
+		uma_zfree(priv->tls.zone, tags[i]);
+	free(tags, M_MLX5E_TLS);
+}
+
 static int
 mlx5e_tls_tag_import(void *arg, void **store, int cnt, int domain, int flags)
 {
 	struct mlx5e_tls_tag *ptag;
-	int i;
+	struct mlx5e_priv *priv = arg;
+	int err, i;
+
+	/*
+	 * mlx5_tls_open_tis() sleeps on a firmware command, so
+	 * zone allocations must be done from a sleepable context.
+	 * Note that the uma_zalloc() in mlx5e_tls_snd_tag_alloc()
+	 * is done with M_NOWAIT so that hitting the zone limit does
+	 * not cause the allocation to pause forever.
+	 */
 
 	for (i = 0; i != cnt; i++) {
 		ptag = malloc_domainset(sizeof(*ptag), M_MLX5E_TLS,
 		    mlx5_dev_domainset(arg), flags | M_ZERO);
+		if (ptag == NULL)
+			return (i);
+		ptag->tls = &priv->tls;
 		mtx_init(&ptag->mtx, "mlx5-tls-tag-mtx", NULL, MTX_DEF);
 		INIT_WORK(&ptag->work, mlx5e_tls_work);
+		err = mlx5_tls_open_tis(priv->mdev, 0, priv->tdn,
+		    priv->pdn, &ptag->tisn);
+		if (err) {
+			MLX5E_TLS_STAT_INC(ptag, tx_error, 1);
+			free(ptag, M_MLX5E_TLS);
+			return (i);
+		}
+
 		store[i] = ptag;
 	}
 	return (i);
@@ -114,7 +154,6 @@ mlx5e_tls_tag_release(void *arg, void **store, int cnt)
 
 		if (ptag->tisn != 0) {
 			mlx5_tls_close_tis(priv->mdev, ptag->tisn);
-			atomic_add_32(&ptls->num_resources, -1U);
 		}
 
 		mtx_destroy(&ptag->mtx);
@@ -136,20 +175,38 @@ mlx5e_tls_tag_zfree(struct mlx5e_tls_tag *ptag)
 	/* avoid leaking keys */
 	memset(ptag->crypto_params, 0, sizeof(ptag->crypto_params));
 
-	/* update number of TIS contexts */
-	if (ptag->tisn == 0)
-		atomic_add_32(&ptag->tls->num_resources, -1U);
-
 	/* return tag to UMA */
 	uma_zfree(ptag->tls->zone, ptag);
 }
 
+static int
+mlx5e_max_tag_proc(SYSCTL_HANDLER_ARGS)
+{
+	struct mlx5e_priv *priv = (struct mlx5e_priv *)arg1;
+	struct mlx5e_tls *ptls = &priv->tls;
+	int err;
+	unsigned int max_tags;
+
+	max_tags = ptls->zone_max;
+	err = sysctl_handle_int(oidp, &max_tags, arg2, req);
+	if (err != 0 || req->newptr == NULL )
+		return err;
+	if (max_tags == ptls->zone_max)
+		return 0;
+	if (max_tags > priv->tls.max_resources || max_tags == 0)
+		return (EINVAL);
+	ptls->zone_max = max_tags;
+	uma_zone_set_max(ptls->zone, ptls->zone_max);
+	return 0;
+}
+
 int
 mlx5e_tls_init(struct mlx5e_priv *priv)
 {
 	struct mlx5e_tls *ptls = &priv->tls;
 	struct sysctl_oid *node;
-	uint32_t x;
+	uint32_t max_dek, max_tis, x;
+	int zone_max = 0, prealloc_tags = 0;
 
 	if (MLX5_CAP_GEN(priv->mdev, tls_tx) == 0 ||
 	    MLX5_CAP_GEN(priv->mdev, log_max_dek) == 0)
@@ -164,13 +221,31 @@ mlx5e_tls_init(struct mlx5e_priv *priv)
 	snprintf(ptls->zname, sizeof(ptls->zname),
 	    "mlx5_%u_tls", device_get_unit(priv->mdev->pdev->dev.bsddev));
 
+
+	TUNABLE_INT_FETCH("hw.mlx5.tls_max_tags", &zone_max);
+	TUNABLE_INT_FETCH("hw.mlx5.tls_prealloc_tags", &prealloc_tags);
+
 	ptls->zone = uma_zcache_create(ptls->zname,
 	     sizeof(struct mlx5e_tls_tag), NULL, NULL, NULL, NULL,
-	     mlx5e_tls_tag_import, mlx5e_tls_tag_release, priv->mdev,
-	     UMA_ZONE_UNMANAGED);
+	     mlx5e_tls_tag_import, mlx5e_tls_tag_release, priv,
+	     UMA_ZONE_UNMANAGED | (prealloc_tags ? UMA_ZONE_NOFREE : 0));
 
 	/* shared between RX and TX TLS */
-	ptls->max_resources = 1U << (MLX5_CAP_GEN(priv->mdev, log_max_dek) - 1);
+	max_dek = 1U << (MLX5_CAP_GEN(priv->mdev, log_max_dek) - 1);
+	max_tis = 1U << (MLX5_CAP_GEN(priv->mdev, log_max_tis) - 1);
+	ptls->max_resources = MIN(max_dek, max_tis);
+
+	if (zone_max != 0) {
+		ptls->zone_max = zone_max;
+		if (ptls->zone_max > priv->tls.max_resources)
+			ptls->zone_max = priv->tls.max_resources;
+	} else {
+		ptls->zone_max = priv->tls.max_resources;
+	}
+
+	uma_zone_set_max(ptls->zone, ptls->zone_max);
+	if (prealloc_tags != 0)
+		mlx5e_prealloc_tags(priv, ptls->zone_max);
 
 	for (x = 0; x != MLX5E_TLS_STATS_NUM; x++)
 		ptls->stats.arg[x] = counter_u64_alloc(M_WAITOK);
@@ -183,6 +258,10 @@ mlx5e_tls_init(struct mlx5e_priv *priv)
 	if (node == NULL)
 		return (0);
 
+	SYSCTL_ADD_PROC(&priv->sysctl_ctx, SYSCTL_CHILDREN(node), OID_AUTO, "tls_max_tag",
+	    CTLFLAG_RW | CTLTYPE_UINT | CTLFLAG_MPSAFE, priv, 0, mlx5e_max_tag_proc,
+	    "IU", "Max number of TLS offload session tags");
+
 	mlx5e_create_counter_stats(&ptls->ctx,
 	    SYSCTL_CHILDREN(node), "stats",
 	    mlx5e_tls_stats_desc, MLX5E_TLS_STATS_NUM,
@@ -206,9 +285,6 @@ mlx5e_tls_cleanup(struct mlx5e_priv *priv)
 	uma_zdestroy(ptls->zone);
 	destroy_workqueue(ptls->wq);
 
-	/* check if all resources are freed */
-	MPASS(priv->tls.num_resources == 0);
-
 	for (x = 0; x != MLX5E_TLS_STATS_NUM; x++)
 		counter_u64_free(ptls->stats.arg[x]);
 }
@@ -225,7 +301,7 @@ mlx5e_tls_st_init(struct mlx5e_priv *priv, struct mlx5e_tls_tag *ptag)
 		    priv->pdn, &ptag->tisn);
 		if (err) {
 			MLX5E_TLS_STAT_INC(ptag, tx_error, 1);
-			return (err);
+			return (-err);
 		}
 	}
 	MLX5_SET(sw_tls_cntx, ptag->crypto_params, progress.pd, ptag->tisn);
@@ -238,7 +314,7 @@ mlx5e_tls_st_init(struct mlx5e_priv *priv, struct mlx5e_tls_tag *ptag)
 	    &ptag->dek_index);
 	if (err) {
 		MLX5E_TLS_STAT_INC(ptag, tx_error, 1);
-		return (err);
+		return (-err);
 	}
 
 	MLX5_SET(sw_tls_cntx, ptag->crypto_params, param.dek_index, ptag->dek_index);
@@ -318,8 +394,7 @@ mlx5e_tls_set_params(void *ctx, const struct tls_session_params *en)
 CTASSERT(MLX5E_TLS_ST_INIT == 0);
 
 int
-mlx5e_tls_snd_tag_alloc(if_t ifp,
-    union if_snd_tag_alloc_params *params,
+mlx5e_tls_snd_tag_alloc(if_t ifp, union if_snd_tag_alloc_params *params,
     struct m_snd_tag **ppmt)
 {
 	union if_snd_tag_alloc_params rl_params;
@@ -334,28 +409,16 @@ mlx5e_tls_snd_tag_alloc(if_t ifp,
 	if (priv->gone != 0 || priv->tls.init == 0)
 		return (EOPNOTSUPP);
 
-	/* allocate new tag from zone, if any */
-	ptag = uma_zalloc(priv->tls.zone, M_WAITOK);
+	ptag = uma_zalloc(priv->tls.zone, M_NOWAIT);
+ 	if (ptag == NULL)
+ 		return (ENOMEM);
 
 	/* sanity check default values */
 	MPASS(ptag->dek_index == 0);
 	MPASS(ptag->dek_index_ok == 0);
 
-	/* setup TLS tag */
-	ptag->tls = &priv->tls;
-
 	/* check if there is no TIS context */
-	if (ptag->tisn == 0) {
-		uint32_t value;
-
-		value = atomic_fetchadd_32(&priv->tls.num_resources, 1U);
-
-		/* check resource limits */
-		if (value >= priv->tls.max_resources) {
-			error = ENOMEM;
-			goto failure;
-		}
-	}
+	KASSERT(ptag->tisn != 0, ("ptag %p w/0 tisn", ptag));
 
 	en = &params->tls.tls->params;
 
@@ -448,17 +511,9 @@ mlx5e_tls_snd_tag_alloc(if_t ifp,
 	/* reset state */
 	ptag->state = MLX5E_TLS_ST_INIT;
 
-	/*
-	 *  Try to immediately init the tag.  We may fail if the NIC's
-	 *  resources are tied up with send tags that are in the work
-	 *  queue, waiting to be freed.  So if we fail, put ourselves
-	 *  on the queue so as to try again after resouces have been freed.
-	 */
 	error = mlx5e_tls_st_init(priv, ptag);
-	if (error != 0) {
-		queue_work(priv->tls.wq, &ptag->work);
-		flush_work(&ptag->work);
-	}
+	if (error != 0)
+		goto failure;
 
 	return (0);
 

sys/dev/mlx5/mlx5_en/mlx5_en_main.c

@@ -3404,6 +3404,51 @@ mlx5e_set_rx_mode(if_t ifp)
 	queue_work(priv->wq, &priv->set_rx_mode_work);
 }
 
+static bool
+mlx5e_is_ipsec_capable(struct mlx5_core_dev *mdev)
+{
+#ifdef IPSEC_OFFLOAD
+	if ((mlx5_ipsec_device_caps(mdev) & MLX5_IPSEC_CAP_PACKET_OFFLOAD) != 0)
+		return (true);
+#endif
+	return (false);
+}
+
+static bool
+mlx5e_is_ratelimit_capable(struct mlx5_core_dev *mdev)
+{
+#ifdef RATELIMIT
+	if (MLX5_CAP_GEN(mdev, qos) &&
+	    MLX5_CAP_QOS(mdev, packet_pacing))
+		return (true);
+#endif
+	return (false);
+}
+
+static bool
+mlx5e_is_tlstx_capable(struct mlx5_core_dev *mdev)
+{
+#ifdef KERN_TLS
+	if (MLX5_CAP_GEN(mdev, tls_tx) != 0 &&
+	    MLX5_CAP_GEN(mdev, log_max_dek) != 0)
+		return (true);
+#endif
+	return (false);
+}
+
+static bool
+mlx5e_is_tlsrx_capable(struct mlx5_core_dev *mdev)
+{
+#ifdef KERN_TLS
+	if (MLX5_CAP_GEN(mdev, tls_rx) != 0 &&
+	    MLX5_CAP_GEN(mdev, log_max_dek) != 0 &&
+	    MLX5_CAP_FLOWTABLE_NIC_RX(mdev,
+	    ft_field_support.outer_ip_version) != 0)
+		return (true);
+#endif
+	return (false);
+}
+
 static int
 mlx5e_ioctl(if_t ifp, u_long command, caddr_t data)
 {
@@ -3507,6 +3552,24 @@ mlx5e_ioctl(if_t ifp, u_long command, caddr_t data)
 		drv_ioctl_data = (struct siocsifcapnv_driver_data *)data;
 		PRIV_LOCK(priv);
 siocsifcap_driver:
+		if (!mlx5e_is_tlstx_capable(priv->mdev)) {
+			drv_ioctl_data->reqcap &= ~(IFCAP_TXTLS4 |
+			    IFCAP_TXTLS6);
+		}
+		if (!mlx5e_is_tlsrx_capable(priv->mdev)) {
+		        drv_ioctl_data->reqcap &= ~(
+			    IFCAP2_BIT(IFCAP2_RXTLS4) |
+			    IFCAP2_BIT(IFCAP2_RXTLS6));
+		}
+		if (!mlx5e_is_ipsec_capable(priv->mdev)) {
+			drv_ioctl_data->reqcap &=
+			    ~IFCAP2_BIT(IFCAP2_IPSEC_OFFLOAD);
+		}
+		if (!mlx5e_is_ratelimit_capable(priv->mdev)) {
+			drv_ioctl_data->reqcap &= ~(IFCAP_TXTLS_RTLMT |
+			    IFCAP_TXRTLMT);
+		}
+
 		mask = drv_ioctl_data->reqcap ^ if_getcapenable(ifp);
 
 		if (mask & IFCAP_TXCSUM) {
@@ -4535,29 +4598,20 @@ mlx5e_create_ifp(struct mlx5_core_dev *mdev)
 	if_setcapabilitiesbit(ifp, IFCAP_TSO | IFCAP_VLAN_HWTSO, 0);
 	if_setcapabilitiesbit(ifp, IFCAP_HWSTATS | IFCAP_HWRXTSTMP, 0);
 	if_setcapabilitiesbit(ifp, IFCAP_MEXTPG, 0);
-#ifdef KERN_TLS
-	if (MLX5_CAP_GEN(mdev, tls_tx) != 0 &&
-	    MLX5_CAP_GEN(mdev, log_max_dek) != 0)
+	if (mlx5e_is_tlstx_capable(mdev))
 		if_setcapabilitiesbit(ifp, IFCAP_TXTLS4 | IFCAP_TXTLS6, 0);
-	if (MLX5_CAP_GEN(mdev, tls_rx) != 0 &&
-	    MLX5_CAP_GEN(mdev, log_max_dek) != 0 &&
-	    MLX5_CAP_FLOWTABLE_NIC_RX(mdev,
-	    ft_field_support.outer_ip_version) != 0)
+	if (mlx5e_is_tlsrx_capable(mdev))
 		if_setcapabilities2bit(ifp, IFCAP2_BIT(IFCAP2_RXTLS4) |
 		    IFCAP2_BIT(IFCAP2_RXTLS6), 0);
-#endif
-#ifdef RATELIMIT
-	if (MLX5_CAP_GEN(mdev, qos) &&
-	    MLX5_CAP_QOS(mdev, packet_pacing))
-		if_setcapabilitiesbit(ifp, IFCAP_TXRTLMT | IFCAP_TXTLS_RTLMT,
-		    0);
-#endif
+	if (mlx5e_is_ratelimit_capable(mdev)) {
+		if_setcapabilitiesbit(ifp, IFCAP_TXRTLMT, 0);
+		if (mlx5e_is_tlstx_capable(mdev))
+			if_setcapabilitiesbit(ifp, IFCAP_TXTLS_RTLMT, 0);
+	}
 	if_setcapabilitiesbit(ifp, IFCAP_VXLAN_HWCSUM | IFCAP_VXLAN_HWTSO, 0);
-#ifdef IPSEC_OFFLOAD
-	if (mlx5_ipsec_device_caps(mdev) & MLX5_IPSEC_CAP_PACKET_OFFLOAD)
+	if (mlx5e_is_ipsec_capable(mdev))
 		if_setcapabilities2bit(ifp, IFCAP2_BIT(IFCAP2_IPSEC_OFFLOAD),
 		    0);
-#endif
 
 	if_setsndtagallocfn(ifp, mlx5e_snd_tag_alloc);
 #ifdef RATELIMIT

sys/dev/sound/dummy.c

@@ -67,6 +67,24 @@ struct dummy_softc {
 	struct mtx *lock;
 };
 
+static bool
+dummy_active(struct dummy_softc *sc)
+{
+	struct dummy_chan *ch;
+	int i;
+
+	snd_mtxassert(sc->lock);
+
+	for (i = 0; i < sc->chnum; i++) {
+		ch = &sc->chans[i];
+		if (ch->run)
+			return (true);
+	}
+
+	/* No channel is running at the moment. */
+	return (false);
+}
+
 static void
 dummy_chan_io(void *arg)
 {
@@ -74,7 +92,9 @@ dummy_chan_io(void *arg)
 	struct dummy_chan *ch;
 	int i = 0;
 
-	snd_mtxlock(sc->lock);
+	/* Do not reschedule if no channel is running. */
+	if (!dummy_active(sc))
+		return;
 
 	for (i = 0; i < sc->chnum; i++) {
 		ch = &sc->chans[i];
@@ -89,8 +109,6 @@ dummy_chan_io(void *arg)
 		snd_mtxlock(sc->lock);
 	}
 	callout_schedule(&sc->callout, 1);
-
-	snd_mtxunlock(sc->lock);
 }
 
 static int
@@ -179,15 +197,15 @@ dummy_chan_trigger(kobj_t obj, void *data, int go)
 
 	switch (go) {
 	case PCMTRIG_START:
-		if (!callout_active(&sc->callout))
-			callout_reset(&sc->callout, 1, dummy_chan_io, sc);
 		ch->ptr = 0;
 		ch->run = 1;
+		callout_reset(&sc->callout, 1, dummy_chan_io, sc);
 		break;
 	case PCMTRIG_STOP:
 	case PCMTRIG_ABORT:
 		ch->run = 0;
-		if (callout_active(&sc->callout))
+		/* If all channels are stopped, stop the callout as well. */
+		if (!dummy_active(sc))
 			callout_stop(&sc->callout);
 	default:
 		break;
@@ -292,6 +310,7 @@ dummy_attach(device_t dev)
 	sc = device_get_softc(dev);
 	sc->dev = dev;
 	sc->lock = snd_mtxcreate(device_get_nameunit(dev), "snd_dummy softc");
+	callout_init_mtx(&sc->callout, sc->lock, 0);
 
 	sc->cap_fmts[0] = SND_FORMAT(AFMT_S32_LE, 2, 0);
 	sc->cap_fmts[1] = SND_FORMAT(AFMT_S24_LE, 2, 0);
@@ -316,7 +335,6 @@ dummy_attach(device_t dev)
 	if (pcm_register(dev, status))
 		return (ENXIO);
 	mixer_init(dev, &dummy_mixer_class, sc);
-	callout_init(&sc->callout, 1);
 
 	return (0);
 }
@@ -327,8 +345,8 @@ dummy_detach(device_t dev)
 	struct dummy_softc *sc = device_get_softc(dev);
 	int err;
 
-	callout_drain(&sc->callout);
 	err = pcm_unregister(dev);
+	callout_drain(&sc->callout);
 	snd_mtxfree(sc->lock);
 
 	return (err);