SVNews

NOTE: This service is experimental and subject to change! Use at your own risk!

Year:   Month: - or - Filter committers: - or -
Filter paths:
Include diffs if smaller than lines.

2015 - 08 - 02   (today)
00:03:08 - r286166
( 8 minutes ago)
by markj
Affects: 
(6 files)
/head/sys/kern/*
/head/sys/sys/proc.h
Don't modify curthread->td_locks unless INVARIANTS is enabled.

This field is only used in a KASSERT that verifies that no locks are held
when returning to user mode. Moreover, the td_locks accounting is only
correct when LOCK_DEBUG > 0, which is implied by INVARIANTS.

Reviewed by:    jhb
MFC after:      1 week
Differential Revision:  https://reviews.freebsd.org/D3205
2015 - 08 - 01   (yesterday)
23:10:36 - r286165
(1 hour ago)
by gonzo
Affects:  /head/sys/arm/ti/ti_gpio.c
Set output pin initial value based on pin's pinmux pullup/pulldown setup

Some of FDT blobs for AM335x-based devices use pinmux pullup/pulldown
flag to setup initial GPIO ouputp value, e.g. 4DCAPE-43 sets LCD DATAEN
signal this way. It works for Linux because Linux driver does not enforce
pin direction until after it's requested by consumer. So input with pullup
flag set acts as output with GPIO_HIGH value

Reviewed by:    loos
22:56:42 - r286164
(1 hour 14 minutes ago)
by rmacklem
Affects: 
(2 files)
/stable/10/sys/fs/nfsserver/nfs_nfsdstate.c
MFC: r286046
This patch fixes a problem where, if the NFSv4 server has a previous
unconfirmed clientid structure for the same client on the last hash list,
this old entry would not be removed/deleted. I do not think this bug would have
caused serious problems, since the new entry would have been before the old one
on the list. This old entry would have eventually been scavenged/removed.
22:00:25 - r286163
(2 hours 11 minutes ago)
by jilles
Affects:  /head/etc/rc.subr
rc.subr: Allow rc.conf.d with multi-directory local_startup.

I also changed ${...%*/rc.d} to ${...%/rc.d} since the shortest match always
has an empty string for the asterisk.

PR:             201641
Submitted by:   Jamie Landeg-Jones (original version)
MFC after:      1 week
20:40:37 - r286162
(3 hours 30 minutes ago)
by hselasky
Affects:  /head/sys/dev/e1000/if_em.c
/head/sys/dev/e1000/if_igb.c
Free mbufs when busdma loading fails.

Reviewed by:    erj, sbruno
MFC after:      1 month
17:48:35 - r286161
(6 hours 22 minutes ago)
by junovitch
Affects:  /head/usr.bin/calendar/calendars/calendar.freebsd
- Add myself to the calendar.freebsd. (fix typo in r286160)

Approved by:    delphij (mentor)
Differential Revision:  https://reviews.FreeBSD.org/D3265
17:29:52 - r286160
(6 hours 41 minutes ago)
by junovitch
Affects:  /head/share/misc/committers-ports.dot
/head/usr.bin/calendar/calendars/calendar.freebsd
- Add myself to the committers-ports.dot and add my mentor relationship.
- Add myself to the calendar.freebsd.

Approved by:    delphij (mentor)
Differential Revision:  https://reviews.FreeBSD.org/D3265
17:27:47 - r286159
(6 hours 43 minutes ago)
by jmg
Affects:  /head/usr.bin/getopt/getopt.1
use : instead of true...

Change file file to file1 file2, partly for igor, and partly because
it's odd to pass the same file to a command twice..
16:27:52 - r286158
(7 hours 43 minutes ago)
by jhb
Affects:  /head/sys/kern/sys_process.c
/head/tests/sys/kern/ptrace_test.c
Clear P_TRACED before reparenting a detached process back to its
original parent. Otherwise the debugee will be set as an orphan of
the debugger.

Add tests for tracing forks via PT_FOLLOW_FORK.

Reviewed by:    kib
MFC after:      2 weeks
Differential Revision:  https://reviews.freebsd.org/D2809
12:20:55 - r286157
(11 hours 50 minutes ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_conf.c
Remove things that crept in after badly checked revert
12:18:48 - r286156
(11 hours 52 minutes ago)
by bapt
Affects: 
(4 files)
/head/usr.sbin/pw/*
Partial revert of r286152

More work needed on the cli validation
11:52:48 - r286155
(12 hours 18 minutes ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_conf.c
Fix build
11:31:59 - r286154
(12 hours 39 minutes ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_conf.c
Fix formatting of new code
Fix sorting or errstr
Remove useless initialisation or errstr

Reported by:    bde
10:40:17 - r286153
(13 hours 31 minutes ago)
by bapt
Affects:  /head/usr.sbin/chkgrp/chkgrp.c
Revert r286148
10:25:55 - r286152
(13 hours 45 minutes ago)
by bapt
Affects: 
(4 files)
/head/usr.sbin/pw/*
Validate expiration days and password days from commmand line and pw.conf
10:10:13 - r286151
(14 hours 1 minute ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_conf.c
Validate the max_uid/max_gid boundaries and entry type in pw.conf
09:55:47 - r286150
(14 hours 15 minutes ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_conf.c
/head/usr.sbin/pw/pw_group.c
/head/usr.sbin/pw/pw_user.c
Cast uid/git to uintmax_t when using printf-like functions so the size of
uid/gid size remains a implementation detail
08:39:55 - r286149
(15 hours 31 minutes ago)
by bapt
Affects:  /head/usr.sbin/chkgrp/chkgrp.c
Handle +:*:: the same way we handle +::: when checking group file

The first is what the documentation recommands for NIS client
08:35:20 - r286148
(15 hours 36 minutes ago)
by bapt
Affects:  /head/usr.sbin/chkgrp/chkgrp.c
Use strtoumax instead of strtoul
07:51:48 - r286147
(16 hours 19 minutes ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_fd.c
Allow CloudABI processes to create shared memory objects.

Summary:
Use the newly created `kern_shm_open()` function to create objects with
just the rights that are actually needed.

Reviewers: jhb, kib

Subscribers: imp

Differential Revision: https://reviews.freebsd.org/D3260
07:21:14 - r286146
(16 hours 50 minutes ago)
by ed
Affects:  /head/sys/kern/uipc_shm.c
/head/sys/sys/syscallsubr.h
Add kern_shm_open().

This allows you to specify the capabilities that the new file descriptor
should have. This allows us to create shared memory objects that only
have the rights we're interested in.

The idea behind restricting the rights is that it makes it a lot easier
for CloudABI to get consistent behaviour across different operating
systems. We only need to make sure that a shared memory implementation
consistently implements the operations that are whitelisted.

Approved by:    kib
Obtained from:  https://github.com/NuxiNL/freebsd
03:37:00 - r286145
(20 hours 34 minutes ago)
by kib
Affects: 
(3 files)
/stable/10/sys/vm/vm_object.c
/stable/10/sys/vm/vm_object.h
MFC r285878:
Revert r173708's modifications to vm_object_page_remove().
This fixes inconsistencies encountered by vm_object_unwire() or
by the buffer cache when the file is truncated.
01:29:55 - r286144
(22 hours 41 minutes ago)
by pfg
Affects:  /head/usr.bin/wall/ttymsg.c
Buffer overflow in wall(1).

Revert r286102 and apply a cleaner fix.
Tested for overflows by FORTIFY_SOURCE GSoC (with clang).

Suggested by:   bde
Reviewed by:    Oliver Pinter
Tested by:      Oliver Pinter
MFC after:      3 days
2015 - 07 - 31   (2 days ago)
23:40:18 - r286143
(1 day ago)
by gnn
Affects:  /head/sbin/setkey/parse.y
/head/sbin/setkey/token.l
Add support for keys that include 4 byte SALT values,
including GCM and ICM/CTR modes for AES.

Reviewed by:    jmg
MFC after:      1 week
Sponsored by:   Rubicon Communications (Netgate)
21:43:27 - r286142
(1 day 2 hours ago)
by loos
Affects:  /head/sys/net/bpf.c
/head/sys/net/bpf.h
Remove two unnecessary sleeps from the hot path in bpf(4).

The first one never triggers because bpf_canfreebuf() can only be true for
zero-copy buffers and zero-copy buffers are not read with read(2).

The second also never triggers, because we check the free buffer before
calling ROTATE_BUFFERS().  If the hold buffer is in use the free buffer
will be NULL and there is nothing else to do besides drop the packet.  If
the free buffer isn't NULL the hold buffer _is_ free and it is safe to
rotate the buffers.

Update the comment in ROTATE_BUFFERS macro to match the logic described
here.

While here fix a few typos in comments.

MFC after:      2 weeks
Sponsored by:   Rubicon Communications (Netgate)
21:31:58 - r286141
(1 day 2 hours ago)
by rmacklem
Affects: 
(2 files)
/stable/10/sys/fs/nfsclient/nfs_clvfsops.c
MFC: r285113
If a "principal" argument isn't provided for a Kerberized NFS mount,
the kernel would generate a bogus one with a ":/<path>" suffix.
This would only occur for the case where there was no explicit
"principal" argument and the getaddrinfo() call in mount_nfs.c failed to a
return a cannonical name for the server.
This patch fixes this unusual case.
20:25:54 - r286140
(1 day 3 hours ago)
by loos
Affects:  /head/sys/net/bpf_buffer.c
Remove the sleep from the buffer allocation routine.

The buffer must be allocated (or even changed) before the interface is set
and thus, there is no need to verify if the buffer is in use.

MFC after:      2 weeks
Sponsored by:   Rubicon Communications (Netgate)
20:02:12 - r286139
(1 day 4 hours ago)
by loos
Affects:  /head/sys/net/bpf.c
Do not allocate the buffers at opening of the descriptor, because once
the buffer is allocated we are committed to a particular buffer method
(BPF_BUFMODE_BUFFER in this case).

If we are using zero-copy buffers, the userland program must register its
buffers before set the interface.

If we are using kernel memory buffers, we can allocate the buffer at the
time that the interface is being set.

This fix allows the usage of BIOCSETBUFMODE after r235746.

Update the comments to reflect the recent changes.

MFC after:      2 weeks
Sponsored by:   Rubicon Communications (Netgate)
17:41:53 - r286138
(1 day 6 hours ago)
by gnn
Affects:  /head/sbin/setkey/setkey.8
Summary: Remove unsupported 3des-deriv encryption algorithm from documentation.
16:46:30 - r286137
(1 day 7 hours ago)
by gjb
Affects:  /releng/10.2/sys/conf/newvers.sh
Forced commit to mark the real 10.2-RC2 point.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
16:45:42 - r286136
(1 day 7 hours ago)
by emaste
Affects:  /head/usr.bin/brandelf/brandelf.1
Use standard 'THE AUTHOR' license text

Approved by:    jmg
16:44:57 - r286135
(1 day 7 hours ago)
by gjb
Affects: 
(3 files)
/releng/10.2/sys/amd64/conf/GENERIC
/releng/10.2/sys/i386/conf/GENERIC
MFS r286132:
MFC r286131:
 Pull pmspcv (pms(4)) from GENERIC.  It has PCI ID conflicts
 with ahd(4), mvs(4), and likely other drivers.

Approved by:    re (implicit, release blocker)
Sponsored by:   The FreeBSD Foundation
15:54:34 - r286134
(1 day 8 hours ago)
by andrew
Affects:  /head/sys/arm64/arm64/machdep.c
Try to put the CPU into a low power state if we failed to otherwise halt
the system.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
15:32:32 - r286133
(1 day 8 hours ago)
by andrew
Affects:  /head/sys/arm64/arm64/stack_machdep.c
/head/sys/conf/files.arm64
Load the stack in stack_save and stack_save_td. This uses the generalised
unwind_frame function to read each stack frame until either the pc or stack
are no longer withing the kernel's address space.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
15:25:07 - r286132
(1 day 8 hours ago)
by gjb
Affects: 
(3 files)
/stable/10/sys/amd64/conf/GENERIC
/stable/10/sys/i386/conf/GENERIC
MFC r286131:
 Pull pmspcv (pms(4)) from GENERIC.  It has PCI ID conflicts
 with ahd(4), mvs(4), and likely other drivers.

With hat:       re
Sponsored by:   The FreeBSD Foundation
15:23:48 - r286131
(1 day 8 hours ago)
by gjb
Affects:  /head/sys/amd64/conf/GENERIC
/head/sys/i386/conf/GENERIC
Pull pmspcv (pms(4)) from GENERIC.  It has PCI ID conflicts
with ahd(4), mvs(4), and likely other drivers.

MFC after:      immediately
With hat:       re
Sponsored by:   The FreeBSD Foundation
14:17:26 - r286130
(1 day 9 hours ago)
by andrew
Affects: 
(4 files)
/head/sys/arm64/arm64/*
/head/sys/arm64/include/vmparam.h
/head/sys/conf/files.arm64
Add support for uma_small_alloc and uma_small_free, and make use of these.
This is copied from the amd64 version with minor changes. These should be
merged into a single file as from a quick look there are other copies of
the same file in other parts of the tree.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
13:39:51 - r286129
(1 day 10 hours ago)
by andrew
Affects:  /head/sys/arm64/arm64/mem.c
Add memrw. This has had minimal testing, and will likely panic the kernel
when trying to read data from outside the DMAP region. I expect this panic
to be from within uiomove_fromphys, which needs to grow support to support
such addresses.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
13:34:43 - r286128
(1 day 10 hours ago)
by andrew
Affects:  /head/sys/arm64/include/atomic.h
Add more atomic_swap_* functions.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
13:32:25 - r286127
(1 day 10 hours ago)
by andrew
Affects:  /head/sys/arm64/include/vmparam.h
Add VIRT_IN_DMAP to check if a virtual address is from the DMAP range.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
12:37:40 - r286126
(1 day 11 hours ago)
by emaste
Affects:  /head/contrib/elftoolchain/brandelf
Add ELF Tool Chain's brandelf(1) to contrib

Noticed by pfg after r286070 (ar and elfdump)
12:02:44 - r286125
(1 day 12 hours ago)
by garga
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf.c
MFC r285945, r285960:

Respect pf rule log option before log dropped packets with IP options or
dangerous v6 headers

Reviewed by:    gnn, eri
Approved by:    gnn, glebius
Obtained from:  pfSense
Sponsored by:   Netgate
Differential Revision:  https://reviews.freebsd.org/D3222
11:10:43 - r286124
(1 day 13 hours ago)
by erwin
Affects: 
(64 files)
/stable/9/contrib/bind9/*
Update BIND to 9.9.7-P2

This is a direct commit to stable/9 as BIND is no longer in head.

Sponsored by:   DK Hostmaster A/S
10:46:45 - r286123
(1 day 13 hours ago)
by ed
Affects:  /head/sys/kern/kern_fork.c
Fix accidental line wrapping introduced in r286122.
10:21:58 - r286122
(1 day 13 hours ago)
by ed
Affects: 
(6 files)
/head/sys/compat/*
/head/sys/kern/*
/head/sys/sys/proc.h
Limit rights on process descriptors.

On CloudABI, the rights bits returned by cap_rights_get() match up with
the operations that you can actually perform on the file descriptor.

Limiting the rights is good, because it makes it easier to get uniform
behaviour across different operating systems. If process descriptors on
FreeBSD would suddenly gain support for any new file operation, this
wouldn't become exposed to CloudABI processes without first extending
the rights.

Extend fork1() to gain a 'struct filecaps' argument that allows you to
construct process descriptors with custom rights. Use this in
cloudabi_sys_proc_fork() to limit the rights to just fstat() and
pdwait().

Obtained from:  https://github.com/NuxiNL/freebsd
10:00:45 - r286121
(1 day 14 hours ago)
by zbb
Affects: 
(4 files)
/head/sys/arm64/*
/head/sys/conf/options.arm64
Apply erratum for mrs ICC_IAR1_EL1 speculative execution on ThunderX

ERRATUM:     22978, 23154
PASS (rev.): 1.0/1.1

Reviewed by:   imp
Obtained from: Semihalf
Sponsored by:  The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3184
09:43:07 - r286120
(1 day 14 hours ago)
by smh
Affects: 
(3 files)
/stable/10/UPDATING
/stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c
MFC: r285946 and r285947

Add warning about low KSTACK_PAGES for ZFS use.

Sponsored by:   Multiplay
09:28:49 - r286119
(1 day 14 hours ago)
by hselasky
Affects: 
(2061 files)
/projects/hps_head/*
Merge ^/head r285084 through 286118.
09:12:31 - r286118
(1 day 14 hours ago)
by hselasky
Affects:  /head/sys/dev/usb/controller/dwc_otg.c
Limit the number of times we loop inside the DWC OTG poll handler to
avoid starving other fast interrupts. Fix a comment while at it.

MFC after:      1 week
Suggested by:   Svatopluk Kraus <onwahe@gmail.com>
09:10:56 - r286117
(1 day 15 hours ago)
by pho
Affects:  /user/pho/stress2/misc/parallelmount.sh
Reduce runtime and clean up script while here.

Sponsored by:   EMC / Isilon storage division
09:09:16 - r286116
(1 day 15 hours ago)
by pho
Affects:  /user/pho/stress2/misc/tmpfs4.sh
Update expected error message filter.

Sponsored by:   EMC / Isilon storage division
09:08:06 - r286115
(1 day 15 hours ago)
by pho
Affects:  /user/pho/stress2/misc/snap7.sh
Reduce runtime and handle umount(1) failure.

Sponsored by:   EMC / Isilon storage division
09:04:22 - r286114
(1 day 15 hours ago)
by ae
Affects:  /head/sys/net/if_stf.c
Ansify if_stf.c
09:03:56 - r286113
(1 day 15 hours ago)
by pho
Affects:  /user/pho/stress2/misc/mmap20.sh
/user/pho/stress2/misc/mmap3.sh
White space and comment updates.

Sponsored by:   EMC / Isilon storage division
09:02:28 - r286112
(1 day 15 hours ago)
by ae
Affects:  /head/sys/netinet/ip_ecn.h
Remove unneded #include "opt_inet.h".
08:45:35 - r286111
(1 day 15 hours ago)
by ed
Affects:  /head/sys/boot/forth/loader.conf
Document the existence of cloudabi_load and cloudabi64_load.
07:48:08 - r286110
(1 day 16 hours ago)
by jmg
Affects:  /head/sys/net/pfkeyv2.h
temporarily fix build..  This isn't the final fix, and testing is
still on going, but it has passed world for mips and powerpc...

I know this has an extra semicolon, but this is the patch that is
tested...

Looks like better fix is to use _Static_assert...
07:04:33 - r286109
(1 day 17 hours ago)
by erwin
Affects:  /vendor/bind9/9.9.7-P2
Tag the 9.9.7-P2 release

Sponsored by:   DK Hostmaster A/S
07:03:06 - r286108
(1 day 17 hours ago)
by erwin
Affects: 
(65 files)
/vendor/bind9/dist/*
Vendor import of BIND 9.9.7-P2

Sponsored by:   DK Hostmaster A/S
04:50:47 - r286107
(1 day 19 hours ago)
by np
Affects:  /head/sys/dev/cxgbe/t4_main.c
cxgbe(4): initialize debug_flags from the kernel environment.

MFC after:      3 days
04:12:51 - r286106
(1 day 19 hours ago)
by kib
Affects:  /head/sys/kern/vfs_vnops.c
vn_io_fault() handling of the LOR for i/o into the file-backed buffers
has observable overhead when the buffer pages are not resident or not
mapped.  The overhead comes at least from two factors, one is the
additional work needed to detect the situation, prepare and execute
the rollbacks.  Another is the consequence of the i/o splitting into
the batches of the held pages, causing filesystems see series of the
smaller i/o requests instead of the single large request.

Note that expected case of the resident i/o buffer does not expose
these issues.  Provide a prefaulting for the userspace i/o buffers,
disabled by default.  I am careful of not enabling prefaulting by
default for now, since it would be detrimental for the applications
which speculatively pass extra-large buffers of anonymous memory to
not deal with buffer sizing (if such apps exist).

Found and tested by:    bde, emaste
Sponsored by:   The FreeBSD Foundation
MFC after:      1 week
03:40:09 - r286105
(1 day 20 hours ago)
by kevlo
Affects: 
(2 files)
/stable/10/sys/netinet/ip.h
MFC r285669:

Since the IETF has redefined the meaning of the tos field to accommodate
a set of differentiated services, set IPTOS_PREC_* macros using
IPTOS_DSCP_* macro definitions.

While here, add IPTOS_DSCP_VA macro according to RFC 5865.

Differential Revision:  https://reviews.freebsd.org/D3119
Reviewed by:    gnn
03:35:27 - r286104
(1 day 20 hours ago)
by kevlo
Affects: 
(2 files)
/stable/10/sys/dev/e1000/e1000_regs.h
MFC r285623:

Fix typo in register definition.

Submitted by:   James Hung
Reviewed by:    sbruno
03:28:02 - r286103
(1 day 20 hours ago)
by jmg
Affects:  /head/share/man/man9/CTASSERT.9
The implementation note isn't true anymore..

Not that anyone reads it, but those that do, remind them that this
isn't usable in userland...  I can't wait till this doc is wrong..
01:12:31 - r286102
(1 day 22 hours ago)
by pfg
Affects:  /head/usr.bin/wall/ttymsg.c
Buffer overflow in wall(1).

This affected syslogd, wall and talkd.
Detected by FORTIFY_SOURCE GSoC (with clang).

Submitted by:   Oliver Pinter
Differential Revision:  https://reviews.freebsd.org/D3254
Reviewed by:    delphij, jmg
MFC after:      3 days
00:31:52 - r286101
(1 day 23 hours ago)
by jmg
Affects:  /head/sys/netipsec/xform_ah.c
/head/sys/netipsec/xform_esp.c
these are comparing authenticators and need to be constant time...
This could be a side channel attack...  Now that we have a function
for this, use it...

jmgurney/ipsecgcm:      24d704cc and 7f37a14
00:23:21 - r286100
(1 day 23 hours ago)
by jmg
Affects:  /head/sys/net/pfkeyv2.h
/head/sys/netipsec/xform_esp.c
Clean up this header file...

use CTASSERTs now that we have them...

Replace a draft w/ RFC that's over 10 years old.

Note that _AALG and _EALG do not need to match what the IKE daemons
think they should be..  This is part of the KABI...  I decided to
renumber AESCTR, but since we've never had working AESCTR mode, I'm
not really breaking anything..  and it shortens a loop by quite
a bit..

remove SKIPJACK IPsec support...  SKIPJACK never made it out of draft
(in 1999), only has 80bit key, NIST recommended it stop being used
after 2010, and setkey nor any of the IKE daemons I checked supported
it...

jmgurney/ipsecgcm: a357a33, c75808b, e008669, b27b6d6

Reviewed by:    gnn (earlier version)
00:21:40 - r286099
(1 day 23 hours ago)
by markj
Affects: 
(2 files)
/stable/10/usr.bin/netstat/main.c
MFC r285782:
Fix counter reads on platforms where sizeof(uint64_t) != sizeof(uint64_t *).

PR:     201700
00:00:59 - r286098
(2 days ago)
by gjb
Affects:  /releng/10.2/sys/conf/newvers.sh
Update releng/10.2 to RC2 in preparation for the next set of builds.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
2015 - 07 - 30   (3 days ago)
23:58:55 - r286097
(2 days ago)
by gjb
Affects:  /releng/10.2/release/doc/en_US.ISO8859-1/relnotes/article.xml
Document r286084, tar(1) sparse file fix.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
23:58:53 - r286096
(2 days ago)
by gjb
Affects:  /releng/10.2/release/doc/en_US.ISO8859-1/relnotes/article.xml
Document r285830, specifically that the default pkg(8) repository
was switched from 'latest' to 'quarterly' for 10.2-RELEASE.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
20:56:27 - r286095
(2 days 3 hours ago)
by eri
Affects:  /head/sys/netipsec/ipsec_output.c
Correct IPSec SA statistic keeping

The IPsec SA statistic keeping is used even for decision making on
expiry/rekeying SAs.
When there are multiple transformations being done the statistic keeping might
be wrong.

This mostly impacts multiple encapsulations on IPsec since the usual scenario
it is not noticed due to the code path not taken.

Differential Revision:  https://reviews.freebsd.org/D3239
Reviewed by:            ae, gnn
Approved by:            gnn(mentor)
19:52:43 - r286094
(2 days 4 hours ago)
by mjg
Affects: 
(4 files)
/head/sys/kern/*
/head/sys/sys/sysctl.h
Revert r285125 until rmlocks get fixed.

Right now there is a chance that sysctl unregister will cause reader to
block on the sx lock associated with sysctl rmlock, in which case kernels
with debug enabled will panic.
19:44:46 - r286093
(2 days 4 hours ago)
by rodrigc
Affects:  /head/usr.bin/ipcrm/ipcrm.c
Use correct number of arguments to semctl() for IPC_RMID.

PR:                     118292
Submitted by:           araujo
Differential Revision:  D2669
19:34:24 - r286092
(2 days 4 hours ago)
by gjb
Affects:  /head/cddl/contrib/opensolaris/cmd/zfs/zfs.8
Fix a rendering issue in the zfs(8) manual.

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation
19:24:49 - r286091
(2 days 4 hours ago)
by hiren
Affects:  /head/sys/netinet/tcp_var.h
Update snd_una description to make it more readable.

Differential Revision:  https://reviews.freebsd.org/D3179
Reviewed by:            gnn
Sponsored by:           Limelight Networks
19:08:23 - r286090
(2 days 5 hours ago)
by brueffer
Affects:  /head/share/man/man4/Makefile
/head/share/man/man4/pms.4
The kernel option and module are actually called pmspcv.

MFC after:      3 days
19:04:14 - r286089
(2 days 5 hours ago)
by gonzo
Affects:  /head/sys/conf/files
/head/sys/dev/gpio/gpiobacklight.c
Add GPIO backlight driver compatible with Linux FDT bindings.
Brightness is controlled through sysctl dev.gpiobacklight.X.brightness:
  - any value greater than 0: backlight is on
  - any value less than or equal to  0: backlight is off

FDT bindings docs in Linux tree:
    Documentation/devicetree/bindings/video/backlight/gpio-backlight.txt
18:59:01 - r286088
(2 days 5 hours ago)
by rodrigc
Affects: 
(4 files)
/head/sys/sys/*
/head/usr.bin/ipcrm/ipcrm.c
Get function prototypes for msg, shm, sem functions
from header files.

Differential Revision: D2669
18:28:37 - r286087
(2 days 5 hours ago)
by markj
Affects:  /head/sys/ofed/drivers/infiniband/core/mad.c
ib mad: fix an incorrect use of list_for_each_entry

In tf_dequeue(), if we reach the end of the list without finding a
non-cancelled element, "tmp" will be a pointer into the list head, so the
tmp->canceled check is bogus. Use a flag instead.

Submitted by:   Tao Liu <Tao.Liu@isilon.com>
Reviewed by:    hselasky
MFC after:      1 week
Sponsored by:   EMC / Isilon Storage Division
Differential Revision: https://reviews.freebsd.org/D3244
18:28:34 - r286086
(2 days 5 hours ago)
by kib
Affects:  /head/sys/vm/vm_fault.c
/head/sys/vm/vm_map.c
/head/sys/vm/vm_map.h
Do not pretend that vm_fault(9) supports unwiring the address.  Rename
the VM_FAULT_CHANGE_WIRING flag to VM_FAULT_WIRE.  Assert that the
flag is only passed when faulting on the wired map entry.  Remove the
vm_page_unwire() call, which should be never reachable.

Since VM_FAULT_WIRE flag implies wired map entry, the TRYPAGER() macro
is reduced to the testing of the fs.object having a default pager.
Inline the check.

Suggested and reviewed by:      alc
Tested by:      pho (previous version)
MFC after:      1 week
18:08:08 - r286085
(2 days 6 hours ago)
by rodrigc
Affects:  /head/usr.bin/ipcrm/ipcrm.c
Make some variables and functions static.
18:00:37 - r286084
(2 days 6 hours ago)
by bdrewery
Affects: 
(5 files)
/releng/10.2/contrib/libarchive/libarchive/archive_read_support_format_tar.c
/releng/10.2/contrib/libarchive/libarchive/test/*
/releng/10.2/lib/libarchive/test/Makefile
MFS r286082:

  MFC r285972:

    MFV r285970:

      Apply upstream changeset bf4f6ec64e:

      Fix issue 356: properly skip a sparse file entry in a tar file.

PR:             201506
Relnotes:       yes
Approved by:    re (gjb)
17:06:28 - r286083
(2 days 7 hours ago)
by gjb
Affects: 
(2 files)
/releng/10.2/sys/netpfil/pf/pf_norm.c
MFS 286079:
MFC r285999 (kp):
  pf: Always initialise pf_fragment.fr_flags

  When we allocate the struct pf_fragment in pf_fillup_fragment() we
  forgot to initialise the fr_flags field. As a result we sometimes
  mistakenly thought the fragment to not be a buffered fragment.
  This resulted in panics because we'd end up freeing the pf_fragment
  but not removing it from V_pf_fragqueue (believing it to be part of
  V_pf_cachequeue).  The next time we iterated V_pf_fragqueue we'd use
  a freed object and panic.

  While here also fix a pf_fragment use after free in pf_normalize_ip().
  pf_reassemble() frees the pf_fragment, so we can't use it any more.

Approved by:    re (glebius)
Sponsored by:   The FreeBSD Foundation
17:04:15 - r286082
(2 days 7 hours ago)
by bdrewery
Affects: 
(5 files)
/stable/10/contrib/libarchive/libarchive/archive_read_support_format_tar.c
/stable/10/contrib/libarchive/libarchive/test/*
/stable/10/lib/libarchive/test/Makefile
MFC r285972:

  MFV r285970:

    Apply upstream changeset bf4f6ec64e:

    Fix issue 356: properly skip a sparse file entry in a tar file.

PR:             201506
Relnotes:       yes
17:02:23 - r286081
(2 days 7 hours ago)
by allanjude
Affects:  /head/share/misc/committers-src.dot
Updat the committers graph

Approved by:    marcel (mentor)
Differential Revision:  https://reviews.freebsd.org/D3251
16:17:44 - r286080
(2 days 7 hours ago)
by andrew
Affects:  /head/sys/arm64/arm64/pmap.c
Add enough of pmap_page_set_memattr to run gstat. It still needs to split
the DMAP 1G pages so we set the attributes only on the specified page.

Obtained from:  ABT Systems Ltd
Sponsored by:   The FreeBSD Foundation
16:06:13 - r286079
(2 days 8 hours ago)
by gjb
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf_norm.c
MFC r285999 (kp):
  pf: Always initialise pf_fragment.fr_flags

  When we allocate the struct pf_fragment in pf_fillup_fragment() we
  forgot to initialise the fr_flags field. As a result we sometimes
  mistakenly thought the fragment to not be a buffered fragment.
  This resulted in panics because we'd end up freeing the pf_fragment
  but not removing it from V_pf_fragqueue (believing it to be part of
  V_pf_cachequeue).  The next time we iterated V_pf_fragqueue we'd use
  a freed object and panic.

  While here also fix a pf_fragment use after free in pf_normalize_ip().
  pf_reassemble() frees the pf_fragment, so we can't use it any more.

X-MFS-To:       releng/10.2
Sponsored by:   The FreeBSD Foundation
15:47:53 - r286078
(2 days 8 hours ago)
by kib
Affects:  /head/sys/amd64/include/atomic.h
/head/sys/i386/include/atomic.h
Improve comments.

Submitted by:   bde
MFC after:      2 weeks
15:43:26 - r286077
(2 days 8 hours ago)
by royger
Affects:  /head/sys/kern/vfs_bio.c
vfs: fill fallout from r286076

This right operator is >= not =>.

Reported by: cem
15:28:06 - r286076
(2 days 8 hours ago)
by royger
Affects:  /head/sys/kern/vfs_bio.c
vfs: fix off-by-one error in vfs_buf_check_mapped

The check added in r285872 can trigger for valid buffers if the buffer space
used happens to be just after unmapped_buf in KVA space.

Discussed with: kib
Sponsored by: Citrix Systems R&D
14:44:13 - r286075
(2 days 9 hours ago)
by bapt
Affects:  /svnadmin/conf/access
/svnadmin/conf/mentors
Add Allan Jude as a new src committer. marcel@ and I will be mentoring him

Approved by:    core
14:31:09 - r286074
(2 days 9 hours ago)
by pfg
Affects: 
(7 files)
/head/contrib/gcc/*
GCC: Add a new option "-fstack-protector-strong"

This includes additional functions to be protected: those that
have local array definitions, or have references to local frame
addresses. This is a new option in GCC-4.9 that was relicensed
by Han Shen from Google under GPLv2 for OpenBSD.

Obtained from:  OpenBSD (2014-01-14)
MFC after:      2 weeks
14:20:36 - r286073
(2 days 9 hours ago)
by emaste
Affects: 
(8 files)
/head/sys/arm64/arm64/*
Add ARM64TODO markers to unimplemented functionality

Reviewed by:    andrew
Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D2389
13:59:38 - r286072
(2 days 10 hours ago)
by zbb
Affects:  /head/sys/arm64/arm64/trap.c
Enable IRQ during syscalls on ARM64

FreeBSD provides a feature called Adaptive Mutexes, which allows
a thread to spin for a while when the mutex is taken instead of
immediately going to sleep. This causes issues when called from
syscall handler if interrupts are masked. If every other core
also attempts to access the same mutex there is a chance that
all of them are spinning on the same lock at the same time.
If interrupts are disabled, no kernel preemtion can occur and
the system becomes unresponsive.

This patch enables interrupts when syscall is being executed
and masks them as soon as it is completed.

Reviewed by:   andrew
Obtained from: Semihalf
Sponsored by:  The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3246
13:45:34 - r286071
(2 days 10 hours ago)
by zbb
Affects:  /head/sys/arm/annapurna/alpine/hal
Remove obsolete vendor code from Alpine platform support

This is a clean-up patch from a serie delivering support for
Annapurna Labs Alpine PoC.
The HAL files have already been added to sys/contrib/alpine-hal
so there is no need for them in the platform directory.
This patch removes obsolete files.

Reviewed by:    andrew
Obtained from:  Semihalf
Sponsored by:   Annapurna Labs
Differential Revision: https://reviews.freebsd.org/D3248
12:41:54 - r286070
(2 days 11 hours ago)
by emaste
Affects:  /head/contrib/elftoolchain/ar
/head/contrib/elftoolchain/elfdump
Add ELF Tool Chain's ar(1) and elfdump(1) to contrib

ELF Tool Chain built on FreeBSD's ar and elfdump, but has a number of
improvements and enhancements. Bring them into contrib in order to start
integrating into our build.
10:26:43 - r286069
(2 days 13 hours ago)
by ae
Affects:  /head/sys/modules/Makefile
/head/sys/modules/if_stf/Makefile
/head/sys/net/if_stf.c
Build if_stf(4) module only when both INET and INET6 support are enabled.
10:09:31 - r286068
(2 days 14 hours ago)
by delphij
Affects:  /releng/8.4/UPDATING
/releng/8.4/crypto/openssh/auth2-chall.c
/releng/8.4/sys/conf/newvers.sh
Fix a regression introduced with previous SA-15:16.openssl
changeset.  The problem is specific to FreeBSD 8.x.

Submitted by:   jkim
Pointy hat to:  delphij
Approved by:    so
10:09:07 - r286067
(2 days 14 hours ago)
by delphij
Affects:  /stable/8/crypto/openssh/auth2-chall.c
Fix a regression introduced with previous OpenSSH by explicitly
initializing devices_done.

Submitted by:   jkim
06:14:47 - r286066
(2 days 17 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/pw.h
/head/usr.sbin/pw/strtounum.c
Improve strtounum

Fix many style bugs
Better variable naming
Use C99 'restrict' were apropriate
Fix potential errno race

Submitted by:   bde
05:13:12 - r286065
(2 days 18 hours ago)
by ngie
Affects:  /user/ngie/more-tests/usr.sbin/syslogd/syslogd.c
Revert local change accidentally committed with r285905
04:53:53 - r286064
(2 days 19 hours ago)
by jamie
Affects: 
(3 files)
/stable/10/usr.sbin/jexec/jexec.8
/stable/10/usr.sbin/jexec/jexec.c
MFC r285420:

  Run a shell in the jail when no command is specified.
  Add a new flag, -l, for a clean environment, same as jail(8) exec.clean.
  Change the GET_USER_INFO macro into a function.

PR:             201300
Submitted by:   Willem Jan Withagen
04:01:00 - r286063
(2 days 20 hours ago)
by cperciva
Affects:  /head/release/tools/ec2.conf
Disable blkif indirect segment I/Os in EC2 by default due to performance
issues on some EC2 instance types.  Users may want to experiment with
removing this from loader.conf and measuring the performance impact on
the EC2 instances they are using.
03:50:01 - r286062
(2 days 20 hours ago)
by cperciva
Affects:  /head/sys/dev/xen/blkfront/blkfront.c
/head/sys/dev/xen/blkfront/block.h
/head/sys/xen/interface/io/blkif.h
Add support for Xen blkif indirect segment I/Os.  This makes it possible for
the blkfront driver to perform I/Os of up to 2 MB, subject to support from
the blkback to which it is connected and the initiation of such large I/Os
by the rest of the kernel.  In practice, the I/O size is increased from 40 kB
to 128 kB.

The changes to xen/interface/io/blkif.h consist merely of merging updates
from the upstream Xen repository.

In dev/xen/blkfront/block.h we add some convenience macros and structure
fields used for indirect-page I/Os: The device records its negotiated limit
on the number of indirect pages used, while each I/O command structure gains
permanently allocated page(s) for indirect page references and the Xen grant
references for those pages.

In dev/xen/blkfront/blkfront.c we now check in xbd_queue_cb whether a request
is small enough to handle without an indirection page, and either follow the
previous behaviour or use new code for issuing an indirect segment I/O.  In
xbd_connect we read the size of indirect segment I/Os supported by the backend
and select the maximum size we will use; then allocate the pages and Xen grant
references for each I/O command structure.  In xbd_free those grants and pages
are released.

A new loader tunable, hw.xbd.xbd_enable_indirect, can be set to 0 in order to
disable this functionality; it works by pretending that the backend does not
support this feature.  Some backends exhibit a loss of performance with large
I/Os, so users may wish to test with and without this functionality enabled.

Reviewed by:    royger
MFC after:      3 days
Relnotes:       yes
03:06:11 - r286061
(2 days 21 hours ago)
by marius
Affects: 
(3 files)
/releng/10.2/sys/dev/uart/uart_bus.h
/releng/10.2/sys/dev/uart/uart_core.c
MFC: r285843 (r286059 in stable/10)

- Since r253161, uart_intr() abuses FILTER_SCHEDULE_THREAD for signaling
  uart_bus_attach() during its test that 20 iterations weren't sufficient
  for clearing all pending interrupts, assuming this means that hardware
  is broken and doesn't deassert interrupts. However, under pressure, 20
  iterations also can be insufficient for clearing all pending interrupts,
  leading to a panic as intr_event_handle() tries to schedule an interrupt
  handler not registered. Solve this by introducing a flag that is set in
  test mode and otherwise restores pre-r253161 behavior of uart_intr(). The
  approach of additionally registering uart_intr() as handler as suggested
  in PR 194979 is not taken as that in turn would abuse special pccard and
  pccbb handling code of intr_event_handle(). [1]
- Const'ify uart_driver_name.
- Fix some minor style bugs.

PR:             194979 [1]
Reviewed by:    marcel (earlier version)
Approved by:    re (gjb)
02:45:35 - r286060
(2 days 21 hours ago)
by marius
Affects: 
(6 files)
/releng/10.2/sys/kern/subr_witness.c
/releng/10.2/sys/sparc64/include/smp.h
/releng/10.2/sys/sparc64/sparc64/*
MFC: r285839 (r286055 in stable/10)

o Revert the other functional half of r239864, i. e. the merge of r134227
  from x86 to use smp_ipi_mtx spin lock not only for smp_rendezvous_cpus()
  but also for the MD cache invalidation, TLB demapping and remote register
  reading IPIs due to the following reasons:
  - The cross-IPI SMP deadlock x86 otherwise is subject to can't happen on
    sparc64. That's because on sparc64, spin locks don't disable interrupts
    completely but only raise the processor interrupt level to PIL_TICK. This
    means that IPIs still get delivered and direct dispatch IPIs such as the
    cache invalidation etc. IPIs in question are still executed.
  - In smp_rendezvous_cpus(), smp_ipi_mtx is held not only while sending an
    IPI_RENDEZVOUS, but until all CPUs have processed smp_rendezvous_action().
    Consequently, smp_ipi_mtx may be locked for an extended amount of time as
    queued IPIs (as opposed to the direct ones) such as IPI_RENDEZVOUS are
    scheduled via a soft interrupt. Moreover, given that this soft interrupt
    is only delivered at PIL_RENDEZVOUS, processing of smp_rendezvous_action()
    on a target may be interrupted by f. e. a tick interrupt at PIL_TICK, in
    turn leading to the target in question trying to send an IPI by itself
    while IPI_RENDEZVOUS isn't fully handled, yet, and, thus, resulting in a
    deadlock.
o As mentioned in the commit message of r245850, on least some sun4u platforms
  concurrent sending of IPIs by different CPUs is fatal. Therefore, hold the
  reintroduced MD ipi_mtx also while delivering cross-traps via MI helpers,
  i. e. ipi_{all_but_self,cpu,selected}().
o Akin to x86, let the last CPU to process cpu_mp_bootstrap() set smp_started
  instead of the BSP in cpu_mp_unleash(). This ensures that all APs actually
  are started, when smp_started is no longer 0.
o In all MD and MI IPI helpers, check for smp_started == 1 rather than for
  smp_cpus > 1 or nothing at all. This avoids races during boot causing IPIs
  trying to be delivered to APs that in fact aren't up and running, yet.
  While at it, move setting of the cpu_ipi_{selected,single}() pointers to
  the appropriate delivery functions from mp_init() to cpu_mp_start() where
  it's better suited and allows to get rid of the global isjbus variable.
o Given that now concurrent IPI delivery no longer is possible, also nuke
  the delays before completely disabling interrupts again in the CPU-specific
  cross-trap delivery functions, previously giving other CPUs a window for
  sending IPIs on their part. Actually, we now should be able to entirely get
  rid of completely disabling interrupts in these functions. Such a change
  needs more testing, though.
o In {s,}tick_get_timecount_mp(), make the {s,}tick variable static. While not
  necessary for correctness, this avoids page faults when accessing the stack
  of a foreign CPU as {s,}tick now is locked into the TLBs as part of static
  kernel data. Hence, {s,}tick_get_timecount_mp() always execute as fast as
  possible, avoiding jitter.

PR:             201245
Approved by:    re (gjb)
02:23:09 - r286059
(2 days 21 hours ago)
by marius
Affects: 
(3 files)
/stable/10/sys/dev/uart/uart_bus.h
/stable/10/sys/dev/uart/uart_core.c
MFC: r285843

- Since r253161, uart_intr() abuses FILTER_SCHEDULE_THREAD for signaling
  uart_bus_attach() during its test that 20 iterations weren't sufficient
  for clearing all pending interrupts, assuming this means that hardware
  is broken and doesn't deassert interrupts. However, under pressure, 20
  iterations also can be insufficient for clearing all pending interrupts,
  leading to a panic as intr_event_handle() tries to schedule an interrupt
  handler not registered. Solve this by introducing a flag that is set in
  test mode and otherwise restores pre-r253161 behavior of uart_intr(). The
  approach of additionally registering uart_intr() as handler as suggested
  in PR 194979 is not taken as that in turn would abuse special pccard and
  pccbb handling code of intr_event_handle(). [1]
- Const'ify uart_driver_name.
- Fix some minor style bugs.

PR:             194979 [1]
Reviewed by:    marcel (earlier version)
02:22:38 - r286058
(2 days 21 hours ago)
by whu
Affects: 
(2 files)
/releng/10.2/sys/dev/hyperv/netvsc/hv_netvsc_drv_freebsd.c
MFC r285928 Do not enable UDP checksum offloading when running on the
Hyper-V on Windows Server 2012 and earlier hosts.

Submitted by: whu
Reviewed by: royger
Approved by: re (gjb)
Sponsored by: Microsoft OSTC
Differential Revision: https://reviews.freebsd.org/D3102
02:09:03 - r286057
(2 days 22 hours ago)
by loos
Affects:  /head/sys/net/radix.c
/head/sys/net/radix.h
/head/sys/net/route.c
Follow r256586 and rename the kernel version of the Free() macro to
R_Free().  This matches the other macros and reduces the chances to clash
with other headers.

This also fixes the build of radix.c outside of the kernel environment.

Reviewed by:    glebius
02:06:34 - r286056
(2 days 22 hours ago)
by marius
Affects: 
(6 files)
/stable/9/sys/kern/subr_witness.c
/stable/9/sys/sparc64/include/smp.h
/stable/9/sys/sparc64/sparc64/*
MFC: r285839

o Revert the other functional half of r239864 (MFCed to stable/9 in r241681),
  i. e. the merge of r134227 from x86 to use smp_ipi_mtx spin lock not only
  for smp_rendezvous_cpus() but also for the MD cache invalidation, TLB
  demapping and remote register reading IPIs due to the following reasons:
  - The cross-IPI SMP deadlock x86 otherwise is subject to can't happen on
    sparc64. That's because on sparc64, spin locks don't disable interrupts
    completely but only raise the processor interrupt level to PIL_TICK. This
    means that IPIs still get delivered and direct dispatch IPIs such as the
    cache invalidation etc. IPIs in question are still executed.
  - In smp_rendezvous_cpus(), smp_ipi_mtx is held not only while sending an
    IPI_RENDEZVOUS, but until all CPUs have processed smp_rendezvous_action().
    Consequently, smp_ipi_mtx may be locked for an extended amount of time as
    queued IPIs (as opposed to the direct ones) such as IPI_RENDEZVOUS are
    scheduled via a soft interrupt. Moreover, given that this soft interrupt
    is only delivered at PIL_RENDEZVOUS, processing of smp_rendezvous_action()
    on a target may be interrupted by f. e. a tick interrupt at PIL_TICK, in
    turn leading to the target in question trying to send an IPI by itself
    while IPI_RENDEZVOUS isn't fully handled, yet, and, thus, resulting in a
    deadlock.
o As mentioned in the commit message of r245850, on least some sun4u platforms
  concurrent sending of IPIs by different CPUs is fatal. Therefore, hold the
  reintroduced MD ipi_mtx also while delivering cross-traps via MI helpers,
  i. e. ipi_{all_but_self,cpu,selected}().
o Akin to x86, let the last CPU to process cpu_mp_bootstrap() set smp_started
  instead of the BSP in cpu_mp_unleash(). This ensures that all APs actually
  are started, when smp_started is no longer 0.
o In all MD and MI IPI helpers, check for smp_started == 1 rather than for
  smp_cpus > 1 or nothing at all. This avoids races during boot causing IPIs
  trying to be delivered to APs that in fact aren't up and running, yet.
  While at it, move setting of the cpu_ipi_{selected,single}() pointers to
  the appropriate delivery functions from mp_init() to cpu_mp_start() where
  it's better suited and allows to get rid of the global isjbus variable.
o Given that now concurrent IPI delivery no longer is possible, also nuke
  the delays before completely disabling interrupts again in the CPU-specific
  cross-trap delivery functions, previously giving other CPUs a window for
  sending IPIs on their part. Actually, we now should be able to entirely get
  rid of completely disabling interrupts in these functions. Such a change
  needs more testing, though.
o In {s,}tick_get_timecount_mp(), make the {s,}tick variable static. While not
  necessary for correctness, this avoids page faults when accessing the stack
  of a foreign CPU as {s,}tick now is locked into the TLBs as part of static
  kernel data. Hence, {s,}tick_get_timecount_mp() always execute as fast as
  possible, avoiding jitter.

PR:             201245
02:06:29 - r286055
(2 days 22 hours ago)
by marius
Affects: 
(6 files)
/stable/10/sys/kern/subr_witness.c
/stable/10/sys/sparc64/include/smp.h
/stable/10/sys/sparc64/sparc64/*
MFC: r285839

o Revert the other functional half of r239864, i. e. the merge of r134227
  from x86 to use smp_ipi_mtx spin lock not only for smp_rendezvous_cpus()
  but also for the MD cache invalidation, TLB demapping and remote register
  reading IPIs due to the following reasons:
  - The cross-IPI SMP deadlock x86 otherwise is subject to can't happen on
    sparc64. That's because on sparc64, spin locks don't disable interrupts
    completely but only raise the processor interrupt level to PIL_TICK. This
    means that IPIs still get delivered and direct dispatch IPIs such as the
    cache invalidation etc. IPIs in question are still executed.
  - In smp_rendezvous_cpus(), smp_ipi_mtx is held not only while sending an
    IPI_RENDEZVOUS, but until all CPUs have processed smp_rendezvous_action().
    Consequently, smp_ipi_mtx may be locked for an extended amount of time as
    queued IPIs (as opposed to the direct ones) such as IPI_RENDEZVOUS are
    scheduled via a soft interrupt. Moreover, given that this soft interrupt
    is only delivered at PIL_RENDEZVOUS, processing of smp_rendezvous_action()
    on a target may be interrupted by f. e. a tick interrupt at PIL_TICK, in
    turn leading to the target in question trying to send an IPI by itself
    while IPI_RENDEZVOUS isn't fully handled, yet, and, thus, resulting in a
    deadlock.
o As mentioned in the commit message of r245850, on least some sun4u platforms
  concurrent sending of IPIs by different CPUs is fatal. Therefore, hold the
  reintroduced MD ipi_mtx also while delivering cross-traps via MI helpers,
  i. e. ipi_{all_but_self,cpu,selected}().
o Akin to x86, let the last CPU to process cpu_mp_bootstrap() set smp_started
  instead of the BSP in cpu_mp_unleash(). This ensures that all APs actually
  are started, when smp_started is no longer 0.
o In all MD and MI IPI helpers, check for smp_started == 1 rather than for
  smp_cpus > 1 or nothing at all. This avoids races during boot causing IPIs
  trying to be delivered to APs that in fact aren't up and running, yet.
  While at it, move setting of the cpu_ipi_{selected,single}() pointers to
  the appropriate delivery functions from mp_init() to cpu_mp_start() where
  it's better suited and allows to get rid of the global isjbus variable.
o Given that now concurrent IPI delivery no longer is possible, also nuke
  the delays before completely disabling interrupts again in the CPU-specific
  cross-trap delivery functions, previously giving other CPUs a window for
  sending IPIs on their part. Actually, we now should be able to entirely get
  rid of completely disabling interrupts in these functions. Such a change
  needs more testing, though.
o In {s,}tick_get_timecount_mp(), make the {s,}tick variable static. While not
  necessary for correctness, this avoids page faults when accessing the stack
  of a foreign CPU as {s,}tick now is locked into the TLBs as part of static
  kernel data. Hence, {s,}tick_get_timecount_mp() always execute as fast as
  possible, avoiding jitter.

PR:             201245
00:28:32 - r286054
(2 days 23 hours ago)
by marius
Affects: 
(3 files)
/stable/9/sys/dev/mii/mii.h
MFC: r284447, r284552

Merge from NetBSD:
o rev. 1.10: Nuke trailing whitespace.
o rev. 1.15: Fix typo in comment.
o rev. 1.16: Add the following registers from IEEE 802.3-2009 Clause 22:
 - PSE control register (0x0b)
 - PSE status register (0x0c)
 - MMD access control register (0x0d)
 - MMD access address data register (0x0e)
o rev. 1.17 (comments only): The bit location of link ability is different
  between 1000Base-X and others (see Annex 28B.2 and 28D).
o rev. 1.18: Nuke dupe word.

Obtained from:  NetBSD
Sponsored by:   genua mbh
00:28:27 - r286053
(2 days 23 hours ago)
by marius
Affects: 
(2 files)
/stable/10/sys/dev/mii/mii.h
MFC: r284447, r284552

Merge from NetBSD:
o rev. 1.10: Nuke trailing whitespace.
o rev. 1.15: Fix typo in comment.
o rev. 1.16: Add the following registers from IEEE 802.3-2009 Clause 22:
 - PSE control register (0x0b)
 - PSE status register (0x0c)
 - MMD access control register (0x0d)
 - MMD access address data register (0x0e)
o rev. 1.17 (comments only): The bit location of link ability is different
  between 1000Base-X and others (see Annex 28B.2 and 28D).
o rev. 1.18: Nuke dupe word.

Obtained from:  NetBSD
Sponsored by:   genua mbh
00:24:21 - r286052
(2 days 23 hours ago)
by marius
Affects: 
(2 files)
/stable/10/sys/arm/allwinner/a20/a20_mp.c
MFC: r281752

Make a comment reflect reality.
00:13:20 - r286051
(2 days 23 hours ago)
by kib
Affects:  /head/sys/i386/i386/vm_machdep.c
/head/sys/i386/include/atomic.h
Use private cache line for the locked nop in *mb() on i386.

Suggested by:   alc
Reviewed by:    alc, bde
Tested by:      pho
Sponsored by:   The FreeBSD Foundation
MFC after:      2 weeks
2015 - 07 - 29   (4 days ago)
23:59:17 - r286050
(3 days ago)
by kib
Affects:  /head/sys/i386/include/atomic.h
MFamd64 r285934: Remove store/load (= full) barrier from the i386
atomic_load_acq_*().

Noted by:       alc (long time ago)
Reviewed by:    alc, bde
Tested by:      pho
Sponsored by:   The FreeBSD Foundation
MFC after:      2 weeks
23:37:15 - r286049
(3 days ago)
by jmg
Affects:  /head/sys/crypto/aesni/aesni.h
/head/sys/crypto/aesni/aesni_ghash.c
const'ify an arg that we don't update...
23:34:38 - r286048
(3 days ago)
by gjb
Affects:  /releng/10.2/release/Makefile.vm
Set the default VHD file format to the fixed-size image for the
10.2-RELEASE, as the issue preventing the dynamic-size image from
booting is not yet resolved.

This is a direct commit to releng/10.2.

Approved by:    re (kib)
Sponsored by:   The FreeBSD Foundation
23:26:14 - r286047
(3 days ago)
by bapt
Affects:  /head/usr.sbin/pw/rm_r.c
Cleanup includes
23:06:30 - r286046
(3 days 1 hour ago)
by rmacklem
Affects:  /head/sys/fs/nfsserver/nfs_nfsdstate.c
This patch fixes a problem where, if the NFSv4 server has a previous
unconfirmed clientid structure for the same client on the last hash list,
this old entry would not be removed/deleted. I do not think this bug would have
caused serious problems, since the new entry would have been before the old one
on the list. This old entry would have eventually been scavenged/removed.
Detected while reading the code looking for another bug.

MFC after:      3 days
22:51:54 - r286045
(3 days 1 hour ago)
by bapt
Affects:  /head/usr.sbin/pw/strtounum.c
Actually set the proper license

Reported by:    trasz
21:41:15 - r286044
(3 days 2 hours ago)
by emaste
Affects:  /head/tools/build/mk/OptionalObsoleteFiles.inc
MK_ELFCOPY_AS_OBJCOPY should be a variable

PR:             201978
Submitted by:   O. Hartmann
Differential Revision:  https://reviews.freebsd.org/D2887
21:29:50 - r286043
(3 days 2 hours ago)
by jimharris
Affects:  /head/sys/dev/nvme/nvme.c
nvme: do not notify a consumer about failures that occur during initialization

MFC after:      3 days
Sponsored by:   Intel
21:15:50 - r286042
(3 days 2 hours ago)
by imp
Affects:  /head/UPDATING
Clarify historical practice of not removing old entries. Add entry for
stable/10 branch that was forgotten when it was created. Update end
date to be correct.
20:50:48 - r286041
(3 days 3 hours ago)
by sbruno
Affects:  /head/sys/dev/mii/brgphy.c
/head/sys/dev/mii/miidevs
Add support for BCM5466 PHY

Differential Revision:  D3232
Submitted by:   kevin.bowling@kev009.com
20:47:27 - r286040
(3 days 3 hours ago)
by sbruno
Affects:  /head/sys/arm/arm/pmap-v6-new.c
/head/sys/i386/i386/pmap.c
/head/sys/mips/mips/pmap.c
Remove dead functions pmap_pvdump and pads.

Differential Revision:  D3206
Submitted by:   kevin.bowling@kev009.com
Reviewed by:    alc
20:17:29 - r286039
(3 days 3 hours ago)
by gjb
Affects:  /releng/10.2/release/doc/en_US.ISO8859-1/hardware/article.xml
Add ixl(4) to the hardware notes.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
20:17:27 - r286038
(3 days 3 hours ago)
by gjb
Affects:  /releng/10.2/release/doc/en_US.ISO8859-1/hardware/article.xml
Update copyright after r286023.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
20:10:36 - r286037
(3 days 4 hours ago)
by eri
Affects:  /head/sys/netinet/ip_output.c
Avoid double reference decrement when firewalls force relooping of packets

When firewalls force a reloop of packets and the caller supplied a route the
reference to the route might be reduced twice creating issues.
This is especially the scenario when a packet is looped because of operation in
the firewall but the new route lookup gives a down route.

Differential Revision:  https://reviews.freebsd.org/D3037
Reviewed by:    gnn
Approved by:    gnn(mentor)
20:02:20 - r286036
(3 days 4 hours ago)
by emaste
Affects:  /head/tools/build/options/WITHOUT_BINUTILS
Remove mention of non-existent gconv tool

I believe this is a typo of gcov, but gcov is not controlled by
WITHOUT_BINUTILS anyhow.

Sponsored by:   The FreeBSD Foundation
19:37:32 - r286035
(3 days 4 hours ago)
by dim
Affects: 
(7 files)
/stable/9/contrib/llvm/include/llvm/CodeGen/SelectionDAGISel.h
/stable/9/contrib/llvm/lib/*
/stable/9/contrib/llvm/patches/patch-r286033-llvm-r219009-x86-codegen-crash.diff
Merge r286033 from stable/10:

Reapply r286007, modified to compile with pre-C++11 compilers:

Pull in r219009 from upstream llvm trunk (by Adam Nemet):

  [ISel] Keep matching state consistent when folding during X86 address match

  In the X86 backend, matching an address is initiated by the 'addr' complex
  pattern and its friends.  During this process we may reassociate and-of-shift
  into shift-of-and (FoldMaskedShiftToScaledMask) to allow folding of the
  shift into the scale of the address.

  However as demonstrated by the testcase, this can trigger CSE of not only the
  shift and the AND which the code is prepared for but also the underlying load
  node.  In the testcase this node is sitting in the RecordedNode and MatchScope
  data structures of the matcher and becomes a deleted node upon CSE.  Returning
  from the complex pattern function, we try to access it again hitting an assert
  because the node is no longer a load even though this was checked before.

  Now obviously changing the DAG this late is bending the rules but I think it
  makes sense somewhat.  Outside of addresses we prefer and-of-shift because it
  may lead to smaller immediates (FoldMaskAndShiftToScale is an even better
  example because it create a non-canonical node).  We currently don't recognize
  addresses during DAGCombiner where arguably this canonicalization should be
  performed.  On the other hand, having this in the matcher allows us to cover
  all the cases where an address can be used in an instruction.

  I've also talked a little bit to Dan Gohman on llvm-dev who added the RAUW for
  the new shift node in FoldMaskedShiftToScaledMask.  This RAUW is responsible
  for initiating the recursive CSE on users
  (http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-September/076903.html) but it
  is not strictly necessary since the shift is hooked into the visited user.  Of
  course it's safer to keep the DAG consistent at all times (e.g. for accurate
  number of uses, etc.).

  So rather than changing the fundamentals, I've decided to continue along the
  previous patches and detect the CSE.  This patch installs a very targeted
  DAGUpdateListener for the duration of a complex-pattern match and updates the
  matching state accordingly.  (Previous patches used HandleSDNode to detect the
  CSE but that's not practical here).  The listener is only installed on X86.

  I tested that there is no measurable overhead due to this while running
  through the spec2k BC files with llc.  The only thing we pay for is the
  creation of the listener.  The callback never ever triggers in spec2k since
  this is a corner case.

  Fixes rdar://problem/18206171

This fixes a possible crash in x86 code generation when compiling recent
llvm/clang trunk sources.

Direct commit to stable/10, since head already has llvm/clang 3.6.1,
which includes this fix.

Reported by:    jonathan, theraven
Upstream PR:    https://llvm.org/bugs/show_bug.cgi?id=24249

Merge r286034 from stable/10:

Add updated llvm patch corresponding to r286033.
19:27:57 - r286034
(3 days 4 hours ago)
by dim
Affects:  /stable/10/contrib/llvm/patches/patch-r286033-llvm-r219009-x86-codegen-crash.diff
Add updated llvm patch corresponding to r286033.
19:25:28 - r286033
(3 days 4 hours ago)
by dim
Affects:  /stable/10/contrib/llvm/include/llvm/CodeGen/SelectionDAGISel.h
/stable/10/contrib/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
/stable/10/contrib/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
Reapply r286007, modified to compile with pre-C++11 compilers:

Pull in r219009 from upstream llvm trunk (by Adam Nemet):

  [ISel] Keep matching state consistent when folding during X86 address match

  In the X86 backend, matching an address is initiated by the 'addr' complex
  pattern and its friends.  During this process we may reassociate and-of-shift
  into shift-of-and (FoldMaskedShiftToScaledMask) to allow folding of the
  shift into the scale of the address.

  However as demonstrated by the testcase, this can trigger CSE of not only the
  shift and the AND which the code is prepared for but also the underlying load
  node.  In the testcase this node is sitting in the RecordedNode and MatchScope
  data structures of the matcher and becomes a deleted node upon CSE.  Returning
  from the complex pattern function, we try to access it again hitting an assert
  because the node is no longer a load even though this was checked before.

  Now obviously changing the DAG this late is bending the rules but I think it
  makes sense somewhat.  Outside of addresses we prefer and-of-shift because it
  may lead to smaller immediates (FoldMaskAndShiftToScale is an even better
  example because it create a non-canonical node).  We currently don't recognize
  addresses during DAGCombiner where arguably this canonicalization should be
  performed.  On the other hand, having this in the matcher allows us to cover
  all the cases where an address can be used in an instruction.

  I've also talked a little bit to Dan Gohman on llvm-dev who added the RAUW for
  the new shift node in FoldMaskedShiftToScaledMask.  This RAUW is responsible
  for initiating the recursive CSE on users
  (http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-September/076903.html) but it
  is not strictly necessary since the shift is hooked into the visited user.  Of
  course it's safer to keep the DAG consistent at all times (e.g. for accurate
  number of uses, etc.).

  So rather than changing the fundamentals, I've decided to continue along the
  previous patches and detect the CSE.  This patch installs a very targeted
  DAGUpdateListener for the duration of a complex-pattern match and updates the
  matching state accordingly.  (Previous patches used HandleSDNode to detect the
  CSE but that's not practical here).  The listener is only installed on X86.

  I tested that there is no measurable overhead due to this while running
  through the spec2k BC files with llc.  The only thing we pay for is the
  creation of the listener.  The callback never ever triggers in spec2k since
  this is a corner case.

  Fixes rdar://problem/18206171

This fixes a possible crash in x86 code generation when compiling recent
llvm/clang trunk sources.

Direct commit to stable/10, since head already has llvm/clang 3.6.1,
which includes this fix.

Reported by:    jonathan, theraven
Upstream PR:    https://llvm.org/bugs/show_bug.cgi?id=24249
19:06:53 - r286032
(3 days 5 hours ago)
by emaste
Affects:  /head/share/mk/src.opts.mk
Use default CLANG build options for ARM

We previously disabled CLANG_FULL on (little-endian) ARM because the
build failed.  This is no longer the case and as of Clang 3.5 we cannot
build any part of the in-tree Clang with in-tree GCC, so it's no longer
necessary to disable CLANG_FULL.

Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D2525
18:55:51 - r286031
(3 days 5 hours ago)
by emaste
Affects:  /head/share/man/man5/src.conf.5
Regenerate src.conf(5) after r286016 and r286030
18:45:38 - r286030
(3 days 5 hours ago)
by emaste
Affects: 
(6 files)
/head/*
Allow ELF Tool Chain elfcopy to be installed as objcopy

ELF Tool Chain elfcopy is nearly a drop-in replacement for GNU objcopy,
but does not currently support PE output which is needed for building
x86 UEFI bits.

Add a src.conf knob to allow installing it as objcopy and set it by
default for aarch64 only, where we don't have a native binutils.

Reviewed by:    bapt
Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D2887
18:33:11 - r286029
(3 days 5 hours ago)
by emaste
Affects:  /head/tools/build/mk/OptionalObsoleteFiles.inc
Update OLD_FILES for tools provided by ELF Tool Chain or Binutils

Sponsored by:   The FreeBSD Foundation
18:04:01 - r286028
(3 days 6 hours ago)
by eri
Affects:  /head/sys/netinet/ip_output.c
ip_output normalization and fixes

ip_output has a big chunk of code used to handle special cases with pfil
consumers which also forces a reloop on it.
Gather all this code together to make it readable and properly handle the
reloop cases.

Some of the issues identified:

M_IP_NEXTHOP is not handled properly in existing code.
route reference leaking is possible with in FIB number change
route flags checking is not consistent in the function

Differential Revision:  https://reviews.freebsd.org/D3022
Reviewed by:    gnn
Approved by:    gnn(mentor)
MFC after:      4 weeks
17:59:13 - r286027
(3 days 6 hours ago)
by pkelsey
Affects: 
(6 files)
/head/sys/netinet/*
/head/sys/sys/mbuf.h
Revert r265338, r271089 and r271123 as those changes do not handle
non-inline urgent data and introduce an mbuf exhaustion attack vector
similar to FreeBSD-SA-15:15.tcp, but not requiring VNETs.

Address the issue described in FreeBSD-SA-15:15.tcp.

Reviewed by:    glebius
Approved by:    so
Approved by:    jmallett (mentor)
Security:       FreeBSD-SA-15:15.tcp
Sponsored by:   Norse Corp, Inc.
17:50:14 - r286026
(3 days 6 hours ago)
by eri
Affects: 
(2 files)
/stable/10/sys/netinet/ip_input.c
MFC 285325
Correct issue presented in r285051 by properly initializing variable.

Differential Revision:   https://reviews.freebsd.org/D3036
17:46:16 - r286025
(3 days 6 hours ago)
by eri
Affects: 
(2 files)
/stable/10/sys/netinet/ip_input.c
MFC r285051
Avoid doing multiple route lookups for the same destination IP during
forwarding.

Differential Revision:  https://reviews.freebsd.org/D2964
17:34:26 - r286024
(3 days 6 hours ago)
by emaste
Affects:  /head/usr.bin/ar/ar.c
ar: Fix deterministic mode default with options other than -q or -r

Reported by:    jhibbits
Reviewed by:    jhibbits
Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D3237
17:28:04 - r286023
(3 days 6 hours ago)
by brueffer
Affects: 
(2 files)
/releng/10.2/release/doc/en_US.ISO8859-1/hardware/article.xml
MFC: r285859 (via r286019)

Auto-generate hardware notes for pms(4).

Approved by:    re (gjb)
17:25:18 - r286022
(3 days 6 hours ago)
by brueffer
Affects: 
(3 files)
/releng/10.2/share/man/man4/Makefile
/releng/10.2/share/man/man4/pms.4
MFC: r285858, r286017

Add a basic manpage for the pms driver.

Approved by:    re (gjb)
17:18:27 - r286021
(3 days 6 hours ago)
by ed
Affects: 
(4 files)
/head/sys/compat/*
/head/sys/kern/sys_pipe.c
/head/sys/sys/syscallsubr.h
Make pipes in CloudABI work.

Summary:
Pipes in CloudABI are unidirectional. The reason for this is that
CloudABI attempts to provide a uniform runtime environment across
different flavours of UNIX.

Instead of implementing a custom pipe that is unidirectional, we can
simply reuse Capsicum permission bits to support this. This is nice,
because CloudABI already attempts to restrict permission bits to
correspond with the operations that apply to a certain file descriptor.

Replace kern_pipe() and kern_pipe2() by a single kern_pipe() that takes
a pair of filecaps. These filecaps are passed to the newly introduced
falloc_caps() function that creates the descriptors with rights in
place.

Test Plan:
CloudABI pipes seem to be created with proper rights in place:

https://github.com/NuxiNL/cloudlibc/blob/master/src/libc/unistd/pipe_test.c#L44

Reviewers: jilles, mjg

Reviewed By: mjg

Subscribers: imp

Differential Revision: https://reviews.freebsd.org/D3236
17:16:53 - r286020
(3 days 6 hours ago)
by ed
Affects:  /head/sys/kern/kern_descrip.c
/head/sys/sys/filedesc.h
Introduce falloc_caps() to create descriptors with capabilties in place.

falloc_noinstall() followed by finstall() allows you to create and
install file descriptors with custom capabilities. Add falloc_caps()
that can do both of these actions in one go.

This will be used by CloudABI to create pipes with custom capabilities.

Reviewed by:    mjg
17:05:42 - r286019
(3 days 7 hours ago)
by brueffer
Affects: 
(2 files)
/stable/10/release/doc/en_US.ISO8859-1/hardware/article.xml
MFC: r285859

Auto-generate hardware notes for pms(4).
16:40:48 - r286018
(3 days 7 hours ago)
by brueffer
Affects: 
(3 files)
/stable/10/share/man/man4/Makefile
/stable/10/share/man/man4/pms.4
MFC: r285858, r286017

Add a basic manpage for the pms driver.
16:37:36 - r286017
(3 days 7 hours ago)
by brueffer
Affects:  /head/share/man/man4/pms.4
Remove the AUTHORS section until it's clear who exactly wrote the driver.
15:42:22 - r286016
(3 days 8 hours ago)
by emaste
Affects:  /head/tools/build/options/WITHOUT_ELFTOOLCHAIN_TOOLS
Include c++filt and readelf in WITHOUT_ELFTOOLCHAIN_TOOLS
15:32:59 - r286015
(3 days 8 hours ago)
by sbruno
Affects: 
(6 files)
/head/sys/mips/conf/*
/head/sys/mips/rmi/*
Make Broadcom XLR use shared ds1374 RTC driver.

Remove its identical and redundant ds1374u version.

Differential Revision:  D3225
Submitted by:   kevin.bowling@kev009.com
14:16:25 - r286014
(3 days 9 hours ago)
by glebius
Affects: 
(4 files)
/releng/10.2/sys/net/pfvar.h
/releng/10.2/sys/netpfil/pf/pf.c
/releng/10.2/sys/netpfil/pf/pf_ioctl.c
Merge r285939-285941,285943,286004 from stable/10:
- Protect against ioctl() vs ioctl() races.
- Always lock hash row of a source node when updating
  its 'states' counter. [1]
- Don't dereference NULL is pf_get_mtag() fails. [2]
- During module unload drop locks before destroying UMA zone.

PR:             182401 [1]
PR:             200222 [2]
Approved by:    re (gjb)
14:07:43 - r286013
(3 days 10 hours ago)
by ae
Affects:  /head/sys/net/if_gif.c
/head/sys/netinet/in_gif.c
/head/sys/netinet6/in6_gif.c
Eliminate the use of m_copydata() in gif_encapcheck().

ip_encap already has inspected mbuf's data, at least an IP header.
And it is safe to use mtod() and do direct access to needed fields.
Add M_ASSERTPKTHDR() to gif_encapcheck(), since the code expects that
mbuf has a packet header.
Move the code from gif_validate[46] into in[6]_gif_encapcheck(), also
remove "martian filters" checks. According to RFC 4213 it is enough to
verify that the source address is the address of the encapsulator, as
configured on the decapsulator.

Reviewed by:    melifaro
Obtained from:  Yandex LLC
Sponsored by:   Yandex LLC
14:07:29 - r286012
(3 days 10 hours ago)
by dim
Affects: 
(11 files)
/stable/10/contrib/llvm/*
/stable/9/contrib/llvm/*
Revert r286007-r286009 for now, until I can figure out how to make the
fix compile with older gcc and libstdc++.
13:49:34 - r286011
(3 days 10 hours ago)
by pluknet
Affects:  /head/tests/sys/kern/unix_seqpacket_test.c
Fixed shutdown(2) unix(4) tests for SOCK_SEQPACKET after r285910 (by ed).
13:36:17 - r286010
(3 days 10 hours ago)
by emaste
Affects:  /head/usr.bin/ar/ar.1
/head/usr.bin/ar/ar.c
ar: enable deterministic mode by default

Ar cannot handle UIDs with more than 6 digits, and storing the mtime,
uid, gid and mode provides little to negative value anyhow for ar's
uses. Turn on deterministic (-D) mode by default; it can be disabled by
the user with -U.

PR:             196929
Relnotes:       Yes
Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D3190
13:14:34 - r286009
(3 days 10 hours ago)
by dim
Affects: 
(7 files)
/stable/9/contrib/llvm/include/llvm/CodeGen/SelectionDAGISel.h
/stable/9/contrib/llvm/lib/*
/stable/9/contrib/llvm/patches/patch-r286007-llvm-r219009-x86-codegen-crash.diff
Merge r286007 from stable/10:

Pull in r219009 from upstream llvm trunk (by Adam Nemet):

  [ISel] Keep matching state consistent when folding during X86 address match

  In the X86 backend, matching an address is initiated by the 'addr' complex
  pattern and its friends.  During this process we may reassociate and-of-shift
  into shift-of-and (FoldMaskedShiftToScaledMask) to allow folding of the
  shift into the scale of the address.

  However as demonstrated by the testcase, this can trigger CSE of not only the
  shift and the AND which the code is prepared for but also the underlying load
  node.  In the testcase this node is sitting in the RecordedNode and MatchScope
  data structures of the matcher and becomes a deleted node upon CSE.  Returning
  from the complex pattern function, we try to access it again hitting an assert
  because the node is no longer a load even though this was checked before.

  Now obviously changing the DAG this late is bending the rules but I think it
  makes sense somewhat.  Outside of addresses we prefer and-of-shift because it
  may lead to smaller immediates (FoldMaskAndShiftToScale is an even better
  example because it create a non-canonical node).  We currently don't recognize
  addresses during DAGCombiner where arguably this canonicalization should be
  performed.  On the other hand, having this in the matcher allows us to cover
  all the cases where an address can be used in an instruction.

  I've also talked a little bit to Dan Gohman on llvm-dev who added the RAUW for
  the new shift node in FoldMaskedShiftToScaledMask.  This RAUW is responsible
  for initiating the recursive CSE on users
  (http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-September/076903.html) but it
  is not strictly necessary since the shift is hooked into the visited user.  Of
  course it's safer to keep the DAG consistent at all times (e.g. for accurate
  number of uses, etc.).

  So rather than changing the fundamentals, I've decided to continue along the
  previous patches and detect the CSE.  This patch installs a very targeted
  DAGUpdateListener for the duration of a complex-pattern match and updates the
  matching state accordingly.  (Previous patches used HandleSDNode to detect the
  CSE but that's not practical here).  The listener is only installed on X86.

  I tested that there is no measurable overhead due to this while running
  through the spec2k BC files with llc.  The only thing we pay for is the
  creation of the listener.  The callback never ever triggers in spec2k since
  this is a corner case.

  Fixes rdar://problem/18206171

This fixes a possible crash in x86 code generation when compiling recent
llvm/clang trunk sources.

Direct commit to stable/10, since head already has llvm/clang 3.6.1,
which includes this fix.

Reported by:    jonathan, theraven
Upstream PR:    https://llvm.org/bugs/show_bug.cgi?id=24249

Merge r286008 from stable/10:

Add llvm patch corresponding to r286007.
13:07:18 - r286008
(3 days 11 hours ago)
by dim
Affects:  /stable/10/contrib/llvm/patches/patch-r286007-llvm-r219009-x86-codegen-crash.diff
Add llvm patch corresponding to r286007.
12:59:16 - r286007
(3 days 11 hours ago)
by dim
Affects:  /stable/10/contrib/llvm/include/llvm/CodeGen/SelectionDAGISel.h
/stable/10/contrib/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
/stable/10/contrib/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
Pull in r219009 from upstream llvm trunk (by Adam Nemet):

  [ISel] Keep matching state consistent when folding during X86 address match

  In the X86 backend, matching an address is initiated by the 'addr' complex
  pattern and its friends.  During this process we may reassociate and-of-shift
  into shift-of-and (FoldMaskedShiftToScaledMask) to allow folding of the
  shift into the scale of the address.

  However as demonstrated by the testcase, this can trigger CSE of not only the
  shift and the AND which the code is prepared for but also the underlying load
  node.  In the testcase this node is sitting in the RecordedNode and MatchScope
  data structures of the matcher and becomes a deleted node upon CSE.  Returning
  from the complex pattern function, we try to access it again hitting an assert
  because the node is no longer a load even though this was checked before.

  Now obviously changing the DAG this late is bending the rules but I think it
  makes sense somewhat.  Outside of addresses we prefer and-of-shift because it
  may lead to smaller immediates (FoldMaskAndShiftToScale is an even better
  example because it create a non-canonical node).  We currently don't recognize
  addresses during DAGCombiner where arguably this canonicalization should be
  performed.  On the other hand, having this in the matcher allows us to cover
  all the cases where an address can be used in an instruction.

  I've also talked a little bit to Dan Gohman on llvm-dev who added the RAUW for
  the new shift node in FoldMaskedShiftToScaledMask.  This RAUW is responsible
  for initiating the recursive CSE on users
  (http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-September/076903.html) but it
  is not strictly necessary since the shift is hooked into the visited user.  Of
  course it's safer to keep the DAG consistent at all times (e.g. for accurate
  number of uses, etc.).

  So rather than changing the fundamentals, I've decided to continue along the
  previous patches and detect the CSE.  This patch installs a very targeted
  DAGUpdateListener for the duration of a complex-pattern match and updates the
  matching state accordingly.  (Previous patches used HandleSDNode to detect the
  CSE but that's not practical here).  The listener is only installed on X86.

  I tested that there is no measurable overhead due to this while running
  through the spec2k BC files with llc.  The only thing we pay for is the
  creation of the listener.  The callback never ever triggers in spec2k since
  this is a corner case.

  Fixes rdar://problem/18206171

This fixes a possible crash in x86 code generation when compiling recent
llvm/clang trunk sources.

Direct commit to stable/10, since head already has llvm/clang 3.6.1,
which includes this fix.

Reported by:    jonathan, theraven
Upstream PR:    https://llvm.org/bugs/show_bug.cgi?id=24249
12:42:45 - r286006
(3 days 11 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_fd.c
Split up Capsicum to CloudABI rights conversion into two separate routines.

CloudABI's openat() ensures that files are opened with the smallest set
of relevant rights. For example, when opening a FIFO, unrelated rights
like CAP_RECV are automatically removed. To remove unrelated rights, we
can just reuse the code for this that was already present in the rights
conversion function.
11:22:19 - r286005
(3 days 12 hours ago)
by zbb
Affects:  /head/sys/arm64/arm64/gic_v3_its.c
/head/sys/arm64/arm64/gic_v3_var.h
Add quirk for ThunderX ITS device table size

Limit the number of supported device IDs to 0x100000
in order to decrease the size of the ITS device table so
that it matches with the HW capabilities.

Obtained from: Semihalf
Sponsored by:  The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3131
11:12:56 - r286004
(3 days 12 hours ago)
by glebius
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf.c
Merge r285944: fix typo: delete nsn if we were the last reference.
10:53:42 - r286003
(3 days 13 hours ago)
by ae
Affects:  /head/sys/netpfil/ipfw/ip_fw2.c
Reduce overhead of ipfw's me6 opcode.

Skip checks for IPv6 multicast addresses.
Use in6_localip() for global unicast.
And for IPv6 link-local addresses do search in the IPv6 addresses list.
Since LLA are stored in the kernel internal form, use
IN6_ARE_MASKED_ADDR_EQUAL() macro with lla_mask for addresses comparison.
lla_mask has zero bits in the second word, where we keep sin6_scope_id.

Obtained from:  Yandex LLC
Sponsored by:   Yandex LLC
09:57:34 - r286002
(3 days 14 hours ago)
by kib
Affects:  /head/sys/kern/vfs_bio.c
Move bufshutdown() out of the #ifdef INVARIANTS block.
08:12:05 - r286001
(3 days 15 hours ago)
by ae
Affects: 
(18 files)
/head/sys/*
Convert in_ifaddr_lock and in6_ifaddr_lock to rmlock.

Both are used to protect access to IP addresses lists and they can be
acquired for reading several times per packet. To reduce lock contention
it is better to use rmlock here.

Reviewed by:    gnn (previous version)
Obtained from:  Yandex LLC
Sponsored by:   Yandex LLC
Differential Revision:  https://reviews.freebsd.org/D3149
07:15:16 - r286000
(3 days 16 hours ago)
by jmg
Affects:  /head/sys/netipsec/xform.h
/head/sys/netipsec/xform_ah.c
/head/sys/netipsec/xform_esp.c
RFC4868 section 2.3 requires that the output be half...  This fixes
problems that was introduced in r285336...  I have verified that
HMAC-SHA2-256 both ah only and w/ AES-CBC interoperate w/ a NetBSD
6.1.5 vm...

Reviewed by:    gnn
06:35:36 - r285999
(3 days 17 hours ago)
by kp
Affects:  /head/sys/netpfil/pf/pf_norm.c
pf: Always initialise pf_fragment.fr_flags

When we allocate the struct pf_fragment in pf_fillup_fragment() we forgot to
initialise the fr_flags field. As a result we sometimes mistakenly thought the
fragment to not be a buffered fragment. This resulted in panics because we'd end
up freeing the pf_fragment but not removing it from V_pf_fragqueue (believing it
to be part of V_pf_cachequeue).
The next time we iterated V_pf_fragqueue we'd use a freed object and panic.

While here also fix a pf_fragment use after free in pf_normalize_ip().
pf_reassemble() frees the pf_fragment, so we can't use it any more.

PR:             201879, 201932
MFC after:      5 days
06:31:44 - r285998
(3 days 17 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_file.c
Implement CloudABI's readdir().

Summary:
CloudABI's readdir() system call could be thought of as a mixture
between FreeBSD's getdents(2) and pread(). Instead of using the file
descriptor offset, userspace provides a 64-bit cloudabi_dircookie_t
continue reading at a given point. CLOUDABI_DIRCOOKIE_START, having
value 0, can be used to return entries at the start of the directory.

The file descriptor offset is not used to store the cookie for the
reason that in a file descriptor centric environment, it would make
sense to allow concurrent use of a single file descriptor.

The remaining space returned by the system call should be filled with a
partially truncated copy of the next entry. The advantage of doing this
is that it gracefully deals with long filenames. If the C library
provides a buffer that is too small to hold a single entry, it can still
extract the directory entry header, meaning that it can retry the read
with a larger buffer or skip it using the cookie.

Test Plan:
This implementation passes the cloudlibc unit tests at:

        https://github.com/NuxiNL/cloudlibc/tree/master/src/libc/dirent

Reviewers: marcel, kib

Reviewed By: kib

Subscribers: imp

Differential Revision: https://reviews.freebsd.org/D3226
06:23:06 - r285997
(3 days 17 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/strtounum.c
Actually add the new code
06:22:42 - r285996
(3 days 17 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/Makefile
/head/usr.sbin/pw/pw.c
/head/usr.sbin/pw/pw.h
Create a strtounum function using the same API as strtonum

This function returns uintmax_t
Use this function to convert to gid_t/uid_t
03:06:08 - r285995
(3 days 21 hours ago)
by jeff
Affects:  /head/sys/fs/ext2fs/ext2_subr.c
 - Remove some dead code copied from ffs.
02:34:25 - r285994
(3 days 21 hours ago)
by imp
Affects:  /head/sbin/sysctl/sysctl.c
/head/share/man/man9/sysctl.9
Teach sysctl about the new optional suffix after IK to specify
precision. Update input as well. Add IK to the manual (it was missing
completely).

Differential Revision: https://reviews.freebsd.org/D3181
02:26:57 - r285993
(3 days 21 hours ago)
by jeff
Affects: 
(6 files)
/head/sys/*
 - Make 'struct buf *buf' private to vfs_bio.c.  Having a global variable
   'buf' is inconvenient and has lead me to some irritating to discover
   bugs over the years.  It also makes it more challenging to refactor
   the buf allocation system.
 - Move swbuf and declare it as an extern in vfs_bio.c.  This is still
   not perfect but better than it was before.
 - Eliminate the unused ffs function that relied on knowledge of the buf
   array.
 - Move the shutdown code that iterates over the buf array into vfs_bio.c.

Reviewed by:    kib
Sponsored by:   EMC / Isilon Storage Division
02:21:35 - r285992
(3 days 21 hours ago)
by araujo
Affects:  /head/usr.bin/ypcat/ypcat.c
/head/usr.bin/ypmatch/ypmatch.c
/head/usr.bin/ypwhich/ypwhich.c
Compilers will complain the usage of obsolescent variable declarations.
Also it will fix the build problem with sparc64.

Submitted by:   ed@
00:57:54 - r285991
(3 days 23 hours ago)
by gjb
Affects: 
(6 files)
/releng/10.2/sys/dev/pccbb/pccbb_pci.c
/releng/10.2/sys/dev/pci/*
MFS r285863 (jhb):
 Partially revert r284034.  In particular, revert the final change in
 this MFC (281874).  It broke suspend and resume on several Thinkpads
 (though not all) in 10 even though it works fine on the same laptops
 in HEAD.

PR:             201239
Reported by:    Kevin Oberman and several others
Approved by:    re (kib)
Sponsored by:   The FreeBSD Foundation
2015 - 07 - 28   (5 days ago)
22:48:58 - r285990
(4 days 1 hour ago)
by bdrewery
Affects:  /head/lib/libc/sys/unlink.2
unlink(2): Note the possibility for ENOSPC to be returned on ZFS.

PR:             154930
21:49:38 - r285989
(4 days 2 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_user.c
Reject usermod and userdel if the user concerned is not on the user database
supposed to be manipulated

This prevent pw usermod creating a new local user when requesting to usermod on
a username is defined in LDAP.

This issue only happens when modifying the local user database (not inpacting
commands when -V or -R are used).

PR:             187653
Submitted by:   tmwalaszek@gmail.com
21:47:37 - r285988
(4 days 2 hours ago)
by dumbbell
Affects:  /head/sys/dev/drm2/i915/i915_drv.h
/head/sys/dev/drm2/i915/i915_gem.c
/head/sys/dev/drm2/i915/i915_gem_gtt.c
drm/i915: Sort functions in i915_gem.c to match Linux 3.8's ordering

While here, reduce the style diff with Linux.

There is no functional change. The goal is to ease the future update to
Linux 3.8's i915 driver.

MFC after:      2 months
21:43:23 - r285987
(4 days 2 hours ago)
by delphij
Affects:  /releng/10.1/sys/conf/newvers.sh
Correct patchlevel.

Noticed by:     Piotr Kubaj
Approved by:    so
21:39:58 - r285986
(4 days 2 hours ago)
by bdrewery
Affects:  /head/usr.sbin/crunch/crunchgen/crunchgen.c
Fix rescue build after r284356 with STRIP= by using proper STRIPBIN per
build(7).

This was causing the following error:

  rescue
  sh: rescue: not found
  *** [rescue] Error code 127

  make[1]: stopped in /usr/obj/usr/src/rescue/rescue

Sponsored by:   EMC / Isilon Storage Division
21:10:58 - r285985
(4 days 3 hours ago)
by bapt
Affects: 
(4 files)
/head/usr.sbin/pw/pw.c
/head/usr.sbin/pw/tests/*
Check uid/gid used when creating a user/group are not larger than
UID_MAX/GID_MAX

PR:             173977
Reported by:    nvass@gmx.com
20:52:10 - r285984
(4 days 3 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/pw_user.c
Fix wrong warning printed after changing or updating NIS users

PR:             37672
Submitted by:   chris+freebsd@chrullrich.de
20:42:36 - r285983
(4 days 3 hours ago)
by gjb
Affects:  /releng/10.2/release/doc/share/xml/security.xml
Document SA-15:14 through SA-15:16.

Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
20:38:52 - r285982
(4 days 3 hours ago)
by gjb
Affects:  /stable/10/release/doc/share/xml/security.xml
/stable/8/release/doc/share/xml/security.xml
/stable/9/release/doc/share/xml/security.xml
Document SA-15:14 through SA-15:17.

Sponsored by:   The FreeBSD Foundation
20:24:09 - r285981
(4 days 3 hours ago)
by jeff
Affects:  /head/sys/kern/vfs_bio.c
 - Eliminate the EMPTYKVA queue.  It served as a cache of KVA allocations
   attached to bufs to avoid the overhead of the vm.  This purposes is now
   better served by vmem.  Freeing the kva immediately when a buf is
   destroyed leads to lower fragmentation and a much simpler scan algorithm.

Reviewed by:    kib
Sponsored by:   EMC / Isilon Storage Division
19:59:22 - r285980
(4 days 4 hours ago)
by delphij
Affects: 
(16 files)
/releng/8.4/*
/releng/9.3/*
Fix resource exhaustion in TCP reassembly. [SA-15:15]

Fix OpenSSH multiple vulnerabilities. [SA-15:16]

Fix BIND remote denial of service vulnerability. [SA-15:17]

Approved by:    so
19:59:11 - r285979
(4 days 4 hours ago)
by delphij
Affects: 
(8 files)
/releng/10.1/*
Fix patch(1) shell injection vulnerability. [SA-15:14]

Fix resource exhaustion in TCP reassembly. [SA-15:15]

Fix OpenSSH multiple vulnerabilities. [SA-15:16]

Approved by:    so
19:59:04 - r285978
(4 days 4 hours ago)
by delphij
Affects: 
(7 files)
/releng/10.2/crypto/openssh/*
/releng/10.2/sys/netinet/*
/releng/10.2/usr.bin/patch/*
Fix patch(1) shell injection vulnerability. [SA-15:14]

Fix resource exhaustion in TCP reassembly. [SA-15:15]

Fix OpenSSH multiple vulnerabilities. [SA-15:16]

Approved by:    re (so blanket)
19:58:54 - r285977
(4 days 4 hours ago)
by delphij
Affects: 
(12 files)
/stable/8/*
/stable/9/*
Fix resource exhaustion in TCP reassembly. [SA-15:15]

Fix OpenSSH multiple vulnerabilities. [SA-15:16]

Fix BIND remote denial of service vulnerability. [SA-15:17]
19:58:44 - r285976
(4 days 4 hours ago)
by delphij
Affects: 
(7 files)
/stable/10/crypto/openssh/*
/stable/10/sys/netinet/*
/stable/10/usr.bin/patch/*
Fix patch(1) shell injection vulnerability. [SA-15:14]

Fix resource exhaustion in TCP reassembly. [SA-15:15]

Fix OpenSSH multiple vulnerabilities. [SA-15:16]
19:58:38 - r285975
(4 days 4 hours ago)
by delphij
Affects:  /head/crypto/openssh/auth2-chall.c
/head/crypto/openssh/sshconnect.c
Fix multiple OpenSSH vulnerabilities.

Security:       CVE-2014-2653
Security:       CVE-2015-5600
Security:       FreeBSD-SA-15:16.openssh
19:58:36 - r285974
(4 days 4 hours ago)
by delphij
Affects:  /head/usr.bin/patch/common.h
/head/usr.bin/patch/inp.c
Fix shell injection vulnerability in patch(1) and drop SCCS
support by replacing system() with execve().

Future revisions may remove the functionality completely.

Obtained from:  Bitrig
Security:       CVE-2015-1416
19:15:44 - r285973
(4 days 4 hours ago)
by davidcs
Affects:  /head/sys/dev/bxe/bxe.c
/head/sys/dev/bxe/bxe.h
- Avoid lock contention in the if_transmit callback by using trylock and
 enqueueing the frames when it fails. This way there is some latency
 removed from the transmitting path.
- If IFF_DRV_OACTIVE is set (and also if IFF_DRV_RUNNING is not) just
 enqueue the desired frames and return successful transmit. This way we
 avoid to return errors on transmit side and resulting in
 possible out-of-order frames. Please note that IFF_DRV_OACTIVE is set
 everytime we get the threshold ring hit, so this can be happening quite
 often.

Submitted by:   Attilio.Rao@isilon.com
MFC after:5 days
18:41:28 - r285972
(4 days 5 hours ago)
by bdrewery
Affects: 
(6 files)
/head/contrib/libarchive/libarchive/archive_read_support_format_tar.c
/head/contrib/libarchive/libarchive/test/*
/head/lib/libarchive/test/Makefile
MFV r285970:

  Apply upstream changeset bf4f6ec64e:

  Fix issue 356: properly skip a sparse file entry in a tar file.

PR:             201506
MFC after:      3 days
Relnotes:       yes
18:37:23 - r285971
(4 days 5 hours ago)
by cem
Affects: 
(2 files)
/stable/10/sys/kern/sys_pipe.c
MFC r285483: pipe_direct_write: Fix mismatched pipelock/unlock

If a signal is caught in pipelock, causing it to fail, pipe_direct_write
should not try to pipeunlock.

Approved by:    markj (mentor)
Sponsored by:   EMC / Isilon Storage Division
17:48:34 - r285970
(4 days 6 hours ago)
by bdrewery
Affects: 
(5 files)
/vendor/libarchive/dist/Makefile.am
/vendor/libarchive/dist/libarchive/archive_read_support_format_tar.c
/vendor/libarchive/dist/libarchive/test/*
Apply upstream changeset bf4f6ec64e:

Fix issue 356: properly skip a sparse file entry in a tar file.
17:32:14 - r285969
(4 days 6 hours ago)
by bdrewery
Affects: 
(2 files)
/head/contrib/libarchive/libarchive
Mark vendor r285968 merged for r280870.
17:20:35 - r285968
(4 days 6 hours ago)
by bdrewery
Affects:  /vendor/libarchive/dist/libarchive/archive_read_disk_posix.c
Apply upstream changeset fa9e61:

Fix --one-file-system to include the directory encountered rather than
excluding it.
17:12:41 - r285967
(4 days 6 hours ago)
by kib
Affects: 
(2 files)
/stable/10/sys/kern/kern_exec.c
MFC r284956:
Do not calculate the stack's bottom address twice.
17:08:32 - r285966
(4 days 7 hours ago)
by kib
Affects: 
(2 files)
/stable/10/sys/kern/vfs_mount.c
MFC r285039:
Remove asserts which might reference freed memory.
17:06:13 - r285965
(4 days 7 hours ago)
by kib
Affects: 
(2 files)
/stable/10/share/man/man9/locking.9
MFC r285173:
Document the locking context for the directly dispatched callouts.
Cross-reference timeout(9).
17:00:03 - r285964
(4 days 7 hours ago)
by kib
Affects: 
(2 files)
/releng/10.2/sys/kern/kern_descrip.c
MFC r285134 (by mjg):
fd: de-k&r-ify functions + some whitespace fixes

MFC r285269:
Handle copyout for the fcntl(F_OGETLK) using oflock structure.

Approved by:    re (gjb)
16:39:36 - r285963
(4 days 7 hours ago)
by kib
Affects: 
(2 files)
/stable/10/sys/kern/kern_descrip.c
MFC r285134 (by mjg):
fd: de-k&r-ify functions + some whitespace fixes

MFC r285269:
Handle copyout for the fcntl(F_OGETLK) using oflock structure.
15:05:19 - r285962
(4 days 9 hours ago)
by gjb
Affects:  /releng/10.2/release/doc/en_US.ISO8859-1/relnotes/article.xml
Fix the description for r274486.

Submitted by:   glebius
Approved by:    re (implicit)
Sponsored by:   The FreeBSD Foundation
15:03:56 - r285961
(4 days 9 hours ago)
by gjb
Affects:  /stable/10/release/doc/en_US.ISO8859-1/relnotes/article.xml
Fix the description for r274486.

Submitted by:   glebius
Sponsored by:   The FreeBSD Foundation
14:59:29 - r285960
(4 days 9 hours ago)
by garga
Affects:  /head/sys/netpfil/pf/pf.c
Simplify logic added in r285945 as suggested by glebius

Approved by:    glebius
MFC after:      3 days
Sponsored by:   Netgate
14:20:33 - r285959
(4 days 9 hours ago)
by zbb
Affects:  /head/sys/contrib/alpine-hal
Import Annapurna Labs Alpine HAL to sys/contrib/

Import from vendor-sys/alpine-hal/2.7
SVN rev.: 285432
HAL version: 2.7

Obtained from:  Semihalf
Sponsored by:   Annapurna Labs
13:48:19 - r285958
(4 days 10 hours ago)
by glebius
Affects:  /head/release/doc/en_US.ISO8859-1/relnotes/article.xml
Fix the r272906 description.
13:16:08 - r285957
(4 days 10 hours ago)
by zbb
Affects:  /head/sys/dev/ofw/ofw_cpu.c
Limit ofw_cpu_early_foreach() to CPUs only

On some platforms, the /cpus node contains cpu-to-cluster
map which deffinitely is not a CPU node. Its presence was
causing incrementing of "id" variable and reporting more
CPUs available than it should.
To make "id" valid, increment it only when an entry really
is a CPU device.

Reviewed by:   andrew
Obtained from: Semihalf
Sponsored by:  The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3216
13:11:31 - r285956
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/releng/10.2/sys/boot/efi/libefi/efipart.c
MFS r285951: Avoid creating invalid UEFI device path

The UEFI loader on the 10.1 release install disk (disc1) modifies an
existing EFI_DEVICE_PATH_PROTOCOL instance in an apparent attempt to
truncate the device path.  In doing so it creates an invalid device
path.

Perform the equivalent action without modification of structures
allocated by firmware.

PR:             197641
Submitted by:   Chris Ruffin <chris.ruffin at intel.com>
Approved by:    re (gjb)
13:09:16 - r285955
(4 days 11 hours ago)
by emaste
Affects:  /head/contrib/llvm/tools/lldb/docs/lldb.1
Remove claim that the OS is Darwin from lldb(1)

Reported by:    bapt
12:57:19 - r285954
(4 days 11 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_file.c
Implement file attribute modification system calls for CloudABI.

CloudABI uses a system call interface to modify file attributes that is
more similar to KPI's/FUSE, namely where a stat structure is passed back
to the kernel, together with a bitmask of attributes that should be
changed. This would allow us to update any set of attributes atomically.

That said, I'd rather not go as far as to actually implement it that
way, as it would require us to duplicate more code than strictly needed.
Let's just stick to the combinations that are actually used by
cloudlibc.

Obtained from:  https://github.com/NuxiNL/freebsd
12:52:22 - r285953
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/releng/10.2/usr.sbin/mfiutil/mfi_foreign.c
MFS r285950: mfiutil: increase buffer size to accommodate sprintf string

PR:             201289
Approved by:    re (gjb)
12:46:37 - r285952
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/stable/10/usr.bin/clang/lldb/Makefile
MFC r285248: lldb: use .PATH to find man page instead of symlinking it
12:45:08 - r285951
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/stable/10/sys/boot/efi/libefi/efipart.c
MFC r285246: Avoid creating invalid UEFI device path

The UEFI loader on the 10.1 release install disk (disc1) modifies an
existing EFI_DEVICE_PATH_PROTOCOL instance in an apparent attempt to
truncate the device path.  In doing so it creates an invalid device
path.

Perform the equivalent action without modification of structures
allocated by firmware.

PR:             197641
Submitted by:   Chris Ruffin <chris.ruffin at intel.com>
12:40:41 - r285950
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/stable/10/usr.sbin/mfiutil/mfi_foreign.c
MFC r201289: mfiutil: increase buffer size to accommodate sprintf string

PR:             201289
12:40:09 - r285949
(4 days 11 hours ago)
by emaste
Affects: 
(2 files)
/stable/9/usr.sbin/mfiutil/mfi_foreign.c
MFC r285067: mfiutil: increase buffer size to accommodate sprintf string

PR:             201289
12:20:57 - r285948
(4 days 11 hours ago)
by bapt
Affects:  /head/usr.sbin/pw/pw.c
/head/usr.sbin/pw/tests/pw_userdel.sh
when -n is passed to any pw subcommand it is always expected to be considered as
a name so do not try to convert it to an id if it is a numeric value

PR:             31933
Reported by:    ted@impulse.net
Sponsored by:   gandi.net
11:21:33 - r285947
(4 days 12 hours ago)
by smh
Affects:  /head/UPDATING
Correct typo in UPDATING message

MFC after:      3 days
Sponsored by:   Multiplay
11:19:38 - r285946
(4 days 12 hours ago)
by smh
Affects:  /head/UPDATING
/head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c
Add warning about low KSTACK_PAGES for ZFS use

As ZFS requires a more kernel stack pages than is the default on some
architectures e.g. i386, warn if KSTACK_PAGES is less than
ZFS_MIN_KSTACK_PAGES (which is 4 at the time of writing).

MFC after:      3 days
Sponsored by:   Multiplay
10:31:34 - r285945
(4 days 13 hours ago)
by garga
Affects:  /head/sys/netpfil/pf/pf.c
Respect pf rule log option before log dropped packets with IP options or
dangerous v6 headers

Reviewed by:    gnn, eri
Approved by:    gnn
Obtained from:  pfSense
MFC after:      3 days
Sponsored by:   Netgate
Differential Revision:  https://reviews.freebsd.org/D3222
09:36:26 - r285944
(4 days 14 hours ago)
by glebius
Affects:  /head/sys/netpfil/pf/pf.c
Fix a typo in r280169. Of course we are interested in deleting nsn only
if we have just created it and we were the last reference.

Submitted by:   dhartmei
09:21:19 - r285943
(4 days 14 hours ago)
by glebius
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf_ioctl.c
Merge r283106:
  During module unload unlock rules before destroying UMA zones, which
  may sleep in uma_drain(). It is safe to unlock here, since we are already
  dehooked from pfil(9) and all pf threads had quit.
09:19:04 - r285942
(4 days 14 hours ago)
by dim
Affects: 
(14 files)
/stable/*
MFC r285340:

Fix swapped copyin(9) arguments in cxgb's iwch_arm_cq() function.
Detected by clang 3.7.0 with the warning:

sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c:309:18: error: variable
'rptr' is uninitialized when used here [-Werror,-Wuninitialized]
                chp->cq.rptr = rptr;
                               ^~~~
09:16:54 - r285941
(4 days 14 hours ago)
by glebius
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf.c
Merge r283061, r283063: don't dereference NULL is pf_get_mtag() fails.

PR:             200222
09:13:55 - r285940
(4 days 14 hours ago)
by glebius
Affects: 
(4 files)
/stable/10/sys/net/pfvar.h
/stable/10/sys/netpfil/pf/pf.c
/stable/10/sys/netpfil/pf/pf_ioctl.c
Merge 280169: always lock the hash row of a source node when updating
its 'states' counter.

PR:             182401
09:09:01 - r285939
(4 days 15 hours ago)
by glebius
Affects: 
(2 files)
/stable/10/sys/netpfil/pf/pf_ioctl.c
Merge r271458:
  - Provide a sleepable lock to protect against ioctl() vs ioctl() races.
  - Use the new lock to protect against simultaneous DIOCSTART and/or
    DIOCSTOP ioctls.
08:50:13 - r285938
(4 days 15 hours ago)
by tuexen
Affects:  /head/sys/netinet/sctp_indata.c
Fix a typo reported by Erik Cederstrand.

MFC after:      1 week
08:25:49 - r285937
(4 days 15 hours ago)
by ae
Affects: 
(2 files)
/stable/9/sbin/geom/class/part/geom_part.c
MFC r285735:
  lseek() allows an offset to be set beyond the end of file. Using
  it to check that partition has enough space to write bootcode doesn't
  work. Use the known size of provider instead.

  PR:           201504
08:22:50 - r285936
(4 days 15 hours ago)
by ae
Affects: 
(2 files)
/stable/10/sbin/geom/class/part/geom_part.c
MFC r285735:
  lseek() allows an offset to be set beyond the end of file. Using
  it to check that partition has enough space to write bootcode doesn't
  work. Use the known size of provider instead.

  PR:           201504
07:30:07 - r285935
(4 days 16 hours ago)
by hselasky
Affects:  /head/sys/dev/usb/controller/dwc_otg.c
/head/sys/dev/usb/controller/dwc_otg.h
Optimise the DWC OTG host mode driver's receive path:

Remove NAKing limit and pause IN and OUT transactions for 125us in
case of NAK response for BULK and CONTROL endpoints. This gets the
receive latency down and improves USB network throughput at the cost
of some CPU usage.

MFC after:      1 month
07:04:51 - r285934
(4 days 17 hours ago)
by kib
Affects:  /head/sys/amd64/include/atomic.h
Remove full barrier from the amd64 atomic_load_acq_*().  Strong
ordering semantic of x86 CPUs makes only the compiler barrier
neccessary to give the acquire behaviour.

Existing implementation ensured sequentially consistent semantic for
load_acq, making much stronger guarantee than required by standard's
definition of the load acquire.  Consumers which depend on the barrier
are believed to be identified and already fixed to use proper
operations.

Noted by:       alc (long time ago)
Reviewed by:    alc, bde
Tested by:      pho
Sponsored by:   The FreeBSD Foundation
MFC after:      2 weeks
06:58:10 - r285933
(4 days 17 hours ago)
by kib
Affects:  /head/sys/rpc/svc_vc.c
Remove useless acquire semantic from the atomic_add operation before
sosend().  The only release on the xp_snt_cnt is done after sosend(),
with an intent to synchronize with load_acq in svc_vc_ack().

Reviewed by:    alc
Tested by:      pho
Sponsored by:   The FreeBSD Foundation
MFC after:      2 weeks
06:55:08 - r285932
(4 days 17 hours ago)
by kib
Affects:  /head/sys/x86/include/specialreg.h
Add bit names for the IA32_MISC_ENABLE msr.

Sponsored by:   The FreeBSD Foundation
MFC after:      1 week
06:50:47 - r285931
(4 days 17 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_file.c
Implement directory and FIFO creation.

The file_create() system call can be used to create files of a given
type. Right now it can only be used to create directories and FIFOs. As
CloudABI does not expose filesystem permissions, this system call lacks
a mode argument. Simply use 0777 or 0666 depending on the file type.
06:36:49 - r285930
(4 days 17 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi/cloudabi_fd.c
/head/sys/compat/cloudabi/cloudabi_file.c
/head/sys/compat/cloudabi/cloudabi_util.h
Make fstat() and friends work.

Summary:
CloudABI provides access to two different stat structures:

- fdstat, containing file descriptor level status: oflags, file
  descriptor type and Capsicum rights, used by cap_rights_get(),
  fcntl(F_GETFL), getsockopt(SO_TYPE).
- filestat, containing your regular file status: timestamps, inode
  number, used by fstat().

Unlike FreeBSD's stat::st_mode, CloudABI file descriptor types don't
have overloaded meanings (e.g., returning S_ISCHR() for kqueues). Add a
utility function to extract the type of a file descriptor accurately.

CloudABI does not work with O_ACCMODEs. File descriptors have two sets
of Capsicum-style rights: rights that apply to the file descriptor
itself ('base') and rights that apply to any new file descriptors
yielded through openat() ('inheriting'). Though not perfect, we can
pretty safely decompose Capsicum rights to such a pair. This is done in
convert_capabilities().

Test Plan: Tests for these system calls are fairly extensive in cloudlibc.

Reviewers: jonathan, mjg, #manpages

Reviewed By: mjg

Subscribers: imp

Differential Revision: https://reviews.freebsd.org/D3171
05:48:58 - r285929
(4 days 18 hours ago)
by bapt
Affects: 
(2 files)
/stable/10/usr.sbin/ctladm/ctladm.c
make ctdladm(8) return 0 is everything was ok.

retval is used to test the return of XML_Parse function which is ok if 1 is
returned and retval it directly returned to the main function and used as an
exit value.

if all the parsing part is done reset retval to 0 so that the command return 0
if everything ok

Differential Revision:  https://reviews.freebsd.org/D3102
Reviewed by:    trasz
Sponsored by:   gandi.net
05:46:37 - r285928
(4 days 18 hours ago)
by whu
Affects: 
(2 files)
/stable/10/sys/dev/hyperv/netvsc/hv_netvsc_drv_freebsd.c
MFC r285785 Do not enable UDP checksum offloading when running on the
Hyper-V on Windows Server 2012 and earlier hosts.

Submitted by: whu
Reviewed by: royger
Approved by: royger
Relnotes: No
Sponsored by: Microsoft OSTC
Differential Revision:  https://reviews.freebsd.org/D3217
04:54:05 - r285927
(4 days 19 hours ago)
by marcel
Affects:  /head/sys/dev/proto/proto_busdma.c
Check the sync operation.
02:32:40 - r285926
(4 days 21 hours ago)
by araujo
Affects: 
(12 files)
/head/libexec/ypxfr/*
/head/usr.bin/*
/head/usr.sbin/*
Staticfy and constify some variables and clean up the code a bit to make it
more readable. No functional change.

Differential Revision:  D3166
Reviewed by:            kib
Sponsored by:           gandi.net
2015 - 07 - 27   (6 days ago)
22:35:54 - r285925
(5 days 1 hour ago)
by tuexen
Affects: 
(7 files)
/head/sys/netinet/*
Provide consistent error causes whenever an ABORT chunk is sent.

MFC after:      1 week
22:20:28 - r285924
(5 days 1 hour ago)
by dim
Affects: 
(249 files)
/projects/clang-trunk/*
Merge ^/head r285793 through r285923.
19:21:23 - r285923
(5 days 4 hours ago)
by delphij
Affects:  /head/usr.bin/ident/ident.c
Rewrite scan procedure with a FSM.  This improves code readability by
making clear transits between different states, and avoids bug with
handling repeated $'s.

Reviewed by:    bapt
Differential Revision: https://reviews.freebsd.org/D3221
17:53:11 - r285922
(5 days 6 hours ago)
by jimharris
Affects: 
(2 files)
/releng/10.2/sys/dev/nvme/nvme_ctrlr.c
MFS r285918:
MFC r285816:

  nvme: ensure csts.rdy bit is cleared before returning from nvme_ctrlr_disable

Approved by:    re (gjb)
Sponsored by:   Intel
17:52:17 - r285921
(5 days 6 hours ago)
by jimharris
Affects: 
(2 files)
/releng/10.2/sys/dev/nvme/nvme_ctrlr.c
MFS r285917:
MFC r285815:

  nvme: properly handle case where pci_alloc_msix does not alloc all vectors

Approved by:    re (gjb)
Sponsored by:   Intel
17:51:12 - r285920
(5 days 6 hours ago)
by jimharris
Affects: 
(2 files)
/releng/10.2/sbin/nvmecontrol/logpage.c
MFS r285916:
MFC r285796:

  nvmecontrol: read controller identify data before any log page operations

Approved by:    re (gjb)
Sponsored by:   Intel
17:50:05 - r285919
(5 days 6 hours ago)
by jimharris
Affects: 
(2 files)
/releng/10.2/sys/dev/nvd/nvd.c
MFS r285915:
MFC r285767:

  nvd: set d_delmaxsize to full capacity of NVMe namespace

  The NVMe specification has no ability to specify a maximum delete size
  that is less than the full capacity of the namespace - so just using the
  namespace size is the correct value here.

  This fixes reported issues where ZFS trim on init looked like it was
  hanging the system - previously the default I/O max size (128KB on
  Intel NVMe controllers) was used for delete operations which worked out
  to only about 8MB/s.  With this patch I can add an 800GB DC P3700
  drive to a ZFS pool in about 15-20 seconds.

Approved by:    re (gjb)
Sponsored by:   Intel
15:37:02 - r285918
(5 days 8 hours ago)
by jimharris
Affects: 
(2 files)
/stable/10/sys/dev/nvme/nvme_ctrlr.c
MFC r285816:

  nvme: ensure csts.rdy bit is cleared before returning from nvme_ctrlr_disable

Sponsored by:   Intel
15:35:36 - r285917
(5 days 8 hours ago)
by jimharris
Affects: 
(2 files)
/stable/10/sys/dev/nvme/nvme_ctrlr.c
MFC r285815:

  nvme: properly handle case where pci_alloc_msix does not alloc all vectors

Sponsored by:   Intel
15:34:02 - r285916
(5 days 8 hours ago)
by jimharris
Affects: 
(2 files)
/stable/10/sbin/nvmecontrol/logpage.c
MFC r285796:

  nvmecontrol: read controller identify data before any log page operations

Sponsored by:   Intel
15:32:37 - r285915
(5 days 8 hours ago)
by jimharris
Affects: 
(2 files)
/stable/10/sys/dev/nvd/nvd.c
MFC r285767:

  nvd: set d_delmaxsize to full capacity of NVMe namespace

  The NVMe specification has no ability to specify a maximum delete size
  that is less than the full capacity of the namespace - so just using the
  namespace size is the correct value here.

  This fixes reported issues where ZFS trim on init looked like it was
  hanging the system - previously the default I/O max size (128KB on
  Intel NVMe controllers) was used for delete operations which worked out
  to only about 8MB/s.  With this patch I can add an 800GB DC P3700
  drive to a ZFS pool in about 15-20 seconds.

Sponsored by:   Intel
15:26:50 - r285914
(5 days 8 hours ago)
by marius
Affects:  /head/sys/dev/usb/controller/xhci.c
/head/sys/dev/usb/controller/xhci_pci.c
- Move the remainder of host controller capability registers reading from
  xhci_start_controller() to xhci_init(). These values don't change at run-
  time so there's no point of acquiring them on every USB_HW_POWER_RESUME
  instead of only once during initialization. In r276717, reading the first
  couple of registers in question already had been moved as a prerequisite
  for the changes in that revision.
- Identify ASMedia ASM1042A controllers.
- Use NULL instead of 0 for pointers.

MFC after:      3 days
14:43:14 - r285913
(5 days 9 hours ago)
by marius
Affects:  /head/etc/devd/usb.conf
/head/sys/dev/usb/net/if_urndis.c
- Fix compilation after r285909 with USB_DEBUG defined.
- Regenerate usb.conf.
14:34:32 - r285912
(5 days 9 hours ago)
by marius
Affects:  /head/sys/dev/vt/colors/vt_termcolors.c
- Use __FBSDID().
- Const'ify cons_to_vga_colors.
- Fix line wrapping.

MFC after:      3 days
14:03:34 - r285911
(5 days 10 hours ago)
by marius
Affects:  /head/sys/dev/vt/hw/efifb/efifb.c
/head/sys/dev/vt/hw/fb/vt_early_fb.c
/head/sys/dev/vt/hw/fb/vt_fb.c
- Nuke dupe $FreeBSD$.
- Fix whitespace.

MFC after:      3 days
13:17:57 - r285910
(5 days 10 hours ago)
by ed
Affects: 
(4 files)
/head/lib/libc/sys/shutdown.2
/head/sys/kern/*
/head/sys/sys/param.h
Make shutdown() return ENOTCONN as required by POSIX, part deux.

Summary:
Back in 2005, maxim@ attempted to fix shutdown() to return ENOTCONN in case the
socket was not connected (r150152). This had to be rolled back (r150155), as it
broke some of the existing programs that depend on this behavior. I reapplied
this change on my system and indeed, syslogd failed to start up. I fixed this
back in February (279016) and MFC'ed it to the supported stable branches. Apart
from that, things seem to work out all right.

Since at least Linux and Mac OS X do the right thing, I'd like to go ahead and
give this another try. To keep old copies of syslogd working, only start
returning ENOTCONN for recent binaries.

I took a look at the XNU sources and they seem to test against both
SS_ISCONNECTED, SS_ISCONNECTING and SS_ISDISCONNECTING, instead of just
SS_ISCONNECTED. That seams reasonable, so let's do the same.

Test Plan:
This issue was uncovered while writing tests for shutdown() in CloudABI:

https://github.com/NuxiNL/cloudlibc/blob/master/src/libc/sys/socket/shutdown_test.c#L26

Reviewers: glebius, rwatson, #manpages, gnn, #network

Reviewed By: gnn, #network

Subscribers: bms, mjg, imp

Differential Revision: https://reviews.freebsd.org/D3039
12:14:14 - r285909
(5 days 11 hours ago)
by marius
Affects:  /head/sys/dev/usb/net/if_urndis.c
- Probe UICLASS_CDC/UISUBCLASS_ABSTRACT_CONTROL_MODEL/0xff again. This
  variant of Microsoft RNDIS, i. e. their unofficial version of CDC ACM,
  has been disabled in r261544 for resolving a conflict with umodem(4).
  Eventually, in r275790 that problem was dealt with in the right way.
  However, r275790 failed to put probing of RNDIS devices in question
  back.
- Initialize the device prior to querying it, as required by the RNDIS
  specification. Otherwise already determining the MAC address may fail
  rightfully.
- On detach, halt the device again.
- Use UCDC_SEND_ENCAPSULATED_{COMMAND,RESPONSE}. While these macros are
  resolving to the same values as UR_{CLEAR_FEATURE,GET_STATUS}, the
  former set is way more appropriate in this context.
- Report unknown - rather: unimplemented - events unconditionally and
  not just in debug mode. This ensures that we'll get some hint of what
  is going wrong instead of the driver silently failing.
- Deal with the Microsoft ActiveSync requirement of using an input buffer
  the size of the expected reply or larger - except for variably sized
  replies - when querying a device.
- Fix some pointless NULL checks, style bugs etc.

This changes allow urndis(4) to communicate with a Microsoft-certified
USB RNDIS test token.

MFC after:      3 days
Sponsored by:   genua mbh
10:07:29 - r285908
(5 days 14 hours ago)
by ed
Affects: 
(4 files)
/head/sys/compat/cloudabi/*
Add a futex implementation for CloudABI.

Summary:
CloudABI provides two different types of futex objects: read-write locks
and condition variables. There is no need to provide separate support
for once objects and thread joining, as these are efficiently simulated
by blocking on a read-write lock. Mutexes simply use read-write locks.

Condition variables always have a lock object associated to them. They
always know to which lock a thread needs to be migrated if woken up.
This allows us to implement requeueing. A broadcast on a condition
variable will never cause multiple threads to be woken up at once. They
will be woken up iteratively.

This implementation still has lots of room for improvement. Locking is
coarse and right now we use linked lists to store all of the locks and
condition variables, instead of using a hash table. The primary goal of
this implementation was to behave correctly. Performance will be
improved as we go.

Test Plan:
This futex implementation has been in use for the last couple of months
and seems to work pretty well. All of the cloudlibc and libc++ unit
tests seem to pass.

Reviewers: dchagin, kib, vangyzen

Subscribers: imp

Differential Revision: https://reviews.freebsd.org/D3148
10:04:28 - r285907
(5 days 14 hours ago)
by ed
Affects: 
(5 files)
/head/sys/compat/cloudabi64/*
Regenerate system call table.
10:04:06 - r285906
(5 days 14 hours ago)
by ed
Affects:  /head/sys/compat/cloudabi64/syscalls.master
/head/sys/contrib/cloudabi/syscalldefs_md.h
/head/sys/contrib/cloudabi/syscalldefs_mi.h
Sync in latest upstream system call definitions.

Futex object scopes have been renamed from using their own constants to
simply reusing the existing CLOUDABI_MAP_{PRIVATE,SHARED} flags, as they
are more accurate in this context.
04:11:41 - r285905
(5 days 19 hours ago)
by ngie
Affects: 
(2277 files)
/user/ngie/more-tests/*
MFhead @ r285904
00:28:51 - r285904
(5 days 23 hours ago)
by rmacklem
Affects: 
(2 files)
/stable/10/sys/fs/nfsclient/nfs_clrpcops.c
MFC: r285066
Alex Burlyga reported a POLA violation for the new NFS client as
compared to the old NFS client via email to the freebsd-fs@ mailing list.
For the new client, when multiple clients attempted to create a symbolic
link concurrently, more that one client would report success instead of
EEXIST. This was caused by code in the new client that mapped EEXIST to
OK assuming it was caused by a retried RPC request.
Since the old client did not do this, the patch defaults to the old
behaviour and permits the new behaviour to be enabled via a sysctl.

 


Powered by Python FreeBSD support by secnetix GmbH & Co. KG

Page generated in 32 ms, 263 revisions printed. Current time is 2015-08-02 00:11:34. All times are in UTC/GMT.