ed [Mon, 30 May 2016 13:51:27 +0000 (13:51 +0000)]
Fix the signature of the psignal() function.
POSIX 2008 added the psignal() function which has already been part of
the BSDs for a long time. The only difference is, the POSIX version uses
an 'int' for the signal number, unlike our version which uses an
'unsigned int'. Fix up the function to use an 'int'. This should not
affect the ABI.
ed [Mon, 30 May 2016 13:37:11 +0000 (13:37 +0000)]
Add missing types and constants to <netdb.h>.
According to POSIX, the netdb.h header must also provide in_addr_t and
in_port_t. It should also provide IPPORT_RESERVED. Copy over the
necessary bits from <netinet/in.h> to achieve that.
grembo [Mon, 30 May 2016 09:05:24 +0000 (09:05 +0000)]
Fix ig4 operation for certain machines
Some machine BIOSes use the I2C bus and leave it in a state that causes
interrupts to not work properly due to a pending interrupt having been
latched.
Refactor the code a bit to clear pending interrupts when I2C is enabled.
This fixes the primary problem.
Also fix a possible race condition in the interrupt handler where the
interrupt was being cleared after reading the status instead of before.
sephe [Mon, 30 May 2016 08:25:09 +0000 (08:25 +0000)]
hyperv/et: Fix STIMER0 operations.
- Make sure that STIMER0 is disabled before writting to it, since
writing to an enabled STIMER will result in undefined behaviour.
- It is unnecessary to reconfigure STIMER0 upon each et_start().
- Make sure that MSR_HV_REF_TIME_COUNT will not return 0, since
writing 0 to STIMER_COUNT will disable the target STIMER.
ed [Mon, 30 May 2016 07:50:57 +0000 (07:50 +0000)]
Add missing declaration of ino_t.
POSIX requires that <dirent.h> provides ino_t in the XSI case. In our
case, this wasn't being exposed, as d_ino is a macro that expands to
d_fileno that is an uint32_t, not an ino_t.
sephe [Mon, 30 May 2016 03:31:37 +0000 (03:31 +0000)]
tcp: Don't prematurely drop receiving-only connections
If the connection was persistent and receiving-only, several (12)
sporadic device insufficient buffers would cause the connection be
dropped prematurely:
Upon ENOBUFS in tcp_output() for an ACK, retransmission timer is
started. No one will stop this retransmission timer for receiving-
only connection, so the retransmission timer promises to expire and
t_rxtshift is promised to be increased. And t_rxtshift will not be
reset to 0, since no RTT measurement will be done for receiving-only
connection. If this receiving-only connection lived long enough
(e.g. >350sec, given the RTO starts from 200ms), and it suffered 12
sporadic device insufficient buffers, i.e. t_rxtshift >= 12, this
receiving-only connection would be dropped prematurely by the
retransmission timer.
We now assert that for data segments, SYNs or FINs either rexmit or
persist timer was wired upon ENOBUFS. And don't set rexmit timer
for other cases, i.e. ENOBUFS upon ACKs.
Discussed with: lstewart, hiren, jtl, Mike Karels
MFC after: 3 weeks
Sponsored by: Microsoft OSTC
Differential Revision: https://reviews.freebsd.org/D5872
lidl [Mon, 30 May 2016 02:09:19 +0000 (02:09 +0000)]
Remove extraneous altq header files
At some point during stable/10, the headers altq_codel.h and
altq_fairq.h where added to /usr/include/altq, before all of altq
was moved to /usr/include/net/altq.
bapt [Sun, 29 May 2016 22:28:39 +0000 (22:28 +0000)]
Regenerate timedef:
For Russian:
- Convert AM/PM which are badly formatted in CLDR to replace it by the proper
cyrillic
- Add a dependency on Text::Iconv so non unicode get the proper encoding for
AM/PM
- fix the date format having 'r.,' and convert it to 'r.' (also fixed in Bulgarian)
For All:
- Use complete Day of Week instead of the abbreviated one
bapt [Sun, 29 May 2016 22:27:42 +0000 (22:27 +0000)]
Add more hacks to correct CLDR data
For Russian:
- Convert AM/PM which are badly formatted in CLDR to replace it by the proper
cyrillic
- Add a dependency on Text::Iconv so non unicode get the proper encoding for
AM/PM
- fix the date format having 'r.,' and convert it to 'r.' (also fixed in Bulgarian)
For All:
- Use complete Day of Week instead of the abbreviated one
dim [Sun, 29 May 2016 20:54:16 +0000 (20:54 +0000)]
Pull in r269908 from upstream llvm trunk (by James Molloy):
[VectorUtils] Fix nasty use-after-free
In truncateToMinimalBitwidths() we were RAUW'ing an instruction then
erasing it. However, that intruction could be cached in the map we're
iterating over. The first check is "I->use_empty()" which in most
cases would return true, as the (deleted) object was RAUW'd first so
would have zero use count. However in some cases the object could
have been polluted or written over and this wouldn't be the case.
Also it makes valgrind, asan and traditionalists who don't like their
compiler to crash sad.
No testcase as there are no externally visible symptoms apart from a
crash if the stars align.
Fixes PR26509.
This should fix crashes when building a number of ports on arm64.
ngie [Sun, 29 May 2016 20:28:01 +0000 (20:28 +0000)]
Follow up to r300932
In the event MK_INET6 != no in userspace, but is disabled in the
kernel, or if there aren't any IPv6 addresses configured in userspace
(for lo0 and all physical interfaces), rpcbind would terminate
immediately instead of silently failing on
Skip over the IPv6 block to its respective cleanup with freeifaddrs if
creating the socket failed instead of terminating rpcbind immediately
MFC after: 6 days
X-MFC with: r300932
Reported by: O. Hartmann <ohartman@zedat.fu-berlin.de>
Sponsored by: EMC / Isilon Storage Division
markj [Sun, 29 May 2016 19:46:34 +0000 (19:46 +0000)]
Fix rpcbind init after r300941.
- getaddrinfo() sets res = NULL on failure and freeaddrinfo() always
dereferences its argument, so we should only free the address list after
a successful call.
- Address a second potential leak caused by getaddrinfo(AF_INET6)
overwriting the address list returned by getaddrinfo(AF_INET).
kib [Sun, 29 May 2016 19:35:55 +0000 (19:35 +0000)]
Stop inlining the struct ucond definition into struct pthread_cond.
This avoids unneccessary casts and make the calls to _thr_ucond_*()
functions less questionable.
The c_spare field was not included into struct pthread_cond, so the
change modifies libthr ABI for shared condvars. But since an off-page
does not legitimately contains any other data past the struct
pthread_cond, the change keeps shared condvars from pre- and post-
changed libthr compatible. Also note that the whole struct ucond was
never copied in or out by kernel.
For private condvars, the privately allocated memory was never exposed
outside libthr.
zbb [Sun, 29 May 2016 17:35:38 +0000 (17:35 +0000)]
Improve ARM debug_monitor for SMP machines
- Reset debug architecture and enable monitor for secondary
CPUs in init_secondary() rather than when configuring watchpoint, etc.
- Disable HW debugging capabilities when one of the CPU cores fails
to set up.
- Use dbg_capable() in a more atomic manner to avoid any mismatch
between CPUs.
zbb [Sun, 29 May 2016 17:33:49 +0000 (17:33 +0000)]
Fix debug_monitor code for older ARMs (ARM11)
- Enable monitor mode prior to accessing watchpoint
registers for v6, v6.1 architectures.
- Fix configuration scheme for v6, v6.1 and v7 Debug Archs
- Enable monitor unconditionally and for good instead
of enabling and disabling it (needed for single stepping
on on v6/v6.1)
dim [Sun, 29 May 2016 17:32:19 +0000 (17:32 +0000)]
Stop exposing the C11 _Atomic() macro in <sys/cdefs.h>, when compiling
for C++. It clashes with the one in libc++'s <atomic> header.
(Previously, the _Atomic() macro was defined in <stdatomic.h>, which is
only for use with C11, but for various reasons it was moved to its
current location in r251804.)
cperciva [Sun, 29 May 2016 17:26:40 +0000 (17:26 +0000)]
Retune SHA2 code for improved performance on CPUs with more ILP and
a preference for memory load instructions over large code footprints
with embedded immediate variables.
On amd64 CPUs from 2007-2008 there is not a significant change, but
amd64 CPUs from 2009-2010 get roughly 10% more throughput with this
code; amd64 CPUs from 2011-2012 get roughly 15% more throughput; and
AMD64 CPUs from 2013-2015 get 20-25% more throughput. The Raspberry
Pi 2 increases its throughput by 6-8%.
ache [Sun, 29 May 2016 16:39:28 +0000 (16:39 +0000)]
Micro optimize: C standard guarantees that right shift for unsigned value
fills left bits with zero, and we have exact 32bit unsigned value
(uint32_t), so there is no reason to add "& 0x7fffffff" here.
ed [Sun, 29 May 2016 16:10:01 +0000 (16:10 +0000)]
Replace local prototype of basename() with an inclusion of <libgen.h>.
libiberty currently defines the prototype for basename() itself instead
of using <libgen.h>. It still uses the BSD-style prototype instead of
the POSIX one, meaning that if FreeBSD would switch over to the POSIX
one, you wouldn't be able to use libiberty.h and libgen.h in a single
source file. It turns out that kgdb does this. Patch up libiberty to
just include <libgen.h>.
I'm currently talking to upstream to see whether we can come up with a
more complete solution that could be integrated, but for our
unmaintained copy of GDB in base, let's just apply the simplest
workaround possible.
ache [Sun, 29 May 2016 13:57:06 +0000 (13:57 +0000)]
1) Unifdef USE_WEAK_SEEDING since it is too obsolete to support and makes
reading hard.
2) Instead of doing range transformation in each and every function here,
do it single time directly in do_rand(). One "mod" operation overhead is not
a big deal, but the code looks nicer and possible future functions additions
or PRNG change do not miss range transformations neither have unneeded ones.
3) Use POSIX argument types for visible functions (cosmetic).
ache [Sun, 29 May 2016 11:54:09 +0000 (11:54 +0000)]
1) Unifdef USE_WEAK_SEEDING it is too obsolete to support and makes reading
harder.
2) ACM paper require seed to be in [1, 2^31-2] range, so use the same range
shifting as already done for rand(3). Also protect srandomdev() + TYPE_0 case
(non default) from negative seeds.
3) Don't check for valid "type" range in setstate(), it is always valid as
calculated. Instead add a check that rear pointer not exceeed end pointer.
ed [Sun, 29 May 2016 10:41:27 +0000 (10:41 +0000)]
Invoke the dirname() function in a POSIX compliant way.
POSIX requires that the argument of dirname() is of type "char *". In
other words, the input buffer can be modified by the function to store
the directory name.
Pull a copy of the string before calling dirname(). We don't care about
freeing up the memory afterwards, as this is done at the very bottom of
main(), right before the program terminates.
mmel [Sun, 29 May 2016 07:39:56 +0000 (07:39 +0000)]
ARM GIC: Allow to setup interrupt without configuration data.
In some cases, like for PCI devices, only interrupt numbers are enumerated
from HW. In this case, use INTR_foo_CONFORM as level and trigger values.
truckman [Sun, 29 May 2016 07:23:56 +0000 (07:23 +0000)]
Cast some expressions that multiply a long long constant by a
floating point constant to int64_t. This avoids the runtime
conversion of the the other operand in a set of comparisons from
int64_t to floating point and doing the comparisions in floating
point.
bdrewery [Sun, 29 May 2016 06:20:36 +0000 (06:20 +0000)]
Libcompat: Swap CXX/CFLAGS.
This is the same as done for the native build in r300770 to ensure that
the libc++ build reads from SYSROOT/usr/include/c++/v1 before reading
from SYSROOT/usr/include.
ngie [Sun, 29 May 2016 04:50:49 +0000 (04:50 +0000)]
Use require.progs with bc instead of require.files with /usr/bin/bc
This will make things more flexible if the program path changes in the future,
and the test in and of itself doesn't call /usr/bin/bc -- it just calls bc
ngie [Sun, 29 May 2016 02:59:03 +0000 (02:59 +0000)]
Make netif REQUIRE hostid
As noted in the PR, if etc/rc.d/zvol is removed, netif will be run before
hostid, and the MAC address generated for any bridge devices will be
non-deterministic. Make the MAC address generated be deterministic for
bridge devices by explicitly REQUIRE'ing hostid.
This fixes up the rest of the PR, inadvertently committed in r299844
bdrewery [Sun, 29 May 2016 01:45:15 +0000 (01:45 +0000)]
Libcompat: Set build tools in environment rather than make overrides.
This allows the CXX hack in r300917 for external GCC to work for
the lib32 build. It is also the same pattern as the native
build uses by adding the tools into CROSSENV for external
toolchain, rather than make overrides.
The C++ header files must be searched before /usr/include.
The original code in Makefile.inc1 did this before the change in r297271 to
use -isystem. The libc++ import in r300770 fixed the bug introduced in
r297271 by swapping XCFLAGS and XCXXFLAGS ordering in CROSSENV.
Moving the code from Makefile.inc1 to bsd.sys.mk in r300886 also made it
more difficult to control the order of the flags. CXXFLAGS is based on
CFLAGS, so any additions to it will come after CFLAGS. The CROSSENV
code from Makefile.inc1 was such that it was ensured the CXXFLAGS came
first by setting them directly in CXX. Using CXXFLAGS+=-I would work
here, but instead continue to use -isystem by adding it to CXX so it
comes before CFLAGS.
avos [Sat, 28 May 2016 18:49:17 +0000 (18:49 +0000)]
net80211: fix use-after-free in frame defragmentation procedure.
- Assign frame sequence/fragment number before frame concatenation;
otherwise, frame header pointer (wh) will be invalid.
- Move this code block upper and eliminate duplicate 'lwh = mtod()'
assignment.
Tested with wpi(4) (transmitter) (STA mode) and urtwn(4) (receiver)
(HOSTAP mode).
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c
Fix the resource.fs.zfs.statechange message. It had a number of
problems:
It was only being emitted on a transition to the HEALTHY state.
That made it impossible for zfsd to take actions based on drives
getting sicker.
It compared the new state to vdev_prevstate, which is the state that
the vdev had the last time it was opened. That doesn't make sense,
because a vdev can change state multiple times without being
reopened.
vdev_set_state contains logic that will change the device's new
state based on various conditions. However, the statechange event
was being posted _before_ that logic took effect. Now it's being
posted after.
allanjude [Sat, 28 May 2016 16:06:07 +0000 (16:06 +0000)]
Implement SHA-512 truncated (224 and 256 bits)
This implements SHA-512/256, which generates a 256 bit hash by
calculating the SHA-512 then truncating the result. A different initial
value is used, making the result different from the first 256 bits of
the SHA-512 of the same input. SHA-512 is ~50% faster than SHA-256 on
64bit platforms, so the result is a faster 256 bit hash.
The main goal of this implementation is to enable support for this
faster hashing algorithm in ZFS. The feature was introduced into ZFS
in r289422, but is disconnected because SHA-512/256 support was missing.
A further commit will enable it in ZFS.
This is the follow on to r292782
Reviewed by: cem
Sponsored by: ScaleEngine Inc.
Differential Revision: https://reviews.freebsd.org/D6061
andrew [Sat, 28 May 2016 13:05:39 +0000 (13:05 +0000)]
Don't panic in hwpmc when stopping sampling.
When hwpmc stops sampling it will set the pm_state to something other
than PMC_STATE_RUNNING. This means the following sequence can happen:
CPU 0: Enter the interrupt handler
CPU 0: Set the thread TDP_CALLCHAIN pflag
CPU 1: Stop sampling
CPU 0: Call pmc_process_samples, sampling is stopped so clears ps_nsamples
CPU 0: Finishes interrupt processing with the TDP_CALLCHAIN flag set
CPU 0: Call pmc_capture_user_callchain to capture the user call chain
CPU 0: Find all the pmc sample are free so no call chains need to be captured
CPU 0: KASSERT because of this
This fixes the issue by checking if any of the samples have been stopped
and including this in te KASSERT.
PR: 204273
Reviewed by: bz, gnn
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D6581
adrian [Sat, 28 May 2016 02:14:24 +0000 (02:14 +0000)]
[ath] add WB335 btcoex for initial testing.
This is like the WB222 coexistence (ie, "MCI", a message bus inside the
chip), and it's currently a cut/paste so I can start using it to flesh
out the differences with WB222.
It doesn't completely /do/ bluetooth coexistence, because it turns out
I need to add some contigmalloc'ed buffers to the btcoex path for this
type of hardware. I'm putting this work in the "people would like
to see functioning-ish btcoex before FreeBSD-11" bucket because I see
this as "broken".
np [Sat, 28 May 2016 00:38:17 +0000 (00:38 +0000)]
cxgbe/t4_tom: Exempt RDMA connections from a TCP sanity test for now, to
avoid panicking debug kernels.
t4_tom does not keep track of a connection once it switches to ULP mode
iWARP. If the connection falls out of ULP mode the driver/hardware seq#
etc. are out of sync. A better fix would be to figure out what the
current seq# are, update the driver's state, and perform all sanity
checks as usual.
bdrewery [Fri, 27 May 2016 23:57:43 +0000 (23:57 +0000)]
exec: Cease tracing if credentials will change with the new image.
This also prevents tracing to a P_INEXEC process since it could race
with other processes attaching to it in filemon_event_process_exec() due
to the filemon_get_proc() race of incrementing ref and then locking the
filemon. With the no-P_INEXEC invariant in place the p_filemon may only
be the same or NULL when trying to drop it in
filemon_event_process_exec().
np [Fri, 27 May 2016 23:44:33 +0000 (23:44 +0000)]
iw_cxgbe: Plug a lock leak in process_mpa_request().
If the parent is DEAD or connect_request_upcall() fails, the parent
mutex is left locked. This leads to a hang when process_mpa_request()
is called again for another child of the listening endpoint.
bdrewery [Fri, 27 May 2016 23:03:44 +0000 (23:03 +0000)]
Move external GCC compiler hacks to bsd.sys.mk.
This allows respecting -nostdinc, -nostdinc++ and -nostdlib before
making the decision to add in -isystem, etc. The -isystem flags
are problematic for building lib/libc++ and lib/libcxxrt which wants
to only use its own headers.
More information the need of these flags can be found at
https://gcc.gnu.org/ml/gcc/2016-03/msg00219.html
asomers [Fri, 27 May 2016 22:40:40 +0000 (22:40 +0000)]
Always create loopback routes on every fib
Always create loopback routes on every fib, for both IPv4 and IPv6
etc/rc.d/routing
Create loopback IPv4 and IPv6 routes on every fib at boot. Revert
278302; now that all FIBs have IPv6 loopback routes, the
"route add -reject" commands won't fail.
tests/etc/rc.d/routing_test.sh
Greatly simplify static_ipv6_loopback_route_for_each_fib. It was
written under the assumption that loopback routes would be added to
a given fib by the kernel as soon as an interface is configured on
that fib. However, the logic can be much simpler now that we simply
add loopback routes to all fibs at boot. This also removes the need
to run the test as root, removes the restriction that
net.add_addr_allfibs=0, and removes the need to configure fibs in
kyua.conf.
asomers [Fri, 27 May 2016 22:32:44 +0000 (22:32 +0000)]
Avoid issuing spa config updates for physical path when not necessary
ZFS's configuration needs to be updated whenever the physical path for a
device changes, but not when a new device is introduced. This is because new
devices necessarily cause config updates, but only if they are actually
accepted into the pool.
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c
Split vdev_geom_set_physpath out of vdev_geom_attrchanged. When
setting the vdev's physical path, only request a config update if
the physical path has changed. Don't request it when opening a
device for the first time, because the config sync will happen
anyway upstack.
sys/geom/geom_dev.c
Split g_dev_set_physpath and g_dev_set_media out of
g_dev_attrchanged
asomers [Fri, 27 May 2016 22:26:43 +0000 (22:26 +0000)]
Strip leading spaces off of a SCSI disk's serial number
sys/cam/scsi/scsi_xpt.c
Strip leading spaces off of a SCSI disk's reported serial number
when populating the CAM serial number. This affects the output of
"diskinfo -v" and the names of /dev/diskid/DISK-* device nodes,
among other things.
SPC5r05 says that the Product Serial Number field from the Unit
Serial Number VPD page is right-aligned. So any leading spaces are
not part of the actual serial number. Most devices don't left-pad
their serial numbers, but some do. In particular, the SN VPD page
that an LSI HBA emulates for a SATA drive contains enough
left-padding to fill a 20-byte field.
UPDATING
Add a note to UPDATING, because some users may have to update
/etc/fstab or geom labels.
dim [Fri, 27 May 2016 20:45:32 +0000 (20:45 +0000)]
After r300770, for libc++ and libcxxrt, use -isystem instead of -I.
This should fix builds with external gcc toolchains from ports, which
also use -isystem to work around problems with gcc's --sysroot
implementation. Thanks to Bryan Drewery for this workaround.