kib [Tue, 3 Feb 2015 11:54:33 +0000 (11:54 +0000)]
MFC r277794:
The sys_quotactl() contract demands that the mount point is
vfs_unbusy()ed when the cmd is Q_QUOTAON, regardless of other input
parameters or error return.
mav [Tue, 3 Feb 2015 08:06:13 +0000 (08:06 +0000)]
MFC r277419:
Allow skipping dmu_buf_will_dirty() call in dsl_dir_transfer_space().
dsl_dir_transfer_space() is mostly called after dsl_dir_diduse_space(),
which already calls dmu_buf_will_dirty() for the same dbuf and tx, so
its duplicate call in those cases will change nothing, only spend time.
Skipping this call by four times reduces time spent in dbuf_write_done()
and descendants, updating dataset statistics with several congested lock
acquisitions. When rewriting 8K zvol blocks at 1GB/s rate, this reduces
CPU time spent inside dbuf_write_done(), according to profiling, from 45%
of 683K samples to 18% of 422K.
nwhitehorn [Tue, 3 Feb 2015 00:29:58 +0000 (00:29 +0000)]
MFC r277792:
Fix bug in mapppings of multiple pages exposed by updates to the VSCSI
support in QEMU. Each page of a many page mapping was getting mapped to
the same physical address, which is not the desired behavior.
loos [Mon, 2 Feb 2015 12:48:13 +0000 (12:48 +0000)]
MFC r277042:
Add support to turn off Beaglebone with poweroff(8) or shutdown(8) -p.
To cut off the power we need to start the shutdown sequence by writing
the OFF bit on PMIC.
Once the PMIC is programmed the SoC needs to toggle the PMIC_PWR_ENABLE
pin when it is ready for the PMIC to cut off the power. This is done by
triggering the ALARM2 interrupt on SoC RTC.
The RTC driver only works in power management mode which means it won't
provide any kind of time keeping functionality. It only implements a way
to trigger the ALARM2 interrupt when requested.
loos [Mon, 2 Feb 2015 11:42:35 +0000 (11:42 +0000)]
MFC r276751:
Remove the check that prevent carp(4) advskew to be set to '0'.
CARP devices are created with advskew set to '0' and once you set it to
any other value in the valid range (0..254) you can't set it back to zero.
The code in question is also used to prevent that zeroed values overwrite
the CARP defaults when a new CARP device is created. Since advskew already
defaults to '0' for newly created devices and the new value is guaranteed
to be within the valid range, it is safe to overwrite it here.
dim [Sun, 1 Feb 2015 01:42:38 +0000 (01:42 +0000)]
MFC r273434 (by bapt):
Do not define bad_array_new_length::bad_array_new_length in libc++ anymore
when used in combinaison with libcxxrt since it is now defined there already.
This fixes building world
mav [Fri, 30 Jan 2015 09:05:43 +0000 (09:05 +0000)]
MFC r277247: Don't count status as sent until CTIO completes successfully.
If we aggregated status sending with data move and got error, allow status
to be updated and resent again separately. Without this command may stuck
without status sent at all.
rstone [Wed, 28 Jan 2015 21:51:34 +0000 (21:51 +0000)]
MFC r277352:
When mountd is creating sockets, it iterates over all addresses specified
in the "hosts" array and eventually looks up the network address with
getaddrinfo(). At one point it checks for a numeric address and if it
sees one, it sets a hint parameter to force getaddrinfo to interpret the
host as a numeric address. However that hint is not cleared for subsequent
iterations of the loop and if any hosts seen after this point are host names,
getaddrinfo will fail on the name. The result of this bug is that you cannot
pass a host name to the -h flag.
Unfortunately, the first iteration will either process ::1 or 127.0.0.1,
so the flag is set on the first iteration and all host names will fail
to be processed.
The same bug applies to rpc.lockd and rpc.statd, so fix them too.
Differential Revision: https://reviews.freebsd.org/D1507
Reported by: Dylan Martin
MFC after: 1 week
Sponsored by: Sandvine Inc.
mav [Wed, 28 Jan 2015 02:55:20 +0000 (02:55 +0000)]
MFC r277169: Reimplement TRIM throttling added in r248577.
Previous throttling implementation approached problem from the wrong side.
It significantly limited useful delaying of TRIM requests and aggregation
potential, while not so much controlled TRIM burstiness under heavy load.
With this change random 4K write benchmarks (probably the worst case for
TRIM) show me IOPS increase by 20%, average latency reduction by 30%, peak
TRIM bursts reduction by 3 times and same peak TRIM map size (memory usage).
Also the new logic does not force map size down so heavily, really allowing
to keep deleted data for 32 TXG or 30 seconds under moderate load. It was
practically impossible with old throttling logic, which pushed map down to
only 64 segments.
bryanv [Tue, 27 Jan 2015 06:19:30 +0000 (06:19 +0000)]
MFC r272886:
Add context pointer and source address to the UDP tunnel callback
These are needed for the forthcoming vxlan implementation. The context
pointer means we do not have to use a spare pointer field in the inpcb,
and the source address is required to populate vxlan's forwarding table.
trasz [Mon, 26 Jan 2015 13:37:18 +0000 (13:37 +0000)]
MFC r272294 by gavin@:
Make clear in the ipheth(4) hardware notes that this driver is for the
tethering functionality only. Add a "bugs" section to give a pointer
to usbconfig set_config if the device isn't automatically detected.
trasz [Mon, 26 Jan 2015 13:21:30 +0000 (13:21 +0000)]
MFC r274791:
Add missing error checking for kernel_port_{add,remove}(). Both can fail
for reasons yet unknown; don't make it increment cumulated_error as a kind
of temporary workaround.
luigi [Mon, 26 Jan 2015 03:26:37 +0000 (03:26 +0000)]
Merge 272659:
Add netmap support to libpcap. Tcpdump and other native pcap clients
can now run directly on netmap ports using netmap:foo or valeXX:YY
as device names.
Modifications to existing code are small and trivial,
the netmap-specific code is all in a new file.
Please be aware that in netmap mode the physical interface is
disconnected from the host stack, so libpcap will steal the traffic
not just make a copy.
For the full version of the code (including linux and autotools support) see
https://code.google.com/p/netmap-libpcap/
mav [Sun, 25 Jan 2015 14:31:44 +0000 (14:31 +0000)]
MFC r276983: When aggregating TRIM segments, move the new one to the end.
New segment at the list head may block all TRIM requests until txg of that
segment can be processed. On my random I/O tests this change reduce peak
TRIM list length from 650 to 450 segments. Hopefully it should reduce TRIM
burstiness when list processing is unblocked.
mav [Sun, 25 Jan 2015 14:29:40 +0000 (14:29 +0000)]
MFC r276952: Add LBA as secondary sort key for synchronous I/O requests.
On FreeBSD gethrtime() implemented via getnanouptime(), that has 1ms (1/hz)
precision. It makes primary sort key (timestamp) collision very possible.
In such situations sorting by secondary key of LBA is much more reasonable
then by totally meaningless zio pointer value.
With this change on multi-threaded synchronous ZVOL read I've measured 10%
throughput increase and average latency reduction.
cperciva [Sun, 25 Jan 2015 08:16:51 +0000 (08:16 +0000)]
MFC r277318:
When disabling C3+ CPU states due to the CPU_QUIRK_NO_C3 quirk, don't
accidentally enable non-existent states.
This bug was triggered if ACPI advertises the presence of a C2 state
which we fail to parse via acpi_PkgGas due to our lack of support for
FFixedHW resources, and causes an immediate panic when an attempt is
made to enter the (NULL) state.
One affected platform is the EC2 c4.8xlarge VM instance type; there
may be others.
ngie [Sun, 25 Jan 2015 00:28:15 +0000 (00:28 +0000)]
MFC r277278:
r277278 (by ngie):
Fix lib/libthr/tests/detach_test
- Eliminate race with liberal use of sleep(3) [1]
- Fix NetBSD-specific implementation way of testing result from pthread_cancel
by testing with `td` instead of `NULL` [2]
delphij [Fri, 23 Jan 2015 18:39:26 +0000 (18:39 +0000)]
MFC r275922: MFV r275914:
As of r270383, the dbuf_compare comparator compares the dbuf
attributes in the following order:
db_level (indirect level)
db_blkid (block number)
db_state (current state)
the address of the element
Because db_state is being considered before the element's state,
changing of db_state would affect balancedness of the AVL tree,
even when the address of element compares differently. For
instance, in dbuf_create, db_state may be altered after the
node is inserted into the AVL tree and may break AVL tree
balancedness.
Instead of using db_state as a comparision critera (introduced
in r270383), consider it only when we are doing a lookup, that
is one of the two dbuf pointers contains DB_SEARCH.
Illumos issue:
5422 preserve AVL invariants in dn_dbufs
delphij [Fri, 23 Jan 2015 18:36:21 +0000 (18:36 +0000)]
MFC r275812: MFV r275784:
Plug a memory leak in libzfs. In zfs_iter_bookmarks, an nvlist is allocated
before calling lzc_get_bookmarks, which allocates the nvlist again (and
overwrites the pointer to previously allocated list).
Illumos issue:
5427 memory leak in libzfs when doing rollback