Commit Graph

91923 Commits

Author SHA1 Message Date
Adrian Chadd
1a85141ad4 Pull out the if_transmit() work and revert back to ath_start().
My changed had some rather significant behavioural changes to throughput.
The two issues I noticed:

* With if_start and the ifnet mbuf queue, any temporary latency
  would get eaten up by some mbufs being queued.  With ath_transmit()
  queuing things to ath_buf's, I'd only get 512 TX buffers before I
  couldn't queue any further frames.

* There's also some non-zero latency involved with TX being pushed
  into a taskqueue via direct dispatch.  Any time the scheduler didn't
  immediately schedule the ath TX task would cause extra latency.
  Various 1ge/10ge drivers implement both direct dispatch (if the TX
  lock can be acquired) and deferred task transmission (if the TX lock
  can't be acquired), with frames being pushed into a drbd queue.
  I'll have to do this at some point, but until I figure out how to
  deal with 802.11 fragments, I'll have to wait a while longer.

So what I saw:

* lots of extra latency, specially under load - if the taskqueue
  wasn't immediately scheduled, things went pear shaped;

* any extra latency would result in TX ath_buf's taking their sweet time
  being replenished, so any further calls to ath_transmit() would drop
  mbufs.

* .. yes, there's no explicit backpressure here - things are just dropped.
  Eek.

With this, the general performance has gone up, but those subtle if_start()
related race conditions are back.  For some reason, this is doubly-obvious
with the AR5416 NIC and I don't quite understand why yet.

There's an unrelated issue with AR5416 performance in STA mode (it's
fine in AP mode when bridging frames, weirdly..) that requires a little
further investigation.  Specifically - it works fine on a Lenovo T40
(single core CPU) running a March 2012 9-STABLE kernel, but a Lenovo T60
(dual core) running an early November 2012 kernel behaves very poorly.
The same hardware with an AR9160 or AR9280 behaves perfectly.
2013-02-13 05:32:19 +00:00
Rui Paulo
eaba9848dd Introduce PLATFORMMETHOD_END and use it. 2013-02-13 02:21:45 +00:00
Marcel Moolenaar
f1acb9d36e Eliminate the PC_CURTHREAD symbol and load the current thread's
thread structure pointer atomically from r13 (the pcpu pointer)
for the current CPU/core.
Add a CTASSERT in machdep.c to make sure that pc_curthread is in
fact the first field in struct pcpu.

The only non-atomic operations left were those related to process-
space operations, such as casuword, subyte, suword16, fubyte,
fuword16, copyin, copyout and their variations.

The casuword function has been re-structured more complete than
the others. This way we have an example of a better bundling
without introducing a lot of risk when we get it wrong. The
other functions can be rebundled in separate commits and with
the appropriate testing.
2013-02-12 17:38:35 +00:00
Marcel Moolenaar
22aa5c496b Eliminate padding by moving 'narg' next to 'code'. Both are 32-bit
entities in the syscall_args structure that otherwise has 64-bit
only fields.
2013-02-12 17:24:41 +00:00
Konstantin Belousov
dd0b4fb6d5 Reform the busdma API so that new types may be added without modifying
every architecture's busdma_machdep.c.  It is done by unifying the
bus_dmamap_load_buffer() routines so that they may be called from MI
code.  The MD busdma is then given a chance to do any final processing
in the complete() callback.

The cam changes unify the bus_dmamap_load* handling in cam drivers.

The arm and mips implementations are updated to track virtual
addresses for sync().  Previously this was done in a type specific
way.  Now it is done in a generic way by recording the list of
virtuals in the map.

Submitted by:	jeff (sponsored by EMC/Isilon)
Reviewed by:	kan (previous version), scottl,
	mjacob (isp(4), no objections for target mode changes)
Discussed with:	     ian (arm changes)
Tested by:	marius (sparc64), mips (jmallet), isci(4) on x86 (jharris),
	amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
Marcel Moolenaar
44c169253d Now that we actually use more memory descriptors, make sure to dump
them as well.
2013-02-12 16:51:43 +00:00
Gleb Smirnoff
23f4fd6da4 Substitute '#ifdef ALIGNED_POINTER' with '#ifndef __NO_STRICT_ALIGNMENT',
since the former is defined everywhere. This cuts off some code not
necessary on non strict aligment arches.

Reviewed by:	adrian
Sponsored by:	Nginx, Inc.
2013-02-12 11:24:37 +00:00
Oleksandr Tymoshenko
bb2902189e Remove debug output 2013-02-12 07:27:40 +00:00
Pawel Jakub Dawidek
04f5c32651 Style. 2013-02-11 23:14:54 +00:00
Marius Strobl
18716f9f4b Update comments to reflect r246689. 2013-02-11 23:05:10 +00:00
Pawel Jakub Dawidek
9270ed9d38 Style. 2013-02-11 22:54:23 +00:00
Marius Strobl
bdc5f0172e Make SYSCTL_{LONG,QUAD,ULONG,UQUAD}(9) work as advertised and also handle
constant values.

Reviewed by:	kib
MFC after:	3 days
2013-02-11 21:50:00 +00:00
Martin Matuska
6a33bbc041 Merge zfs_ioctl.c code that should have been merged together with ZFS v28.
Fixes several problems if working with read-only pools.

Changed code originaly introduced in onnv-gate 13061:bda0decf867b
Contains changes up to illumos-gate 13700:4bc0783f6064

PR:		kern/175897
Suggested by:	avg

MFC after:	2 weeks
2013-02-11 21:10:55 +00:00
Michael Tuexen
2416af26a0 Send the adaptation layer indication only if set by the user.
MFC after: 3 days
Discussed with: rrs
2013-02-11 21:02:49 +00:00
Neel Natu
485b3300cc Implement guest vcpu pinning using 'pthread_setaffinity_np(3)'.
Prior to this change pinning was implemented via an ioctl (VM_SET_PINNING)
that called 'sched_bind()' on behalf of the user thread.

The ULE implementation of 'sched_bind()' bumps up 'td_pinned' which in turn
runs afoul of the assertion '(td_pinned == 0)' in userret().

Using the cpuset affinity to implement pinning of the vcpu threads works with
both 4BSD and ULE schedulers and has the happy side-effect of getting rid
of a bunch of code in vmm.ko.

Discussed with:	grehan
2013-02-11 20:36:07 +00:00
Martin Matuska
9689178c3f MFV r246633:
Import vendor bugfixes regarding SA rounding, header size and layout.
This was already partially fixed by avg.

Illumos ZFS issues:
  3512 rounding discrepancy in sa_find_sizes()
  3513 mismatch between SA header size and layout

References:
  https://www.illumos.org/issues/3512
  https://www.illumos.org/issues/3513

MFC after:	2 weeks
2013-02-11 14:29:38 +00:00
Martin Matuska
bb03847418 MFV r246394:
Add tunable to allow block allocation on degraded vdevs.

Illumos ZFS issues:
  3507 Tunable to allow block allocation even on degraded vdevs

References:
  https://www.illumos.org/issues/3507

MFC after:	2 weeks
2013-02-11 13:59:57 +00:00
Michael Tuexen
c53f854a17 Don't send kernel provided information in the User Initiated
ABORT cause, since the user can also provide this kind of
information. So the receiver doesn't know who provided the
information.
While there: Fix a bug where the stack would send a malformed
ABORT chunk when using a send() call with SCTP_ABORT|SCT_SENDALL
flags.

MFC after: 3 days
2013-02-11 13:57:03 +00:00
Martin Matuska
ff20578569 MFV r246392:
Import vendor ZFS bugfix fixing a possible deadlock in arc_read().

Illumos ZFS issues:
  3498 panic in arc_read(): !refcount_is_zero(&pbuf->b_hdr->b_refcnt)

References:
  https://www.illumos.org/issues/3498

MFC after:	2 weeks
2013-02-11 12:42:11 +00:00
Oleksandr Tymoshenko
2655166c71 A10 reset mechanism is the same for all boards in this family so remove
redundant reset function implementation pointer. We might want to
ressurect it later when support for other Allwinner chips is introduced.
2013-02-11 11:33:56 +00:00
Oleksandr Tymoshenko
3748deb755 Add watchdog driver for Allwinner A10 2013-02-11 11:31:23 +00:00
Gleb Smirnoff
24421c1c32 Resolve source address selection in presense of CARP. Add a couple
of helper functions:

- carp_master()   - boolean function which is true if an address
		    is in the MASTER state.
- ifa_preferred() - boolean function that compares two addresses,
		    and is aware of CARP.

  Utilize ifa_preferred() in ifa_ifwithnet().

  The previous version of patch also changed source address selection
logic in jails using carp_master(), but we failed to negotiate this part
with Bjoern. May be we will approach this problem again later.

Reported & tested by:	Anton Yuzhaninov <citrin citrin.ru>
Sponsored by:		Nginx, Inc
2013-02-11 10:58:22 +00:00
Rui Paulo
392c880066 Use DEVMETHOD_END. 2013-02-11 08:20:21 +00:00
Adrian Chadd
53a835d2db Put this back into the ath taskqueue rather than the ath TX taskqueue.
This now should mean all the entry points into the software TX
scheduler are back in the same taskqueue.
2013-02-11 07:49:40 +00:00
Martin Matuska
8a2dc7faae MFV r246390:
Import minor type change in refcount.h header from vendor (illumos).

MFC after:	2 weeks
2013-02-11 07:48:57 +00:00
Adrian Chadd
a40880ade4 Go back to direct-dispatch of the software queue and frame TX paths
when they're being called from the TX completion handler.

Going (back) through the taskqueue is just adding extra locking and
latency to packet operations.  This improves performance a little bit
on most NICs.

It still hasn't restored the original performance of the AR5416 NIC
but the AR9160, AR9280 and later NICs behave very well with this.

Tested:

* AR5416 STA (still tops out at ~ 70mbit TCP, rather than 150mbit TCP..)
* AR9160 hostap (good for both TX and RX)
* AR9280 hostap (good for both TX and RX)
2013-02-11 07:48:26 +00:00
Adrian Chadd
81561d0459 Extend the timestamp to be a timeval, rather than ticks.
This makes it easier to see TX and RX buffer latencies.
2013-02-11 02:48:49 +00:00
Tim Kientzle
53d21f0bbe Provide verbose help for fdt commands on platforms that use it. 2013-02-11 02:15:26 +00:00
Konstantin Belousov
2871baa49a Remove the ia64-specific code fragment, which effect is more cleanly
done by the call to trans_prot() function a line before.

Discussed with:	Oliver Pinter <oliver.pntr@gmail.com>
MFC after:	1 week
2013-02-10 20:08:33 +00:00
Michael Tuexen
f0d44a49a0 Make sure that received packets for removed addresses are handled
consistently. While there, make variable names consistent.

MFC after: 3 days
2013-02-10 19:57:19 +00:00
Pedro F. Giffuni
a9d1b29995 ext2fs: Use prototype declarations for function definitions
Submitted by:	Christoph Mallon
MFC after:	2 weeks
2013-02-10 19:49:37 +00:00
Martin Matuska
fd9778c236 MFV r246388:
Import vendor bugfixes

Illumos ZFS issues:
  3422 zpool create/syseventd race yield non-importable pool
  3425 first write to a new zvol can fail with EFBIG

References:
  https://www.illumos.org/issues/3422
  https://www.illumos.org/issues/3425

MFC after:	2 weeks
2013-02-10 19:32:55 +00:00
Andrey V. Elsukov
c6e31e81e0 Add bootcamp support to the loader.
Tested by:	dchagin
MFC after:	1 week
2013-02-10 19:27:17 +00:00
Hans Petter Selasky
6950c75f40 - Move scratch data from the USB bus structure to the USB device structure
so that simultaneous access cannot happen. Protect scratch area using
the enumeration lock. Also reduce stack usage in usbd_transfer_setup()
by moving some big stack members to the scratch area. This saves around
200 bytes of stack.
- Fix a whitespace.

MFC after:	1 week
2013-02-10 10:56:13 +00:00
Hans Petter Selasky
74b50dd979 Fix correct use of USB header files. 2013-02-10 10:55:20 +00:00
Hans Petter Selasky
645e4d17f0 - Streamline detach logic in wlan drivers, so that
freed memory cannot be used during detach.
- Remove all panic() calls from the urtw driver because
  panic() is not appropriate here.
- Remove redundant checks for device detached in
  device detach callbacks.
- Use DEVMETHOD_END to mark end of device methods.

MFC after:	2 weeks
2013-02-10 10:36:16 +00:00
Konstantin Belousov
9604a7f1b8 Fix several unsafe pointer dereferences in the buffered_write()
function, implementing the sysctl vfs.ffs.set_bufoutput (not used in
the tree yet).

- The current directory vnode dereference is unsafe since fd_cdir
  could be changed and unreferenced, lock the filedesc around and vref
  the fd_cdir.
- The VTOI() conversion of the fd_cdir is unsafe without first
  checking that the vnode is indeed from an FFS mount, otherwise
  the code dereferences a random memory.
- The cdir could be reclaimed from under us, lock it around the
  checks.
- The type of the fp vnode might be not a disk, or it might have
  changed while the thread was in flight, check the type.

Reviewed and tested by:	mckusick
MFC after:	2 weeks
2013-02-10 10:17:33 +00:00
Attilio Rao
5e60cb948e Remove a racy checks on resident and cached pages for
tmpfs_mapped{read, write}() functions:
- tmpfs_mapped{read, write}() are only called within VOP_{READ, WRITE}(),
  which check before-hand to work only on valid VREG vnodes.  Also the
  vnode is locked for the duration of the work, making vnode reclaiming
  impossible, during the operation. Hence, vobj can never be NULL.
- Currently check on resident pages and cached pages without vm object
  lock held is racy and can do even more harm than good, as a page could
  be transitioning between these 2 pools and then be skipped entirely.
  Skip the checks as lookups on empty splay trees are very cheap.

Discussed with:	alc
Tested by:	flo
MFC after:	2 weeks
2013-02-10 01:04:10 +00:00
Tim Kientzle
3d98a28bea Fix breakage introduced in r246318. 2013-02-09 21:36:14 +00:00
Michael Tuexen
a1cb341b5d Cleanup the handling of address scopes. Announce in the INIT/INIT-ACK
only the supported address types. While there, do some whitespace
cleanups.

MFC after: 1 week
2013-02-09 17:26:14 +00:00
Michael Tuexen
c39cfa1f7e Fix a bug where HEARTBEATs were still sent in SHUTDOWN_SENT or
SHUTDOWN_ACK_SENT state. While there, make the corresponding
code consistent.

MFC after: 1 week
2013-02-09 08:27:08 +00:00
Xin LI
ef17620fc8 MFV r245512:
* Illumos zfs issue #3035 [1] LZ4 compression support in ZFS.

LZ4 is a new high-speed BSD-licensed compression algorithm created
by Yann Collet that delivers very high compression and decompression
performance compared to lzjb (>50% faster on compression, >80% faster
on decompression and around 3x faster on compression of incompressible
data), while giving better compression ratio [1].

This version of LZ4 corresponds to upstream's [2] revision 85.

Please note that for obvious reasons this is not backward read
compatible.  This means once a pool have LZ4 compressed data, these
data can no longer be read by older ZFS implementations.

Local changes:

 - On-stack hash table disabled and using kernel slab allocator
   instead, at this time.  This requires larger kernel thread stack
   for zio workers.  This may change in the future should we adjusted
   the zio workers' thread stack size.
 - likely and unlikely will be undefined if they are already defined,
   this is required for i386 XEN build.
 - Removed De Bruijn sequence based __builtin_ctz family of builtins
   in favor of the latter.  Both GCC and clang supports these builtins.
 - Changed the way the LZ4 code detects endianness.
 - Manual pages modifications to mention the feature based on Illumos
   counterpart.
 - Boot loader changes to make it support LZ4 decompression.

[1] https://www.illumos.org/issues/3035
[2] http://code.google.com/p/lz4/source/list

Obtained from:	Illumos (13921:9d721847e469)
Tested on:	FreeBSD/amd64
MFC after:	1 month
2013-02-09 06:39:28 +00:00
Xin LI
18c573e000 Fix LINT build for ARM. 2013-02-09 06:31:22 +00:00
Xin LI
3b9ba32d88 Fix LINT build on amd64. 2013-02-09 04:13:45 +00:00
Adrian Chadd
650da23095 The encryption type field needs to be preserved for each descriptor
making up a frame, in both a sub-frame and for all frames in an
aggregate.

Tested:

* AR5416, STA mode
2013-02-09 02:42:01 +00:00
Navdeep Parhar
bf3db9ebd8 Do not hold locks around hardware context reads.
MFC after:	3 days
2013-02-09 00:35:28 +00:00
Hans Petter Selasky
c3400997c4 Correctly list the usbloader dependencies. 2013-02-08 23:13:46 +00:00
Hans Petter Selasky
056e0b5dcd Make sure we don't leak command buffers when a USB
command transfer fails.

MFC after:	1 week
Reported by:	Ian FREISLICH
2013-02-08 22:51:09 +00:00
Xin LI
08c49d655e In r246282 the KTR_ENTRIES was specified with syntax error, fix it so 'make
universe' would work.

MFC after:	12 days
X-MFC-with:	r246282
2013-02-08 22:41:48 +00:00
Hans Petter Selasky
2d8a425be0 Fix regression issue after r244503:
Correct init order to fix a NULL pointer access.

MFC after:	1 week
Reported by:	Ian FREISLICH
2013-02-08 21:15:47 +00:00