hselasky [Wed, 13 Feb 2013 12:35:17 +0000 (12:35 +0000)]
Resolve a LOR after r246616. Protect control requests using the USB device
enumeration lock. Make sure all callers of usbd_enum_lock() check the return
value. Remove the control transfer specific lock. Bump the FreeBSD version
number, hence external USB modules may need to be recompiled due to a USB
device structure change.
adrian [Wed, 13 Feb 2013 05:32:19 +0000 (05:32 +0000)]
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.
marcel [Tue, 12 Feb 2013 17:38:35 +0000 (17:38 +0000)]
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.
kib [Tue, 12 Feb 2013 16:57:20 +0000 (16:57 +0000)]
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>)
pluknet [Tue, 12 Feb 2013 13:01:34 +0000 (13:01 +0000)]
o Bring in sync decoding the first nfssvc(2) parameter (flags) with
the current definitions location.
o Respect numbers in NFSSVC_* (e.g. NFSSVC_V4ROOTEXPORT).
glebius [Tue, 12 Feb 2013 11:24:37 +0000 (11:24 +0000)]
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.
andrew [Tue, 12 Feb 2013 06:04:51 +0000 (06:04 +0000)]
When clang builds libc it may insert calls to __aeabi_* functions. Normally
this is not a problem as they are resolved by libgcc. The exception is for
the __aeabi_mem* functions. These call back into libc to the appropriate
function. This causes issues for static binaries as we only link against
libc once so there is no way for it to call into libgcc and back.
The fix for this is to include these symbols in libc but keep them hidden
so binaries use the libgcc version.
andrew [Tue, 12 Feb 2013 05:56:00 +0000 (05:56 +0000)]
Allow us to build clang for ARM EABI. Clang and llvm use the
arm-gnueabi-freebsd10.0 triple for EABI. Use this when we are on arm or
armv6 and are building for EABI.
neel [Mon, 11 Feb 2013 20:36:07 +0000 (20:36 +0000)]
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.
tuexen [Mon, 11 Feb 2013 13:57:03 +0000 (13:57 +0000)]
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.
pluknet [Mon, 11 Feb 2013 12:56:23 +0000 (12:56 +0000)]
Major update for unix_cmsg from Andrey Simonenko.
Quoting the submitter:
- Added tests for SCM_BINTIME, LOCAL_PEERCRED, cmsghdr.cmsg_len
- Code that checks correctness of groups was corrected (getgroups(2) change)
- unix_cmsg.c was completely redesigned and simplified
- Use less timeout value in unix_cmsg.c for faster work
- Added support for not sending data in a message, not sending data and
data array associated with a cmsghdr structure in a message
- Existent tests were improved
- unix_cmsg.t was redesigned and simplified
Correctness of unix_cmsg verified on 7.1-STABLE, 9.1-STABLE and 10-CURRENT.
gonzo [Mon, 11 Feb 2013 11:33:56 +0000 (11:33 +0000)]
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.
glebius [Mon, 11 Feb 2013 10:58:22 +0000 (10:58 +0000)]
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.
brooks [Mon, 11 Feb 2013 09:45:24 +0000 (09:45 +0000)]
Add -lheimntlm to LDADD directly.
With the current binutils, symbols from libheimtlm.so are loaded because
it is referenced by DT_NEEDED. This feature is not implemented in
mclinker (https://code.google.com/p/mclinker/issues/detail?id=104).
I encountered the same issue when linking with a recent devel/binutils
invoked via clang. This was the only use of DT_NEEDED in the tree so
removing it simplifies toolchain requirements.
Submitted by: Pete Chou <petechou@gmail.com> (mclinker issue)
adrian [Mon, 11 Feb 2013 07:48:26 +0000 (07:48 +0000)]
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)
jilles [Sun, 10 Feb 2013 23:09:15 +0000 (23:09 +0000)]
fts: Use O_DIRECTORY when opening name that might be changed by attacker.
There are uncommon cases where fts_safe_changedir() may be called with a
non-NULL name that is not "..". Do not block or worse if an attacker put (a
(symlink to) a fifo or device where a directory used to be.
jilles [Sun, 10 Feb 2013 18:56:37 +0000 (18:56 +0000)]
find: Run when cwd cannot be opened, except with -execdir or -delete.
fts(3) can run (albeit more slowly and imposing the {PATH_MAX} limit) when
the current directory cannot be opened. Therefore, do not make a failure to
open the current directory (for returning to it later in -exec) fatal.
If -execdir or -delete are used, the expectation is that fts(3) will use
chdir to avoid race conditions (except for -execdir with -L). Do not break
this expectation any more than it already is by still failing if the current
directory cannot be opened.
cperciva [Sun, 10 Feb 2013 17:58:44 +0000 (17:58 +0000)]
Don't try to suppress the inclusion of the build date in named's version
string by undefining __DATE__, since (unlike gcc) clang doesn't allow us
to do that. Instead, define NO_VERSION_DATE, which was helpfully added
to the named source code for exactly this purpose.
jilles [Sun, 10 Feb 2013 13:28:02 +0000 (13:28 +0000)]
find: In -execdir ... {} +, only pass one file per invocation.
This is inefficient but ensures that -execdir ... {} + does not mix files
from different directories in one invocation; the command could not access
some files. Files from the same directory should really be handled in one
invocation but this is somewhat more complicated.
hselasky [Sun, 10 Feb 2013 10:56:13 +0000 (10:56 +0000)]
- 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.
hselasky [Sun, 10 Feb 2013 10:36:16 +0000 (10:36 +0000)]
- 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.
kib [Sun, 10 Feb 2013 10:17:33 +0000 (10:17 +0000)]
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
attilio [Sun, 10 Feb 2013 01:04:10 +0000 (01:04 +0000)]
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.
brooks [Sat, 9 Feb 2013 23:17:28 +0000 (23:17 +0000)]
Add nmtree to ITOOLS if it is installed on the host instead of keying off
the BOOTSTRAPPING variable. The previous test was wrong because
BOOTSTRAPPING is 0 in most cases.
delphij [Sat, 9 Feb 2013 06:39:28 +0000 (06:39 +0000)]
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.
pfg [Fri, 8 Feb 2013 20:58:00 +0000 (20:58 +0000)]
ext2fs: make e2fs_maxcontig local and remove tautological check.
e2fs_maxcontig was modelled after UFS when bringing the
"Orlov allocator" to ext2. On UFS fs_maxcontig is kept in the
superblock and is used by userland tools (fsck and growfs),
In ext2 this information is volatile so it is not available
for userland tools, so in this case it doesn't have sense
to carry it in the in-memory superblock.
pfg [Fri, 8 Feb 2013 19:39:15 +0000 (19:39 +0000)]
patch: Follow original versioning convention.
According to the README file [1] the 12u variant, unlike
the 12g variant, contains no copyleft code. It is therefore
convenient to keep using the original versioning scheme to
prevent confusions.
des [Fri, 8 Feb 2013 14:14:00 +0000 (14:14 +0000)]
Print a warning if not setuid root.
Document the need for the setuid bit and how to set it.
Explain why it isn't set by default, and suggest simply adding users
to groups instead.
dteske [Fri, 8 Feb 2013 11:14:01 +0000 (11:14 +0000)]
Remove NO_OBJ from Makefiles that generate manuals because this causes the
GZIP compressed manuals to appear in ./src instead of the appropriate obj dir.