marius [Tue, 29 Dec 2015 17:07:28 +0000 (17:07 +0000)]
- Add entries for the more prominent members of the Digi International
Neo series, which are based on Exar PCI chips.
- Mark some unused parameters as such.
- Fix style
theraven [Tue, 29 Dec 2015 16:29:42 +0000 (16:29 +0000)]
Improvements to BSD-licensed DTC.
- Added an expression parser so that expressions from headers are now working
- Fixed missing null terminators on cross references
- Disabled exceptions / RTTI in the build for smaller binaries
- Changed phandle order generation to be identical to GPL'd dtc
ian [Tue, 29 Dec 2015 15:23:03 +0000 (15:23 +0000)]
Correct the code for sign-extending a 16 bit value. As near as I can tell
this is effectively a no-op -- the addend term in MOVT/MOVW relocations
always seems to be zero. But this is correct and the old code wasn't.
delphij [Tue, 29 Dec 2015 08:19:43 +0000 (08:19 +0000)]
hyperv: vmbus: run non-blocking message handlers in vmbus_msg_swintr()
We'll remove the per-channel control_work_queue because it can't properly
do serialization of message handling, e.g., when there are 2 NIC devices,
vmbus_channel_on_offer() -> hv_queue_work_item() has a race condition:
for an SMP VM, vmbus_channel_process_offer() can run concurrently on
different CPUs and if the second NIC's
vmbus_channel_process_offer() -> hv_vmbus_child_device_register() runs
first, the second NIC's name will be hn0 and the first NIC's name will
be hn1!
We can fix the race condition by removing the per-channel control_work_queue
and run all the message handlers in the global
hv_vmbus_g_connection.work_queue -- we'll do this in the next patch.
With the coming next patch, we have to run the non-blocking handlers
directly in the kernel thread vmbus_msg_swintr(), because the special
handling of sub-channel: when a sub-channel (e.g., of the storvsc driver)
is received and being handled in vmbus_channel_on_offer() running on the
global hv_vmbus_g_connection.work_queue, vmbus_channel_process_offer()
invokes channel->sc_creation_callback, i.e., storvsc_handle_sc_creation,
and the callback will invoke hv_vmbus_channel_open() -> hv_vmbus_post_message
and expect a further reply from the host, but the handling of the further
messag can't be done because the current message's handling hasn't finished
yet; as result, hv_vmbus_channel_open() -> sema_timedwait() will time out
and th device can't work.
Also renamed the handler type from hv_pfn_channel_msg_handler to
vmbus_msg_handler: the 'pfn' and 'channel' in the old name make no sense.
Submitted by: Dexuan Cui <decui microsoft com>
Reviewed by: royger
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D4596
marius [Mon, 28 Dec 2015 20:24:08 +0000 (20:24 +0000)]
- Add an entry for the SIIG Cyber 2SP1 PCIe adapter, which is based
on an Oxford Semiconductor OX16PCI954 but uses only two ports and
a non-default clock rate.
- Fix style/whitespace
hselasky [Mon, 28 Dec 2015 18:50:18 +0000 (18:50 +0000)]
Add support for CQE zipping. CQE zipping reduces PCI overhead by
coalescing and zipping multiple CQEs into a single merged CQE. The
feature is enabled by default and can be disabled by a sysctl.
Implementing this feature mlx5_cqwq_pop() has been separated from
mlx5e_get_cqe().
wollman [Mon, 28 Dec 2015 18:29:47 +0000 (18:29 +0000)]
in6_if2idlen: treat bridge(4) interfaces like other Ethernet interfaces
bridge(4) interfaces have an if_type of IFT_BRIDGE, rather than
IFT_ETHER, even though they only support Ethernet-style links. This
caused in6_if2idlen to emit an "unknown link type (209)" warning to
the console every time it was called. Add IFT_BRIDGE to the case
statement in the appropriate place, indicating that it uses the same
IPv6 address format as other Ethernet-like interfaces.
hselasky [Mon, 28 Dec 2015 18:20:05 +0000 (18:20 +0000)]
Reduce memory consumption when allocating kobject strings in the
LinuxKPI. Compute string length before allocating memory instead of
using fixed size allocations. Make kobject_set_name_vargs() global
instead of inline to save some bytes when compiling.
araujo [Mon, 28 Dec 2015 05:48:22 +0000 (05:48 +0000)]
The sdp opens the database with PERM_SECURE mode and it is different than
dp that opens the database with PERM_INSECURE, so we need to check sdp->put
against sdp instead of use dp->put.
ngie [Mon, 28 Dec 2015 02:34:59 +0000 (02:34 +0000)]
- Remove unused but set ssize in shutdown_send_sigpipe
- Add #ifdef TEST_SEQ_PACKET_SOURCE_ADDRESS` for untestable code
because FreeBSD doesn't have a means to map source addresses for
SEQ_PACKET AF_UNIX sockets (paraphrased). Put pathname variable
under the #ifdef to mute another unused but set variable warning
ngie [Mon, 28 Dec 2015 02:18:22 +0000 (02:18 +0000)]
Fix style(9) a bit and ensure that error from initializing kqueue(2) is
sane
- Push the kqueue(2) initialization down so the errno will correspond with
the failure instead of potentially being stomped on by functions called
by `PLAIN_REQUIRE_KERNEL_MODULE`
- Delete trailing whitespace
- Add spaces between braces for conditional and control blocks (for/if)
- Use err/errx instead of perror+printf+exit/printf+exit.
- Remove braces for single-line conditionals
ngie [Mon, 28 Dec 2015 00:53:37 +0000 (00:53 +0000)]
- Explicitly initialize ch to 0
- Delete some spurious whitespace
- Use calloc instead of malloc in the last test to ensure that
sendspace is properly zero'ed out
imp [Sun, 27 Dec 2015 23:04:12 +0000 (23:04 +0000)]
Use a macro to create the names for the library path names. This will
allow later substitution at run time instead of compile time of the
environment variable name prefix.
imp [Sun, 27 Dec 2015 23:04:11 +0000 (23:04 +0000)]
The FILE structure has a mbstate_t in it. This structure needs to be
aligned on a int64_t boundary. However, when we allocate the array of
these structures, we use ALIGNBYTES which defaults to sizeof(int) on
arm, i386 and others. The i386 stuff can handle unaligned accesses
seemlessly. However, arm cannot. Take this into account when creating
the array of FILEs, and add some comments about why.
allanjude [Sun, 27 Dec 2015 17:09:23 +0000 (17:09 +0000)]
Fix incorrect error message in geom map
If geom_map fails to find the end of a mapped partition based on a search, it would return the incorrect error message, stating it could not parse the START value
Reviewed by: adrian
Approved by: bapt (mentor)
Sponsored by: ScaleEngine Inc.
Differential Revision: https://reviews.freebsd.org/D4187
dchagin [Sun, 27 Dec 2015 15:37:07 +0000 (15:37 +0000)]
Verify that tv_sec value specified in settimeofday() and clock_settime()
(CLOCK_REALTIME case) system calls is non negative.
This commit hides a kernel panic in atrtc_settime() as the clock_ts_to_ct()
does not properly convert negative tv_sec.
ps. in my opinion clock_ts_to_ct() should be rewritten to properly handle
negative tv_sec values.
ed [Sun, 27 Dec 2015 07:50:11 +0000 (07:50 +0000)]
Replace implementation of hsearch() by one that scales.
Traditionally the hcreate() function creates a hash table that uses
chaining, using a fixed user-provided size. The problem with this
approach is that this often either wastes memory (table too big) or
yields bad performance (table too small). For applications it may not
always be easy to estimate the right hash table size. A fixed number
only increases performance compared to a linked list by a constant
factor.
This problem can be solved easily by dynamically resizing the hash
table. If the size of the hash table is at least doubled, this has no
negative on the running time complexity. If a dynamically sized hash
table is used, we can also switch to using open addressing instead of
chaining, which has the advantage of just using a single allocation for
the entire table, instead of allocating many small objects.
Finally, a problem with the existing implementation is that its
deterministic algorithm for hashing makes it possible to come up with
fixed patterns to trigger an excessive number of collisions. We can
easily solve this by using FNV-1a as a hashing algorithm in combination
with a randomly generated offset basis.
Measurements have shown that this implementation is about 20-25% faster
than the existing implementation (even if the existing implementation is
given an excessive number of buckets). Though it allocates more memory
through malloc() than the old implementation (between 4-8 pointers per
used entry instead of 3), process memory use is similar to the old
implementation as if the estimated size was underestimated by a factor
10. This is due to the fact that malloc() needs to perform less
bookkeeping.
mav [Sun, 27 Dec 2015 06:28:31 +0000 (06:28 +0000)]
Allocate separate scratch space for scanner purposes.
This space does not require DMA syncing. It reduces lock scope of the DMA
scratch space. It allows whole DMA scratch space to be used to I/O, so now
we can fetch up to ~1000 ports from SNS.
Due to the last fact, increase maximal number of ports from 256 to 1024.
jamie [Sat, 26 Dec 2015 23:01:34 +0000 (23:01 +0000)]
Let old-style (shell-based) jail configuration handle jail names that
contain characters not allowed in a shell variable (such as "-").
These will be replaced by an underscore in jail config variables,
e.g. for jail "foo-bar" you would set "jail_foo_bar_hostname".
This is separate from the current code that changes the jail names
if they contain "." or "/". It also doesn't apply to jails defined
in a jail.conf file.
adrian [Sat, 26 Dec 2015 19:14:24 +0000 (19:14 +0000)]
Disable the firwmare licence check for rsu and urtwn.
The licence grant says something exactly the same as the atheros patent
grant, which is "As long as you use this firmware on our chips, everything
is totally okay." Now, I'm pretty sure if that we /have/ to have this,
we're going to have to have it for every other firmware for every other
device in the tree.
So, I'll flip this off in -HEAD for now so people stop asking about
why rsu/urtwn don't work out of the box, and I'll kick off a larger
discussion about this in the new year.
ian [Sat, 26 Dec 2015 18:21:32 +0000 (18:21 +0000)]
Enhance rc.d/netwait script to wait for late-attaching interfaces such as
USB NICs.
USB network hardware may not be enumerated and available when the rc.d
networking scripts run. Eventually the USB attachment completes and devd
events cause the network initialization to happen, but by then other rc.d
scripts have already failed, because services which depend on NETWORKING
(such as mountcritremote) may end up running before the network is actually
ready.
There is an existing netwait script, but because it is dependent on
NETWORKING it runs too late to prevent failure of some other rc
scripts. This change flips the order so that NETWORKING depends on netwait,
and netwait now depends on devd and routing (the former is needed to make
interfaces appear, and the latter is needed to run the ping tests in
netwait).
The netwait script used to be oriented primarily towards "as soon as any
host is reachable the network is fully functional", so you gave it a list of
IPs to try and you could optionally name an interface and it would wait for
carrier on that interface. That functionality still works the same, but now
you can provide a list of interfaces to wait for and it waits until each one
of them is available. The ping logic still completes as soon as the first IP
on the list responds.
These changes were submitted by Brenden Molloy <brendan+freebsd@bbqsrc.net>
in PR 205186, and lightly modified by me to allow a list of interfaces
instead of just one.
adrian [Sat, 26 Dec 2015 17:22:02 +0000 (17:22 +0000)]
[mdio] Add MDIO support for "extended" registers as defined by IEEE 802.3 Clause 45.
IEEE 802.3 Clause 45 added backwards-compatible support for 2^16 PHY registers
through the addition of an additional device address frame.
Clause 45 addressing is used in 10Gbe PHYs, 802.3az EEE registers, etc. It may
make sense to provide a similar extension to the miibus interface, but I've
refrained from unilaterally doing so here.
np [Sat, 26 Dec 2015 06:05:21 +0000 (06:05 +0000)]
cxgbei: Hardware accelerated iSCSI target and initiator for TOE capable
cards supported by cxgbe(4).
On the host side this driver interfaces with the storage stack via the
ICL (iSCSI Common Layer) in the kernel. On the wire the traffic is
standard iSCSI (SCSI over TCP as per RFC 3720/7143 etc.) that
interoperates with all other standards compliant implementations. The
driver is layered on top of the TOE driver (t4_tom) and promotes
connections being handled by t4_tom to iSCSI ULP (Upper Layer Protocol)
mode. Hardware assistance in this mode includes:
- Full TCP processing.
- iSCSI PDU identification and recovery within the TCP stream.
- Header and/or data digest insertion (tx) and verification (rx).
- Zero copy (both tx and rx).
Man page will follow in a separate commit in a couple of weeks.
mav [Sat, 26 Dec 2015 04:26:32 +0000 (04:26 +0000)]
Make virtual ports control asynchronous.
Before this change virtual ports control IOCBs were executed synchronously
via Execute IOCB mailbox command. It required exclusive use of scratch
space of driver and mailbox registers of the hardware. Because of that
shared resources use this code could not really sleep, having to spin for
completion, blocking any other operation.
This change introduces new asynchronous design, sending the IOCBs directly
on request queue and gracefully waiting for their return on response queue.
Returned IOCBs are identified with unified handle space from r292725.
adrian [Sat, 26 Dec 2015 02:31:39 +0000 (02:31 +0000)]
[mdio] migrate mdiobus out of etherswitch and into a top-level device of its own.
The mdio driver interface is generally useful for devices that require
MDIO without the full MII bus interface. This lifts the driver/interface
out of etherswitch(4), and adds a mdio(4) man page.
tuexen [Fri, 25 Dec 2015 18:11:40 +0000 (18:11 +0000)]
Don't implicitly terminate a user message when moving it to the
send_queue and the socket is closed. This results in strange
race conditions for the application.
While there, remove a stray character.
mav [Fri, 25 Dec 2015 13:03:18 +0000 (13:03 +0000)]
Unify handles allocation for initiator and target IOCBs.
I am not sure why this was split long ago, but I see no reason for it.
At this point this unification just slightly reduces memory usage, but
as next step I plan to reuse shared handle space for other IOCB types.
adrian [Thu, 24 Dec 2015 18:40:10 +0000 (18:40 +0000)]
[rt305x] Add initial RT5350 and MT7620 glue.
* Add in chipset awareness to the obio bus layout (ie, which devices are
where);
* Add in some USB OTG changes to be aware of the newer stuff;
* Add in a configurable primary console - some chips use the normal UART,
some use UARTLITE.
Tested (by Stanislav);
* RT3050 (NFS)
* RT5350 (NFS, MFS)
* MT7620 (USB)
Submitted by: Stanislav Galabov <sgalabov@gmail.com>
projects / FreeBSD / FreeBSD.git / log