dd(1): Enable access to SIZE_T_MAX character devices
On machines where SIZE_T_MAX exceeds OFF_MAX (signed 64-bit), permit seeking
character devices to negative off_t values. This enables dd(1) to interact
with kernel KVA in /dev/kmem on amd64, for example.
Alexander Motin [Sat, 9 Jul 2016 14:59:16 +0000 (14:59 +0000)]
Improve checksum "offload" support.
For compatibility reasons make driver not report any checksum offload by
default, since there is indeed none. But if administrator knows that
interface is used only for local traffic, he can enable fake checksum
offload manually on both sides to save some CPU cycles, since the data
are already protected by CRC32 of PCIe link.
pf: Map hook returns onto the correct error values
pf returns PF_PASS, PF_DROP, ... in the netpfil hooks, but the hook callers
expect to get E<foo> error codes.
Map the returns values. A pass is 0 (everything is OK), anything else means
pf ate the packet, so return EACCES, which tells the stack not to emit an ICMP
error message.
Alexander Motin [Sat, 9 Jul 2016 11:57:21 +0000 (11:57 +0000)]
Reimplement doorbell register emulation for NTB_SB01BASE_LOCKUP.
This allows at least first three doorbells to work very close to normal
hardware, properly signaling events to upper layers without spurious or
lost events. Doorbells above the first three may still report spurious
events due to lack of reliable information, but they are rarely used.
Alexander Motin [Sat, 9 Jul 2016 11:47:52 +0000 (11:47 +0000)]
Switch ctx_lock from mutex to rmlock.
It is odd idea to serialize different MSI-X vectors. Use of rmlocks
here allows them to execute in parallel, but still protects ctx.
If upper layers require any additional serialization -- they can
do it by themselves.
Alexander Motin [Sat, 9 Jul 2016 11:20:42 +0000 (11:20 +0000)]
NewBus'ify NTB subsystem.
This follows NTB subsystem modularization in Linux, tuning it to FreeBSD
native NewBus interfaces. This change allows to support different types
of hardware with different drivers, support multiple NTB instances in a
system, ntb_transport module use for needs other then if_ntb, etc.
Alexander Motin [Sat, 9 Jul 2016 09:34:24 +0000 (09:34 +0000)]
Fix NTB_SDOORBELL_LOCKUP workaround.
Since SBARxSZ register can be write-once, it can be unusable for disabling
the SBAR. For such case also set SBARxBASE to zero to not intersect with
config BAR.
On FreeBSD there is a setsockopt option SO_USER_COOKIE which allows
setting a 32 bit value on each socket. This can be used by applications
and DTrace as a rendezvous point so that an applicaton's data can
more easily be captured at run time. Expose the user cookie via
DTrace by updating the translator in tcp.d and add a quick test
program, a TCP server, that sets the cookie on each connection
accepted.
Adrian Chadd [Fri, 8 Jul 2016 22:59:15 +0000 (22:59 +0000)]
[ath_hal] add in locationing timer support and capabilities.
* add support to read the timer and capability
* add support to enable/disable the location timer.
On AR9380 at least, enabling the location timer is required to make
the timer tick, otherwise location packets return a timestamp of 0.
However, it then makes /all/ RX packets use the RX location timestamp
instead of the TSF timestamp.
So, unless I find another magical way to do location timestamping,
we will have to dynamically switch things on/off and ensure the
TX/RX path handles the "different" timestamps correctly.
Adrian Chadd [Fri, 8 Jul 2016 22:39:37 +0000 (22:39 +0000)]
[ath_hal] add locationing / tx-beamforming RX status descriptor decoding.
* LOC_INFO is mostly just "did this packet come with a locationing
timestamp instead of TSF";
* Decode not-sounding, uploaded-data, data-valid, data type and
number of extension spatial streams.
Adrian Chadd [Fri, 8 Jul 2016 22:37:52 +0000 (22:37 +0000)]
[ath_hal] Add setting positioning bit and clear sounding-disable bit.
* If fast_ts is set then the TX timestamp is the fast timestamp, not
normal TSF.
* If the TX descriptor has the position bit set then request locationing
and clear sounding-disable. This way we (a) get the response with
the TX timestamp from the location side of things, and (b) we get
a CSI dump of the response ACK, which we will eventually use in the
locationing path.
Adrian Chadd [Fri, 8 Jul 2016 21:34:39 +0000 (21:34 +0000)]
[ath_hal] retire a "long RX desc" flag, store/use the TX/RX timestamp length.
* the code already stored the length of the RX desc, which I never used.
So, use that and retire the new flag I introduced a while ago.
* Introduce a TX timestamp length field and capability.
Ed Schouten [Fri, 8 Jul 2016 20:09:21 +0000 (20:09 +0000)]
Don't forget to set sa->narg for CloudABI system calls.
It turns out that this value is not used within the system call code
under normal conditions, except when using tracing tools like ktrace.
If we forget to set this value, it is set to random garbage. This may
cause ktrace to hang indefinitely, making it impossible to kill.
Reported by: Michael Plass
PR: 210800
MFC before: 11.0-RELEASE
Adrian Chadd [Fri, 8 Jul 2016 19:16:50 +0000 (19:16 +0000)]
[ath_hal] extend the TX/RX descriptor layout to include location/beamforming fields.
* extend the TX timestamp to 32 bits, as the AR5416 and later does a full
32 bit TX timestamp instead of 15 or 16 bits.
* add RX descriptor fields for PHY uploaded information (coming soon)
* add flags for RX/TX fast timestamp, hardware upload, etc
* add a flag for TX to request ToD/ToA location information.
Enji Cooper [Fri, 8 Jul 2016 16:29:45 +0000 (16:29 +0000)]
Revert r302403
lang/gcc{48,49,5} lacks -fformat-extensions support (causing build errors, which
is what prompted r302403 to be committed). devel/amd64-gcc on the other hand
(which is used by Jenkins), has the support.
This fixes the Jenkins failure emails due to excessive warnings being produced
with "make buildkernel".
Alexander Motin [Thu, 7 Jul 2016 22:10:10 +0000 (22:10 +0000)]
Fix ahci(4) driver attach to controller with 32 ports.
Incorrect sign expansion in variables that supposed to be a bit fields
caused infinite loop. Fixing this allows system properly detect maximal
possible 32 devices configured on AHCI HBA of BHyVe. That case did not
happen in a wild before due to lack of hardware AHCI HBAs with 32 ports.
fcntl(2): Document interrupt/restart for file locks.
Since r302216, thread suspension causes advisory file locks to restart
(instead of continuing to wait) and for a long time SA_RESTART has
affected advisory file locks. These are both not compliant to POSIX.1.
To clarify that restarting means something, add a paragraph about fair
queuing. Note that the network lock manager does not implement fair
queuing.
Reviewed by: kib (previous version)
Approved by: re (gjb)
Alan Cox [Thu, 7 Jul 2016 20:58:16 +0000 (20:58 +0000)]
Change the type of the map entry's next_read field from a vm_pindex_t to a
vm_offset_t. (This field is used to detect sequential access to the virtual
address range represented by the map entry.) There are three reasons to
make this change. First, a vm_offset_t is smaller on 32-bit architectures.
Consequently, a struct vm_map_entry is now smaller on 32-bit architectures.
Second, a vm_offset_t can be written atomically, whereas it may not be
possible to write a vm_pindex_t atomically on a 32-bit architecture. Third,
using a vm_pindex_t makes the next_read field dependent on which object in
the shadow chain is being read from.
Sometimes the software loses the race when appending more descriptors to
the tx ring and the tx queue stops.
This commit detects this condition and restart the tx queue whenever it stall.
Tested by: sobomax@, Keith White <kwhite@site.uottawa.ca>,
Paul Mather <paul@gromit.dlib.vt.edu>
Sponsored by: Rubicon Communications (Netgate)
Approved by: re (kib)
Colin Percival [Thu, 7 Jul 2016 18:37:12 +0000 (18:37 +0000)]
Autotune the number of pages set aside for UMA startup based on the number
of CPUs present. On amd64 this unbreaks the boot for systems with 92 or
more CPUs; the limit will vary on other systems depending on the size of
their uma_zone and uma_cache structures.
The major consumer of pages during UMA startup is the 19 zone structures
which are set up before UMA has bootstrapped itself sufficiently to use
the rest of the available memory: UMA Slabs, UMA Hash, 4 / 6 / 8 / 12 /
16 / 32 / 64 / 128 / 256 Bucket, vmem btag, VM OBJECT, RADIX NODE, MAP,
KMAP ENTRY, MAP ENTRY, VMSPACE, and fakepg. If the zone structures occupy
more than one page, they will not share pages and the number of pages
currently needed for startup is 19 * pages_per_zone + N, where N is the
number of pages used for allocating other structures; on amd64 N = 3 at
present (2 pages are allocated for UMA Kegs, and one page for UMA Hash).
This patch adds a new definition UMA_BOOT_PAGES_ZONES, currently set to 32,
and if a zone structure does not fit into a single page sets boot_pages to
UMA_BOOT_PAGES_ZONES * pages_per_zone instead of UMA_BOOT_PAGES (which
remains at 64). Consequently this patch has no effect on systems where the
zone structure fits into 2 or fewer pages (on amd64, 59 or fewer CPUs), but
increases boot_pages sufficiently on systems where the large number of CPUs
makes this structure larger. It seems safe to assume that systems with 60+
CPUs can afford to set aside an additional 128kB of memory per 32 CPUs.
The vm.boot_pages tunable continues to override this computation, but is
unlikely to be necessary in the future.
Tested on: EC2 x1.32xlarge
Relnotes: FreeBSD can now boot on 92+ CPU systems without requiring
vm.boot_pages to be manually adjusted.
Reviewed by: jeff, alc, adrian
Approved by: re (kib)
Add new unmount(2) flag, MNT_NONBUSY, to check whether there are
any open vnodes before proceeding. Make autounmound(8) use this flag.
Without it, even an unsuccessfull unmount causes filesystem flush,
which interferes with normal operation.
Steven Hartland [Wed, 6 Jul 2016 23:49:19 +0000 (23:49 +0000)]
Fix ZFS ARC min / max tunable
Due to ARC initial configuration not being done and kmem information
not being available we need to blindly set zfs_arc_max and zfs_arc_min
when configured via the tunable.
This fixes vfs.zfs.arc_(min|max) configuration via loader.conf broken by
r302265.
Andrew Turner [Wed, 6 Jul 2016 16:20:10 +0000 (16:20 +0000)]
Remove the old pre-INTRNG arm64 interrupt framework. GENERIC was switched
to INTRNG in r301565 with the old code no longer being built by default with
no reports of issues on any supported hardware.
Approved by: re (gjb)
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
The TCPPCAP debugging feature caches recently-used mbufs for use in
debugging TCP connections. This commit provides a mechanism to free those
mbufs when the system is under memory pressure.
Because this will result in lost debugging information, the behavior is
controllable by a sysctl. The default setting is to free the mbufs.
Enji Cooper [Wed, 6 Jul 2016 16:02:15 +0000 (16:02 +0000)]
Fix CTASSERT issue in a more clean way
- Replace all CTASSERT macro instances with static_assert's.
- Remove the WRAPPED_CTASSERT macro; it's now an unnecessary obfuscation.
- Localize all static_assert's to the structures being tested.
- Sort some headers per-style(9).
Replace a number of conflations of mp_ncpus and mp_maxid with either
mp_maxid or CPU_FOREACH() as appropriate. This fixes a number of places in
the kernel that assumed CPU IDs are dense in [0, mp_ncpus) and would try,
for example, to run tasks on CPUs that did not exist or to allocate too
few buffers on systems with sparse CPU IDs in which there are holes in the
range and mp_maxid > mp_ncpus. Such circumstances generally occur on
systems with SMT, but on which SMT is disabled. This patch restores system
operation at least on POWER8 systems configured in this way.
There are a number of other places in the kernel with potential problems
in these situations, but where sparse CPU IDs are not currently known
to occur, mostly in the ARM machine-dependent code. These will be fixed
in a follow-up commit after the stable/11 branch.
PR: kern/210106
Reviewed by: jhb
Approved by: re (glebius)
Only set the ipfilter running state to 'not running' if we are
doing the teardown. ipf_destroy_all() may free ipfmain in case
of ipf_dynamic_softc being true, thus we are avoiding a possible
memory modified after free as well.
Reported by: Coverity
Coverity CID: 1357320
Approved by: re (hrs)
MFC after: 10 days
Enji Cooper [Wed, 6 Jul 2016 05:17:07 +0000 (05:17 +0000)]
Fix gcc build errors with SSE 4.2 detection and gcc warnings
- Remove -Wunused-but-set-variable's (`len`, etc).
- Replace clang-specific tests in sse42_supported(..) with generic,
FreeBSD-supported CPU feature tests, using macros and functions
from machine/cpufunc.h and machine/specialreg.h . The previous method
for determining SSE4.2 availability was only compatible with clang.
- Sort #includes per style(9).
Enji Cooper [Wed, 6 Jul 2016 05:09:13 +0000 (05:09 +0000)]
Fix gcc warnings
- Put parentheses around bitwise OR'ed values in the `FIELD_COPY(..)` and
`FIELD_REPLACE(..)` macros to mute warning from gcc 4.2.1.
- Remove -Wunused-but-set-variable's (`setup_addr`, `status_addr`).
Enji Cooper [Wed, 6 Jul 2016 05:02:59 +0000 (05:02 +0000)]
Fix gcc warnings
Add `WRAPPED_CTASSERT` macro by annotating CTASSERTs with __unused
to deal with -Wunused-local-typedefs warnings from gcc 4.8+.
All other compilers (clang, etc) use CTASSERT as-is. A more generic
solution for this issue will be proposed after ^/stable/11 is forked.
Consolidate all CTASSERTs under one block instead of inlining them in
functions.
Remove incorrect attributes from posix_memalign(3) declaration.
Both __alloc_align and __alloc_size can't be used when the function
returns a pointer to memory. This fixes breakage when building with
clang 3.4:
In file included from /usr/src/svn/usr.sbin/bhyve/atkbdc.c:40:
/usr/include/stdlib.h:176:6: error: '__alloc_size__' attribute only
applies to functions that return a pointer [-Werror,-Wignored-attributes]
projects / FreeBSD / FreeBSD.git / log