up to now.
The new sendfile is the code that Netflix uses to send their multiple tens
of gigabits of data per second. The new implementation features asynchronous
I/O, when I/O operations are launched, but not awaited to be complete. An
explanation of why such behavior is beneficial compared to old one is
going to be too long for a commit message, so we will skip it here.
Additional features of new syscall are extra flags, which provide an
application more control over data sent. The SF_NOCACHE flag tells
kernel that data shouldn't be cached after it was sent. The SF_READAHEAD()
macro allows to specify readahead size in pages.
The new syscalls is a drop in replacement. No modifications are required
to applications. One can take nginx binary for stable/10 and run it
successfully on head. Although SF_NODISKIO lost its original sense, as now
sendfile doesn't block, and now means something completely different (tm),
using the new sendfile the old way is absolutely safe.
Celebrates: Netflix global launch!
Sponsored by: Nginx, Inc.
Sponsored by: Netflix
Relnotes: yes
ioat_acquire_reserve() is an extended version of ioat_acquire(). It
allows users to reserve space in the channel for some number of
descriptors. If this succeeds, it guarantees that at least submission
of N valid descriptors will succeed.
Sponsored by: EMC / Isilon Storage Division
Due to FreeBSD system-wide limits on number of MSI-X vectors
(https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=199321),
it may be desirable to allocate fewer than the maximum number
of vectors for an NVMe device, in order to save vectors for
other devices (usually Ethernet) that can take better
advantage of them and may be probed after NVMe.
This tunable is expressed in terms of minimum number of CPUs
per I/O queue instead of max number of queues per controller,
to allow for a more even distribution of CPUs per queue. This
avoids cases where some number of CPUs have a dedicated queue,
but other CPUs need to share queues. Ideally the PR referenced
above will eventually be fixed and the mechanism implemented
here becomes obsolete anyways.
While here, fix a bug in the CPUs per I/O queue calculation to
properly account for the admin queue's MSI-X vector.
Reviewed by: gallatin
MFC after: 3 days
Sponsored by: Intel
This helps immensily with our ability to operate in the Amazon Cloud.
Discussed on Intel Networking Community call this morning.
Submitted by: Jarrod Petz(petz@nisshoko.net)
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D4788
the system is booted and running.
Add PHY detection logic to ixgbe_handle_mod() and add locking to
ixgbe_handle_msf() as well.
PR: 150251
Submitted by: aboyer@averesystems.com
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D3188
e1000/e1000e split in linux.
Split rxbuffer and txbuffer apart to support the new RX descriptor format
structures. Move rxbuffer manipulation to em_setup_rxdesc() to unify the
new behavior changes.
Add a RSSKEYLEN macro for help in generating the RSSKEY data structures
in the card.
Change em_receive_checksum() to process the new rxdescriptor format
status bit.
MFC after: 2 weeks
Sponsored by: Limelight Networks
Differential Revision: https://reviews.freebsd.org/D3447
Previously nvme(4) would revert to a signle I/O queue if it could not
allocate enought interrupt vectors or NVMe submission/completion queues
to have one I/O queue per core. This patch determines how to utilize a
smaller number of available interrupt vectors, and assigns (as closely
as possible) an equal number of cores to each associated I/O queue.
MFC after: 3 days
Sponsored by: Intel
Instead just use num_io_queues to make this determination.
This prepares for some future changes enabling use of multiple
queues when we do not have enough queues or MSI-X vectors
for one queue per CPU.
MFC after: 3 days
Sponsored by: Intel
This significantly improves parallelism in the most common case.
The taskqueue is still used whenever BIO_ORDERED bios are in flight.
This patch is based heavily on a patch from gallatin@.
MFC after: 3 days
Sponsored by: Intel
This ensures the bio flags are not read after biodone().
The ordering will still be enforced, after the bio is
submitted successfully.
MFC after: 3 days
Sponsored by: Intel
Still wait until all in-flight bios (including the ordered bio)
complete before processing more bios from the queue.
MFC after: 3 days
Sponsored by: Intel
and t_maxseg. This dualism emerged with T/TCP, but was not properly cleaned
up after T/TCP removal. After all permutations over the years the result is
that t_maxopd stores a minimum of peer offered MSS and MTU reduced by minimum
protocol header. And t_maxseg stores (t_maxopd - TCPOLEN_TSTAMP_APPA) if
timestamps are in action, or is equal to t_maxopd otherwise. That's a very
rough estimate of MSS reduced by options length. Throughout the code it
was used in places, where preciseness was not important, like cwnd or
ssthresh calculations.
With this change:
- t_maxopd goes away.
- t_maxseg now stores MSS not adjusted by options.
- new function tcp_maxseg() is provided, that calculates MSS reduced by
options length. The functions gives a better estimate, since it takes
into account SACK state as well.
Reviewed by: jtl
Differential Revision: https://reviews.freebsd.org/D3593
This optimization is not proper (and causes kernel panic),
since driver checks fw_status to optimize away parsing stage
if it was already done.
Reported by: dchagin
It was returning a pointer to stack-allocated memory, so make the
allocation at the caller instead.
Found by: clang static analyzer
Coverity: CID 1245774
Reviewed by: ed, rpaulo
Review URL: https://reviews.freebsd.org/D4740
The fd is closed later in this case. This fixes a "SS_NOFDREF on enter"
panic.
Submitted by: Krishnamraju Eraparaju @ Chelsio
Reviewed by: Steve Wise @ Open Grid Computing
- Separate 'firmware_put(sc->fw_fp, FIRMWARE_UNLOAD); sc->fw_fp = NULL;'
into iwn_unload_firmware().
- Move error handling to the end of iwn_read_firmware().
No functional changes.
Approved by: adrian (mentor)
Differential Revision: https://reviews.freebsd.org/D4768
iwn(4) / wpi(4) works in the same way
(read_firmware() -> hw_init() -> firmware_put())
Approved by: adrian (mentor)
Differential Revision: https://reviews.freebsd.org/D4766
- Simplify defragmentation code.
- Use proper number of dma segments for data.
Approved by: adrian (mentor)
Obtained from: DragonFlyBSD (mostly)
Differential Revision: https://reviews.freebsd.org/D4754
stable/10 doesn't have the if_getdrvflags(9) KPI. Reference the field in the
structure directly if the __FreeBSD_version is < 1100022, so the driver can
be built with PCI_IOV support on stable/10, without backporting all of
r266974 (which requires additional changes due to projects/ifnet, etc)
Differential Revision: https://reviews.freebsd.org/D4759
Reviewed by: erj, sbruno
Sponsored by: EMC / Isilon Storage Division
These are going to be much more efficient on low end embedded systems
but unfortunately they make it .. less convenient to implement correct
bus barriers and debugging. They also didn't implement the register
serialisation workaround required for Owl (AR5416.)
So, just remove them for now. Later on I'll just inline the routines
from ah_osdep.c.
- Change order of data in if_iwmvar.h
(like it is in other drivers: defines, data structures,
vap/node structures, softc struct and locks); use indentation.
- Fix IWM_LOCK(_sc) / IWM_UNLOCK(_sc) macro.
- Add IWM_LOCK_INIT / DESTROY(sc) + fix mtx_init() usage.
- Wrap iwm_node casts into IWM_NODE() macro.
- Drop some fields:
* wt_hwqueue from Tx radiotap header;
* macaddr[6] from iwm_vap;
Approved by: adrian
Differential Revision: https://reviews.freebsd.org/D4753
tree parsing opt-out rather than opt-in. All FDT-based systems as well as
PowerPC systems with real Open Firmware use the CHRP-derived binding that
includes it, which makes SPARC the odd man out here. Making it opt-out
avoids astonishment on new platform bring up.
The ath hal and driver code all assume the world is an x86 or the
bus layer does an explicit bus flush after each operation (eg netbsd.)
However, we don't do that.
So, to be "correct" on platforms like sparc64, mips and ppc (and maybe
ARM, I am not sure), just do explicit barriers after each operation.
Now, this does slow things down a tad on embedded platforms but I'd
rather things be "correct" versus "fast." At some later point if someone
wishes it to be fast then we should add the barrier calls to the HAL and
driver.
Tested:
* carambola 2 (AR9331.)
rounding) has better spread. Implement fp16_sin() to go along with
fp16_cos(). In the rendering loop, switch from addition to subtraction
so the center of the pattern will be a trough rather than a peak. This
is completely arbitrary, of course, but looks better to me.
This is a port from openbsd. It's incomplete and unstable, but it's better
than nothing. I have no plans to MFC this until it's complete and stable.
Submitted by: kevlo
