- Add RATELIMIT kernel configuration keyword which must be set to
enable the new functionality.
- Add support for hardware driven, Receive Side Scaling, RSS aware, rate
limited sendqueues and expose the functionality through the already
established SO_MAX_PACING_RATE setsockopt(). The API support rates in
the range from 1 to 4Gbytes/s which are suitable for regular TCP and
UDP streams. The setsockopt(2) manual page has been updated.
- Add rate limit function callback API to "struct ifnet" which supports
the following operations: if_snd_tag_alloc(), if_snd_tag_modify(),
if_snd_tag_query() and if_snd_tag_free().
- Add support to ifconfig to view, set and clear the IFCAP_TXRTLMT
flag, which tells if a network driver supports rate limiting or not.
- This patch also adds support for rate limiting through VLAN and LAGG
intermediate network devices.
- How rate limiting works:
1) The userspace application calls setsockopt() after accepting or
making a new connection to set the rate which is then stored in the
socket structure in the kernel. Later on when packets are transmitted
a check is made in the transmit path for rate changes. A rate change
implies a non-blocking ifp->if_snd_tag_alloc() call will be made to the
destination network interface, which then sets up a custom sendqueue
with the given rate limitation parameter. A "struct m_snd_tag" pointer is
returned which serves as a "snd_tag" hint in the m_pkthdr for the
subsequently transmitted mbufs.
2) When the network driver sees the "m->m_pkthdr.snd_tag" different
from NULL, it will move the packets into a designated rate limited sendqueue
given by the snd_tag pointer. It is up to the individual drivers how the rate
limited traffic will be rate limited.
3) Route changes are detected by the NIC drivers in the ifp->if_transmit()
routine when the ifnet pointer in the incoming snd_tag mismatches the
one of the network interface. The network adapter frees the mbuf and
returns EAGAIN which causes the ip_output() to release and clear the send
tag. Upon next ip_output() a new "snd_tag" will be tried allocated.
4) When the PCB is detached the custom sendqueue will be released by a
non-blocking ifp->if_snd_tag_free() call to the currently bound network
interface.
Reviewed by: wblock (manpages), adrian, gallatin, scottl (network)
Differential Revision: https://reviews.freebsd.org/D3687
Sponsored by: Mellanox Technologies
MFC after: 3 months
sources to return timestamps when SO_TIMESTAMP is enabled. Two additional
clock sources are:
o nanosecond resolution realtime clock (equivalent of CLOCK_REALTIME);
o nanosecond resolution monotonic clock (equivalent of CLOCK_MONOTONIC).
In addition to this, this option provides unified interface to get bintime
(equivalent of using SO_BINTIME), except it also supported with IPv6 where
SO_BINTIME has never been supported. The long term plan is to depreciate
SO_BINTIME and move everything to using SO_TS_CLOCK.
Idea for this enhancement has been briefly discussed on the Net session
during dev summit in Ottawa last June and the general input was positive.
This change is believed to benefit network benchmarks/profiling as well
as other scenarios where precise time of arrival measurement is necessary.
There are two regression test cases as part of this commit: one extends unix
domain test code (unix_cmsg) to test new SCM_XXX types and another one
implementis totally new test case which exchanges UDP packets between two
processes using both conventional methods (i.e. calling clock_gettime(2)
before recv(2) and after send(2)), as well as using setsockopt()+recv() in
receive path. The resulting delays are checked for sanity for all supported
clock types.
Reviewed by: adrian, gnn
Differential Revision: https://reviews.freebsd.org/D9171
gtaskqueue bits at SI_SUB_INIT_IF instead of waiting until SI_SUB_SMP
which is far too late.
Add an assertion in taskqgroup_attach() to catch startup initialization
failures in the future.
Reported by: kib bde
Replace archaic "busses" with modern form "buses."
Intentionally excluded:
* Old/random drivers I didn't recognize
* Old hardware in general
* Use of "busses" in code as identifiers
No functional change.
http://grammarist.com/spelling/buses-busses/
PR: 216099
Reported by: bltsrc at mail.ru
Sponsored by: Dell EMC Isilon
dropped then reacquired due to using M_WAITOK, which opens a window in
which the tty device can disappear. Check for this and return ENXIO
back up the call chain so that callers can cope.
This closes a race where TF_GONE would get set while buffers were being
allocated as part of ttydev_open(), causing a subsequent call to
ttydevsw_modem() later in ttydev_open() to assert.
Reported by: pho
Reviewed by: kib
drain timeout handling to historical freebsd behavior.
The primary reason for these changes is the need to have tty_drain() call
ttydevsw_busy() at some reasonable sub-second rate, to poll hardware that
doesn't signal an interrupt when the transmit shift register becomes empty
(which includes virtually all USB serial hardware). Such hardware hangs
in a ttyout wait, because it never gets an opportunity to trigger a wakeup
from the sleep in tty_drain() by calling ttydisc_getc() again, after
handing the last of the buffered data to the hardware.
While researching the history of changes to tty_drain() I stumbled across
some email describing the historical BSD behavior of tcdrain() and close()
on serial ports, and the ability of comcontrol(1) to control timeout
behavior. Using that and some advice from Bruce Evans as a guide, I've
put together these changes to implement the hardware polling and restore
the historical timeout behaviors...
- tty_drain() now calls ttydevsw_busy() in a loop at 10 Hz to accomodate
hardware that requires polling for busy state.
- The "new historical" behavior for draining during close(2) is retained:
the drain timeout is "1 second without making any progress". When the
1-second timeout expires, if the count of bytes remaining in the tty
layer buffer is smaller than last time, the timeout is extended for
another second. Unfortunately, the same logic cannot be extended all
the way down to the hardware, because the interface to that layer is a
simple busy/not-busy indication.
- Due to the previous point, an application that needs a guarantee that
all data has been transmitted must use TIOCDRAIN/tcdrain(3) before
calling close(2).
- The historical behavior of honoring the drainwait setting for TIOCDRAIN
(used by tcdrain(3)) is restored.
- The historical kern.drainwait sysctl to control the global default
drainwait time is restored, but is now named kern.tty_drainwait.
- The historical default drainwait timeout of 300 seconds is restored.
- Handling of TIOCGDRAINWAIT and TIOCSDRAINWAIT ioctls is restored
(this also makes the comcontrol(1) drainwait verb work again).
- Manpages are updated to document these behaviors.
Reviewed by: bde (prior version)
This was meant to be used by a future FORTIFY_SOURCE implementation.
Probably for good, FORTIFY_SOURCE and this particular GCCism were never
well supported by clang or other compilers. Furthermore, the technology
has long since been replaced by either static checkers, sanitizers, or
even just the strong stack protector that was enabled by default.
Drop __gnu_inline to avoid cluttering the headers.
MFC after: 5 days
sched_*(2) syscalls might be not available at runtime. Defining this
constant as zero directs POSIX-compliant code to call sysconf(3) to
detect the feature at runtime, and forces libc sysconf(3) to ask
kernel.
Noted by: ngie
Reviewed by: jilles, ngie
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D9055
Add a MSG_MOREOTOCOME message flag. When this flag is set, sosend*
set PRUS_MOREOTOCOME when invoking the protocol send method. The aio
worker tasks for sending on a socket set this flag when there are
additional write jobs waiting on the socket buffer.
Reviewed by: adrian
MFC after: 1 month
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D8955
Right now size of the structure is 472 bytes on amd64, which is
already large and stack allocations are indesirable. With the ino64
work, MNAMELEN is increased to 1024, which will make it impossible to have
struct statfs on the stack.
Extracted from: ino64 work by gleb
Discussed with: mckusick
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Upon each execve, we allocate a KVA range for use in copying data to the
new image. Pages must be faulted into the range, and when the range is
freed, the backing pages are freed and their mappings are destroyed. This
is a lot of needless overhead, and the exec_map management becomes a
bottleneck when many CPUs are executing execve concurrently. Moreover, the
number of available ranges is fixed at 16, which is insufficient on large
systems and potentially excessive on 32-bit systems.
The new allocator reduces overhead by making exec_map allocations
persistent. When a range is freed, pages backing the range are marked clean
and made easy to reclaim. With this change, the exec_map is sized based on
the number of CPUs.
Reviewed by: kib
MFC after: 1 month
Differential Revision: https://reviews.freebsd.org/D8921
Instead of spuriously re-reading the lock value, read it once.
This change also has a side effect of fixing a performance bug:
on failed _mtx_obtain_lock, it was possible that re-read would find
the lock is unowned, but in this case the primitive would make a trip
through turnstile code.
This is diff reduction to a variant which uses atomic_fcmpset.
Discussed with: jhb (previous version)
Tested by: pho (previous version)
On systems without a configured swap device, an attempt to launder pages
from a swap object will always fail and result in the page being
reactivated. This means that the page daemon will continuously scan pages
that can never be evicted. With this change, anonymous pages are instead
moved to PQ_UNSWAPPABLE after a failed laundering attempt when no swap
devices are configured. PQ_UNSWAPPABLE is not scanned unless a swap device
is configured, so unreferenced unswappable pages are excluded from the page
daemon's workload.
Reviewed by: alc
Add two new qualifiers for use by the static checkers:
_Nonnull
The _Nonnull nullability qualifier indicates that null is not a meaningful
value for a value of the _Nonnull pointer type.
_Nullable
The _Nullable nullability qualifier indicates that a value of the
_Nullable pointer type can be null.
These were introduced in Clang 3.7. For more information, see:
http://clang.llvm.org/docs/AttributeReference.html#nonnull
We add these now without using them so that the GCC ports have time to
pick up the header change.
Hinted by: Android Bionic libc [1]
Also seen in: Apple's Libc-1158.20.4
[1]
baa2a973bd
This argument is not a bitmask of flags, but only accepts a single value.
Fail with EINVAL if an invalid value is passed to 'flag'. Rename the
'flags' argument to getmntinfo(3) to 'mode' as well to match.
This is a followup to r308088.
Reviewed by: kib
MFC after: 1 month
closed by r310302 for knote().
If KN_INFLUX | KN_SCAN flags are set for the note passed to knote() or
knote_fork(), i.e. the knote is scanned, we might erronously clear
INFLUX when finishing notification. For normal knote() it was fixed
in r310302 simply by remembering the fact that we do not own
KN_INFLUX, since there we own knlist lock and scan thread cannot clear
KN_INFLUX until we drop the lock. For knote_fork(), the situation is
more complicated, e must drop knlist lock AKA the process lock, since
we need to register new knotes.
Change KN_INFLUX into counter and allow shared ownership of the
in-flux state between scan and knote_fork() or knote(). Both in-flux
setters need to ensure that knote is not dropped in parallel. Added
assert about kn_influx == 1 in knote_drop() verifies that in-flux state
is not shared when knote is destroyed.
Since KBI of the struct knote is changed by addition of the int
kn_influx field, reorder kn_hook and kn_hookid to fill pad on LP64
arches [1]. This keeps sizeof(struct knote) to same 128 bytes as it
was before addition of kn_influx, on amd64.
Reviewed by: markj
Suggested by: markj [1]
Tested by: pho (previous version)
Sponsored by: The FreeBSD Foundation
Differential revision: https://reviews.freebsd.org/D8898
There is no need to do two allocations per kqueue timer. Gather all
data needed by the timer callout into the structure and allocate it at
once.
Use the structure to preserve the result of timer2sbintime(), to not
perform repeated 64bit calculations in callout.
Remove tautological casts.
Remove now unused p_nexttime [1].
Noted by: markj [1]
Reviewed by: markj (previous version)
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
X-MFC note: do not remove p_nexttime
Differential revision: https://reviews.freebsd.org/D8901
This 6 times gettimeofday performance, as measured by
tools/tools/syscall_timing
Reviewed by: kib
MFC after: 1 week
Sponsored by: Microsoft
Differential Revision: https://reviews.freebsd.org/D8789
Without this change, every individual FEATURE() declaration would have
an individual metric in Prometheus. Though this wouldn't be harmful, it
would look very cluttered.
By letting it use a single metric with the name of the feature attached
as a label, it also becomes easier to search, as you can apply regex
matching, etc.
Reviewed by: cem
Differential Revision: https://reviews.freebsd.org/D8775
I'm currently working on writing a metrics exporter for the Prometheus
monitoring system to provide access to sysctl metrics. Prometheus and
sysctl have some structural differences:
- sysctl is a tree of string component names.
- Prometheus uses a flat namespace for its metrics, but allows you to
attach labels with values to them, so that you can do aggregation.
An initial version of my exporter simply translated
hw.acpi.thermal.tz1.temperature
to
sysctl_hw_acpi_thermal_tz1_temperature_celcius
while we should ideally have
sysctl_hw_acpi_thermal_temperature_celcius{thermal_zone="tz1"}
allowing you to graph all thermal zones on a system in one go.
The change presented in this commit adds support for accomplishing this,
by providing the ability to attach labels to nodes. In the example I
gave above, the label "thermal_zone" would be attached to "tz1". As this
is a feature that will only be used very rarely, I decided to not change
the KPI too aggressively.
Discussed on: hackers@
Reviewed by: cem
Differential Revision: https://reviews.freebsd.org/D8775
This allows blind increment of relevant counters which under contention
is cheaper than inc-not-zero loops at least on amd64.
Use it in some of the places which are guaranteed to see already active
vnodes.
Reviewed by: kib (previous version)
Changes include modifications in kernel crash dump routines, dumpon(8) and
savecore(8). A new tool called decryptcore(8) was added.
A new DIOCSKERNELDUMP I/O control was added to send a kernel crash dump
configuration in the diocskerneldump_arg structure to the kernel.
The old DIOCSKERNELDUMP I/O control was renamed to DIOCSKERNELDUMP_FREEBSD11 for
backward ABI compatibility.
dumpon(8) generates an one-time random symmetric key and encrypts it using
an RSA public key in capability mode. Currently only AES-256-CBC is supported
but EKCD was designed to implement support for other algorithms in the future.
The public key is chosen using the -k flag. The dumpon rc(8) script can do this
automatically during startup using the dumppubkey rc.conf(5) variable. Once the
keys are calculated dumpon sends them to the kernel via DIOCSKERNELDUMP I/O
control.
When the kernel receives the DIOCSKERNELDUMP I/O control it generates a random
IV and sets up the key schedule for the specified algorithm. Each time the
kernel tries to write a crash dump to the dump device, the IV is replaced by
a SHA-256 hash of the previous value. This is intended to make a possible
differential cryptanalysis harder since it is possible to write multiple crash
dumps without reboot by repeating the following commands:
# sysctl debug.kdb.enter=1
db> call doadump(0)
db> continue
# savecore
A kernel dump key consists of an algorithm identifier, an IV and an encrypted
symmetric key. The kernel dump key size is included in a kernel dump header.
The size is an unsigned 32-bit integer and it is aligned to a block size.
The header structure has 512 bytes to match the block size so it was required to
make a panic string 4 bytes shorter to add a new field to the header structure.
If the kernel dump key size in the header is nonzero it is assumed that the
kernel dump key is placed after the first header on the dump device and the core
dump is encrypted.
Separate functions were implemented to write the kernel dump header and the
kernel dump key as they need to be unencrypted. The dump_write function encrypts
data if the kernel was compiled with the EKCD option. Encrypted kernel textdumps
are not supported due to the way they are constructed which makes it impossible
to use the CBC mode for encryption. It should be also noted that textdumps don't
contain sensitive data by design as a user decides what information should be
dumped.
savecore(8) writes the kernel dump key to a key.# file if its size in the header
is nonzero. # is the number of the current core dump.
decryptcore(8) decrypts the core dump using a private RSA key and the kernel
dump key. This is performed by a child process in capability mode.
If the decryption was not successful the parent process removes a partially
decrypted core dump.
Description on how to encrypt crash dumps was added to the decryptcore(8),
dumpon(8), rc.conf(5) and savecore(8) manual pages.
EKCD was tested on amd64 using bhyve and i386, mipsel and sparc64 using QEMU.
The feature still has to be tested on arm and arm64 as it wasn't possible to run
FreeBSD due to the problems with QEMU emulation and lack of hardware.
Designed by: def, pjd
Reviewed by: cem, oshogbo, pjd
Partial review: delphij, emaste, jhb, kib
Approved by: pjd (mentor)
Differential Revision: https://reviews.freebsd.org/D4712
kinfo_proc::ki_tdname is three characters shorter than
thread::td_name. Add a ki_moretdname field for these three
extra characters. Add the new field to kinfo_proc32, as well.
Update all in-tree consumers to read the new field and assemble
the full name, except for lldb's HostThreadFreeBSD.cpp, which
I will handle separately. Bump __FreeBSD_version.
Reviewed by: kib
MFC after: 1 week
Relnotes: yes
Sponsored by: Dell EMC
Differential Revision: https://reviews.freebsd.org/D8722
buf_ring contains an assert that checks whether an item being
enqueued already exists on the ring. There is a subtle bug in
this assert. An item can be returned by a peek() function and
freed, and then the consumer thread can be preempted before
calling advance(). If this happens the item appears to still be
on the queue, but another thread may allocate the item from the
free pool and wind up trying to enqueue it again, causing the
assert to trigger incorrectly.
Fix this by skipping the head of the consumer's portion of the
ring, as this index is what will be returned by peek().
Sponsored by: Dell EMC Isilon
MFC After: 1 week
Differential Revision: https://reviews.freebsd.org/D8685
Reviewed by: hselasky
the vnode is inactivated. This contradicts with the nullfs caching
which keeps upper vnode around, as consequence keeping the use
reference to lower vnode.
Add a filesystem flag to request nullfs to not cache when mounted over
that filesystem, and set the flag for nfs v4 mounts.
Reported by: asomers
Reviewed by: rmacklem
Tested by: asomers, rmacklem
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
The swap pager enqueues laundered pages near the head of the inactive queue
to avoid another trip through LRU before reclamation. This change adds
support for this behaviour to the vnode pager and makes use of it in UFS and
ext2fs. Some ioflag handling is consolidated into a common subroutine so
that this support can be easily extended to other filesystems which make use
of the buffer cache. No changes are needed for ZFS since its putpages
routine always undirties the pages before returning, and the laundry
thread requeues the pages appropriately in this case.
Reviewed by: alc, kib
Differential Revision: https://reviews.freebsd.org/D8589
longer used. More precisely, they are always zero because the code that
decremented and incremented them no longer exists.
Bump __FreeBSD_version to mark this change.
Reviewed by: kib, markj
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D8583
do any speculations about readahead, and use exactly the amount of readahead
specified by user. E.g. setting SF_FLAGS(0, SF_USER_READAHEAD) will guarantee
that no readahead at all will be performed.
