Move software iSCSI tunables/sysctls into kern.icl.soft subtree.
Replace several hardcoded length constants there with variables.
While there, stretch the limits to better match Linux' open-iscsi
and our own initiator with new MAXPHYS of 1MB. Our CTL target is
also optimized for up to 1MB I/Os, so there is also a match now.
For Windows 10 and VMware 6.7 initiators at default settings it
should make no change, since previous limits were sufficient there.
Tests of QD1 1MB writes from FreeBSD over 10GigE link show throughput
increase by 29% on idle connection and 132% with concurrent QD8 reads.
MFC after: 3 days
Sponsored by: iXsystems, Inc.
RFC 7143 (11.2.1.8):
An ITT value of 0xffffffff is reserved and MUST NOT be assigned for a
task by the initiator. The only instance in which it may be seen on
the wire is in a target-initiated NOP-In PDU (Section 11.19) and in
the initiator response to that PDU, if necessary.
MFC after: 1 month
For hardware offload solicited data may potentially be handled more
efficiently than unsolicited due to direct data placement. Or there
can be some unsolicited write buffering limitations. It may create
situations where FirstBurstLength limit is really useful.
Software driver though has no those factors, having to do memcopy()
any way and having no so hard limit on the temporary storage. Same
time more active use of unsolicited transfers allows to avoid some
of Ready To Transfer (R2T) PDU round-trip times and processing.
This change effectively doubles from 64KB to 128KB the maximum size
of write command that can be transferred within one link RTT. Tests
of (64KB, 128KB] QD1 writes mixed with simultaneous QD8 reads over
the same connection, increasing RTT, shows almost double write speed
and half latency, while we should be able to afford few megabytes of
RAM for additional buffering on a target these days.
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
Replace MAXPHYS by runtime variable maxphys. It is initialized from
MAXPHYS by default, but can be also adjusted with the tunable kern.maxphys.
Make b_pages[] array in struct buf flexible. Size b_pages[] for buffer
cache buffers exactly to atop(maxbcachebuf) (currently it is sized to
atop(MAXPHYS)), and b_pages[] for pbufs is sized to atop(maxphys) + 1.
The +1 for pbufs allow several pbuf consumers, among them vmapbuf(),
to use unaligned buffers still sized to maxphys, esp. when such
buffers come from userspace (*). Overall, we save significant amount
of otherwise wasted memory in b_pages[] for buffer cache buffers,
while bumping MAXPHYS to desired high value.
Eliminate all direct uses of the MAXPHYS constant in kernel and driver
sources, except a place which initialize maxphys. Some random (and
arguably weird) uses of MAXPHYS, e.g. in linuxolator, are converted
straight. Some drivers, which use MAXPHYS to size embeded structures,
get private MAXPHYS-like constant; their convertion is out of scope
for this work.
Changes to cam/, dev/ahci, dev/ata, dev/mpr, dev/mpt, dev/mvs,
dev/siis, where either submitted by, or based on changes by mav.
Suggested by: mav (*)
Reviewed by: imp, mav, imp, mckusick, scottl (intermediate versions)
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential revision: https://reviews.freebsd.org/D27225
Foundation copyrights, approved by emaste@. It does not include
files which carry other people's copyrights; if you're one
of those people, feel free to make similar change.
Reviewed by: emaste, imp, gbe (manpages)
Differential Revision: https://reviews.freebsd.org/D26980
SAM-3 specification introduced concept of Task Priority, that was renamed
to Command Priority in SAM-4, and supported by all modern SCSI transports.
It provides 15 levels of relative priorities: 1 - highest, 15 - lowest and
0 - default. SAT specification for SATA devices translates priorities 1-3
into NCQ high priority.
This change adds new "priority" field into empty spots of struct ccb_scsiio
and struct ccb_accept_tio of CAM and struct ctl_scsiio of CTL. Respective
support is added into iscsi(4), isp(4), mpr(4), mps(4) and ocs_fc(4) drivers
for both initiator and where applicable target roles. Minimal support was
added to CTL to receive the priority value from different frontends, pass it
between HA controllers and report in few places.
This patch does not add consumers of this functionality, so nothing should
really change yet, since the field is still set to 0 (default) on initiator
and not actively used on target. Those are to be implemented separately.
I've confirmed priority working on WD Red SATA disks connected via mpr(4)
and properly transferred to CTL target via iscsi(4), isp(4) and ocs_fc(4).
While there, added missing tag_action support to ocs_fc(4) initiator role.
MFC after: 1 month
Relnotes: yes
Sponsored by: iXsystems, Inc.
Make the Ethernet PCP codepoint configurable
for L2 local traffic, to allow lower latency for
iSCSI block IO. This addresses the initiator
side only.
Reviewed by: mav, trasz, bcr
Sponsored by: NetApp, Inc.
Differential Revision: https://reviews.freebsd.org/D26739
It does not change anything immediately, but allows further support of
Command Priority, Status Qualifier and new task management functions.
MFC after: 1 month
Sponsored by: iXsystems, Inc.
another thread has started to send in a CCB and already checked
the queue wasn't frozen, we would end up with iscsi_action()
being called despite the queue is now frozen.
Add a check to make sure this doesn't happen . Perhaps this should
be fixed at the CAM level instead, but given how the send queue and
SIM are governed by two separate mutexes, it is somewhat hard to do.
Reviewed by: imp, mav
MFC after: 2 weeks
Sponsored by: NetApp, Inc.
Sponsored by: Klara, Inc.
Differential Revision: https://reviews.freebsd.org/D26750
Allow the DSCP codepoint also to be configurable
for the traffic in the direction from the initiator
to the target, such that writes and any requests
are also treated in the appropriate QoS class.
Reviewed by: mav
MFC after: 2 weeks
Sponsored by: NetApp, Inc.
Differential Revision: https://reviews.freebsd.org/D26714
Add ICL_NOCOPY flag to icl_pdu_append_data(), specifying that the method
can just reference the data buffer instead of immediately copying it.
Extend the offload KPI with optional PDU queue method, allowing to specify
completion callback, called when all the data referenced by above has been
transferred and won't be accessed any more (the buffers can be freed).
Implement the above functionality in software iSCSI driver using mbufs
with external storage and reference counter. Note that some NICs (ixl(4))
may keep the mbuf in TX queue for a long time, so CTL has to be ready.
Add optional method to struct ctl_scsiio for buffer reference counting.
Implement it for CTL block backend, allowing to delay free of the struct
ctl_be_block_io and memory it references as needed. In first reincarnation
of the patch I tried to delay whole I/O as it is done for FibreChannel,
that was cleaner, but due to the above callback delays I had to rewrite
it this way to not leave LUN referenced potentially for hours or more.
All together on sequential read from ZFS ARC this saves about 30% of CPU
time and memory bandwidth by avoiding one of 3 memory copies (the other
two are from ZFS ARC to DMU cache and then from DMU cache to CTL buffers).
On tests with 2x Xeon Silver 4114 this allows to reach full line rate of
100GigE NIC. Tests with Gold CPUs and two 100GigE NICs are stil TBD,
but expectations to saturate them are pretty high. ;)
Discussed with: Chelsio
Sponsored by: iXsystems, Inc.
Setting so_snd.sb_lowat to at least 1/8 of the socket buffer size allows
send thread more actively use PDUs coalescing, that dramatically reduces
TCP lock congestion and number of context switches, when the socket is
full and PDUs are small.
MFC after: 1 week
Sponsored by: iXsystems, Inc.
r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are
still not MPSAFE (or already are but aren’t properly marked). Use it in
preparation for a general review of all nodes.
This is non-functional change that adds annotations to SYSCTL_NODE and
SYSCTL_PROC nodes using one of the soon-to-be-required flags.
Reviewed by: kib, trasz
Approved by: kib (mentor)
Differential Revision: https://reviews.freebsd.org/D23640
rename the source to gsb_crc32.c.
This is a prerequisite of unifying kernel zlib instances.
PR: 229763
Submitted by: Yoshihiro Ota <ota at j.email.ne.jp>
Differential Revision: https://reviews.freebsd.org/D20193
This allows replacing "sys/eventfilter.h" includes with "sys/_eventfilter.h"
in other header files (e.g., sys/{bus,conf,cpu}.h) and reduces header
pollution substantially.
EVENTHANDLER_DECLARE and EVENTHANDLER_LIST_DECLAREs were moved out of .c
files into appropriate headers (e.g., sys/proc.h, powernv/opal.h).
As a side effect of reduced header pollution, many .c files and headers no
longer contain needed definitions. The remainder of the patch addresses
adding appropriate includes to fix those files.
LOCK_DEBUG and LOCK_FILE_LINE_ARG are moved to sys/_lock.h, as required by
sys/mutex.h since r326106 (but silently protected by header pollution prior
to this change).
No functional change (intended). Of course, any out of tree modules that
relied on header pollution for sys/eventhandler.h, sys/lock.h, or
sys/mutex.h inclusion need to be fixed. __FreeBSD_version has been bumped.
this check on open, but "iscsictl -M", or an iSCSI redirect received by
iscsid(8) could end up with two sessions with the same target name and
portal.
MFC after: 2 weeks
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.
The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
can be used prior to the ISCSIDHANDOFF IOCTL which set the negotiated values.
Else the login PDU will fail when passing the "-r" option to "iscsictl" which
means iSCSI over RDMA instead of TCP/IP.
Discussed with: np@ and trasz@
Sponsored by: Mellanox Technologies
MFC after: 1 week
CAM_REQ_ABORTED sounds natural for aborting outstanding requests when
tearing down a session, but that status actually causes eligible
requests to be tried again. That's completely useless, so let's use
CAM_DEV_NOT_THERE instead. Perhaps there is a better status, but this
should be good enough. The change should affect only the session
termination.
Tested by: Ben RUBSON <ben.rubson@gmail.com>
Reviewed by: mav, trasz
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D12653
It should be sufficient to hold the lock just for removing the session
from the session list. Everything else should be covered by the session
specific lock.
On top of that, at present we can get a deadlock caused by waiting on
the CAM SIM reference count while holding the global lock. A specific
scenario involving ZFS is this:
- concurrent termination of two sessions, S1 and S2
- session S1 completed all I/Os and sleeps in CAM waiting for device
close by ZFS;
- session S2 is also dead now, but can not forcefully complete
outstanding requests by calling iscsi_session_cleanup() from
iscsi_maintenance_thread_terminate(), since it can't get the same
global sc_lock;
- as soon as there are unfinished requests, ZFS can not do
spa_config_enter() as writer, and so can not close the device for
session S1;
- deadlock.
Reported by: Ben RUBSON <ben.rubson@gmail.com>
Tested by: Ben RUBSON <ben.rubson@gmail.com>
Reviewed by: mav, trasz
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D12652
o Separate fields of struct socket that belong to listening from
fields that belong to normal dataflow, and unionize them. This
shrinks the structure a bit.
- Take out selinfo's from the socket buffers into the socket. The
first reason is to support braindamaged scenario when a socket is
added to kevent(2) and then listen(2) is cast on it. The second
reason is that there is future plan to make socket buffers pluggable,
so that for a dataflow socket a socket buffer can be changed, and
in this case we also want to keep same selinfos through the lifetime
of a socket.
- Remove struct struct so_accf. Since now listening stuff no longer
affects struct socket size, just move its fields into listening part
of the union.
- Provide sol_upcall field and enforce that so_upcall_set() may be called
only on a dataflow socket, which has buffers, and for listening sockets
provide solisten_upcall_set().
o Remove ACCEPT_LOCK() global.
- Add a mutex to socket, to be used instead of socket buffer lock to lock
fields of struct socket that don't belong to a socket buffer.
- Allow to acquire two socket locks, but the first one must belong to a
listening socket.
- Make soref()/sorele() to use atomic(9). This allows in some situations
to do soref() without owning socket lock. There is place for improvement
here, it is possible to make sorele() also to lock optionally.
- Most protocols aren't touched by this change, except UNIX local sockets.
See below for more information.
o Reduce copy-and-paste in kernel modules that accept connections from
listening sockets: provide function solisten_dequeue(), and use it in
the following modules: ctl(4), iscsi(4), ng_btsocket(4), ng_ksocket(4),
infiniband, rpc.
o UNIX local sockets.
- Removal of ACCEPT_LOCK() global uncovered several races in the UNIX
local sockets. Most races exist around spawning a new socket, when we
are connecting to a local listening socket. To cover them, we need to
hold locks on both PCBs when spawning a third one. This means holding
them across sonewconn(). This creates a LOR between pcb locks and
unp_list_lock.
- To fix the new LOR, abandon the global unp_list_lock in favor of global
unp_link_lock. Indeed, separating these two locks didn't provide us any
extra parralelism in the UNIX sockets.
- Now call into uipc_attach() may happen with unp_link_lock hold if, we
are accepting, or without unp_link_lock in case if we are just creating
a socket.
- Another problem in UNIX sockets is that uipc_close() basicly did nothing
for a listening socket. The vnode remained opened for connections. This
is fixed by removing vnode in uipc_close(). Maybe the right way would be
to do it for all sockets (not only listening), simply move the vnode
teardown from uipc_detach() to uipc_close()?
Sponsored by: Netflix
Differential Revision: https://reviews.freebsd.org/D9770
Renumber cluase 4 to 3, per what everybody else did when BSD granted
them permission to remove clause 3. My insistance on keeping the same
numbering for legal reasons is too pedantic, so give up on that point.
Submitted by: Jan Schaumann <jschauma@stevens.edu>
Pull Request: https://github.com/freebsd/freebsd/pull/96
I found that at least with Chelsio NICs TOE sockets quite often report
negative sbspace() values. Using unsigned variable to store it resulted
in attempts to aggregate too much data in one sosend() call, that caused
errors and following connection termination.
MFC after: 2 weeks
All this code is based on assumption that data will be stored in one piece,
and since buffer size if known and fixed, it is easier to hardcode it.
MFC after: 2 weeks
In general case m_pullup() does not really guarantee any data alignment.
Instead of depenting on side effects caused by data being always copied
out of mbuf cluster (which is probably a bug by itself), always allocate
aligned BHS buffer and read data there directly from socket.
While there, reuse new icl_conn_receive_buf() function to read digests.
The code could probably be even more optimized to aggregate those reads,
but until that done, this is still easier then the way it was before.
MFC after: 2 weeks
ip_data_mbuf is always appended to ip_bhs_mbuf, so it does not need own
packet header. This change first avoids allocation/initialization of the
header, and then avoids dropping one when it later gets to socket buffer.
MFC after: 2 weeks
Decouple the send and receive limits on the amount of data in a single
iSCSI PDU. MaxRecvDataSegmentLength is declarative, not negotiated, and
is direction-specific so there is no reason for both ends to limit
themselves to the same min(initiator, target) value in both directions.
Allow iSCSI drivers to report their send, receive, first burst, and max
burst limits explicitly instead of using hardcoded values or trying to
derive all of them from the receive limit (which was the only limit
reported by the drivers prior to this change).
Display the send and receive limits separately in the userspace iSCSI
utilities.
Reviewed by: jpaetzel@ (earlier version), trasz@
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D7279
discovery without attaching to the targets ("iscsictl -Ad ... -e off"),
and then attach to selected ones ("iscsictl -Mi ... -e on").
PR: 204129
MFC after: 1 month
Relnotes: yes
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D6633
