Commit 95bd09eb27 ("tcp: TSO packets automatic sizing") tried to
control TSO size, but did this at the wrong place (sendmsg() time)
At sendmsg() time, we might have a pessimistic view of flow rate,
and we end up building very small skbs (with 2 MSS per skb).
This is bad because :
- It sends small TSO packets even in Slow Start where rate quickly
increases.
- It tends to make socket write queue very big, increasing tcp_ack()
processing time, but also increasing memory needs, not necessarily
accounted for, as fast clones overhead is currently ignored.
- Lower GRO efficiency and more ACK packets.
Servers with a lot of small lived connections suffer from this.
Lets instead fill skbs as much as possible (64KB of payload), but split
them at xmit time, when we have a precise idea of the flow rate.
skb split is actually quite efficient.
Patch looks bigger than necessary, because TCP Small Queue decision now
has to take place after the eventual split.
As Neal suggested, introduce a new tcp_tso_autosize() helper, so that
tcp_tso_should_defer() can be synchronized on same goal.
Rename tp->xmit_size_goal_segs to tp->gso_segs, as this variable
contains number of mss that we can put in GSO packet, and is not
related to the autosizing goal anymore.
Tested:
40 ms rtt link
nstat >/dev/null
netperf -H remote -l -2000000 -- -s 1000000
nstat | egrep "IpInReceives|IpOutRequests|TcpOutSegs|IpExtOutOctets"
Before patch :
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/s
87380 2000000 2000000 0.36 44.22
IpInReceives 600 0.0
IpOutRequests 599 0.0
TcpOutSegs 1397 0.0
IpExtOutOctets 2033249 0.0
After patch :
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
87380 2000000 2000000 0.36 44.27
IpInReceives 221 0.0
IpOutRequests 232 0.0
TcpOutSegs 1397 0.0
IpExtOutOctets 2013953 0.0
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
... making both non-draining. That means that tcp_recvmsg() becomes
non-draining. And _that_ would break iscsit_do_rx_data() unless we
a) make sure tcp_recvmsg() is uniformly non-draining (it is)
b) make sure it copes with arbitrary (including shifted)
iov_iter (it does, all it uses is iov_iter primitives)
c) make iscsit_do_rx_data() initialize ->msg_iter only once.
Fortunately, (c) is doable with minimal work and we are rid of one
the two places where kernel send/recvmsg users would be unhappy with
non-draining behaviour.
Actually, that makes all but one of ->recvmsg() instances iov_iter-clean.
The exception is skcipher_recvmsg() and it also isn't hard to convert
to primitives (iov_iter_get_pages() is needed there). That'll wait
a bit - there's some interplay with ->sendmsg() path for that one.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Just use copy_from_iter(). That's what this method is trying to do
in all cases, in a very convoluted fashion.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Note that the code _using_ ->msg_iter at that point will be very
unhappy with anything other than unshifted iovec-backed iov_iter.
We still need to convert users to proper primitives.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
it'll die soon enough - now that kvec-backed iov_iter works regardless
of set_fs(), both instances will become copy_from_iter() as soon as
we introduce ->msg_iter...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pull arm64 updates from Will Deacon:
"Here's the usual mixed bag of arm64 updates, also including some
related EFI changes (Acked by Matt) and the MMU gather range cleanup
(Acked by you).
Changes include:
- support for alternative instruction patching from Andre
- seccomp from Akashi
- some AArch32 instruction emulation, required by the Android folks
- optimisations for exception entry/exit code, cmpxchg, pcpu atomics
- mmu_gather range calculations moved into core code
- EFI updates from Ard, including long-awaited SMBIOS support
- /proc/cpuinfo fixes to align with the format used by arch/arm/
- a few non-critical fixes across the architecture"
* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (70 commits)
arm64: remove the unnecessary arm64_swiotlb_init()
arm64: add module support for alternatives fixups
arm64: perf: Prevent wraparound during overflow
arm64/include/asm: Fixed a warning about 'struct pt_regs'
arm64: Provide a namespace to NCAPS
arm64: bpf: lift restriction on last instruction
arm64: Implement support for read-mostly sections
arm64: compat: align cacheflush syscall with arch/arm
arm64: add seccomp support
arm64: add SIGSYS siginfo for compat task
arm64: add seccomp syscall for compat task
asm-generic: add generic seccomp.h for secure computing mode 1
arm64: ptrace: allow tracer to skip a system call
arm64: ptrace: add NT_ARM_SYSTEM_CALL regset
arm64: Move some head.text functions to executable section
arm64: jump labels: NOP out NOP -> NOP replacement
arm64: add support to dump the kernel page tables
arm64: Add FIX_HOLE to permanent fixed addresses
arm64: alternatives: fix pr_fmt string for consistency
arm64: vmlinux.lds.S: don't discard .exit.* sections at link-time
...
Move the port check [ipvlan_dev_master()] and device check
[ipvlan_dev_slave()] functions to netdevice.h and rename them
netif_is_ipvlan_port() and netif_is_ipvlan() resp. to be
consistent with macvlan api naming.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similar to a check for macvlan device, netif_is_macvlan(), add
another function to check if a device is used as macvlan port.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit f886497212 ("ipv4: fix dst race in sk_dst_get()")
DST_NOCACHE dst_entries get freed by RCU. So there is no need to get a
reference on them when we are in rcu protected sections.
Cc: Eric Dumazet <edumazet@google.com>
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reviewed-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull Altera Nios II processor support from Ley Foon Tan:
"Here is the Linux port for Nios II processor (from Altera) arch/nios2/
tree for v3.19.
The patchset has been discussed on the kernel mailing lists since
April and has gone through 6 revisions of review. The additional
changes since then have been mostly further cleanups and fixes when
merged with other trees.
The arch code is in arch/nios2 and one asm-generic change (acked by
Arnd)"
Arnd Bergmann says:
"I've reviewed the architecture port in the past and it looks good in
its latest version"
Acked-by: Arnd Bergmann <arnd@arndb.de>
* tag 'nios2-v3.19-rc1' of git://git.rocketboards.org/linux-socfpga-next: (40 commits)
nios2: Make NIOS2_CMDLINE_IGNORE_DTB depend on CMDLINE_BOOL
nios2: Add missing NR_CPUS to Kconfig
nios2: asm-offsets: Remove unused definition TI_TASK
nios2: Remove write-only struct member from nios2_timer
nios2: Remove unused extern declaration of shm_align_mask
nios2: include linux/type.h in io.h
nios2: move include asm-generic/io.h to end of file
nios2: remove include asm-generic/iomap.h from io.h
nios2: remove unnecessary space before define
nios2: fix error handling of irq_of_parse_and_map
nios2: Use IS_ENABLED instead of #ifdefs to check config symbols
nios2: Build infrastructure
Documentation: Add documentation for Nios2 architecture
MAINTAINERS: Add nios2 maintainer
nios2: ptrace support
nios2: Module support
nios2: Nios2 registers
nios2: Miscellaneous header files
nios2: Cpuinfo handling
nios2: Time keeping
...
When deploying FQ pacing, one thing we noticed is that CUBIC Hystart
triggers too soon.
Having SNMP counters to have an idea of how often the various Hystart
methods trigger is useful prior to any modifications.
This patch adds SNMP counters tracking, how many time "ack train" or
"Delay" based Hystart triggers, and cumulative sum of cwnd at the time
Hystart decided to end SS (Slow Start)
myhost:~# nstat -a | grep Hystart
TcpExtTCPHystartTrainDetect 9 0.0
TcpExtTCPHystartTrainCwnd 20650 0.0
TcpExtTCPHystartDelayDetect 10 0.0
TcpExtTCPHystartDelayCwnd 360 0.0
->
Train detection was triggered 9 times, and average cwnd was
20650/9=2294,
Delay detection was triggered 10 times and average cwnd was 36
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is never called and implementations are void. So just remove it.
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
legacy_only flag is now unused, drop it from core.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Commit ce1a4ea3f1 ("net: avoid one atomic operation in skb_clone()")
took the wrong way to save one atomic operation.
It is actually possible to avoid two atomic operations, if we
do not change skb->fclone values, and only rely on clone_ref
content to signal if the clone is available or not.
skb_clone() can simply use the fast clone if clone_ref is 1.
kfree_skbmem() can avoid the atomic_dec_and_test() if clone_ref is 1.
Note that because we usually free the clone before the original skb,
this particular attempt is only done for the original skb to have better
branch prediction.
SKB_FCLONE_FREE is removed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Chris Mason <clm@fb.com>
Cc: Sabrina Dubroca <sd@queasysnail.net>
Cc: Vijay Subramanian <subramanian.vijay@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit 56bfa7ee7c ("unregister_netdevice : move RTM_DELLINK to
until after ndo_uninit") tried to do this ealier but while doing so
it created a problem. Unfortunately the delayed rtmsg_ifinfo() also
delayed call to fill_info(). So this translated into asking driver
to remove private state and then query it's private state. This
could have catastropic consequences.
This change breaks the rtmsg_ifinfo() into two parts - one takes the
precise snapshot of the device by called fill_info() before calling
the ndo_uninit() and the second part sends the notification using
collected snapshot.
It was brought to notice when last link is deleted from an ipvlan device
when it has free-ed the port and the subsequent .fill_info() call is
trying to get the info from the port.
kernel: [ 255.139429] ------------[ cut here ]------------
kernel: [ 255.139439] WARNING: CPU: 12 PID: 11173 at net/core/rtnetlink.c:2238 rtmsg_ifinfo+0x100/0x110()
kernel: [ 255.139493] Modules linked in: ipvlan bonding w1_therm ds2482 wire cdc_acm ehci_pci ehci_hcd i2c_dev i2c_i801 i2c_core msr cpuid bnx2x ptp pps_core mdio libcrc32c
kernel: [ 255.139513] CPU: 12 PID: 11173 Comm: ip Not tainted 3.18.0-smp-DEV #167
kernel: [ 255.139514] Hardware name: Intel RML,PCH/Ibis_QC_18, BIOS 1.0.10 05/15/2012
kernel: [ 255.139515] 0000000000000009 ffff880851b6b828 ffffffff815d87f4 00000000000000e0
kernel: [ 255.139516] 0000000000000000 ffff880851b6b868 ffffffff8109c29c 0000000000000000
kernel: [ 255.139518] 00000000ffffffa6 00000000000000d0 ffffffff81aaf580 0000000000000011
kernel: [ 255.139520] Call Trace:
kernel: [ 255.139527] [<ffffffff815d87f4>] dump_stack+0x46/0x58
kernel: [ 255.139531] [<ffffffff8109c29c>] warn_slowpath_common+0x8c/0xc0
kernel: [ 255.139540] [<ffffffff8109c2ea>] warn_slowpath_null+0x1a/0x20
kernel: [ 255.139544] [<ffffffff8150d570>] rtmsg_ifinfo+0x100/0x110
kernel: [ 255.139547] [<ffffffff814f78b5>] rollback_registered_many+0x1d5/0x2d0
kernel: [ 255.139549] [<ffffffff814f79cf>] unregister_netdevice_many+0x1f/0xb0
kernel: [ 255.139551] [<ffffffff8150acab>] rtnl_dellink+0xbb/0x110
kernel: [ 255.139553] [<ffffffff8150da90>] rtnetlink_rcv_msg+0xa0/0x240
kernel: [ 255.139557] [<ffffffff81329283>] ? rhashtable_lookup_compare+0x43/0x80
kernel: [ 255.139558] [<ffffffff8150d9f0>] ? __rtnl_unlock+0x20/0x20
kernel: [ 255.139562] [<ffffffff8152cb11>] netlink_rcv_skb+0xb1/0xc0
kernel: [ 255.139563] [<ffffffff8150a495>] rtnetlink_rcv+0x25/0x40
kernel: [ 255.139565] [<ffffffff8152c398>] netlink_unicast+0x178/0x230
kernel: [ 255.139567] [<ffffffff8152c75f>] netlink_sendmsg+0x30f/0x420
kernel: [ 255.139571] [<ffffffff814e0b0c>] sock_sendmsg+0x9c/0xd0
kernel: [ 255.139575] [<ffffffff811d1d7f>] ? rw_copy_check_uvector+0x6f/0x130
kernel: [ 255.139577] [<ffffffff814e11c9>] ? copy_msghdr_from_user+0x139/0x1b0
kernel: [ 255.139578] [<ffffffff814e1774>] ___sys_sendmsg+0x304/0x310
kernel: [ 255.139581] [<ffffffff81198723>] ? handle_mm_fault+0xca3/0xde0
kernel: [ 255.139585] [<ffffffff811ebc4c>] ? destroy_inode+0x3c/0x70
kernel: [ 255.139589] [<ffffffff8108e6ec>] ? __do_page_fault+0x20c/0x500
kernel: [ 255.139597] [<ffffffff811e8336>] ? dput+0xb6/0x190
kernel: [ 255.139606] [<ffffffff811f05f6>] ? mntput+0x26/0x40
kernel: [ 255.139611] [<ffffffff811d2b94>] ? __fput+0x174/0x1e0
kernel: [ 255.139613] [<ffffffff814e2129>] __sys_sendmsg+0x49/0x90
kernel: [ 255.139615] [<ffffffff814e2182>] SyS_sendmsg+0x12/0x20
kernel: [ 255.139617] [<ffffffff815df092>] system_call_fastpath+0x12/0x17
kernel: [ 255.139619] ---[ end trace 5e6703e87d984f6b ]---
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Reported-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Roopa Prabhu <roopa@cumulusnetworks.com>
Cc: David S. Miller <davem@davemloft.net>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
These were useful when I was tracking down a race condition between
svc_xprt_do_enqueue and svc_get_next_xprt.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Testing has shown that the pool->sp_lock can be a bottleneck on a busy
server. Every time data is received on a socket, the server must take
that lock in order to dequeue a thread from the sp_threads list.
Address this problem by eliminating the sp_threads list (which contains
threads that are currently idle) and replacing it with a RQ_BUSY flag in
svc_rqst. This allows us to walk the sp_all_threads list under the
rcu_read_lock and find a suitable thread for the xprt by doing a
test_and_set_bit.
Note that we do still have a potential atomicity problem however with
this approach. We don't want svc_xprt_do_enqueue to set the
rqst->rq_xprt pointer unless a test_and_set_bit of RQ_BUSY returned
zero (which indicates that the thread was idle). But, by the time we
check that, the bit could be flipped by a waking thread.
To address this, we acquire a new per-rqst spinlock (rq_lock) and take
that before doing the test_and_set_bit. If that returns false, then we
can set rq_xprt and drop the spinlock. Then, when the thread wakes up,
it must set the bit under the same spinlock and can trust that if it was
already set then the rq_xprt is also properly set.
With this scheme, the case where we have an idle thread no longer needs
to take the highly contended pool->sp_lock at all, and that removes the
bottleneck.
That still leaves one issue: What of the case where we walk the whole
sp_all_threads list and don't find an idle thread? Because the search is
lockess, it's possible for the queueing to race with a thread that is
going to sleep. To address that, we queue the xprt and then search again.
If we find an idle thread at that point, we can't attach the xprt to it
directly since that might race with a different thread waking up and
finding it. All we can do is wake the idle thread back up and let it
attempt to find the now-queued xprt.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Tested-by: Chris Worley <chris.worley@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
In a later patch, we'll be removing some spinlocking around the socket
and thread queueing code in order to fix some contention problems. At
that point, the stats counters will no longer be protected by the
sp_lock.
Change the counters to atomic_long_t fields, except for the
"sockets_queued" counter which will still be manipulated under a
spinlock.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Tested-by: Chris Worley <chris.worley@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
In a later patch, we'll want to be able to handle this flag without
holding the sp_lock. Change this field to an unsigned long flags
field, and declare a new flag in it that can be managed with atomic
bitops.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
There are a couple of holes in the svc_rqst field on x86_64. Move the
rq_cachetype to a different location to eliminate both of them.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
In a later patch, we're going to need some atomic bit flags. Since that
field will need to be an unsigned long, we mitigate that space
consumption by migrating some other bitflags to the new field. Start
with the rq_secure flag.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Mainly what I need is 860a0d9e51 "sunrpc: add some tracepoints in
svc_rqst handling functions", which subsequent server rpc patches from
jlayton depend on. I'm merging this later tag on the assumption that's
more likely to be a tested and stable point.
transports need to be able to detect legacy-only
devices (ATM balloon only) to use legacy path
to drive them.
Add a core API to do just that.
The implementation just blacklists balloon:
not too pretty, but let's not over-engineer.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
We have no plans to support virtio 1.0 in balloon driver. Add an
explicit flag to mark it legacy only.
This will be used by follow-up patches.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Replace uXX by __uXX and _packed by __attribute((packed))
as seems to be the norm for userspace headers.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Guests need to use virtio scsi API, so export it to uapi,
nice to e.g. qemu and will help us remember this file
affects ABI.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Note: for consistency, and to avoid sparse errors,
convert all fields, even those no longer in use
for virtio v1.0.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
virtio 1.0 modified virtio net header format,
making all fields little endian.
Users can tweak header format before submitting it to tun,
but this means more data copies where none were necessary.
And if the iovec is in RO memory, this means we might
need to split iovec also means we might in theory overflow
iovec max size.
This patch adds a simpler way for applications to handle this,
using new "little endian" flag in tun.
As a result, tun simply byte-swaps header fields as appropriate.
This is a NOP on LE architectures.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
TUN_ flags are internal and never exposed
to userspace. Any application using it is almost
certainly buggy.
Move them out to tun.c.
Note: we remove these completely in follow-up patches,
this code movement is split out for ease of review.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Based on patch by Cornelia Huck.
Note: for consistency, and to avoid sparse errors,
convert all fields, even those no longer in use
for virtio v1.0.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Based on patches by Rusty Russell, Cornelia Huck.
Note: more code changes are needed for 1.0 support
(due to different header size).
So we don't advertize support for 1.0 yet.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
virtio-blk has some legacy feature bits that modern drivers
must not negotiate, but are needed for old legacy hosts
(that e.g. don't support virtio-scsi).
Allow a separate legacy feature table for such cases.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
For virtio-1, we can theoretically have a more complex virtqueue
layout with avail and used buffers not on a contiguous memory area
with the descriptor table. For now, it's fine for a transport driver
to stay with the old layout: It needs, however, a way to access
the locations of the avail/used rings so it can register them with
the host.
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We (ab)use virtio conversion functions for device-specific
config space accesses.
Based on original patches by Cornelia and Rusty.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.com>
