| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
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 <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
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 <[email protected]>
Tested-by: Chris Worley <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
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 <[email protected]>
Tested-by: Chris Worley <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
...also make the manipulation of sp_all_threads list use RCU-friendly
functions.
Signed-off-by: Jeff Layton <[email protected]>
Tested-by: Chris Worley <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Currently all svc_create callers pass in NULL for the shutdown parm,
which then gets fixed up to be svc_rpcb_cleanup if the service uses
rpcbind.
Simplify this by instead having the the only caller that requires it
(lockd) pass in svc_rpcb_cleanup and get rid of the special casing.
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
The way that svc_wake_up works is a bit inefficient. It walks all of the
available pools for a service and either wakes up a task in each one or
sets the SP_TASK_PENDING flag in each one.
When svc_wake_up is called, there is no need to wake up more than one
thread to do this work. In practice, only lockd currently uses this
function and it's single threaded anyway. Thus, this just boils down to
doing a wake up of a thread in pool 0 or setting a single flag.
Eliminate the for loop in this function and change it to just operate on
pool 0. Also update the comments that sit above it and get rid of some
code that has been commented out for years now.
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
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 <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Signed-off-by: Jeff Layton <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
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 <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
|
|
Mainly what I need is 860a0d9e511f "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.
|
|
This merely fixes sparse warnings, without actually
adding support for the new APIs.
Still working out the best way to enable the new
functionality.
Signed-off-by: Michael S. Tsirkin <[email protected]>
|
|
net/bluetooth/smp.c:2650:9-16: WARNING: ERR_CAST can be used with tfm_aes
Use ERR_CAST inlined function instead of ERR_PTR(PTR_ERR(...))
Generated by: scripts/coccinelle/api/err_cast.cocci
Signed-off-by: Fengguang Wu <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next
Steffen Klassert says:
====================
pull request (net-next): ipsec-next 2014-12-03
1) Fix a set but not used warning. From Fabian Frederick.
2) Currently we make sequence number values available to userspace
only if we use ESN. Make the sequence number values also available
for non ESN states. From Zhi Ding.
3) Remove socket policy hashing. We don't need it because socket
policies are always looked up via a linked list. From Herbert Xu.
4) After removing socket policy hashing, we can use __xfrm_policy_link
in xfrm_policy_insert. From Herbert Xu.
5) Add a lookup method for vti6 tunnels with wildcard endpoints.
I forgot this when I initially implemented vti6.
Please pull or let me know if there are problems.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
This patch extends the set/get_rxfh ethtool-options for getting or
setting the RSS hash function.
It modifies drivers implementation of set/get_rxfh accordingly.
This change also delegates the responsibility of checking whether a
modification to a certain RX flow hash parameter is supported to the
driver implementation of set_rxfh.
User-kernel API is done through the new hfunc bitmask field in the
ethtool_rxfh struct. A bit set in the hfunc field is corresponding to an
index in the new string-set ETH_SS_RSS_HASH_FUNCS.
Got approval from most of the relevant driver maintainers that their
driver is using Toeplitz, and for the few that didn't answered, also
assumed it is Toeplitz.
Cc: Tom Lendacky <[email protected]>
Cc: Ariel Elior <[email protected]>
Cc: Prashant Sreedharan <[email protected]>
Cc: Michael Chan <[email protected]>
Cc: Hariprasad S <[email protected]>
Cc: Sathya Perla <[email protected]>
Cc: Subbu Seetharaman <[email protected]>
Cc: Ajit Khaparde <[email protected]>
Cc: Jeff Kirsher <[email protected]>
Cc: Jesse Brandeburg <[email protected]>
Cc: Bruce Allan <[email protected]>
Cc: Carolyn Wyborny <[email protected]>
Cc: Don Skidmore <[email protected]>
Cc: Greg Rose <[email protected]>
Cc: Matthew Vick <[email protected]>
Cc: John Ronciak <[email protected]>
Cc: Mitch Williams <[email protected]>
Cc: Amir Vadai <[email protected]>
Cc: Solarflare linux maintainers <[email protected]>
Cc: Shradha Shah <[email protected]>
Cc: Shreyas Bhatewara <[email protected]>
Cc: "VMware, Inc." <[email protected]>
Cc: Ben Hutchings <[email protected]>
Signed-off-by: Eyal Perry <[email protected]>
Signed-off-by: Amir Vadai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
since head->handle == handle (checked before), just assign handle.
Signed-off-by: Jiri Pirko <[email protected]>
Acked-by: Jamal Hadi Salim <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Signed-off-by: Jiri Pirko <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
rcu variant is not correct here. The code is called by updater (rtnl
lock is held), not by reader (no rcu_read_lock is held).
Signed-off-by: Jiri Pirko <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
rcu variant is not correct here. The code is called by updater (rtnl
lock is held), not by reader (no rcu_read_lock is held).
Signed-off-by: Jiri Pirko <[email protected]>
ACKed-by: Jamal Hadi Salim <[email protected]>
Acked-by: Daniel Borkmann <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Signed-off-by: Jiri Pirko <[email protected]>
Acked-by: Jamal Hadi Salim <[email protected]>
Acked-by: Daniel Borkmann <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Signed-off-by: Jiri Pirko <[email protected]>
Acked-by: Jamal Hadi Salim <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Convert tipc name table read-write lock to RCU. After this change,
a new spin lock is used to protect name table on write side while
RCU is applied on read side.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
When a list_head variable is seen as a new entry to be added to a
list head, it's unnecessary to be initialized with INIT_LIST_HEAD().
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
When tipc name sequence is published, name table lock is released
before name sequence buffer is delivered to remote nodes through its
underlying unicast links. However, when name sequence is withdrawn,
the name table lock is held until the transmission of the removal
message of name sequence is finished. During the process, node lock
is nested in name table lock. To prevent node lock from being nested
in name table lock, while withdrawing name, we should adopt the same
locking policy of publishing name sequence: name table lock should
be released before message is sent.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
As tipc_nametbl_lock is used to protect name_table structure, the lock
must be held while all members of name_table structure are accessed.
However, the lock is not obtained while a member of name_table
structure - local_publ_count is read in tipc_nametbl_publish(), as
a consequence, an inconsistent value of local_publ_count might be got.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
TIPC internally created a name table which is used to store name
sequences. Now there is a read-write lock - tipc_nametbl_lock to
protect the table, and each name sequence saved in the table is
protected with its private lock. When a name sequence is inserted
or removed to or from the table, its members might need to change.
Therefore, in normal case, the two locks must be held while TIPC
operates the table. However, there are still several places where
we only hold tipc_nametbl_lock without proprerly obtaining name
sequence lock, which might cause the corruption of name sequence.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
As TIPC subscriber server is terminated before name table, no user
depends on subscription list of name sequence when name table is
stopped. Therefore, all name sequences stored in name table should
be released whatever their subscriptions lists are empty or not,
otherwise, memory leak might happen.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Name table locking policy is going to be adjusted from read-write
lock protection to RCU lock protection in the future commits. But
its essential precondition is to convert the allocation way of name
table from static to dynamic mode.
Signed-off-by: Ying Xue <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The size variable is introduced in publ_list struct to help us exactly
calculate SKB buffer sizes needed by publications when all publications
in name table are delivered in bulk in named_distribute(). But if
publication SKB buffer size is assumed to MTU, the size variable in
publ_list struct can be completely eliminated at the cost of wasting
a bit memory space for last SKB.
Signed-off-by: Ying Xue <[email protected]>
Signed-off-by: Tero Aho <[email protected]>
Reviewed-by: Erik Hugne <[email protected]>
Reviewed-by: Jon Maloy <[email protected]>
Tested-by: Erik Hugne <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
|
|
The compute_score functions are a bit difficult to read.
Neaten them a bit to reduce object sizes and make them a
bit more intelligible.
Return early to avoid indentation and avoid unnecessary
initializations.
(allyesconfig, but w/ -O2 and no profiling)
$ size net/ipv[46]/udp.o.*
text data bss dec hex filename
28680 1184 25 29889 74c1 net/ipv4/udp.o.new
28756 1184 25 29965 750d net/ipv4/udp.o.old
17600 1010 2 18612 48b4 net/ipv6/udp.o.new
17632 1010 2 18644 48d4 net/ipv6/udp.o.old
Signed-off-by: Joe Perches <[email protected]>
Acked-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Allow reading of timestamps and cmsg at the same time on all relevant
socket families. One use is to correlate timestamps with egress
device, by asking for cmsg IP_PKTINFO.
on AF_INET sockets, call the relevant function (ip_cmsg_recv). To
avoid changing legacy expectations, only do so if the caller sets a
new timestamping flag SOF_TIMESTAMPING_OPT_CMSG.
on AF_INET6 sockets, IPV6_PKTINFO and all other recv cmsg are already
returned for all origins. only change is to set ifindex, which is
not initialized for all error origins.
In both cases, only generate the pktinfo message if an ifindex is
known. This is not the case for ACK timestamps.
The difference between the protocol families is probably a historical
accident as a result of the different conditions for generating cmsg
in the relevant ip(v6)_recv_error function:
ipv4: if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
ipv6: if (serr->ee.ee_origin != SO_EE_ORIGIN_LOCAL) {
At one time, this was the same test bar for the ICMP/ICMP6
distinction. This is no longer true.
Signed-off-by: Willem de Bruijn <[email protected]>
----
Changes
v1 -> v2
large rewrite
- integrate with existing pktinfo cmsg generation code
- on ipv4: only send with new flag, to maintain legacy behavior
- on ipv6: send at most a single pktinfo cmsg
- on ipv6: initialize fields if not yet initialized
The recv cmsg interfaces are also relevant to the discussion of
whether looping packet headers is problematic. For v6, cmsgs that
identify many headers are already returned. This patch expands
that to v4. If it sounds reasonable, I will follow with patches
1. request timestamps without payload with SOF_TIMESTAMPING_OPT_TSONLY
(http://patchwork.ozlabs.org/patch/366967/)
2. sysctl to conditionally drop all timestamps that have payload or
cmsg from users without CAP_NET_RAW.
Signed-off-by: David S. Miller <[email protected]>
|
|
One line change, in response to catching an occurrence of this bug.
See also fix f4713a3dfad0 ("net-timestamp: make tcp_recvmsg call ...")
Signed-off-by: Willem de Bruijn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless
|
|
* device-properties:
leds: leds-gpio: Fix multiple instances registration without 'label' property
leds: leds-gpio: Fix legacy GPIO number case
ACPI / property: Drop size_prop from acpi_dev_get_property_reference()
leds: leds-gpio: Convert gpio_blink_set() to use GPIO descriptors
ACPI / GPIO: Document ACPI GPIO mappings API
net: rfkill: gpio: Add default GPIO driver mappings for ACPI
ACPI / GPIO: Driver GPIO mappings for ACPI GPIOs
input: gpio_keys_polled: Make use of device property API
leds: leds-gpio: Make use of device property API
gpio: Support for unified device properties interface
Driver core: Unified interface for firmware node properties
input: gpio_keys_polled: Add support for GPIO descriptors
leds: leds-gpio: Add support for GPIO descriptors
gpio: sch: Consolidate core and resume banks
gpio / ACPI: Add support for _DSD device properties
misc: at25: Make use of device property API
ACPI: Allow drivers to match using Device Tree compatible property
Driver core: Unified device properties interface for platform firmware
ACPI: Add support for device specific properties
|
|
When the host decides to use a non-resolvable private address, it
must ensure that this generated address does not match the public
address of the controller. Add an extra check to ensure this required
behavior.
In addition rename the variable from urpa to nrpa and fix all of the
comments in the code that use the term unresolvable instead of the
term non-resolvable as used in the Bluetooth specification.
Signed-off-by: Marcel Holtmann <[email protected]>
Signed-off-by: Johan Hedberg <[email protected]>
|
|
xfrm_decode_session() was originally designed for the
usage in the receive path where the correct nexthdr offset
is stored in IP6CB(skb)->nhoff. Over time this function
spread to code that is used in the output path (netfilter,
vti) where IP6CB(skb)->nhoff is not set. As a result, we
get a wrong nexthdr and the upper layer flow informations
are wrong. This can leed to incorrect policy lookups.
Signed-off-by: Steffen Klassert <[email protected]>
|
|
skb->transport_header might not be valid when we do a reverse
decode because the ipv6 tunnel error handlers don't update it
to the inner transport header. This leads to a wrong offset
calculation and to wrong layer 4 informations. We fix this
by using the size of the ipv6 header as the first offset.
Signed-off-by: Steffen Klassert <[email protected]>
|
|
Fix various spelling errors in the comments of the CAN modules.
Signed-off-by: Jeremiah Mahler <[email protected]>
Acked-by: Oliver Hartkopp <[email protected]>
Signed-off-by: Marc Kleine-Budde <[email protected]>
|
|
Several CAN modules use a design pattern with a banner[] variable at the
top which defines a string that is used once during init to print the
banner. The string is also embedded with KERN_INFO which makes it
printk() specific.
Improve the code by eliminating the banner[] variable and moving the
string to where it is printed. Then switch from printk(KERN_INFO to
pr_info() for the lines that were changed.
Signed-off-by: Jeremiah Mahler <[email protected]>
Acked-by: Oliver Hartkopp <[email protected]>
Signed-off-by: Marc Kleine-Budde <[email protected]>
|
|
The SMP over BR/EDR requests for cross-transport pairing should also
accepted when the debugfs setting force_lesc_support has been enabled.
Signed-off-by: Marcel Holtmann <[email protected]>
Signed-off-by: Johan Hedberg <[email protected]>
|
|
The SMP over BR/EDR support for cross-transport pairing should also be
enabled when the debugfs setting force_lesc_support has been enabled.
Signed-off-by: Marcel Holtmann <[email protected]>
Signed-off-by: Johan Hedberg <[email protected]>
|
|
introduce new setsockopt() command:
setsockopt(sock, SOL_SOCKET, SO_ATTACH_BPF, &prog_fd, sizeof(prog_fd))
where prog_fd was received from syscall bpf(BPF_PROG_LOAD, attr, ...)
and attr->prog_type == BPF_PROG_TYPE_SOCKET_FILTER
setsockopt() calls bpf_prog_get() which increments refcnt of the program,
so it doesn't get unloaded while socket is using the program.
The same eBPF program can be attached to multiple sockets.
User task exit automatically closes socket which calls sk_filter_uncharge()
which decrements refcnt of eBPF program
Signed-off-by: Alexei Starovoitov <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Following patch fixes typo in the flow validation. This prevented
installation of ARP and IPv6 flows.
Fixes: 19e7a3df72 ("openvswitch: Fix NDP flow mask validation")
Signed-off-by: Pravin B Shelar <[email protected]>
Reviewed-by: Thomas Graf <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Set the inner mac header to point to the GRE payload when
doing GRO. This is needed if we proceed to send the packet
through GRE GSO which now uses the inner mac header instead
of inner network header to determine the length of encapsulation
headers.
Fixes: 14051f0452a2 ("gre: Use inner mac length when computing tunnel length")
Reported-by: Wolfgang Walter <[email protected]>
Signed-off-by: Tom Herbert <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Pablo Neira Ayuso says:
====================
Netfilter updates for net-next
The following batch contains netfilter updates for net-next. Basically,
enhancements for xt_recent, skip zeroing of timer in conntrack, fix
linking problem with recent redirect support for nf_tables, ipset
updates and a couple of cleanups. More specifically, they are:
1) Rise maximum number per IP address to be remembered in xt_recent
while retaining backward compatibility, from Florian Westphal.
2) Skip zeroing timer area in nf_conn objects, also from Florian.
3) Inspect IPv4 and IPv6 traffic from the bridge to allow filtering using
using meta l4proto and transport layer header, from Alvaro Neira.
4) Fix linking problems in the new redirect support when CONFIG_IPV6=n
and IP6_NF_IPTABLES=n.
And ipset updates from Jozsef Kadlecsik:
5) Support updating element extensions when the set is full (fixes
netfilter bugzilla id 880).
6) Fix set match with 32-bits userspace / 64-bits kernel.
7) Indicate explicitly when /0 networks are supported in ipset.
8) Simplify cidr handling for hash:*net* types.
9) Allocate the proper size of memory when /0 networks are supported.
10) Explicitly add padding elements to hash:net,net and hash:net,port,
because the elements must be u32 sized for the used hash function.
Jozsef is also cooking ipset RCU conversion which should land soon if
they reach the merge window in time.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next
|
|
When the LE Read Local P-256 Public Key command is supported, then
enable its corresponding complete event. And when the LE Generate DHKey
command is supported, enable its corresponding complete event as well.
Signed-off-by: Marcel Holtmann <[email protected]>
Signed-off-by: Johan Hedberg <[email protected]>
|