ip_vs_dest_hold will be used under RCU lock
while ip_vs_dest_put can be called even after dest
is removed from service, as it happens for conns and
some schedulers.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Allow schedulers to use rcu_dereference when
returning destination on lookup. The RCU read-side critical
section will allow ip_vs_bind_dest to get dest refcnt as
preparation for the step where destinations will be
deleted without an IP_VS_WAIT_WHILE guard that holds the
packet processing during update.
Add new optional scheduler methods add_dest,
del_dest and upd_dest. For now the methods are called
together with update_service but update_service will be
removed in a following change.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
The global list with schedulers ip_vs_schedulers
is accessed only from user context - configuration and
scheduler module [un]registration. Use ip_vs_sched_mutex
instead.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
We have many fields to set and few to reset,
use kmem_cache_alloc instead to save some cycles.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
__ip_vs_conn_in_get and ip_vs_conn_out_get are
hot places. Optimize them, so that ports are matched first.
By moving net and fwmark below, on 32-bit arch we can fit
caddr in 32-byte cache line and all addresses in 64-byte
cache line.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Convert __ip_vs_conntbl_lock_array as follows:
- readers that do not modify conn lists will use RCU lock
- updaters that modify lists will use spinlock_t
Now for conn lookups we will use RCU read-side
critical section. Without using __ip_vs_conn_get such
places have access to connection fields and can
dereference some pointers like pe and pe_data plus
the ability to update timer expiration. If full access
is required we contend for reference.
We add barrier in __ip_vs_conn_put, so that
other CPUs see the refcnt operation after other writes.
With the introduction of ip_vs_conn_unlink()
we try to reorganize ip_vs_conn_expire(), so that
unhashing of connections that should stay more time is
avoided, even if it is for very short time.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Allow the readers to use RCU lock and for
PE module registrations use global mutex instead of
spinlock. All PE modules need to use synchronize_rcu
in their module exit handler.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
rs_lock was used to protect rs_table (hash table)
from updaters (under global mutex) and readers (packet handlers).
We can remove rs_lock by using RCU lock for readers. Reclaiming
dest only with kfree_rcu is enough because the readers access
only fields from the ip_vs_dest structure.
Use hlist for rs_table.
As we are now using hlist_del_rcu, introduce in_rs_table
flag as replacement for the list_empty checks which do not
work with RCU. It is needed because only NAT dests are in
the rs_table.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
We use locks like tcp_app_lock, udp_app_lock,
sctp_app_lock to protect access to the protocol hash tables
from readers in packet context while the application
instances (inc) are [un]registered under global mutex.
As the hash tables are mostly read when conns are
created and bound to app, use RCU for readers and reclaim
app instance after grace period.
Simplify ip_vs_app_inc_get because we use usecnt
only for statistics and rely on module refcounting.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Currently when forwarding requests to real servers
we use dst_lock and atomic operations when cloning the
dst_cache value. As the dst_cache value does not change
most of the time it is better to use RCU and to lock
dst_lock only when we need to replace the obsoleted dst.
For this to work we keep dst_cache in new structure protected
by RCU. For packets to remote real servers we will use noref
version of dst_cache, it will be valid while we are in RCU
read-side critical section because now dst_release for replaced
dsts will be invoked after the grace period. Packets to
local real servers that are passed to local stack with
NF_ACCEPT need a dst clone.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Consolidate the PMTU checks, ICMP sending and
skb_dst modification in __ip_vs_get_out_rt and
__ip_vs_get_out_rt_v6. Now skb_dst is changed early
to simplify the transmitters.
Make sure update_pmtu is called only for local clients.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
We run in contexts like ip_rcv, ipv6_rcv, br_handle_frame,
do not expect shared skbs.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
After commit 70e7341673 (ipv4: Show that ip_send_reply()
is purely unicast routine.) we do not need to reroute DNAT-ed
traffic over loopback because reply uses iph daddr and not
rt_spec_dst.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Move and give better names to two functions:
- ip_vs_dst_reset to __ip_vs_dst_cache_reset
- __ip_vs_dev_reset to ip_vs_forget_dev
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
It was a bad idea to hide return statements in macros.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
The real server becomes unreachable on down event,
no need to wait device unregistration. Should help in
releasing dsts early before dst->dev is replaced with lo.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Avoid replacing the cached route for real server
on every packet with different TOS. I doubt that routing
by TOS for real server is used at all, so we should be
better with such optimization.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Rename skb_dst_set_noref to __skb_dst_set_noref and
add force flag as suggested by David Miller. The new wrapper
skb_dst_set_noref_force will force dst entries that are not
cached to be attached as skb dst without taking reference
as long as provided dst is reclaimed after RCU grace period.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Simon Horman <horms@verge.net.au>
PCIe. On some devices this means that Ethernet will go uninitialized
and other devices will fail. This pull request fixes it with a single
patch to pass the proper base address in the Tegra clock driver.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=WU9c
-----END PGP SIGNATURE-----
Merge tag 'clk-fixes-for-linus' of git://git.linaro.org/people/mturquette/linux
Pull tegra clock driver fix from Mike Turquette:
"Missing base address in Tegra clock driver results in non-operational
PCIe. On some devices this means that Ethernet will go uninitialized
and other devices will fail. This pull request fixes it with a single
patch to pass the proper base address in the Tegra clock driver."
* tag 'clk-fixes-for-linus' of git://git.linaro.org/people/mturquette/linux:
clk: tegra: Allow PLLE training to succeed
The Kconfig symbol SND_SOC_OF_SIMPLE got removed in commit
f0fba2ad1b ("ASoC: multi-component - ASoC
Multi-Component Support"). But that commit missed one instance. Remove
it now, together with the prompt it has effectively hidden ever since.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
When clearing the walked flags there is no need to clear all paths, we
only need to clear the paths we actually walked. This means we can split
dapm_clear_walk() into input and output versions and rather than going
through all DAPM paths we can recurse down the path until we encounter
paths we have not yet walked.
This reduces the number of operations we need to perform and improves
cache locality.
[Pulled out of the vendor tree that the patch was originally generated
for by me, any bugs were introduced in that process -- broonie]
Signed-off-by: Ryo Tsutsui <Ryo.Tsutsui@wolfsonmicro.com>
Acked-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
We already clear the walked state in dapm_widget_power_check(), no need
to do it again.
Signed-off-by: Ryo Tsutsui <Ryo.Tsutsui@wolfsonmicro.com>
Acked-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
In some pathological use cases users may insert an accessory very slowly
causing multiple indeterminate measurements. Handle this by retrying many
measurements before we give up and declare a headphone.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
These patches modify the common probe routine to recognize the RTL8188EE
chip and implement asynchronous firmware reading in the callback routine
to initialize the sw variables.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: jcheung@suse.com
Cc: machen@suse.com
Cc: mmarek@suse.cz
Cc: zhiyuan_yang@realsil.com.cn
Cc: page_he@realsil.com.cn
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch combines the remaining changes in the rtlwifi family to handle
the addition of rtl8188ee. A number of these changes eliminate some CamelCase
variable names, and other shorten common variable names so that long lines
in the new driver could be shortened.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: jcheung@suse.com
Cc: machen@suse.com
Cc: mmarek@suse.cz
Cc: zhiyuan_yang@realsil.com.cn
Signed-off-by: John W. Linville <linville@tuxdriver.com>
In commit a5ffbe0, some of the calls to rtl_lps_leave() were switched
to be called from a work queue to avoid a scheduling while atomic bug.
This patch converts the remaining calls to use the work queue. In
addition, the call to rtl_lps_enter() is also switched to the work
queue. None of these newly converted calls had triggered the bug (yet),
but this change make all of them fit a single pattern.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: jcheung@suse.com
Cc: machen@suse.com
Cc: mmarek@suse.cz
Cc: zhiyuan_yang@realsil.com.cn
Cc: page_he@realsil.com.cn
Signed-off-by: John W. Linville <linville@tuxdriver.com>
These changes add the new variables for P2P and modify the various struct
definitions for other new features.
This patch updates files base.{c,h} for the changes in the newest vendor
driver.
This patch updates files ps.{c,h} for the changes in the newest vendor
driver.
This patch updates files debug.{c,h}, efuse.c, pci.{c,h}, and wifi.h for
the changes in the newest vendor driver.
This patch updates files core.c, ps.c, rc.c, and wifi.h for
the changes in the newest vendor driver.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: jcheung@suse.com
Cc: machen@suse.com
Cc: mmarek@suse.cz
Cc: zhiyuan_yang@realsil.com.cn
Cc: page_he@realsil.com.cn
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Some platforms only provide their PCI ROM via a platform-specific interface.
Fall back to attempting that if all other sources fail.
Signed-off-by: Matthew Garrett <matthew.garrett@nebula.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Some platforms only provide their PCI ROM via a platform-specific interface.
Fall back to attempting that if all other sources fail.
Signed-off-by: Matthew Garrett <matthew.garrett@nebula.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
The 'rt2800_validate_eeprom' function uses the type of
the RT chipset for verifying the number of RX streams
on RT28x0 devices. However the type of the RT chipset
is not yet detected when the 'rt2800_validate_eeprom'
function is called.
Move the RT chipset detection code into a separate helper
function, and call it before rt2800_validate_eeprom.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Use the newly introduced rt2x00_set_{rf,rt} helpers
to set the RT and RF chipset separately.
This change makes it possible to move the RT
detection code into another function which will
be done in a subseqent patch.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The new helpers can be used to set the type of the
RT and RF chipsets separately.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This fixes some sparse warnings.
b43_nphy_set_rxantenna() was not used anywhere.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Newer 11ac enabled chipsets have more TX and RX buffers in FW
and hardware; so they may support larger TX and RX window sizes
for BA. Reset BA settings during association, adhoc join/start
or start_ap() if we are joining/creating 11ac network.
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Sagar Bijwe <bsagar@marvell.com>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch fixes an issue where RX throughput values observed
were substantially lower than TX counterparts for PCIe8897 STA.
PCIe8897 supports larger rx_win_size. After changing these values
we see big improvement for TX and RX throughput values.
Different tx_win_size and rx_win_size are used for AP mode.
All BA setup related initialization has been moved to separate
function.
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Sagar Bijwe <bsagar@marvell.com>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Limit the channel number in scan request, or the driver scan
config structure memory will be overflowed.
Cc: <stable@vger.kernel.org> # 3.5+
Signed-off-by: Stone Piao <piaoyun@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Adds a function to read the various GPMC chip-select settings from
device-tree and store them in the gpmc_settings structure.
Update the GPMC device-tree binding documentation to describe these
options.
Signed-off-by: Jon Hunter <jon-hunter@ti.com>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
With the addition of the gpmc_cs_program_settings(), we no longer need
or use gpmc_cs_configure() to configure some of the GPMC chip-select
options. So rename the function to gpmc_configure() and remove code that
modifies options in the CONFIG1 register.
Signed-off-by: Jon Hunter <jon-hunter@ti.com>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Convert the OMAP2+ TUSB code to use the gpmc_cs_program_settings()
function for configuring the various GPMC options instead of directly
programming the CONFIG1 register.
Signed-off-by: Jon Hunter <jon-hunter@ti.com>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Convert the OMAP2+ SMC91x code to use the gpmc_cs_program_settings()
function for configuring the various GPMC options instead of directly
programming the CONFIG1 register.
Move configuration of the GPMC settings outside retime function and
this does not need to be done if the timings are changed dynamically
at runtime.
Signed-off-by: Jon Hunter <jon-hunter@ti.com>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Convert the OMAP2+ NAND code to use the gpmc_cs_program_settings()
function for configuring the various GPMC options instead of directly
programming the CONFIG1 register.
This moves the configuration of some GPMC options outside the
nand_gpmc_retime() because these options should only need to be set once
regardless of whether the gpmc timing is changing dynamically at runtime.
The programming of where the wait-pin is also moved slightly, but this
will not have any impact to existing devices as no boards are currently
setting the dev_ready variable.
Signed-off-by: Jon Hunter <jon-hunter@ti.com>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>