Grant Likely [Tue, 10 Apr 2012 23:48:42 +0000 (17:48 -0600)]
irq_domain: Fix debug output to use "irq" instead of "virq"
Back when irq_domain was a powerpc-only thing, the term "virq", or
virtual irq, was used to differentiate irq numbers from hardware irq
numbers. However, virqs are synonymous with Linux irq numbers, so
using a different term is confusing. This patch changes the debug
output to use the header "irq" instead of "virq" for the irq number
output column.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Grant Likely [Tue, 3 Apr 2012 13:11:04 +0000 (07:11 -0600)]
irq: Kill pointless irqd_to_hw export
It makes no sense to export this trivial function. Make it a static inline
instead.
This patch also drops virq_to_hw from arch/c6x since it is unused by that
architecture.
v2: Move irq_hw_number_t into types.h to fix ARM build failure
Signed-off-by: Grant Likely <grant.likely@secretlab.ca> Acked-by: Thomas Gleixner <tglx@linutronix.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
David Daney [Thu, 5 Apr 2012 23:52:13 +0000 (16:52 -0700)]
irq/irq_domain: Quit ignoring error returns from irq_alloc_desc_from().
In commit 4bbdd45a (irq_domain/powerpc: eliminate irq_map; use
irq_alloc_desc() instead) code was added that ignores error returns
from irq_alloc_desc_from() by (silently) casting the return value to
unsigned. The negitive value error return now suddenly looks like a
valid irq number.
Commits cc79ca69 (irq_domain: Move irq_domain code from powerpc to
kernel/irq) and 1bc04f2c (irq_domain: Add support for base irq and
hwirq in legacy mappings) move this code to its current location in
irqdomain.c
The result of all of this is a null pointer dereference OOPS if one of
the error cases is hit.
The fix: Don't cast away the negativeness of the return value and then
check for errors.
Signed-off-by: David Daney <david.daney@cavium.com> Acked-by: Rob Herring <rob.herring@calxeda.com>
[grant.likely: dropped addition of new 'irq' variable] Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Dan Williams [Fri, 6 Apr 2012 20:41:15 +0000 (13:41 -0700)]
kobject: provide more diagnostic info for kobject_add_internal() failures
1/ convert open-coded KERN_ERR+dump_stack() to WARN(), so that automated
tools pick up this warning.
2/ include the 'child' and 'parent' kobject names. This information was
useful for tracking down the case where scsi invoked device_del() on a
parent object and subsequently invoked device_add() on a child. Now the
warning looks like:
Dan Williams [Fri, 6 Apr 2012 20:41:06 +0000 (13:41 -0700)]
sysfs: handle 'parent deleted before child added'
In scsi at least two cases of the parent device being deleted before the
child is added have been observed.
1/ scsi is performing async scans and the device is removed prior to the
async can thread running (can happen with an in-opportune / unlikely
unplug during initial scan).
2/ libsas discovery event running after the parent port has been torn
down (this is a bug in libsas).
In this scenario the parent is still valid (because we have a
reference), but it has been device_del()'d which means its kobj->sd
pointer is NULL'd via:
device_del()->kobject_del()->sysfs_remove_dir()
...and then sysfs_create_dir() (without this fix) goes ahead and
de-references parent_sd via sysfs_ns_type():
Thomas Abraham [Sun, 25 Mar 2012 15:02:49 +0000 (20:32 +0530)]
of/irq: add empty irq_of_parse_and_map() for non-dt builds
Add a empty irq_of_parse_and_map() function that returns 0 for non-dt
builds and avoid having #ifdef CONFIG_OF around all calls to
irq_of_parse_and_map(). In addition to that, the irq_of_parse_and_map()
function declaration is made available only if CONFIG_OF_IRQ is defined,
which is the same config option that makes the irq_of_parse_and_map()
function definition available. While at it, fix a typo as well.
Changes since v1:
- Moved irq_of_parse_and_map() function declaration under CONFIG_OF_IRQ.
- Fix a minor typo in comments.
Suggested-by: Grant Likely <grant.likely@secretlab.ca> Signed-off-by: Thomas Abraham <thomas.abraham@linaro.org> Acked-by: Rob Herring <rob.herring@calxeda.com> Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Johan Hovold [Tue, 20 Mar 2012 15:59:33 +0000 (16:59 +0100)]
USB: serial: fix race between probe and open
Fix race between probe and open by making sure that the disconnected
flag is not cleared until all ports have been registered.
A call to tty_open while probe is running may get a reference to the
serial structure in serial_install before its ports have been
registered. This may lead to usb_serial_core calling driver open before
port is fully initialised.
With ftdi_sio this result in the following NULL-pointer dereference as
the private data has not been initialised at open:
Roland Stigge [Wed, 4 Apr 2012 00:02:58 +0000 (02:02 +0200)]
gpio: Fix range check in of_gpio_simple_xlate()
of_gpio_simple_xlate() has an off-by-one bug where it checks to see if
args[0] is > ngpio instead of >=. args[0] must always be less than
ngpio because it is a zero-based enumeration.
Signed-off-by: Roland Stigge <stigge@antcom.de>
[grant.likely: beef up commit text] Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Grant Likely [Wed, 28 Mar 2012 20:55:04 +0000 (14:55 -0600)]
gpio/sodaville: Convert sodaville driver to new irqdomain API
The irqdomain api changed significantly in v3.4 which caused a build
failure for this driver.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca> Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Hans J. Koch <hjk@linutronix.de> Cc: Torben Hohn <torbenh@linutronix.de>
Javier Cardona [Sat, 31 Mar 2012 18:31:33 +0000 (11:31 -0700)]
{nl,cfg}80211: Support for mesh synchronization
Report Toffset to userspace.
Let userspace select the mesh synchronization method.
Signed-off-by: Marco Porsch <marco.porsch@s2005.tu-chemnitz.de> Signed-off-by: Pavel Zubarev <pavel.zubarev@gmail.com> Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch adds MBSS extensible synchronization framework (Sec.
13.13.2 of IEEE Std. 802.11-2012).
The framework is implemented via an ops table which defines the
following functions:
rx_bcn_presp() - this is called every time a mesh beacon is
received.
adjust_tbtt() - this is called immediately before a beacon is about
to be transmitted.
The default neighbor offset synchronization defined in the standard is
implemented. We also provide template functions for vendor specific
methods.
When neighbor offset synchronization is active (which is the default)
mesh neighbors in the same MBSS will track timing offsets to each other
and compensate clock drift.
In our tests we observed that this mesh synchronization implementation
successfully corrected drifts between stations of ~2PPM while
introducing a jitter of ~20us.
It is also possible to test this framework on mac80211_hwsim simulated
phys to see how it behaves under different topologies, over poor links,
etc.
Signed-off-by: Marco Porsch <marco.porsch@s2005.tu-chemnitz.de> Signed-off-by: Pavel Zubarev <pavel.zubarev@gmail.com> Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Javier Cardona [Sat, 31 Mar 2012 18:31:31 +0000 (11:31 -0700)]
mac80211: Allow tsf increments via debugfs
Reading and writing back the tsf value via tsf is too slow if one wants
to make small increments to this timer. With this change you can use
the syntax "+=<some value>" or "-=<some value>" to add or substract a
value from the tsf counter.
Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Javier Cardona [Sat, 31 Mar 2012 18:31:30 +0000 (11:31 -0700)]
mac80211_hwsim: Fill timestamp beacon at the time it is transmitted
Generate more acurate tsf values in hwsim by setting the tsf value on
trasmitted beacons immediately before they are moved to the rx path.
Also, adjust the beacon timestamp to be the time at which the first byte
of the timestamp is transmitted.
With these changes the observed tsf offset between two hwsim/mesh peers
is 0 (unless the offset is modified via debugfs)
Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
While associated we should never have empty SSID, but life can be full
of surprises, and is allways better to print a warning than crash.
Before memcpy() in ieee80211_probereq_get() check ssid_len instead of
ssid pointer, sice pointer it always passed by "ssidie + 2" expression
to send probe functions, so practically never can be NULL.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 09:04:28 +0000 (11:04 +0200)]
mac80211: lazily stop queues in add_pending
When adding pending SKBs there's no need to
stop all queues, we only need to stop those
that we're adding frames to. Implement that
by lazily stopping a queue as we add an SKB.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 09:04:26 +0000 (11:04 +0200)]
mac80211: decouple # of netdev queues from HW queues
When we get more hardware queues, we'll still want
to only have netdev queues per AC, so set it up in
that way. If the hardware doesn't support QoS (by
not supporting at least 4 queues) the netdevs get
a single queue only (this is no change in behavior
as there are no drivers with 2 or 3 queues today.)
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 09:04:25 +0000 (11:04 +0200)]
mac80211: refuse TX queue configuration on non-QoS HW
Drivers that don't support QoS also don't support
setting up their ACs, catch that early. While at
it, remove the input check since cfg80211 does it
now.
Also fix up the restart code to not try to set up
the queues in this case.
Finally also change the tx_conf array to have
IEEE80211_NUM_ACS entries instead of # of queues
since that's what it really needs.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 09:04:24 +0000 (11:04 +0200)]
cfg80211/nl80211: clarify TX queue API
With the plan to change mac80211's queue API to
not map ACs to queues 1:1, it seems necessary to
clarify some APIs that act on ACs rather than on
queues to spell that out explicitly. Do this.
Also verify that the AC number given is valid.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 09:04:23 +0000 (11:04 +0200)]
mac80211: remove antenna_sel_tx TX info field
This field is never set to anything non-zero in
mac80211, so we should be able to remove it.
Unfortunately though, the iwlwifi and iwlegacy
drivers use it for their internal TX status
processing (which shouldn't be using the rate
control API to start with), so add a new field
"status.antenna" for them, at least for now.
In the future, I plan to use the new field to
hold the hardware queue, while the SKB's queue
mapping holds the AC.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Fri, 30 Mar 2012 06:43:32 +0000 (08:43 +0200)]
mac80211: notify driver of rate control updates
Devices that have internal rate control need to be
notified when the bandwidth or SMPS state changes
just like external rate control algorithms get a
notification now.
Add this notification and clarify the change bits
while at it, the HT_CHANGED bit really meant only
bandwidth changed.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 08:58:38 +0000 (10:58 +0200)]
mac80211: remove queue stop on rate control update
We currently stop the queue when changing the rate
control between 20/40 MHz in the BSS. This seems to
have been necessary when we actually changed the
channel, but now that we just update the station it
doesn't seem right any more. Remove it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 08:58:37 +0000 (10:58 +0200)]
mac80211: remove channel type argument from rate_update
The channel type argument to the rate_update()
callback isn't really the correct way to give
the rate control algorithm about the desired
RX bandwidth of the peer.
Remove this argument, and instead update the
STA capabilities with 20/40 appropriately. The
SMPS update done by this callback works in the
same way, so this makes the callback cleaner.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Wed, 28 Mar 2012 08:58:36 +0000 (10:58 +0200)]
mac80211: set HT channel before association
Changing the channel type during operation is
confusing to some drivers and will be hard to
handle in multi-channel scenarios. Instead of
changing the channel, set it to the right HT
channel before authenticating/associating and
don't change it -- just update the 20/40 MHz
restrictions in rate control as needed when
changed by the AP.
This also fixes a problem that Paul missed in
his fix for the "regulatory makes us deaf"
issue -- when we couldn't use 40 MHz we still
associated saying we were using 40 MHz, which
could in similarly broken APs make us never
even connect successfully.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Tue, 27 Mar 2012 12:07:59 +0000 (14:07 +0200)]
mwifiex: don't use IEEE80211_MAX_QUEUES
IEEE80211_MAX_QUEUES is an internal mac80211 value,
it is not guaranteed to be always 4. The firmware
API in mwifiex almost certainly doesn't care about
mac80211 changing though, so mwifiex shouldn't use
this value.
Maybe it should use IEEE80211_NUM_ACS instead and
that is what I'm doing here as at least that value
will probably never change, but maybe it should
have its own define instead.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Acked-by: Bing Zhao <bzhao@marvell.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Not all devices are really capable of implementing
remain-on-channel, even if it is implemented in SW,
as they can't necessarily deal with channel changes
while associated.
Remove the WIPHY_FLAG_HAS_REMAIN_ON_CHANNEL and add
it only if either the driver has remain_on_channel
implemented in the driver/device.
Also add it to all drivers that advertise P2P right
now since those definitely have to have it working.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Acked-by: Luciano Coelho <coelho@ti.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Larry Finger [Mon, 26 Mar 2012 15:48:20 +0000 (10:48 -0500)]
rtlwifi: Add missing DMA buffer unmapping for PCI drivers
In https://bugzilla.kernel.org/show_bug.cgi?id=42976, a system with driver
rtl8192se used as an AP suffers from "Out of SW-IOMMU space" errors. These
are caused by the DMA buffers used for beacons never being unmapped.
This bug was also reported at
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/961618
Reported-and-Tested-by: Da Xue <da@lessconfused.com> Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Cc: Stable <stable@vger.kernel.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Oliver Hartkopp [Sun, 25 Mar 2012 06:43:24 +0000 (08:43 +0200)]
iwlwifi: fix unused variable warning
In the case of disabled CONFIG_IWLWIFI_DEBUGFS option the compiler complains
about the unused variable 'img'. Fix this by moving the 'img' definition.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
tcp: fix tcp_rcv_rtt_update() use of an unscaled RTT sample
Fix a code path in tcp_rcv_rtt_update() that was comparing scaled and
unscaled RTT samples.
The intent in the code was to only use the 'm' measurement if it was a
new minimum. However, since 'm' had not yet been shifted left 3 bits
but 'new_sample' had, this comparison would nearly always succeed,
leading us to erroneously set our receive-side RTT estimate to the 'm'
sample when that sample could be nearly 8x too high to use.
The overall effect is to often cause the receive-side RTT estimate to
be significantly too large (up to 40% too large for brief periods in
my tests).
Signed-off-by: Neal Cardwell <ncardwell@google.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Tue, 10 Apr 2012 00:56:42 +0000 (00:56 +0000)]
tcp: restore correct limit
Commit c43b874d5d714f (tcp: properly initialize tcp memory limits) tried
to fix a regression added in commits 4acb4190 & 3dc43e3,
but still get it wrong.
Result is machines with low amount of memory have too small tcp_rmem[2]
value and slow tcp receives : Per socket limit being 1/1024 of memory
instead of 1/128 in old kernels, so rcv window is capped to small
values.
Fix this to match comment and previous behavior.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Glauber Costa <glommer@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Users have reported connection failures in 3.3.1 and suspend/resume
failures in 3.4-rcX. Revert this commit for now - PS IDLE can be
fixed in a clean manner later on.
Cc: stable@vger.kernel.org Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Chen, Chien-Chia [Thu, 29 Mar 2012 10:21:47 +0000 (18:21 +0800)]
rt2x00: Fix rfkill_polling register function.
Move rt2x00rfkill_register(rt2x00dev) to rt2x00lib_probe_dev
function. It fixes of starting rfkill_poll function at the
right time if sets hard rfkill block and reboot. rt2x00mac_rfkill_poll
should be starting before bringing up the wireless interface.
Signed-off-by: Chen, Chien-Chia <machen@suse.com> Acked-by: Helmut Schaa <helmut.schaa@googlemail.com> CC: Kevin Chou <kevin.chou@mediatek.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Paul Gortmaker [Mon, 26 Mar 2012 00:02:55 +0000 (20:02 -0400)]
bcma: fix build error on MIPS; implicit pcibios_enable_device
The following is seen during allmodconfig builds for MIPS:
drivers/bcma/driver_pci_host.c:518:2: error: implicit declaration
of function 'pcibios_enable_device' [-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors
make[3]: *** [drivers/bcma/driver_pci_host.o] Error 1
Add the header instead of implicitly assuming it will be present.
Sounds like a good idea, but that alone doesn't fix anything.
The real problem is that the Kconfig has settings related to whether
PCI is possible, i.e.
config BCMA_HOST_PCI_POSSIBLE
bool
depends on BCMA && PCI = y
default y
config BCMA_HOST_PCI
bool "Support for BCMA on PCI-host bus"
depends on BCMA_HOST_PCI_POSSIBLE
...but what is missing is that BCMA_DRIVER_PCI_HOSTMODE doesn't
have any dependencies on the above. Add one.
CC: Hauke Mehrtens <hauke@hauke-m.de> CC: John W. Linville <linville@tuxdriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Acked-by: Hauke Mehrtens <hauke@hauke-m.de> Signed-off-by: John W. Linville <linville@tuxdriver.com>
2) Make the sparc side changes that go along with the infrastructure to
retry faults when blocking on a disk transfer. From Kautuk Consul.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
sparc32,leon: fix leon build
sparc/mm/fault_32.c: Port OOM changes to do_sparc_fault
sparc/mm/fault_64.c: Port OOM changes to do_sparc64_fault
projects / karo-tx-linux.git / log