If cable is not connected to peripheral only board when initializing the
gadget driver, then runtime pm calls are out-of-sync and the musb cannot
idle with omap2430.c. This was noted on Nokia N900 where musb prevented the
CPU to be able to enter deeper retention idle state.
This was working in 2.6.38 before runtime pm conversions but there musb
smart standby/idle modes were configured statically where they are now
updated runtime depending on use and cable status.
Reason for out-of-sync is that runtime pm is activated in function
musb_gadget.c: usb_gadget_probe_driver but suspended only in OTG mode if
cable is not connected when initializing. In peripheral only mode this leads
to out-of-sync runtime pm since runtime pm remain active and is activated
another time in omap2430.c: musb_otg_notifications for VBUS Connect event
and thus cannot suspend for VBUS Disconnect event since the use count remains
active.
Fix this by moving cable status check and pm_runtime_put call in
usb_gadget_probe_driver out of is_otg_enabled block.
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Recent runtime pm and hwmod conversions for 2.6.39 broke the musb peripheral
mode OMAP retention idle on boards where the board mode in struct
musb_hdrc_platform_data is set to MUSB_PERIPHERAL.
These conversions changed the way how the OTG_SYSCONFIG register is
configured and used in runtime. Before 2.6.39 smart standby/idle modes were
activated statically in OTG_SYSCONFIG. Those modes allow that the musb is
able to idle when peripheral device is not connected to host.
In 2.6.39 the OTG_SYSCONFIG is updated runtime depending on VBUS status.
No standby/idle modes are used when device is connected and force
standby/idle when disconnected.
Unfortunately VBUS disconnect event that handles the disconnect case lets
the peripheral musb to idle only when board mode is MUSB_OTG. Fix this by
checking the peripheral mode also.
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This reverts commit c6b358748e.
It turned out that there are different pin configurations for this
PCI SSID, including multi-channel modes. And more proper fix for
allowing line-out mutes will come up in 2.6.40 tree, so we won't need
this fixup any more there.
Reported-by: Andrew Clayton <andrew@digital-domain.net>
Reported-by: Emmanuel Benisty <benisty.e@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
By reading the "driver_state" debugfs value we get all the important
state information from the wl12xx driver. This helps testing and
debugging, particularly in situations where the driver seems "stuck".
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When operating as AP, the TX queues are not stopped when we start
recovery. mac80211 is notified only after the fact. When there is
pending TX, it will be queued even after the FW is down. This leads to
situations where the TX queues are stopped (because of the TX-watermark
mechanism), and are never woken up when we return from recovery.
Fix this by explicitly stopping the TX queues when before initiating
recovery.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
This entry is useful for debugging the driver state machine during
recovery.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When there's a change in the basic rates of the AP, reconfigure relevant
templates with the new rates.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Use the minimal rate configured in the basic rates set as the AP
broadcast and multicast rate. The minimal rate is used to ensure weak
links can still communicate.
When the basic rates contains at least one OFDM rate, configure all
unicast TX rates to OFDM only.
Unify rate configuration on initialization and on change notification
into a single function.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When performing recovery, print the firmware version and program
counter (by reading the SCR_PAD4 register). The value of the firmware
program counter during assert can be useful for debugging.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
We use a long timeout (2 seconds) when sending commands to the FW.
When a command times out, it means the FW is stuck, and we should
commence recovery.
This should make recovery times shorter as we'll recover on the first
timeout indication.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
New AP-mode FWs filter external beacons by default. Disable this
filtering on start up so we can properly configure ERP protection.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Use the wiphy RTS and fragmentation thresholds for initializing the FW
when possible. This mitigates a bug where previously set values are
forgotten after interface down/up.
Add checks before settings these values to ensure they are valid. Use
default values when invalid thresholds are configured.
Update the default RTS threshold to the maximum value given by the
specification.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Initialize AP specific BT coexitance parameters to default values and
enable them in AP mode.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
It is preferred to use the setter that to set queue_mapping directly.
This also helps backporting in compat-wireless.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When packets arrive with a RX descriptor indicating corruption, discard
them.
In general white-list the RX descriptor status to prevent rouge data
from being sent up.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Frames are considered pending when they reside in the driver TX queue or
already queued in the FW.
This notion of "pending" is appropriate for power save considerations in
STA mode, but not necessarily in other modes (for instance P2P-GO).
[Fixed a sparse warning about missing "static" in a function
declaration -- Luca]
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The FW can dynamically manage its internal TX & RX memory pools, moving
blocks from one pool to another when necessary. This can significantly
improve performance. Currently this feature is enabled only for 128x.
Enable dynamic memory for 127x as well. Other parameters in the memory
configuration structure may need to be fine tuned, as the optimal values
for these may change once dynamic memory is enabled.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The driver stops sending TX packets when either there aren't enough
memory blocks, or it runs out of TX descriptors. The driver continues to
send packets to the FW only when more memory blocks are available.
The FW might free TX descriptors without freeing the corresponding
memory blocks, especially when dynamic memory is enabled. In cases where
memory blocks are not freed at all, the driver will keep waiting for
more memory blocks indefinitely.
Fix this by clearing the WL1271_FLAG_FW_TX_BUSY flag when there are
available TX descriptors.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The 128x/AP firmware does not yet support dynamic memory. Temporarily,
the memory configuration for the 127x was used both for 127x/AP as well
as 128x/AP. Since the two chips don't have the same number of memory
blocks, TP was significantly degraded.
This hasn't been fine tuned yet, but using the base 128x numbers
(without dynamic memory) seems to yield much better results (around 30%
more). Additional fine tuning will be required in the future.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When configuring ACX_WAKE_UP_CONDITIONS (before entering psm), we
tell the firmware to wake up once in N DTIMs/beacons.
Allow control of this value via debugfs (for debugging purposes).
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Reducing the retries of sending PS exit packets to the peer AP.
That fix is to avoid sending unrealizable number of PS exit packets
in case of ap lost.
Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Add support to FM WLAN coexistence (STA only). Some WiFi harmonics
may interfere with FM operation, to avoid this problem special
coexistence techniques are activated around some FM frequencies.
Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When working in ibss mode, we don't configure rate policy per station
(as we use the same link for multiple stations), so currently the
1mb/s rate is being used.
Instead, configure the firmware to use the whole 11b rates by default.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
ieee80211_reconfig() sets most of the "changed" flags regardless
of the actual change (e.g. BSS_CHANGED_ASSOC will be set even if
the interface is still not associated). in this case the driver
will issue some unneeded commands.
Since the driver relies solely on the BSS_CHANGED_ASSOC flag,
without checking if there was an actual change, it will end up
issuing unjoin() and dummy_join() commands, although it was
never associated and should just remain idle.
Avoid it by checking the actual state change, in addition to the
"changed" flag.
(there seem to be more redundant configuration commands being
issued, but they shouldn't harm)
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When debugging, reduce the padding size from each rx packet, to
get the actual packet size (so comparing it against a cap file
will be easier)
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Fix mismatch between the REF CLK and TCXO CLK information that is
set in the platform data and the NVS, so we override what comes
from the NVS and replace it with what comes from the platform data.
[Small fix in a comment -- Luca]
Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The elp_work is being enqueued on wl1271_ps_elp_sleep, but doesn't get
cancelled on wl1271_ps_elp_wakeup. This might cause immediate entrance
to elp when the wl->mutex is being released, rather than using the delayed
enqueueing optimization.
Cancel elp_work on wakeup request, and add a new WL1271_FLAG_ELP_REQUESTED
flag to further synchronize the elp actions.
[Fixed a couple of typos in some comments -- Luca]
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
commit d05c806 ("wl12xx: rearrange some ELP wake_up/sleep calls")
introduced a bug in which wl1271_ps_elp_wakeup() was called instead
of wl1271_ps_elp_sleep() after completing the tx work.
Reported-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
End-of-transaction flag should be set when working with
wl127x chip on AP mode.
Thanks Ido Yariv <ido@wizery.com> for the guidance with that.
Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
fib_trie_table() is called during netns creation and
Chromium uses clone(CLONE_NEWNET) to sandbox renderer process.
Don't print anything.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes the following oops discovered by Dan Aloni:
> Anyway, the following is the output of the Oops that I got on the
> Ubuntu kernel on which I first detected the problem
> (2.6.37-12-generic). The Oops that followed will be more useful, I
> guess.
>[ 5594.669852] BUG: unable to handle kernel NULL pointer dereference
> at (null)
> [ 5594.681606] IP: [<ffffffff81550b7b>] unix_dgram_recvmsg+0x1fb/0x420
> [ 5594.687576] PGD 2a05d067 PUD 2b951067 PMD 0
> [ 5594.693720] Oops: 0002 [#1] SMP
> [ 5594.699888] last sysfs file:
The bug was that unix domain sockets use a pseduo packet for
connecting and accept uses that psudo packet to get the socket.
In the buggy seqpacket case we were allowing unconnected
sockets to call recvmsg and try to receive the pseudo packet.
That is always wrong and as of commit 7361c36c5 the pseudo
packet had become enough different from a normal packet
that the kernel started oopsing.
Do for seqpacket_recv what was done for seqpacket_send in 2.5
and only allow it on connected seqpacket sockets.
Cc: stable@kernel.org
Tested-by: Dan Aloni <dan@aloni.org>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If an application program does not make any changes to the indirect
blocks or extent tree, i_datasync_tid will not get updated. If there
are enough commits (i.e., 2**31) such that tid_geq()'s calculations
wrap, and there isn't a currently active transaction at the time of
the fdatasync() call, this can end up triggering a BUG_ON in
fs/jbd2/commit.c:
J_ASSERT(journal->j_running_transaction != NULL);
It's pretty rare that this can happen, since it requires the use of
fdatasync() plus *very* frequent and excessive use of fsync(). But
with the right workload, it can.
We fix this by replacing the use of tid_geq() with an equality test,
since there's only one valid transaction id that we is valid for us to
wait until it is commited: namely, the currently running transaction
(if it exists).
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
percpu_counter_sum_positive() never returns a negative value.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
This patch fixes a bug where proper queue id's were not passed to
chip for cleanup.
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This patch fixes a bug where the interrupts were not rearmed if
the Interrupt Mask was enabled. This patch checks for Interrupt mask
enabled and still rearms interrupt generation even if the Interrupt
mask was enabled
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The code currently computes an offset into a dma_alloc_coherent() area
on the assumption that the alignment is imprecise. In fact, the API
guarantees PAGE_SIZE alignment, so the offset calculation is always
zero: remove it.
[jejb: make description actually descriptive]
Signed-off-by: Tomas henzl <thenzl@redhat.com>
Acked-by: Nick Cheng<nick.cheng@areca.com.tw>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
I removed outer loops in ...wait_msgint_ready the sleeptime and
retrycount are in fact never changed so I changed them into
defines. In arcmsr_flush_hba_cache is a loop removed, which printed
the same printk 100 times, one line in log is enough I think. The
arcmsr_sleep_for_bus_reset has lost a functionality with the latest
patches, The only thing the function does is a long sleep, so it's
replaced with a ssleep.
[jejb: checkpatch fixes]
Signed-off-by: Tomas henzl <thenzl@redhat.com>
Acked-by: Nick Cheng <nick.cheng@areca.com.tw>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Use fc_exch_mgr_list_clone to setup em for npiv port.
Also remove redundant vport and lport initializations.
Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Acked-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The rcu protected macros rcu_dereference() and rcu_assign_pointer()
for the bat_priv->primary_if need to be used, as well as spin/rcu locking.
Otherwise we might end up using a primary_if pointer pointing to already
freed memory.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
This is a regression from c4aac1ab9b
- gw_node_update() doesn't add a new gw_node in case of empty curr_gw.
This means that at the beginning no gw_node is added, leading to an
empty gateway list.
- gw_election() is terminating in case of curr_gw == NULL. It has to
terminate in case of curr_gw != NULL
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
atomic_dec_not_zero() is very useful and it is currently defined
multiple times. So it is possible to move it in main.h
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
orig_hash_find() manages rcu_lock/unlock internally and doesn't need to
be surrounded by rcu_read_lock() / rcu_read_unlock() anymore
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Acked-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
On systems with very large numbers (> 1600 or so) of SCSI devices,
cat /proc/scsi/scsi ends up failing with -ENOMEM. This is due to
the show routine simply iterating over all of the devices with
bus_for_each_dev(), and trying to dump all of them into the buffer
at the same time. On my test system (using scsi_debug with 4064 devices),
the output ends up being ~ 632k, far more than kmalloc will typically allow.
This patch defines its own seq_file opreations to iterate over the scsi
devices.The result is that each show() operation only dumps ~ 180 bytes
into the buffer at a time so we don't run out of memory.
If the "Attached devices" header isn't required, we can dump the
sfile->private bit completely.
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>