When we allocate queues, we currently don't
use kzalloc() right now. When we then free
those queues again without having used all
entries, we may end up trying to free random
pointers found in the txb array since it was
never initialised. This fixes it simply by
using kzalloc().
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
This can be cleanly applied to wireless-2.6 and iwlwifi git trees.
=
From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Subject: [PATCH] iwlwifi: use the DMA state API instead of the pci equivalents
This replace the PCI DMA state API (include/linux/pci-dma.h) with the
DMA equivalents since the PCI DMA state API will be obsolete.
No functional change.
For further information about the background:
http://marc.info/?l=linux-netdev&m=127037540020276&w=2
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This function is only used by the agn drivers,
so doesn't have to be in core and exported.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
We used to free all the Tx queues memory when interface is brought
down and reallocate them again in interface up. This requires
order-4 allocation for txq->cmd[]. In situations like s2ram, this
usually leads to allocation failure in the memory subsystem. The
patch fixed this problem by allocating the Tx queues memory only at
the first time. Later iwl_down/iwl_up only initialize but don't
free and reallocate them. The memory is freed at the device removal
time. BTW, we have already done this for the Rx queue.
This fixed bug https://bugzilla.kernel.org/show_bug.cgi?id=15551
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Update to include additional tx command response status for "_agn"
devices.
The following status indicate the transmission was postponed:
TX_STATUS_POSTPONE_DELAY
TX_STATUS_POSTPONE_FEW_BYTES
TX_STATUS_POSTPONE_BT_PRIO
TX_STATUS_POSTPONE_QUIET_PERIOD
TX_STATUS_POSTPONE_CALC_TTAK
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Includes minor improvements in debugging messages in iwl-4965.c,
function iwl4965_is_temp_calib_needed().
Signed-off-by: Frans Pop <elendil@planet.nl>
Cc: Zhu Yi <yi.zhu@intel.com>
Cc: Reinette Chatre <reinette.chatre@intel.com>
Cc: Intel Linux Wireless <ilw@linux.intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
We used to free all the Tx queues memory when interface is brought
down and reallocate them again in interface up. This requires
order-4 allocation for txq->cmd[]. In situations like s2ram, this
usually leads to allocation failure in the memory subsystem. The
patch fixed this problem by allocating the Tx queues memory only at
the first time. Later iwl_down/iwl_up only initialize but don't
free and reallocate them. The memory is freed at the device removal
time. BTW, we have already done this for the Rx queue.
This fixed bug https://bugzilla.kernel.org/show_bug.cgi?id=15551
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Identify the tx functions only used by agn driver and move those from
iwlcore to iwlagn.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Monitors the internal TX queues periodically. When a queue is stuck
for some unknown conditions causing the throughput to drop and the
transfer is stop, the driver will force firmware reload and bring the
system back to normal operational state.
The iwlwifi devices behave differently in this regard so this feature is
made part of the ops infrastructure so we can have more control on how to
monitor and recover from tx queue stall case per device.
Signed-off-by: Trieu 'Andrew' Nguyen <trieux.t.nguyen@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Commit a239a8b47c introduced a
noisy message, that fills up the log very fast.
The error seems not to be fatal (the connection is stable and
performance is ok), so make it IWL_DEBUG_TX rather than IWL_ERR.
Signed-off-by: Adel Gadllah <adel.gadllah@gmail.com>
Cc: stable@kernel.org
Acked-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
4965 hardware has 7 queues reserved and the
remaining ones used for aggregation, 5000
and higher need to have 10 reserved. This
is not very clear in the code right now,
unfortunately.
Introduce a new IWL_TX_FIFO_UNUSED constant
and make the queue/FIFO mapping arrays able
to hold that value, and change the setup
code to reserve all queues in the arrays
(the queue number is the index) and use the
new unused constant to not map those queues
to any FIFO.
Additionally, clear up the AC/queue mapping
code to be more understandable. The mapping
is the identity mapping right now, but with
the mapping function I think it's easier to
understand what happens there.
Finally, HCCA isn't implemented at all and
I think newer microcode removed it, so let's
remove all mention of it in the code, some
comments remain for 4965.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Shanyu Zhao <shanyu.zhao@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
With some of the new code in mac80211, public action
frames can be exchanged as non-injected frames even
while not associated.
Aside from that, dropping frames here is pointless
since we do deal with arbitrary frames that were
injected already, so let mac80211 make the decision
about which frames to allow or not.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Increase the buffer size for commands with "huge"
bit set. This has been recently observed for 6050 cards where
for even with huge bit set few commands were not properly allocated
memory with the command overwriting the buffer allocated for it..
Also add a check to see if command size exceeds the
maximum allowable size.
Signed-off-by: Abhijeet Kolekar <abhijeet.kolekar@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Check the frame control for ieee80211_is_data_qos() is true before
counting the number of tfds can be free, the tfds_in_queue only
increment when ieee80211_is_data_qos() is true before transmit; so it
should only decrement if the type match.
Remove ieee80211_is_data_qos check for frame_ctrl in tx_resp to avoid
invalid information pass from uCode.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
CC: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
When receive reply_tx and ready to decrement the count for number of
tfds in queue, do error checking to prevent error condition and
tfds_in_queue become negative number.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
CC: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Until now it was only possible to have one synchronous command running at
any time. If a synchronous command is in progress when a second request
arrives then the second command will fail. Create a new mutex specific for
this purpose to only allow one synchronous command at a time, but enable
other commands to wait instead of fail if a synchronous command is in
progress.
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Change pci_alloc_consistent() to dma_alloc_coherent() so we can use
GFP_KERNEL flag.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Error checking for aggregation frames should go into aggregation queue,
if aggregation queue not available, use legacy queue instead.
Also make sure the aggregation queue is available to activate,
if driver and mac80211 is out-of-sync, try to disable the queue and
sync-up with mac80211.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
drivers/net/wireless/iwlwifi/iwl-tx.c: In function `iwl_hw_txq_ctx_free':
drivers/net/wireless/iwlwifi/iwl-tx.c:410: warning: suggest explicit braces to avoid ambiguous `else'
Cc: Zhu Yi <yi.zhu@intel.com>
Cc: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
CC [M] drivers/net/wireless/iwlwifi/iwl-tx.o
drivers/net/wireless/iwlwifi/iwl-tx.c: In function ‘iwl_tx_agg_stop’:
drivers/net/wireless/iwlwifi/iwl-tx.c:1356: warning: passing argument 1 of ‘ieee80211_stop_tx_ba_cb_irqsafe’ from incompatible pointer type
include/net/mac80211.h:2128: note: expected ‘struct ieee80211_vif *’ but argument is of type ‘struct ieee80211_hw *’
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Lennert Buytenhek noticed that delBA handling in mac80211
was broken and has remotely triggerable problems, some of
which are due to some code shuffling I did that ended up
changing the order in which things were done -- this was
commit d75636ef9c
Author: Johannes Berg <johannes@sipsolutions.net>
Date: Tue Feb 10 21:25:53 2009 +0100
mac80211: RX aggregation: clean up stop session
and other parts were already present in the original
commit d92684e660
Author: Ron Rindjunsky <ron.rindjunsky@intel.com>
Date: Mon Jan 28 14:07:22 2008 +0200
mac80211: A-MPDU Tx add delBA from recipient support
The first problem is that I moved a BUG_ON before various
checks -- thereby making it possible to hit. As the comment
indicates, the BUG_ON can be removed since the ampdu_action
callback must already exist when the state is != IDLE.
The second problem isn't easily exploitable but there's a
race condition due to unconditionally setting the state to
OPERATIONAL when a delBA frame is received, even when no
aggregation session was ever initiated. All the drivers
accept stopping the session even then, but that opens a
race window where crashes could happen before the driver
accepts it. Right now, a WARN_ON may happen with non-HT
drivers, while the race opens only for HT drivers.
For this case, there are two things necessary to fix it:
1) don't process spurious delBA frames, and be more careful
about the session state; don't drop the lock
2) HT drivers need to be prepared to handle a session stop
even before the session was really started -- this is
true for all drivers (that support aggregation) but
iwlwifi which can be fixed easily. The other HT drivers
(ath9k and ar9170) are behaving properly already.
Reported-by: Lennert Buytenhek <buytenh@marvell.com>
Cc: stable@kernel.org
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Change error message for command queue full
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Acked-by: Ben Cahill <ben.m.cahill@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The entire aggregation code currently operates on the
hw pointer and station addresses, but that needs to
change to make stations purely per-vif; As one step
preparing for that make the aggregation code callable
with the station, or by the combination of virtual
interface and station address.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Using the new mac80211 functionality, this makes
iwlwifi handle unicast PS buffering correctly.
The device works like this:
* when a station goes to sleep, the microcode notices
this and marks the station as asleep
* when the station is marked asleep, the microcode
refuses to transmit to the station and rejects all
frames queued to it with the failure status code
TX_STATUS_FAIL_DEST_PS (a previous patch handled
this correctly)
* when we need to send frames to the station _although_
it is asleep, we need to tell the ucode how many,
and this is asynchronous with sending so we cannot
just send the frames, we need to wait for all other
frames to be flushed, and then update the counter
before sending out the poll response frames. This
is handled partially in the driver and partially in
mac80211.
In order to do all this correctly, we need to
* keep track of how many frames are pending for each
associated client station (avoid doing it for other
stations to avoid the atomic ops)
* tell mac80211 that we driver-block the PS status
while there are still frames pending on the queues,
and once they are all rejected (due to the dest sta
being in PS) unblock mac80211
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add txq_id info to "Tx queue requesting wakeup" debug message
Add "Rx queue requesting wakeup" debug message
Move clear of CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ to be after nearby
iwl_write_prph(), since iwl_write_prph() sets it and clears it. Almost
removed it entirely, but just making sure in case someone removes the
iwl_write_prph()! Also remove unneeded priv->lock usage; this is now
handled by priv->reg_lock within iwl_clear_bit().
Join a couple of lines that had unneeded line returns.
Signed-off-by: Ben Cahill <ben.m.cahill@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
When I added that code setting the swq_id, I evidently
did not understand the distinction between FIFO and TX
queue yet and added code to compare a queue ID and a
FIFO number, which is bogus.
However, the code there need not be this specific, it
can just set all queues to the identity mapping which
will be overwritten by the aggregation queue code. As
a bit of defensive coding, don't assign an swq_id to
the command queue so that if we ever use it for frames
we notice quickly.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
More information than the "-EIO" return code will be useful here.
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Handling responses to driver originated commands include passing the
original command buffer to the caller. At this time it is possible for a
callback to be invoked that is passed this command buffer and thus likely
to access it.
We need to make sure that the memory associated with that buffer is not DMA
mapped at the time.
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
parameters "len" is not used in both iwl_tx_queue_free() and
iwl_cmd_queue_free() functions
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This switches the iwlwifi driver to use paged skb from linear skb for Rx
buffer. So that it relieves some Rx buffer allocation pressure for the
memory subsystem. Currently iwlwifi (4K for 3945) requests 8K bytes for
Rx buffer. Due to the trailing skb_shared_info in the skb->data,
alloc_skb() will do the next order allocation, which is 16K bytes. This
is suboptimal and more likely to fail when the system is under memory
usage pressure. Switching to paged Rx skb lets us allocate the RXB
directly by alloc_pages(), so that only order 1 allocation is required.
It also adjusts the area spin_lock (with IRQ disabled) protected in the
tasklet because tasklet guarentees to run only on one CPU and the new
unprotected code can be preempted by the IRQ handler. This saves us from
spawning another workqueue to make skb_linearize/__pskb_pull_tail happy
(which cannot be called in hard irq context).
Finally, mac80211 doesn't support paged Rx yet. So we linearize the skb
for all the management frames and software decryption or defragmentation
required data frames before handed to mac80211. For all the other frames,
we __pskb_pull_tail 64 bytes in the linear area of the skb for mac80211
to handle them properly.
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Remove the ununsed variable data_retry_limit
from priv.
Signed-off-by: Abhijeet Kolekar <abhijeet.kolekar@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Instead of always allocate the max number of tx queue structure,
use dynamic allocation based on the number of queues in device
configuration. With these changes, device does not have to allocate more
memory than the h/w can support.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
After m68k's task_thread_info() doesn't refer to current,
it's possible to remove sched.h from interrupt.h and not break m68k!
Many thanks to Heiko Carstens for allowing this.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
In order to have an easier way to debug issues, create
trace events (using the ftrace framework) that will
allow us to follow exactly what the driver is doing
with the device.
The text format isn't all that useful, but the binary
format can also be obtained easily via debugfs and
then analysed on the fly or offline with debugging
tools.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg