smatch correctly complains:
iwl-trans-pcie.c +1528 iwl_trans_pcie_start_hw(50) warn: 'trans->irq' was not released on error
Fix it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
This trans_ops->stop_hw leaves the RFKILL interrupt enabled,
we can call that one instead of enable_rfkill_int. By that,
we reduce the numbers of acceesses to the NIC from the upper
layers.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The HW revision is now read by the transport layer in its allocation.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Get this information from the transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Get this information from the transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Get this information from the transport layer which is now in charge
of the APM too.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This handler was called from the transport layer only. Merge it
to the transport's apm_init.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Now there is only one transport function that launch a specific fw:
trans_ops->start_fw. This one replaces trans_ops->start_device and
trans_ops->kick_nic. The code that actually loads the fw to the
device has been moved to the transport specific code.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This is another clean up of the proble flow.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This handler stops the HW and puts it in low power state.
It will allow to clean up the flows in the upper layers.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Kill the trans_ops->prepare_card_hw which is now useless.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This handler will become thicker, reflect its real role now.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
From now on, the transport layer in charge of providing access to the
device. So change all the driver to give a pointer to the transport
to all the low level functions that actually access the device.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Most of the accesses to the registers are done from the transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
All the bus configuration is now done in the transport
allocation fucntion.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Change the way we alloc the transport on the way.
Since the transport is allocated from a bus specific area, we can
give the bus specific parameters (i.e. pci_dev for PCI) to the
transport. This will be useful when the bus layer will be killed.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This patch connects IDI transport to driver. It does so
by using a number of ifdefs at this stage.
IDI is a new transport that is under development.
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Define a new handler in the transport layer API: fw_alive.
Move iwl_reset_ict to this new handler, and move the content
of tx_start to this handler.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
In my AMPDU rework, I rely on the sequence numbers of frames. But
I didn't check that the frame has a valid tid before updating the
tracking counters. As a result, the Tx queues were stalled. People
who hit this bug saw that we simply didn't let any data out.
This bug was introduced in 3.3.
This patch fixes that and checks that the frame is a QoS frame before
looking at its tid and changing the counters.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Replace dma_alloc_coherent()+memset() with the new dma_zalloc_coherent()
Signed-off-by: Djalal Harouni <tixxdz@opendz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Conflicts:
net/bluetooth/l2cap_core.c
Just two overlapping changes, one added an initialization of
a local variable, and another change added a new local variable.
Signed-off-by: David S. Miller <davem@davemloft.net>
Since we configure all the queues as CHAINABLE, we need to update the
byte count for all the queues, not only the AGGREGATABLE ones.
Not doing so can confuse the SCD and make the fw assert.
Cc: stable@vger.kernel.org
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This will allow us to catch bad cases in which the packets aren't in
the right place on the ring.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This is another step towards the move of tid_data from the shared
area.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The tid_data is not related to the transport layer, so move
the logic that depends on it to the upper layer.
This patch deals with the seq_number.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The tid_data is not related to the transport layer, so move
the logic that depends on it to the upper layer.
This patch deals with the mapping of RA / TID to HW queues in AGG.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The tid_data is not related to the transport layer, so move
the logic that depends on it to the upper layer.
This patch deals with the code that checks if there are still
pending packets for an RA / TID.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The tid_data is not related to the transport layer, so move
the logic that depends on it to the upper layer.
This patch deals with tx AGG stop.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
In the same spirit as the previous patch. Eventually this will
allow us to remove the tid_data knowledge from the transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
In the very first implementation of HT, the driver was responsible
for the queueing: stopping and waking the queues while the HW queues
where being drained. In this implementation, we had to deal with the
case where we were draining the AGG queue because we wanted to tear
down the BA agreement.
In the normal flow (when we don't drain any HW queue), when packets
are reclaimed, we wake the SW queue in case the SW queue was stopped
which can happen when the HW queues are too full.
While draining a HW queue, we must make sure that we don't wake the
SW queue, since the whole point of the draining is to empty totally
the HW queue and not only get below a certain threshold.
This is why there is condition in the reclaim function:
if (NOT EMPTYING DELBA)
wake the SW queue is applicable
Since then, a lot has changed and mac80211 is now able to buffer
packets that are being sent to a packet list that will be spliced
after the driver has reported it has drained its HW queues.
Hence, there is no need for the for aforementioned if, and we can
safely wake up the queue even if we are draining HW queues.
Removing this if, also allows us to remove the wake_queue in
check_empty that was there in order to deal with a corner case
created by the if.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Since packets sent to an RA / TID in AGG are sent from a
separate HW Tx queue, we may get into a race:
the regular queue isn't empty while we already begin to
send packets from the AGG queue. This would result in sending
packets out of order.
In order to cope with this, mac80211 waits until the driver
reports that the legacy queue is drained before it can send
packets to the AGG queue. During that time, mac80211 buffers
packets for the driver. These packets will be sent in order
after the driver reports it is ready.
The way this was implemented in the driver is as follows:
We held a counter that monitors the number of packets for
an RA / TID in the HW queues. When this counter reached 0,
we knew that the HW queues were drained and we reported to
mac80211 that were ready to proceed.
This patch changes the implementation described above. We
now remember what is the wifi sequence number of the first
packet that will be sent in the AGG queue (lets' call it
ssn). When we reclaim the packet before ssn, we know that
the queue is drained, and we are ready to proceed.
This will allow us to move this logic in the upper layer and
eventually remove the tid_data from the shared area.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Move the calib_results list from the upper layer iwl_priv structure
to the lower layer iwl_trans structure.
Signed-off-by: Don Fry <donald.h.fry@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Add more data when inconsistencies occur in the AGG state machine.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Some information was redundation, other was missing.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add more information when a queue is stuck and actually get
information from the scheduler instead of looking at internal
variables.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Users complain that the traffic gets stalled sometimes. This will
allow easier debugging.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
As Stanislaw pointed out, my patch
iwlagn: fix a race in the unmapping of the TFDs
solved only part of the problem. The race still exists for TFDs of
the host commands. Fix that too.
Reported-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Use the same calling style for all the mac80211 callback functions
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This one can be _very_ noisy.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
While inspecting the code, I saw that iwl_tx_queue_unmap modifies
the read pointer of the Tx queue without taking any locks. This means
that it can race with the reclaim flow. This can possibly lead to
a DMA warning complaining that we unmap the same buffer twice.
This is more a W/A than a fix since it is really weird to take
sta_lock inside iwl_tx_queue_unmap, but it can help until we revamp
the locking model in the transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Occasionally, the device will send interrupts
while it is resuming, at a point where we are
not set up again to handle them. This causes
the core IRQ handling to completely disable
the IRQ, and then the driver won't work again
until it is reloaded/rebound.
To fix this issue disable the IRQ on suspend,
this will cause us to only get interrupts
again after we've setup everything on resume.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
802.11 says:
"Sequence numbers for QoS (+)Null frames may be
set to any value."
However, if we use the normal counters then peers
will get confused with aggregation since there'll
be holes in the sequence number sequence.
To avoid that, don't assign sequence numbers to
QoS Null frames.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi GUy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
As everybody knows kcalloc checks the multiplication is safe and
that we don't run into overflow.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
John W. Linville