Instead of using a global lock, the PCIe transport
can use an own lock for its IRQ. This will make it
possible to not disable IRQs for the shared lock.
The lock is currently used throughout the code but
this can be improved even further by splitting up
the locking for the queues.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
In order to separate the different parts of the
driver better, we are reducing the shared data.
This moves the workqueue to "priv", and removes
it from the transport. To do this, simply use
schedule_work() in the transport.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Make iwlwifi record all debug messages into
tracing, even if debug_level is not enabled.
Due to the lack of APIs, the debug messages
are now recorded up to a max length of 100,
the only one above that is the RXON which is
not needed if you trace the commands as well
as it only dumps the command contents.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
The testmode command for host command send now replies
with a nl80211 message and the response it recieved from
the device.
This does not change the API directly, but adds a reply
to the testmode call.
Signed-off-by: Amit Beka <amit.beka@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Replaced the old SRAM and periphery indirect access functions
with a unified indirect memory access functions. These include
new IWL_TM_CMDs for buffer read/write/dump which replace the
SRAM read/dump commands, but the API for IWL_TM_CMD_INDIRECT_REG
read/write will now not be supported (returns error).
This also handles writing to periphery registers in 1-3 bytes.
Requires the corresponding patch in the library for the API change.
Signed-off-by: Amit Beka <amit.beka@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
A lot of error conditions in testmode log as IWL_DEBUG_INFO which is not
logged by default. Change it
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Added a check on the direct register access.
Checks that the address is in the lower ragnge (0x0-0x2000),
which belongs to CSR, HBUS and FH registers.
Signed-off-by: Amit Beka <amit.beka@intel.com>
Signed-off-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
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>
The routines dealing with the ucode are spread through several files.
Move them all to the same file and create a iwl-ucode.h file with the
ucode file definitions.
Signed-off-by: Don Fry <donald.h.fry@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
There is nothing device specific in the initialization of the
bcast_sta_id so move it to the common inititalization routine.
Signed-off-by: Don Fry <donald.h.fry@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The message was misleading when a queue is deactivated. The fifo
number is irrelevant then, so don't print it.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi 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>
Allocating the shrd area dynamically will allow more agility
while revamping the flows.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
There is no link between the two. Ensure that the NIC is on outside
the code of the EEPROM handling.
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>
Whenever the PAN (P2P) context is active, it
has timers in the uCode that prevent sleep,
so scanning can't be out of channel for more
than the beacon interval programmed into the
device.
Before this patch, a full scan including any
passive channels when P2P was active would
stall forever because it wouldn't find time
to execute the passive requests (for default
beacon intervals of 100 TU.)
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>
Led has no use for some platform.
Add additional module parameter option to disable LED
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Fix multiple bugs in event tracing:
1) If you enable uCode tracing with the device down,
it will still attempt to access the device and
continuously log "MAC is in deep sleep!" errors.
Fix this by only starting logging when the device
is actually alive.
2) Now you can set the flag when the device is down,
but logging doesn't happen when you bring it up.
To fix that, start logging when the device comes
alive. This means we don't log before -- we could
do that but I don't need it right now.
3) For some reason we read the error instead of the
event log -- use the right pointer.
4) Optimise SRAM reading of event log header.
5) Fix reading write pointer == capacity, which can
happen due to racy SRAM access
6) Most importantly: fix an error where we would try
to read WAY too many events (like 2^32-300) when
we read the wrap counter before it is updated by
the uCode -- this does happen in practice and will
cause the driver to hang the machine.
7) Finally, change the timer to 10ms instead of 100ms
as 100ms is too slow to capture all data with a
normal event log and with 100ms the log will wrap
multiple times before we have a chance to read it.
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>
Whenever the PAN (P2P) context is active, it
has timers in the uCode that prevent sleep,
so scanning can't be out of channel for more
than the beacon interval programmed into the
device.
Before this patch, a full scan including any
passive channels when P2P was active would
stall forever because it wouldn't find time
to execute the passive requests (for default
beacon intervals of 100 TU.)
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Fix multiple bugs in event tracing:
1) If you enable uCode tracing with the device down,
it will still attempt to access the device and
continuously log "MAC is in deep sleep!" errors.
Fix this by only starting logging when the device
is actually alive.
2) Now you can set the flag when the device is down,
but logging doesn't happen when you bring it up.
To fix that, start logging when the device comes
alive. This means we don't log before -- we could
do that but I don't need it right now.
3) For some reason we read the error instead of the
event log -- use the right pointer.
4) Optimise SRAM reading of event log header.
5) Fix reading write pointer == capacity, which can
happen due to racy SRAM access
6) Most importantly: fix an error where we would try
to read WAY too many events (like 2^32-300) when
we read the wrap counter before it is updated by
the uCode -- this does happen in practice and will
cause the driver to hang the machine.
7) Finally, change the timer to 10ms instead of 100ms
as 100ms is too slow to capture all data with a
normal event log and with 100ms the log will wrap
multiple times before we have a chance to read it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
When an interrupt comes in, we read the reason
bits and collect them into "trans_pcie->inta".
This happens with the spinlock held. However,
there's a bug resetting this variable -- that
happens after the spinlock has been released.
This means that it is possible for interrupts
to be missed if the reset happens after some
other interrupt reasons were already added to
the variable.
I found this by code inspection, looking for a
reason that we sometimes see random commands
time out. It seems possible that this causes
such behaviour, but I can't say for sure right
now since it happens extremely infrequently on
my test systems.
Cc: stable@vger.kernel.org [3.2]
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>
Don Fry