The previous P2P implementation turned out to
not work well and new uCode capabilities were
added to support P2P. Modify the driver to
take advantage of those, and also discover P2P
support automatically based on a uCode flag
instead of having a Kconfig symbol for P2P.
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>
The availability of contexts depends on the
firmware capabilities. Currently only the
presence of the second context depends on it,
but soon P2P support will also be different.
Move the context initialisation code to the
firmware-dependent setup before registering
with mac80211 to make it easier to handle.
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>
We sometimes need to support new firmware API for
a while before we can publish them since testing
them fully takes a long time. We could keep all
the new code private, but that causes plenty of
problems and sometimes we can give a pre-release
version of firmware to people who need to test.
However, when we just bump the API version, the
driver will warn everybody that their firmware is
outdated, when in fact it isn't. (Currently our
case for this doesn't really change the API but
bumping the API version is necessary because the
firmware isn't fully backward compatible)
In order to handle this in the future, add a new
"api_ok" version; only below this will the driver
warn that the uCode is too old.
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>
linux/wireless.h and net/iw_handler.h headers are
for wireless extensions only, so mac80211 drivers
shouldn't be including them.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This is since my patch:
iwlagn: introduce transport layer and implement rx_init
The IRQ is requested before the locks are initialized, hence the crash.
Initialize the tasklet before we request the IRQ on the way.
Reported-by: Johannes Berg <johannes.berg@intel.com>
Tested-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Implement WoWLAN support in iwlagn. The device
supports a number of wakeup triggers and can do
GTK rekeying when asleep (if HW crypto is used).
Unfortunately, we need to disconnect from the AP
after resume since we can't yet get all the info
out of the wowlan uCode to stay connected safely.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
As I just discovered while doing WoWLAN, HW crypto
is done wrong for GTKs: they should be programmed
for the AP station ID (in the managed mode case)
and the HW can actually deal with multiple group
keys per station as well (which is useful in IBSS
RSN but that I've chosen not to use this).
To fix all this, modify the way keys are sent to
the device and key offsets are allocated. After
these changes, key offsets are stored into the
hw_key_idx which we can then track for the key
lifetime, not relying on our sta_cmd array. WEP
default keys get special treatment, of course.
Additionally, since I had the API for it, we can
now pre-fill TKIP phase 1 keys for RX now that we
can obtain the P1K from mac80211, a capability I
had added for WoWLAN initially.
Finally, some keys simply don't need to be added
into the device's key cache -- a key that won't
be used for RX is only needed in the TX header,
so "pretend" to have accepted any key without
adding it into the device -- no need to use up
key space there for it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The price to pay is the access to the log system. Therefore logs from bus layer
are sent by dev_printk instead of IWL_XXXX.
Rename bus->priv to bus->drv_data to make the separation even clearer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Call iwl_probe with a ready iwl_bus struct. This means that the bus layer
assigns the irq, dev and iwl_bus_ops pointers to iwl_bus before giving it to
iwl_probe.
The device specific struct is allocated together with the common iwl_bus struct
by the bus specific layer. The pointer to the aggregate struct is passed to the
upper layer that holds a pointer to iwl_bus instead of an embedded iw_bus.
The private data given to the PCI subsystem is now iwl_bus and not iwl_priv.
Provide bus_* inliners on the way in order to simplify the syntax.
Rename iwl-pci.h -> iwl-bus.h since it is bus agnostic and represent the
external of the bus layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
It still holds a pointer to iwl_priv. But hopefully this will disappear at some point.
Also add the multiple inclusion protection to iwl-trans.h that was forgotten.
Move iwl-trans structures to iwl-trans.h
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
This function is really related to the transport layer - move it.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Basically all the nic_init flow should be in the transport layer.
iwl_prepare_card_hw will move to the transport too in a separate patch.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
All the configurations of the HW for AMPDU are now 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>
Since iwlagn_setup_deferred_work is always called, fold it into
iwl_setup_deferred_work. BT related works are setup by the new
bt_setup_deferred_work lib_ops.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.guy@intel.com>
Since the ICT is transport related, move all its functions to the transport
layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Since iwlagn_stop_device was the only caller to the rx_stop / tx_stop,
these two don't need to be API any more.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.guy@intel.com>
Also create a new file: iwl-trans-int-pcie.h which will include
the non static functions that are shared among the current pcie transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The upper layer receives a pointer to an iwl_rx_mem_buffer. I would prefer the
upper layer to receive a pointer to an iwl_rx_packet, but this is impossible
since the Rx path needs to add the address of the page to the skb.
I may find a solution later.
All the pre_rx_handler and notification code has been moved to the upper layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Since all the irq / tasklet is now handled in the transport layer, it should
give an API to ensure that all the irq / tasklet have finished running. This
will allow the upper layer to release all its resources.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
PCIe doesn't provide any ISR registration API, whereas other buses do.
Hence, we need to move the tasklet and irq to the transport layer to allow this
flexibility.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Multiple quoted strings are concatenated without comma separators.
Make the array const while there.
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
WiFi throughput drops drastically when BT is turned on, BT and WiFi
are simultaneously transmitting/receiving traffic. This is particularly true
when BT has higher priority over WiFi, and hence the device defers TX frames.
The AP assumes that the channel is bad and reduces the data rate, implying
longer airtime, which exacerbates the problem further, resulting ultimately
in what is popularly called the "death-spiral" phenomenon. The use of PS-poll
in such scenarios guarantees a low but consistent throughput.
Since the death-spiral phenomenon is observed only when the RSSI is low, use
PS-poll only when RSSI is low and disable when high, with a known hysterisis.
This feature specifies the high and low thresholds and implements the
callbacks registered with mac80211, which will be called when threshold events
occur.
iwlwifi: dynamic pspoll: optimize rssi monitor code
Signed-off-by: Meenakshi Venkataraman <meenakshi.venkataraman@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This removes the for priv->trans.ops->...
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>
Now, there are only two functions to send a host command:
* send_cmd that receives a iwl_host_cmd
* send_cmd_pdu that builds the iwl_host_cmd itself and received flags
The flags CMD_ASYNC / CMD_SYNC / CMD_WANT_SKB are not changed by the API
functions.
Kill the unused flags CMD_SIZE_NORMAL / CMD_NO_SKB on the way.
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>
Tx free functions move to the transport layer. Unify the functions that deal with tx queues and cmd queue.
Since the CMD queue is not fully allocated, but uses the q->n_bd / q->window trick, the release flow of TX queue and CMD queue was different.
iwlagn_txq_free_tfd receives now the index of the TFD to be freed, which allows to unify the release flow for all the queues.
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>
The transport layer ness to release all rx ressources. This function is an API for it.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The transport layer is responsible for all the queues, DMA rings etc...
This is the beginning of the separation of all the code that is tighly
related to HW design to the aforementioned transport layer.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Add the parameter to disable stuck queue watchdog timer, different platforms
might have different timing. Provide the option to disable the timer to prevent
un-necessary firmware reload.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Use the rate that mac80211 requested to fill the
uCode TX command for the beacon. Unfortunately,
the uCode is (currently?) ignoring it, but now
at least fill it properly.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Add power_level module parameter to set the default power save level.
Power save level has range from 1 - 5, default power save level is 1.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Add power_save module parameter to enable power management if needed
Default power management is disabled.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Since the irq number is just an unsigned int, store it inside iwl_bus
instead of calling the get_irq ops every time it is needed.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
All pci related stuff is in iwl-pci.c.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
In order to remove a few more dereference to priv->pdev that will be killed
[Asoon, there is now a method to get the IRQ number.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
The PCI_REVISION_ID is read and printed in iwl_pci_probe anyway using pr_info
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Get a pointer to the struct device during probe and get the rid of all the PCI
specific DMA wrappers.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Since we have now a PCI layer, all the init and deinit code that is PCI
related should move to there.
Also move the IO functions: read8/read32/write32. They need hw_base which
is killed from priv.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Bus specific layer must know how to return the struct device* of the device.
Implement that as a callback of iwl_bus_ops and use that callback instead of
using the priv->pdev pointer which is meant to disappear soon.
Since the struct device * is needed in hot path, iwl_bus holds a pointer to it
instead of calling get_dev all the time.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
iwl_bus will represent a bus, and iwl_bus_ops all the operations that can be
done on this bus.
For the moment only set_prv_data is implemented. More to come...
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Move some PCI functionality to the new iwl_pci.[ch] files:
* the PCI_DEVICE_TABLE
* the pci_driver struct definition
* the PCI probe / remove functions
* the PCI suspend / resume functions
All these functions are now split: the trigger comes from the PCI layer which
calls to the bus generic code located in the other files.
This is the beginning only. There are still a lot of PCI related code needs
to be gathered.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
