If the device is not started, we can't read its
SRAM and attempting to do so will cause issues.
Protect the debugfs read.

Cc: stable@vger.kernel.org
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

iwl_dbgfs_fh_reg_read() can cause crashes and/or
BUG_ON in slub because the ifdefs are wrong, the
code in iwl_dump_fh() should use DEBUGFS, not
DEBUG to protect the buffer writing code.

Also, while at it, clean up the arguments to the
function, some code and make it generally safer.

Cc: stable@vger.kernel.org
Reported-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

There's a bug that causes the rate scaling to get stuck
when it has to use single-stream rates with a peer that
can do GF and SGI; the two are incompatible so we can't
use them together, but that causes the algorithm to not
work at all, it always rejects updates.

Disable greenfield for now to prevent that problem.

Cc: stable@vger.kernel.org
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Tested-by: NCesar Eduardo Barros <cesarb@cesarb.net>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Since the kmem cache API doesn't internally allocate
the name but just points to the name that was passed
in we can't use stack memory for it. Move the name
into the transport struct.
Reported-by: NStanislaw Gruszka <sgruszka@redhat.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Some drivers (iwlegacy, iwlwifi and rt2x00) today use the
bss_conf.last_tsf value. By itself though that value is
completely worthless since it may be ancient. What really
is needed is synchronisation between some device time and
the TSF.

To clarify this, rename bss_conf.last_tsf to sync_tsf and
add sync_device_ts which is obtained from rx_status which
gets a new field device_timestamp for this purpose. This
is intentionally not using the mactime field since that
is used for other things and in IBSS is expected to sync
with the IBSS's TSF which isn't necessarily true for the
device timestamp.

Also, since we have the information and it's useful even
before the connection has been established, give all the
timing details to the driver before authenticating.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Newer devices use 32 bit for boost register,
set the correct value for it.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Remove this dead code, it is unused for device newer than
4965.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

'echo 1 > log_event' generates the bogus "MAC is in deep sleep"
or "Timeout waiting for hardware access" log messages when
the interface is down, we should just disallow accessing the
device through debugfs when it is down.
Signed-off-by: NRichard A. Griffiths <richardx.a.griffiths@intel.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's no need to check trans for non-null
here as it has already been checked in the
caller. This fixes an smatch warning that we
check after having dereferenced it.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Debug messages should be printed using dev_dbg() not
dev_err() which requires DEBUG to be defined.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Some messages were missing a trailing newline, add it.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Signed-off-by: NPaul Bolle <pebolle@tiscali.nl>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

When authentication/association timed out, the driver would
complain bitterly, printing the message
ACTIVATE a non DRIVER active station id ... addr ...

The cause turns out to be that when the AP station is added
but we don't associate, the IWL_STA_UCODE_INPROGRESS is set
but never cleared. This then causes iwl_restore_stations()
to attempt to resend it because it uses the flag internally
and uploads even if it didn't set it itself.

To fix this issue and not upload the station again when it
has already been removed by mac80211, clear the flag after
adding it in case we add it only for association.

Cc: stable@vger.kernel.org
Reviewed-by: NMeenakshi Venkataraman <meenakshi.venkataraman@intel.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

I saw that when the watchdog triggers, the packets do go
through if we wait enough time. So we still have an issue
where packets are blocked in the Tx queue for a short while
and this needs to be debugged separately. For now, don't
restart the driver when it is not needed.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The dwell time for scanning is currently limited
so that it fits into the timings inside the ucode
when that is tracking DTIM/beacon periods for the
AP(s) it's connected to.

However, when it's connected to two APs, those
may be in lockstep, for example if they both have
a DTIM interval of 100 TU, then one could be 50
TU after the other, leaving only 50 TU free to
be used by scanning.

Since we can't know how far apart they are the
only option is to restrict to 1/2 of the minium
of the two APs.

In theory, it would be possible to not use 1/2 of
the minimum but take into account that if they
have different intervals then there will be a bit
more time since they can't be in lockstep, but as
they will have 100 TU intervals in practice that
complex calculation will probably just result in
hard-to-find bugs.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

__get_str() is identical to (char *)__get_dynamic_array()
that is in the code now, substitute __get_str to make the
code more readable.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's one message that goes just over the
limit of 100 characters, so bump the limit to
110 to get rid of the warning from that.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Since the op_mode defines the queue mapping, let it do it
completely through the API functions.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Currently when mac80211 asks to change the interface
type, we will accept it for both the BSS and PAN
contexts. This is not terribly complicated today,
but with the addition of the P2P Device abstraction
the PAN context handling will get more complex, so
restrict mac_change_interface to the BSS context.

Also fix a small locking issue and use is_active
instead of the vif pointer to check if the other
context is activated, guarding exclusive interface
types on the BSS context (IBSS) against the PAN
context being used for something else.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When the first interface is active, then scanning
on it or the second interface can take a little
longer than 7s (I observed around 8s.) Bump the
timeout to 15s to avoid aborting a scan that is
still running, just taking more time.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

My previous commit to shorten the radio reset time
caused issues as the firmware checks the active
dwell time against the quiet time, asserting that
the dwell is >= quiet time. This isn't really
needed in case of passive scanning like here, but
of course we need to pass that check.

To fix this, override the quiet time to be the
same as the radio reset dwell time.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When debugging is disabled, the event log functions aren't
functional in the way that the debugfs file expects. This
leads to the debugfs access crashing. Since the event log
functions aren't functional then, remove the debugfs file
when CONFIG_IWLWIFI_DEBUG is not set.

Cc: stable@kernel.org
Reported-by: NLekensteyn <lekensteyn@gmail.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Now that the eeprom parsing code overrides the sku
field directly with 11n_disable parameters, there's
no longer a need to keep a copy of this field.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The effect of using a short single-channel scan
to reset the radio is that scanning a channel
that isn't in use needs to re-tune the radio.
This means that the dwell time is irrelevant,
so use a shorter time.

While at it, clean up the code for this a bit.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

My commit 26a7ca9a ("iwlwifi: refactor EEPROM reading/parsing")
broke the 11n_disable module parameter's BIT(0) to disable all HT
operation (using the other bits to disable aggregation only was
unaffected). Restore this by overriding the SKU when parsing the
EEPROM if the module parameter is set.
Reported-by: NMatthijs Kooijman <matthijs@stdin.nl>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When the device is doing an internal radio reset
scan, ROC can be rejected to the supplicant with
busy status which confuses it.

One option would be to queue the ROC and handle
it later, but since the radio reset scan is very
quick we can just wait for it to finish instead.

Also add a warning since we shouldn't run into
the case of having a scan active when requesting
a ROC in any other case since mac80211 will not
scan while ROC or ROC while scanning.
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This feature needs to be disabled for all NICs.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This variable was accessed without taking the lock.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This is really not needed, we already have a lock inside
the accesses to the prph.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This is not needed since notif_wait_lock is never accessed
from IRQ.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This is required by the op_mode API.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

A few op_mode of the op_mode API functions have requirements
on the running context of the caller. Document that.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

We introduced a lock here in ff1ffb85 ("iwlwifi: fix dynamic
loading").  But we missed an error path which needs an unlock.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This flow can actually happen due to a corner case in
mac80211: the station is deleted before we get a chance
to reclaim all the packets in flight in AGG queue.
The tid_data for this station is zeroed, and we lose
the match with the Tx queue.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Since request_module_nowait() can't be backported
use request_module() instead -- we don't need the
asynchronous behaviour of request_module_nowait()
here since we're running in the firmware request
work struct.
Tested-by: NDonald H Fry <donald.h.fry@intel.com>
Reviewed-by: NDonald H Fry <donald.h.fry@intel.com>
Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Due to commit 26a7ca9a ("iwlwifi: refactor EEPROM
reading/parsing") adding a new parameter, while commit
d2c8b15d ("iwlwifi: use correct supported firmware
for 6035 and 6000g2") added a new device structure we
need to add the parameter to the new device structure
to make 6035 device work.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

For AGG queues, we must match between the WiFi sequence
number and the TFD number. This is a HW (SCD) requirement.

This is a take two of my

    iwlwifi: add debug in Tx path in AGG flow

    This will allow us to catch bad cases in which the packets aren't in
    the right place on the ring.

which disappeared during code move.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

In fragmentation we don't update the write pointer of the
HW immediately. So we shouldn't modify the timer in that
case.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>