The ACX_TSF_INFO command now takes role_id as param.
change the struct accordingly, and pass the wlvif
to the wl1271_acx_tsf_info() function.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Use device role for scans when the sta is not
associated.
sched_scan is used only when the sta is
not associated, and thus should use
the dev role (instead of sta role).
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The set_template commands now takes the role_id as
parameter.
Usually, we'll use the vif's main role_id.
However, sometimes we'll want to use
wlvif->dev_role_id instead of wlvif->role_id,
so pass the wanted role_id as param.
Update WL127X_FW_NAME/WL128X_FW_NAME.
(This commit starts a series of fw update patches,
and changes the start() callback to return an error
in order to prevent the use of the driver during the
transition. This change will be reverted in the
last patch of series)
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Since we reverted to claiming the host only when needed,
we no longer need to release/claim the host on suspend/resume.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
This reverts commit 393fb560d3.
Commit b6ad726 ("mmc: core: Prevent too long response times
for suspend") fails the suspend if the mmc host can't be
claimed before suspend. As the host is claimed by us as long
as the chip is powered on, suspend will always fail.
Revert to claiming the sdio bus only when needed.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
wl1271_tx_update_filters() is used as some workaround
to open filters while roaming on the same channel.
However, it doesn't handle roaming to a different channel,
and it might also sleep in the tx path, which is a bug.
With the new auth/assoc redesign, roaming is much simpler,
and this function is not needed anymore.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
with the new auth/assoc redesign, we get CHANGED_BSSID
indication before CHANGED_ASSOC indication, while our
CHANGED_BSSID handling block assumes we are already
associated.
Fix it by checking we are either in ibss mode, or
already associated.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The vif variable was being declared inside one of the internal blocks
of wl1271_event_process. This is not necessary, since this variable
is already declared in the function context.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Add a testmode command to retrieve the BD_ADDR that is stored in the
Fuse ROM in newer PGs. In old PGs this operation is not supported.
The caller can then derive the MAC addresses from it.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Add support for two MAC addresses. If the NVS has a valid MAC
address, that takes precedence and we use two sequential address
starting from the one specified.
If the NVS doesn't contain a valid MAC address (ie. if it is set to
00:00:00:00:00:00), we check if the HW PG version in use has the
BD_ADDR written in the fuse ROM. If it does, we read it and derive
the two subsequent addresses for WLAN.
During production, 3 addresses are reserved per device. The first for
Bluetooth (burnt in the fuse ROM) and the following two for WLAN.
This patch has some code by Igal and Arik (squashed from internal
patches).
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Igal Chernobelsky <igalc@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
In order to read the MAC addresses from the fuse ROM, we need to know
the chip ID and the HW PG version. We need to know the MAC address
during probe, because that's when we register our HW with mac80211.
To prepare for that, this patch reads the chip ID and HW PG version
during probe instead of doing it at boot time. We power the chip on
briefly in order to do that.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Up till now we only needed to access the partition table in boot.c.
But to add support for reading the MAC address from the FUSE in
testmode, we will have to change the partition in testmode.c.
Thus, we move the partition table to io.c and export it via io.h. It
makes more sense to have it in the io part anyway.
Signed-off-by: Luciano Coelho <coelho@ti.com>
In some cases a race condition can happen if we don't cancel any
pending ELP work before stopping PLT. With this commit we cancel ELP
work and clear the wl->flags bitmask. Also clean up the wl elements
after powering off.
Signed-off-by: Luciano Coelho <coelho@ti.com>
The interrupt threaded handler exits immediately if the driver's state
is WL1271_STATE_OFF. As a result, the interrupt status is not read. If
the interrupt is level triggered, it will be fired again.
Fix this by disabling interrupts before setting the state to OFF.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
When stopping plt, the chip is powered off before all current work items
are flushed and interrupts are disabled. This might introduce a race in
which the driver tries to communicate with a powered off chip.
Fix this by powering off the device only after interrupts are disabled
and all work items are flushed.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
__wl1271_plt_stop is called from both wl1271_plt_stop and
wl1271_unregister_hw. While wl1271_plt_stop acquires a mutex,
wl1271_unregister_hw does not.
Fix this by calling wl1271_plt_stop instead of __wl1271_plt_stop from
wl1271_unregister_hw.
Signed-off-by: Ido Yariv <ido@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
The sched_scan_(stop|start) ops fails to check for WL1271_STATE_OFF.
This can lead to a race where the driver tries to access the HW
while it's off.
Fix this by checking for WL1271_STATE_OFF before accessing the HW.
Signed-off-by: Pontus Fuchs <pontus.fuchs@gmail.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Following the good example of the Intel (and more recently Atheros)
drivers, enable endianess check by default when running sparse.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Following the good example of the Intel (and more recently Atheros)
drivers, enable endianess check by default when running sparse.
Signed-off-by: Luciano Coelho <coelho@ti.com>
The 32-bit values were not converted before writing them to the chip.
Change the wl1251_read32() and wl1251_write32() so that they always
read and write le32 values and convert to and from the CPU endianess.
Signed-off-by: Luciano Coelho <coelho@ti.com>
The wl1251 driver was generating the following warning:
drivers/net/wireless/wl1251/boot.c:467:21: warning: incorrect type in assignment (different base types)
drivers/net/wireless/wl1251/boot.c:467:21: expected unsigned int [unsigned] [assigned] [usertype] val
drivers/net/wireless/wl1251/boot.c:467:21: got restricted __le32 [usertype] <noident>
Fix this by removing one cpu_to_le32() call in the wrong place.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Jean Delvare reported bonding on top of 3c59x adapters was not detecting
network cable removal fast enough.
3c59x indeed uses a 60 seconds timer to check link status if carrier is
on, and 5 seconds if carrier is off.
This patch reduces timer period to 5 seconds if device is a bonding
slave.
Reported-by: Jean Delvare <jdelvare@suse.de>
Acked-by: Jean Delvare <jdelvare@suse.de>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
'poll' was a debugging option, but turning it on these days leads to
kernel panic. Remove it.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add unicast steering entries to resource tracker.
Do qp_detach also for these entries when VF doesn't shut down gracefully.
Otherwise there is leakage of these resources, since they are not tracked.
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Reviewed-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
mlx4_uc_steer_add/release() should not add MLX4_UC_STEER flag to gid.
It is added in mlx4_unicast_attach/detach().
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Reviewed-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
mlx4_qp_attach/detach_common() don't use hash variable, move it to find_entry()
static find_entry() in mcg.c doesn't use steer argument
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Reviewed-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
When adding new unicast steer entry, before moving qp to state ready,
actually before calling mlx4_RST2INIT_QP_wrapper(), there were added
a lot of entries with local_qpn=0 into radix tree.
This fact impacted the get_res() function and proper functioning
of resource tracker in addition to adding trash entries into radix tree.
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Reviewed-by: Yevgeny Petrilin <yevgenyp@melllanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
When passing MLX4_UC_STEER=1 it was translated to value 2
after mlx4_QP_ATTACH_wrapper. Therefore in new_steering_entry()
unicast steer entries were added to index 2 of array of size 2.
Fixing this bug by shift right to one position.
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Reviewed-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch updates the copyright dates in the tg3 driver.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch replaces tg3's private SPEED_INVALID and DUPLEX_INVALID
definitions with SPEED_UNKNOWN and DUPLEX_UNKNOWN respectively.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch removes the unneeded link_config.orig_* members. When the
phy is in a low power state, the TG3_PHYFLG_IS_LOW_POWER flag solely
determines how the link is configured. When the phy is powered back up,
it can resume using the original settings.
For the phylib case, the link configuration still needs to be saved, but
since the phylib maintains its own configuration, we can repurpose the
(unused in this case) tg3 link configuration members.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When tg3 devices are shutdown, the driver uses the
TG3_PHYFLG_IS_LOW_POWER flag to enable a static phy configuration.
Any attempt to use the link configuration variables is dead code. This
patch removes such code.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Detecting the ASIC revision of a device is getting to be an increasingly
complex process. This patch consolidates all the ASIC rev detection
code to a single routine for better maintainability.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The tg3 driver needs to submit a few phy register values to the UMP
firmware each time the link state changes. Up until now, the driver
would wait for the previous event to complete, then proceed to gather
data through a series of phy accesses. Since phy accesses are
relatively slow, it is possible for another thread to attempt to submit
its own event while the UMP code is still construction its message.
This patch seeks to minimize the collision window as much as possible by
preloading the phy data.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On newer devices, the hardware expects the NVRAM address register
to be written only once per NVRAM page. To do otherwise causes NVRAM
corruption. This patch fixes the problem.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When checking the autoneg advertisements, the driver failed to include
the master and master enable bits for the bcm5701. This patch fixes the
problem.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If the tg3 interface is down, the driver will return ethtool stats
uninitialized. This patch zeroes out the destination stat buffer in
such a case.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A bug was found in the new DMA engine for the 57766. This patch
disables it, which causes the device to fallback to the old DMA engine.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Log an explicit warning if we are unable to create MTDs for a net
device. Also correct the comment about why mtd_device_register() may
fail; there is no longer an MTD table to fill up.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
The 'page size' for PCIe DMA, i.e. the alignment of boundaries at
which DMA must be broken, is 4KB. Name this value as EFX_PAGE_SIZE
and use it in efx_max_tx_len(). Redefine EFX_BUF_SIZE as
EFX_PAGE_SIZE since its value is also a result of that requirement,
and use it in efx_init_special_buffer().
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
If efx_pci_probe_main() schedules an INVISIBLE or ALL reset (but
nothing more drastic), we retry it up to 5 times. So far as I'm
aware, this was a workaround for bugs in Falcon A0 which were fixed
in production silicon. Remove the retry.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
The code in efx_process_channel() to update the RX queue after each
batch of RX completions works out as a no-op on a TX-only channel
where the RX queue structure is set to all-zeroes, but
(1) efx_channel_get_rx_queue() will BUG() if DEBUG is defined, and
(2) it's a waste of time.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Per ./Documentation/CodingStyle, goto statements are acceptable for the
centralized exiting of functions when there are multiple exit points which
share common work such as cleanup. When no common work is required for
multiple exit points, the function should just return at these exit points
instead of doing an unnecessary jump to a centralized return. This patch
cleans up the inappropriate use of goto statements, and removes unnecessary
variables (or move to a smaller scope) where possible as a result of the
cleanups.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In the unlikely event that e1000_poll_fiber_serdes_link_generic() is called
and it returns an error, the returned error code value is not propagated to
the caller of e1000e_setup_fiber_serdes_link().
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In the unlikely event that e1000_setup_link_ich8lan() returns an error,
the returned error code value is not propagated to the caller of
e1000_init_hw_ich8lan().
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
