Smatch complains:
drivers/staging/wfx/hif_rx.c:26 hif_generic_confirm() warn: negative user subtract: 0-u16max - 4
20 static int hif_generic_confirm(struct wfx_dev *wdev,
21 const struct hif_msg *hif, const void *buf)
22 {
23 // All confirm messages start with status
24 int status = le32_to_cpup((__le32 *)buf);
25 int cmd = hif->id;
26 int len = le16_to_cpu(hif->len) - 4; // drop header
^^^^^
27
28 WARN(!mutex_is_locked(&wdev->hif_cmd.lock), "data locking error");
In fact, rx_helper() already make the necessary checks on the value of
hif->len. Never mind, add an explicit check to make Smatch happy.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201009171307.864608-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Smatch complains:
bus_spi.c:228 wfx_spi_probe() warn: 'bus->core' could be an error pointer
bus_sdio.c:221 wfx_sdio_probe() warn: 'bus->core' could be an error pointer
bus->core contains the result of wfx_init_common(). With this patch,
wfx_init_common() returns a valid pointer or NULL.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201009171307.864608-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Smatch complains:
drivers/staging/wfx/hif_rx.c:177 hif_scan_complete_indication() warn: potential NULL parameter dereference 'wvif'
drivers/staging/wfx/data_tx.c:576 wfx_flush() warn: potential NULL parameter dereference 'wvif'
Indeed, if the vif id returned by the device does not exist anymore,
wdev_to_wvif() could return NULL.
In add, the error is not handled uniformly in the code, sometime a
WARN() is displayed but code continue, sometime a dev_warn() is
displayed, sometime it is just not tested, ...
This patch standardize that.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201009171307.864608-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The device can be connected on SPI or on SDIO. The original file
described the two options separately. So, most of the file had to be
rewritten in order to match with the Yaml requirements.
Some device requirements are still written in the comments since they
cannot been expressed with the current scheme (e.g. reg must be set to 1
with SDIO, interrupt is mandatory with SPI, reset-gpio in SPI is
replaced by mmc-pwrseq in SDIO, etc...).
The examples provided have also been reworked in order to make
dt_binding_check happy.
Finally, also fix typo in the name of the file (siliabs instead of
silabs)
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201007101943.749898-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The device is in charge of respecting the QoS constraints. The driver
have to ensure that all the queues contain data and the device choose
the right queue to send.
The things starts to be more difficult when the bandwidth of the bus is
lower than the bandwidth of the WiFi. The device quickly sends the
frames of the highest priority queue. Then, it starts to send frames
from a lower priority queue. Though, there are still some high priority
frames waiting in the driver.
To work around this problem, this patch add some priorities to each
queue. The weigh of the queue was (roughly) calculated experimentally by
checking the speed ratio of each queue when the bus does not limit the
traffic:
- Be/Bk -> 20Mbps/10Mbps
- Vi/Be -> 36Mbps/180Kbps
- Vo/Be -> 35Mbps/600Kbps
- Vi/Vo -> 24Mbps/12Mbps
So, if we fix the weigh of the Background to 1, the weight of Best
Effort should be 2. The weight of Video should be 116. However, since
there is only 32 queues, it make no sense to use a value greater than
64[1]. And finally, the weight of the Voice is set to 128.
[1] Because of this approximation, with very slow bus, we can still
observe frame starvation when we measure the speed ratio of Vi/Be. It is
around 35Mbps/1Mbps (instead of 36Mbps/180Kbps). However, it is still in
accepted error range.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201007101943.749898-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Firmwares with API < 3.6 do not forward DELBA requests. Thus, when a
Block Ack session is restarted, the reordering buffer is not flushed and
the received sequence number is not contiguous. Therefore, mac80211
starts to wait some missing frames that it will never receive.
This patch disables the reordering buffer for old firmware. It is
harmless when the network is unencrypted. When the network is encrypted,
the non-contiguous frames will be thrown away by the TKIP/CCMP replay
protection. So, the user will observe some packet loss with UDP and
performance drop with TCP.
Fixes: e5da5fbd77 ("staging: wfx: fix CCMP/TKIP replay protection")
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201007101943.749898-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit e8d607ce0c ("staging: wfx: drop 'secure link' feature") had
removed the 'secure link' feature. However, a few lines of codes were
yet here.
Fixes: e8d607ce0c ("staging: wfx: drop 'secure link' feature")
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201007101943.749898-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As expected, when the device detect a MMIC error, it returns a specific
status. However, it also strip IV from the frame (don't ask me why).
So, with the current code, mac80211 detects a corrupted frame and it
drops it before it handle the MMIC error. The expected behavior would be
to detect MMIC error then to renegotiate the EAP session.
So, this patch correctly informs mac80211 that IV is not available. So,
mac80211 correctly takes into account the MMIC error.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20201007101943.749898-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewing sram_write_dma_safe(), there are two
identical calls to virt_addr_valid(). The second
call can be simplified by a comparison of variables
set from the first call.
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Signed-off-by: Tom Rix <trix@redhat.com>
Link: https://lore.kernel.org/r/20200912144719.13929-1-trix@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver is now close to leave the staging directory. Update the TODO
list to reflect the work done.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-32-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The parameter "async" in wfx_cmd_send() allows to send command without
waiting for the reply. In this case, the mutex hif_cmd.lock is released
asynchronously in the context of the receiver workqueue.
However, "kbuild test robot" complains about this architecture[1] since
it is not able to follow the lock duration of hif_cmd.lock (and indeed,
the state of the driver if the hardware wouldn't reply is not well
defined).
Besides, this feature is not really necessary. It is only used by
hif_shutdown(). This function hijack the 'async' flag to run a command
that won't answer.
So, this patch removes the 'async' flag and introduces a 'no_reply' flag.
Thus, the mutex hif_cmd.lock is only acquired/released from
hif_cmd_send(). Therefore:
- hif_shutdown() does not have to touch the private data of the struct
hif_cmd
- Kbuild test robot should be happy
- the resulting code is simpler
[1] https://lore.kernel.org/driverdev-devel/alpine.DEB.2.21.1910041317381.2992@hadrien/
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-31-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
hif_tx_mib.c contains functions that format data to be sent to the
hardware. In this file, sometime the struct to be sent is named 'arg',
sometime 'val'. In some other function 'val' is used for the argument of
the function.
This patch uniformize the things and choose to call all the data in
destination to the hardware 'arg' (note this choice is only dictated by
the number of lines to change in the code)
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-30-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is no reason to place two spaces between the field tx_conf_payload
and its type.
In the same vein, remove duplicate empty lines between declarations.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-29-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the wfx driver, the prefix 'hif_mib_' is normally used for structures
that represent a hardware message. hif_mib_tx_rate_retry_policy does not
fall in this category. So, rename it.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-28-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The maximum length of a SSID is defined by 802.11 specification. It is
already defined in mac80211: IEEE80211_MAX_SSID_LEN. Therefore, use this
generic definition.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-27-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This struct hif_ie_tlv is definitively an Information Element (IE). This
struct is defined by 802.11 specification and already exists in
mac80211. Reuse this definition instead of struct hif_ie_tlv.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-26-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The union hif_indication_data is never used in the driver. So, it is not
necessary to declare it separately from hif_ind_generic.
In add, drop prefix 'indication_' from the names 'indication_type' and
'indication_data' since it is redundant with the name of the struct.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-25-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The union hif_event_data is never used in the driver. So, it is
not necessary to declare it separately from hif_ind_event.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-24-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The union hif_privacy_key_data is never used in the driver. So, it is
not necessary to declare it separately from hif_req_add_key.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-23-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The structs hif_capabilities, hif_otp_regul_sel_mode_info and
hif_otp_phy_info have no real reasons to exist. Drop them and simplify
access to fields of struct hif_ind_startup.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-22-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The structs hif_scan_type, hif_scan_flags and hif_auto_scan_param have
no real reasons to exist (apart maybe defining namespaces). Moreover,
the names of the fields within these structs are not all meaningful.
Drop the structs and rename the fields.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-21-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The structs hif_queue, hif_data_flags, hif_tx_flags and
hif_ht_tx_parameters have no real reasons to exist. Drop them and
simplify access to fields of struct hif_req_tx.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-20-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_tx_result_flags has no reason to exist. Drop it and simplify
access to struct hif_cnf_tx.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-19-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_pm_mode has no reason to exist. Drop it and simplify access
to struct hif_req_set_pm_mode.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-17-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_suspend_resume_flags has no reason to exist. Drop it and
simplify access to struct hif_ind_suspend_resume_tx.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-16-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_map_link_flags has no reason to exist. Drop it and simplify
access to struct hif_req_map_link.
Also rename the field 'map_direction' in 'unmap'. It is more
meaningful and allows to drop enum hif_sta_map_direction.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-15-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_ie_flags has no reason to exist. Drop it and simplify
access to struct hif_req_update_ie.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Struct hif_reset_flags has no reason to exist. Drop it and simplify
access to struct hif_req_reset.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Enum hif_beacon is not used. Moreover, it is just another definition of
a boolean. Absolutely useless.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-9-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since the code for multicast filtering has been dropped, the function
hif_set_data_filtering() is only called to disable multicast filtering.
In fact, the multicast filtering is already disabled by default. So,
this function is useless and can be dropped.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-8-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The device allows to filter multicast frames. The driver has the
necessary code to take advantage of this feature. However, some bugs
has been reported on this feature. So, it was temporary disabled.
Finally, the things work well as-is for more than one year now. So there
is no plan to enable this feature in near future.
Since we dislike to maintain dead code, drop it.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Secure Link (slk) feature allows to encrypt (and authenticate) the
traffic between the host and the device. The official implementation of
this feature relies on mbedTLS. For that reason, this implementation is
not included in the current driver. To be included, the implementation
has to rely on kernel crypto API instead of mbedTLS.
So, for now, the driver contained stub functions waiting for the new
implementation based on kernel crypto API.
This new implementation represent a bunch of work and it is unlikely to
be done in near future. Moreover, we strongly dislike to maintain
useless code. So, drop all the code related to secure link.
Whoever wants to implement secure link should revert this patch before
starting to work on it.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A new kind of error has appeared in API 3.4.
The Linux driver is not concerned by this new error, but let's keep the
API in sync with the firmware.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The file hif_tx_mib.c expects to contain functions that format messages
for the hardware. It is unexpected to find function that manipulate
RCU and structures from mac80211.
Keep hif_set_association_mode() with the code necessary for message
formatting and relocate the smart part in wfx_join_finalize().
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_join() and wfx_join_finalize() are the two halves of the association
process. Group them.
In addition, for better uniformity of the code, rename wfx_do_join() in
wfx_join().
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The statements in wfx_bss_info_changed() has no particular order.
For better readability, group and sort the statements relative to the
association processing.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200907101521.66082-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The host and the device can be connected with a called Wake-Up GPIO.
When the host fall down this GPIO, it allows the device to enter in deep
sleep and no communication with the device is no more possible (the
device wakes up automatically on DTIM and fetch data if necessary).
So, before to communicate with the device, the driver have to raise the
Wake-up GPIO and then wait for an IRQ from the device.
Unfortunately, old firmwares have a race in sleep/wake-up process and
the device may never wake up. In this case, the IRQ is not sent and
driver complains with "timeout while wake up chip". Then, the driver
tries anyway to access the bus and an other error is raised by the bus.
Fortunately, when the bug occurs, it is possible to fall down the IRQ
and the device will eventually finish the sleep process. Then the driver
can wake it up normally.
The patch implements that workaround and add a retry limit in case
something goes very wrong.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The initial developer has feared msecs_to_jiffies() could round down the
result. However, the documentation of msecs_to_jiffies() says that the
result is rounded upward. So the increment of the result of
msecs_to_jiffies() is not necessary.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the old days, ieee80211 powersave has some impact on the Rx speed.
These problems are solved for a long time now. There is no more reason
to not enabling it.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jérôme Pouiller