Remove frame logic from start_pool cb using the new iface
for async send.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
Remove frame logic from transceive cb using new iface
for async send.
For pn533_wq_mi_recv() use pn533_send_cmd_direct_async which
sends the cmd directly to the hardware, skipping cmd queue.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
Remove frame logic from InJmumpForDEP command using
the new iface for async send.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
Remove frame logic from TgSetData and TgGetData commands
using the new iface for async send.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
The reference count bump on the llcp Rx path is leading to a memory leak
whenever we're not receiving an I frame.
We fix that by removing the refcount bump (drivers must not free their
received skb) and using it only in the I frame path, when the frame is
actually queued. In that case, the skb will only be freed when someone
fetches it from userspace. in all other cases, LLCP received frames will
be freed when leaving the Rx work queue.
Reported-by: Eric Lapuyade <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
Not only it was improperly use to queue backlogged RX skbuffs, but it was
also not processed at all.
If the socket receive queue is full we simply drop the incoming packets.
Signed-off-by: Samuel Ortiz <[email protected]>
Remove obsolate pn533_send_cmd_frame_sync() and use
previously added new iface for sync send.
The new interface require the use of individual skb for each
cmd which removes some memcpy calls and hides frame logic.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
It is intended to replace pn533_send_cmd_frame_sync() iface which
requires from the caller to create complete frame.
The new function constructs a complete frame itself and sends it out
in sync manner. This way frame logic is hidden from the caller.
pn533_send_cmd_sync() returns ERR_PTR in case of an error or a pointer
to valid response sk_buff otherwise. The pointer must be freed by the
caller when it's been consumed.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
This iface is intended to be used with DEP transfers. It differs
from pn533_send_cmd_async() in the way the response skb is allocated.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
This is intendend to replace pn533_send_cmd_frame_async() which
requires from the caller to create a complete frame.
The new function constructs a frame and sends it out which hides the
frame logic and avoid code duplication.
The caller has to allocate skb and put its payload there, and finally
provide the skb together with a complete cb to pn533_send_cmd_async().
Response skb is allocated by the core part and pass to the caller cb.
Next, the caller has to free it when is not needed anymore or pass it
up to the stack.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
__pn533_send_cmd_frame_async() is called when lock is held so GFP_KERNEL
flag will be always used. Thus, having extra param does not optimise the
code.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
PN533_CMD_DATAEXCH_HEAD_LEN includes a frame header length which is not
seen at a glance. It can be missleading, so split it and define the
frame header length explicitly.
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
in_maxlen and out_maxlen was replaced with PN533_NORMAL_FRAME_MAX_LEN
Signed-off-by: Waldemar Rymarkiewicz <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
Set timestamp in sent and received sk_buffs. timestamp is then put in
msghdr structure in llcp_sock_recvmsg().
Signed-off-by: Thierry Escande <[email protected]>
Signed-off-by: Samuel Ortiz <[email protected]>
The length parameter should be sizeof(req->name) - 1 because there is no
guarantee that string provided by userspace will contain the trailing
'\0'.
Can be easily reproduced by manually setting req->name to 128 non-zero
bytes prior to ioctl(HIDPCONNADD) and checking the device name setup on
input subsystem:
$ cat /sys/devices/pnp0/00\:04/tty/ttyS0/hci0/hci0\:1/input8/name
AAAAAA[...]AAAAAAAAf0:af:f0:af:f0:af
("f0:af:f0:af:f0:af" is the device bluetooth address, taken from "phys"
field in struct hid_device due to overflow.)
Cc: [email protected]
Signed-off-by: Anderson Lizardo <[email protected]>
Acked-by: Marcel Holtmann <[email protected]>
Signed-off-by: Gustavo Padovan <[email protected]>
Adds the spectral scan feature for ath9k. AR92xx and AR93xx chips
are supported for now. The spectral scan is triggered by configuring
a mode through a debugfs control file. Samples can be gathered via
another relay debugfs file.
Essentially, to try it out:
echo chanscan > /sys/kernel/debug/ieee80211/phy0/ath9k/spectral_scan_ctl
iw dev wlan0 scan
cat /sys/kernel/debug/ieee80211/phy0/ath9k/spectral_scan0 > samples
echo disable > /sys/kernel/debug/ieee80211/phy0/ath9k/spectral_scan_ctl
This feature is still experimental.
The special "chanscan" mode is used to perform spectral scan while
mac80211 is scanning for channels. To allow this,
sw_scan_start/complete() ops have been added.
The patch contains code snippets and information from Zefir Kurtisi and
information provided by Adrian Chadd and Felix Fietkau.
Signed-off-by: Simon Wunderlich <[email protected]>
Signed-off-by: Mathias Kretschmer <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
For AR9550, program the synth value based on the ref. clock.
The logic for AR9550 is similar to AR9330, but keep the code
separate since changes for AR9330 are required - which would be
done later.
Signed-off-by: Sujith Manoharan <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
If queues are stopped in mwl8k_stop,
these should be started in mwl8k_start
Signed-off-by: Nishant Sarmukadam <[email protected]>
Signed-off-by: Yogesh Powar <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
When a station deauths, we do not delete the streams with state
AMPDU_STREAM_NEW and these remain created forever. Fix this issue by
removing such streams in the driver
Signed-off-by: Nishant Sarmukadam <[email protected]>
Signed-off-by: Yogesh Ashok Powar <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
Whenever TX ring is drained, priv->tx_wait still points
to a valid completion. Making sure that it points to
NULL before returning from the mwl8k_tx_wait_empty
check.
Signed-off-by: Nishant Sarmukadam <[email protected]>
Signed-off-by: Yogesh Ashok Powar <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
When AGC calibration is to be done, a GPM message with the
payload, MCI_GPM_WLAN_CAL_REQ has to be sent. Currently this falls
within the IQ-CAL code block which is incorrect. Fix this by using
a separate variable to decide when IQ-CAL is to be done separately
and call ar9003_mci_init_cal_req correctly.
Signed-off-by: Sujith Manoharan <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
IQ calibration doesn't complete and times out for half/quarter
rates, so skip it correctly.
Signed-off-by: Sujith Manoharan <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
My commit 379b82f4c9
("regulatory: pass new regdomain to reset function")
broke the restore_regulatory_settings() function due
to a logic change. Consider this change:
- reset_regdomains(true);
- cfg80211_regdomain = cfg80211_world_regdom;
+ reset_regdomains(true, cfg80211_world_regdom);
This looks innocent enough, until you realise that the
called function (reset_regdomains) also resets the
cfg80211_world_regdom pointer, so that the old version
of the code would use the new object it pointed to and
the new version of the code uses the old object. This
lead to a double-free of this object.
Since reset_regdomains() sets it to &world_regdom, use
that directly.
Reported-by: Sujith Manoharan <[email protected]>
Tested-by: Sujith Manoharan <[email protected]>
Reported-by: Bob Copeland <[email protected]>
Reported-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
The uevent callback doesn't protect its access to
last_request, which now causes a warning since
the conversion to get_last_request(). Fix this by
allowing to use RCU protection for last_request.
Reported-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
This patch is an enhacement to mwifiex_pcie driver to use
map/unmap PCI memory APIs. This reduces one memcpy each in TX
path and RX path, and enhances throughput.
Signed-off-by: Avinash Patil <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
This patch adds handler to clean PCIe TX rings after disconnect
or bss stop is called for PCIe based mwifiex driver.
Signed-off-by: Avinash Patil <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
This patch uses pci_alloc_consistent and pci_free_consistent
APIs for mwifiex_pcie driver. Consistent DMA memory is allocated
for TX, RX and event rings. Command buffer and command response
buffer also uses map/unmap memory APIs to download commands and
get command responses.
Signed-off-by: Avinash Patil <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
This patch adds support for init_fw_port handler for PCIe
interface, which resets RXBD read pointer for PCIe.
This fixes issue where RX doesn't work until some TX from
driver happens.
Signed-off-by: Avinash Patil <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
This patch fixes a bug for few instances where PCIe interrupt
status variable is accessed without holding spin lock.
This can result into missing interrupts.
Fix this by copying interrupt status to a local variable and
then using it for calling specific routine.
Signed-off-by: Avinash Patil <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
The new name better matches the use of this variable.
Signed-off-by: Hauke Mehrtens <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
Some cores do not have a IRQ assigned and some do not support when an
IRQ is assigned to them, this is now handled and printed out in the
code.
Signed-off-by: Hauke Mehrtens <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
The assignment of the IRQs to the cores of the chips by iterating over
the cores is complicated and causes problems with SoC like the BCM4706
with two GMAC core where just one should get a dedicated IRQ number.
Now the code assigns the same IRQs to the cores as the code from the
Broadcom SDK does. If the SoC is not know the current assigned IRQs are
only read out and an error message is printed.
Signed-off-by: Hauke Mehrtens <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
This bcma_find_core_unit() is needed by the mips driver in the next
patch.
Signed-off-by: Hauke Mehrtens <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
TX_IQ_ON_AGC_CAL should not be enabled for AR9340. TX-IQ calibration
is run as part of AGC calibration only for AR9485, AR9462 and AR9565.
For the others (AR9300, AR9330, AR9340), TX-IQ cal is done independent
of AGC-cal.
Signed-off-by: Sujith Manoharan <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
