This isolates the common code that is required to use an mei bus nfc
device from an NFC HCI drivers. This prepares for future drivers for
NFC chips connected behind an Intel Management Engine controller.
The microread_mei HCI driver is also modified to use that common code.
Signed-off-by: NEric Lapuyade <eric.lapuyade@intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

All NFC devices will now get proper RFKILL support as long as they provide
some dev_up and dev_down hooks. Rfkilling an NFC device will bring it down
while it is left to userspace to bring it back up when being rfkill unblocked.
This is very similar to what Bluetooth does.
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

And return the proper string for it.
Acked-by: NJohannes Berg <johannes@sipsolutions.net>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Some devices turn radio on whenever they're asked to start a poll.
To prevent that from happening, we just don't call into the driver
start_poll hook when the NFC device is down.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Add missing usb_put_dev on failure path in pn533_probe().

Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: NMarina Makienko <makienko@ispras.ru>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

According to the LLCP specs, we must terminate the LLCP link when receiving
a DISC with both ssap and dsap set to 0.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

local_cleanup is always called with device set to false as it means the
local LLCP is going away. So no need to pass this switch as an argument.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

When the MAC goes down, connected and connection less sockets should be
notified, but raw sockets should be kept alive.
They will get notified only when the physical devices goes away.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Major features added in 0.2 version:

* frame ops added to support wider set of devices
* support of ACS ACR122U
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

pn533_abort_cmd() aborts last command sent to the controller
and cancels already requested urb.

As ACR122U does not support any mechanism (as ACK for standard
PN533) which aborts last command this cannot be issued for this
device. Otherwise, acr122u will behave in an unstable way.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

ACS ACR122U is an USB NFC reader, PC/SC and CCID compilant, based
on NXP PN532 chip.

Internally, it's build of MCU, PN532 and an antenna. MCU makes the
device CCID and PC/SC compilant and provide USB connection.

In this achitecture, a host cannot talk directly to PN532 and must
rely on MCU. Luckily, MCU exposes pseud-APDU through PC/SC Escape
mechanism which let the host to transmit standard PN532 commands
directly to PN532 chip with some limitations.

The frame roughly looks like:

    CCID header    |          APDU header           |    PN532 header
(pc_to_rdr_escape) |  (pseudo apdu Direct Tramsmit) | (len, TFI, cmd, params)

Accordign to limitations, ACR122U does't provide any mechanism to
abort last issued command.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

As not all devices require ACK confirmation of every request sent to the
controller, differentiate two protocol types.

First one, request-ack-response and the second one request-response type.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Rename 'wq_in_error' field to more relevant 'status'
and move it to cmd context struct.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Reorder code to avoid functions declaration.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

In all cases (send_cmd_async, send_data_async and send_sync)
pn533_send_async_complete() handles all responses internally,
so there is no need to pass this as a callback.

Cmd context is passed to __pn533_send_frame_async in all the
cases as well. It's already kept in struct pn533 which is
available all the time the device is attached. So we can make
use of it instead.

Therefore, cmd_complete and cmd_complete_arg are no needed any more.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

We must free 'cmd_complete_mi_arg' and not 'cmd_complete_arg'
when getting send error handling fragmented response.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

'cmd->code' looks better then 'cmd->cmd_code'
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Keep cmd context in pn533 struct instead of only cmd code.
The context already includes cmd_code.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Use struct pn533_cmd instead of pn533_send_async_complete_arg
to track the context of the issued cmd.

This way pn533_send_async_complete_arg struct is no needed
anymore. Just move issuer complete callback to pn533_cmd struct.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

In case of error from __pn533_send_frame_async() while sending
next cmd from the queue (cmd_wq), cmd->req, cmd->resp and
cmd->arg pointers won't be freed.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Define explicitely it is Pasori specific reset command.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Remove duplicated authors info from the header as well.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Depends on timing division by zero can happen when user stops
polling.

pn533_stop_poll() resets modulation counter on stop_poll, but
meanwhile we get response for last poll request and try, despite
of stop poll request, to schedule next modulation for polling.

Log message:
[345.922515] pn533 1-1.3:1.0: pn533_stop_poll
[345.928314] pn533 1-1.3:1.0: pn533_send_ack
[345.932769] pn533 1-1.3:1.0: Received a frame.
[345.937438] PN533 RX: 00 00 ff 03 fd d5 4b 00 e0 00
[345.942840] pn533 1-1.3:1.0: pn533_poll_complete
[345.947753] pn533 1-1.3:1.0: pn533_start_poll_complete
[345.953186] Division by zero in kernel.
[345.957244] [<c001b38c>] (unwind_backtrace+0x0/0xf0)
[345.965698] [<c0260e78>] (Ldiv0+0x8/0x10)
[345.974060] [<c0260e60>] (__aeabi_idivmod+0x8/0x18)
[345.983978] [<c02e0064>] (pn533_poll_complete+0x3c0/0x500)
[345.994903] [<c02df690>] (pn533_send_async_complete+0x7c/0xc0)
[346.005828] [<c02e028c>] (pn533_wq_cmd_complete+0x1c/0x34)
[346.016113] [<c005d134>] (process_one_work+0x1ac/0x57c)
[346.025848] [<c005d85c>] (worker_thread+0x168/0x42c)
[346.034576] [<c00620c0>] (kthread+0xa4/0xb0)
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

For better debugging as the codes are defined in hex in the spec.
Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Signed-off-by: NWaldemar Rymarkiewicz <waldemar.rymarkiewicz@tieto.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Useful for LLCP validation tests.
Signed-off-by: NThierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

This resets remote parameters in both local and socket llcp structures when the
link goes down. That way, nfc_llcp_getsockopt won't return values corresponding
to the previous link parameters.
Signed-off-by: NThierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

If remote_miu value is not set in the socket (i.e. connection-less socket) the
value stored in the local is used.
Signed-off-by: NThierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

This adds support for AGF PDUs. For each PDU contained in the AGF, a new sk_buff
is allocated and dispatched to its corresponding handler.
Signed-off-by: NThierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

LLCP Validation test #2 (Connection-less information transfer) send a
service data unit of zero octets length. This is now handled correctly.
Signed-off-by: NOlivier Guiter <olivier.guiter@linux.intel.com>
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

If a socket option has not been set by the user, fall back to the LLCP
local ones.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

The MIUX must be transmitted in big endian and as such we have to convert
it properly.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

Those can be very verbose and we only want them when debugging pn533.
Signed-off-by: NSamuel Ortiz <sameo@linux.intel.com>

There is better to use kstrto* instead of simple_strtoul. In this case it
applies a bit stricter rules for input as well.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

In routine _rtl8188e_read_power_value_fromprom(), there are loops initializing
index_cck_base and index_bw40_base from the PROM. As the result of a typo, the
second loop is ended one element too soon.
Signed-off-by: NLarry Finger <Larry.Finger@lwfinger.net>
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Inside one FIFO slot queue, EDMA chipsets have the same link pointer
re-read race condition as older chipsets, so the same buffer holding
logic needs to be used in order to avoid use-after-free bugs.
Unlike on older chips, it can be skipped for the end of the queue.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

If AR_CRCErr, AR_PHYErr, AR_DecryptCRCErr or AR_MichaelErr is indicated
in the rx status word, but AR_RxFrameOK is also set, the descriptor
contents are typically invalid. This can show up as a warning about
invalid MCS rates in a frame. Even with those checks in place, a
descriptor with invalid MCS rates can still sometimes make it through to
the driver (mostly on older hardware like AR91xx).

Detect such errors in the last descriptor of a frame and discard the
whole frame if present.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>