When a bad frame is detected for a bad crc.
We were reallocating and loosing the previous frame pointer.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Stabilize communication by using a mutex.
This avoids running a write transaction during a read retry or a read
transaction during a write retry.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

A frame starts with ST21NFCA_SOF_EOF(0x7e) + 0x00.
A frame ends with ST21NFCA_SOF_EOF(0x7e).
It is possible that the i2c macrocell is stopped for other
communication interfaces with highest priority(RF or SWP).
This can be seen with some 0xFF data at the end of a received shdlc buffer.
If this happen we need to discard the frame because the CLF will repeat it.

In order to push accurate data to hci layer, we add the following fix:
- Instead of looking for the first 0x7e in the frame, check that the last
received byte is 0x7e.
- Check that the first frame reception block start with start of
frame(0x7e 0x00). If not, clear the buffer.
- Check that the next frame reception block do not start with start of
frame(0x7e). If so, clear the buffer.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

There is no byte stuffing when data are stored in skb.
TAILROOM is 2 byte crc + 1 byte eof.

st21nfca_hci_remove_len_crc was doing an incorrect operation on
the TAILROOM data.
If shdlc timer T2 is triggered, it will request to send the same data.
Before every hci data was lost after st21nfca_hci_remove_len_crc.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Byte stuffing was not correctly removed after a i2c read operation.
This was causing improper crc calculation when byte stuffing was
applied to more than 1 byte.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Remove unneeded includes from i2c.c
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Change in st21nfca_hci_platform_init in order to handle in a better way the
internal reboot command.
Once the reboot is completed, the driver expect to receive a 0x7e filled
buffer.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Remove struct st21nfca_i2c_phy* as this parameter can be retrieve
through i2c_get_clientdata(client)
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

st21nfca_hci_probe return 0 in case of successful call and a different
value in any other cases.

There is no need to check for st21nfca_hci_probe return as this will be
checked after st21nfca_hci_i2c_probe is completed.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

wait_tab variable is already global and may create conflicts.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Fix "warning: array subscript is above array bounds" in load_session
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Fixing "sparse: cast to restricted __be16" message when building with
make C=1 CF=-D__CHECK_ENDIAN__
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Fixing "sparse: cast to restricted __be16" message when building with
make C=1 CF=-D__CHECK_ENDIAN__
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Fixes a null pointer exception occurring when the IRQ request in
pn544_hci_i2c_probe fails and no platform data is available.
Signed-off-by: Clément Perrochaud <clement.perrochaud@nxp.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Now that the NFC digital layer has support for the ISO/IEC 14443-B
protocol and type 4B tags, add the corresponding support to the
trf7970a driver.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Add support for the ISO/IEC 14443-B protocol and Type 4B tags.
It is expected that there will be only one tag within range so the full
anticollision scheme is not implemented. Only the SENSB_REQ/SENSB_RES
and ATTRIB_REQ/ATTRIB_RES are implemented.

CC: Thierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Add RF tech and framing macros for the ISO/IEC 14443-B Protocol.

Cc: Thierry Escande <thierry.escande@linux.intel.com>
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

The trf7970a driver recently had support added for the 'autosuspend-delay'
property so document it.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

The trf7970a driver currently uses a fixed autosuspend delay of 30 seconds.
To enable users to customize the delay as they see fit, add support for the
new 'autosuspend-delay' DTS property (part of the nfc node) which can
override the default 30 seconds.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

trf7970a_switch_rf_on() no longer returns anything other than 0 so make
it void and clean up the code that checks for errors when its called.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Add pm_runtime support by moving the code that enables the trf7970a to
the pm_runtime hook routines. The pm_runtime 'autosuspend' feature is
used so that the device isn't disabled until at least 30 seconds have
passed since trf7970a_switch_rf_off() was last called.

The result is that when trf7970a_switch_rf_on() is called, the device
will be enabled and initialized (if it isn't already).  When
trf7970a_switch_rf_off() is called, it will turn off the RF immediately
but leave the device enabled for at least 30 seconds.
If 30 seconds have passed and the pm_runtime facility decides to suspend
the driver, the device will be disabled then.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

The NFC digital layer calls the 'switch_rf' hook to turn the RF on before
it configures the RF technology and framing. There is potential to confuse
a tag doing it this way so don't enable the RF until the RF technology and
initial framing have been configured.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Currently, the trf7970a driver assumes that the Vin voltage is 5V when
it writes to the 'Chip Status Control' register. That may not be correct
so use the regulator facility to get the Vin voltage and set the VRS5_3
bit correctly when writing to that register.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Currently the driver writes the same value to the 'Modulator and SYS_CLK
Control' register no matter what RF technology is being used. That works
for now but new RF technologies (e.g., ISO/IEC 14443-B) will require
different values to be written to that register. To support this, add a
member to the trf7970a structure which is set by the RF technology handling
code and used by the framing code when it writes to that register.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Writing to the 'ISO Control' register may cause the contents of the
'Modulator and SYS_CLK Control' register to change so be sure to write
to 'Modulator and SYS_CLK Control' after writing to 'ISO Control'.

Note that writing to the 'Modulator and SYS_CLK Control' register
shouldn't be necessary at all according to the trf790a manual but testing
shows that it is necessary.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

The current code always writes to the 'ISO Control' register when the
RF framing is set.  That's not necessary since the register's value
doesn't always change. Instead, only write to it when its value is
actually being changed.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Sometimes after sending a frame there is tx data leftover in the FIFO
which the driver will think is part of the receive frame. That data can
be cleared when an 'End of TX' interrupt is received by issuing the
'FIFO Reset' command.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Due to a trf7970a erratum, the 'NFC Target Detection Level' register
(0x18) must be cleared after power-up.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Only initiate the abort command process when there is an active command.
If the abort process were started and there wasn't an active command
then the next command issued by the digital layer would be incorrectly
aborted.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

After further testing periods of ~16 ms have been observed
between interrupts indicating that there is receive data in
the FIFO.  To accomodate that, increase the time the driver
waits before deciding there is no more data to receive to
20 ms.  The macro that represents that delay is
'TRF7970A_WAIT_FOR_RX_DATA_TIMEOUT'.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Describe the properties used by the PN544 NFC controller driver.
Signed-off-by: Clément Perrochaud <clement.perrochaud@nxp.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Add functions to recover hardware resources from the device-tree when not
provided by the platform data.
Signed-off-by: Clément Perrochaud <clement.perrochaud@nxp.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

This implementation rely on the ST21NFCA_DEVICE_MGNT_GATE and
ST21NFCA_DM_GETINFO proprietary gates commands.

First we are retrieving a pipe list available on the CLF with the
ST21NFCA_DM_GETINFO_PIPE_LIST parameter. A gate<->pipe table match
is done with ST21NFCA_DM_GETINFO_PIPE_INFO for each pipe.
If the pipe is created and open, we fill st21nfca_gates table.
If the pipe is create but closed or is not created we keep the gate
with NFC_HCI_INVALID_PIPE.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

load_session allows a CLF to restore the gate <-> pipe table from some
proprietary location.
The main advantage to add this function is to reduce the memory wear by
running pipe creation (and storing) only once.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Extend it up to the maximum FWI value 4949 ms defined by the
ISO14443-3 specification.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Add driver for STMicroelectronics ST21NFCA NFC controller.
ST21NFCA is using HCI protocol, shdlc as LLC layer & I2C as
communication protocol.

Adding support for Reader/Writer mode with Tag type 1/2/3/4 A & B.
It is using proprietary gate 15 for ISO14443-3 such as type 1 &
type 2 tags. It is using proprietary gate 14 for type F tags.
ST21NFCA_DEVICE_MGNT_GATE gives access to proprietary CLF configuration.
Standard gate for ISO14443-4 A (13) & B (11) are also used.

ST21NFCA specific mecanism:

One particular point to notice for the data handling is that frame
does not contain any length value. Therefore the i2c part of this driver
is managing the reception with a read length sequence until the end of
frame (0x7e) is reached.

In order to avoid conflict between sof & eof a mecanism
called byte stuffing concist of an escape byte (0x7d) insertion before
special byte (0x7e, 0x7d). The special byte is then xored with 0x20.

In this driver, When data are available in the CLF, the interrupt
gpio is driven to active state and triggered an interrupt.
Once the i2c_master_recv start, the interrupt gpio is driven to idle
state until its complete. If the frame is incomplete or data are still
available, interrupts will be triggered again.
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>

Pull misc kbuild changes from Michal Marek:
 "Here is the non-critical part of kbuild:
   - One bogus coccinelle check removed, one check fixed not to suggest
     the obsolete PTR_RET macro
   - scripts/tags.sh does not index the generated *.mod.c files
   - new objdiff tool to list differences between two versions of an
     object file
   - A fix for scripts/bootgraph.pl"

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  scripts/coccinelle: Use PTR_ERR_OR_ZERO
  scripts/bootgraph.pl: Add graphic header
  scripts: objdiff: detect object code changes between two commits
  Coccicheck: Remove memcpy to struct assignment test
  scripts/tags.sh: Ignore *.mod.c

This patch fixes I/O errors with the sym53c8xx_2 driver when the disk
returns QUEUE FULL status.

When the controller encounters an error (including QUEUE FULL or BUSY
status), it aborts all not yet submitted requests in the function
sym_dequeue_from_squeue.

This function aborts them with DID_SOFT_ERROR.

If the disk has full tag queue, the request that caused the overflow is
aborted with QUEUE FULL status (and the scsi midlayer properly retries
it until it is accepted by the disk), but the sym53c8xx_2 driver aborts
the following requests with DID_SOFT_ERROR --- for them, the midlayer
does just a few retries and then signals the error up to sd.

The result is that disk returning QUEUE FULL causes request failures.

The error was reproduced on 53c895 with COMPAQ BD03685A24 disk
(rebranded ST336607LC) with command queue 48 or 64 tags.  The disk has
64 tags, but under some access patterns it return QUEUE FULL when there
are less than 64 pending tags.  The SCSI specification allows returning
QUEUE FULL anytime and it is up to the host to retry.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Commit 8f619b54 ("powerpc/ppc64: Do not turn AIL (reloc-on
interrupts) too early") added code to set the AIL bit in the LPCR
without checking whether the kernel is running in hypervisor mode.  The
result is that when the kernel is running as a guest (i.e., under
PowerKVM or PowerVM), the processor takes a privileged instruction
interrupt at that point, causing a panic.  The visible result is that
the kernel hangs after printing "returning from prom_init".

This fixes it by checking for hypervisor mode being available before
setting LPCR.  If we are not in hypervisor mode, we enable relocation-on
interrupts later in pSeries_setup_arch using the H_SET_MODE hcall.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>