Sometimes under load, the Blackfin core is able to send SPI register
updates out before the controller is actually disabled.  So when we
go to reprogram the entire state (to switch to a different slave),
make sure we sync after disabling the controller.
Signed-off-by: NBarry Song <barry.song@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

We can't rely on the SPI_CTL/SPI_FLG registers retaining their state when
suspending, so save/restore their entire values.
Signed-off-by: NBarry Song <barry.song@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The common SPI layers take care of detecting CS conflicts and preventing
two devices from claiming the same CS.  This causes problems for the GPIO
CS support we currently have as we are using CS0 to mean "GPIO CS".  But
if we have multiple devices using a GPIO CS, the common SPI layers see
multiple devices using the virtual "CS0" and reject any such attempts.

To make both work, we introduce an offset define.  This represents the
max number of hardware CS values that the SPI peripheral supports.  If
the CS is below this limit, we know we can use the hardware CS.  If it's
above, we treat it as a GPIO CS.  This keeps the CS unique as seen by
the common code and prevents conflicts.
Signed-off-by: NBarry Song <barry.song@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

This lets us push the short SPI MMR bit names out of the global namespace.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The driver that we based ours on uses a little extra memory behind the
normal driver state, but we don't.  So drop this useless bit of memory.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The current structure names are a bit confusing as to what they represent,
so use better names.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

Rather than having to look up the same 3 sets of functions at the same
time, just use an ops structure so we only need to set one pointer.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

No point in creating our own version of true/false defines when there is
already a standard stdbool available to us.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The chip ops should always be initialized, so having null fallback
functions are useless.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

As David points out, the cs_change_per_word option isn't standard, nor is
anyone actually using it.  So punt all of the dead code considering it
makes up ~10% of the code size.
Reported-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The CS helper functions were toggling both the Flag Enable and the Flag
Value bits, but the Flag Value bit is ignored if the corresponding Flag
Enable bit is cleared.  So under high speed transactions, the CS sometimes
would not toggle properly.

Since it makes no sense to toggle the Flag Enable bit dynamically when we
actually want to control the Flag Value, do this when setting up the device
and then only handle toggling of the CS value during runtime.
Signed-off-by: NBarry Song <barry.song@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

The current behavior in PIO mode is to poll the SPI status registers which
can obviously lead to higher latencies when doing a lot of SPI traffic.
There is a SPI interrupt which can be used instead to signal individual
completion of transactions.
Signed-off-by: NYi Li <yi.li@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

We should make sure the SPI controller is in a sane state in case the
boot loader left it in a crappy state.  Such as DMA pending which causes
interrupts to fire on us.

When setting a sane initial state, do not default to slave mode.  If we
do, then the SPI peripheral may implicitly take over the SPISS pin which
other things might be using.

For example, the BF533-STAMP uses this pin as a GPIO to control switching
between ethernet and flash.  If the SPI peripheral controls the output
state instead, the ethernet is no longer accessible.

URL: http://blackfin.uclinux.org/gf/tracker/5630Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NMichael Hennerich <michael.hennerich@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

Anomaly 05000119 states that the DMA_RUN bit with peripherals isn't
reliable.  However, the way the driver is currently written (DMA IRQ
callback), we don't need the polling in the first place, so drop it.
Signed-off-by: NSonic Zhang <sonic.zhang@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

Re-order setup() a bit so we don't leak memory/dma/gpio resources upon
errors.  Also make sure we don't call kfree() twice on the same object.
Signed-off-by: NDaniel Mack <daniel@caiaq.de>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Signed-off-by: NYi Li <yi.li@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h

percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files.  percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.

percpu.h -> slab.h dependency is about to be removed.  Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability.  As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.

  http://userweb.kernel.org/~tj/misc/slabh-sweep.py

The script does the followings.

* Scan files for gfp and slab usages and update includes such that
  only the necessary includes are there.  ie. if only gfp is used,
  gfp.h, if slab is used, slab.h.

* When the script inserts a new include, it looks at the include
  blocks and try to put the new include such that its order conforms
  to its surrounding.  It's put in the include block which contains
  core kernel includes, in the same order that the rest are ordered -
  alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
  doesn't seem to be any matching order.

* If the script can't find a place to put a new include (mostly
  because the file doesn't have fitting include block), it prints out
  an error message indicating which .h file needs to be added to the
  file.

The conversion was done in the following steps.

1. The initial automatic conversion of all .c files updated slightly
   over 4000 files, deleting around 700 includes and adding ~480 gfp.h
   and ~3000 slab.h inclusions.  The script emitted errors for ~400
   files.

2. Each error was manually checked.  Some didn't need the inclusion,
   some needed manual addition while adding it to implementation .h or
   embedding .c file was more appropriate for others.  This step added
   inclusions to around 150 files.

3. The script was run again and the output was compared to the edits
   from #2 to make sure no file was left behind.

4. Several build tests were done and a couple of problems were fixed.
   e.g. lib/decompress_*.c used malloc/free() wrappers around slab
   APIs requiring slab.h to be added manually.

5. The script was run on all .h files but without automatically
   editing them as sprinkling gfp.h and slab.h inclusions around .h
   files could easily lead to inclusion dependency hell.  Most gfp.h
   inclusion directives were ignored as stuff from gfp.h was usually
   wildly available and often used in preprocessor macros.  Each
   slab.h inclusion directive was examined and added manually as
   necessary.

6. percpu.h was updated not to include slab.h.

7. Build test were done on the following configurations and failures
   were fixed.  CONFIG_GCOV_KERNEL was turned off for all tests (as my
   distributed build env didn't work with gcov compiles) and a few
   more options had to be turned off depending on archs to make things
   build (like ipr on powerpc/64 which failed due to missing writeq).

   * x86 and x86_64 UP and SMP allmodconfig and a custom test config.
   * powerpc and powerpc64 SMP allmodconfig
   * sparc and sparc64 SMP allmodconfig
   * ia64 SMP allmodconfig
   * s390 SMP allmodconfig
   * alpha SMP allmodconfig
   * um on x86_64 SMP allmodconfig

8. percpu.h modifications were reverted so that it could be applied as
   a separate patch and serve as bisection point.

Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: NTejun Heo <tj@kernel.org>
Guess-its-ok-by: NChristoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>

Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Bryan Wu <cooloney@kernel.org>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Move some common spi_setup() error checks into the SPI framework from the
spi_master controller drivers:

 - Add a new "mode_bits" field to spi_master

 - Use that in spi_setup to validate the spi->mode value being
   requested.  Setting this new field is now mandatory for any
   controller supporting more than vanilla SPI_MODE_0.

 - Update all spi_master drivers to:

     * Initialize that field
     * Remove current spi_setup() checks using that value.

This is a net minor code shrink.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Start moving some spi_setup() functionality into the SPI core from the
various spi_master controller drivers:

 - Make that function stop being an inline;

 - Move two common idioms from drivers into that new function:
    * Default bits_per_word to 8 if that field isn't set
    * Issue a standardized dev_dbg() message

This is a net minor source code shrink, and supports enhancments found in
some follow-up patches.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

bfin_spi_flush() returns limit, which reaches -1 upon timeout.  but in
function bfin_spi_pump_transfers() it is compared with 0.
Signed-off-by: NRoel Kluin <roel.kluin@gmail.com>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: Bryan Wu <cooloney@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The latter naming convention is much more common.
Signed-off-by: NJie Zhang <jie.zhang@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>

Signed-off-by: NMike Rapoport <mike@compulab.co.il>
Cc: Bryan Wu <bryan.wu@analog.com>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

1. Rewrite of the non-dma data transfer functions to use only ONE mode
   of TIMOD (TIMOD=0x1).  With TIMOD=0, it was not possible to set the TX
   bit pattern.  So the TDBR = 0xFFFF inside the read calls won't work.

2. Clear SPI_RDBR before reading and before duplex transfer.
   Otherwise the garbage data in RDBR will get read.  Since mmc_spi uses a
   lot of duplex transfers, this is the main cause of mmc_spi failure.

3. Poll RXS for transfer completion.  Polling SPIF or TXS cannot
   guarantee transfer completion.  This may interrupt a transfer before it
   is finished.  Also this may leave garbage data in buffer and affect
   next transfer.

[Yi Li <yi.li@analog.com>: add a field "u16 idle_tx_val" in "struct
bfin5xx_spi_chip" to specify the value to transmit if no TX value
is supplied.]
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NYi Li <yi.li@analog.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add support for GPIO controlled SPI Chip Selects.  To make use of this
feature, set chip_select = 0 and add a proper cs_gpio to your
controller_data.

struct spi_board_info
        .chip_select = 0

struct bfin5xx_spi_chip
        .cs_gpio = GPIO_P###

There are various SPI devices that require SPI MODE_0, and need to have
the Chip Selects asserted during the entire transfer.  Consider using
SPI_MODE_3 (SPI_CPHA | SPI_CPOL) if your device allows it.
Signed-off-by: NMichael Hennerich <michael.hennerich@analog.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fix NULL pointer crash when cleaning up from invalid platform resources
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Do this because when things crash, we get simple names like "setup" and
"start_queue" which is pretty difficult to trace back to the real thing:
the spi driver
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

According to comments in linux/spi/spi.h:

 * All SPI transfers start with the relevant chipselect active.  Normally
 * it stays selected until after the last transfer in a message. Drivers
 * can affect the chipselect signal using cs_change.
 *
 * (i) If the transfer isn't the last one in the message, this flag is
 * used to make the chipselect briefly go inactive in the middle of the
 * message.  Toggling chipselect in this way may be needed to terminate
 * a chip command, letting a single spi_message perform all of group of
 * chip transactions together.
 *
 * (ii) When the transfer is the last one in the message, the chip may
 * stay selected until the next transfer.  On multi-device SPI busses
 * with nothing blocking messages going to other devices, this is just
 * a performance hint; starting a message to another device deselects
 * this one.  But in other cases, this can be used to ensure correctness.
 * Some devices need protocol transactions to be built from a series of
 * spi_message submissions, where the content of one message is determined
 * by the results of previous messages and where the whole transaction
 * ends when the chipselect goes intactive.
Signed-off-by: NYi Li <yi.li@analog.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This bug can be observed when two SPI devices are sharing the spi bus: One
device is set as SPI CS 7, another one is using SPI CS 4.

In spi_bfin5xx.c: cs_active(), cs_deactive() are used to control SPI_FLG
register.  From the debug bellow:

cs_active: flag: 0x7f91, chip->flag: 0x7f80, cs: 7
cs_active: flag: 0xef91, chip->flag: 0xef10, cs: 4

When device A (cs_7) activate CS 7, SPI_FLG is set as 0x7f91 (however,
SPI_FLG should be set as 0x7f80, or 0x6f91 if in broadcast mode).

Due to some HW bug (very possibly), if SPI_FLG is set as 0x7f91, SPISSEL7
is asserted, however SPISSEL4 will be asserted too (I can see this using
the scope).  This is unreasonable according to HRM.
Signed-off-by: NYi Li <yi.li@analog.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Without this change, SPI DMA is not reliably under stress tests.
Obiviously it's a hardware issue which is not addressed by any document.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The "while" endless loop will cause the system hang if hardware error, so
we add timeout control to make the system alive.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When using a BF533-STAMP here with a W25X10 SPI flash.  It works fine when
enable_dma is disabled, but doesn't work at all when turning DMA on.  We
get just 0xff bytes back when trying to read the device.

Change the code around so that it programs the SPI first and then enables
DMA, it seems to work a lot better ...
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- remove duplicated definition MAX_SPI_SSEL

- remove unnecessary array size
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

use the properl BIT_CTL_...  defines rather than the internal driv er
CFG_SPI_...  defines
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We only need to check SPI error when DMA failes, cause that is the DMA IRQ
handling routine.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Because of DMA hardware issue, we were trying to use software workaround.
This patch add some useful debug messages to help us debugging the DMA
code.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Blackfin's related DMA callback API doesn't need void * cast, so drop it.
And this driver is for all Blackfin processors not only for BF53x, we
update the DMA request label for more meaningful information.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

For DMA TX/RX operation in pump_transfers, DMA contriguration code in TX
and RX paths are almost the same.  This patch unify the duplicated DMA
code to make it more readable.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If the SPI bus registers a receive overflow error, pass the result back up
to the higher levels.
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use len_in_bytes when we care about the number of bytes transferred rather
than the number of spi transactions.  (this value will be the same for
8bit transfers, but not any other sizes)
Signed-off-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NBryan Wu <cooloney@kernel.org>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>