Merge branch 'fix/intel' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into asoc-intel

Conflicts:
	sound/soc/intel/sst-baytrail-dsp.c

Commit 10df3509 ("ASoC: Intel: Fix Audio DSP usage when IOMMU is
enabled.") caused following regression in Baytrail SST:

baytrail-pcm-audio baytrail-pcm-audio: error: DMA alloc failed
baytrail-pcm-audio baytrail-pcm-audio: error: failed to load firmware

Fix this by calling dma_coerce_mask_and_coherent() in sst_byt_init() with
the same dma_dev device what is now used in sst_fw_new() when allocating the
DMA buffer.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

This follows the same idea than commit 10df3509
("ASoC: Intel: Fix Audio DSP usage when IOMMU is enabled.") by using only
ACPI device for all DMA allocations. Since DMA masking is already done in
firmware loading it can be removed from here.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Since there is no support for compressed audio in Baytrail ADSP firmware
there is no need to leave it on during suspend since ALSA PCM buffers are
too small for leaving ADSP on for playing or recording.

Implement PM callbacks to Baytrail byt-rt5640.c machine driver that call
snd_soc_suspend and snd_soc_resume functions and unset the ignore_suspend
fields in DAI links.

This makes soc-core and ALSA core gracefully suspend and resume active
stream and call sst_byt_pcm_trigger() during suspend-resume cycle.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Add suspend and resume support to Baytrail SST DSP. This is implemented by
unloading firmware modules and putting DSP into reset prior suspend and
restarting DSP again in normal boot state after resume.

Context restore for running streams is implemented by scheduling a work from
sst_byt_pcm_trigger() that will allocate a stream with existing parameters
and start it from last known buffer position before suspend.

[Jarkko: Squashed together 5 WIP patches from Liam and 1 from me]
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Suspend/resume requires reloading FW to boot state so we need to also make
sure that the driver matches the FW state at boot.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

We have to save the physical address of extended firmware block in the
beginning of mailbox every time when we boot the DSP firmware since that
mailbox address is re-used after DSP firmware is running. Otherwise DSP
firmware will get bogus extended firmware block address during next DSP
boot.

Currently this is not problem but becomes when DSP runtime rebooting is
implemented. Prepare for that by moving extended firmware address saving
from sst_byt_init() to sst_byt_boot().
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Stream start position will be needed in resume code. Prepare for it by
adding start offset argument to sst_byt_stream_start().
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Baytrail ADSP stream IPC simplifies a little by moving IPC_IA_START_STREAM
construction and sending directly into sst_byt_stream_start() from
sst_byt_stream_operations(). This is because IPC_IA_START_STREAM is only
stream IPC with extra message data so this move saves a few code lines.

Main motivation for this is to prepare for passing stream start position
to sst_byt_stream_start() which will be needed in resume code.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

This is for preparing suspend/resume support but can give also more
safeguard against concurrent timestamp structure access between DSP firmware
and host.

Now DSP DMA pointer is sampled in each pcm pointer callback in
sst_byt_pcm_pointer() but that is unneeded since DSP updates the timestamp
period basis and can potentially be racy if sst_byt_pcm_pointer() is called
when DSP is updating the timestamp.

By taking DSP DMA pointer only after period elapsed IPC messages in
byt_notify_pointer() and returning stored hw pointer in
sst_byt_pcm_pointer() there is less risk for concurrent access.

The same stored hw pointer can be also used in suspend/resume code for
restarting the stream at the same position.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Since commit 4b68b4e1 (ASoC: Intel: split the pcm and compress to
different files) the compressed ops haven't been built causing link
failures on allyesconfig and making the driver unbuildable.  Add the
object to the Makefile to fix that.
Signed-off-by: Mark Brown <broonie@linaro.org>
Acked-by Vinod Koul <vinod.koul@intel.com>

this is for further updates to driver which supports DPCM :)
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

With DPCM we have media dai used and no seperate headset and speaker dai so
remove the speaker dai
The vibra is no longer supported thru audio, so remove
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

For manging them and adding support for more platforms
Code move only
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

as this will be used in compressed split file in subsequent patch
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

to sst-mfld-platform-pcm.c so that we can split pcm and compress to different
files for upcoming changes to support more platforms
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

As this address can move
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

to the place near it is used
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Add some SST API calls to unload and reload firmware modules. This can be used
by PM code to restore state and also allow modular FW to unload and release
memory blocks.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Block offset calculations are done in the contiguous allocator so
are not required here.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

This patch fixes the following dereference check ordering.

 sound/soc/intel/sst-haswell-pcm.c:749 hsw_pcm_probe() warn: variable dereferenced before check 'pdata' (see line 746)

 git remote add asoc git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git
 git remote update asoc
 git checkout 0b708c87
 vim +/pdata +749 sound/soc/intel/sst-haswell-pcm.c

 a4b12990 Mark Brown    2014-03-12  740  };
 a4b12990 Mark Brown    2014-03-12  741
 a4b12990 Mark Brown    2014-03-12  742  static int hsw_pcm_probe(struct snd_soc_platform *platform)
 a4b12990 Mark Brown    2014-03-12  743  {
 a4b12990 Mark Brown    2014-03-12  744  	struct sst_pdata *pdata = dev_get_platdata(platform->dev);
 a4b12990 Mark Brown    2014-03-12  745  	struct hsw_priv_data *priv_data;
 0b708c87 Liam Girdwood 2014-05-02 @746  	struct device *dma_dev = pdata->dma_dev;
 0b708c87 Liam Girdwood 2014-05-02  747  	int i, ret = 0;
 a4b12990 Mark Brown    2014-03-12  748
 a4b12990 Mark Brown    2014-03-12 @749  	if (!pdata)
 a4b12990 Mark Brown    2014-03-12  750  		return -ENODEV;
 a4b12990 Mark Brown    2014-03-12  751
 a4b12990 Mark Brown    2014-03-12  752  	priv_data = devm_kzalloc(platform->dev, sizeof(*priv_data), GFP_KERNEL);
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Read the stream offset and presentation position from DSP memory rather
than using the old estimated position. This fixes timing issues with
pulseaudio.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

hw_params() can be called multiple times. Make sure we release the DSP
stream that was allocated on previous hw_params() calls before allocating
a new DSP stream.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

The Intel IOMMU requires that the ACPI device is used to allocate all
DMA memory buffers. This means we need to pass the DMA device pointer into child
component devices that allocate DMA memory.

We also only set the DMA mask for the ACPI device now instead of for each
component device.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Fix page table creation on Haswell and Broadwell to remove unsafe
virt_to_phys mappings and use more portable SG buffer. Use audio buffer
APIs to allocate DMA buffers.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Make sure we add the allocated blocks to the modules list of blocks.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Make sure we dont alloc blocks twice with requests spanning more
than one block.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

I suppose there is a possibility that hsw_notification_work() may run after
sst_hsw_stream_free() which can lead to a kernel crash since struct
sst_hsw_stream is freed at that point and
stream = container_of(work, struct sst_hsw_stream, notify_work) is not valid
when hsw_notification_work() is run.
Reported-by: Derek Basehore <dbasehore@chromium.org>
Reported-by: Wenkai Du <wenkai.du@intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

There is a race between sst_byt_stream_free() and sst_byt_get_stream()
if sst_byt_get_stream() called from sst_byt_irq_thread() context is
accessing the byt->stream_list while a stream is deleted from the list.

A stream is added to byt->stream_list in sst_byt_stream_new() and deleted in
sst_byt_stream_free(). sst_byt_get_stream() is always protected by
sst->spinlock, but the stream addition and deletion are not protected.

The patch adds spinlock to both stream addition and deletion.

[Jarkko: Same fix added to sst-haswell-ipc.c too]
Signed-off-by: Wenkai Du <wenkai.du@intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

There were occasional ADSP crash during reboot testing:

[   11.883364] BUG: unable to handle kernel paging request at ffffc90121700000
[   11.883380] IP: [<ffffffffc024d8bc>] sst_module_insert_fixed_block+0x24f/0x26d [snd_soc_sst_dsp]
[   11.883397] PGD 7800b067 PUD 0
[   11.883405] Oops: 0002 [#1] SMP
[   11.886418] gsmi: Log Shutdown Reason 0x03

The virtual address, ffffc90121700000, was out of range. The virtual
address is calculated by adding LPE base address with an offset:

sst_memcpy32(dsp->addr.lpe + data->offset, data->data, data->size);

The offset is calculated in sst_byt_parse_module, by subtraction of
two virtual addresses dsp->addr.fw_ext and dsp->addr.lpe:

block_data.offset = block->ram_offset + (dsp->addr.fw_ext - dsp->addr.lpe);

These virtual addresses are assigned by kernel from ioremap:

sst->addr.lpe = ioremap(pdata->lpe_base, pdata->lpe_size);
sst->addr.fw_ext = ioremap(pdata->fw_base, pdata->fw_size);

In current driver code, offset is defined as unsigned int32:

struct sst_module_data {
...
	u32 offset;		/* offset in FW file */
};

Most of the time kernel assigned virtual addresses with addr.fw_ext
greater than addr.lpe. But sometimes it was the other way round.

Fix the problem by declaring offset as signed int32_t.
Signed-off-by: Wenkai Du <wenkai.du@intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Fix an incorrect sizeof() usage in sst_hsw_stream_get_volume(). sst_dsp_read()
is called to read into a variable of type u32, but is passed sizeof(u32 *) for
argument 'size_t bytes'. Detected by Coverity: CID 1195260.
Signed-off-by: Christian Engelmayer <cengelma@gmx.at>
Signed-off-by: Mark Brown <broonie@linaro.org>

The intent was to say "sizeof(*pos)" and not "sizeof(pos)".

The sizeof(*pos) is 8 bytes so the bug won't show up on 64 bit systems.
The sizeof(*dx) is 172 bytes so that will be a bugfix.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mark Brown <broonie@linaro.org>

Remove a self assignment in sst_mem_block_alloc_scratch(). When calculating
buffer sizes there is no need for statements without effect. Detected by
Coverity: CID 1195249.
Signed-off-by: Christian Engelmayer <cengelma@gmx.at>
Signed-off-by: Mark Brown <broonie@linaro.org>

Fix an incorrect sizeof() usage in sst_hsw_stream_get_volume(). sst_dsp_read()
is called to read into a variable of type u32, but is passed sizeof(u32 *) for
argument 'size_t bytes'. Detected by Coverity: CID 1195260.
Signed-off-by: Christian Engelmayer <cengelma@gmx.at>
Signed-off-by: Mark Brown <broonie@linaro.org>

Some versions of gcc even warn about it:

  mm/shmem.c: In function ‘shmem_file_aio_read’:
  mm/shmem.c:1414: warning: ‘error’ may be used uninitialized in this function

If the loop is aborted during the first iteration by one of the two
first break statements, error will be uninitialized.

Introduced by commit 6e58e79d ("introduce copy_page_to_iter, kill
loop over iovec in generic_file_aio_read()").
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

On 32 bit, size_t is "unsigned int", not "unsigned long", causing the
following warning when comparing with PAGE_SIZE, which is always "unsigned
long":

  fs/cifs/file.c: In function ‘cifs_readdata_to_iov’:
  fs/cifs/file.c:2757: warning: comparison of distinct pointer types lacks a cast

Introduced by commit 7f25bba8 ("cifs_iovec_read: keep iov_iter
between the calls of cifs_readdata_to_iov()"), which changed the
signedness of "remaining" and the code from min_t() to min().
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull slab changes from Pekka Enberg:
 "The biggest change is byte-sized freelist indices which reduces slab
  freelist memory usage:

    https://lkml.org/lkml/2013/12/2/64"

* 'slab/next' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/linux:
  mm: slab/slub: use page->list consistently instead of page->lru
  mm/slab.c: cleanup outdated comments and unify variables naming
  slab: fix wrongly used macro
  slub: fix high order page allocation problem with __GFP_NOFAIL
  slab: Make allocations with GFP_ZERO slightly more efficient
  slab: make more slab management structure off the slab
  slab: introduce byte sized index for the freelist of a slab
  slab: restrict the number of objects in a slab
  slab: introduce helper functions to get/set free object
  slab: factor out calculate nr objects in cache_estimate

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>