When using poll() to wait for the next period -- or avail_min samples --
one gets a consistent delay for each system call that is usually just a
little short of the selected period time. However, When using
snd_pcm_read/write(), one gets a jittery delay that alternates between
less than a millisecond and approximately two period times. This is
caused by snd_pcm_lib_{read,write}1() transferring any available samples
to the user's buffer and adjusting the application pointer prior to
sleeping to the end of the current period. When the next period
interrupt occurs, there is then less than avail_min samples remaining to
be transferred in the period, so we end up sleeping until a second
period occurs.

This is solved by using runtime->twake as the number of samples needed
for a wakeup in addition to selecting the proper wait queue to wake in
snd_pcm_update_state(). This requires twake to be non-zero when used
by snd_pcm_lib_{read,write}1() even if avail_min is zero.
Signed-off-by: NDave Dillow <dave@thedillows.org>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

This patch changes the string based list management to a handle base
implementation to help with the hot path use of pm-qos, it also renames
much of the API to use "request" as opposed to "requirement" that was
used in the initial implementation.  I did this because request more
accurately represents what it actually does.

Also, I added a string based ABI for users wanting to use a string
interface.  So if the user writes 0xDDDDDDDD formatted hex it will be
accepted by the interface.  (someone asked me for it and I don't think
it hurts anything.)

This patch updates some documentation input I got from Randy.
Signed-off-by: Nmarkgross <mgross@linux.intel.com>
Signed-off-by: NRafael J. Wysocki <rjw@sisk.pl>

Clemens Ladisch noted for hw_ptr_removal in "cleanup & merge hw_ptr
update functions" commit:

"It is possible for the status/delay ioctls to be called when the sound
card's pointer register alreay shows a position at the beginning of the
new period, but immediately before the interrupt is actually executed.
(This happens regularly on a SMP machine with mplayer.)  When that
happens, the code thinks that the position must be at least one period
ahead of the current position and drops an entire buffer of data."

Return back the hw_ptr_interrupt variable. The last interrupt pointer
is always computed from the latest hw_ptr instead of tracking it
separately (in this case all hw_ptr checks and modifications might
influence also hw_ptr_interrupt and it is difficult to keep it
consistent).
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

This change fixes the "ALSA: pcm_lib - optimize wake_up() calls for PCM I/O"
commit. New sleeping queue is introduced to separate user space and kernel
space wake_ups. runtime->nowake is renamed to twake (transfer wake).
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

pgprot_noncached() can be set for vmalloc'ed buffers safely, and we'd
need non-cached behavior more or less, even for the intermediate ring-
buffers.

Now snd_pcm_lib_mmap_vmalloc() is added as the common PCM mmap callback
that is coupled with snd_pcm_lib_alloc_vmalloc_buffer() & co.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

As noted by pl bossart <bossart.nospam@gmail.com>, the PCM I/O routines
(snd_pcm_lib_write1, snd_pcm_lib_read1) should block wake_up() calls
until all samples are not processed.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Do general cleanup in snd_pcm_update_hw_ptr*() routines and merge them.
The main change is hw_ptr_interrupt variable removal to simplify code
logic. This variable can be computed directly from hw_ptr.

Ensure that updated hw_ptr is not lower than previous one (it was possible
with old code in some obscure situations when interrupt was delayed or
the lowlevel driver returns wrong ring buffer position value).
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

In some debug cases, it might be usefull to see previous ring buffer
positions to determine position problems from the lowlevel drivers.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

There are now five copies of the code to allocate a PCM buffer using
vmalloc().  Add a sixth in the core so that the others can be removed.
Signed-off-by: NClemens Ladisch <clemens@ladisch.de>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Record the pid of the task that opened a PCM substream. For sound
cards with hardware mixing, this allows determining which process
is associated with a specific substream's volume control.
Signed-off-by: NClemens Ladisch <clemens@ladisch.de>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Re-export snd_pcm_format_name() function to be used outside the PCM core.
As a first example, usbaudio is changed to use it now again.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Using SG-buffers with dma_alloc_coherent() is often very inefficient
on non-coherent architectures because a tracking record could be
allocated in addition for each dma_alloc_coherent() call.
Instead, simply disable SG-buffers but just allocate normal continuous
buffers on non-supported (currently all but x86) architectures.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Replace the house-made div64_32() with the standard div_u64*() functions.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Move the fifo_size assignment to hw->ioctl callback to allow lowlevel
drivers overwrite the default behaviour.

fifo_size is in frames not bytes as specified in asound.h and alsa-lib's
documentation, but most hardware have fixed byte based FIFOs. Introduce
internal SNDRV_PCM_INFO_FIFO_IN_FRAMES.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Added runtime->delay field to adjust the delayed samples for snd_pcm_delay().
Typically a hardware FIFO length is stored in this field, so that the
extra delay between hwptr and applptr can be computed.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Some drivers like Intel8x0 or Intel HDA are broken for some hardware variants.
This patch adds more strict buffer position checks based on jiffies when
internal hw_ptr is updated. Enable xrun_debug to see mangling of wrong
positions.

As a side effect, the hw_ptr interrupt update routine might do slightly better
job when many interrupts are lost.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

The timer callbacks are called in the protected status by the lock
of the timer instance, so there is no need for an extra lock in the
PCM substream.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Impact: cleanup

snd_pcm_new takes a char *id argument, although it is not modifying
the string. it can therefore be declared as const char *id.
Signed-off-by: NTim Blechmann <tim@klingt.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Most hardwares have limited buffer-descriptor table length.  This
also restricts the max buffer size of the sound driver.
For example, snd-hda-intel has 1MB buffer size limit, and this is
because it can have at most 256 BDL entries.  For supporting larger
buffers, we need to allocate larger pages even for sg-buffers.

This patch changes the sgbuf allocation code to try to allocate
larger pages first.  At each head of the allocated pages, the
number of allocated pages is stored in the lowest bits of the
corresponding entry of the table addr field.  This change isn't
visible as long as the driver uses snd_sgbuf_get_addr() helper.

Also, the patch adds a new function, snd_pcm_sgbuf_get_chunk_size().
This returns the size of the chunk on continuous pages starting at
the given position offset.  If the chunk reaches to a non-continuous
page, it returns the size to the boundary.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Clean up SG-buffer helper functions and macros.  Helpers take substream
as arguments now.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Kill snd_assert() in sound/core/*, either removed or replaced with
if () with snd_BUG_ON().
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

When compiled with CONFIG_SND_DYNAMIC_MINORS the ALSA core is fine
to have more than 8 PCM devices per card, except one place - the
SNDRV_CTL_IOCTL_PCM_NEXT_DEVICE ioctl, which will not enumerate
devices > 7. This patch fixes the issue, changing the devices list
organisation.

Instead of adding new device to the tail, the list is now kept always
ordered (by card number, then device number). Thus, during enumeration,
it is easy to discover the fact that there is no more given card's
devices.

Additionally the device field of struct snd_pcm had to be changed to int,
as its "unsignednity" caused a lot of problems when comparing it to
potentially negative signed values. (-1 is 0xffffffff or even more then ;-)
Signed-off-by: NPawel Moll <pawel.moll@st.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

This reverts commit fb3d6f2b77bdec75d45aa9d4464287ed87927866.

New, updated patch with same subject replaces this commit.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

When compiled with CONFIG_SND_DYNAMIC_MINORS the ALSA core is fine
to have more than 8 PCM devices per card, except one place - the
SNDRV_CTL_IOCTL_PCM_NEXT_DEVICE ioctl, which will not enumerate
devices > 7. This patch fixes the issue, changing the devices list
organisation.

Instead of adding new device to the tail, the list is now kept always
ordered (by card number, then device number). Thus, during enumeration,
it is easy to discover the fact that there is no more given card's
devices. The same limit was present in OSS emulation code. It has
been fixed as well.

Additionally the device field of struct snd_pcm is now int, instead of
unsigned int, as there is no obvious reason for keeping it unsigned.
This caused a lot of problems with comparing this value with other
(almost always signed) variables. There is just one more place where
device number is unsigned - in struct snd_pcm_info, which should be
also sorted out in future.
Signed-off-by: NPawel MOLL <pawel.moll@st.com>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

This reverts commit 36b34d2437104f323e09d7c6af6451d3c0b9c0cd.

From: Al Viro <viro@ZenIV.linux.org.uk>

WIW, *all* this stuff is not bitwise at all.  For crying out loud, half
of these types are routinely used as array indices and loop variables...

If anything, we want a different set of allowed operations - subtraction
between elements of type (yielding integer), addition/subtraction of
integer types not bigger than ours (yielding our type), comparisons,
assignments (=, +=, -=, passing to function as argument, return from
function, initializers) and second/third arguments in ?:.  With 0 *not*
being allowed as a constant of such type.

It's not bitwise; we may use the same infrastructure in sparse, but it
should be a separate class of types (__attribute__((affine))).

dma_addr_t is another candidate for the same treatment, but there we'll
need helpers for conversions to hw-acceptable form (dma_to_le32(), etc.)
and gradual conversion of drivers.

ALSA ones and pm mess are absolutely straightforward cases, though.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Fully half of all alsa sparse warnings are from snd_pcm_hw_param_t degrading
to integer type, this goes a long way towards eliminating them.
Signed-off-by: NHarvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

The 'tick' in PCM is set (again) via sw_params.  And, nobody uses
this feature at all except for a command line option of aplay.
(This is literally 'nobody', as I checked alsa-lib API calls in all
 programs in major distros.)
Above all, if we need finer wake-ups for the position update, it's
basically an issue that the driver should solve, not tuned by each
application.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

The xfer_align sw_params parameter has never been used in a sane manner,
and no one understands what this does exactly.  The current
implementation looks also buggy because it allows write of shorter size
than xfer_align.  So, if you do partial writes, the write isn't actually
aligned at all.
Removing this parameter will make some pcm_lib_* code more readable
(and less buggy).
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

We need an accurate and continuous (monotonic) time sources to do
accurate synchronization among more timing sources. This patch allows
to enable monotonic timestamps for ALSA PCM devices and enables monotonic
timestamps for ALSA timer devices.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Signed-off-by: NJaroslav Kysela <perex@perex.cz>

The volatile prefix is just useless there.  Let's kill them, and then
gcc will be happier, too.
   sound/acore/pcm.c:867: warning: passing argument 1 of ‘__constant_c_and_count_memset’ discards qualifiers from pointer target type
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Fixed cast messes in pcm.h.
    include/sound/pcm.h: In function ‘hw_param_interval_c’:
    include/sound/pcm.h:800: warning: passing argument 1 of ‘hw_param_interval’ discards qualifiers from pointer target type
Simply redefine the inline functions again for const pointers.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Add a snd_pcm_rate_to_rate_bit() function to factor out common code used
by several drivers.
Signed-off-by: NClemens Ladisch <clemens@ladisch.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Merge the rates[] arrays from pcm_misc.c and pcm_native.c because they
are both the same.
Signed-off-by: NClemens Ladisch <clemens@ladisch.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Added a new macro snd_pcm_group_for_each_entry() just for code cleanup.
Old macros, snd_pcm_group_for_each() and snd_pcm_group_substream_entry(),
are removed.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Many struct file_operations in the kernel can be "const".  Marking them const
moves these to the .rodata section, which avoids false sharing with potential
dirty data.  In addition it'll catch accidental writes at compile time to
these shared resources.
Signed-off-by: NArjan van de Ven <arjan@linux.intel.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

fixes: include/sound/pcm.h:62: warning: 'struct snd_pcm_substream' declared
inside parameter list
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Users ask us many times about the maximum DMA size for PCM devices. This
file gives them a hint in KB.
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

This patch adds a struct device pointer to struct snd_pcm in order to be
able to give it a different device than the card. It defaults to the card's
device, however, so it should behave identically for drivers not touching
the field.
Signed-off-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NJaroslav Kysela <perex@suse.cz>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>