Currently, snd_hdac_power_up()/down() helpers checks whether the codec
is being in pm (suspend/resume), and skips the call of runtime get/put
during it. This is needed as there are lots of power up/down
sequences called in the paths that are also used in the PM itself. An
example is found in hda_codec.c::codec_exec_verb(), where this can
power up the codec while it may be called again in its power up
sequence, too.
The above works in most cases, but sometimes we really want to wait
for the real power up. For example, the control element get/put may
want explicit power up so that the value change is assured to reach to
the hardware. Using the current snd_hdac_power_up(), however,
results in a race, e.g. when it's called during the runtime suspend is
being performed. In the worst case, as found in patch_ca0132.c, it
can even lead to the deadlock because the code assumes the power up
while it was skipped due to the check above.
For dealing with such cases, this patch makes snd_hdac_power_up() and
_down() to two variants: with and without in_pm flag check. The
version with pm flag check is named as snd_hdac_power_up_pm() while
the version without pm flag check is still kept as
snd_hdac_power_up(). (Just because the usage of the former is fewer.)
Then finally, the patch replaces each call potentially done in PM with
the new _pm() variant.
In theory, we can implement a unified version -- if we can distinguish
the current context whether it's in the pm path. But such an
implementation is cumbersome, so leave the code like this a bit messy
way for now...
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96271
Signed-off-by: Takashi Iwai <tiwai@suse.de>
we are dereferencing pcm first then checking pcm. instead now lets put
them in same if condition so that pcm is checked first.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Like the previous patches, this patch converts also to the regmap, at
this time, the cached verb writes are the target. But this conversion
needs a bit more caution than before.
- In the old code, we just record any verbs as is, and restore them at
resume. For the regmap scheme, this doesn't work, since a few verbs
like AMP or DIGI_CONVERT are asymmetrical. Such verbs are converted
either to the dedicated function (snd_hda_regmap_xxx_amp()) or
changed to the unified verb.
- Some verbs have to be declared as vendor-specific ones before
accessing via regmap.
Also, the minor optimization with codec->cached_write flag is dropped
in a few places, as this would confuse the operation. Further
optimizations will be brought in the later patches, if any.
This conversion ends up with a drop of significant amount of codes,
mostly the helper codes that are no longer used.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The amp hash table was used for recording the cached reads of some
capability values like pin caps or amp caps. Now all these are moved
to regmap as well.
One addition to the regmap helper is codec->caps_overwriting flag.
This is set in snd_hdac_override_parm(), and the regmap helper accepts
any register while this flag is set, so that it can overwrite even the
read-only verb like AC_VERB_PARAMETERS. The flag is cleared
immediately in snd_hdac_override_parm(), as it's a once-off flag.
Along with these changes, the no longer needed amp hash and relevant
fields are removed from hda_codec struct now.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This patch converts the amp access functions to the regmap helpers.
The amp values were formerly cached in the own hash table. Now it's
dropped by the regmap's cache.
The only tricky conversion is snd_hda_codec_amp_init(). This function
shouldn't do anything if the amp was already initialized. For
achieving this behavior, a value is read once at first temporarily in
the cache-only mode. Only if it returns an error, i.e. the item
still doesn't exist in the cache, it proceeds to the update.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Sometimes we need the uncached reads, e.g. for refreshing the tree.
This patch provides the helper function for that and uses it for
refreshing widgets, reading subtrees and the whole proc reads.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This patch adds an infrastructure to support regmap-based verb
accesses. Because o the asymmetric nature of HD-audio verbs,
especially the amp verbs, we need to translate the verbs as a sort of
pseudo registers to be mapped uniquely in regmap.
In this patch, a pseudo register is built from the NID, the
AC_VERB_GET_* and 8bit parameters, i.e. almost in the form to be sent
to HD-audio bus but without codec address field. OTOH, for writing,
the same pseudo register is translated to AC_VERB_SET_* automatically.
The AC_VERB_SET_AMP_* verb is re-encoded from the corresponding
AC_VERB_GET_AMP_* verb and parameter at writing.
Some verbs has a single command for read but multiple for writes. A
write for such a verb is split automatically to multiple verbs.
The patch provides also a few handy helper functions. They are
designed to be accessible even without regmap. When no regmap is set
up (e.g. before the codec device instantiation), the direct hardware
access is used. Also, it tries to avoid the unnecessary power-up.
The power up/down sequence is performed only on demand.
The codec driver needs to call snd_hdac_regmap_exit() and
snd_hdac_regmap_exit() at probe and remove if it wants the regmap
access.
There is one flag added to hdac_device. When the flag lazy_cache is
set, regmap helper ignores a write for a suspended device and returns
as if it was actually written. It reduces the hardware access pretty
much, e.g. when adjusting the mixer volume while in idle. This
assumes that the driver will sync the cache later at resume properly,
so use it carefully.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When the driver is unloaded before the codec is bound, it still keeps
the runtime PM refcount up, and results in the unbalance. This patch
covers these cases by introducing a flag indicating the runtime PM
initialization and handling the codec registration procedure more
properly. It also fixes the missing input beep device as a gratis,
too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add an overriding exec_verb op to struct hdac_device so that the call
via snd_hdac_exec_verb() can switch to a different route depending on
the setup. The codec driver sets this field so that it can handle the
errors or applying quirks appropriately. Furthermore, this mechanism
will be used for smooth transition for the regmap support in later
patches.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This patch changes the sysfs files assigned to the codec device on the
bus which were formerly identical with hwdep sysfs files. Now it
shows only a few core parameter, vendor_id, subsystem_id, revision_id,
afg, mfg, vendor_name and chip_name.
In addition, now a widget tree is added to the bus device sysfs
directory for showing the widget topology and attributes. It's just a
flat tree consisting of subdirectories named as the widget NID
including various attributes like widget capability bits. The AFG
(usually NID 0x01) is always found there, and it contains always
amp_in_caps, amp_out_caps and power_caps files. Each of these
attributes show a single value. The rest are the widget nodes
belonging to that AFG. Note that the child node might not start from
0x02 but from another value like 0x0a.
Each child node may contain caps, pin_caps, amp_in_caps, amp_out_caps,
power_caps and connections files. The caps (representing the widget
capability bits) always contain a value. The rest may contain
value(s) if the attribute exists on the node. Only connections file
show multiple values while other attributes have zero or one single
value.
An example of ls -R output is like below:
% ls -R /sys/bus/hdaudio/devices/hdaudioC0D0/
/sys/bus/hdaudio/devices/hdaudioC0D0/widgets/:
01/ 04/ 07/ 0a/ 0d/ 10/ 13/ 16/ 19/ 1c/ 1f/ 22/
02/ 05/ 08/ 0b/ 0e/ 11/ 14/ 17/ 1a/ 1d/ 20/ 23/
03/ 06/ 09/ 0c/ 0f/ 12/ 15/ 18/ 1b/ 1e/ 21/
/sys/bus/hdaudio/devices/hdaudioC0D0/widgets/01:
amp_in_caps amp_out_caps power_caps
/sys/bus/hdaudio/devices/hdaudioC0D0/widgets/02:
amp_in_caps amp_out_caps caps connections pin_caps pin_cfg
power_caps
/sys/bus/hdaudio/devices/hdaudioC0D0/widgets/03:
.....
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Now some codes and functionalities of hda_codec struct are moved to
hdac_device struct. A few basic attributes like the codec address,
vendor ID number, FG numbers, etc are moved to hdac_device, and they
are accessed like codec->core.addr. The basic verb exec functions are
moved, too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
A few basic codes for communicating over HD-audio bus are moved to
struct hdac_bus now. It has only command and get_response ops in
addition to the unsolicited event handling.
Note that the codec-side tracing support is disabled temporarily
during this transition due to the code shuffling. It will be
re-enabled later once when all pieces are settled down.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Define the common hd-audio driver and device types to bind over
snd_hda_bus_type publicly. This allows to implement other type of
device and driver code over hd-audio bus.
Now both struct hda_codec and struct hda_codec_driver inherit these
new struct hdac_device and struct hdac_driver, respectively.
The bus registration is done in subsys_initcall() to assure it
before any other driver registrations.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This patch enables the finer power state control of each widget
depending on the jack plug state and streaming state in addition to
the existing power_down_unused power optimization. The new feature is
enabled only when codec->power_mgmt flag is set.
Two new flags, pin_enabled and stream_enabled, are introduced in
nid_path struct for marking the two individual power states: the pin
plug/unplug and DAC/ADC stream, respectively. They can be set
statically in case they are static routes (e.g. some mixer paths),
too.
The power up and down events for each pin are triggered via the
standard hda_jack table. The call order is hard-coded, relying on the
current implementation of jack event chain (a la FILO/stack order).
One point to be dealt carefully is that DAC/ADC cannot be powered
on/off while streaming. They are pinned as long as the stream is
running. For controlling the power of DAC/ADC, a new patch_ops is
added. The generic parser provides the default callback for that.
As of this patch, only IDT/Sigmatel codec driver enables the flag.
The support on other codecs will follow.
An assumption we made in this code is that the widget state (e.g. amp,
pinctl, connections) remains after the widget power transition (not
about FG power transition). This is true for IDT codecs, at least.
But if the widget state is lost at widget power transition, we'd need
to implement additional code to sync the cached amp/verbs for the
specific NID.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The driver shutdown ops is simpler than registering reboot notifier
manually. There should be no functional change by this -- the codec
driver calls its own callback while the bus driver just calls
azx_stop() like before.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Move the destructor code to device release callback for the codec
object instead. This is a safer place to release the resources than
dev_free callback in general.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The events that are handled by HD-audio drivers are no frequent and
urgent ones, so we can use the standard workqueue without any problem
nowadays.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This makes the code flow easier -- instead of the controller driver
calling snd_hda_build_pcms() and snd_hda_build_controls() explicitly,
the codec driver itself builds PCMs and controls at probe time. Then
the controller driver only needs to call snd_card_register().
Also, this allows us the full bind/unbind control, too. Even when a
codec driver is bound later, it automatically registers the new PCM
and controls by itself.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Now we have all pieces ready, and put them into places:
- add the hda_pcm refcount to azx_pcm_open() and azx_pcm_close(),
- call the most of cleanup code in hda_codec_reset() from the codec
driver remove,
- call the same code also from the hda_codec object free.
Then the codec driver can be unbound more safely now.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The PCM ops might be set NULL, or cleared to NULL when the driver is
unbound. Give a proper NULL check at each place to be more robust.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
So far, the hda_codec object kept the hda_pcm list in an array, and
the codec driver was expected to assign the array. However, this
makes the object life cycle management harder, because the assigned
array is freed at the codec driver detach while it might be still
accessed by the opened streams.
In this patch, we allocate each hda_pcm object dynamically and manage
it as a linked list. Each object has a kref refcount, and both the
codec driver binder and the PCM open/close touches it, so that the
object won't be freed while in use.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The memory allocators should have already given the kernel warning
messages, thus we don't have to annoy again.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Allow the codec object to have an individual card pointer. Not only
this simplifies the redirections in many places, also this will allow
us to make each codec assigned to a different card object.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_hda_build_pcms() does actually three things: let the codec driver
build up hda_pcm list, set the PCM default values, and call the
attach_pcm bus ops for each hda_pcm instance. The former two are
basically independent from the bus implementation, so it'd make the
code a bit more readable.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
So far, we let the controller driver power down the all codecs at the
end of probe. But this can be done better in the codec's dev_register
callback. This results in the reduction of duplicated codes in each
control driver.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Now the final bit of runtime PM cleanup: instead of manual
notification of the power up/down of the codec via hda_bus pm_notify
ops, use the standard runtime PM feature.
The child codec device will kick off the runtime PM of the parent
(PCI) device upon suspend/resume automatically. For managing whether
the link can be really turned off, we use the bit flags
bus->codec_powered instead of the earlier bus->power_keep_link_on.
flag. Each codec driver is responsible to set/clear the bit flag, and
the controller device can be turned off only when all these bits are
cleared.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We used to pass the power_save option value to hda_bus via a given
pointer. This was needed to refer to the value from the HD-audio core
side. However, after the transition to the runtime PM, this is no
longer needed.
This patch drops the power_save value indirection in hda_bus above,
and let the controller driver reprograms the autosuspend value
explicitly by a new helper, snd_hda_set_power_save(). Without this
call, the HD-audio core doesn't set up the autosuspend and flip the
runtime PM. (User may still be able to set up via sysfs, though.)
Along with this change, the pointer argument of azx_bus_create() is
dropped as well.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Like the previous transition of suspend/resume, now move the
power-save code to the standard runtime PM. As usual for runtime PM,
it's a bit tricky, but this simplified codes a lot in the end.
For keeping the usage compatibility, power_save module option still
controls the whole power-saving behavior on all codecs. The value is
translated to pm_runtime_*_autosuspend() and pm_runtime_allow() /
pm_runtime_forbid() calls.
snd_hda_power_up() and snd_hda_power_down() are translated to
pm_runtime_get_sync() and pm_runtime_put_autosuspend(), respectively.
Since we can do call pm_runtime_get_sync() more reliably, the sync
version is used always and snd_hda_power_up_d3wait() is dropped.
Another slight difference is that snd_hda_power_up()/down() don't call
runtime_pm code during the suspend/resume transition phase. Calling
them there isn't safe unlike our own code, resulted in unexpected
behavior (endless wakeups).
The hda_power_count tracepoint was removed, as it doesn't match well
with the new code.
Last but not least, we need to set ignore_children flag in the parent
dev.power field so that the runtime PM of the controller chip won't
get confused. The notification is still done in the bus pm_notify
callback. We'll get rid of this hack in the later patch.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This patch moves the suspend/resume mechanisms down to each codec
driver level, as we have a proper codec driver bound on the bus now.
Then we get the asynchronous PM gratis without fiddling much in the
driver level.
As a soft-landing transition, implement the common suspend/resume pm
ops for hda_codec_driver and keep the each codec driver intact. Only
the callers of suspend/resume in the controller side (azx_suspend()
and azx_resume()) are removed.
Another involved place is azx_bus_reset() calling the temporary
suspend and resume as a hackish method of bus reset. The HD-audio
core provide a helper function snd_hda_bus_reset() instead.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Now we create the standard HD-audio bus (/sys/bus/hdaudio), and bind
the codec driver with the codec device over there. This is the first
step of the whole transition so that the changes to each codec driver
are kept as minimal as possible.
Each codec driver needs to register hda_codec_driver struct containing
the currently existing preset via the new helper macro
module_hda_codec_driver(). The old hda_codec_preset_list is replaced
with this infrastructure. The generic parsers (for HDMI and other)
are also included in the preset with the special IDs to bind
uniquely.
In HD-audio core side, the device binding code is split to
hda_bind.c. It provides the snd_hda_bus_type implementation to match
the codec driver with the given codec vendor ID. It also manages the
module auto-loading by itself like before: when the matching isn't
found, it tries to probe the corresponding codec modules, and finally
falls back to the generic drivers. (The special ID mentioned above is
set at this stage.)
The only visible change to outside is that the hdaudio sysfs entry now
appears in /sys/bus/devices, not as a sound class device.
More works to move the suspend/resume and remove ops will be
(hopefully) done in later patches.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When the mixer amp is touched by control elements, we don't have to
power up always; if the codec was suspended at the time, we can just
update the amp cache and it's reflected to the hardware upon resume.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There is no big merit to handle hda_bus_unsolicited object
individually, as it's tightly coupled with the hda_bus object itself.
Embedding it makes the code simpler in the end.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Instead of copying from the given template, let the caller fills the
fields after creation. This simplifies the code after all.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In olden times the snd_hda_param_read() function always set "*start_id"
but in 2007 we introduced a new return and it causes uninitialized data
bugs in a couple of the callers: print_codec_info() and
hdmi_parse_codec().
Fixes: e8a7f136f5 ('[ALSA] hda-intel - Improve HD-audio codec probing robustness')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The functions kfree() and release_firmware() test whether their argument
is NULL and then return immediately. Thus the test around the call
is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Put more kerneldoc comments to the exported functions.
Still the generic parser code and the HD-audio controller code aren't
covered yet, though.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
request_module() handles the printf style arguments, so we don't have
to render strings in the caller side. Not only it reduces the
unnecessary temporary string buffer, it's even safer from the security
POV.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The next patch will use it, so make it visible across modules.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The helper functions snd_hda_check_board_config() and
snd_hda_check_board_codec_sid_config() are no longer used since the
transition to the generic parser and all quirks have been replaced
with fixups. Let's kill these dead codes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Previously, calling one quirk function first and another later
would make the latter one take priority, this is now changed
to make the former take priority.
By adding two special values for fixup_id we can also get rid of the
fixup_forced flag.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The recent commit [6194b99d: ALSA: hda - Kill the rest of snd_print*()
usages] changed the callback map_slaves(), but one call was forgotten
to be replaced due to the cast, which leads to kernel Oops due to
invalid function. This patch replaces it with a proper function.
Fixes: 6194b99de9 ('ALSA: hda - Kill the rest of snd_print*() usages')
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Pass the codec object so that we can replace all the rest of
snd_print*() usages with the proper device-specific print helpers.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The beep input device is registered via input_register_device(), but
this is called in snd_hda_attach_beep_device() where the sound devices
aren't registered yet. This leads to the binding to non-existing
object, thus results in failure. And, even if the binding worked
(against the PCI object), it's still racy; the input device appears
before the sound objects.
For fixing this, register the input device properly at dev_register
ops of the codec object it's bound with. Also, call
snd_hda_detach_beep_device() at dev_disconnection so that it's
detached at the right timing. As a bonus, since it's called in the
codec's ops, we can get rid of the further call from the other codec
drivers.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Use dev_err() and co for messages from HD-audio controller and codec
drivers. The codec drivers are mostly bound with codec objects, so
some helper macros, codec_err(), codec_info(), etc, are provided.
They merely wrap the corresponding dev_xxx().
There are a few places still calling snd_printk() and its variants
as they are called without the codec or device context.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We have currently sysfs attributes for each hwdep, but basically these
should belong to the codec itself, per se. Let's add them to the
codec object while keeping them for hwdep as is for compatibility.
While we are at it, split the sysfs-related stuff into a separate
source file, hda_sysfs.c, and keep only the stuff necessary for hwdep
in hda_hwdep.c.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
As the HD-audio is treated individually in each codec driver, it's
more convenient to assign an own struct device to each codec object.
Then we'll be able to use dev_err() more easily for each codec, for
example.
For achieving it, this patch just creates an object "hdaudioCxDy".
It belongs to sound class instead of creating a new bus, just for
simplicity, at this stage. No pm ops is implemented in the device
struct level but currently it's merely a container. The PCM and hwdep
devices are now children of this codec device.
Signed-off-by: Takashi Iwai <tiwai@suse.de>