soundwire: bus: add bpt_stream pointer

Add a convenience pointer to the 'sdw_bus' structure. BPT is a
dedicated stream which will typically not be handled by DAIs or
dailinks. Since there's only one

soundwire: bus: add bpt_stream pointer

Add a convenience pointer to the 'sdw_bus' structure. BPT is a
dedicated stream which will typically not be handled by DAIs or
dailinks. Since there's only one BPT stream per link, storing the
pointer at the link level seems rather natural.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Reviewed-by: Liam Girdwood <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Tested-by: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: bus: add send_async/wait APIs for BPT protocol

Add definitions and helpers for the BPT/BRA protocol. Peripheral
drivers (aka ASoC codec drivers) can use this API to send bulk data
such as

soundwire: bus: add send_async/wait APIs for BPT protocol

Add definitions and helpers for the BPT/BRA protocol. Peripheral
drivers (aka ASoC codec drivers) can use this API to send bulk data
such as firmware or tables. The design intent is however NOT to
directly use this API but to rely on an intermediate regmap layer.

The API is only available when no other audio streams have been
allocated, and only one BTP/BRA stream is allowed per link. To avoid
the addition of yet another lock, the refcount tests are handled in
the stream master_runtime alloc/free routines where the bus_lock is
already held. Another benefit of this approach is that the same
bus_lock is used to handle runtime and port linked lists, which
reduces the potential for misaligned configurations.

In addition to exclusion with audio streams, BPT transfers have a lot
of overhead, specifically registers writes are needed to enable
transport in DP0. Most DMAs don't handle too well very small data sets
and they may have alignment limitations.

The size and alignment requirements are for now not handled by the
core but must be checked by platform-specific drivers.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Reviewed-by: Liam Girdwood <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Tested-by: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: stream: extend sdw_alloc_stream() to take 'type' parameter

In the existing definition of sdw_stream_runtime, the 'type' member is
never set and defaults to PCM. To prepare for the BPT/BRA

soundwire: stream: extend sdw_alloc_stream() to take 'type' parameter

In the existing definition of sdw_stream_runtime, the 'type' member is
never set and defaults to PCM. To prepare for the BPT/BRA support, we
need to special-case streams and make use of the 'type'.

No functional change for now, the implicit PCM type is now explicit.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Reviewed-by: Liam Girdwood <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Tested-by: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: extend sdw_stream_type to BPT

BPT/BRA need to be special cased, i.e. there's no point in using the
bandwidth allocation since the entire frame can be used.

Signed-off-by: Pierre-Louis Bo

soundwire: extend sdw_stream_type to BPT

BPT/BRA need to be special cased, i.e. there's no point in using the
bandwidth allocation since the entire frame can be used.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Reviewed-by: Liam Girdwood <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Tested-by: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

SoundWire: pass stream to compute_params()

The stream parameter will be used in the follow up commit.
No function change.

Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Ran

SoundWire: pass stream to compute_params()

The stream parameter will be used in the follow up commit.
No function change.

Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

Soundwire: stream: program BUSCLOCK_SCALE

We need to program bus clock scale to adjust the bus clock if current
bus clock doesn't fit the bandwidth.

Signed-off-by: Bard Liao <yung-chuan.liao@linux.

Soundwire: stream: program BUSCLOCK_SCALE

We need to program bus clock scale to adjust the bus clock if current
bus clock doesn't fit the bandwidth.

Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

Soundwire: add sdw_slave_get_scale_index helper

Currently, we only set peripheral frequency when the peripheral is
initialized. However, curr_dr_freq may change to get required bandwidth.
For exampl

Soundwire: add sdw_slave_get_scale_index helper

Currently, we only set peripheral frequency when the peripheral is
initialized. However, curr_dr_freq may change to get required bandwidth.
For example, curr_dr_freq may increase from 4.8MHz to 9.6MHz when the
4th stream is opened. Add a helper to get the scale index so that we can
get the scale index and program it.

Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Ranjani Sridharan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: add lane_used_bandwidth in struct sdw_bus

To support multi-lane, we need to know how much bandwidth
is used on each lane. And to use the lane that has enough
bandwidth.

Signed-off-by: Ba

soundwire: add lane_used_bandwidth in struct sdw_bus

To support multi-lane, we need to know how much bandwidth
is used on each lane. And to use the lane that has enough
bandwidth.

Signed-off-by: Bard Liao <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Reviewed-by: Pierre-Louis Bossart <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: mipi_disco: read lane mapping properties from ACPI

The DisCo for SoundWire 2.0 added support for the
'mipi-sdw-lane-<n>-mapping' property.

Co-developed-by: Chao Song <[email protected]

soundwire: mipi_disco: read lane mapping properties from ACPI

The DisCo for SoundWire 2.0 added support for the
'mipi-sdw-lane-<n>-mapping' property.

Co-developed-by: Chao Song <[email protected]>
Signed-off-by: Chao Song <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: Minor formatting fixups in sdw.h header

Fixup some minor formatting and whitespace in the sdw.h header file.

Signed-off-by: Charles Keepax <[email protected]>
Link: https://l

soundwire: Minor formatting fixups in sdw.h header

Fixup some minor formatting and whitespace in the sdw.h header file.

Signed-off-by: Charles Keepax <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: Update the includes on the sdw.h header

There are quite a few things used in the sdw.h header that it relies on
the consumer to include. If something is used directly in the header it
sho

soundwire: Update the includes on the sdw.h header

There are quite a few things used in the sdw.h header that it relies on
the consumer to include. If something is used directly in the header it
should be included by the header. Update the includes to cover the
missing items, or add forward declarations for things that are only used
as pointers. Whilst making the change also alphabetise the list of
includes.

Signed-off-by: Charles Keepax <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

ASoC: SDCA: add initial module

Add new module for SDCA (SoundWire Device Class for Audio) support.
For now just add a parser to identify the SDCA revision and the
function mask.

Note that the SDCA

ASoC: SDCA: add initial module

Add new module for SDCA (SoundWire Device Class for Audio) support.
For now just add a parser to identify the SDCA revision and the
function mask.

Note that the SDCA definitions and related MIPI DisCo properties are
defined only for ACPI platforms and extracted with _DSD helpers. There
is currently no support for Device Tree in the specification, the
'depends on ACPI' reflects this design limitation. This might change
in a future revision of the specification but for SDCA 1.0 ACPI is the
only supported type of platform firmware.

The SDCA library is defined with static inline fallbacks, which will
allow for unconditional addition of SDCA support in common parts of
the code.

The design follows a four-step process:

1) Basic information related to Functions is extracted from MIPI DisCo
tables and stored in the 'struct sdw_slave'. Devm_ based memory
allocation is not allowed at this point prior to a driver probe, so we only
store the function node, address and type.

2) When a codec driver probes, it will register subdevices for each
Function identified in phase 1)

3) a driver will probe for each subdevice and addition parsing/memory
allocation takes place at this level. devm_ based allocation is highly
encouraged to make error handling manageable.

4) Before the peripheral device becomes physically attached, register
access is not permitted and the regmaps are cache-only. When
peripheral device is enumerated, the bus level uses the
'update_status' notification; after optional device-level
initialization, the codec driver will notify each of the subdevices so
that they can start interacting with the hardware.

Note that the context extracted in 1) should be arguably be handled
completely in the codec driver probe. That would however make it
difficult to use the ACPI information for machine quirks, and
e.g. select different machine driver and topologies as done for the
RT712_VB handling later in the series. To make the implementation of
quirks simpler, this patchset extracts a minimal amount of context
(interface revision and number/type of Functions) before the codec
driver probe, and stores this context in the scope of the 'struct
sdw_slave'.

The SDCA library can also be used in a vendor-specific driver without
creating subdevices, e.g. to retrieve the 'initialization-table'
values to write platform-specific values as needed.

For more technical details, the SDCA specification is available for
public downloads at https://www.mipi.org/mipi-sdca-v1-0-download

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Mark Brown <[email protected]>

show more ...

ASoC/soundwire: remove sdw_slave_extended_id

This structure is used to copy information from the 'sdw_slave'
structures, it's better to create a flexible array of 'sdw_slave'
pointers and directly a

ASoC/soundwire: remove sdw_slave_extended_id

This structure is used to copy information from the 'sdw_slave'
structures, it's better to create a flexible array of 'sdw_slave'
pointers and directly access the information. This will also help
access additional information stored in the 'sdw_slave' structure,
such as an SDCA context.

This patch does not add new functionality, it only modified how the
information is retrieved.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Reviewed-by: Péter Ujfalusi <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Mark Brown <[email protected]>

show more ...

soundwire: mipi-disco: add support for DP0/DPn 'lane-list' property

The SoundWire specification did not clearly require that ports could
use all Lanes. Some SoundWire/SDCA peripheral adopters added

soundwire: mipi-disco: add support for DP0/DPn 'lane-list' property

The SoundWire specification did not clearly require that ports could
use all Lanes. Some SoundWire/SDCA peripheral adopters added
restrictions on which lanes can be used by what port, and the DisCo
for SoundWire 2.1 specification added a 'lane-list' property to model
this hardware limitation.

When not specified, the ports can use all Lanes. Otherwise, the
'lane-list' indicates which Lanes can be used, sorted by order of
preference (most-preferred-first).

This patch only reads the properties, the use of this property will
come at a later time with multi-lane support.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: mipi-disco: add new properties from 2.0 spec

The DisCo for SoundWire 2.0 spec adds support for new
'mipi-sdw-sdca-interrupt-register-list' and
'mipi-sdw-commit-register-supported'.

This

soundwire: mipi-disco: add new properties from 2.0 spec

The DisCo for SoundWire 2.0 spec adds support for new
'mipi-sdw-sdca-interrupt-register-list' and
'mipi-sdw-commit-register-supported'.

This patch only adds the definitions and property reads, but the use
of these properties will come at some point in the future when needed.

Note a slight conceptual disconnect between the MIPI DisCo definition
of a boolean property and the Linux implementation. The latter only
checks the presence of the property to set its value to 'true',
whereas the MIPI definitions allow for a property with a 'false'
value. This patch uses the new introduced mipi_device_property_read_bool()
to handle it.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: mipi-disco: remove DPn audio-modes

The concept of DPn audio-modes was never used by anyone, and was
removed from the DisCo for SoundWire 2.0 specification.

Remove the definitions and TOD

soundwire: mipi-disco: remove DPn audio-modes

The concept of DPn audio-modes was never used by anyone, and was
removed from the DisCo for SoundWire 2.0 specification.

Remove the definitions and TODO.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_dpn_prop

before:
struct sdw_dpn_prop {
u32 num; /* 0 4 */
u32 max_word; /* 4 4 */

soundwire: optimize sdw_dpn_prop

before:
struct sdw_dpn_prop {
u32 num; /* 0 4 */
u32 max_word; /* 4 4 */
u32 min_word; /* 8 4 */
u32 num_words; /* 12 4 */
u32 * words; /* 16 8 */
enum sdw_dpn_type type; /* 24 4 */
u32 max_grouping; /* 28 4 */
bool simple_ch_prep_sm; /* 32 1 */

/* XXX 3 bytes hole, try to pack */

u32 ch_prep_timeout; /* 36 4 */
u32 imp_def_interrupts; /* 40 4 */
u32 max_ch; /* 44 4 */
u32 min_ch; /* 48 4 */
u32 num_channels; /* 52 4 */
u32 * channels; /* 56 8 */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 num_ch_combinations; /* 64 4 */

/* XXX 4 bytes hole, try to pack */

u32 * ch_combinations; /* 72 8 */
u32 modes; /* 80 4 */
u32 max_async_buffer; /* 84 4 */
bool block_pack_mode; /* 88 1 */
bool read_only_wordlength; /* 89 1 */

/* XXX 2 bytes hole, try to pack */

u32 port_encoding; /* 92 4 */
struct sdw_dpn_audio_mode * audio_modes; /* 96 8 */

/* size: 104, cachelines: 2, members: 22 */
/* sum members: 95, holes: 3, sum holes: 9 */
/* last cacheline: 40 bytes */
};

after:

struct sdw_dpn_prop {
struct sdw_dpn_audio_mode * audio_modes; /* 0 8 */
u32 num; /* 8 4 */
u32 max_word; /* 12 4 */
u32 min_word; /* 16 4 */
u32 num_words; /* 20 4 */
u32 * words; /* 24 8 */
enum sdw_dpn_type type; /* 32 4 */
u32 max_grouping; /* 36 4 */
u32 ch_prep_timeout; /* 40 4 */
u32 imp_def_interrupts; /* 44 4 */
u32 max_ch; /* 48 4 */
u32 min_ch; /* 52 4 */
u32 num_channels; /* 56 4 */
u32 num_ch_combinations; /* 60 4 */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 * channels; /* 64 8 */
u32 * ch_combinations; /* 72 8 */
u32 modes; /* 80 4 */
u32 max_async_buffer; /* 84 4 */
u32 port_encoding; /* 88 4 */
bool block_pack_mode; /* 92 1 */
bool read_only_wordlength; /* 93 1 */
bool simple_ch_prep_sm; /* 94 1 */

/* size: 96, cachelines: 2, members: 22 */
/* padding: 1 */
/* last cacheline: 32 bytes */
};

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_dp0_prop

Move pointers and booleans.

Before:

struct sdw_dp0_prop {
u32 max_word; /* 0 4 */
u32 min_word;

soundwire: optimize sdw_dp0_prop

Move pointers and booleans.

Before:

struct sdw_dp0_prop {
u32 max_word; /* 0 4 */
u32 min_word; /* 4 4 */
u32 num_words; /* 8 4 */

/* XXX 4 bytes hole, try to pack */

u32 * words; /* 16 8 */
bool BRA_flow_controlled; /* 24 1 */
bool simple_ch_prep_sm; /* 25 1 */

/* XXX 2 bytes hole, try to pack */

u32 ch_prep_timeout; /* 28 4 */
bool imp_def_interrupts; /* 32 1 */

/* size: 40, cachelines: 1, members: 8 */
/* sum members: 27, holes: 2, sum holes: 6 */
/* padding: 7 */
/* last cacheline: 40 bytes */
};

after:

struct sdw_dp0_prop {
u32 * words; /* 0 8 */
u32 max_word; /* 8 4 */
u32 min_word; /* 12 4 */
u32 num_words; /* 16 4 */
u32 ch_prep_timeout; /* 20 4 */
bool BRA_flow_controlled; /* 24 1 */
bool simple_ch_prep_sm; /* 25 1 */
bool imp_def_interrupts; /* 26 1 */

/* size: 32, cachelines: 1, members: 8 */
/* padding: 5 */
/* last cacheline: 32 bytes */
};

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_slave_prop

move pointers first, and move booleans together.

before:

struct sdw_slave_prop {
u32 mipi_revision; /* 0 4 */
bool

soundwire: optimize sdw_slave_prop

move pointers first, and move booleans together.

before:

struct sdw_slave_prop {
u32 mipi_revision; /* 0 4 */
bool wake_capable; /* 4 1 */
bool test_mode_capable; /* 5 1 */
bool clk_stop_mode1; /* 6 1 */
bool simple_clk_stop_capable; /* 7 1 */
u32 clk_stop_timeout; /* 8 4 */
u32 ch_prep_timeout; /* 12 4 */
enum sdw_clk_stop_reset_behave reset_behave; /* 16 4 */
bool high_PHY_capable; /* 20 1 */
bool paging_support; /* 21 1 */
bool bank_delay_support; /* 22 1 */

/* XXX 1 byte hole, try to pack */

enum sdw_p15_behave p15_behave; /* 24 4 */
bool lane_control_support; /* 28 1 */

/* XXX 3 bytes hole, try to pack */

u32 master_count; /* 32 4 */
u32 source_ports; /* 36 4 */
u32 sink_ports; /* 40 4 */

/* XXX 4 bytes hole, try to pack */

struct sdw_dp0_prop * dp0_prop; /* 48 8 */
struct sdw_dpn_prop * src_dpn_prop; /* 56 8 */
/* --- cacheline 1 boundary (64 bytes) --- */
struct sdw_dpn_prop * sink_dpn_prop; /* 64 8 */
u8 scp_int1_mask; /* 72 1 */

/* XXX 3 bytes hole, try to pack */

u32 quirks; /* 76 4 */
bool clock_reg_supported; /* 80 1 */
bool use_domain_irq; /* 81 1 */

/* size: 88, cachelines: 2, members: 23 */
/* sum members: 71, holes: 4, sum holes: 11 */
/* padding: 6 */
/* last cacheline: 24 bytes */
};

after:

truct sdw_slave_prop {
struct sdw_dp0_prop * dp0_prop; /* 0 8 */
struct sdw_dpn_prop * src_dpn_prop; /* 8 8 */
struct sdw_dpn_prop * sink_dpn_prop; /* 16 8 */
u32 mipi_revision; /* 24 4 */
bool wake_capable; /* 28 1 */
bool test_mode_capable; /* 29 1 */
bool clk_stop_mode1; /* 30 1 */
bool simple_clk_stop_capable; /* 31 1 */
u32 clk_stop_timeout; /* 32 4 */
u32 ch_prep_timeout; /* 36 4 */
enum sdw_clk_stop_reset_behave reset_behave; /* 40 4 */
bool high_PHY_capable; /* 44 1 */
bool paging_support; /* 45 1 */
bool bank_delay_support; /* 46 1 */
bool lane_control_support; /* 47 1 */
enum sdw_p15_behave p15_behave; /* 48 4 */
u32 master_count; /* 52 4 */
u32 source_ports; /* 56 4 */
u32 sink_ports; /* 60 4 */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 quirks; /* 64 4 */
u8 scp_int1_mask; /* 68 1 */
bool clock_reg_supported; /* 69 1 */
bool use_domain_irq; /* 70 1 */

/* size: 72, cachelines: 2, members: 23 */
/* padding: 1 */
/* last cacheline: 8 bytes */
};

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_bus structure

The sdw_bus structure has seen multiple additions over the years. It's
one of the most used structures in this subsystem, so there's merit in
reshuffling the me

soundwire: optimize sdw_bus structure

The sdw_bus structure has seen multiple additions over the years. It's
one of the most used structures in this subsystem, so there's merit in
reshuffling the members a bit with 'pahole' to reduce holes and
structures across cache lines.

before:

struct sdw_bus {
struct device * dev; /* 0 8 */
struct sdw_master_device * md; /* 8 8 */
int controller_id; /* 16 4 */
unsigned int link_id; /* 20 4 */
int id; /* 24 4 */

/* XXX 4 bytes hole, try to pack */

struct list_head slaves; /* 32 16 */
long unsigned int assigned[1]; /* 48 8 */
struct mutex bus_lock; /* 56 160 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct lock_class_key bus_lock_key; /* 216 16 */
struct mutex msg_lock; /* 232 160 */
/* --- cacheline 6 boundary (384 bytes) was 8 bytes ago --- */
struct lock_class_key msg_lock_key; /* 392 16 */
int (*compute_params)(struct sdw_bus *); /* 408 8 */
const struct sdw_master_ops * ops; /* 416 8 */
const struct sdw_master_port_ops * port_ops; /* 424 8 */
struct sdw_bus_params params; /* 432 36 */

/* XXX 4 bytes hole, try to pack */

/* --- cacheline 7 boundary (448 bytes) was 24 bytes ago --- */
struct sdw_master_prop prop; /* 472 72 */

/* XXX last struct has 6 bytes of padding */

/* --- cacheline 8 boundary (512 bytes) was 32 bytes ago --- */
void * vendor_specific_prop; /* 544 8 */
struct list_head m_rt_list; /* 552 16 */
struct dentry * debugfs; /* 568 8 */
/* --- cacheline 9 boundary (576 bytes) --- */
struct irq_chip irq_chip; /* 576 264 */
/* --- cacheline 13 boundary (832 bytes) was 8 bytes ago --- */
struct irq_domain * domain; /* 840 8 */
struct sdw_defer defer_msg; /* 848 112 */
/* --- cacheline 15 boundary (960 bytes) --- */
unsigned int clk_stop_timeout; /* 960 4 */
u32 bank_switch_timeout; /* 964 4 */
bool multi_link; /* 968 1 */

/* XXX 3 bytes hole, try to pack */

int hw_sync_min_links; /* 972 4 */
int stream_refcount; /* 976 4 */

/* size: 984, cachelines: 16, members: 27 */
/* sum members: 969, holes: 3, sum holes: 11 */
/* padding: 4 */
/* paddings: 1, sum paddings: 6 */
/* last cacheline: 24 bytes */
};

after:

struct sdw_bus {
struct device * dev; /* 0 8 */
struct sdw_master_device * md; /* 8 8 */
struct lock_class_key bus_lock_key; /* 16 16 */
struct mutex bus_lock; /* 32 160 */
/* --- cacheline 3 boundary (192 bytes) --- */
struct list_head slaves; /* 192 16 */
struct lock_class_key msg_lock_key; /* 208 16 */
struct mutex msg_lock; /* 224 160 */
/* --- cacheline 6 boundary (384 bytes) --- */
struct list_head m_rt_list; /* 384 16 */
struct sdw_defer defer_msg; /* 400 112 */
/* --- cacheline 8 boundary (512 bytes) --- */
struct sdw_bus_params params; /* 512 36 */
int stream_refcount; /* 548 4 */
const struct sdw_master_ops * ops; /* 552 8 */
const struct sdw_master_port_ops * port_ops; /* 560 8 */
struct sdw_master_prop prop; /* 568 72 */

/* XXX last struct has 6 bytes of padding */

/* --- cacheline 10 boundary (640 bytes) --- */
void * vendor_specific_prop; /* 640 8 */
int hw_sync_min_links; /* 648 4 */
int controller_id; /* 652 4 */
unsigned int link_id; /* 656 4 */
int id; /* 660 4 */
int (*compute_params)(struct sdw_bus *); /* 664 8 */
long unsigned int assigned[1]; /* 672 8 */
unsigned int clk_stop_timeout; /* 680 4 */
u32 bank_switch_timeout; /* 684 4 */
struct irq_chip irq_chip; /* 688 264 */
/* --- cacheline 14 boundary (896 bytes) was 56 bytes ago --- */
struct irq_domain * domain; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct dentry * debugfs; /* 960 8 */
bool multi_link; /* 968 1 */

/* size: 976, cachelines: 16, members: 27 */
/* padding: 7 */
/* paddings: 1, sum paddings: 6 */
/* last cacheline: 16 bytes */
};

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_master_prop

Make pahole happy by moving pointers and u64 first instead of
interleaving them.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signe

soundwire: optimize sdw_master_prop

Make pahole happy by moving pointers and u64 first instead of
interleaving them.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: optimize sdw_stream_runtime memory layout

pahole suggestion: swap position of 'm_rt_count'

before: pahole -C sdw_stream_runtime drivers/soundwire/soundwire-bus.ko
struct sdw_stream_runti

soundwire: optimize sdw_stream_runtime memory layout

pahole suggestion: swap position of 'm_rt_count'

before: pahole -C sdw_stream_runtime drivers/soundwire/soundwire-bus.ko
struct sdw_stream_runtime {
const char * name; /* 0 8 */
struct sdw_stream_params params; /* 8 12 */
enum sdw_stream_state state; /* 20 4 */
enum sdw_stream_type type; /* 24 4 */

/* XXX 4 bytes hole, try to pack */

struct list_head master_list; /* 32 16 */
int m_rt_count; /* 48 4 */

/* size: 56, cachelines: 1, members: 6 */
/* sum members: 48, holes: 1, sum holes: 4 */
/* padding: 4 */
/* last cacheline: 56 bytes */
};

after: pahole --reorganize -C sdw_stream_runtime drivers/soundwire/soundwire-bus.ko
struct sdw_stream_runtime {
const char * name; /* 0 8 */
struct sdw_stream_params params; /* 8 12 */
enum sdw_stream_state state; /* 20 4 */
enum sdw_stream_type type; /* 24 4 */
int m_rt_count; /* 28 4 */
struct list_head master_list; /* 32 16 */

/* size: 48, cachelines: 1, members: 6 */
/* last cacheline: 48 bytes */
}; /* saved 8 bytes! */

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: bus: drop unused driver name field

The soundwire driver name field is not currently used by any driver (and
even appears to never have been used) so drop it.

Signed-off-by: Johan Hovold

soundwire: bus: drop unused driver name field

The soundwire driver name field is not currently used by any driver (and
even appears to never have been used) so drop it.

Signed-off-by: Johan Hovold <[email protected]>
Reviewed-by: Pierre-Louis Bossart <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: bus: add stream refcount

The notion of stream is by construction based on a multi-bus
capability, to allow for aggregation of Peripheral devices or
functions located on different segments

soundwire: bus: add stream refcount

The notion of stream is by construction based on a multi-bus
capability, to allow for aggregation of Peripheral devices or
functions located on different segments. We currently count how many
master_rt contexts are used by a stream, but we don't have the dual
refcount of how many streams are allocated on a given bus. This
refcount will be useful to check if BTP/BRA streams can be allocated.

Note that the stream_refcount is modified in sdw_master_rt_alloc() and
sdw_master_rt_free() which are both called with the bus_lock mutex
held, so there's no need for refcount_ primitives for additional
protection.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Reviewed-by: Rander Wang <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...

soundwire: intel_ace2x: use DOAIS and DODS settings from firmware

Starting with LNL, the recommendation is to use settings read from DSD
properties instead of hard-coding the values.

The DOAIS and

soundwire: intel_ace2x: use DOAIS and DODS settings from firmware

Starting with LNL, the recommendation is to use settings read from DSD
properties instead of hard-coding the values.

The DOAIS and DODS values are completely-specific to Intel and are
stored in a vendor-specific property structure.

Signed-off-by: Pierre-Louis Bossart <[email protected]>
Signed-off-by: Bard Liao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>

show more ...