When a Slave device becomes synchronized with the bus, it may report its presence in PING frames, as well as optionally asserting an in-band PREQ signal.

The bus driver will detect a new Device0, start the enumeration process and assign it a non-zero device number. The SoundWire enumeration provides an arbitration to deal with multiple Slaves reporting ATTACHED at the same time. The bus driver will also invoke the driver .probe() callback associated with this device. The probe() depends on the Linux device core, which handles the match operations and may result in modules being loaded.

Once the non-zero device number is programmed, the Slave will report its new status in PING frames and the Master hardware will typically report this status change with an interrupt. At this point, the .update_status() callback of the codec driver will be invoked (usually from an interrupt thread or workqueue scheduled from the interrupt thread).

The first race condition which can happen is between the .probe(), which allocates the resources, and .update_status() where initializations are typically handled. The .probe() is only called once during the initial boot, while .update_status() will be called for every bus hardware reset and if the Slave device loses synchronization (an unlikely event but with non-zero probability).

The time difference between the end of the enumeration process and a change of status reported by the hardware may be as small as one SoundWire PING frame. The scheduling of the interrupt thread, which invokes .update_status() is not deterministic, but can be small enough to create a race condition. With a 48 kHz frame rate and ideal scheduling cases, the .probe() may be pre-empted within double-digit microseconds.

Since there is no guarantee that the .probe() completes by the time .update_status() is invoked as a result of an interrupt, it's not unusual for the .update_status() to rely on data structures that have not been allocated yet, leading to kernel oopses.

This patch adds a probe_complete utility, which is used in the sdw_update_slave_status() routine. The codec driver does not need to do anything and can safely assume all resources are allocated in its update_status() callback.

Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- include/linux/soundwire/sdw.h | 6 ++++++ 1 file changed, 6 insertions(+)

diff --git a/include/linux/soundwire/sdw.h b/include/linux/soundwire/sdw.h index 28745b9ba279..cb1db4a7475d 100644 --- a/include/linux/soundwire/sdw.h +++ b/include/linux/soundwire/sdw.h @@ -547,6 +547,10 @@ struct sdw_slave_ops { * @node: node for bus list * @port_ready: Port ready completion flag for each Slave port * @dev_num: Device Number assigned by Bus + * @probed: boolean tracking driver state + * @probe_complete: completion utility to control potential races + * on startup between driver probe/initialization and SoundWire + * Slave state changes/implementation-defined interrupts */ struct sdw_slave { struct sdw_slave_id id; @@ -561,6 +565,8 @@ struct sdw_slave { struct list_head node; struct completion *port_ready; u16 dev_num; + bool probed; + struct completion probe_complete; };

#define dev_to_sdw_dev(_dev) container_of(_dev, struct sdw_slave, dev)

Slave drivers may have different ways of handling their settings, with or without regmap.

During the integration of codec drivers, done in partnership between Intel and Realtek, it became desirable to implement a predictable order between low-level initializations performed in .update_status() (invoked by an interrupt thread) and the settings restored in the resume steps (invoked by the PM core).

This patch builds on the previous solution to wait for the Slave device to be fully enumerated. The complete() in this case is signaled not before the .update_status() is called, but after .update_status() returns. Without this patch, the settings were not properly restored, leading to timing-dependent 'no sound after resume' or 'no headset detected after resume' bug reports.

Depending on how initialization is handled, a Slave device driver may wait for enumeration_complete, or for initialization_complete, both are valid synchronization points. They are initialized at the same time, they only differ on when complete() is invoked.

Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- include/linux/soundwire/sdw.h | 3 +++ 1 file changed, 3 insertions(+)

diff --git a/include/linux/soundwire/sdw.h b/include/linux/soundwire/sdw.h index 3fa8d875b16b..ed42cd79eab7 100644 --- a/include/linux/soundwire/sdw.h +++ b/include/linux/soundwire/sdw.h @@ -554,6 +554,8 @@ struct sdw_slave_ops { * @enumeration_complete: completion utility to control potential races * on startup between device enumeration and read/write access to the * Slave device + * @initialization_complete: completion utility to control potential races + * on startup between device enumeration and settings being restored */ struct sdw_slave { struct sdw_slave_id id; @@ -571,6 +573,7 @@ struct sdw_slave { bool probed; struct completion probe_complete; struct completion enumeration_complete; + struct completion initialization_complete; };

#define dev_to_sdw_dev(_dev) container_of(_dev, struct sdw_slave, dev)

The current interfaces between ASoC and SoundWire are limited by the platform_device infrastructure to an init() and exit() (mapped to the platform driver.probe and .remove)

To help with the platform detection, machine driver selection and management of power dependencies between DSP and SoundWire IP, the ASoC side requires:

a) an ACPI scan helper, to report if any devices are exposed in the DSDT tables, and if any links are disabled by the BIOS.

b) a probe helper that allocates the resources without actually starting the bus.

c) a startup helper which does start the bus when all power dependencies are settled.

d) an exit helper to free all resources

e) an interrupt_enable/disable helper, typically invoked after the startup helper but also used in suspend routines.

This patch moves all required interfaces to sdw_intel.h, mainly to allow SoundWire and ASoC parts to be merged separately once the header files are shared between trees.

To avoid compilation issues, the conflicts in intel_init.c are blindly removed. This would in theory prevent the code from working, but since there are no users of the Intel Soundwire driver this has no impact. Functionality will be restored when the removal of platform devices is complete.

Support for SoundWire + SOF builds will only be provided once all the required pieces are upstream.

Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- drivers/soundwire/intel.h | 9 ++-- drivers/soundwire/intel_init.c | 31 +++--------- include/linux/soundwire/sdw_intel.h | 77 +++++++++++++++++++++++++++-- 3 files changed, 85 insertions(+), 32 deletions(-)

diff --git a/drivers/soundwire/intel.h b/drivers/soundwire/intel.h index d923b6262330..e4cc1d3804ff 100644 --- a/drivers/soundwire/intel.h +++ b/drivers/soundwire/intel.h @@ -5,17 +5,20 @@ #define __SDW_INTEL_LOCAL_H

/** - * struct sdw_intel_link_res - Soundwire link resources + * struct sdw_intel_link_res - Soundwire Intel link resource structure, + * typically populated by the controller driver. + * @pdev: platform_device + * @mmio_base: mmio base of SoundWire registers * @registers: Link IO registers base * @shim: Audio shim pointer * @alh: ALH (Audio Link Hub) pointer * @irq: Interrupt line * @ops: Shim callback ops * @arg: Shim callback ops argument - * - * This is set as pdata for each link instance. */ struct sdw_intel_link_res { + struct platform_device *pdev; + void __iomem *mmio_base; /* not strictly needed, useful for debug */ void __iomem *registers; void __iomem *shim; void __iomem *alh; diff --git a/drivers/soundwire/intel_init.c b/drivers/soundwire/intel_init.c index 2a2b4d8df462..bc739a38916d 100644 --- a/drivers/soundwire/intel_init.c +++ b/drivers/soundwire/intel_init.c @@ -27,19 +27,9 @@ static int link_mask; module_param_named(sdw_link_mask, link_mask, int, 0444); MODULE_PARM_DESC(sdw_link_mask, "Intel link mask (one bit per link)");

-struct sdw_link_data { - struct sdw_intel_link_res res; - struct platform_device *pdev; -}; - -struct sdw_intel_ctx { - int count; - struct sdw_link_data *links; -}; - static int sdw_intel_cleanup_pdev(struct sdw_intel_ctx *ctx) { - struct sdw_link_data *link = ctx->links; + struct sdw_intel_link_res *link = ctx->links; int i;

if (!link) @@ -62,7 +52,7 @@ static struct sdw_intel_ctx { struct platform_device_info pdevinfo; struct platform_device *pdev; - struct sdw_link_data *link; + struct sdw_intel_link_res *link; struct sdw_intel_ctx *ctx; struct acpi_device *adev; int ret, i; @@ -123,14 +113,12 @@ static struct sdw_intel_ctx continue; }

- link->res.irq = res->irq; - link->res.registers = res->mmio_base + SDW_LINK_BASE + link->registers = res->mmio_base + SDW_LINK_BASE + (SDW_LINK_SIZE * i); - link->res.shim = res->mmio_base + SDW_SHIM_BASE; - link->res.alh = res->mmio_base + SDW_ALH_BASE; + link->shim = res->mmio_base + SDW_SHIM_BASE; + link->alh = res->mmio_base + SDW_ALH_BASE;

- link->res.ops = res->ops; - link->res.arg = res->arg; + link->ops = res->ops;

memset(&pdevinfo, 0, sizeof(pdevinfo));

@@ -138,8 +126,6 @@ static struct sdw_intel_ctx pdevinfo.name = "int-sdw"; pdevinfo.id = i; pdevinfo.fwnode = acpi_fwnode_handle(adev); - pdevinfo.data = &link->res; - pdevinfo.size_data = sizeof(link->res);

pdev = platform_device_register_full(&pdevinfo); if (IS_ERR(pdev)) { @@ -216,7 +202,6 @@ void *sdw_intel_init(acpi_handle *parent_handle, struct sdw_intel_res *res)

return sdw_intel_add_controller(res); } -EXPORT_SYMBOL(sdw_intel_init);

/** * sdw_intel_exit() - SoundWire Intel exit @@ -224,10 +209,8 @@ EXPORT_SYMBOL(sdw_intel_init); * * Delete the controller instances created and cleanup */ -void sdw_intel_exit(void *arg) +void sdw_intel_exit(struct sdw_intel_ctx *ctx) { - struct sdw_intel_ctx *ctx = arg; - sdw_intel_cleanup_pdev(ctx); kfree(ctx); } diff --git a/include/linux/soundwire/sdw_intel.h b/include/linux/soundwire/sdw_intel.h index c9427cb6020b..034eca8df748 100644 --- a/include/linux/soundwire/sdw_intel.h +++ b/include/linux/soundwire/sdw_intel.h @@ -16,24 +16,91 @@ struct sdw_intel_ops { };

/** - * struct sdw_intel_res - Soundwire Intel resource structure + * struct sdw_intel_acpi_info - Soundwire Intel information found in ACPI tables + * @handle: ACPI controller handle + * @count: link count found with "sdw-master-count" property + * @link_mask: bit-wise mask listing links enabled by BIOS menu + * + * this structure could be expanded to e.g. provide all the _ADR + * information in case the link_mask is not sufficient to identify + * platform capabilities. + */ +struct sdw_intel_acpi_info { + acpi_handle handle; + int count; + u32 link_mask; +}; + +struct sdw_intel_link_res; + +/** + * struct sdw_intel_ctx - context allocated by the controller + * driver probe + * @count: link count + * @mmio_base: mmio base of SoundWire registers, only used to check + * hardware capabilities after all power dependencies are settled. + * @link_mask: bit-wise mask listing SoundWire links reported by the + * Controller + * @handle: ACPI parent handle + * @links: information for each link (controller-specific and kept + * opaque here) + */ +struct sdw_intel_ctx { + int count; + void __iomem *mmio_base; + u32 link_mask; + acpi_handle handle; + struct sdw_intel_link_res *links; +}; + +/** + * struct sdw_intel_res - Soundwire Intel global resource structure, + * typically populated by the DSP driver + * + * @count: link count * @mmio_base: mmio base of SoundWire registers * @irq: interrupt number * @handle: ACPI parent handle * @parent: parent device * @ops: callback ops - * @arg: callback arg + * @dev: device implementing hwparams and free callbacks + * @link_mask: bit-wise mask listing links selected by the DSP driver + * This mask may be a subset of the one reported by the controller since + * machine-specific quirks are handled in the DSP driver. */ struct sdw_intel_res { + int count; void __iomem *mmio_base; int irq; acpi_handle handle; struct device *parent; const struct sdw_intel_ops *ops; - void *arg; + struct device *dev; + u32 link_mask; };

-void *sdw_intel_init(acpi_handle *parent_handle, struct sdw_intel_res *res); -void sdw_intel_exit(void *arg); +/* + * On Intel platforms, the SoundWire IP has dependencies on power + * rails shared with the DSP, and the initialization steps are split + * in three. First an ACPI scan to check what the firmware describes + * in DSDT tables, then an allocation step (with no hardware + * configuration but with all the relevant devices created) and last + * the actual hardware configuration. The final stage is a global + * interrupt enable which is controlled by the DSP driver. Splitting + * these phases helps simplify the boot flow and make early decisions + * on e.g. which machine driver to select (I2S mode, HDaudio or + * SoundWire). + */ +int sdw_intel_acpi_scan(acpi_handle *parent_handle, + struct sdw_intel_acpi_info *info); + +struct sdw_intel_ctx * +sdw_intel_probe(struct sdw_intel_res *res); + +int sdw_intel_startup(struct sdw_intel_ctx *ctx); + +void sdw_intel_exit(struct sdw_intel_ctx *ctx); + +void sdw_intel_enable_irq(void __iomem *mmio_base, bool enable);

#endif

From: Bard Liao yung-chuan.liao@linux.intel.com

The existing use of 6 handlers is problematic in MSI mode. Update headers so that all shared interrupts can be handled with a single handler.

Signed-off-by: Bard Liao yung-chuan.liao@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- include/linux/soundwire/sdw_intel.h | 4 ++++ 1 file changed, 4 insertions(+)

diff --git a/include/linux/soundwire/sdw_intel.h b/include/linux/soundwire/sdw_intel.h index 3ccb38d48eef..2ce3e9ecc4b6 100644 --- a/include/linux/soundwire/sdw_intel.h +++ b/include/linux/soundwire/sdw_intel.h @@ -4,6 +4,8 @@ #ifndef __SDW_INTEL_H #define __SDW_INTEL_H

+#include <linux/irqreturn.h> + /** * struct sdw_intel_stream_params_data: configuration passed during * the @params_stream callback, e.g. for interaction with DSP @@ -127,4 +129,6 @@ void sdw_intel_exit(struct sdw_intel_ctx *ctx);

void sdw_intel_enable_irq(void __iomem *mmio_base, bool enable);

+irqreturn_t sdw_intel_thread(int irq, void *dev_id); + #endif

From: Rander Wang rander.wang@intel.com

In ClockStop mode, the PCI device will be notified of a wake, which will be handled from an interrupt thread.

Signed-off-by: Rander Wang rander.wang@intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- include/linux/soundwire/sdw_intel.h | 2 ++ 1 file changed, 2 insertions(+)

diff --git a/include/linux/soundwire/sdw_intel.h b/include/linux/soundwire/sdw_intel.h index 2a56180bc9dc..073121c49695 100644 --- a/include/linux/soundwire/sdw_intel.h +++ b/include/linux/soundwire/sdw_intel.h @@ -122,6 +122,8 @@ struct sdw_intel_res { int sdw_intel_acpi_scan(acpi_handle *parent_handle, struct sdw_intel_acpi_info *info);

+void sdw_intel_process_wakeen_event(struct sdw_intel_ctx *ctx); + struct sdw_intel_ctx * sdw_intel_probe(struct sdw_intel_res *res);

Due to power rail dependencies, the SoundWire Master driver cannot make decisions on its own when entering pm runtime suspend.

Add quirk mask for each link, so that the SOF parent driver can inform the SoundWire master driver of the desired behavior: a) leave clock on b) power-off instead of clock stop c) power-off if all devices cannot generate wakes d) force bus reset on clock restart

Note that for now the interface with the SOF driver relies on a single mask for all links. If needed, the interface might be modified at a later point to provide more freedom. The code at the lower level does not assume any commonality between links.

Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- include/linux/soundwire/sdw_intel.h | 37 +++++++++++++++++++++++++++++ 1 file changed, 37 insertions(+)

diff --git a/include/linux/soundwire/sdw_intel.h b/include/linux/soundwire/sdw_intel.h index 45fa6d93197f..93b83bdf8035 100644 --- a/include/linux/soundwire/sdw_intel.h +++ b/include/linux/soundwire/sdw_intel.h @@ -59,6 +59,40 @@ struct sdw_intel_acpi_info {

struct sdw_intel_link_res;

+/* Intel clock-stop/pm_runtime quirk definitions */ + +/* + * Force the clock to remain on during pm_runtime suspend. This might + * be needed if Slave devices do not have an alternate clock source or + * if the latency requirements are very strict. + */ +#define SDW_INTEL_CLK_STOP_NOT_ALLOWED BIT(0) + +/* + * Stop the bus during pm_runtime suspend. If set, a complete bus + * reset and re-enumeration will be performed when the bus + * restarts. This mode shall not be used if Slave devices can generate + * in-band wakes. + */ +#define SDW_INTEL_CLK_STOP_TEARDOWN BIT(1) + +/* + * Stop the bus during pm_suspend if Slaves are not wake capable + * (e.g. speaker amplifiers). The clock-stop mode is typically + * slightly higher power than when the IP is completely powered-off. + */ +#define SDW_INTEL_CLK_STOP_WAKE_CAPABLE_ONLY BIT(2) + +/* + * Require a bus reset (and complete re-enumeration) when exiting + * clock stop modes. This may be needed if the controller power was + * turned off and all context lost. This quirk shall not be used if a + * Slave device needs to remain enumerated and keep its context, + * e.g. to provide the reasons for the wake, report acoustic events or + * pass a history buffer. + */ +#define SDW_INTEL_CLK_STOP_BUS_RESET BIT(3) + /** * struct sdw_intel_ctx - context allocated by the controller * driver probe @@ -97,6 +131,8 @@ struct sdw_intel_ctx { * @link_mask: bit-wise mask listing links selected by the DSP driver * This mask may be a subset of the one reported by the controller since * machine-specific quirks are handled in the DSP driver. + * @clock_stop_quirks: mask array of possible behaviors requested by the + * DSP driver. The quirks are common for all links for now. */ struct sdw_intel_res { int count; @@ -107,6 +143,7 @@ struct sdw_intel_res { const struct sdw_intel_ops *ops; struct device *dev; u32 link_mask; + u32 clock_stop_quirks; };

/*