[PATCH v11 0/4] firmware: ti_sci: Introduce system suspend support

[PATCH v11 0/4] firmware: ti_sci: Introduce system suspend support

[Markus Schneider-Pargmann]

Abstract
********

This series introduces necessary ti_sci driver functionality to support
various Suspend-to-RAM modes on TI AM62 family of devices. These Low
Power Modes include Deep Sleep and MCU Only as described in section
"6.2.4 Power Modes" of the AM62P Technical Reference Manual [0].

Summary
*******

This series is a restructuring and rebase of the patch series by
Dave Gerlach [1] and Dhruva Gole [2]. It applies on top of Linux
6.11-rc1.

The kernel triggers entry to Low Power Mode through the mem suspend
transition with the following:

* At the  bootloader stage, one is  expected to package the  TIFS stub
  which then gets  pulled into the Tightly coupled memory  of the Device
  Mgr  (DM) R5  when it  starts up.  If using  U-Boot, then  it requires
  tispl.bin to contain the TIFS stub. Refer to documentation in upstream
  u-boot[3] for further details. The  supported firmware version is from
  TI Processor SDK >= 10.00 ie. tag 10.00.04 from ti-linux-firmware [4].

* Use a TF-A binary that supports PSCI_SYSTEM_SUSPEND call. This causes
  system to use PSCI system suspend as last step of mem sleep.

* We add support for the TISCI_MSG_QUERY_FW_CAPS message, used to retrieve
  the firmware capabilities of the currently running system firmware [6].
  Sysfw version >= 10.00.04 support LPM_DM_MANAGED capability, where
  Device Mgr firmware now manages which low power mode is chosen. Going
  forward, this is the default configuration supported for TI AM62 family
  of devices. The state chosen by the DM can be influenced by sending
  constraints using the new LPM constraint APIs. (Patch 1)

* The firmware requires that the OS sends a TISCI_MSG_PREPARE_SLEEP
  message in order to provide details about suspend. The ti_sci driver
  must send this message to firmware with the above information
  included, which it does during the driver suspend handler when
  PM_MEM_SUSPEND is the determined state being entered. The mode being
  sent depends on whether firmware capabilities have support for
  LPM_DM_MANAGED feature. Legacy firmware or those supporting other
  modes can extend the mode selection logic as needed. (Patch 2)

* We also add the remaining TISCI Low Power Mode messages required for
  inquiring wake reason and managing LPM constraints as part of a new PM
  ops. These messages are part of the TISCI PM Low Power Mode API [5].
  (Patch 3)

* Finally if any CPUs have PM QoS resume latency constraints set, we
  aggregate these and set the TISCI system-wide latency constraint.
  (Patch 4)

Testing
*******

This series can for example be tested with a am62a-lp-sk board.

For am62a-lp-sk all usb nodes have to be disabled at the moment (usbss0,
usb0, usbss1 and usb1). There is currently an issue with USB Link Power
Management and turning off the USB device which is being worked on.

Once booted suspend/resume can be tested with rtcwake:
  $ rtcwake -m mem -s 10 -d /dev/rtc0

Make sure /dev/rtc0 corresponds to rtc-ti-k3:
  $ dmesg | grep rtc-ti-k3
  rtc-ti-k3 2b1f0000.rtc: registered as rtc0

Base commit:
************

v6.11-rc1

Changelog:
**********
v11:
- Add dev_err() in many error cases
- Fix a typo in the commit message of 'firmware: ti_sci: Add support for
  querying the firmware caps'
- Fix URL in commit message
- Unify error handling and printing over the added ti_sci_cmd_*
  functions
- Use dev_* instead of pr_*
- Add a debug message listing the capabilities the firmware returned
- Add a small comment describing why context address is 0x0
- Removed unnecessary debug prints
- Restructure the ifdefs for pm ops as suggested
- Only print the wake reason if there was no error. If there is an
  error, the ti_sci_cmd_lpm_wake_reason already printed an error. As a
  missing reason is not critical, the error is ignored and the system
  continues to resume
- Remove wakeup source macros from sci_protocol.h

v10:
- add "wake reason" handling which is also supported by 10.x
  firmware update, but was mistakenly left out of v9
- fix debug print of which CPU caused max CPU latency
- update TRM pointer[0] to point to AM62P TRM which has better
  description of low-power modes shared across AM62P family
- update u-boot documentaion pointer to point to upstream u-boot
  commit.

v9:
- Include Kevin's patch to add CPU latency constraint management into
  this series. Posted here in v3:
  https://lore.kernel.org/lkml/20240802214220.3472221-1-khilman@baylibre.com/
- reorder patches to avoid any build warnings
- ti_sci_cmd_prepare_sleep was moved into the patch that adds suspend
  and resume calls
- pmops wake_reason is now only set if the capabilities exist
- Use pmops pointer instead of full path

v8:
- Restructuring of code to include all TISCI PM LPM ops
- Removing malloc related to TIFS Stub as it is managed by DM
- Dropping has_lpm check as suggested by Nishanth
- Using LPM_DM_MANAGED capability for mode selection
- Updating the suspend and resume callback handlers

v7:
- Address Andrew's concerns on SYSFW fw_caps API
- Remove all the unused functions and variables including
  set_io_isolation and wake_reason calls
- use dma_free_attrs
- remove IO isolation related code from linux side,

v6:
- link to v6 [5]
- Loading of FS Stub from linux no longer needed, hence drop that patch,
- Drop 1/6 and 5/6 from the previous series [4].
- Add system suspend resume callbacks which were removed in
commit 9225bcdedf16297a346082e7d23b0e8434aa98ed ("firmware: ti_sci: Use
system_state to determine polling")
- Use IO isolation while putting the system in suspend to RAM

v5:
- Add support (patch 3) for detecting the low power modes (LPM) of the
  FW/SoC with a recently introduced core TISCI_MSG_QUERY_FW_CAPS message.
- Use TISCI_MSG_QUERY_FW_CAPS instead of misusing the
  TISCI_MSG_PREPARE_SLEEP to detect the FW/SoC low power caps (patch 4).
- Take into account the supported LPMs in ti_sci_prepare_system_suspend()
  and handle the case when CONFIG_SUSPEND is not enabled (patch 6) that
  was reported by Roger Quadros and LKP.
- Pick up Rob Herring's "Reviewed-by" tag for the binding patch.

v4:
- Fix checkpacth warnings in patches 2 and 3.
- Drop the links with anchors in patch 2.

v3:
- Fix the compile warnings on 32-bit platforms reported by the kernel
  test robot in patches (3,5).
- Pick up Roger's "Tested-by" tags.

v2:
- Addressed comments received for v1 series [1].
- Updated v1 patch 5 to use pm notifier to avoid firmware loading
  issues.
- Dropped the reserved region requirement and allocate DMA memory
  instead. The reserved region binding patch is also removed.
- Introduce two more TISCI LPM messages that are supported in SysFW.
- Fixes in error handling.

References:
***********

[0] https://www.ti.com/lit/pdf/spruiv7
[1] https://lore.kernel.org/lkml/20220421203659.27853-1-d-gerlach@ti.com
[2] https://lore.kernel.org/lkml/20230804115037.754994-1-d-gole@ti.com
[3] https://source.denx.de/u-boot/u-boot/-/commit/962f60abca82bb11501bc0c627abacda15bed076
[4] https://git.ti.com/cgit/processor-firmware/ti-linux-firmware/commit/?h=10.00.06&id=193f7d7570583a41ddc50a221e37c32be6be583e
[5] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html
[6] https://downloads.ti.com/tisci/esd/latest/2_tisci_msgs/general/core.html#tisci-msg-query-fw-caps

Dave Gerlach (1):
  firmware: ti_sci: Introduce Power Management Ops

Georgi Vlaev (1):
  firmware: ti_sci: Add support for querying the firmware caps

Kevin Hilman (1):
  firmware: ti_sci: add CPU latency constraint management

Vibhore Vardhan (1):
  firmware: ti_sci: Add system suspend and resume call

 drivers/firmware/ti_sci.c              | 490 ++++++++++++++++++++++++-
 drivers/firmware/ti_sci.h              | 143 +++++++-
 include/linux/soc/ti/ti_sci_protocol.h |  30 ++
 3 files changed, 661 insertions(+), 2 deletions(-)

-- 
2.45.2

[PATCH v11 1/4] firmware: ti_sci: Add support for querying the firmware caps

[Markus Schneider-Pargmann]

From: Georgi Vlaev <g-vlaev@ti.com>

Add support for the TISCI_MSG_QUERY_FW_CAPS message, used to retrieve
the firmware capabilities of the currently running system firmware. The
message belongs to the TISCI general core message API [1] and is
available in SysFW version 08.04.03 and above. Currently, the message is
supported on devices with split architecture of the system firmware (DM
+ TIFS) like AM62x. Old revisions or not yet supported platforms will
NACK this request.

We're using this message locally in ti_sci.c to get the low power
features of the FW/SoC. As there's no other kernel consumers yet, this
is not added to struct ti_sci_core_ops.

Sysfw version >= 10.00.04 support LPM_DM_MANAGED capability [2], where
Device Mgr firmware now manages which low power mode is chosen. Going
forward, this is the default configuration supported for TI AM62 family
of devices. The state chosen by the DM can be influenced by sending
constraints using the new LPM constraint APIs.

[1] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/general/core.html
[2] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/general/core.html#tisci-msg-query-fw-caps

Signed-off-by: Georgi Vlaev <g-vlaev@ti.com>
[vibhore@ti.com: Support for LPM_DM_MANAGED mode]
Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
---
 drivers/firmware/ti_sci.c | 74 ++++++++++++++++++++++++++++++++++++++-
 drivers/firmware/ti_sci.h | 22 ++++++++++++
 2 files changed, 95 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index 160968301b1f..bed7c8892341 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -2,10 +2,11 @@
 /*
  * Texas Instruments System Control Interface Protocol Driver
  *
- * Copyright (C) 2015-2022 Texas Instruments Incorporated - https://www.ti.com/
+ * Copyright (C) 2015-2024 Texas Instruments Incorporated - https://www.ti.com/
  *	Nishanth Menon
  */
 
+#include "linux/dev_printk.h"
 #define pr_fmt(fmt) "%s: " fmt, __func__
 
 #include <linux/bitmap.h>
@@ -24,6 +25,7 @@
 #include <linux/slab.h>
 #include <linux/soc/ti/ti-msgmgr.h>
 #include <linux/soc/ti/ti_sci_protocol.h>
+#include <linux/sys_soc.h>
 #include <linux/reboot.h>
 
 #include "ti_sci.h"
@@ -98,6 +100,7 @@ struct ti_sci_desc {
  * @minfo:	Message info
  * @node:	list head
  * @host_id:	Host ID
+ * @fw_caps:	FW/SoC low power capabilities
  * @users:	Number of users of this instance
  */
 struct ti_sci_info {
@@ -114,6 +117,7 @@ struct ti_sci_info {
 	struct ti_sci_xfers_info minfo;
 	struct list_head node;
 	u8 host_id;
+	u64 fw_caps;
 	/* protected by ti_sci_list_mutex */
 	int users;
 };
@@ -1651,6 +1655,67 @@ static int ti_sci_cmd_clk_get_freq(const struct ti_sci_handle *handle,
 	return ret;
 }
 
+/**
+ * ti_sci_msg_cmd_query_fw_caps() - Get the FW/SoC capabilities
+ * @handle:		Pointer to TI SCI handle
+ * @fw_caps:		Each bit in fw_caps indicating one FW/SOC capability
+ *
+ * Check if the firmware supports any optional low power modes.
+ * Old revisions of TIFS (< 08.04) will NACK the request which results in
+ * -ENODEV being returned.
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_msg_cmd_query_fw_caps(const struct ti_sci_handle *handle,
+					u64 *fw_caps)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_xfer *xfer;
+	struct ti_sci_msg_resp_query_fw_caps *resp;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_QUERY_FW_CAPS,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(struct ti_sci_msg_hdr),
+				   sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_resp_query_fw_caps *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to get capabilities\n");
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	if (fw_caps)
+		*fw_caps = resp->fw_caps;
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
 static int ti_sci_cmd_core_reboot(const struct ti_sci_handle *handle)
 {
 	struct ti_sci_info *info;
@@ -3390,6 +3455,13 @@ static int ti_sci_probe(struct platform_device *pdev)
 		goto out;
 	}
 
+	ti_sci_msg_cmd_query_fw_caps(&info->handle, &info->fw_caps);
+	dev_dbg(dev, "Detected firmware capabilities: %s%s%s\n",
+		info->fw_caps & MSG_FLAG_CAPS_GENERIC ? "Generic" : "",
+		info->fw_caps & MSG_FLAG_CAPS_LPM_PARTIAL_IO ? " Partial-IO" : "",
+		info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED ? " DM-Managed" : ""
+	);
+
 	ti_sci_setup_ops(info);
 
 	ret = devm_register_restart_handler(dev, tisci_reboot_handler, info);
diff --git a/drivers/firmware/ti_sci.h b/drivers/firmware/ti_sci.h
index 5846c60220f5..73ca9503606b 100644
--- a/drivers/firmware/ti_sci.h
+++ b/drivers/firmware/ti_sci.h
@@ -19,6 +19,7 @@
 #define TI_SCI_MSG_WAKE_REASON	0x0003
 #define TI_SCI_MSG_GOODBYE	0x0004
 #define TI_SCI_MSG_SYS_RESET	0x0005
+#define TI_SCI_MSG_QUERY_FW_CAPS	0x0022
 
 /* Device requests */
 #define TI_SCI_MSG_SET_DEVICE_STATE	0x0200
@@ -132,6 +133,27 @@ struct ti_sci_msg_req_reboot {
 	struct ti_sci_msg_hdr hdr;
 } __packed;
 
+/**
+ * struct ti_sci_msg_resp_query_fw_caps - Response for query firmware caps
+ * @hdr:	Generic header
+ * @fw_caps:	Each bit in fw_caps indicating one FW/SOC capability
+ *		MSG_FLAG_CAPS_GENERIC: Generic capability (LPM not supported)
+ *		MSG_FLAG_CAPS_LPM_PARTIAL_IO: Partial IO in LPM
+ *		MSG_FLAG_CAPS_LPM_DM_MANAGED: LPM can be managed by DM
+ *
+ * Response to a generic message with message type TI_SCI_MSG_QUERY_FW_CAPS
+ * providing currently available SOC/firmware capabilities. SoC that don't
+ * support low power modes return only MSG_FLAG_CAPS_GENERIC capability.
+ */
+struct ti_sci_msg_resp_query_fw_caps {
+	struct ti_sci_msg_hdr hdr;
+#define MSG_FLAG_CAPS_GENERIC		TI_SCI_MSG_FLAG(0)
+#define MSG_FLAG_CAPS_LPM_PARTIAL_IO	TI_SCI_MSG_FLAG(4)
+#define MSG_FLAG_CAPS_LPM_DM_MANAGED	TI_SCI_MSG_FLAG(5)
+#define MSG_MASK_CAPS_LPM		GENMASK_ULL(4, 1)
+	u64 fw_caps;
+} __packed;
+
 /**
  * struct ti_sci_msg_req_set_device_state - Set the desired state of the device
  * @hdr:		Generic header
-- 
2.45.2

[PATCH v11 2/4] firmware: ti_sci: Add system suspend and resume call

[Markus Schneider-Pargmann]

From: Vibhore Vardhan <vibhore@ti.com>

Introduce system suspend call that enables the ti_sci driver to support
low power mode when the user space issues a suspend to mem.

The following power management operations defined in the TISCI
Low Power Mode API [1] are implemented to support suspend and resume:

1) TISCI_MSG_PREPARE_SLEEP
Prepare the SOC for entering into a low power mode and
provide details to firmware about the state being entered.

2) TISCI_MSG_SET_IO_ISOLATION
Control the IO isolation for Low Power Mode.

Also, write a ti_sci_prepare_system_suspend call to be used in the driver
suspend handler to allow the system to identify the low power mode being
entered and if necessary, send TISCI_MSG_PREPARE_SLEEP with information
about the mode being entered.

Sysfw version >= 10.00.04 support LPM_DM_MANAGED capability [2], where
Device Mgr firmware now manages which low power mode is chosen. Going
forward, this is the default configuration supported for TI AM62 family
of devices. The state chosen by the DM can be influenced by sending
constraints using the new LPM constraint APIs.

In case the firmware does not support LPM_DM_MANAGED mode, the mode
selection logic can be extended as needed. If no suspend-to-RAM modes
are supported, return without taking any action.

We're using "pm_suspend_target_state" to map the kernel's target suspend
state to SysFW low power mode. Make sure this is available only when
CONFIG_SUSPEND is enabled.

Suspend has to be split into two parts, ti_sci_suspend() will send
the prepare sleep message to prepare suspend. ti_sci_suspend_noirq()
sets IO isolation which needs to be done as late as possible to avoid
any issues. On resume this has to be done as early as possible.

[1] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html

Co-developed-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Georgi Vlaev <g-vlaev@ti.com>
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
---
 drivers/firmware/ti_sci.c              | 196 +++++++++++++++++++++++++
 drivers/firmware/ti_sci.h              |  45 +++++-
 include/linux/soc/ti/ti_sci_protocol.h |   4 +
 3 files changed, 244 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index bed7c8892341..8a772b378ffa 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -25,6 +25,7 @@
 #include <linux/slab.h>
 #include <linux/soc/ti/ti-msgmgr.h>
 #include <linux/soc/ti/ti_sci_protocol.h>
+#include <linux/suspend.h>
 #include <linux/sys_soc.h>
 #include <linux/reboot.h>
 
@@ -1655,6 +1656,68 @@ static int ti_sci_cmd_clk_get_freq(const struct ti_sci_handle *handle,
 	return ret;
 }
 
+/**
+ * ti_sci_cmd_prepare_sleep() - Prepare system for system suspend
+ * @handle:		pointer to TI SCI handle
+ * @mode:		Device identifier
+ * @ctx_lo:		Low part of address for context save
+ * @ctx_hi:		High part of address for context save
+ * @debug_flags:	Debug flags to pass to firmware
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_cmd_prepare_sleep(const struct ti_sci_handle *handle, u8 mode,
+				    u32 ctx_lo, u32 ctx_hi, u32 debug_flags)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_msg_req_prepare_sleep *req;
+	struct ti_sci_msg_hdr *resp;
+	struct ti_sci_xfer *xfer;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_PREPARE_SLEEP,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(*req), sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+
+	req = (struct ti_sci_msg_req_prepare_sleep *)xfer->xfer_buf;
+	req->mode = mode;
+	req->ctx_lo = ctx_lo;
+	req->ctx_hi = ctx_hi;
+	req->debug_flags = debug_flags;
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_hdr *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to prepare sleep\n");
+		ret = -ENODEV;
+	}
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
 /**
  * ti_sci_msg_cmd_query_fw_caps() - Get the FW/SoC capabilities
  * @handle:		Pointer to TI SCI handle
@@ -1716,6 +1779,61 @@ static int ti_sci_msg_cmd_query_fw_caps(const struct ti_sci_handle *handle,
 	return ret;
 }
 
+/**
+ * ti_sci_cmd_set_io_isolation() - Enable IO isolation in LPM
+ * @handle:		Pointer to TI SCI handle
+ * @state:		The desired state of the IO isolation
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_cmd_set_io_isolation(const struct ti_sci_handle *handle,
+				       u8 state)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_msg_req_set_io_isolation *req;
+	struct ti_sci_msg_hdr *resp;
+	struct ti_sci_xfer *xfer;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_SET_IO_ISOLATION,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(*req), sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+	req = (struct ti_sci_msg_req_set_io_isolation *)xfer->xfer_buf;
+	req->state = state;
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_hdr *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to set IO isolation\n");
+		ret = -ENODEV;
+	}
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
 static int ti_sci_cmd_core_reboot(const struct ti_sci_handle *handle)
 {
 	struct ti_sci_info *info;
@@ -3327,6 +3445,83 @@ static int tisci_reboot_handler(struct sys_off_data *data)
 	return NOTIFY_BAD;
 }
 
+static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
+{
+	/*
+	 * Map and validate the target Linux suspend state to TISCI LPM.
+	 * Default is to let Device Manager select the low power mode.
+	 */
+	switch (pm_suspend_target_state) {
+	case PM_SUSPEND_MEM:
+		if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
+			/*
+			 * For the DM_MANAGED mode the context is reserved for
+			 * internal use and can be 0
+			 */
+			return ti_sci_cmd_prepare_sleep(&info->handle,
+							TISCI_MSG_VALUE_SLEEP_MODE_DM_MANAGED,
+							0, 0, 0);
+		} else {
+			/* DM Managed is not supported by the firmware. */
+			dev_err(info->dev, "Suspend to memory is not supported by the firmware\n");
+			return -EOPNOTSUPP;
+		}
+		break;
+	default:
+		/*
+		 * Do not fail if we don't have action to take for a
+		 * specific suspend mode.
+		 */
+		return 0;
+	}
+
+	return 0;
+}
+
+static int __maybe_unused ti_sci_suspend(struct device *dev)
+{
+	struct ti_sci_info *info = dev_get_drvdata(dev);
+	int ret;
+
+	ret = ti_sci_prepare_system_suspend(info);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int __maybe_unused ti_sci_suspend_noirq(struct device *dev)
+{
+	struct ti_sci_info *info = dev_get_drvdata(dev);
+	int ret = 0;
+
+	ret = ti_sci_cmd_set_io_isolation(&info->handle, TISCI_MSG_VALUE_IO_ENABLE);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int __maybe_unused ti_sci_resume_noirq(struct device *dev)
+{
+	struct ti_sci_info *info = dev_get_drvdata(dev);
+	int ret = 0;
+
+	ret = ti_sci_cmd_set_io_isolation(&info->handle, TISCI_MSG_VALUE_IO_DISABLE);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static const struct dev_pm_ops ti_sci_pm_ops = {
+#ifdef CONFIG_PM_SLEEP
+	.suspend = ti_sci_suspend,
+	.suspend_noirq = ti_sci_suspend_noirq,
+	.resume_noirq = ti_sci_resume_noirq,
+#endif
+};
+
 /* Description for K2G */
 static const struct ti_sci_desc ti_sci_pmmc_k2g_desc = {
 	.default_host_id = 2,
@@ -3495,6 +3690,7 @@ static struct platform_driver ti_sci_driver = {
 		   .name = "ti-sci",
 		   .of_match_table = of_match_ptr(ti_sci_of_match),
 		   .suppress_bind_attrs = true,
+		   .pm = &ti_sci_pm_ops,
 	},
 };
 module_platform_driver(ti_sci_driver);
diff --git a/drivers/firmware/ti_sci.h b/drivers/firmware/ti_sci.h
index 73ca9503606b..8efe4d0e61fb 100644
--- a/drivers/firmware/ti_sci.h
+++ b/drivers/firmware/ti_sci.h
@@ -6,7 +6,7 @@
  * The system works in a message response protocol
  * See: https://software-dl.ti.com/tisci/esd/latest/index.html for details
  *
- * Copyright (C)  2015-2016 Texas Instruments Incorporated - https://www.ti.com/
+ * Copyright (C)  2015-2024 Texas Instruments Incorporated - https://www.ti.com/
  */
 
 #ifndef __TI_SCI_H
@@ -36,6 +36,10 @@
 #define TI_SCI_MSG_QUERY_CLOCK_FREQ	0x010d
 #define TI_SCI_MSG_GET_CLOCK_FREQ	0x010e
 
+/* Low Power Mode Requests */
+#define TI_SCI_MSG_PREPARE_SLEEP	0x0300
+#define TI_SCI_MSG_SET_IO_ISOLATION	0x0307
+
 /* Resource Management Requests */
 #define TI_SCI_MSG_GET_RESOURCE_RANGE	0x1500
 
@@ -567,6 +571,45 @@ struct ti_sci_msg_resp_get_clock_freq {
 	u64 freq_hz;
 } __packed;
 
+/**
+ * struct tisci_msg_req_prepare_sleep - Request for TISCI_MSG_PREPARE_SLEEP.
+ *
+ * @hdr				TISCI header to provide ACK/NAK flags to the host.
+ * @mode			Low power mode to enter.
+ * @ctx_lo			Low 32-bits of physical pointer to address to use for context save.
+ * @ctx_hi			High 32-bits of physical pointer to address to use for context save.
+ * @debug_flags			Flags that can be set to halt the sequence during suspend or
+ *				resume to allow JTAG connection and debug.
+ *
+ * This message is used as the first step of entering a low power mode. It
+ * allows configurable information, including which state to enter to be
+ * easily shared from the application, as this is a non-secure message and
+ * therefore can be sent by anyone.
+ */
+struct ti_sci_msg_req_prepare_sleep {
+	struct ti_sci_msg_hdr	hdr;
+
+#define TISCI_MSG_VALUE_SLEEP_MODE_DM_MANAGED				0xfd
+	u8			mode;
+	u32			ctx_lo;
+	u32			ctx_hi;
+	u32			debug_flags;
+} __packed;
+
+/**
+ * struct tisci_msg_set_io_isolation_req - Request for TI_SCI_MSG_SET_IO_ISOLATION.
+ *
+ * @hdr:	Generic header
+ * @state:	The deseared state of the IO isolation.
+ *
+ * This message is used to enable/disable IO isolation for low power modes.
+ * Response is generic ACK / NACK message.
+ */
+struct ti_sci_msg_req_set_io_isolation {
+	struct ti_sci_msg_hdr hdr;
+	u8 state;
+} __packed;
+
 #define TI_SCI_IRQ_SECONDARY_HOST_INVALID	0xff
 
 /**
diff --git a/include/linux/soc/ti/ti_sci_protocol.h b/include/linux/soc/ti/ti_sci_protocol.h
index bd0d11af76c5..1f1871e23f76 100644
--- a/include/linux/soc/ti/ti_sci_protocol.h
+++ b/include/linux/soc/ti/ti_sci_protocol.h
@@ -195,6 +195,10 @@ struct ti_sci_clk_ops {
 			u64 *current_freq);
 };
 
+/* TISCI LPM IO isolation control values */
+#define TISCI_MSG_VALUE_IO_ENABLE			1
+#define TISCI_MSG_VALUE_IO_DISABLE			0
+
 /**
  * struct ti_sci_resource_desc - Description of TI SCI resource instance range.
  * @start:	Start index of the first resource range.
-- 
2.45.2

[PATCH v11 3/4] firmware: ti_sci: Introduce Power Management Ops

[Markus Schneider-Pargmann]

From: Dave Gerlach <d-gerlach@ti.com>

Introduce power management ops supported by the TISCI
Low Power Mode API [1].

1) TISCI_MSG_LPM_WAKE_REASON
Get which wake up source woke the SoC from Low Power Mode.
The wake up source IDs will be common for all K3 platforms.

2) TISCI_MSG_LPM_SET_DEVICE_CONSTRAINT
Set LPM constraint on behalf of a device. By setting a constraint, the
device ensures that it will not be powered off or reset in the selected
mode.

3) TISCI_MSG_LPM_SET_LATENCY_CONSTRAINT
Set LPM resume latency constraint. By setting a constraint, the host
ensures that the resume time from selected mode will be less than the
constraint value.

[1] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html

Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
[g-vlaev@ti.com: LPM_WAKE_REASON and IO_ISOLATION support]
Signed-off-by: Georgi Vlaev <g-vlaev@ti.com>
[a-kaur@ti.com: SET_DEVICE_CONSTRAINT support]
Signed-off-by: Akashdeep Kaur <a-kaur@ti.com>
[vibhore@ti.com: SET_LATENCY_CONSTRAINT support]
Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Reviewed-by: Akashdeep Kaur <a-kaur@ti.com>
Tested-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
---
 drivers/firmware/ti_sci.c              | 198 +++++++++++++++++++++++++
 drivers/firmware/ti_sci.h              |  76 ++++++++++
 include/linux/soc/ti/ti_sci_protocol.h |  26 ++++
 3 files changed, 300 insertions(+)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index 8a772b378ffa..8482b7077eef 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -1834,6 +1834,186 @@ static int ti_sci_cmd_set_io_isolation(const struct ti_sci_handle *handle,
 	return ret;
 }
 
+/**
+ * ti_sci_msg_cmd_lpm_wake_reason() - Get the wakeup source from LPM
+ * @handle:		Pointer to TI SCI handle
+ * @source:		The wakeup source that woke the SoC from LPM
+ * @timestamp:		Timestamp of the wakeup event
+ * @pin:		The pin that has triggered wake up
+ * @mode:		The last entered low power mode
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_msg_cmd_lpm_wake_reason(const struct ti_sci_handle *handle,
+					  u32 *source, u64 *timestamp, u8 *pin, u8 *mode)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_xfer *xfer;
+	struct ti_sci_msg_resp_lpm_wake_reason *resp;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_LPM_WAKE_REASON,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(struct ti_sci_msg_hdr),
+				   sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_resp_lpm_wake_reason *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to get wake reason\n");
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	if (source)
+		*source = resp->wake_source;
+	if (timestamp)
+		*timestamp = resp->wake_timestamp;
+	if (pin)
+		*pin = resp->wake_pin;
+	if (mode)
+		*mode = resp->mode;
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
+/**
+ * ti_sci_cmd_set_device_constraint() - Set LPM constraint on behalf of a device
+ * @handle:	pointer to TI SCI handle
+ * @id:	Device identifier
+ * @state:	The desired state of device constraint: set or clear
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_cmd_set_device_constraint(const struct ti_sci_handle *handle,
+					    u32 id, u8 state)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_msg_req_lpm_set_device_constraint *req;
+	struct ti_sci_msg_hdr *resp;
+	struct ti_sci_xfer *xfer;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_LPM_SET_DEVICE_CONSTRAINT,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(*req), sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+	req = (struct ti_sci_msg_req_lpm_set_device_constraint *)xfer->xfer_buf;
+	req->id = id;
+	req->state = state;
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_hdr *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to set device constraint\n");
+		ret = -ENODEV;
+	}
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
+/**
+ * ti_sci_cmd_set_latency_constraint() - Set LPM resume latency constraint
+ * @handle:	pointer to TI SCI handle
+ * @latency:	maximum acceptable latency (in ms) to wake up from LPM
+ * @state:	The desired state of latency constraint: set or clear
+ *
+ * Return: 0 if all went well, else returns appropriate error value.
+ */
+static int ti_sci_cmd_set_latency_constraint(const struct ti_sci_handle *handle,
+					     u16 latency, u8 state)
+{
+	struct ti_sci_info *info;
+	struct ti_sci_msg_req_lpm_set_latency_constraint *req;
+	struct ti_sci_msg_hdr *resp;
+	struct ti_sci_xfer *xfer;
+	struct device *dev;
+	int ret = 0;
+
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+	if (!handle)
+		return -EINVAL;
+
+	info = handle_to_ti_sci_info(handle);
+	dev = info->dev;
+
+	xfer = ti_sci_get_one_xfer(info, TI_SCI_MSG_LPM_SET_LATENCY_CONSTRAINT,
+				   TI_SCI_FLAG_REQ_ACK_ON_PROCESSED,
+				   sizeof(*req), sizeof(*resp));
+	if (IS_ERR(xfer)) {
+		ret = PTR_ERR(xfer);
+		dev_err(dev, "Message alloc failed(%d)\n", ret);
+		return ret;
+	}
+	req = (struct ti_sci_msg_req_lpm_set_latency_constraint *)xfer->xfer_buf;
+	req->latency = latency;
+	req->state = state;
+
+	ret = ti_sci_do_xfer(info, xfer);
+	if (ret) {
+		dev_err(dev, "Mbox send fail %d\n", ret);
+		goto fail;
+	}
+
+	resp = (struct ti_sci_msg_hdr *)xfer->xfer_buf;
+
+	if (!ti_sci_is_response_ack(resp)) {
+		dev_err(dev, "Failed to set device constraint\n");
+		ret = -ENODEV;
+	}
+
+fail:
+	ti_sci_put_one_xfer(&info->minfo, xfer);
+
+	return ret;
+}
+
 static int ti_sci_cmd_core_reboot(const struct ti_sci_handle *handle)
 {
 	struct ti_sci_info *info;
@@ -2976,6 +3156,7 @@ static void ti_sci_setup_ops(struct ti_sci_info *info)
 	struct ti_sci_core_ops *core_ops = &ops->core_ops;
 	struct ti_sci_dev_ops *dops = &ops->dev_ops;
 	struct ti_sci_clk_ops *cops = &ops->clk_ops;
+	struct ti_sci_pm_ops *pmops = &ops->pm_ops;
 	struct ti_sci_rm_core_ops *rm_core_ops = &ops->rm_core_ops;
 	struct ti_sci_rm_irq_ops *iops = &ops->rm_irq_ops;
 	struct ti_sci_rm_ringacc_ops *rops = &ops->rm_ring_ops;
@@ -3015,6 +3196,13 @@ static void ti_sci_setup_ops(struct ti_sci_info *info)
 	cops->set_freq = ti_sci_cmd_clk_set_freq;
 	cops->get_freq = ti_sci_cmd_clk_get_freq;
 
+	if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
+		pr_debug("detected DM managed LPM in fw_caps\n");
+		pmops->lpm_wake_reason = ti_sci_msg_cmd_lpm_wake_reason;
+		pmops->set_device_constraint = ti_sci_cmd_set_device_constraint;
+		pmops->set_latency_constraint = ti_sci_cmd_set_latency_constraint;
+	}
+
 	rm_core_ops->get_range = ti_sci_cmd_get_resource_range;
 	rm_core_ops->get_range_from_shost =
 				ti_sci_cmd_get_resource_range_from_shost;
@@ -3506,11 +3694,21 @@ static int __maybe_unused ti_sci_resume_noirq(struct device *dev)
 {
 	struct ti_sci_info *info = dev_get_drvdata(dev);
 	int ret = 0;
+	u32 source;
+	u64 time;
+	u8 pin;
+	u8 mode;
 
 	ret = ti_sci_cmd_set_io_isolation(&info->handle, TISCI_MSG_VALUE_IO_DISABLE);
 	if (ret)
 		return ret;
 
+	ret = ti_sci_msg_cmd_lpm_wake_reason(&info->handle, &source, &time, &pin, &mode);
+	/* Do not fail to resume on error as the wake reason is not critical */
+	if (!ret)
+		dev_info(dev, "ti_sci: wakeup source:0x%x, pin:0x%x, mode:0x%x\n",
+			 source, pin, mode);
+
 	return 0;
 }
 
diff --git a/drivers/firmware/ti_sci.h b/drivers/firmware/ti_sci.h
index 8efe4d0e61fb..053387d7baa0 100644
--- a/drivers/firmware/ti_sci.h
+++ b/drivers/firmware/ti_sci.h
@@ -38,7 +38,10 @@
 
 /* Low Power Mode Requests */
 #define TI_SCI_MSG_PREPARE_SLEEP	0x0300
+#define TI_SCI_MSG_LPM_WAKE_REASON	0x0306
 #define TI_SCI_MSG_SET_IO_ISOLATION	0x0307
+#define TI_SCI_MSG_LPM_SET_DEVICE_CONSTRAINT	0x0309
+#define TI_SCI_MSG_LPM_SET_LATENCY_CONSTRAINT	0x030A
 
 /* Resource Management Requests */
 #define TI_SCI_MSG_GET_RESOURCE_RANGE	0x1500
@@ -610,6 +613,79 @@ struct ti_sci_msg_req_set_io_isolation {
 	u8 state;
 } __packed;
 
+/**
+ * struct ti_sci_msg_resp_lpm_wake_reason - Response for TI_SCI_MSG_LPM_WAKE_REASON.
+ *
+ * @hdr:		Generic header.
+ * @wake_source:	The wake up source that woke soc from LPM.
+ * @wake_timestamp:	Timestamp at which soc woke.
+ * @wake_pin: The pin that has triggered wake up.
+ * @mode: The last entered low power mode.
+ * @rsvd:	Reserved for future use.
+ *
+ * Response to a generic message with message type TI_SCI_MSG_LPM_WAKE_REASON,
+ * used to query the wake up source, pin and entered low power mode.
+ */
+struct ti_sci_msg_resp_lpm_wake_reason {
+	struct ti_sci_msg_hdr hdr;
+	u32 wake_source;
+	u64 wake_timestamp;
+	u8 wake_pin;
+	u8 mode;
+	u32 rsvd[2];
+} __packed;
+
+/**
+ * struct ti_sci_msg_req_lpm_set_device_constraint - Request for
+ * TISCI_MSG_LPM_SET_DEVICE_CONSTRAINT.
+ *
+ * @hdr:	TISCI header to provide ACK/NAK flags to the host.
+ * @id:	Device ID of device whose constraint has to be modified.
+ * @state:	The desired state of device constraint: set or clear.
+ * @rsvd:	Reserved for future use.
+ *
+ * This message is used by host to set constraint on the device. This can be
+ * sent anytime after boot before prepare sleep message. Any device can set a
+ * constraint on the low power mode that the SoC can enter. It allows
+ * configurable information to be easily shared from the application, as this
+ * is a non-secure message and therefore can be sent by anyone. By setting a
+ * constraint, the device ensures that it will not be powered off or reset in
+ * the selected mode. Note: Access Restriction: Exclusivity flag of Device will
+ * be honored. If some other host already has constraint on this device ID,
+ * NACK will be returned.
+ */
+struct ti_sci_msg_req_lpm_set_device_constraint {
+	struct ti_sci_msg_hdr hdr;
+	u32 id;
+	u8 state;
+	u32 rsvd[2];
+} __packed;
+
+/**
+ * struct ti_sci_msg_req_lpm_set_latency_constraint - Request for
+ * TISCI_MSG_LPM_SET_LATENCY_CONSTRAINT.
+ *
+ * @hdr:	TISCI header to provide ACK/NAK flags to the host.
+ * @wkup_latency:	The maximum acceptable latency to wake up from low power mode
+ *			in milliseconds. The deeper the state, the higher the latency.
+ * @state:	The desired state of wakeup latency constraint: set or clear.
+ * @rsvd:	Reserved for future use.
+ *
+ * This message is used by host to set wakeup latency from low power mode. This can
+ * be sent anytime after boot before prepare sleep message, and can be sent after
+ * current low power mode is exited. Any device can set a constraint on the low power
+ * mode that the SoC can enter. It allows configurable information to be easily shared
+ * from the application, as this is a non-secure message and therefore can be sent by
+ * anyone. By setting a wakeup latency constraint, the host ensures that the resume time
+ * from selected low power mode will be less than the constraint value.
+ */
+struct ti_sci_msg_req_lpm_set_latency_constraint {
+	struct ti_sci_msg_hdr hdr;
+	u16 latency;
+	u8 state;
+	u32 rsvd;
+} __packed;
+
 #define TI_SCI_IRQ_SECONDARY_HOST_INVALID	0xff
 
 /**
diff --git a/include/linux/soc/ti/ti_sci_protocol.h b/include/linux/soc/ti/ti_sci_protocol.h
index 1f1871e23f76..fd104b666836 100644
--- a/include/linux/soc/ti/ti_sci_protocol.h
+++ b/include/linux/soc/ti/ti_sci_protocol.h
@@ -199,6 +199,31 @@ struct ti_sci_clk_ops {
 #define TISCI_MSG_VALUE_IO_ENABLE			1
 #define TISCI_MSG_VALUE_IO_DISABLE			0
 
+/* TISCI LPM constraint state values */
+#define TISCI_MSG_CONSTRAINT_SET			1
+#define TISCI_MSG_CONSTRAINT_CLR			0
+
+/**
+ * struct ti_sci_pm_ops - Low Power Mode (LPM) control operations
+ * @lpm_wake_reason: Get the wake up source that woke the SoC from LPM
+ *		- source: The wake up source that woke soc from LPM.
+ *		- timestamp: Timestamp at which soc woke.
+ * @set_device_constraint: Set LPM constraint on behalf of a device
+ *		- id: Device Identifier
+ *		- state: The desired state of device constraint: set or clear.
+ * @set_latency_constraint: Set LPM resume latency constraint
+ *		- latency: maximum acceptable latency to wake up from low power mode
+ *		- state: The desired state of latency constraint: set or clear.
+ */
+struct ti_sci_pm_ops {
+	int (*lpm_wake_reason)(const struct ti_sci_handle *handle,
+			       u32 *source, u64 *timestamp, u8 *pin, u8 *mode);
+	int (*set_device_constraint)(const struct ti_sci_handle *handle,
+				     u32 id, u8 state);
+	int (*set_latency_constraint)(const struct ti_sci_handle *handle,
+				      u16 latency, u8 state);
+};
+
 /**
  * struct ti_sci_resource_desc - Description of TI SCI resource instance range.
  * @start:	Start index of the first resource range.
@@ -543,6 +568,7 @@ struct ti_sci_ops {
 	struct ti_sci_core_ops core_ops;
 	struct ti_sci_dev_ops dev_ops;
 	struct ti_sci_clk_ops clk_ops;
+	struct ti_sci_pm_ops pm_ops;
 	struct ti_sci_rm_core_ops rm_core_ops;
 	struct ti_sci_rm_irq_ops rm_irq_ops;
 	struct ti_sci_rm_ringacc_ops rm_ring_ops;
-- 
2.45.2

[PATCH v11 4/4] firmware: ti_sci: add CPU latency constraint management

[Markus Schneider-Pargmann]

From: Kevin Hilman <khilman@baylibre.com>

During system-wide suspend, check if any of the CPUs have PM QoS
resume latency constraints set.  If so, set TI SCI constraint.

TI SCI has a single system-wide latency constraint, so use the max of
any of the CPU latencies as the system-wide value.

Note: DM firmware clears all constraints at resume time, so
constraints need to be checked/updated/sent at each system suspend.

Co-developed-by: Vibhore Vardhan <vibhore@ti.com>
Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
Reviewed-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Dhruva Gole <d-gole@ti.com>
Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
---
 drivers/firmware/ti_sci.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index 8482b7077eef..d5d64e5ffbd7 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -10,6 +10,7 @@
 #define pr_fmt(fmt) "%s: " fmt, __func__
 
 #include <linux/bitmap.h>
+#include <linux/cpu.h>
 #include <linux/debugfs.h>
 #include <linux/export.h>
 #include <linux/io.h>
@@ -20,6 +21,7 @@
 #include <linux/of.h>
 #include <linux/of_platform.h>
 #include <linux/platform_device.h>
+#include <linux/pm_qos.h>
 #include <linux/property.h>
 #include <linux/semaphore.h>
 #include <linux/slab.h>
@@ -3669,7 +3671,27 @@ static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
 static int __maybe_unused ti_sci_suspend(struct device *dev)
 {
 	struct ti_sci_info *info = dev_get_drvdata(dev);
-	int ret;
+	struct device *cpu_dev, *cpu_dev_max = NULL;
+	s32 val, cpu_lat = 0;
+	int i, ret;
+
+	if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
+		for_each_possible_cpu(i) {
+			cpu_dev = get_cpu_device(i);
+			val = dev_pm_qos_read_value(cpu_dev, DEV_PM_QOS_RESUME_LATENCY);
+			if (val != PM_QOS_RESUME_LATENCY_NO_CONSTRAINT) {
+				cpu_lat = max(cpu_lat, val);
+				cpu_dev_max = cpu_dev;
+			}
+		}
+		if (cpu_dev_max) {
+			dev_dbg(cpu_dev_max, "%s: sending max CPU latency=%u\n", __func__, cpu_lat);
+			ret = ti_sci_cmd_set_latency_constraint(&info->handle,
+								cpu_lat, TISCI_MSG_CONSTRAINT_SET);
+			if (ret)
+				return ret;
+		}
+	}
 
 	ret = ti_sci_prepare_system_suspend(info);
 	if (ret)
-- 
2.45.2

Re: [PATCH v11 1/4] firmware: ti_sci: Add support for querying the firmware caps

[Dhruva Gole]

On Aug 29, 2024 at 22:16:02 +0200, Markus Schneider-Pargmann wrote:
> From: Georgi Vlaev <g-vlaev@ti.com>
> 
> Add support for the TISCI_MSG_QUERY_FW_CAPS message, used to retrieve
> the firmware capabilities of the currently running system firmware. The
> message belongs to the TISCI general core message API [1] and is
> available in SysFW version 08.04.03 and above. Currently, the message is
> supported on devices with split architecture of the system firmware (DM
> + TIFS) like AM62x. Old revisions or not yet supported platforms will
> NACK this request.
> 
> We're using this message locally in ti_sci.c to get the low power
> features of the FW/SoC. As there's no other kernel consumers yet, this
> is not added to struct ti_sci_core_ops.
> 
> Sysfw version >= 10.00.04 support LPM_DM_MANAGED capability [2], where
> Device Mgr firmware now manages which low power mode is chosen. Going
> forward, this is the default configuration supported for TI AM62 family
> of devices. The state chosen by the DM can be influenced by sending
> constraints using the new LPM constraint APIs.
> 
> [1] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/general/core.html
> [2] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/general/core.html#tisci-msg-query-fw-caps
> 
> Signed-off-by: Georgi Vlaev <g-vlaev@ti.com>
> [vibhore@ti.com: Support for LPM_DM_MANAGED mode]
> Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
> Signed-off-by: Kevin Hilman <khilman@baylibre.com>
> Tested-by: Dhruva Gole <d-gole@ti.com>
> Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
> ---
>  drivers/firmware/ti_sci.c | 74 ++++++++++++++++++++++++++++++++++++++-
>  drivers/firmware/ti_sci.h | 22 ++++++++++++
>  2 files changed, 95 insertions(+), 1 deletion(-)
> 

[..snip..]

No remaining concerns,
Acked-by: Dhruva Gole <d-gole@ti.com>

-- 
Best regards,
Dhruva Gole
Texas Instruments Incorporated

Re: [PATCH v11 3/4] firmware: ti_sci: Introduce Power Management Ops

[Dhruva Gole]

On Aug 29, 2024 at 22:16:04 +0200, Markus Schneider-Pargmann wrote:
> From: Dave Gerlach <d-gerlach@ti.com>
> 
> Introduce power management ops supported by the TISCI
> Low Power Mode API [1].
> 
> 1) TISCI_MSG_LPM_WAKE_REASON
> Get which wake up source woke the SoC from Low Power Mode.
> The wake up source IDs will be common for all K3 platforms.
> 
> 2) TISCI_MSG_LPM_SET_DEVICE_CONSTRAINT
> Set LPM constraint on behalf of a device. By setting a constraint, the
> device ensures that it will not be powered off or reset in the selected
> mode.
> 
> 3) TISCI_MSG_LPM_SET_LATENCY_CONSTRAINT
> Set LPM resume latency constraint. By setting a constraint, the host
> ensures that the resume time from selected mode will be less than the
> constraint value.
> 
> [1] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html
> 
> Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
> [g-vlaev@ti.com: LPM_WAKE_REASON and IO_ISOLATION support]
> Signed-off-by: Georgi Vlaev <g-vlaev@ti.com>
> [a-kaur@ti.com: SET_DEVICE_CONSTRAINT support]
> Signed-off-by: Akashdeep Kaur <a-kaur@ti.com>
> [vibhore@ti.com: SET_LATENCY_CONSTRAINT support]
> Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
> Signed-off-by: Kevin Hilman <khilman@baylibre.com>
> Reviewed-by: Akashdeep Kaur <a-kaur@ti.com>
> Tested-by: Dhruva Gole <d-gole@ti.com>
> Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
> ---
>  drivers/firmware/ti_sci.c              | 198 +++++++++++++++++++++++++
>  drivers/firmware/ti_sci.h              |  76 ++++++++++
>  include/linux/soc/ti/ti_sci_protocol.h |  26 ++++
>  3 files changed, 300 insertions(+)
> 

<removing older email addresses from CC>

LGTM,
Acked-by: Dhruva Gole <d-gole@ti.com>

-- 
Best regards,
Dhruva Gole
Texas Instruments Incorporated

Re: [PATCH v11 0/4] firmware: ti_sci: Introduce system suspend support

[Nishanth Menon]

Hi Markus Schneider-Pargmann,

On Thu, 29 Aug 2024 22:16:01 +0200, Markus Schneider-Pargmann wrote:
> Abstract
> ********
> 
> This series introduces necessary ti_sci driver functionality to support
> various Suspend-to-RAM modes on TI AM62 family of devices. These Low
> Power Modes include Deep Sleep and MCU Only as described in section
> "6.2.4 Power Modes" of the AM62P Technical Reference Manual [0].
> 
> [...]

I have applied the following to branch ti-drivers-soc-next on [1].
Thank you!

[1/4] firmware: ti_sci: Add support for querying the firmware caps
      commit: 371af6a83b580081d2ed76671f2184c5bb52c5b6
[2/4] firmware: ti_sci: Add system suspend and resume call
      commit: 6b48779503a6a080664e917fd91a71bd3f5ae8b5
[3/4] firmware: ti_sci: Introduce Power Management Ops
      commit: 235468957c099707d96e30d91f1885afbc7f4175
[4/4] firmware: ti_sci: add CPU latency constraint management
      commit: 458d22d2e064ded193d697b0629822dc72814933

NOTE: since all other comments are incorporated, I did hand edit patch #1 to
drop the inclusion of redundant linux/dev_printk.h which was the cause of
kernel test robot warning
https://lore.kernel.org/all/202408311026.LluPK1A2-lkp@intel.com/ )
Let me know if you disagree with the change and I can drop the series from
my branch.

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent up the chain during
the next merge window (or sooner if it is a relevant bug fix), however if
problems are discovered then the patch may be dropped or reverted.

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/ti/linux.git
-- 
Regards,
Nishanth Menon
Key (0xDDB5849D1736249D) / Fingerprint: F8A2 8693 54EB 8232 17A3  1A34 DDB5 849D 1736 249D

Re: [PATCH v11 4/4] firmware: ti_sci: add CPU latency constraint management

[Nathan Chancellor]

Hi Markus,

On Thu, Aug 29, 2024 at 10:16:05PM +0200, Markus Schneider-Pargmann wrote:
> From: Kevin Hilman <khilman@baylibre.com>
> 
> During system-wide suspend, check if any of the CPUs have PM QoS
> resume latency constraints set.  If so, set TI SCI constraint.
> 
> TI SCI has a single system-wide latency constraint, so use the max of
> any of the CPU latencies as the system-wide value.
> 
> Note: DM firmware clears all constraints at resume time, so
> constraints need to be checked/updated/sent at each system suspend.
> 
> Co-developed-by: Vibhore Vardhan <vibhore@ti.com>
> Signed-off-by: Vibhore Vardhan <vibhore@ti.com>
> Reviewed-by: Dhruva Gole <d-gole@ti.com>
> Signed-off-by: Dhruva Gole <d-gole@ti.com>
> Signed-off-by: Kevin Hilman <khilman@baylibre.com>
> Tested-by: Dhruva Gole <d-gole@ti.com>
> Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
> ---
>  drivers/firmware/ti_sci.c | 24 +++++++++++++++++++++++-
>  1 file changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
> index 8482b7077eef..d5d64e5ffbd7 100644
> --- a/drivers/firmware/ti_sci.c
> +++ b/drivers/firmware/ti_sci.c
> @@ -10,6 +10,7 @@
>  #define pr_fmt(fmt) "%s: " fmt, __func__
>  
>  #include <linux/bitmap.h>
> +#include <linux/cpu.h>
>  #include <linux/debugfs.h>
>  #include <linux/export.h>
>  #include <linux/io.h>
> @@ -20,6 +21,7 @@
>  #include <linux/of.h>
>  #include <linux/of_platform.h>
>  #include <linux/platform_device.h>
> +#include <linux/pm_qos.h>
>  #include <linux/property.h>
>  #include <linux/semaphore.h>
>  #include <linux/slab.h>
> @@ -3669,7 +3671,27 @@ static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
>  static int __maybe_unused ti_sci_suspend(struct device *dev)
>  {
>  	struct ti_sci_info *info = dev_get_drvdata(dev);
> -	int ret;
> +	struct device *cpu_dev, *cpu_dev_max = NULL;
> +	s32 val, cpu_lat = 0;
> +	int i, ret;
> +
> +	if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
> +		for_each_possible_cpu(i) {
> +			cpu_dev = get_cpu_device(i);
> +			val = dev_pm_qos_read_value(cpu_dev, DEV_PM_QOS_RESUME_LATENCY);

This change is now in -next as commit 458d22d2e064 ("firmware: ti_sci:
add CPU latency constraint management"), where it breaks the build
when this driver is built as a module because dev_pm_qos_read_value() is
not exported to modules:

  ERROR: modpost: "dev_pm_qos_read_value" [drivers/firmware/ti_sci.ko] undefined!

Obviously exporting it would fix the build but sometimes that is
controversial, hence just the report.

> +			if (val != PM_QOS_RESUME_LATENCY_NO_CONSTRAINT) {
> +				cpu_lat = max(cpu_lat, val);
> +				cpu_dev_max = cpu_dev;
> +			}
> +		}
> +		if (cpu_dev_max) {
> +			dev_dbg(cpu_dev_max, "%s: sending max CPU latency=%u\n", __func__, cpu_lat);
> +			ret = ti_sci_cmd_set_latency_constraint(&info->handle,
> +								cpu_lat, TISCI_MSG_CONSTRAINT_SET);
> +			if (ret)
> +				return ret;
> +		}
> +	}
>  
>  	ret = ti_sci_prepare_system_suspend(info);
>  	if (ret)
> -- 
> 2.45.2
> 

Re: [PATCH v11 4/4] firmware: ti_sci: add CPU latency constraint management

[Nishanth Menon]

On 17:54-20240902, Nathan Chancellor wrote:
[...]

> > @@ -3669,7 +3671,27 @@ static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
> >  static int __maybe_unused ti_sci_suspend(struct device *dev)
> >  {
> >  	struct ti_sci_info *info = dev_get_drvdata(dev);
> > -	int ret;
> > +	struct device *cpu_dev, *cpu_dev_max = NULL;
> > +	s32 val, cpu_lat = 0;
> > +	int i, ret;
> > +
> > +	if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
> > +		for_each_possible_cpu(i) {
> > +			cpu_dev = get_cpu_device(i);
> > +			val = dev_pm_qos_read_value(cpu_dev, DEV_PM_QOS_RESUME_LATENCY);
> 
> This change is now in -next as commit 458d22d2e064 ("firmware: ti_sci:
> add CPU latency constraint management"), where it breaks the build
> when this driver is built as a module because dev_pm_qos_read_value() is
> not exported to modules:
> 
>   ERROR: modpost: "dev_pm_qos_read_value" [drivers/firmware/ti_sci.ko] undefined!
> 
> Obviously exporting it would fix the build but sometimes that is
> controversial, hence just the report.

Thank you for the report. I will drop the series from my queue for now.
That should give us some time to sort things out properly for the next
window.

-- 
Regards,
Nishanth Menon
Key (0xDDB5849D1736249D) / Fingerprint: F8A2 8693 54EB 8232 17A3  1A34 DDB5 849D 1736 249D

Re: [PATCH v11 4/4] firmware: ti_sci: add CPU latency constraint management

[Markus Schneider-Pargmann]

Hi,

On Tue, Sep 03, 2024 at 07:45:40AM GMT, Nishanth Menon wrote:
> On 17:54-20240902, Nathan Chancellor wrote:
> [...]
> 
> > > @@ -3669,7 +3671,27 @@ static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
> > >  static int __maybe_unused ti_sci_suspend(struct device *dev)
> > >  {
> > >  	struct ti_sci_info *info = dev_get_drvdata(dev);
> > > -	int ret;
> > > +	struct device *cpu_dev, *cpu_dev_max = NULL;
> > > +	s32 val, cpu_lat = 0;
> > > +	int i, ret;
> > > +
> > > +	if (info->fw_caps & MSG_FLAG_CAPS_LPM_DM_MANAGED) {
> > > +		for_each_possible_cpu(i) {
> > > +			cpu_dev = get_cpu_device(i);
> > > +			val = dev_pm_qos_read_value(cpu_dev, DEV_PM_QOS_RESUME_LATENCY);
> > 
> > This change is now in -next as commit 458d22d2e064 ("firmware: ti_sci:
> > add CPU latency constraint management"), where it breaks the build
> > when this driver is built as a module because dev_pm_qos_read_value() is
> > not exported to modules:
> > 
> >   ERROR: modpost: "dev_pm_qos_read_value" [drivers/firmware/ti_sci.ko] undefined!
> > 
> > Obviously exporting it would fix the build but sometimes that is
> > controversial, hence just the report.
> 
> Thank you for the report. I will drop the series from my queue for now.
> That should give us some time to sort things out properly for the next
> window.

Thanks as well for reporting. I looked into this issue and it looks like
many of the dev_pm_qos_* functions are already exported.
Documentation/power/pm_qos_interface.rst also already lists the function
for kernel-internal use along with many others. So I think adding the
export is the right way and I prepared a patch for that.

Also from what I saw and tested, I was only able to reproduce this issue
with ARCH_KEYSTONE when the TI_SCI_PROTOCOL is selected as module.
multi_v7_defconfig and keystone_defconfig both select it as built-in as
well as ARCH_K3. For all other architectures, TI_SCI_PROTOCOL can not be
selected as it depends on TI_MESSAGE_MANAGER which depends on KEYSTONE
or K3.

Best
Markus