2022-11-08 23:06:51

by Georgi Vlaev

[permalink] [raw]
Subject: [PATCH v2 0/5] firmware: ti_sci: Introduce system suspend support

This series introduces necessary ti_sci driver functionality in
preparation of supporting DeepSleep mode for suspend to mem on TI
K3 AM62x. This version is a fixup and rebase of the patch series by
Dave Gerlach [1]. It applies on top of v6.1-rc4.

Deep Sleep mode is described in section "5.2.4.4 DeepSleep" of the
AM62x Technical Reference Manual [2].

The kernel triggers entry to Deep Sleep mode through the mem suspend
transition with the following:

* 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.

* The firmware requires that the OS sends a TISCI_MSG_PREPARE_SLEEP
message in order to provide details about suspend, so we must add the
ability to send this message. We also add TISCI_MSG_LPM_WAKE_REASON
and TISCI_MSG_SET_IO_ISOLATION messages as part of a new PM ops. These
messages are part of the TISCI PM Low Power Mode API [3]. (Patch 2)

* A memory address must be provided to the firmware using the above
message, which is allocated and managed by dma_alloc_coherent()
and friends. (Patch 3)

* System must load firmware to a specific location before Deep Sleep is
entered, and this is accomplished using a memory region in device
tree to indicate where this firmware should be loaded, and also a
"firmware-name" property to indicate the name of the firmware
to load. The ti_sci driver checks in its pm handler to see if
the firmware has been loaded and if not, loads it. (Patch 4)

* Finally, the ti_sci driver must actually send TISCI_MSG_PREPARE_SLEEP
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. (Patch 5)

This is tested on am625-sk using a limited dts with all devices disabled
apart from cpu0, main_uart0, i2c, rtc, mmc/sd, dmsc, and secure_proxy_main.

Testing this sequence requires K3 sdhci suspend/resume support [4],
enable the wkup_rtc in the am625-sk.dts, disable devices that don't
support system suspend/resume like OSPI and CPSW3G.

In can be tested on the following branch:
https://github.com/gvlaev/linux/tree/upstream-v6.2/lpm-ti-sci-v1

Changelog:
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.

[1] https://lore.kernel.org/lkml/[email protected]
[2] https://www.ti.com/lit/pdf/spruiv7
[3] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html
[4] https://lore.kernel.org/lkml/[email protected]

Dave Gerlach (5):
dt-bindings: ti, sci: Add lpm region and firmware-name
firmware: ti_sci: Introduce Power Management Ops
firmware: ti_sci: Allocate memory for the LPM modes
firmware: ti_sci: Use dt provided fw name and address to load at
suspend time
firmware: ti_sci: Introduce prepare system suspend call

.../bindings/arm/keystone/ti,sci.yaml | 21 +-
drivers/firmware/ti_sci.c | 356 ++++++++++++++++++
drivers/firmware/ti_sci.h | 64 +++-
include/linux/soc/ti/ti_sci_protocol.h | 44 +++
4 files changed, 480 insertions(+), 5 deletions(-)


base-commit: f0c4d9fc9cc9462659728d168387191387e903cc
--
2.30.2



2022-11-08 23:07:00

by Georgi Vlaev

[permalink] [raw]
Subject: [PATCH v2 4/5] firmware: ti_sci: Use dt provided fw name and address to load at suspend time

From: Dave Gerlach <[email protected]>

Use request_firmware_direct to load the fs stub LPM firmware to a
provided memory region. The filename for the firmware is provided
in the device tree as "firmware-name".

Use a pm_notifier for loading of the low power firmware during
PM_SUSPEND_PREPARE phase so that it can be loaded from the rootfs
before it is suspended. It is possible in the future for this
firmware to require reload, so add a check to indicate that the
firmware is currently loaded so it is only loaded once.

Signed-off-by: Dave Gerlach <[email protected]>
Signed-off-by: Vibhore Vardhan <[email protected]>
Signed-off-by: Georgi Vlaev <[email protected]>
---
drivers/firmware/ti_sci.c | 97 +++++++++++++++++++++++++++++++++++++++
1 file changed, 97 insertions(+)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index 6a07c7276c24..4986df1b8570 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -12,6 +12,7 @@
#include <linux/debugfs.h>
#include <linux/dma-mapping.h>
#include <linux/export.h>
+#include <linux/firmware.h>
#include <linux/io.h>
#include <linux/iopoll.h>
#include <linux/kernel.h>
@@ -22,6 +23,7 @@
#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/reboot.h>

#include "ti_sci.h"
@@ -89,10 +91,13 @@ struct ti_sci_desc {
* @dev: Device pointer
* @desc: SoC description for this instance
* @nb: Reboot Notifier block
+ * @pm_nb: PM notifier block
* @d: Debugfs file entry
* @debug_region: Memory region where the debug message are available
* @debug_region_size: Debug region size
* @debug_buffer: Buffer allocated to copy debug messages.
+ * @lpm_region: Memory region where the FS Stub LPM Firmware will be stored
+ * @lpm_region_size: LPM region size
* @handle: Instance of TI SCI handle to send to clients.
* @cl: Mailbox Client
* @chan_tx: Transmit mailbox channel
@@ -104,15 +109,20 @@ struct ti_sci_desc {
* @ctx_mem_buf: Low power context memory buffer
* @users: Number of users of this instance
* @is_suspending: Flag set to indicate in suspend path.
+ * @lpm_firmware_loaded: Flag to indicate if LPM firmware has been loaded
+ * @lpm_firmware_name: Name of firmware binary to load from fw search path
*/
struct ti_sci_info {
struct device *dev;
struct notifier_block nb;
+ struct notifier_block pm_nb;
const struct ti_sci_desc *desc;
struct dentry *d;
void __iomem *debug_region;
char *debug_buffer;
size_t debug_region_size;
+ void __iomem *lpm_region;
+ size_t lpm_region_size;
struct ti_sci_handle handle;
struct mbox_client cl;
struct mbox_chan *chan_tx;
@@ -125,11 +135,14 @@ struct ti_sci_info {
/* protected by ti_sci_list_mutex */
int users;
bool is_suspending;
+ bool lpm_firmware_loaded;
+ const char *lpm_firmware_name;
};

#define cl_to_ti_sci_info(c) container_of(c, struct ti_sci_info, cl)
#define handle_to_ti_sci_info(h) container_of(h, struct ti_sci_info, handle)
#define reboot_to_ti_sci_info(n) container_of(n, struct ti_sci_info, nb)
+#define pm_nb_to_ti_sci_info(n) container_of(n, struct ti_sci_info, pm_nb)

#ifdef CONFIG_DEBUG_FS

@@ -3465,6 +3478,32 @@ static int tisci_reboot_handler(struct notifier_block *nb, unsigned long mode,
return NOTIFY_BAD;
}

+static int ti_sci_load_lpm_firmware(struct device *dev, struct ti_sci_info *info)
+{
+ const struct firmware *firmware;
+ int ret = 0;
+
+ /* If no firmware name is set, do not attempt to load. */
+ if (!info->lpm_firmware_name)
+ return -EINVAL;
+
+ ret = request_firmware_direct(&firmware, info->lpm_firmware_name, dev);
+ if (ret) {
+ dev_warn(dev, "Cannot load %s\n", info->lpm_firmware_name);
+ return ret;
+ }
+
+ if (firmware->size > info->lpm_region_size) {
+ release_firmware(firmware);
+ return -ENOMEM;
+ }
+
+ memcpy_toio(info->lpm_region, firmware->data, firmware->size);
+
+ release_firmware(firmware);
+
+ return ret;
+}
static void ti_sci_set_is_suspending(struct ti_sci_info *info, bool is_suspending)
{
info->is_suspending = is_suspending;
@@ -3494,10 +3533,33 @@ static int ti_sci_resume(struct device *dev)

static DEFINE_SIMPLE_DEV_PM_OPS(ti_sci_pm_ops, ti_sci_suspend, ti_sci_resume);

+static int tisci_pm_handler(struct notifier_block *nb, unsigned long pm_event,
+ void *unused)
+{
+ struct ti_sci_info *info = pm_nb_to_ti_sci_info(nb);
+ int ret;
+
+ /* Load the LPM firmware on PM_SUSPEND_PREPARE if not loaded yet */
+ if (pm_event != PM_SUSPEND_PREPARE || info->lpm_firmware_loaded)
+ return NOTIFY_DONE;
+
+ ret = ti_sci_load_lpm_firmware(info->dev, info);
+ if (ret) {
+ dev_err(info->dev, "Failed to LPM firmware, suspend is disabled (%d)\n",
+ ret);
+ return NOTIFY_BAD;
+ }
+
+ info->lpm_firmware_loaded = true;
+
+ return NOTIFY_OK;
+}
+
static int ti_sci_init_suspend(struct platform_device *pdev,
struct ti_sci_info *info)
{
struct device *dev = &pdev->dev;
+ struct resource *res;
int ret;

dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64));
@@ -3521,6 +3583,38 @@ static int ti_sci_init_suspend(struct platform_device *pdev,
if (ret)
goto err;

+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "lpm");
+ if (!res) {
+ dev_warn(dev,
+ "lpm region is required for suspend but not provided.\n");
+ ret = -EINVAL;
+ goto err;
+ }
+
+ info->lpm_region = devm_ioremap_resource(dev, res);
+ if (IS_ERR(info->lpm_region)) {
+ ret = PTR_ERR(info->lpm_region);
+ goto err;
+ }
+ info->lpm_region_size = resource_size(res);
+
+ if (of_property_read_string(dev->of_node, "firmware-name",
+ &info->lpm_firmware_name)) {
+ dev_warn(dev,
+ "firmware-name is required for suspend but not provided.\n");
+ ret = -EINVAL;
+ goto err;
+ }
+
+ info->pm_nb.notifier_call = tisci_pm_handler;
+ info->pm_nb.priority = 128;
+
+ ret = register_pm_notifier(&info->pm_nb);
+ if (ret) {
+ dev_err(dev, "pm_notifier registration fail(%d)\n", ret);
+ goto err;
+ }
+
return 0;
err:
dma_free_coherent(info->dev, LPM_CTX_MEM_SIZE,
@@ -3718,6 +3812,9 @@ static int ti_sci_remove(struct platform_device *pdev)

info = platform_get_drvdata(pdev);

+ if (info->pm_nb.notifier_call)
+ unregister_pm_notifier(&info->pm_nb);
+
if (info->nb.notifier_call)
unregister_restart_handler(&info->nb);

--
2.30.2


2022-11-08 23:08:21

by Georgi Vlaev

[permalink] [raw]
Subject: [PATCH v2 5/5] firmware: ti_sci: Introduce prepare system suspend call

From: Dave Gerlach <[email protected]>

Introduce 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 TI_SCI_MSG_BEGIN_SLEEP with information
about the mode is being entered and the address for allocated memory for
storing the context during Deep Sleep.

Signed-off-by: Dave Gerlach <[email protected]>
Signed-off-by: Vibhore Vardhan <[email protected]>
Signed-off-by: Georgi Vlaev <[email protected]>
---
drivers/firmware/ti_sci.c | 30 ++++++++++++++++++++++++++++++
1 file changed, 30 insertions(+)

diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
index 4986df1b8570..a6d982685aca 100644
--- a/drivers/firmware/ti_sci.c
+++ b/drivers/firmware/ti_sci.c
@@ -3509,9 +3509,39 @@ static void ti_sci_set_is_suspending(struct ti_sci_info *info, bool is_suspendin
info->is_suspending = is_suspending;
}

+static int ti_sci_prepare_system_suspend(struct ti_sci_info *info)
+{
+ int ret = 0;
+ int mode;
+
+ switch (pm_suspend_target_state) {
+ case PM_SUSPEND_MEM:
+ mode = TISCI_MSG_VALUE_SLEEP_MODE_DEEP_SLEEP;
+ break;
+ default:
+ ret = -EINVAL;
+ }
+
+ /*
+ * Do not fail if we don't have action to take for a
+ * specific suspend mode.
+ */
+ if (ret)
+ return 0;
+
+ return ti_sci_cmd_prepare_sleep(&info->handle, mode,
+ (u32)(info->ctx_mem_addr & 0xffffffff),
+ (u32)(info->ctx_mem_addr >> 32), 0);
+}
+
static int 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;
/*
* We must switch operation to polled mode now as drivers and the genpd
* layer may make late TI SCI calls to change clock and device states
--
2.30.2


2022-11-09 15:40:03

by Roger Quadros

[permalink] [raw]
Subject: Re: [PATCH v2 0/5] firmware: ti_sci: Introduce system suspend support



On 09/11/2022 00:45, Georgi Vlaev wrote:
> This series introduces necessary ti_sci driver functionality in
> preparation of supporting DeepSleep mode for suspend to mem on TI
> K3 AM62x. This version is a fixup and rebase of the patch series by
> Dave Gerlach [1]. It applies on top of v6.1-rc4.
>
> Deep Sleep mode is described in section "5.2.4.4 DeepSleep" of the
> AM62x Technical Reference Manual [2].
>
> The kernel triggers entry to Deep Sleep mode through the mem suspend
> transition with the following:
>
> * 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.
>
> * The firmware requires that the OS sends a TISCI_MSG_PREPARE_SLEEP
> message in order to provide details about suspend, so we must add the
> ability to send this message. We also add TISCI_MSG_LPM_WAKE_REASON
> and TISCI_MSG_SET_IO_ISOLATION messages as part of a new PM ops. These
> messages are part of the TISCI PM Low Power Mode API [3]. (Patch 2)
>
> * A memory address must be provided to the firmware using the above
> message, which is allocated and managed by dma_alloc_coherent()
> and friends. (Patch 3)
>
> * System must load firmware to a specific location before Deep Sleep is
> entered, and this is accomplished using a memory region in device
> tree to indicate where this firmware should be loaded, and also a
> "firmware-name" property to indicate the name of the firmware
> to load. The ti_sci driver checks in its pm handler to see if
> the firmware has been loaded and if not, loads it. (Patch 4)
>
> * Finally, the ti_sci driver must actually send TISCI_MSG_PREPARE_SLEEP
> 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. (Patch 5)
>
> This is tested on am625-sk using a limited dts with all devices disabled
> apart from cpu0, main_uart0, i2c, rtc, mmc/sd, dmsc, and secure_proxy_main.
>
> Testing this sequence requires K3 sdhci suspend/resume support [4],
> enable the wkup_rtc in the am625-sk.dts, disable devices that don't
> support system suspend/resume like OSPI and CPSW3G.
>
> In can be tested on the following branch:
> https://github.com/gvlaev/linux/tree/upstream-v6.2/lpm-ti-sci-v1
>
> Changelog:
> 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.
>
> [1] https://lore.kernel.org/lkml/[email protected]
> [2] https://www.ti.com/lit/pdf/spruiv7
> [3] https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html
> [4] https://lore.kernel.org/lkml/[email protected]
>
> Dave Gerlach (5):
> dt-bindings: ti, sci: Add lpm region and firmware-name
> firmware: ti_sci: Introduce Power Management Ops
> firmware: ti_sci: Allocate memory for the LPM modes
> firmware: ti_sci: Use dt provided fw name and address to load at
> suspend time
> firmware: ti_sci: Introduce prepare system suspend call
>
> .../bindings/arm/keystone/ti,sci.yaml | 21 +-
> drivers/firmware/ti_sci.c | 356 ++++++++++++++++++
> drivers/firmware/ti_sci.h | 64 +++-
> include/linux/soc/ti/ti_sci_protocol.h | 44 +++
> 4 files changed, 480 insertions(+), 5 deletions(-)
>
>
> base-commit: f0c4d9fc9cc9462659728d168387191387e903cc

Tested-by: Roger Quadros <[email protected]>

Tested this with am65-CPSW suspend/resume patches [1]
[1] - https://lore.kernel.org/netdev/[email protected]/T/

--