SK-AM62P-LP: Is Deep Sleep actually working?

Ivan Simonavicius

Part Number: SK-AM62P-LP
Other Parts Discussed in Thread: AM62P, AM625

Hi!,

I am working with the EVM board, and I'm trying to use the Deep Sleep mode.

I've followed this step: https://software-dl.ti.com/processor-sdk-android/esd/AM62X/09_01_00/docs/devices/AM62X/android/Application_Notes_Android_Low_Power.html
I've followed these stepts: https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/09_01_00_08/exports/docs/linux/Foundational_Components/Kernel/Kernel_Drivers/Power_Management/pm_low_power_modes.html
I've followed these stepts: https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/09_01_00_08/exports/docs/linux/Foundational_Components/Kernel/Kernel_Drivers/Power_Management/pm_wakeup_sources.html

These are the logs I get:

console:/ # rtcwake -s 10 -m mem
wakeup using "mem" from /dev/rtc0 at Thu Feb  1 20:57:03 2024
[  121.525624][ T2623] PM: suspend entry (deep)
[  121.533531][ T2623] Filesystems sync: 0.003 seconds
[  121.556552][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ctrl
[  121.560118][ T2623] remoteproc remoteproc0: stopped remote processor 79000000.r5f
[  121.564800][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ns
[  121.572372][ T2623] Freezing user space processes
[  121.579586][  T125] ueventd: LoadWithAliases was unable to load rpmsg:ti.ipc4.ping-pong
[  121.594747][ T2623] Freezing user space processes completed (elapsed 0.011 seconds)
[  121.602453][ T2623] OOM killer disabled.
[  121.606377][ T2623] Freezing remaining freezable tasks
[  121.615203][ T2623] Freezing remaining freezable tasks completed (elapsed 0.003 seconds)
[  121.623333][ T2623] printk: Suspending console(s) (use no_console_suspend to debug)
[  121.650442][ T2623] omap8250 2800000.serial: PM domain pd:146 will not be powered off
[  121.661408][   T76] android_work: sent uevent USB_STATE=DISCONNECTED
[  121.663157][ T2623] PM: suspend devices took 0.036 seconds
[  121.668897][ T2623] Disabling non-boot CPUs ...
[  121.671297][ T2623] psci: CPU1 killed (polled 0 ms)
[  121.677286][ T2623] psci: CPU2 killed (polled 0 ms)
[  121.686031][ T2623] psci: CPU3 killed (polled 0 ms)
[  121.691297][    C0] PM: pm_system_irq_wakeup: 244 triggered 2b1f0000.rtc
[  121.691378][ T2623] Enabling non-boot CPUs ...
[  121.691792][    T0] Detected VIPT I-cache on CPU1
[  121.691862][    T0] GICv3: CPU1: found redistributor 1 region 0:0x00000000018a0000
[  121.691928][    T0] CPU1: Booted secondary processor 0x0000000001 [0x410fd034]
[  121.693319][ T2623] CPU1 is up
[  121.693648][    T0] Detected VIPT I-cache on CPU2
[  121.693709][    T0] GICv3: CPU2: found redistributor 2 region 0:0x00000000018c0000
[  121.693774][    T0] CPU2: Booted secondary processor 0x0000000002 [0x410fd034]
[  121.695465][ T2623] CPU2 is up
[  121.695824][    T0] Detected VIPT I-cache on CPU3
[  121.695890][    T0] GICv3: CPU3: found redistributor 3 region 0:0x00000000018e0000
[  121.695961][    T0] CPU3: Booted secondary processor 0x0000000003 [0x410fd034]
[  121.698003][ T2623] CPU3 is up
[  121.698568][ T2623] ti-sci 44043000.system-controller: ti_sci_resume: wakeup source: 0x50
[  122.195324][ T2630] usb 1-1: reset low-speed USB device number 2 using xhci-hcd
[  122.229331][    T6] android_work: sent uevent USB_STATE=CONNECTED
[  122.335423][    T6] android_work: sent uevent USB_STATE=CONFIGURED
[  122.502231][ T2623] am65-cpsw-nuss 8000000.ethernet: set new flow-id-base 19
[  122.508045][ T2623] am65-cpsw-nuss 8000000.ethernet eth0: PHY [8000f00.mdio:00] driver [Generic PHY] (irq=POLL)
[  122.508073][ T2623] am65-cpsw-nuss 8000000.ethernet eth0: configuring for phy/rgmii-rxid link mode
[  122.512436][ T2623] am65-cpsw-nuss 8000000.ethernet eth1: PHY [8000f00.mdio:01] driver [Generic PHY] (irq=POLL)
[  122.512451][ T2623] am65-cpsw-nuss 8000000.ethernet eth1: configuring for phy/rgmii-rxid link mode
[  122.514619][ T2623] PM: resume devices took 0.728 seconds
[  122.710407][ T2623] OOM killer enabled.
[  122.714243][ T2623] Restarting tasks ...
[  122.717442][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_chrdev
[  122.723363][ T2623] done.
[  122.731311][  T125] ueventd: LoadWithAliases was unable to load virtio:d00000007v00000000
[  122.732269][ T2623] random: crng reseeded on system resumption
[  122.745035][  T125] ueventd: LoadWithAliases was unable to load platform:rproc-virtio
[  122.763299][ T2623] Resume caused by IRQ 244, 2b1f0000.rtc
[  122.769088][ T2623] remoteproc remoteproc0: powering up 79000000.r5f
[  122.776370][ T2623] remoteproc remoteproc0: Booting fw image am62p-mcu-r5f0_0-fw, size 55272
[  122.780118][  T125] ueventd: LoadWithAliases was unable to load of:Nethernet-phyT(null)
[  122.786635][ T2623] rproc-virtio rproc-virtio.4.auto: assigned reserved memory node mcu-r5fss-dma-memory-region@9b800000
[  122.797169][  T125] ueventd: LoadWithAliases was unable to load of:Nethernet-phyT(null)
[  122.801584][  T592] dwc3 31000000.usb: request 0000000000000000 was not queued to ep0out
[  122.801818][   T50] android_work: sent uevent USB_STATE=DISCONNECTED
[  122.804668][ T2623] virtio_rpmsg_bus virtio1: rpmsg host is online
[  122.812155][ T2640] virtio_rpmsg_bus virtio1: creating channel ti.ipc4.ping-pong addr 0xd
[  122.813285][  T607] read descriptors
[  122.813315][  T607] read strings
[  122.818606][  T125] ueventd: LoadWithAliases was unable to load platform:rproc-virtio
console:/ # [  122.820151][ T2623] rproc-virtio rproc-virtio.4.auto: registered virtio1 (type 7)
[  122.822082][  T125] ueventd: LoadWithAliases was unable to load platform:rproc-virtio
[  122.826960][ T2640] virtio_rpmsg_bus virtio1: creating channel rpmsg_chrdev addr 0xe
[  122.829855][  T125] ueventd: LoadWithAliases was unable to load virtio:d00000007v00000000
[  122.832682][ T2623] remoteproc remoteproc0: remote processor 79000000.r5f is now up
[  122.832746][ T2623] PM: suspend exit
[  122.833163][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ctrl
[  122.836212][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ns
[  122.839110][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ns
[  122.842045][  T125] ueventd: LoadWithAliases was unable to load virtio:d00000007v00000000
[  122.932627][  T125] ueventd: LoadWithAliases was unable to load rpmsg:ti.ipc4.ping-pong
[  122.943994][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_chrdev
[  122.958146][  T125] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_chrdev
[  123.851375][  T607] read descriptors
[  123.855081][  T607] read strings
[  124.797708][ T2640] android_work: sent uevent USB_STATE=CONNECTED
[  124.904231][ T2640] android_work: sent uevent USB_STATE=CONFIGURED

I can see that the screen and the peripherals such as a mouse, are disabled.

But, in: https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/09_01_00_08/exports/docs/linux/Foundational_Components/Kernel/Kernel_Drivers/Power_Management/pm_low_power_modes.html says:

"This partially indicates that linux has finished it’s deep sleep sequence. For further confirmation, one can take a look at the TP25/PMIC_LPM_EN pin on the EVM (after programming the PMCTRL_SYS register (0x43018080) to 0x15). Here, if the pin is 3.3V when active and 0V when in deep sleep."

The test point according to the schematic is: TP46
The value es 3v3 and never changes.

Also, If I reed the register:

am62p:/ # devmem 0x43018080
0x00000010

In this question: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1310915/am625-device-not-able-to-wakeup-from-deep-sleep Anush says: "It should be done by default, but if not, a Linux Command you can run is "devmem2 0x43018080" which will read the register and "devmem2 0x43018080 w 0x15" to write to the register."

But if I write 0x15 to the register and send the device to DP, it reboots, so It does not work.

Q1: Is Deep Sleep actually working and turning things off? How could I know?

Q2: Where is set the 0x10 of the register? Should It be 0x10? (It in the Table 14-7826, page 7071 spruj83)

Thank for your response.

over 1 year ago

0 Anshu Madwesh over 1 year ago

TI__Mastermind 18530 points

Hello,

I'm not familiar with the differences between Deep Sleep on Android SDK versus Linux SDK. Please allow a few days for a response.

Thanks,
Anshu

0 Guillaume La Roque over 1 year ago

TI__Intellectual 2545 points

Hello,

you give link to am62X but you message was on AM62P ,can you clarify ?

not sure devmem command on am625 should works on am62P.
i will check.

i will give you more complete feedback monday.

Guillaume

0 Ivan Simonavicius over 1 year ago in reply to Guillaume La Roque

Prodigy 50 points

Sorry for the delay, I was out of the office these days.

The link provided is for am62x, but the link for am62p says exactly the same step for the Android SDK.
software-dl.ti.com/.../Application_Notes_Android_Low_Power.html

Regarding am625, I took that forum as a reference since it was the only one I found debating a similar topic.

Thank you,
Ivan.

0 Mattijs Korpershoek over 1 year ago in reply to Ivan Simonavicius

TI__Intellectual 1635 points

Hey Ivan,

> Q1: Is Deep Sleep actually working and turning things off? How could I know?

For testing, we mostly use the `rtcwake` command as documented in the application note:

https://software-dl.ti.com/processor-sdk-android/esd/AM62PX/09_01_00/docs/devices/AM62PX/android/Application_Notes_Android_Low_Power.html

I can confirm that the above commands work fine on 09.01.00.
We have not tried to program PMCTRL_SYS register (0x43018080) to 0x15.
We will try to do so on release 09.01.00 and see if we can reproduce your issue.

> Q2: Where is set the 0x10 of the register? Should It be 0x10? (It in the Table 14-7826, page 7071 spruj83)

Reading spruj83, we can see that 0x10 is the *default* value upon bootup.

In fact, only bit 5 defaults to 1, which gives us 0x10.
So, to answer the question, the 0x10 is not set, because it's the default bootup value for register 0x43018080

0 Mattijs Korpershoek over 1 year ago in reply to Mattijs Korpershoek

TI__Intellectual 1635 points

Hi Ivan,

I confirm that we can reproduce your issue on SDK 09.01.00 when setting PMCTRL_SYS to 0x15:

console:/ # devmem 0x43018080
0x00000010
console:/ # devmem 0x43018080 4 0x15                                         
# devmem 0x43018080                                                
0x00000015

After, attempting to suspend with:

console:/ # rtcwake -m mem -s 10                                               
wakeup using "mem" from /dev/rtc0 at Wed Mar 13 11:27:24 2024
[  210.249419][ T2151] PM: suspend entry (deep)
[  210.255018][ T2151] Filesystems sync: 0.000 seconds
[  210.261429][  T126] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ctrl
[  210.264894][ T2151] remoteproc remoteproc0: stopped remote processor 79000000.r5f
[  210.269753][  T126] ueventd: LoadWithAliases was unable to load rpmsg:rpmsg_ns
[  210.277214][ T2151] Freezing user space processes
[  210.284388][  T126] ueventd: LoadWithAliases was unable to load rpmsg:ti.ipc4.ping-pong
[  210.297385][ T2151] Freezing user space processes completed (elapsed 0.009 seconds)
[  210.305115][ T2151] OOM killer disabled.
[  210.309077][ T2151] Freezing remaining freezable tasks
[  212.139001][ T2151] Freezing remaining freezable tasks completed (elapsed 1.824 seconds)
[  212.147124][ T2151] printk: Suspending console(s) (use no_console_suspend to debug)

[...] // reboot happened

U-Boot SPL 2023.04 (Dec 07 2023 - 14:32:20 +0100)
SYSFW ABI: 3.1 (firmware rev 0x0009 '9.1.8--v09.01.08 (Kool Koala)')
SPL initial stack usage: 17056 bytes

0 Ivan Simonavicius over 1 year ago in reply to Mattijs Korpershoek

Prodigy 50 points

Hi Mattijs,
Assuming that the board is indeed going into DeepSleep, using rtcwake we were able to measure the current consumption through a DC source. The consumption is approximately 1 A and when we go to DeepSleep it drops to approximately 200 mA.
The evm has several extra things that remain on, but is it possible to lower consumption further?
We need to determine if we could bring them to a consumption of 1 mA, which is what is required in Automotive projects.
Thank you. Greetings.

Ivan.

0 Owen Xiao over 1 year ago in reply to Ivan Simonavicius

Prodigy 20 points

Hi Ivan,

May I know if you got the answer to this point or not? I'm also interesting in the consumption at the DeepSleep mode.

Cheers,

Owen.

0 Mattijs Korpershoek over 1 year ago in reply to Owen Xiao

TI__Intellectual 1635 points

Sorry for the delay.

Checking with the BSP team.

0 Anshu Madwesh over 1 year ago in reply to Mattijs Korpershoek

TI__Mastermind 18530 points

Hi,

I will start working on this. Please allow some time for me to catch up and respond.

Thanks,

Anshu

0 Anshu Madwesh over 1 year ago in reply to Anshu Madwesh

TI__Mastermind 18530 points

Hi Ivan,

Ivan Simonavicius said:
is it possible to lower consumption further?

I recommend that you removed device tree nodes of peripherals that you do not need.

See this thread: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1301570/sk-am62a-lp-power-consumption-by-turning-off-mcu-domain/4950661#4950661

Ivan Simonavicius said:
bring them to a consumption of 1 mA

Even using LPDDR4 and reducing leakage current, 1mA is out of reach. The lowest Deep Sleep for Linux is 14.62mW / 12.5mA.

Best Regards,
Anshu

0 Ivan Simonavicius over 1 year ago in reply to Anshu Madwesh

Prodigy 50 points

Hi Anshu, I hope this message finds you well.

Is this consumption measured at a voltage of 1.1 V?
For our specific needs, we require maintaining a current of 1 mA at 14V.
Calculating the power consumption as 14.62 mW at 14 V results in approximately 1.04 mA.
This leaves us with very little margin. Is this calculation correct?
Why is there a specific mention of LPDDR4? If we were to use a different memory type/mode or configuration, could this potentially improve our situation?

Thank you for your time and assistance.

0 Anshu Madwesh over 1 year ago in reply to Ivan Simonavicius

TI__Mastermind 18530 points

Hi Ivan,

Ivan Simonavicius said:
Is this consumption measured at a voltage of 1.1 V?

Can you clarify what this means? The voltage supply for LPDDR?

Ivan Simonavicius said:
we require maintaining a current of 1 mA at 14V

Can you also clarify 14V?

Ivan Simonavicius said:
Why is there a specific mention of LPDDR4?

LPDDR has less power consumption than DDR.

Thanks,

Anshu

Processors

Processors forum

SK-AM62P-LP: Is Deep Sleep actually working?