SSH'ing into a.. suspended system?

[Hu]

I think some versions of Windows have the ability to automatically wake the system so that Windows Update can run. Perhaps the system was in S3 when OP walked out the door, woke itself later to run some updates, then went back to sleep before he could get home to see it.

[pjp]

That largely seems to be what is at work. It appears that this originated from Intel's mobile device market push¹ ².

Naming appears to be Modern Standby³ (or InstantGo) and was previously called Connected Standby. Microsoft seems to refer to it as Modern Standby⁴.

ACPI⁵ includes mention of the Device States (d0-d3) and Global / Sleep states (s0-s6). The one being discussed in this thread appears to be s0ix. The Anandtech article² also mentions the sleep states.

The Microsoft link⁴ includes references to "related articles" which may be of interest.

It appears that AMD may support these, but I'm not certain of that, and there may be "usability" problems with "s0ix" in general (not CPU specific). A recent Linux kernel seems to have caused some issues with s3⁶ ⁷.


1. https://www.zdnet.com/article/d3-s0ix-and-other-bay-trail-tricks-intels-using-to-be-a-device-player-again/
2. https://www.anandtech.com/show/6355/intels-haswell-architecture/
3. https://en.wikipedia.org/wiki/InstantGo
4. https://docs.microsoft.com/en-us/windows-hardware/design/device-experiences/modern-standby
5. https://en.wikipedia.org/wiki/Advanced_Configuration_and_Power_Interface
6. https://www.phoronix.com/forums/forum/linux-graphics-x-org-drivers/x-org-drm/1243549-s0ix-support-for-amd-is-broken
7. https://forums.lenovo.com/t5/Other-Linux-Discussions/Linux-S0ix-support-on-AMD-platforms/m-p/5098254

Including for reference, but not sure if they're worth much:
https://software.intel.com/content/www/us/en/develop/documentation/energy-analysis-user-guide/top/energy-analysis-metrics-reference/d0ix-states.html
https://software.intel.com/content/www/us/en/develop/documentation/vtune-help/top/reference/energy-analysis-metrics-reference/s0ix-states.html
_________________
Quis separabit? Quo animo?

[eccerr0r]

Actually I think anything with standard ATX supplies fail the "Modern Standby" requirement because PG itself takes that 500ms. Clearly anything that takes HDDs will fail the requirement too.

So this is only good for custom PCs/tablets that supplies power all the time (and stays quiet so you can't tell it's always powered - no fans so there's a limit to power consumption to begin with) and have no expansion slots (usb excepted)...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[pjp]

At least some laptops support it, but I didn't notice or look for details.

It makes sense why that would be useful for mobile devices. Other factors aside, i can understand the "benefit" in being able to talk to the computer to wake it up so that it can search or perform some kind of task. Less understandable is preventing users from choosing traditional standby and off modes.
_________________
Quis separabit? Quo animo?

[dmpogo]

[Hu]

Yes, that is what I posited above. Windows Update would schedule a timer to wake the system, usually (but not necessarily) in the middle of the night.

[eccerr0r]

This high speed suspend/resume completely nullifies need for S3. However hibernate is still useful for these machines.

For a traditional ACPI machine, Windows already does use the timer when a computer's on battery and S3, it will check the battery occasionally and hibernate if the battery is going low. It may use that time to check for updates depending on battery/AC power status.

As far as I know Linux DEs do not do the battery paranoia check, seems like a good idea.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[pjp]

[eccerr0r]

[Zucca]