Immediate Shutdown/Reboot with flash modded iPod Classic

[iPodVT]

This morning I installed the current dev build - f3d6196016-241021 - onto a flash modded iPod Classic and was surprised to find that there is no longer a ~20 second delay between when I have the iPod shutdown and when it actually powers off.  It is now so immediate that I don't even see the "shutting down" message displayed on the screen, and the reboot command now works.  I don't recall seeing any mention of that change on the "Recent activity - code" webpage.  I assume that the power management concerns about this have been resolved, and I'm wondering how was it accomplished?

Thanks.

[speachy]

You didn't mention what version you were coming _from_, but I suspect what you are seeing is due to e0df9952fd13904da646adae0409b5c76c67f298 from about a month ago.

Another recent change is was shrinking the ipod6g's storage idle->shutdown timeout from 20s to 7s, matching the other HDD-based targets.

[iPodVT]

Thanks for your reply.

I had updated from 607adf00b5-240915.

I did see the code update note re the reduction of the ipod6g's storage idle->shutdown timeout from 20s to 7s, but I didn't associate that with the instantaneous shutdown I was seeing.  In any case, I'm just curious if something is being handled differently (and what that is, if it's not complicated or time consuming to explain) or if it was decided that the delay was unnecessary or maybe ineffective.

[speachy]

I had updated from 607adf00b5-240915.

What you are seeing is a side effect of e0df9952fd13904da646adae0409b5c76c67f298 landed four days later.

When shutting down, waited for the drive to be "spun down" which isn't a thing when the drive doesn't support power management. That commit switched how we treat those drives to "not spinning" instead of "always spinning".

I've been working on the 6g's ATA driver to bring it functionally/behaviourally in line with the standard one, but this is a quirk now shared by both.  I need to look into this more carefully; to make sure we don't inadvertently trigger data loss when shutting down.

[Frankenpod]

Does this situation have any implications for battery life?  That is, does the 6g now actually power-down when shut down, or does the drive continue to drain battery power even when the device is turned off?

  I have the impression that the flash adaptors' problems with power-management commands maybe mean they never fully turn off (so batteries will drain while the things are out-of-use)?  Or is it only an issue with respect to the drive powering down between accesses while the device itself is in use?

[As I have several flash-modded ipods, and some of them sit unused for longish periods, I'm a bit bothered at the implications for battery life...though I guess the fault lies with the adaptors themselves rather than being anything RB can do much about]

[speachy]

Does this situation have any implications for battery life?  That is, does the 6g now actually power-down when shut down, or does the drive continue to drain battery power even when the device is turned off?

These two changes have no implications for battery live beyond that 20s of additional runtime. 

  I have the impression that the flash adaptors' problems with power-management commands maybe mean they never fully turn off (so batteries will drain while the things are out-of-use)?  Or is it only an issue with respect to the drive powering down between accesses while the device itself is in use?

The latter.    When the device is "turned off" all power supplies and regulators are switched off, this includes the storage device.

[As I have several flash-modded ipods, and some of them sit unused for longish periods, I'm a bit bothered at the implications for battery life...though I guess the fault likes with the adaptors themselves rather than being anything RB can do much about]

The problems are due to not properly implementing the ATA power management commands (first defined by ATA-3 in 1997, mandatory since ATA-4 in 1998)  -- When issued the STANDBY IMMEDIATE command to put the device to sleep (which includes an implicit flush operation) it no longer responds to _anything_, requiring a full power cycle to get it going again.  They also claim to not support the FLUSH CACHE command (introduced in ATA-4, mandatory since ATA-5 in 2000, capability flag added in ATA-6 in 2002.  These devices claim to be ATA-8 devices) so we can't ever be sure when it's safe to kill power.

It also turns out these SD adapters don't properly implement LBA48 (ATA-6) -- they claim to, but truncate sector addresses to only 32 bits which limits total capacity to 2TiB.

tl;dr:  Use a SATA adapter intstead of an SD adapter.  They are fast, reliable, and implement proper power management.

[Frankenpod]

Does this situation have any implications for battery life?  That is, does the 6g now actually power-down when shut down, or does the drive continue to drain battery power even when the device is turned off?

These two changes have no implications for battery live beyond that 20s of additional runtime. 

  I have the impression that the flash adaptors' problems with power-management commands maybe mean they never fully turn off (so batteries will drain while the things are out-of-use)?  Or is it only an issue with respect to the drive powering down between accesses while the device itself is in use?

The latter.    When the device is "turned off" all power supplies and regulators are switched off, this includes the storage device.

[As I have several flash-modded ipods, and some of them sit unused for longish periods, I'm a bit bothered at the implications for battery life...though I guess the fault likes with the adaptors themselves rather than being anything RB can do much about]

The problems are due to not properly implementing the ATA power management commands (first defined by ATA-3 in 1997, mandatory since ATA-4 in 1998)  -- When issued the STANDBY IMMEDIATE command to put the device to sleep (which includes an implicit flush operation) it no longer responds to _anything_, requiring a full power cycle to get it going again.  They also claim to not support the FLUSH CACHE command (introduced in ATA-4, mandatory since ATA-5 in 2000, capability flag added in ATA-6 in 2002.  These devices claim to be ATA-8 devices) so we can't ever be sure when it's safe to kill power.

It also turns out these SD adapters don't properly implement LBA48 (ATA-6) -- they claim to, but truncate sector addresses to only 32 bits which limits total capacity to 2TiB.

tl;dr:  Use a SATA adapter intstead of an SD adapter.  They are fast, reliable, and implement proper power management.

Thank you - that pretty-much addresses everything I was wondering about.  Quite annoying that the devices don't fully follow the spec, but as long as it doesn't mean the battery being drained dry when the things aren't in use at all, I can live with it.