You are not logged in.

#26 2021-06-04 04:14:05

Gadgethm
Member
Registered: 2020-12-29
Posts: 8

Re: HP Envy x360 no S states available

Pyrrho wrote:

Have you managed to solve this problem?

Unfortunately, I gave up on S3 with discovering this Gitlab issue.  I was able to get s2idle to work on my machine (15z-ee000) with a 5.12 kernel and the six patches applied mentioned in the issue.  That was actually a while ago now and I've found I get more consistent and reliable suspends using the 5.13rc1 kernel instead.  Next on my list: try it on 5.13rc4!

S2idle actually works pretty well on this laptop and has much better power savings than before.  From what I hear from the above gitlab issue, most of the patches I applied will be mainlined in 5.14, so the good news is that our Renoir laptops will have working modern suspend pretty soon!

From what I can tell though, the power savings are still not as good as a true S3 sleep would be...

nasaiya wrote:

I then of course was UNABLE to compile the DSDT.dsl due to compiler errors that were not related to anything I had changed... I eventually decided to just remove the offending lines to resolve the errors... I don't know if they were necessary or not but nothing has exploded yet so I think it's okay! ymmv... Here is a diff showing the lines I removed:

...which means it may be time for me to try this again, especially with your findings nasaiya!  I'll try to give this a shot sometime this week and report back.

Offline

#27 2021-06-04 04:28:50

nasaiya
Member
Registered: 2021-05-28
Posts: 4

Re: HP Envy x360 no S states available

Gadgethm wrote:

...which means it may be time for me to try this again, especially with your findings nasaiya!  I'll try to give this a shot sometime this week and report back.

I can report that it's been working great for me so far since that last post! Both S3 and S4. The only bug was the screen orientation got flipped upside down after resume twice early on (I'm using kde/plasma), but it hasn't done that again after many successful suspend/resume cycles.

I'm not using any special kernel just the current one pacman installs 5.12.6-arch1-1 (I'm still an arch newbie it's all default config). Can't confirm whether s2idle actually works other than it draws the battery down rapidly so it's effectively useless... S3 works with very little battery drain and it appears to drop down to S4 if I leave it long enough...

Good luck!

Offline

#28 2021-06-04 15:27:16

Gadgethm
Member
Registered: 2020-12-29
Posts: 8

Re: HP Envy x360 no S states available

nasaiya wrote:

The only bug was the screen orientation got flipped upside down after resume twice early on (I'm using kde/plasma), but it hasn't done that again after many successful suspend/resume cycles.

Oh wow! I've seen this too with s2idle.  Up until the issue first came up (about a week ago), I had never seen monitor-sensor report any orientation other than 'undefined', but now it reports 'bottom-up' when the issue occurs.  Rotating the laptop doesn't result in any change in monitor-sensor, not does it flip the screen.  Instead I have to manually set it back to the normal orientation.  This has happened twice now out of about 4-5 suspend/resume cycles.  I'm using Gnome.

Total shot in the dark here, but I suspect that there's some sort of edge case during resume (specifically powering up the sensor fusion hub) which causes it to send a message that's being interpreted as an orientation.  Super cool since I haven't been able to get my accelerometer to work yet, maybe there's hope?

Offline

#29 2021-06-07 13:34:37

nasaiya
Member
Registered: 2021-05-28
Posts: 4

Re: HP Envy x360 no S states available

Gadgethm wrote:

Total shot in the dark here, but I suspect that there's some sort of edge case during resume (specifically powering up the sensor fusion hub) which causes it to send a message that's being interpreted as an orientation.  Super cool since I haven't been able to get my accelerometer to work yet, maybe there's hope?

That's an interesting thought... I also haven't gotten that working yet... I don't recall exactly where but I believe I found some kernel dev talk about issues with the sensor fusion hub hopefully being resolved in a future kernel... That bug does seem to indicate hope though!

If you get it to work, I also noticed that while I CAN get it to resume from touching the keyboard or touchpad, usually when I resume that way the touchpad stops working... I have to close the lid to suspend again and resume by hitting the power button -- resuming that way seems to always be successful.

I'm guessing there are probably some more minor tweaks needed in the acpi tables, or perhaps some udev rules to handle bringing up the devices more cleanly on resume, but that's a bit above my knowledge level to figure out...

Offline

#30 2021-06-07 14:34:31

Gadgethm
Member
Registered: 2020-12-29
Posts: 8

Re: HP Envy x360 no S states available

nasaiya wrote:

If you get it to work, I also noticed that while I CAN get it to resume from touching the keyboard or touchpad, usually when I resume that way the touchpad stops working... I have to close the lid to suspend again and resume by hitting the power button -- resuming that way seems to always be successful.

I think I saw this before.  Try removing and then modprobing the i2c-hid module when it happens and see if that gets the touchpad to work again:

rmmod i2c_hid
modprobe i2c_hid

This might be a sequence thing?  Like the module is being loaded before the touchpad is powered up fully or something? I remember reading about it ages ago but I don't remember the details unfortunately.

Last edited by Gadgethm (2021-06-07 14:35:36)

Offline

#31 2021-06-09 21:29:49

Gadgethm
Member
Registered: 2020-12-29
Posts: 8

Re: HP Envy x360 no S states available

Gadgethm wrote:
nasaiya wrote:

I then of course was UNABLE to compile the DSDT.dsl due to compiler errors that were not related to anything I had changed... I eventually decided to just remove the offending lines to resolve the errors... I don't know if they were necessary or not but nothing has exploded yet so I think it's okay! ymmv... Here is a diff showing the lines I removed:

...which means it may be time for me to try this again, especially with your findings nasaiya!  I'll try to give this a shot sometime this week and report back.

Gave this a shot, still no luck on my 15z-ee000...  I used 5.12.9 for this test.  Think I'll stick with Modern Suspend for now, since that works significantly better for me

Offline

#32 2021-06-10 13:04:24

nasaiya
Member
Registered: 2021-05-28
Posts: 4

Re: HP Envy x360 no S states available

Gadgethm wrote:

Gave this a shot, still no luck on my 15z-ee000...  I used 5.12.9 for this test.  Think I'll stick with Modern Suspend for now, since that works significantly better for me

I'm guessing you probably know this, but just in case you don't... I found on mine it would not work unless I got the aml files to disassemble without ANY errors -- that means making sure to add the required external includes with the iasl command... It will produce an output file (presumably an incomplete one) even if you don't do that, but it didn't work for me until I got the includes figured out... It was especially confusing because multiple files seem to provide the same external symbols, but only the right combination of files discovered through trial and error -- a lot of trial and error -- allowed it to disassemble without error. Including all files did not work either that just created redefinition conflicts...

Offline

#33 2021-06-10 16:25:30

Gadgethm
Member
Registered: 2020-12-29
Posts: 8

Re: HP Envy x360 no S states available

nasaiya wrote:

I'm guessing you probably know this, but just in case you don't... I found on mine it would not work unless I got the aml files to disassemble without ANY errors -- that means making sure to add the required external includes with the iasl command... It will produce an output file (presumably an incomplete one) even if you don't do that, but it didn't work for me until I got the includes figured out... It was especially confusing because multiple files seem to provide the same external symbols, but only the right combination of files discovered through trial and error -- a lot of trial and error -- allowed it to disassemble without error. Including all files did not work either that just created redefinition conflicts...

Might try again tonight, but I don't remember seeing any compilation errors... It just wouldn't resume when suspending using S3.  Good tip though!

Offline

Board footer

Powered by FluxBB