Intel Has Been Working To Improve Linux Suspend/Resume, Calls For More Testing

[phoronix]

Phoronix: Intel Has Been Working To Improve Linux Suspend/Resume, Calls For More Testing

With Linux suspend/resume support still sometimes being problematic, it's great to here Intel's Open-Source Technology Center has a team working on continuing to improve the Linux support for this power-saving functionality...

http://www.phoronix.com/scan.php?page=news_item&px=Linux-Suspend-Resume-Intel

[sheldonl]

s/here/hear/

[Danniello]

For me suspend/resume works quite well (X11 crash happens but not very often), but there are strange side effects. Some of them are... good actually.

Fedora 27 on Intel Z170 with i7-6700K + GeForce GTX 1070 (connected via DisplayPort with monitor)
Fresh system boot (before first suspend):
*[BAD] Accelerated Firefox (layers.acceleration.force-enabled;true) not working correctly - full screen video extreme stuttering - in fact Firefox is not usable
*[OK] Software turn off/on monitor working OK (for example after Win+L) - monitor start working normally after key press or mouse move

After resume:
*[OK!] Accelerated Firefox working OK - no full screen issues at all!
*[BAD] Software turn off/on monitor working BAD (for example after Win+L). I mean turn off monitor works as expected, but turn on via click something on keyboard or move mouse causes only turn on monitor backlight (with message "No DisplayPort signal"). I need change monitor input to HDMI and back to DisplayPort - after this "reinitialisation" screen will start working correctly again (until next software turn off monitor).

I know that probably it has nothing to do with Intel suspend/resume improvements - probably it is related with closed source nVidia driver bugs and/or Gnome3. Anyway I'm not sure if I want this strange behaviour to be fixed - at least thanks to this nVidia driver resume issue I could use Firefox with acceleration (without disabling Gnome3 acceleration via lg -> Meta.disable_unredirect_for_screen(global.screen)) .

[macemoneta]

My Chromebook is Intel-based and obviously runs Linux. It suspends / resumes perfectly, and faster than I can raise the screen. I usually use it two-headed at my desk, with an external monitor connected via USB-C to HDMI adapter. It also works great 99% of the time. Occassionally, the external monitor spontaneously goes to sleep, but can be rapidly restored by ctrl-alt-f2 (forward), ctrl-alt-f1 (backward).

Someone has obviously spent time correcting issues with Intel suspend/resume (Google?). Why haven't those patches made it upstream?

[dkasak]

My experience with suspending and hibernating is that it works best with the least Intel devices possible. I used to have no issues hibernating a desktop back 10 years ago, with a Radeon GPU and other assorted non-Intel hardware. Before that, I was even able to hibernate a desktop with an nVidia GPU. But the last string of laptops ( work-provided ) have never even been able to suspend properly, and it's always an Intel device giving the trouble. The most annoying issue is the occasional ( like every 10 attempts ) GPU lockup. Then there are lower-level but more common ( ie every time ) annoyances, like bluetooth never coming back up ... again, Intel hardware.

So Intel is now interested in what's up with suspending in Linux, eh? About time ...

[Tomin]

For whatever reason my Dell XPS13 (model 9333, those with Haswell processor and coil whine) doesn't always suspend correctly. Sometimes it just turns display off and seems to go to some kind of deadlock where it burns power (the computer gets hot after a while, fans are spinning) and it won't suspend. Particularly annoying when I happen to leave it to charge and since the same light tells me that it's on and that it's charging I can't tell if it suspended or not. The only way, that I know of, to recover is to press the power button for four seconds. I should investigate whether this is a user space or kernel space problem. At this point maybe even a reinstall of Arch Linux would not be a bad idea (maybe I'll get a bigger SSD first and then try that).

That same laptop also has an issue with shutting down. For some reason sometimes when I turn it off it seems to work fine until it says "[timestamp] Power down." or something like that, AFAIK the very last message before turning off, and then it just stops without shutting down completely. I have to shut it down by pressing down the power button for four seconds, luckily by this point the filesystems have been unmounted and synced so it's fine to turn it off.

On desktops suspending (and powering off) seems to work for me right now. Though, I don't use suspend on desktops as often.

[samdraz]

mine suspends and resume within 1-2second, suspend:700ms, resume 1500ms

[DanL]

My experience with suspending and hibernating is that it works best with the least Intel devices possible.

I disagree with your theory. I used to triage bugs for Debian and Ubuntu and I've seen so many suspend/resume complaints on all sorts of devices. I don't think you can pin it on one manufacturer.

[RussianNeuroMancer]

Dear Intel, upon suspend freeze in Linux, devices with your hardware stay in S0 and doesn't go into S0i3. Please do something about it.

[Azrael5]

it's one of the matter linux continues to fail.