(SOLVED) Odd sleep/ACPI behaviour

If none of the more specific forums is the right place to ask

Re: Odd sleep/ACPI behaviour

Postby Deb-fan » 2020-07-28 23:36

Not even going to try outlining all the many ways it's ignorant, too damn many. Using release candidate kernels, gazillion other things. The kernel is not just another pkg, not the same as others on a gnu/nix Os, shouldn't be handled as if it is. Really no biggie, as it's harmless overall and noted above actually totally encourage people to try different kernels and/or tweak the things. When discussing something like getting best audio, real-time and etc, there are kernel configs(oops, edit: and/or patches) which could make a world of difference, are distro's for audiophiles around, could be a good resource for someone with a big interest in the subject.

No worries, not meaning to argue or insult here. Ignorance means not knowing, when it comes to gnu/nix and the kernel also I'll remain extremely ignorant to a huge degree for life. Imo there's simply too much involved for anyone not to stay ignorant to some (great) extent. :)
Last edited by Deb-fan on 2020-07-29 03:53, edited 1 time in total.
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 944
Joined: 2012-08-14 12:27

Re: Odd sleep/ACPI behaviour

Postby Mr. Lumbergh » 2020-07-29 02:10

Deb-fan wrote:When discussing something like getting best audio, real-time and etc, there are kernel configs which could make a world of difference, are distro's for audiophiles around, could be a good resource for someone with a big interest in the subject.

That's exactly what I'm after. I gravitated to the RT kernel because with it I can't hear a delay between a keypress on my MIDI controller and when the sound is generated in the DAW, but I could on the mainline. Going from the 4.19RT to the 5.4RT played even nicer with my particular setup, so I didn't have reason to think that the 5.6 would be a regression, I s'pose. I also came to Debian from a rolling-release distro so it became habitual for me to install the latest, at least after giving it a week or so. Live and learn. If no other good comes from it I filed a bug report on the 5.6RT so hopefully it gets fixed.
I'll probably just stick with the 5.4 for the time being; it and 4.19 will both be getting 6 years of LTS: https://fossbytes.com/linux-lts-kernel-4-19-and-5-4-will-now-be-supported-for-6-years-instead-of-2/, so that's cool.
I know there are a couple of distros out there with the kind of stuff that I like to do in mind such as KXStudio and Ubuntu Studio, and I actually do get some things from KX but I won't use Ubuntu Studio, because Ubuntu. I figured on this box I'd start with a barebones Debian and add some of those same tweaks myself that make it better for music work because Debian is so tweakable and there are so many packages available for it. Work in progress but it's been fun. Anywho, I've noticed I'm rambling a bit.
Deb-fan wrote:No worries, not meaning to argue or insult here. Ignorance means not knowing, when it comes to gnu/nix and the kernel also I'll remain extremely ignorant to a huge degree for life. Imo there's simply too much involved for anyone not to stay ignorant to some (great) extent. :)

OK, no worries and pardon the way I interpreted it.
Mr. Lumbergh
 
Posts: 97
Joined: 2019-08-02 04:28

Re: Odd sleep/ACPI behaviour

Postby Deb-fan » 2020-07-29 02:31

Totally understandable, lately been mega-dosing caffeine, guzzling coffee and staying awake long periods, 36-48hrs at a time, sleep 10-12hrs, rinse/repeat, get to the point where I become a highly impatient/irritable azzhat. Not that the crazy sleep schedule I've been doing is much of an excuse to justify being an ahole(isn't). Sounds like you're making some good progress and a real effort to study gnu/nix for what you're interested in. Which of course is really cool. Haven't devoted a huge amount of effort in studying audio or gaming or etc on gnu/nix yet. Only bits n pieces.

The kernel someone is using surely has to be a big factor there. Though kernel has to be a huge consideration while discussing anything nix. It's the friggin Linux kernel after all. The Linux part of errr gnu/Linux. :)
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 944
Joined: 2012-08-14 12:27

Re: Odd sleep/ACPI behaviour

Postby Mr. Lumbergh » 2020-07-29 05:13

Deb-fan wrote:Totally understandable, lately been mega-dosing caffeine, guzzling coffee and staying awake long periods, 36-48hrs at a time, sleep 10-12hrs, rinse/repeat, get to the point where I become a highly impatient/irritable azzhat. Not that the crazy sleep schedule I've been doing is much of an excuse to justify being an ahole(isn't). Sounds like you're making some good progress and a real effort to study gnu/nix for what you're interested in. Which of course is really cool. Haven't devoted a huge amount of effort in studying audio or gaming or etc on gnu/nix yet. Only bits n pieces.

The kernel someone is using surely has to be a big factor there. Though kernel has to be a huge consideration while discussing anything nix. It's the friggin Linux kernel after all. The Linux part of errr gnu/Linux. :)

It isn't you, just the way things don't come across as intended online sometimes. Really no worries now that we've had a real convo about it. But get some sleep man!
I'm still drilling down to find what tweaks I can do that will make this a better music platform, but so far the RT kernel and some massaging of JACK Audio have had the biggest impact. This Debian install has become not only my main music production rig but has become my everyday driver over that other distro I mentioned. I occasionally have to restart my DAW, but I don't think it's Debian that's the issue so much have having to use WINE when there isn't a native Linux plugin for virtual instruments.
Mr. Lumbergh
 
Posts: 97
Joined: 2019-08-02 04:28

Re: Odd sleep/ACPI behaviour

Postby Deb-fan » 2020-07-29 05:54

Can be and sometimes am a butthole. Whenever a person calls another person ignorant, they can rest assured it's going to be somewhat offensive and taken that way, sorry. One thing about spending too much time in gnu/Linux forums a person gets burnt out and it's easy to start considering people simply too lazy and/or stupid to competently admin their systems or show the common sense, in taking steps to learn for themselves. That isn't aimed at you, only when I'm already in that azzhat state of mind, have a tendency to act like a jerk even in cases where it's unwarranted. Some posts in thread for example. :)

Have studied things which parallel what would (or some things which could) be involved in getting best audio recording, sound quality etc in gnu/Linux. Might try looking into a Windows install on a real hypervisor ie: Kvm vs use of an emulator like WINE. I've looked it over but could never bring myself to install it. The idea of mixing a bunch of windows software into a gnu/Linux OS, has always somewhat disgusted me on some level. So I've read but could never commit to actually using it/WINE. Might give you some joy(a Windows guest os installed on a virtual machine) on software not natively supported on Nix and may even impart some degree of better performance of such software vs what can be done with WINE. That's pure speculation on my part though, I honestly don't know for sure.

Another poss place to look is what your system is using as the power governor. On new Intel (they have this pstate thing) on older Intel and on AMD generally default to the ondemand governor. One of the kernel config's jacked up on RT kernels and many desktop "performance" kernels is the timer interrupt freq, how often the kernel checks for tasks it should be doing, on RT and performance kerns often you'll find it set at 1000hz vs 250hz that many-most distro's kernels come with. Reason I bring up power governor selected on an OS, it plays a big role in how high, how quickly your cpu(s) step up their clock speeds. This can have an impact on how live time apps etc perform. If the clock speeds are capped too low(by power governor), even though the kernel is RT and has preemption and timer interrupt set to 1000hz, such a setup could still clash and come up lacking in performance potential. You've likely already covered this to whatever extent too, if not am just noting it. Are you already using the performance governor ? I like the ondemand governor but have tweaked it from what's generally found set default.

On my Os's it's set to step up the cpu-cores clock speeds(cycles)more quickly and keep it raised a bit longer whenever the cores get put under load. Most the time, if not needed I want my cores to stay running at the lowest clock speeds. Conserves power, reduces heat and other wear n tear on the proc's. Anyway don't doubt getting best audio out of gnu/Linux is a fairly wide subject itself, so am going to shut it as I've never focused on it and don't doubt there's a TON of good resources available with which people could learn much on the subject. You can find which config options are set/used in whichever of your kernels looking at their files kept in /boot ie: config-4.19 etc in that location.


Lol ... thanks, will have to sleep eventually, earliest est crash-time, approx 10-11pm tonight. My kernel(brain) is still running pretty smoothy at this point. Once my uptime hits around the 20hrs mark, errrrr, not so much. :)
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 944
Joined: 2012-08-14 12:27

Re: Odd sleep/ACPI behaviour

Postby Mr. Lumbergh » 2020-07-29 20:54

Deb-fan wrote:Have studied things which parallel what would (or some things which could) be involved in getting best audio recording, sound quality etc in gnu/Linux. Might try looking into a Windows install on a real hypervisor ie: Kvm vs use of an emulator like WINE. I've looked it over but could never bring myself to install it. The idea of mixing a bunch of windows software into a gnu/Linux OS, has always somewhat disgusted me on some level. So I've read but could never commit to actually using it/WINE. Might give you some joy(a Windows guest os installed on a virtual machine) on software not natively supported on Nix and may even impart some degree of better performance of such software vs what can be done with WINE. That's pure speculation on my part though, I honestly don't know for sure.

The bummer about virtualization though is the additional overhead of running Windows, plus the fighting it always seems to need to pass through my audio interface to the guest OS. Then there's that whole Windows thing...
There are a couple of synth plugins I have that have demo versions that ran fine in WINE, but once I paid for the full version demanded crap like iLok piracy protection that I just can't get WINE to run. That worked my nerves for obvious reasons, and I've avoided anything by AIR or iZotope since that hard lesson. Live and learn I guess. Since I can't my money back I'll go ahead and install those on my Macbook using their OSX versions, once I get the DAW and stuff installed there. I'm still waiting on a couple upgrades.
Deb-fan wrote:Another poss place to look is what your system is using as the power governor. On new Intel (they have this pstate thing) on older Intel and on AMD generally default to the ondemand governor. One of the kernel config's jacked up on RT kernels and many desktop "performance" kernels is the timer interrupt freq, how often the kernel checks for tasks it should be doing, on RT and performance kerns often you'll find it set at 1000hz vs 250hz that many-most distro's kernels come with. Reason I bring up power governor selected on an OS, it plays a big role in how high, how quickly your cpu(s) step up their clock speeds. This can have an impact on how live time apps etc perform. If the clock speeds are capped too low(by power governor), even though the kernel is RT and has preemption and timer interrupt set to 1000hz, such a setup could still clash and come up lacking in performance potential. You've likely already covered this to whatever extent too, if not am just noting it. Are you already using the performance governor ? I like the ondemand governor but have tweaked it from what's generally found set default.

On my Os's it's set to step up the cpu-cores clock speeds(cycles)more quickly and keep it raised a bit longer whenever the cores get put under load. Most the time, if not needed I want my cores to stay running at the lowest clock speeds. Conserves power, reduces heat and other wear n tear on the proc's. Anyway don't doubt getting best audio out of gnu/Linux is a fairly wide subject itself, so am going to shut it as I've never focused on it and don't doubt there's a TON of good resources available with which people could learn much on the subject. You can find which config options are set/used in whichever of your kernels looking at their files kept in /boot ie: config-4.19 etc in that location.


No, I hadn't actually thought about governor tweaks, I'll have to look into it. Thanks for the suggestion.
Mr. Lumbergh
 
Posts: 97
Joined: 2019-08-02 04:28

Re: Odd sleep/ACPI behaviour

Postby Mr. Lumbergh » 2020-08-03 15:19

pendrachken wrote:
It's also remotely possible that xen is grabbing the USB connection when waking from sleep for some reason. Personally, I think a kernel regression is way more likely, but this is also a possibility.


Well, I thought I was out of the woods using the 5.4 kernel but it happened again last night. I had to plug in a keyboard from another box this morning to wake from sleep and restart so that the USB was back online. You mentioned Xen and I did run a VM briefly over the weekend; how would I check to see if it's still binding to the USB3 ports?
Mr. Lumbergh
 
Posts: 97
Joined: 2019-08-02 04:28

Previous

Return to General Questions

Who is online

Users browsing this forum: No registered users and 13 guests

fashionable