linux-kernel - Re: PM / Suspend: Print wall time at suspend entry and exit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 6 Jun 2017 00:40:11 +0200
From:   Pavel Machek <pavel@....cz>
To:     Mark Salyzyn <salyzyn@...roid.com>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux PM <linux-pm@...r.kernel.org>,
        Todd Poynor <toddpoynor@...gle.com>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Len Brown <len.brown@...el.com>
Subject: Re: PM / Suspend: Print wall time at suspend entry and exit

Hi!

> >>>On Mon, Jun 5, 2017 at 9:31 PM, Mark Salyzyn <salyzyn@...roid.com>
> >>>wrote:
> >>>>Permits power state and battery life diagnosis.
> >>>Which is possible even without this patch and we have tools for that
> >>>(analyze_suspend.py, anyone?).
> >>>
> >>>Honestly, I don't see why this change is necessary or even useful.
> >>I ran dmesg after resume and now I...
> >>
> >>[1019729.486249] ACPI : EC: EC stopped
> >>[1019729.486255] PM: Saving platform NVS memory
> >>[1019729.486276] Disabling non-boot CPUs ...
> >>[1019729.490423] smpboot: CPU 1 is now offline
> >>[1019729.531015] smpboot: CPU 2 is now offline
> >>[1019729.577742] Broke affinity for irq 16
> >>[1019729.577747] Broke affinity for irq 17
> >>[1019729.577751] Broke affinity for irq 19
> >>[1019729.577756] Broke affinity for irq 23
> >>[1019729.578771] smpboot: CPU 3 is now offline
> >>[1019729.604987] ACPI: Low-level resume complete
> >>[1019729.605052] ACPI : EC: EC started
> >>[1019729.605054] PM: Restoring platform NVS memory
> >>[1019729.605352] Suspended for 3196.166 seconds
> >>[1019729.605376] Enabling non-boot CPUs ...
> >>[1019729.631246] x86: Booting SMP configuration:
> >>[1019729.631249] smpboot: Booting Node 0 Processor 1 APIC 0x1
> >>[1019729.631496] Initializing CPU#1
> >>[1019729.631514] Disabled fast string operations
> >>[1019729.636643]  cache: parent cpu1 should not be sleeping
> >>[1019729.637286] CPU1 is up
> >>
> >>Aha. I missed the "suspended for 3196" message before. So I believe
> >>you are right; there's already enough information in the kernel logs.
> >>
> >>                                    Pavel
> >
> >We do not get the Suspended for messages on failure to suspend (interrupt
> >wakeup), but that is a weak argument ;-/ since it is discoverable from
> >context and we can count the aborted suspends switching the tools. The
> >Suspended for messages are in ms which is not accurate enough for Android
> >user space logs (in ns) long term continuous tracking of monotonic

Nanoseconds do not really matter here, right?

And userspace should not really be parsing kernel logs for time
keeping, right?

> >realtime and makes no accounting for ntp adjustments. Currently, post
> >mortem data only on Android, we have no other way to align monotonic
> >kernel logs with user space logs without this.

> Also, not all devices have a persistent time that can be collected at
> Suspend at that level, so they have to turn off CONFIG_SUSPEND_TIME because
> 'Suspended for' is meaningless on those platforms, but alas these prints
> turn into a backup mechanism for those platforms as they are printed befopre
> and after the hardware is up and running.

The devices you are talking about... is any of them running something
close to mainline kernel, or is planned for merge? I'd like to get a
phone that is supported by mainline...
									
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Download attachment "signature.asc" of type "application/pgp-signature" (182 bytes)