[Bug 530035] New: [Virtualization/xen] v3.4.1 power management setup not stable @ boot
http://bugzilla.novell.com/show_bug.cgi?id=530035 Summary: [Virtualization/xen] v3.4.1 power management setup not stable @ boot Classification: openSUSE Product: openSUSE.org Version: unspecified Platform: x86-64 OS/Version: openSUSE 11.1 Status: NEW Severity: Normal Priority: P5 - None Component: 3rd party software AssignedTo: carnold@novell.com ReportedBy: pgnet.dev@gmail.com QAContact: opensuse-communityscreening@forge.provo.novell.com Found By: Customer on openSUSE 11.1 Dom0, uname -ri 2.6.27.29-7-xen x86_64 where, fyi, CPU is an AMD Phenom II X4 920 @2.8 GHz. i've installed from https://build.opensuse.org/package/show?package=xen&project=Virtualization Xen 3.4.1 rpms, xen-3.4.1_19711_01-9.1.src.rpm xen-3.4.1_19711_01-9.1.x86_64.rpm xen-devel-3.4.1_19711_01-9.1.x86_64.rpm xen-doc-html-3.4.1_19711_01-9.1.x86_64.rpm xen-doc-pdf-3.4.1_19711_01-9.1.x86_64.rpm xen-libs-3.4.1_19711_01-9.1.x86_64.rpm xen-tools-3.4.1_19711_01-9.1.x86_64.rpm i've booted, xm dmesg ... (XEN) Xen version 3.4.1_19711_01-9.1 (abuild@) (gcc version 4.3.2 [gcc-4_3-branch revision 141291] (SUSE Linux) ) Wed Aug 5 01:48:18 UTC 2009 (XEN) Latest ChangeSet: 19711 (XEN) Command line: dom0_mem=768M cpufreq=xen:performance cpuidle loglvl=all loglvl_guest=all vga=gfx-1280x1024x32 console=vga,com1 com1=57600,8n1 ... with kernel-xen grub settings of, cpufreq=xen:performance cpuidle @ boot, checking @ Dom0, cat /proc/cpuinfo | grep MHz cpu MHz : 2800.000 cpu MHz : 2800.000 cpu MHz : 2800.000 cpu MHz : 2800.000 after simply waiting ~ 20 seconds, repeating, cat /proc/cpuinfo | grep MHz cpu MHz : 800.000 cpu MHz : 800.000 cpu MHz : 800.000 cpu MHz : 800.000 and, xenpm get-cpufreq-para [CPU0] failed to get cpufreq parameter [CPU1] failed to get cpufreq parameter [CPU2] failed to get cpufreq parameter [CPU3] failed to get cpufreq parameter but, then, xenpm set-scaling-governor performance xenpm get-cpufreq-para cpu id : 0 affected_cpus : *0 1 2 3 cpuinfo frequency : max [2800000] min [800000] cur [800000] scaling_driver : scaling_avail_gov : userspace performance powersave ondemand current_governor : performance scaling_avail_freq : 2800000 2100000 1600000 *800000 scaling frequency : max [2800000] min [800000] cur [800000] cpu id : 1 affected_cpus : 0 *1 2 3 cpuinfo frequency : max [2800000] min [800000] cur [800000] scaling_driver : scaling_avail_gov : userspace performance powersave ondemand current_governor : performance scaling_avail_freq : 2800000 2100000 1600000 *800000 scaling frequency : max [2800000] min [800000] cur [800000] cpu id : 2 affected_cpus : 0 1 *2 3 cpuinfo frequency : max [2800000] min [800000] cur [800000] scaling_driver : scaling_avail_gov : userspace performance powersave ondemand current_governor : performance scaling_avail_freq : 2800000 2100000 1600000 *800000 scaling frequency : max [2800000] min [800000] cur [800000] cpu id : 3 affected_cpus : 0 1 2 *3 cpuinfo frequency : max [2800000] min [800000] cur [800000] scaling_driver : scaling_avail_gov : userspace performance powersave ondemand current_governor : performance scaling_avail_freq : 2800000 2100000 1600000 *800000 scaling frequency : max [2800000] min [800000] cur [800000] changing, xenpm set-scaling-governor ondemand xenpm get-cpufreq-para | grep "cur \[" cpuinfo frequency : max [2800000] min [800000] cur [800000] sampling_rate : max [10000000] min [10000] cur [20000] scaling frequency : max [2800000] min [800000] cur [800000] cpuinfo frequency : max [2800000] min [800000] cur [800000] sampling_rate : max [10000000] min [10000] cur [20000] scaling frequency : max [2800000] min [800000] cur [800000] cpuinfo frequency : max [2800000] min [800000] cur [800000] sampling_rate : max [10000000] min [10000] cur [20000] scaling frequency : max [2800000] min [800000] cur [800000] cpuinfo frequency : max [2800000] min [800000] cur [800000] sampling_rate : max [10000000] min [10000] cur [20000] scaling frequency : max [2800000] min [800000] cur [800000] then back, xenpm set-scaling-governor performance xenpm get-cpufreq-para | grep "cur \[" cpuinfo frequency : max [2800000] min [800000] cur [2800000] scaling frequency : max [2800000] min [800000] cur [2800000] cpuinfo frequency : max [2800000] min [800000] cur [2800000] scaling frequency : max [2800000] min [800000] cur [2800000] cpuinfo frequency : max [2800000] min [800000] cur [2800000] scaling frequency : max [2800000] min [800000] cur [2800000] cpuinfo frequency : max [2800000] min [800000] cur [2800000] scaling frequency : max [2800000] min [800000] cur [2800000] seems to get the CPU to correct current cpu freq ... and, afaict, it stays there without chaning again (e.g., after 20 seconds ... as above). i'm clearly not testing this on Factory, but suspect this may manifest there as well ... -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 Jan Beulich <jbeulich@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED AssignedTo|carnold@novell.com |jbeulich@novell.com -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User jbeulich@novell.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c1 Jan Beulich <jbeulich@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P4 - Low Status|ASSIGNED |NEEDINFO Info Provider| |pgnet.dev@gmail.com Severity|Normal |Minor --- Comment #1 from Jan Beulich <jbeulich@novell.com> 2009-08-18 00:32:28 MDT --- Am I getting it right that the only issue you're reporting is that "xenpm get-cpufreq-para" doesn't work the first time it is being run on that system? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c2 --- Comment #2 from pgnet Dev <pgnet.dev@gmail.com> 2009-08-18 09:30:49 MDT --- no. (1) setting, cpufreq=xen:performance cpuidle @ boot/grub. clearly that doesn't 'stick'. it needs to. (2) after boot, getting "failed to get cpufreq parameter" is clearly a problem. (3) manually setting xenpm set-scaling-governor performance xenpm get-cpufreq-para cpu id : 0 affected_cpus : *0 1 2 3 cpuinfo frequency : max [2800000] min [800000] cur [800000] after boot 'fixes' the "failed to get cpufreq parameter", but fails to set the freq correctly. note -> "cur [800000]" (4) have to first manually set xenpm set-scaling-governor ondemand before xenpm set-scaling-governor performance to get 'performance' governor to set/report frequencty correctly AND 'stick' is a problem. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User jbeulich@novell.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c3 Jan Beulich <jbeulich@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |RESOLVED CC| |joachim.deguara@amd.com Info Provider|pgnet.dev@gmail.com | Resolution| |FEATURE --- Comment #3 from Jan Beulich <jbeulich@novell.com> 2009-08-19 06:01:51 MDT --- Just to clarify: The governors other than on-demand are (even in -unstable) only supported/used on Intel CPUs (using the generic ACPI cpufreq driver). The PowerNow! driver is hardwired to the on-demand governor. I don't think we're going to fix this on our own - we'll want AMD to adjust their code (Joachim, could you find out if there are any plans to do so, and if so, when to expect functional code in -unstable?), and only then we can consider backporting it. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c4 --- Comment #4 from pgnet Dev <pgnet.dev@gmail.com> 2009-08-19 10:03:19 MDT --- (In reply to comment #3)
Just to clarify: The governors other than on-demand are (even in -unstable) only supported/used on Intel CPUs (using the generic ACPI cpufreq driver). The PowerNow! driver is hardwired to the on-demand governor. I don't think we're going to fix this on our own - we'll want AMD to adjust their code (Joachim, could you find out if there are any plans to do so, and if so, when to expect functional code in -unstable?), and only then we can consider backporting it.
PowerNow's _hardwired_ to a single governor ?! Then, you're a better sleuth than I ... where on earth did you find that fact documented? I've clearly missed it :-/ Wholly agree that if it _is_ the case, then it should be an upstream fix. I'm not clear what effect, if any, the cpufreq=xen:ondemand(or performance) in grub's having ... is it simply being ignored in the PowerNow case? Even if so, that still leaves items (1) & (2) as issues ... or, perhaps _all_ of this depends on AMD fixes? I'll happily retest the above behaviors once that's in place. p.s. I don't think "RESOLVED:FEATURE" is wholly appropriate here -- as the goal needs to be "works consistently with Xen's features and across CPUs", but not clear how ELSE one would categorize other than to leave OPEN ... -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User jbeulich@novell.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c5 --- Comment #5 from Jan Beulich <jbeulich@novell.com> 2009-08-20 05:14:42 MDT --- (In reply to comment #4)
PowerNow's _hardwired_ to a single governor ?! Then, you're a better sleuth than I ... where on earth did you find that fact documented? I've clearly missed it :-/
I'm reading the sources.
I'm not clear what effect, if any, the cpufreq=xen:ondemand(or performance) in grub's having ... is it simply being ignored in the PowerNow case?
Correct.
Even if so, that still leaves items (1) & (2) as issues ... or, perhaps _all_ of this depends on AMD fixes?
Very likely, as xenpm assumes the general governor infrastructure was properly initialized, which it is only partly (at best) for the powernow driver. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c6 --- Comment #6 from pgnet Dev <pgnet.dev@gmail.com> 2009-08-20 08:24:22 MDT --- thanks. hopefully Joachim will follow-up here, even though you've closed the bug ... -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c7 pgnet Dev <pgnet.dev@gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FEATURE | --- Comment #7 from pgnet Dev <pgnet.dev@gmail.com> 2009-09-22 09:37:32 MDT --- (In reply to comment #3)
Just to clarify: The governors other than on-demand are (even in -unstable) only supported/used on Intel CPUs (using the generic ACPI cpufreq driver). The PowerNow! driver is hardwired to the on-demand governor. I don't think we're going to fix this on our own - we'll want AMD to adjust their code (Joachim, could you find out if there are any plans to do so, and if so, when to expect functional code in -unstable?), and only then we can consider backporting it.
Reopening. Still trying to get an answer to the question posed. Direct email followups have, unfortunately, not been responded to at all. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User jbeulich@novell.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c8 Jan Beulich <jbeulich@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FEATURE --- Comment #8 from Jan Beulich <jbeulich@novell.com> 2009-09-22 09:44:35 MDT --- (In reply to comment #7)
Reopening. Still trying to get an answer to the question posed. Direct email followups have, unfortunately, not been responded to at all.
Sorry, but I can't see what this changes with respect to this bug's status: This continues to be a feature request, and will be treated as such. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c9 --- Comment #9 from pgnet Dev <pgnet.dev@gmail.com> 2009-09-22 10:24:30 MDT ---
and will be treated as such.
Since it's _you_ that posed the question, suggesting that AMD needs to address the issue, and you've not responded to any off-bug, direct follow-up email inquiries on the matter, what would you suggest I do? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User pgnet.dev@gmail.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c10 --- Comment #10 from pgnet Dev <pgnet.dev@gmail.com> 2009-09-22 10:25:03 MDT ---
and will be treated as such.
Since it's _you_ that posed the question, suggesting that AMD needs to address the issue, and you've not responded to any off-bug, direct follow-up email inquiries on the matter, what would you suggest I do? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User jbeulich@novell.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c11 --- Comment #11 from Jan Beulich <jbeulich@novell.com> 2009-09-23 00:42:29 MDT --- (In reply to comment #10)
and will be treated as such.
Since it's _you_ that posed the question, suggesting that AMD needs to address the issue, and you've not responded to any off-bug, direct follow-up email inquiries on the matter, what would you suggest I do?
Honestly, I'd suggest to simply be patient. After all, this is not some critical problem we're talking about - frequency scaling is working (if you want it), all that's missing is proper selection and switching between governors. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=530035 User mark.langsdorf@amd.com added comment http://bugzilla.novell.com/show_bug.cgi?id=530035#c12 --- Comment #12 from Mark Langsdorf <mark.langsdorf@amd.com> 2009-10-01 12:38:07 MDT --- Created an attachment (id=320715) --> (http://bugzilla.novell.com/attachment.cgi?id=320715) patch to update PowerNow! to latest Xen cpufreq The attached patch updates the Xen PowerNow! driver to match the latest Xen cpufreq infrastructure and add support for capabilities that were added to cpufreq after the driver was initially written. It passes all the test cases the original reporter mentioned. It has been submitted to xen-unstable but has not yet made it into the tree. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com