This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


[Xen-devel] *massive* slowdown when running oprofile in dom0

To: Xen Devel <xen-devel@xxxxxxxxxxxxxxxxxxx>, Jose Renato Santos <jsantos@xxxxxxxxxx>
Subject: [Xen-devel] *massive* slowdown when running oprofile in dom0
From: Gianni Tedesco <gianni.tedesco@xxxxxxxxxx>
Date: Wed, 1 Sep 2010 10:06:16 +0100
Delivery-date: Wed, 01 Sep 2010 02:09:52 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx

I am running oprofile in dom0 (just profiling dom0, not VM's) and
whenever statistics collection is enabled (via opcontrol --start) the
performance of VM's drops through the floor.

For example a fresh windows(r) XP(tm) SP3 install takes 2 minutes to
boot without oprofile but a short hair under 5 minutes when stats are
being collected. It seemed even worse for installing, I had left it
going for hours before realising oprofile was the culprit (and still at
the yellow progress bar on blue background 'copying files' stage). In
general there is just a noticable 'lagginess' about everything.

I have ruled out the obvious culprits, eg: xenoprof interrupt is only
firing 800 or so times a second, oprofiled is not using all the CPU
time. So it's got to be a hypervisor/emulation thing right?

Furthermore when I disable collection with opcontrol --stop I get:

(XEN) nmi_stop: APIC not good 66560l
(XEN) CA-24901: CPU 0 fixed up 1 lost platform-timer cycles
(XEN) CPU01:              @ 00000ce3b1e78f55, 00000ce3b1e794d1
(XEN)   t 000026a6f8ef3731, 00000ce26efa7a65, 00000ce26efa79a4 <<-1 *aaac435f
(XEN)   c 000026aac1accdc5, 00000ce3b1e75cab, 00000ce3b1e73927 <<-1 *aaabcb5b
(XEN) CPU00:              @ 00000ce3b1e7ea2b, 00000ce3b1e7f2ab
(XEN)   t 000026a6f8ef35ba, 00000ce26efa7b28, 00000ce26efa79a4 <<-1 *aaac4113
(XEN)   c 000026aac1accc21, 00000ce3b1e759d3, 00000ce3b1e73927 <<-1 *aaabd388

And everything hangs for 5 seconds due to mdelay(5000) !?!?!?

Please tell me it's bad hardware...

# cat /proc/cpuinfo 
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 6
model name      : Genuine Intel(R) CPU 3.00GHz
stepping        : 2
cpu MHz         : 2999.892
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 6
wp              : yes
flags           : fpu de tsc msr pae mce cx8 apic sep mtrr mca cmov pat clflush
acpi mmx fxsr sse sse2 ss ht nx constant_tsc up pni vmx est cid
bogomips        : 6004.03
clflush size    : 64
power management:

Xen-devel mailing list

<Prev in Thread] Current Thread [Next in Thread>