Description of problem: In RHEL-5.3, running the Xen kernel kernel-xen-2.6.18-128.el5 and the oprofile package oprofile-0.9.3-18.el5.x86_64.rpm, I am successfully able to use xenoprof to collect profile data from the dom0 and hypervisor using this script: #!/bin/bash opcontrol --reset opcontrol --start-daemon --xen=/usr/lib/debug/boot/xen-syms-2.6.18-128.el5.debug --vmlinux=/usr/lib/debug/lib/modules/2.6.18-128.el5xen/vmlinux opcontrol --start sleep 10 opcontrol --stop opcontrol --shutdown opreport -l However, when running RHEL-5.4 with kernel-xen-2.6.18-156.el5 and oprofile-0.9.4-10.el5, I can no longer get profile data, and the last step (opreport -l) reports: opreport error: No sample file found: try running opcontrol --dump or specify a session containing sample files Also, running "opcontrol -l" on RHEL-5.3 shows the CPU_CLK_UNHALTED event available, while "opcontrol -l" on RHEL-5.4 doesn't show that event available anymore. If I downgrade *just* the oprofile package to the 5.3 version, reboot (which seems to be necessary, for some reason), and run the test again, it runs successfully.
It seems odd that the machine needed to be rebooted to get things working again. I tried to replicate this problem on 2.6.18-156.el5xen kernel and was unable to replicate it. # uname -a; rpm -q xen oprofile Linux calfee.devel.redhat.com 2.6.18-156.el5xen #1 SMP Mon Jun 29 18:24:43 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux xen-3.0.3-80.el5 oprofile-0.9.4-10.el5 I corrected a the script with line continuations, print out a bit less info (opreport -t1) and proper --xen and --vmlinux: #!/bin/bash opcontrol --reset opcontrol --start-daemon \ --xen=/usr/lib/debug/boot/xen-syms-2.6.18-156.el5.debug \ --vmlinux=/usr/lib/debug/lib/modules/`uname -r`/vmlinux opcontrol --start sleep 10 opcontrol --stop opcontrol --shutdown opreport -l -t1 I definitely obtained data with oprofile-0.9.4-10. # opcontrol --version opcontrol: oprofile 0.9.4 compiled on Jun 21 2009 23:09:40 # opreport -l -t1 .... CPU: P4 / Xeon with 2 hyper-threads, speed 2793.1 MHz (estimated) Counted GLOBAL_POWER_EVENTS events (time during which processor is not stopped) with a unit mask of 0x01 (mandatory) count 100000 samples % image name app name symbol name 1403 15.3652 vmlinux-unknown vmlinux-unknown /vmlinux-unknown 327 3.5812 bash bash /bin/bash 245 2.6832 xen-syms-2.6.18-152.el5.debug xen-syms-2.6.18-152.el5.debug toggle_guest_mode 239 2.6175 ext3 ext3 /ext3
I just tried it out on two different machines here, and it worked just fine. Unfortunately the original machine I was having a problem with is in RHTS, so I'll have to go back on that machine and re-reserve it. I'll put this back to waiting on myself until I can get back on the machine and see if I can reproduce. Thanks for taking a look, Chris Lalancette
OK. This must have been user cockpit error, since this now works for me on the original problem machine with kernel-xen 2.6.18-157.el5 and oprofile 0.9.4-10. Sorry for the noise, I'll close this as NOTABUG. Chris Lalancette
Clearing out bogus flags for reporting purposes. Chris Lalancette