Created attachment 28741 [details] .config oprofile doesn't work, but 'perf record -e cycles' does. $ sudo opcontrol --start --verbose Parameters used: SESSION_DIR /var/lib/oprofile LOCK_FILE /var/lib/oprofile/lock SAMPLES_DIR /var/lib/oprofile/samples CURRENT_SAMPLES_DIR /var/lib/oprofile/samples/current CPUTYPE x86-64/family10 BUF_SIZE default value BUF_WATERSHED default value CPU_BUF_SIZE default value SEPARATE_LIB 0 SEPARATE_KERNEL 0 SEPARATE_THREAD 0 SEPARATE_CPU 0 CALLGRAPH 0 VMLINUX /home/edwin/builds/linux-2.6/vmlinux KERNEL_RANGE ffffffff80200000,ffffffff805b4e35 XENIMAGE none XEN_RANGE Error: counter 0 not available dmesg says: [12052.123057] oprofile: AMD IBS detected (0x0000001f) [12052.123065] oprofile: using NMI interrupt. [12124.203495] oprofile: AMD IBS detected (0x0000001f) [12124.203504] oprofile: using NMI interrupt. Attached .config
Created attachment 28751 [details] dmesg
Seems to work with 2.6.35.4. Time to bisect.
I bisected a bit, here is my bisection log. Kind of hard to bisect because reiserfs evict errors keep popping up, or md resync is started and oopses, etc. Always have to go back to a released kernel to run 'git bisect good/bad' and compile new kernel. I'll have to see if I can reproduce the bug in KVM, should be easier to bisect there. git bisect start # bad: [2bfc96a127bc1cc94d26bfaa40159966064f9c8c] Linux 2.6.36-rc3 git bisect bad 2bfc96a127bc1cc94d26bfaa40159966064f9c8c # good: [1506707a6c740db316e422239a53ae5df1727591] Linux 2.6.35.4 git bisect good 1506707a6c740db316e422239a53ae5df1727591 # good: [9fe6206f400646a2322096b56c59891d530e8d51] Linux 2.6.35 git bisect good 9fe6206f400646a2322096b56c59891d530e8d51 # good: [ab69bcd66fb4be64edfc767365cb9eb084961246] Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6 git bisect good ab69bcd66fb4be64edfc767365cb9eb084961246 # good: [b4eda9cb48eac1b7369ad33e5d015c33b4376431] stallion: prune lock_kernel calls git bisect good b4eda9cb48eac1b7369ad33e5d015c33b4376431 # good: [b4eda9cb48eac1b7369ad33e5d015c33b4376431] stallion: prune lock_kernel calls git bisect good b4eda9cb48eac1b7369ad33e5d015c33b4376431 # good: [b4eda9cb48eac1b7369ad33e5d015c33b4376431] stallion: prune lock_kernel calls git bisect good b4eda9cb48eac1b7369ad33e5d015c33b4376431 # good: [16bb85bc12e7d0b066b1471185f3889cb1ceb382] Merge git://git.infradead.org/battery-2.6 git bisect good 16bb85bc12e7d0b066b1471185f3889cb1ceb382 # good: [16bb85bc12e7d0b066b1471185f3889cb1ceb382] Merge git://git.infradead.org/battery-2.6 git bisect good 16bb85bc12e7d0b066b1471185f3889cb1ceb382 # good: [16bb85bc12e7d0b066b1471185f3889cb1ceb382] Merge git://git.infradead.org/battery-2.6 git bisect good 16bb85bc12e7d0b066b1471185f3889cb1ceb382
Not sure I have have the same issue: # opcontrol --start-daemon --verbose Parameters used: SESSION_DIR /var/lib/oprofile LOCK_FILE /var/lib/oprofile/lock SAMPLES_DIR /var/lib/oprofile/samples CURRENT_SAMPLES_DIR /var/lib/oprofile/samples/current CPUTYPE x86-64/family10 BUF_SIZE default value BUF_WATERSHED default value CPU_BUF_SIZE default value SEPARATE_LIB 0 SEPARATE_KERNEL 0 SEPARATE_THREAD 0 SEPARATE_CPU 0 CALLGRAPH 0 VMLINUX none KERNEL_RANGE XENIMAGE none XEN_RANGE Using default event: CPU_CLK_UNHALTED:100000:0:1:1 Error: counter 0 not available # uname -a Linux ada 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 i686 athlon i386 GNU/Linux in /var/log/messages: Apr 2 22:44:54 ada kernel: [40985.251154] [Firmware Bug]: cpu 0, try to setup vector 0x10400, but vector 0xf9 was already reserved by another core, APIC500=0xf9 Apr 2 22:44:54 ada kernel: [40985.251157] [Firmware Bug]: cpu 0, IBS interrupt offset 0 not available (MSRC001103A=0x0000000000000100) Apr 2 22:44:54 ada kernel: [40985.251167] [Firmware Bug]: workaround enabled for IBS LVT offset Apr 2 22:44:54 ada kernel: [40985.251168] oprofile: AMD IBS detected (0x0000001f) Apr 2 22:44:54 ada kernel: [40985.251178] oprofile: using NMI interrupt.
Closing as obsolete. If this is still seen on modern kernels (3.2+) please update/reopen