On Tuesday 29 January 2013 10:35 PM, James Hogan wrote: > Hi Vineet, > > You don't appear to define CONFIG_HW_PERF_EVENTS, so > include/linux/oprofile.h will presumably define oprofile_perf_init as > just a pr_info(...); return -ENODEV; > > Similarly drivers/oprofile/oprofile_perf.o doesn't seem to be being built. > > I'm probably missing something somewhere? Not much :-) oprofile_arch_init() failure causes oprofile to fall back to timer based PC only sampling - for coarse grained profiling. I'll soon be starting on integratign the hardware counter support to both oprofile/perf. FWIW, I just re-verified that latest upstream OProfile works fine with 3.8 kernel. [ARCLinux]$ uname -a Linux ARCLinux 3.8.0-rc4+ #2 Wed Jan 30 11:13:21 IST 2013 arc GNU/Linux [ARCLinux]$ ./opcontrol -v opcontrol: oprofile 0.9.9git compiled on Jan 30 2013 11:40:15 [ARCLinux]$ ./opcontrol --dump [ARCLinux]$ ./opreport -l /mnt/arc/next-linux/vmlinux Using /var/lib/oprofile/samples/ for samples directory. CPU: CPU with timer interrupt Profiling through timer interrupt samples % symbol name 856 69.3679 cpu_idle 61 4.9433 arch_local_irq_enable .... Thx for taking a look and if you are content - a Reviewed-by please (and for DeviceTree patch as well if that looks sane now). I'll start doing the same for metag patches after this one bug I'm debugging right now. -Vineet -- To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
