From: William C. <wc...@re...> - 2020-07-21 13:49:01
|
On 7/20/20 9:03 PM, J Lumby wrote: > Oprofiling an intensive postgresql workload running on a Lenovo P52 and lniux kernel 5.7.4, I see very high numbers of both of these : approximately two-thirds of all user-space samples are being lost. see operf.log below > > > The operf command I've used is > > operf --vmlinux /mnt/bullbild/linux-5.7.4-bullarch/vmlinux --session-dir /tmp/oprof_session.200720-195024 --events cpu_clk_unhalted:30000045:thread:1:1 --callgraph --separate-cpu --system-wide > > > and it works but tells me > > WARNING: Lost samples detected! See /tmp/oprof_session.200720-195024/samples/operf.log for details. > > > Lowering the sampling rate as suggested does not help : I tried with a count of 99999999 and same high lost samples. > > > I assume oprofile somehow is unable to map the instruction pointer of the event to any process? How does it do this? Is there any setting or control or hint that I can set to help it? or , failing that, is there any way I can ask opreport to give me the results as raw instruction addresses? > > > Help! Hi, Which version of oprofile was being used? I just did a release of oprofile-1.4.0 it has commit [a3742f]: Use the mmap offset to correctly compute the IP location in a file Newer versions of binutils are now now splitting the file into multiple mmap loads. The assumption that the mmap for the executable code in the file starts at the beginning of the file (that the file offset is always zero) is no longer true. The computation to convert the IP address into an offset needs to also use the offset. I wonder if maybe that a number of the samples might be discarded because they are not being mapped to reasonable locations in executables. Could you give that new release of oprofile a try? -Will > > > Cheers, John Lumby > > ___________________________________________________________ > > Profiling started at Mon Jul 20 19:50:24 2020 > Profiling stopped at Mon Jul 20 19:51:25 2020 > > -- OProfile/operf Statistics -- > Nr. non-backtrace samples: 9435 > Nr. kernel samples: 1924 > Nr. user space samples: 7511 > Nr. samples lost due to sample address not in expected range for domain: 0 > Nr. lost kernel samples: 0 > Nr. samples lost due to sample file open failure: 0 > Nr. samples lost due to no permanent mapping: 4967 > Nr. user context kernel samples lost due to no app info available: 0 > Nr. user samples lost due to no app info available: 0 > Nr. backtraces skipped due to no file mapping: 4967 > Nr. hypervisor samples dropped due to address out-of-range: 0 > Nr. samples lost reported by perf_events kernel: 0 > > > > _______________________________________________ > oprofile-list mailing list > opr...@li... > https://lists.sourceforge.net/lists/listinfo/oprofile-list |