13:47 <irker481> systemtap: mcermak systemtap.git:refs/heads/master * release-3.2-152-g09cd42d / systemtap.spec: Fix the naming convention problem for python-setuptools. http://tinyurl.com/ybyouyum

16:15 <irker481> systemtap: smakarov systemtap.git:refs/heads/master * release-3.2-153-g81e03c8 / elaborate.cxx staptree.cxx staptree.h testsuite/runok/regex_noelide.stp: PR22193: don't elide regex match '=~' operator since it's now side-effecting http://tinyurl.com/yalwzdcd

22:09 <pk34663> hi, are there any known issues with stack unwinding on 4.16 kernels? i'm currently getting

22:09 <pk34663> running /opt/bin/stap -v -DDEBUG_UNWIND=3 -e 'probe kernel.function("invalidate_mapping_pages") {printf("stack:%s\n", print_backtrace());}'

22:09 <pk34663> when

22:09 <pk34663> unwind:1533: unwinding failed: -5

22:09 <pk34663> unwind_frame:1178: Module /usr/lib/debug/lib/modules/4.16.0-rc3+/vmlinux: no unwind frame data

22:10 <fche> we've received a few reports, yea

22:11 <fche> haven't been able to narrow down the cause; there may be multiple issues

22:12 <fche> https://sourceware.org/ml/systemtap/2018-q1/msg00090.html <-- e.g.

22:12 <fche> https://bugzilla.redhat.com/show_bug.cgi?id=1552745 <--- e.g.

22:15 <pk34663> thanks. i'll try building without CONFIG_RANDOMIZE_BASE set to y.

22:16 <pk34663> if there's anything i can try to narrow down the root cause let me know

22:31 <fche> there were earlier indications that retpoline or such machinery, built with old compilers, could also mess things up

22:40 <pk34663> im building with gcc version 5.4.0, i'll try upgrading to the latest as well as building without randomize_base

22:40 <fche> thanks for trying it out

22:40 <fche> we'll try to get some focus on it shortly

22:41 <pk34663> np

22:51 <pk34663> without CONFIG_RANDOMIZE_BASE=y i certainly get more of a stack trace, not sure it's a complete stack but when i trace inside the ip stack it definitely looks better. i'll try upgrading gcc tomorrow and see what results i get then.

22:51 <fche> ok, that's a valuable tip

22:52 <fche> if you could send your findings to the mailing list, it'll help put together a case

23:04 <pk34663> np i just sent an email to the mailing list, tomorrow i'll try rebuilding my kernel with a more recent gcc

23:13 <pk34663> actually looking through the stacks i get when tracing invalidate_mapping_pages they look pretty good now without CONFIG_RANDOMIZE_BASE set to y

23:22 <fche> thanks a lot

23:22 <fche> that's good news

23:24 <pk34663> i'll try using gcc-7 and CONFIG_RANDOMIZE_BASE=y next