Skip to content

rr record is stuck and doesnt proceed  #2627

Description

@rajeshgangam

my rr record is stuck waiting for a response. Not sure what is the issue. Here is backtrace when it is stuck.

Attaching to process 20306
[New LWP 20313]
[New LWP 20317]
[New LWP 20318]
[New LWP 20319]
[New LWP 20320]
[New LWP 20321]
[New LWP 20322]
[New LWP 20323]
[New LWP 20324]
[New LWP 20325]
[New LWP 20327]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f5442f34f7b in __waitpid (pid=30031, stat_loc=0x7ffdb221951c, options=1073741824) at ../sysdeps/unix/sysv/linux/waitpid.c:29
29 ../sysdeps/unix/sysv/linux/waitpid.c: No such file or directory.
(gdb) bt
#0 0x00007f5442f34f7b in __waitpid (pid=30031, stat_loc=0x7ffdb221951c, options=1073741824) at ../sysdeps/unix/sysv/linux/waitpid.c:29
#1 0x00000000005d31ee in rr::Task::wait(double) ()
#2 0x00000000005b436f in rr::Scheduler::reschedule(rr::Switchable) ()
#3 0x00000000005400ec in rr::RecordSession::record_step() ()
#4 0x0000000000535e21 in rr::RecordCommand::run(std::vector<std::__cxx11::basic_string<char, std::char_traits, std::allocator >, std::allocator<std::__cxx11::basic_string<char, std::char_traits, std::allocator > > >&) ()
#5 0x00000000004ac743 in main ()
(gdb) n
31 in ../sysdeps/unix/sysv/linux/waitpid.c
(gdb) c
Continuing.

Activity

  1. khuey commented on Jul 29, 2020

    @khuey
    Collaborator

    What is the tracee program doing when things get stuck?

  2. rajeshgangam commented on Jul 29, 2020

    @rajeshgangam
    Author

    Thanks kyle for the response. I think the tracee program isdoing mem-write to a pci-e mapped device. Not sure of exact function yet. I am looking into it though. problem is rr has already debugging with ptrace, so I am not able to attach the gdb to my process.

  3. khuey commented on Jul 29, 2020

    @khuey
    Collaborator

    If you have rr in gdb you can call emergency_debug(<the correct Task* here>) which will stop tracee execution and switch rr to a gdbserver mode (which you can then connect to with a different gdb and inspect the tracee state).

    Running rr record up to the hang with RR_LOG=all:debug might be interesting too.

  4. GitMensch commented on Jul 16, 2021

    @GitMensch
    Contributor

    Looks like #2640 is a duplicate of this. I'm_likely_ able to run the requested commands and logging on Monday... (still happening with RR 5.4.0).

  5. added a commit that references this issue on Jul 22, 2021
    bef9803
  6. GitMensch commented on Jul 22, 2021

    @GitMensch
    Contributor

    @rajeshgangam Is there any chance that you can query uname -a for this machine or remember what distro that was?

  7. GitMensch commented on Oct 28, 2021

    @GitMensch
    Contributor

    As there was no response and #2640 is a very likely duplicate I suggest to close this issue as duplicate.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions