[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-bugs-dist
Subject:    [valgrind] [Bug 332876] New: valgrind (memcheck) hangs intermittently
From:       Bill Torpey <btorpey () nyx ! com>
Date:       2014-03-31 19:09:48
Message-ID: bug-332876-17878 () http ! bugs ! kde ! org/
[Download RAW message or body]

https://bugs.kde.org/show_bug.cgi?id=332876

            Bug ID: 332876
           Summary: valgrind (memcheck) hangs intermittently
    Classification: Unclassified
           Product: valgrind
           Version: 3.9.0
          Platform: RedHat RPMs
                OS: Linux
            Status: UNCONFIRMED
          Severity: major
          Priority: NOR
         Component: memcheck
          Assignee: jseward@acm.org
          Reporter: btorpey@nyx.com

valgrind/memcheck appears to hang intermittently. 

This is a new problem w/3.8.1, 3.9.0.  We did not observe this with earlier
versions of valgrind (up to version 3.7.0).

Some observations:

- the problem appears to be intermittent: sometimes the same program with the
same options, running under valgrind with the same options, will run to
completion and produce a report, sometimes it will hang "forever"

- the program makes heavy use of threads

- when the program hangs, strace will report either of the following:
Process 28369 attached - interrupt to quit
futex(0x802006914, FUTEX_WAIT_PRIVATE, 25, NULL

- when the program hangs, it will use 90-100% CPU, even though it appears to
have no activity

- when I attach to the process with gdb, it appears that there is only a single
thread, with a call stack similar to following (athough that changes depending
on where the hang occurs):
(gdb) bt
#0  0x0000000038081aac in run_thread_for_a_while (two_words=0x80947ff50, 
    dispatchCtrP=0x80947ff7c, tid=1, alt_host_addr=0, 
    use_alt_host_addr=0 '\000') at m_scheduler/scheduler.c:952
#1  0x0000000038083785 in vgPlain_scheduler (tid=1)
    at m_scheduler/scheduler.c:1267
#2  0x00000000380b1de6 in thread_wrapper (tidW=1)
    at m_syswrap/syswrap-linux.c:103
#3  run_a_thread_NORETURN (tidW=1) at m_syswrap/syswrap-linux.c:156
#4  0x0000000000000000 in ?? ()
(gdb) info threads
  Id   Target Id         Frame 
* 1    process 2901 "memcheck-amd64-" 0x0000000038081aac in
run_thread_for_a_while (two_words=0x80947ff50, dispatchCtrP=0x80947ff7c, tid=1,
alt_host_addr=0, 
    use_alt_host_addr=0 '\000') at m_scheduler/scheduler.c:952
(gdb) 

- I've tried with both settings for -fair-sched -- there doesn't appear to be
much, if any, difference in results

Any insights would be much appreciated.  Please let me know if additional
information would be helpful.

Regards,

Bill Torpey

-- 
You are receiving this mail because:
You are watching all bug changes.
[prev in list] [next in list] [prev in thread] [next in thread]