<html><head><meta http-equiv="Content-Type" content="text/html; charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class=""><div class=""><br class=""></div>  I don't know why valgrind is not putting in line numbers but something is clearly very wrong with data structure initialization. I am not sure if it is really directly related to PETSc but somehow comes about in the C++ constructor business, which seems simple enough in your code but I don't understand. When working with MPI and C++ one must be careful that no constructors of global objects that reference MPI get called before MPI is initialized but that doesn't seem to be happening in your code.<div class=""><br class=""></div><div class="">  Sorry, I can debug it but I have no clue how to build your code it seems to use  "armadillo-matrices" stuff that somehow doesn't come in through any include files or libraries. If you gave me a full makefile that builds everything I might be able to build it.<div><br class=""><blockquote type="cite" class=""><div class="">On Apr 1, 2022, at 9:56 AM, Roland Richter <<a href="mailto:roland.richter@ntnu.no" class="">roland.richter@ntnu.no</a>> wrote:</div><br class="Apple-interchange-newline"><div class="">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" class="">
  
  <div class=""><p class="">Hei,</p><p class="">It defines the namespace used for armadillo-matrices. A
      valgrind-log is attached for the case with removed destructor
      (can't run valgrind if my program fails with a segfault).<br class="">
    </p><p class="">Regards,</p><p class="">Roland<br class="">
    </p>
    <div class="moz-cite-prefix">Am 01.04.22 um 15:23 schrieb Barry
      Smith:<br class="">
    </div>
    <blockquote type="cite" cite="mid:6C65CC1D-365D-43AE-B159-AFB3DCF3D4C0@petsc.dev" class="">
      
      <div class=""><br class="">
      </div>
        I recommend first running with valgrind. 
      <div class=""><br class="">
      </div>
      <div class="">  I tried to build your code but got compile errors
        from arma:: being unknown. Where does it come from? Is it only
        in a super new version of Boost?</div>
      <div class=""><br class="">
      </div>
      <div class=""><br class="">
        <div class=""><br class="">
          <blockquote type="cite" class="">
            <div class="">On Apr 1, 2022, at 6:50 AM, Roland Richter
              <<a href="mailto:roland.richter@ntnu.no" class="moz-txt-link-freetext" moz-do-not-send="true">roland.richter@ntnu.no</a>>
              wrote:</div>
            <br class="Apple-interchange-newline">
            <div class="">
              <div class=""><p class="">I re-run my code with a debug version of
                  PETSc, resulting in <br class="">
                </p><p class=""><span style="font-family:monospace" class=""><span style="font-weight:bold;color:#ff5454;background-color:#ffffff;" class="">[0]PETSC ERROR: ---------------------
                      Error Message
                      --------------------------------------------------------------</span><span style="background-color: rgb(255, 255, 255);" class=""> </span><br class="">
                    [0]PETSC ERROR: Corrupt argument: <a class="moz-txt-link-freetext" href="https://petsc.org/release/faq/#valgrind" moz-do-not-send="true">https://petsc.org/release/faq/#valgrind</a>
                    <br class="">
                    [0]PETSC ERROR: Object already free: Parameter # 1 <br class="">
                    [0]PETSC ERROR: See <a class="moz-txt-link-freetext" href="https://petsc.org/release/faq/" moz-do-not-send="true">https://petsc.org/release/faq/</a>
                    for trouble shooting. <br class="">
                    [0]PETSC ERROR: Petsc Development GIT revision:
                    v3.17.0-8-gf6d6129e50  GIT Date: 2022-03-31 18:10:33
                    +0000 <br class="">
                    [0]PETSC ERROR: #1 VecGetSize() at
/home/roland/Downloads/git-files/petsc/src/vec/vec/interface/vector.c:670<br class="">
                    0       64 <br class="">
                    [0]PETSC ERROR: #2 VecDestroy() at
/home/roland/Downloads/git-files/petsc/src/vec/vec/interface/vector.c:375<br class="">
                    0       64 <br class="">
                    [0]PETSC ERROR: #3 VecGetArray() at
/home/roland/Downloads/git-files/petsc/src/vec/vec/interface/rvector.c:1780<br class="">
                    <br class="">
                  </span></p><p class="">I do not understand why it tries to access
                  the vector, even though it has been set to PETSC_NULL
                  in the previous free-call.</p><p class="">Regards,</p><p class="">Roland<br class="">
                </p>
                <div class="moz-cite-prefix">Am 31.03.22 um 15:50
                  schrieb Matthew Knepley:<br class="">
                </div>
                <blockquote type="cite" cite="mid:CAMYG4GnoJnQmGeCTiBZoN6QMXw5bNm+EbM0-P_DzM2S45UyxVQ@mail.gmail.com" class="">
                  <div dir="ltr" class="">
                    <div dir="ltr" class="">On Thu, Mar 31, 2022 at 9:47
                      AM Roland Richter <<a href="mailto:roland.richter@ntnu.no" moz-do-not-send="true" class="moz-txt-link-freetext">roland.richter@ntnu.no</a>>
                      wrote:<br class="">
                    </div>
                    <div class="gmail_quote">
                      <blockquote class="gmail_quote" style="margin:0px
                        0px 0px 0.8ex;border-left:1px solid
                        rgb(204,204,204);padding-left:1ex">
                        <div class=""><p class="">The backtrace is</p><p class=""><span style="font-family:monospace" class=""><span style="background-color: rgb(255, 255,
                                255);" class="">#0  </span><span style="color:rgb(24,24,178);background-color:rgb(255,255,255)" class="">0x00007fffeec4ba97</span><span style="background-color: rgb(255, 255,
                                255);" class=""> in </span><span style="color:rgb(178,104,24);background-color:rgb(255,255,255)" class="">VecGetSize_Seq</span><span style="background-color: rgb(255, 255,
                                255);" class=""> () from </span><span style="color:rgb(24,178,24);background-color:rgb(255,255,255)" class="">/opt/petsc/lib/libpetsc.so.3.016</span><span style="background-color: rgb(255, 255,
                                255);" class=""> </span><br class="">
                              #1  <span style="color:rgb(24,24,178);background-color:rgb(255,255,255)" class="">0x00007fffeec78f5a</span><span style="background-color: rgb(255, 255,
                                255);" class=""> in </span><span style="color:rgb(178,104,24);background-color:rgb(255,255,255)" class="">VecGetSize</span><span style="background-color: rgb(255, 255,
                                255);" class=""> () from </span><span style="color:rgb(24,178,24);background-color:rgb(255,255,255)" class="">/opt/petsc/lib/libpetsc.so.3.016</span><span style="background-color: rgb(255, 255,
                                255);" class=""> </span><br class="">
                              #2  <span style="color:rgb(24,24,178);background-color:rgb(255,255,255)" class="">0x0000000000410b73</span><span style="background-color: rgb(255, 255,
                                255);" class=""> in </span><span style="color:rgb(178,104,24);background-color:rgb(255,255,255)" class="">test_ts_arma_with_pure_petsc_preconfigured_clean(unsigned
                                long, unsigned long,
                                arma::Col<std::complex<doubl</span><br class="">
                              e> > const&,
                              arma::Col<std::complex<double>
                              >&, double, double, double) [clone
                              .constprop.0]<span style="background-color: rgb(255, 255,
                                255);" class=""> () </span><br class="">
                              #3  <span style="color:rgb(24,24,178);background-color:rgb(255,255,255)" class="">0x0000000000414384</span><span style="background-color: rgb(255, 255,
                                255);" class=""> in </span><span style="color:rgb(178,104,24);background-color:rgb(255,255,255)" class="">test_RK4_solvers_clean(unsigned
                                long, unsigned long, unsigned long,
                                bool) [clone .constprop.0]</span><span style="background-color: rgb(255, 255,
                                255);" class=""> () </span><br class="">
                              #4  <span style="color:rgb(24,24,178);background-color:rgb(255,255,255)" class="">0x0000000000405c6c</span><span style="background-color: rgb(255, 255,
                                255);" class=""> in </span><span style="color:rgb(178,104,24);background-color:rgb(255,255,255)" class="">main</span><span style="background-color: rgb(255, 255,
                                255);" class=""> ()</span></span></p>
                        </div>
                      </blockquote>
                      <div class="">It looks like you are passing an
                        invalid vector. If you compiled in debug mode,
                        it would tell you. I would run</div>
                      <div class="">in debug until my code was running
                        like I expect, then switch to optimized. You can
                        do that by using two</div>
                      <div class="">different PETSC_ARCH configures, and
                        switch at runtime with that variable.</div>
                      <div class=""><br class="">
                      </div>
                      <div class="">  Thanks,</div>
                      <div class=""><br class="">
                      </div>
                      <div class="">    Matt </div>
                      <blockquote class="gmail_quote" style="margin:0px
                        0px 0px 0.8ex;border-left:1px solid
                        rgb(204,204,204);padding-left:1ex">
                        <div class=""><p class="">Regards,</p>
                          <div class="">Roland Richter</div>
                          <div class=""><br class="">
                          </div>
                          <div class="">Am 31.03.22 um 15:35 schrieb
                            Matthew Knepley:<br class="">
                          </div>
                          <blockquote type="cite" class="">
                            <div dir="ltr" class="">
                              <div dir="ltr" class="">On Thu, Mar 31,
                                2022 at 9:01 AM Roland Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank" moz-do-not-send="true" class="moz-txt-link-freetext">roland.richter@ntnu.no</a>>
                                wrote:<br class="">
                              </div>
                              <div class="gmail_quote">
                                <blockquote class="gmail_quote" style="margin:0px 0px 0px
                                  0.8ex;border-left:1px solid
                                  rgb(204,204,204);padding-left:1ex">
                                  <div class=""><p class="">Hei,</p><p class="">Thanks for the idea! I
                                      added a simple std::cout for both
                                      constructor and destructor, and
                                      found out that my destructor is
                                      called multiple times, while the
                                      constructor is called only once.
                                      This could explain the error
                                      (double free), but I do not know
                                      why segfault is thrown even though
                                      I explicitly check if the vector
                                      has been used. Are there
                                      explanations for that?</p>
                                  </div>
                                </blockquote>
                                <div class="">Run with
                                  -start_in_debugger and get the stack
                                  trace when it faults. Right now, I
                                  have no idea where it is faulting.</div>
                                <div class=""><br class="">
                                </div>
                                <div class="">  Thanks,</div>
                                <div class=""><br class="">
                                </div>
                                <div class="">    Matt</div>
                                <div class=""> <br class="">
                                </div>
                                <blockquote class="gmail_quote" style="margin:0px 0px 0px
                                  0.8ex;border-left:1px solid
                                  rgb(204,204,204);padding-left:1ex">
                                  <div class=""><p class="">Regards,</p><p class="">Roland Richter<br class="">
                                    </p>
                                    <div class="">Am 31.03.22 um 12:14
                                      schrieb Matthew Knepley:<br class="">
                                    </div>
                                    <blockquote type="cite" class="">
                                      <div dir="ltr" class="">
                                        <div dir="ltr" class="">On Thu,
                                          Mar 31, 2022 at 5:58 AM Roland
                                          Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank" moz-do-not-send="true" class="moz-txt-link-freetext">roland.richter@ntnu.no</a>>
                                          wrote:<br class="">
                                        </div>
                                        <div class="gmail_quote">
                                          <blockquote class="gmail_quote" style="margin:0px 0px 0px
                                            0.8ex;border-left:1px solid
rgb(204,204,204);padding-left:1ex">
                                            <div class=""><p class="">Hei,</p><p class="">For a project
                                                I wanted to combine
                                                boost::odeint for
                                                timestepping and
                                                PETSc-based vectors and
                                                matrices for calculating
                                                the right hand side. As
                                                comparison for both
                                                timing and correctness I
                                                set up an
                                                armadillo-based right
                                                hand side (with the
                                                main-function being in <b class="">main.cpp</b>,
                                                and the test code in <b class="">test_timestepping_clean.cpp</b>)<br class="">
                                              </p><p class="">In theory, the
                                                code works fine, but I
                                                have some issues with
                                                cleaning up afterwards
                                                in my struct <i class="">Petsc_RHS_state_clean</i>.
                                                My initial intention was
                                                to set up all involved
                                                matrices and vectors
                                                within the constructor,
                                                and free the memory in
                                                the destructor. To avoid
                                                freeing vectors I have
                                                not used I initially set
                                                them to <i class="">PETSC_NULL</i>,
                                                and check if this value
                                                has been changed before
                                                calling <i class="">VecDestroy().</i></p>
                                            </div>
                                          </blockquote>
                                          <div class="">You do not need
                                            to check. Destroy()
                                            functions already check for
                                            NULL.</div>
                                          <blockquote class="gmail_quote" style="margin:0px 0px 0px
                                            0.8ex;border-left:1px solid
rgb(204,204,204);padding-left:1ex">
                                            <div class=""><p class=""> However, when
                                                doing that I get the
                                                following error:</p><p class=""><span style="font-family:monospace" class=""><span style="background-color:
                                                    rgb(255, 255, 255);" class="">[0]PETSC
                                                    ERROR:
                                                    ------------------------------------------------------------------------
                                                  </span><br class="">
                                                  [0]PETSC ERROR: Caught
                                                  signal number 11 SEGV:
                                                  Segmentation
                                                  Violation, probably
                                                  memory access out of
                                                  range <br class="">
                                                  [0]PETSC ERROR: Try
                                                  option
                                                  -start_in_debugger or
-on_error_attach_debugger <br class="">
                                                  [0]PETSC ERROR: or see
                                                  <a href="https://petsc.org/release/faq/#valgrind" target="_blank" moz-do-not-send="true" class="moz-txt-link-freetext">https://petsc.org/release/faq/#valgrind</a>
                                                  <br class="">
                                                  [0]PETSC ERROR: or try
                                                  <a href="http://valgrind.org/" target="_blank" moz-do-not-send="true" class="moz-txt-link-freetext">http://valgrind.org</a> on GNU/linux and
                                                  Apple Mac OS X to find
                                                  memory corruption
                                                  errors <br class="">
                                                  [0]PETSC ERROR:
                                                  configure using
                                                  --with-debugging=yes,
                                                  recompile, link, and
                                                  run  <br class="">
                                                  [0]PETSC ERROR: to get
                                                  more information on
                                                  the crash. <br class="">
                                                  <span style="font-weight:bold;color:rgb(255,84,84);background-color:rgb(255,255,255)" class="">[0]PETSC
                                                    ERROR:
                                                    ---------------------
                                                    Error Message
--------------------------------------------------------------</span></span></p><p class="">If I comment
                                                out that code in
                                                ~Petsc_RHS_state_clean(),
                                                the program runs, but
                                                will use ~17 GByte of
                                                RAM during runtime. As
                                                the memory is not used
                                                immediately in full, but
                                                rather increases during
                                                running, I assume a
                                                memory leak somewhere.
                                                Where does it come from,
                                                and how can I avoid it?</p>
                                            </div>
                                          </blockquote>
                                          <div class="">It must be that
                                            your constructor is called
                                            multiple times without
                                            calling your destructor. I
                                            cannot understand this code
                                            in order</div>
                                          <div class="">to see where
                                            that happens, but you should
                                            just be able to run in the
                                            debugger and put a break
                                            point at the creation and<br class="">
                                          </div>
                                          <div class="">destruction
                                            calls.</div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class="">  Thanks,</div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class="">      Matt</div>
                                          <blockquote class="gmail_quote" style="margin:0px 0px 0px
                                            0.8ex;border-left:1px solid
rgb(204,204,204);padding-left:1ex">
                                            <div class=""><p class="">Thanks!</p><p class="">Regards,</p><p class="">Roland Richter<br class="">
                                              </p>
                                              <div class=""><span style="font-family:monospace" class=""></span><br class="webkit-block-placeholder">
                                              </div>
                                            </div>
                                          </blockquote>
                                        </div>
                                        <br class="" clear="all">
                                        <div class=""><br class="">
                                        </div>
                                        -- <br class="">
                                        <div dir="ltr" class="">
                                          <div dir="ltr" class="">
                                            <div class="">
                                              <div dir="ltr" class="">
                                                <div class="">
                                                  <div dir="ltr" class="">
                                                    <div class="">What
                                                      most experimenters
                                                      take for granted
                                                      before they begin
                                                      their experiments
                                                      is infinitely more
                                                      interesting than
                                                      any results to
                                                      which their
                                                      experiments lead.<br class="">
                                                      -- Norbert Wiener</div>
                                                    <div class=""><br class="">
                                                    </div>
                                                    <div class=""><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank" moz-do-not-send="true" class="">https://www.cse.buffalo.edu/~knepley/</a><br class="">
                                                    </div>
                                                  </div>
                                                </div>
                                              </div>
                                            </div>
                                          </div>
                                        </div>
                                      </div>
                                    </blockquote>
                                  </div>
                                </blockquote>
                              </div>
                              <br class="" clear="all">
                              <div class=""><br class="">
                              </div>
                              -- <br class="">
                              <div dir="ltr" class="">
                                <div dir="ltr" class="">
                                  <div class="">
                                    <div dir="ltr" class="">
                                      <div class="">
                                        <div dir="ltr" class="">
                                          <div class="">What most
                                            experimenters take for
                                            granted before they begin
                                            their experiments is
                                            infinitely more interesting
                                            than any results to which
                                            their experiments lead.<br class="">
                                            -- Norbert Wiener</div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class=""><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank" moz-do-not-send="true" class="">https://www.cse.buffalo.edu/~knepley/</a><br class="">
                                          </div>
                                        </div>
                                      </div>
                                    </div>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </blockquote>
                        </div>
                      </blockquote>
                    </div>
                    <br class="" clear="all">
                    <div class=""><br class="">
                    </div>
                    -- <br class="">
                    <div dir="ltr" class="gmail_signature">
                      <div dir="ltr" class="">
                        <div class="">
                          <div dir="ltr" class="">
                            <div class="">
                              <div dir="ltr" class="">
                                <div class="">What most experimenters
                                  take for granted before they begin
                                  their experiments is infinitely more
                                  interesting than any results to which
                                  their experiments lead.<br class="">
                                  -- Norbert Wiener</div>
                                <div class=""><br class="">
                                </div>
                                <div class=""><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank" moz-do-not-send="true" class="">https://www.cse.buffalo.edu/~knepley/</a><br class="">
                                </div>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </blockquote>
              </div>
            </div>
          </blockquote>
        </div>
        <br class="">
      </div>
    </blockquote>
  </div>

<span id="cid:CFA2F8E1-BF13-4EF8-B340-A2FA12F963FB"><out_file.txt></span></div></blockquote></div><br class=""></div></body></html>