<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<div dir="ltr">
<div>Hello,</div>
<div> I am investigating petsc issue <a href="https://bitbucket.org/petsc/petsc/issues/306/snes-example-ex9-sometimes-fails-in">
306</a>. One can produce the problem with src/snes/examples/tutorials/ex9.c and mpirun -n 3 ./ex9 -snes_grid_sequence 3 -snes_converged_reason -pc_type mg</div>
<div> The program can run to finish, or crash, or hang. The error appears either in PetscGatherMessageLengths or PetscCommBuildTwoSided(). From from my debugging, the following routine is suspicious. It claims not collective, but it might call DMKSPCreate,
which can indirectly call a collective MPI_Comm_dup().</div>
<div> Can someone familiar with the code explain it? Thanks.</div>
<div><br>
</div>
<div> /*@C<br>
</div>
DMGetDMKSPWrite - get write access to private DMKSP context from a DM<br>
<br>
Not Collective<br>
<br>
Input Argument:<br>
. dm - DM to be used with KSP<br>
<br>
Output Argument:<br>
. kspdm - private DMKSP context<br>
<br>
Level: developer<br>
<br>
.seealso: DMGetDMKSP()<br>
@*/<br>
PetscErrorCode DMGetDMKSPWrite(DM dm,DMKSP *kspdm)<br>
{<br>
PetscErrorCode ierr;<br>
DMKSP kdm;<br>
<br>
PetscFunctionBegin;<br>
PetscValidHeaderSpecific(dm,DM_CLASSID,1);<br>
ierr = DMGetDMKSP(dm,&kdm);CHKERRQ(ierr);<br>
if (!kdm->originaldm) kdm->originaldm = dm;<br>
if (kdm->originaldm != dm) { /* Copy on write */<br>
DMKSP oldkdm = kdm;<br>
ierr = PetscInfo(dm,"Copying DMKSP due to write\n");CHKERRQ(ierr);<br>
ierr = DMKSPCreate(PetscObjectComm((PetscObject)dm),&kdm);CHKERRQ(ierr);<br>
ierr = DMKSPCopy(oldkdm,kdm);CHKERRQ(ierr);<br>
ierr = DMKSPDestroy((DMKSP*)&dm->dmksp);CHKERRQ(ierr);<br>
dm->dmksp = (PetscObject)kdm;<br>
}<br>
*kspdm = kdm;<br>
PetscFunctionReturn(0);<br>
}<br>
<div><br>
</div>
<div>The calling stack is</div>
<div>...</div>
<div>#20 0x00007fcf3e3d23e2 in PetscCommDuplicate (comm_in=comm_in@entry=-2080374782, comm_out=comm_out@entry=0x557a18304db0, first_tag=first_tag@entry=0x557a18304de4) at /home/jczhang/petsc/src/sys/objects/tagm.c:162<br>
</div>
<div>#21 0x00007fcf3e3d7730 in PetscHeaderCreate_Private (h=0x557a18304d70, classid=<optimized out>, class_name=class_name@entry=0x7fcf3f7f762a "DMKSP", descr=descr@entry=0x7fcf3f7f762a "DMKSP", mansec=mansec@entry=0x7fcf3f7f762a "DMKSP", comm=comm@entry=-2080374782,
destroy=0x7fcf3f350570 <DMKSPDestroy>, view=0x0) at /home/jczhang/petsc/src/sys/objects/inherit.c:64<br>
#22 0x00007fcf3f3504c9 in DMKSPCreate (comm=-2080374782, kdm=kdm@entry=0x7ffc1d4d00f8) at /home/jczhang/petsc/src/ksp/ksp/interface/dmksp.c:24<br>
#23 0x00007fcf3f35150f in DMGetDMKSPWrite (dm=0x557a18541a10, kspdm=kspdm@entry=0x7ffc1d4d01a8) at /home/jczhang/petsc/src/ksp/ksp/interface/dmksp.c:155<br>
#24 0x00007fcf3f1bb20e in PCSetUp_MG (pc=<optimized out>) at /home/jczhang/petsc/src/ksp/pc/impls/mg/mg.c:682<br>
#25 0x00007fcf3f204bea in PCSetUp (pc=0x557a17dc1860) at /home/jczhang/petsc/src/ksp/pc/interface/precon.c:894<br>
#26 0x00007fcf3f32ba4b in KSPSetUp (ksp=0x557a17d73500) at /home/jczhang/petsc/src/ksp/ksp/interface/itfunc.c:377<br>
#27 0x00007fcf3f41e43e in SNESSolve_VINEWTONRSLS (snes=0x557a17bff210) at /home/jczhang/petsc/src/snes/impls/vi/rs/virs.c:502<br>
#28 0x00007fcf3f3fa191 in SNESSolve (snes=0x557a17bff210, b=0x0, x=<optimized out>) at /home/jczhang/petsc/src/snes/interface/snes.c:4433<br>
#29 0x0000557a16432095 in main (argc=<optimized out>, argv=<optimized out>) at /home/jczhang/petsc/src/snes/examples/tutorials/ex9.c:105<br>
</div>
<div><br>
</div>
<div>
<div>
<div dir="ltr" class="m_-926364523377175141gmail_signature" data-smartmail="gmail_signature">
<div dir="ltr">--Junchao Zhang</div>
</div>
</div>
</div>
</div>
</body>
</html>