Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

MPICH 3.2 + Clang 8.0 + OSX Sierra + MPI_Alltoallv = MPID_Request_create segfault #200

Closed
poulson opened this issue Nov 23, 2016 · 3 comments

Comments

@poulson
Copy link
Member

poulson commented Nov 23, 2016

My lldb session is showing a segfault within an MPI_Alltoallv call of the form

Process 67961 stopped
* thread #1: tid = 0x55d6af, 0x00000001040b70e3 libpmpi.12.dylib`MPID_Request_create + 275, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=EXC_I386_GPFLT)
    frame #0: 0x00000001040b70e3 libpmpi.12.dylib`MPID_Request_create + 275
libpmpi.12.dylib`MPID_Request_create:
->  0x1040b70e3 <+275>: movaps %xmm0, 0x230(%rax)
    0x1040b70ea <+282>: movq   $0x0, 0x240(%rax)
    0x1040b70f5 <+293>: movl   $0x2c000000, 0x210(%rax)  ; imm = 0x2C000000 
    0x1040b70ff <+303>: popq   %rbp

This toolchain issue appears to be related to http://stackoverflow.com/questions/38155812/segmentation-fault-while-using-mpi-barrier-in-libpmpi-12-dylib

@jeffhammond
Copy link
Member

See http://lists.mpich.org/pipermail/discuss/2016-May/004764.html and http://lists.mpich.org/pipermail/discuss/2016-August/004861.html for root cause analysis.

@poulson
Copy link
Member Author

poulson commented Nov 23, 2016

It seems that OS X alignment issues strike again.

@poulson
Copy link
Member Author

poulson commented Nov 24, 2016

I have confirmed that building Elemental using the git version of MPICH (compiling using clang) resolves the issue.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants