Maybe the distribution tar ball at https://download.open-mpi.org/release/open-mpi/v3.1/openmpi-3.1.3.tar.gz
did not get refreshed after the fix in https://github.com/bosilca/ompi/commit/b902cd5eb765ada57f06c75048509d0716953549 was implemented? I downloaded the tarball from open-mpi.org today, 22 Dec, and compiled and I get the warnings. ibv_exp_query_device: invalid comp_mask !!! (comp_mask = 0xd8000000002 valid_mask = 0x1) [bn01][[37143,17005],0][btl_openib_component.c:1670:init_one_device] error obtaining device attributes for mlx4_0 errno says Invalid argument ibv_exp_query_device: invalid comp_mask !!! (comp_mask = 0xd8100000002 valid_mask = 0x1) [bn01][[37143,17005],1][btl_openib_component.c:1670:init_one_device] error obtaining device attributes for mlx4_0 errno says Invalid argument -------------------------------------------------------------------------- WARNING: There was an error initializing an OpenFabrics device. Local host: bn01 Local device: mlx4_0 -------------------------------------------------------------------------- -------------------------------------------------------------------------- WARNING: There was an error initializing an OpenFabrics device. Local host: bn01 Local device: mlx4_0 -------------------------------------------------------------------------- It looks like Howard merged the fix on Dec 4, but the date listed for the 3.1.3 tarball on the open-mpi.org site is in Oct. Relevant lines in opal/mca/btl/openib/btl_openib_component.c from the tar ball are these. Missing the memset(&device->ib_exp_dev_attr, 0, sizeof(device->ib_exp_dev_attr)); that should have been inserted at 1667. 1666 #if HAVE_DECL_IBV_EXP_QUERY_DEVICE 1667 device->ib_exp_dev_attr.comp_mask = IBV_EXP_DEVICE_ATTR_RESERVED - 1; 1668 if(ibv_exp_query_device(device->ib_dev_context, &device->ib_exp_dev_att r)){ 1669 BTL_ERROR(("error obtaining device attributes for %s errno says %s" , 1670 ibv_get_device_name(device->ib_dev), strerror(errno))); 1671 goto error; 1672 } 1673 #endif I added a comment to the GitHub issue, but it was closed and I am not sure that will be noticed. Sorry for the double-posting if that was sufficient. -- bennet _______________________________________________ users mailing list users@lists.open-mpi.org https://lists.open-mpi.org/mailman/listinfo/users
