Hui Zhou

> I just now did a quick check and found that all ADIO drivers are not doing communication in their ADIOI_xxx_Close, except for DAOS, in which ADIOI_DAOS_Close does call MPI_Barrier...

> @hzhou Is this a known issue to you? Thanks for reporting. Since it is libfabric issue, it is of lower priority for us. Is it important to you to...

The assertion is inside a macro: https://github.com/pmodels/mpich/blob/00b8129600cf3b89c15e319f5a5e97c233f1c2d9/src/include/mpir_datatype.h#L277 The assertions should be turned off with configure option `--enable-fast=ndebug` or `--enable-fast=all`. Please contact Rocky Linux maintainer for how its packages are built.

Assertions are not supposed to replace run-time error checking. @the-german-guy I am closing this issue for now. But if you have a reproducer, please reopen or create a new issue...

I guess we can generate `mpich-c.pc`, `mpich-cxx.pc`, and `mpich-fort.pc`. Then it is up to distribution to create links to `mpi-c.pc`, `mpi-cxx.pc`, and `mpi-fort.pc`.

@lykke-li Thanks you for reporting the issue. I recall you mentioned that the current `intra_tree` algorithm will run into failures when configured with `--disable-error-checking`, right?

Oh, I see. Your modification *relies* on receiving different amounts of data from the original size, thus you need to disable the error checking, right?

I see. So it is the IPC path. We need to make IPC GPU copy nonblocking.

@jxy Thanks. Could you confirm that the blocking nature of `MPI_Test` is observed for both intra-node and inter-node? I'll focus on the intra-node first and ping you for testing when...

@psteinbrecher @jxy This PR https://github.com/pmodels/mpich/pull/6841 should fix the blocking issues of MPI_Test at least for contiguous datatype intra-node. Could you test?