Bug 2384580

Summary: espresso: FTBFS in Fedora rawhide/f43
Product: [Fedora] Fedora Reporter: Fedora Release Engineering <releng>
Component: espressoAssignee: Jean-Noël Grad <jgrad>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: rawhideCC: jgrad, junghans, orion
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: espresso-4.2.2-12.fc43 Doc Type: ---
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2025-08-01 01:06:07 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 2385779    
Bug Blocks: 2339432, 2373702    
Attachments:
Description Flags
build.log
none
root.log
none
state.log none

Description Fedora Release Engineering 2025-07-30 08:53:24 UTC 
espresso failed to build from source in Fedora rawhide/f43

https://koji.fedoraproject.org/koji/taskinfo?taskID=135166105


For details on the mass rebuild see:

https://fedoraproject.org/wiki/Fedora_43_Mass_Rebuild
Please fix espresso at your earliest convenience and set the bug's status to
ASSIGNED when you start fixing it. If the bug remains in NEW state for 8 weeks,
espresso will be orphaned. Before branching of Fedora 44,
espresso will be retired, if it still fails to build.

For more details on the FTBFS policy, please visit:
https://docs.fedoraproject.org/en-US/fesco/Fails_to_build_from_source_Fails_to_install/
Comment 1 Fedora Release Engineering 2025-07-30 08:53:32 UTC 
Created attachment 2099162 [details]
build.log

file build.log too big, will only attach last 32768 bytes
Comment 2 Fedora Release Engineering 2025-07-30 08:53:36 UTC 
Created attachment 2099163 [details]
root.log

file root.log too big, will only attach last 32768 bytes
Comment 3 Fedora Release Engineering 2025-07-30 08:53:39 UTC 
Created attachment 2099164 [details]
state.log
Comment 4 Jean-Noël Grad 2025-07-30 13:01:50 UTC 
In a release build, running the full testsuite multiple times on ppc64le reveals illegal instructions with OpenMPI 5.0.8-2.fc43:
https://koji.fedoraproject.org/koji/taskinfo?taskID=135481842

In a debug build on aarch64 and ppc64le, a known floating-point accuracy bug surfaces, for which I can easily backport a bugfix from the development version of ESPResSo.

The segmentation fault reported in the build log attached to this issue is not reproducible locally on my workstation in a Docker container, nor on Koji.
```
[ade83cb4db414b0ba6dc8e83be585645:14460:0:14460] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x556d17ab4b60)
[ade83cb4db414b0ba6dc8e83be585645:14466:0:14466] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x5573aedc9b60)
[ade83cb4db414b0ba6dc8e83be585645:14463:0:14463] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x557d38323b60)
[ade83cb4db414b0ba6dc8e83be585645:14458:0:14458] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x55ef9fd19520)
==== backtrace (tid:  14460) ====
 0  /lib64/libucs.so.0(ucs_handle_error+0x2ec) [0x7f292b0d692c]
 1  /lib64/libucs.so.0(+0x1cb6e) [0x7f292b0d9b6e]
 2  /lib64/libucs.so.0(+0x1cd3d) [0x7f292b0d9d3d]
 3  /lib64/libc.so.6(+0x19cd0) [0x7f2b30e26cd0]
 4  /usr/lib64/openmpi/lib/libopen-pal.so.80(mca_btl_sm_poll_handle_frag+0x194) [0x7f292a581844]
 5  /usr/lib64/openmpi/lib/libopen-pal.so.80(+0x82ec4) [0x7f292a586ec4]
 6  /usr/lib64/openmpi/lib/libopen-pal.so.80(opal_progress+0x34) [0x7f292a50e104]
 7  /usr/lib64/openmpi/lib/libmpi.so.40(ompi_comm_nextcid+0x5c) [0x7f292a61e7ac]
 8  /usr/lib64/openmpi/lib/libmpi.so.40(ompi_comm_create_w_info+0x21b) [0x7f292a61448b]
 9  /usr/lib64/openmpi/lib/libmpi.so.40(mca_topo_base_cart_create+0x1b0) [0x7f292a82edf0]
10  /usr/lib64/openmpi/lib/libmpi.so.40(PMPI_Cart_create+0x107) [0x7f292a6596d7]
11  /builddir/build/BUILD/espresso-4.2.2-build/BUILDROOT/usr/lib64/python3.14/site-packages/openmpi/espressomd/Espresso_core.so(_ZN13Communication4initESt10shared_ptrIN5boost3mpi11environmentEE+0xbc) [0x7f292ae4131c]
12  /builddir/build/BUILD/espresso-4.2.2-build/espresso/openmpi/src/python/espressomd/_init.so(+0x2646) [0x7f2b2fb90646]
13  /lib64/libpython3.14.so.1.0(PyModule_ExecDef+0x97) [0x7f2b31265d27]
14  /lib64/libpython3.14.so.1.0(+0x29b3a4) [0x7f2b3129b3a4]
15  /lib64/libpython3.14.so.1.0(_PyEval_EvalFrameDefault+0x600d) [0x7f2b3117411d]
16  /lib64/libpython3.14.so.1.0(+0x1a0705) [0x7f2b311a0705]
17  /lib64/libpython3.14.so.1.0(+0x193141) [0x7f2b31193141]
18  /lib64/libpython3.14.so.1.0(PyObject_CallMethodObjArgs+0x105) [0x7f2b311d3ab5]
19  /lib64/libpython3.14.so.1.0(PyImport_ImportModuleLevelObject+0xe96) [0x7f2b311d3856]
20  /lib64/libpython3.14.so.1.0(+0x1e5bcd) [0x7f2b311e5bcd]
21  /lib64/libpython3.14.so.1.0(_PyEval_EvalFrameDefault+0x600d) [0x7f2b3117411d]
22  /lib64/libpython3.14.so.1.0(+0x1a0705) [0x7f2b311a0705]
23  /lib64/libpython3.14.so.1.0(+0x193141) [0x7f2b31193141]
24  /lib64/libpython3.14.so.1.0(PyObject_CallMethodObjArgs+0x105) [0x7f2b311d3ab5]
25  /lib64/libpython3.14.so.1.0(PyImport_ImportModuleLevelObject+0xc95) [0x7f2b311d3655]
26  /lib64/libpython3.14.so.1.0(_PyEval_ImportName+0xb4) [0x7f2b3125bb74]
27  /lib64/libpython3.14.so.1.0(_PyEval_EvalFrameDefault+0x8f7c) [0x7f2b3117708c]
28  /lib64/libpython3.14.so.1.0(+0x16ab05) [0x7f2b3116ab05]
29  /lib64/libpython3.14.so.1.0(PyEval_EvalCode+0xb7) [0x7f2b3125ab17]
30  /lib64/libpython3.14.so.1.0(+0x279401) [0x7f2b31279401]
31  /lib64/libpython3.14.so.1.0(_PyEval_EvalFrameDefault+0x600d) [0x7f2b3117411d]
32  /lib64/libpython3.14.so.1.0(+0x1a0705) [0x7f2b311a0705]
33  /lib64/libpython3.14.so.1.0(+0x193141) [0x7f2b31193141]
34  /lib64/libpython3.14.so.1.0(PyObject_CallMethodObjArgs+0x105) [0x7f2b311d3ab5]
35  /lib64/libpython3.14.so.1.0(PyImport_ImportModuleLevelObject+0xe96) [0x7f2b311d3856]
36  /lib64/libpython3.14.so.1.0(_PyEval_ImportName+0xb4) [0x7f2b3125bb74]
37  /lib64/libpython3.14.so.1.0(_PyEval_EvalFrameDefault+0x8f7c) [0x7f2b3117708c]
38  /lib64/libpython3.14.so.1.0(+0x16ab05) [0x7f2b3116ab05]
39  /lib64/libpython3.14.so.1.0(PyEval_EvalCode+0xb7) [0x7f2b3125ab17]
40  /lib64/libpython3.14.so.1.0(+0x29a4a7) [0x7f2b3129a4a7]
41  /lib64/libpython3.14.so.1.0(+0x29999a) [0x7f2b3129999a]
42  /lib64/libpython3.14.so.1.0(+0x299254) [0x7f2b31299254]
43  /lib64/libpython3.14.so.1.0(+0x299085) [0x7f2b31299085]
44  /lib64/libpython3.14.so.1.0(Py_RunMain+0x490) [0x7f2b3124ef20]
45  /lib64/libpython3.14.so.1.0(Py_BytesMain+0x3b) [0x7f2b3124907b]
46  /lib64/libc.so.6(+0x35b5) [0x7f2b30e105b5]
47  /lib64/libc.so.6(__libc_start_main+0x88) [0x7f2b30e10668]
48  /usr/bin/python3(_start+0x25) [0x555bebde93d5]
```
The segfault is triggered at random at program startup, where we initialize the MPI communication and the boost::mpi::environment object.
Comment 5 Jean-Noël Grad 2025-07-31 18:18:34 UTC 
I can now confirm ESPResSo works as intended with Python 3.14. The bug reported above is not reproducible on Koji. A bugfix for the completely orthogonal precision loss bug is ready, but cannot be pushed yet, because ESPResSo is affected by 2385779 on ppc64le.
Comment 6 Orion Poplawski 2025-08-01 01:06:07 UTC 
We've fixed this by disabling UCX in openmpi on ppc64le
Comment 7 Jean-Noël Grad 2025-08-01 09:41:21 UTC 
Rebuild was successful: https://bodhi.fedoraproject.org/updates/FEDORA-2025-6a203d2f17