Control: notfixed -1 sphinx/9.1.0-4 Control: reassign -1 python3.13 3.13.5-2+deb13u3 Control: tags -1 - pending Control: tags 1141977 + trixie Control: affects -1 - src:python3.13 Control: affects 1141977 src:blender-doc src:cp2k src:flint src:groonga src:kitty src:openmpi Control: merge 1141977 -1
On Fri, Jul 17, 2026 at 12:54:35AM +0300, Dmitry Shachnev wrote: > Well, the segfaults are a python3.13 bug. But we still don't know about the > EOFError: it appeared recently so it may be related to python3.13 update, but > it may still be a bug in Sphinx. > > (I still did not have time to look deeper, sorry.) Actually I have just tried to build openmpi on your second VM, and I see in dmesg output that we have a segfault here too. [ 516.735856] sphinx-build[247755]: segfault at 1 ip 00000000005739c3 sp 00007ffeca2084c0 error 4 in python3.13[1739c3,420000+31f000] likely on CPU 1 (core 1, socket 0) [ 516.742635] Code: 00 00 00 00 00 66 66 2e 0f 1f 84 00 00 00 00 00 66 66 2e 0f 1f 84 00 00 00 00 00 66 66 2e 0f 1f 84 00 00 00 00 00 90 48 89 f1 <48> 8b 3e 48 8b 51 10 48 8b 41 08 48 89 fe 85 d2 0f 88 c2 6d ec ff [ 2630.616676] systemd-ssh-generator[248529]: Failed to query local AF_VSOCK CID: Cannot assign requested address [85928.309361] sphinx-build[495026]: segfault at ffffffffffffffff ip 000000000059a56a sp 00007ffc6d59b4e0 error 5 in python3.13[19a56a,420000+31f000] likely on CPU 0 (core 0, socket 0) [85928.314974] Code: 01 00 00 0f 87 b5 01 00 00 8d 75 ff c1 ee 04 8d 04 36 49 8b 04 c7 48 8b 78 10 48 39 f8 0f 84 1f 02 00 00 48 8b 58 08 83 00 01 <48> 8b 0b 48 89 48 08 48 85 c9 0f 84 f4 00 00 00 8b 82 10 1d 00 00 The child process segfaults, and then the main process prints a traceback with EOFError because it could not receive anything from the child process. So this is most probably the same issue as #1141977, so I am merging them. I have run `init 0` on both VMs now, thanks again for providing them. -- Dmitry Shachnev
signature.asc
Description: PGP signature
