Perl general protection fault

anyone has been finding the metadata.pl generating perl crashes and malformed /tmp/config.in?

./script/feeds install -a

is generating

[  427.649047] traps: perl[6438] general protection fault ip:64f9b2368cd3 sp:7ffd3dab2b58 error:0 in perl[157cd3,64f9b2254000+1df000]
[  440.344911] perl[6683]: segfault at 9 ip 0000608e638977a5 sp 00007ffd9cdc34e0 error 4 in perl[14a7a5,608e63790000+1df000] likely on CPU 10 (core 10, socket 0)
[  440.344921] Code: 00 00 00 04 75 13 49 8b 84 24 f8 00 00 00 48 89 03 49 89 9c 24 f8 00 00 00 49 83 ac 24 f0 00 00 00 01 48 89 f3 48 85 f6 74 13 <8b> 53 08 85 d2 74 64 83 ea 01 89 53 08 0f 84 18 ff ff ff 4d 85 ff
[  441.562871] traps: perl[6696] general protection fault ip:5f1d1ad5aa0b sp:7fff8eb28650 error:0 in perl[14fa0b,5f1d1ac4e000+1df000]
[  471.633158] traps: perl[6892] general protection fault ip:58457f048a0b sp:7fffbaef9db0 error:0 in perl[14fa0b,58457ef3c000+1df000]
[  473.244676] traps: perl[6906] general protection fault ip:62dabfd32a0b sp:7ffd2a56fae0 error:0 in perl[14fa0b,62dabfc26000+1df000]
./scripts/feeds install -a
Segmentation fault (core dumped)
Failed to build tmp/.config-package.in
WARNING: Makefile 'package/feeds/packages/mqttled/Makefile' has a dependency on 'python3-netifaces', which does not exist
Use of uninitialized value $fh in <HANDLE> at /home/Development/bananapi4/scripts/metadata.pm line 23, <FILE> line 150829

also

./scripts/feeds install -a
Segmentation fault (core dumped)
Failed to build tmp/.config-package.in
WARNING: Makefile 'package/feeds/packages/mqttled/Makefile' has a dependency on 'python3-netifaces', which does not exist
Use of uninitialized value $fh in <HANDLE> at /home/Development/bananapi4/scripts/metadata.pm line 23, <FILE> line 150829

it only disappears after i do a distclean ... but eventually reappears

this both happens on Fedora 42 and Ubuntu 24.10

can't reproduce on FC41, ./script/feeds install -a completes just fine.

2 Likes

... you have local auto-debug tools to help with standard packages crasging.

maybe start with helping us out whether it is system z or power8 youre struggling with?

how odd!!!... i was trying it out on virtualbox noticed that host i/o cache was unselected ... selecting it seems to have fixed it ... a 16cpu vm i guess can't cope if it is not assisted by the host i/o

sorry for the bother

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.