You are not logged in.
I eventually solved this problem myself, but the solution was rather obscure, so I thought I'd post it here for people to find...
Symptoms
The CUPS home page at http://localhost:631/ loads fine, but any link you click results in Internal Server Error.
journald shows that cups-exec is crashing with a segmentation fault (segfault, SEGV, SIGSEGV).
Completely uninstalling and reinstalling CUPS does not help, even if you purge /etc/cupsd as well.
pacman -Syu does not help.
Example message from the journal:
kernel: cups-exec[6540]: segfault at 14 ip 0000577b8470a18f sp 00007fff51f9bb90 error 4 in cups-exec[118f,577b8470a000+1000] likely on CPU 8 (core 4, socket >
kernel: Code: 0f 85 21 01 00 00 ff 15 07 2e 00 00 85 c0 75 38 8b 7d c8 ff 15 2a 2e 00 00 85 c0 0f 85 f5 00 00 00 8b 7d cc ff 15 e1 2d 00 00 <8b> 70 14 48 8b >
systemd-coredump[6541]: Process 6540 (cups-exec) of user 0 terminated abnormally with signal 11/SEGV, processing...
systemd[1]: Started Process Core Dump (PID 6541/UID 0).
systemd-coredump[6542]: [?] Process 6540 (cups-exec) of user 0 dumped core.
Stack trace of thread 6540:
#0 0x0000577b8470a18f n/a (/usr/lib/cups/daemon/cups-exec + 0x118f)
#1 0x0000758296afb488 n/a (libc.so.6 + 0x27488)
#2 0x0000758296afb54c __libc_start_main (libc.so.6 + 0x2754c)
#3 0x0000577b8470a325 n/a (/usr/lib/cups/daemon/cups-exec + 0x1325)
ELF object binary architecture: AMD x86-64Cause
(At least in my case...) It's a permissions issue. The PKGBUILD for the cups package creates a user named "cups" with a fixed user id of 209, and hardcodes that user id in several places.
However, if you already have a "cups" user at the time of installation, with a different user id, the new one won't be created, and the hardcoded user id of 209 will be wrong.
How this causes segfaults is anyone's guess. I've filed an issue for the CUPS maintainers about this.
Offline