D4GL (4js) COFF get segmentation violation
The LLM had thoughts on this one too. It recons it's a bug in the SCO binary, not ibcs-us. It might be right - but it can be very confident even with little evidence. I close this now because I don't think there is much hope. Reopen it again it necessary. Here is what the LLM said: A. Capture envs from a known-working and a failing context and diff: # in cron: * * * * * env > /tmp/cron.env # via CUPS pipeline: same diff -u /tmp/cups.env /tmp/cron.env A missing variable that the binary touches (e.g....
Closing this - but I took another look, and yes it's the fact the ibcs-us modifies the binary while it's in memory that causes licensing to break. If a version 4.4 is ever done, it may have a fix. I'll leave this open so you get notified if that happens.
read(7, ERROR(-6071):Cannot use directory '/usr/fgl2c/lock'.
I put this reply on Ticket #5, but it was meant as a reply here: It's been 4 years, LLM's are a thing, and they are far smarter than me. This is Opus's take: This isn't an ibcs-us bug: fglWrt is a statically-linked 32-bit native Linux ELF binary, so when your COFF program fork+execs it, ibcs-us drops out of the picture entirely — fglWrt then talks directly to the kernel. Your fglWrt_u.strace shows it using the legacy stat()/fstat() syscalls, which fail with EOVERFLOW on modern XFS because the inode...
Oracle Froms 3.0 (SCO) segfault
It's very hard to diagnose what is going on here, but I did find something in per-svr4/socksys.c that may be related and fixed it. Version 4.3 will have the change - give it a try.
Foxbase+