Port-vax archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

(Cross) compiling for VAX in 9.1



TLDR; wordy, long-winded user can't (cross) compile BIND from pkcsrc due to internal compiler errors, has 3 questions (see 3 bullets near EoM)...

Dear port-VAX:
Thank you for your prior responses on other subjects -- notably my boot and kernel init troubles with my KA410.  I can't work on that one further until I check for and replace those dreadful RIFA (now Kemet) brand polypropylene impregnated, metalized paper filter capacitors, however.
 
In the meantime, I was absolutely delighted to find -- after pulling my BA123 and 16MB KA630 (with a repaired RD53 for swap on ra0c) out of storage -- that a) it's power supply does not have any RIFAs (at least not that can be seen by looking in from the fan) and that b) it boots NetBSD 9.1 beautifully!  (Even if the clock loses a lot of time.)
 
Great, so onto enabling ntpd, disabling postfix and cron and compiling BIND from pkgsrc for it, as it's a major priority to bring a VAX production system online at NODOMAIN.NET in 2021...  So, when the kernel doesn't panic during the compile job (I may have a bad memory cell somewhere), I found that there were internal compiler errors.  So, I checked in a not terribly recent simh on an emulated KA630 (16MB RAM/96MB swap) and also on John Klos' simh KA655 from August (with updated pkgsrc) and both of them exhibit internal compiler errors, too (albeit not all at the same point in the compile, IIRC.)
 
So, I thought I'd just cross-compile the blessed thing on a large-ish NetBSD-amd64 VM that I provisioned just for that purpose.  But the instructions for cross compiling (https://ftp.netbsd.org/pub/pkgsrc/current/pkgsrc/doc/HOWTO-use-crosscompile) are either inaccurate or out of date (they reference NetBSD 6.1.)  I'll not go into that on this list, though, except to say that -- under some circumstances -- I saw internal compiler errors on the Intel machine, too.  While my real KA630 *MAY* have a bad RAM cell or other hardware problem, my build VM and host and the two simh VAXen cannot all be suffering from "hardware issues" that would manifest so similarly.

This all leaves me wondering what to do next.  I know from this list that major work has just been completed on the compiler backend for VAX (and I regret not supporting that work when I could.)  But we all know that it will be at least one release before that makes it into NetBSD.  So, I am unsure of whether to invest the effort to provide the requested bug reports on the internal compiler errors (and tackling the problem myself may be beyond my abilities in the near term), especially as there are now two compiler backends that may need fixing.
 
Cross compiling seems like the option to pursue -- especially as those KA630s are so majestically 11/780 in their perf -- but I wanted to query the list for input:
- Has anyone had success in (cross) compiling from pkgsrc for VAX under 9.1?  If so, what instructions did you use?
- Should I expend the effort of gathering and testing repro steps for internal compiler errors from the legacy backend?
- Both during the kernel panics and the internal compiler errors, there are insinuations and statements to the effect that stack unwinding is not available, implemented or is broken.  Is that expected?
 
Your input is greatly appreciated.
 
Humbly,
Josh/NODOMAIN.NET
Love | Responsibility | Justice
Leibe | Verantwortung | Gerechtigkeit


Home | Main Index | Thread Index | Old Index