On 2016-09-20 23:57, Mark Pizzolato wrote:
On Tuesday, September 20, 2016 at 2:00 PM, Johnny Billquist wrote:Hi. On 2016-09-20 20:54, John Klos wrote:Hi,But it seems as if this is not worth doing anymore. NetBSD/vax does not build natively, and have not done so for years. And noone seems to even notice.Oh, we notice. I've been trying to build packages (and fixing what I can, and even reporting fixes upstream) using the toolchain from NetBSD 6.1.5. It would be nice if someone who understood things could get gcc in to a working state, or if someone would get support for VAX in to clang / LLVM.Anything would be an improvement at this point. Like I said. Trying to run build.sh on current bombs out right away, even in simh, so anyone can reproduce this within seconds on a booted system.Running on a real 8650, a "cvs update" on /usr/src nowadays take days, and usually the machine hangs after about a day, because of some bugs in the system which just seem to stop all I/O activity.I've never seen any of my systems hang. Regardless of toolchain issues, the hardware has always been incredibly stable. Trying to build certain packages can cause a panic, but even then I've never seen a hang. Then again, all of my hardware are VAXstations (4000/30, /60, /90), so perhaps some driver for hardware in your 8650 is problematic.The hardware is fine. It runs VMS and Ultrix without issues. And with NetBSD, I have had it stall as well as panic.He said "perhaps some driver for hardware", not the hardware itself. The 8650 shares no disk drivers with any of the systems John lists.... That you see this also with a simh system supports the theory that there is a driver problem....
I didn't say that I had the same problems with simh. The problems I've seen (so far) with simh is that gcc conks out. Which is a known different problem.
The hanging and crashing might be possible to provoke in simh as well. I don't know. Right now I'm running Ultrix on the system to get some comparable numbers for some things, and even though I'm far from done, there seems to be some interesting observations I can do about how the system appears even now. I'm currently doing a cvs checkout of the netbsd source tree. Under NetBSD, cvs spends between 50& and 70% of CPU time in system. Under Ultrix, I seem to be spending about 15% in system. Which is a dramatical difference. But I want to do more testing, and also run some I/O performance testing to get some more solid numbers. It will take a while before I'm done, so expect a better report in a couple of days.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt%softjar.se@localhost || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol