Closed jguillaumes closed 6 years ago
I don't remember, does locking memory work better if you run as root?
Just toke a look at that.
Memory locking is not implemented in MacOS, so that message is expected and normal.
Regarding the VDE issue: You can follow along in issue #2 how that came to klh10. Basically I only compile-tested it in a slow vm that was only setup for compiling and nothing more. The amount of code for it is rather trivial, though. Basically only lines 2286...2337 in src/osdnet.c are VDE-only. Anything that goes wrong there is either trivial (permissions or something) or unfixable-serious
Hello, I've found the problem. When using VDE the "interface name" is actually a filesystem path, which can be longer than the 16 characters used as limit for the "regular" interfaces. In OSX the /tmp directory, used usually to put the vde sockets, is really a symlink to /private/tmp, so "/tmp/vde.ctl" is truncated to "/private/tmp/vd".
I've hacked a quick (and quite dirty) patch to fix the problem. Unfortunately it looks like the xCode editor has scrambled part of the format, so there is more stuff in the path that it should be.
To cut the story short, I've added an "alternate name" field to the dpni20_s struct, wich can be as long as MAX_PATH, and the logic to select the appropiate name when using IFMETH=vde. The logging messages still use the truncated version, but that is just an esthetic problem.
Good! So we can close this?
Once https://github.com/PDP-10/klh10/pull/27 gets merged it will be ok to close.
I'm closing the issue, since the problem was fixed in commit 9d13c2b5e55916c0ec69d3435fcb63f01fb70bab
Hello,
I've found a couple of problems:
The boot sequence continues but the simulator hangs and it has to be killed.
/tmp/vde exists and it is under vde control:
The dpni20 program is set said:
And this are the contents of my INI file: