Subject: Re: another kdump problem
To: None <tech-kern@NetBSD.ORG>
From: der Mouse <mouse@Collatz.McRCIM.McGill.EDU>
Date: 08/25/1995 20:44:57
> I've done an implementation of the following sysctl mib-variables:
> as well as changed sysctl(8), and kdump(1) to take advantage of them.
> It all works like a charm.
Something occurred to me while reading this. This works only if you
run kdump on the same system you ran ktrace on.
Wouldn't it be better to have ktrace emit a special record the first
time any given <emulation,syscall> pair is emitted, giving the name of
that syscall? I wouldn't expect this means too much bookkeeping in the
kernel, though I must admit I haven't looked at it. But it would make
ktrace.out files entirely self-contained.
Of course, still better - at least in some respects - would be for me
to get off my tush and port my ptrace changes to -current, so we can
have _real_ syscall tracing :-/