bill@twwells.com (T. William Wells) (09/30/89)
In article <988@dtscp1.UUCP> scott@dtscp1.UUCP (Scott Barman) writes:
: How about us System V Release 3 people with that bug that doesn't
: free inodes under certain circumstances. I just ran into this one
: as recently as yesterday
Some of us System V Release 3 people just fix the bug. If you have a
'386 box, I may have a fix for you. I have a working patch for
Microport SysV/386 3.0e; I just did a fix for Interactive 2.0.2 (I
think, I'll know when the vict..., uh tester, has used it for a
while). These are binary patches: you modify an object file in your
kernel configuration directory and rebuild your kernel.
It takes me about two hours to find and fix this bug. All you need is
a disassembler (adb will do, if you have it); and some kind of binary
patching program (again, adb will do). You also need a little luck,
but so far I've had it.
If you want either of these fixes, drop me a note. And while you are
at it, blast your vendor for not fixing the problem. It's a simple
fix. (Try: each patch has been, in essence, to a single branch. Truth,
it doesn't really fix the underlying problem, but it eliminates the
effects of the actual bug. Why not fix that? Because the correct fix
for that would be very difficult in binary.)
If you have a '386 that uses COFF format executables (I think they
all do) and is not one of those two, I might be able to track down
the bug and swat it for you. If you have another kind of system, I
might be able to work with you to fix the bug. Let me know and I'll
see what I can do.
If you have a fix for other machines, I'd like to hear about it; I'm
thinking of setting up a server to distribute these things and if I
get a few more of these patches, I'll do that.
Followups have been directed to news.sysadmin.
---
Bill { uunet | novavax | ankh | sunvice } !twwells!bill
bill@twwells.com