I am using Kernel 1.3.37, a.out system, libc 4.6.27. Everything is in
modules but the ext2fs.
Now here is a problem that I am able to recreate, however, not find
the source to.
The problem seems easy to re-create when the following modules are
installed and in this order:
slhc.0
ppp.0
lp.0
sound.o - SB16 LDMA: 1 HDMA: 5 IRQ: 10 (0x220)
And the following software running:
PPPD 2.2.0c
X11R6: XF86_MACH8
fvwm
fvwmpager
color_xterm
(This is a 486/80 system with 16Mb memory)
Here is what I did:
I was compiling (as root) NAS (Network Audio System
ftp.x.org/contrib/audio) and a couple times, I had to kill the make
process. The makefile tree was setup to run child make processes with -k,
so when an error occured, I had to press CTRL-C many times to kill the damn
thing. After having done this 5 times, I could no longer kill processes
in my system.
'ps ax' would show up processes like: make, rm, makedepend All of
which where processes generated by the makefiles for NAS. If I killed
them expictly with 'kill -9' or used 'killall make', for example, they
would not die. They were marked as having not been swapped and running.
I would also like to point out, that using ps under 1.3.37 generates a
strange listing:
PID TTY STAT TIME COMMAND
74 v01 S < 0:00 /sbin/agetty 38400 tty1
76 v10 S < 0:00 /sbin/agetty 38400 tty10
77 v11 S < 0:00 /sbin/agetty 38400 tty11
78 v12 S < 0:00 /sbin/agetty 38400 tty12
80 s01 S < 0:00 gpm -t msc -m /dev/ttyS1
117 pp1 S < 0:00 -bash
170 s02 S < 0:00 pppd /dev/cua2 38400 connect chat -f /etc/ppp.conf$
183 pp1 S < 0:02 pine
^
|
+----------+
|
+- This '>' character is new, and doesn't seem to have any
particular purpose, since _all_ process have that mark. When I go
looking around the /proc system, I see nothing to indicate why this
character is here. Yet if I go back down to 1.3.36 or 1.2.13, the
character dissapears.
----
On a different note:
This kernel is very nice. runs smooth. I have not had any problems,
until I picked up NAS =) So I have been running for 8 days w/o a
hitch. However, I do have one concern. There was absolutely _NO_
mention of the fact that QIC117 support was taken out of the kernel. For
those that do not understand that it is simply a "#define" statement,
this can be very confusing. And to top it off, the Configure.help file,
as supplied with the kernel, indicates that there is QIC117 support and
that there "should" be a question in the configure script asking you if
you want it. There is no such question.
The solution, and a good one at that, is to use the
zftape-2.03b.tar.gz package on Sunsite. This package uses dynamic buffer
allocation and therefore has no depencies on the "old" defines orginally
put into the kernel for ftape drivers. zftape also supports the QIC-80
format and incorporates LZW3 compression.
So the only thing that need be done, is to announce in the
Changes.log the removal of the QIC117 support /OR/ put it back in.
If QIC117 support is to be removed on a permanate basis, then there
should be mention in the Kernel docs, somewhere noticable, that one
should get and compile zftape-2.03b for QIC117 support.
It may not matter however, from what I gather from conversations and
email, I seem to be one of the only people using a QIC117 floppy tape
drive :/
Best of wishes to all Linuxers,
Sincerely,
Chad C Giffin
TNET Information Systems, Canada
(204) 857-5754
(aka typo)
cgiffin@portage.net