Re: New 2.0.31 pre patch is out (still problems)

Greg Patterson (
Sun, 15 Jun 1997 16:16:35 -0400 (EDT)


On Mon, 2 Jun 1997, David S. Miller wrote:

> That should be about it. Unfortunately I did not get a chance to
> merge in Stephen Tweedie's ext2 I/O error recovery fixes, that
> certainly will make it into 2.0.31 so do not worry. ;-)
> I would love it if only one more pre-patch is needed for a final
> 2.0.31 so people please test this well.

I have reported a problem with ioctl(TIOCGWINSZ) in the kernel a few weeks
ago to the list as well as to Linus. He asked me to try and pin-point it
to the specific kernel. I sent my response to him and to the list a
couple of times but received no feedback.

What I responded with is as follows:

======= cut here =======

Okay, it is definitely 2.0.30 that is at fault. This is easily producable
on my system from tty1 virtual console:

1. login as gomer
2. rlogin -l gomer localhost
3. stty size
4. if size == ok, type "exit" and goto step 2
5. stty size -> 0, 0

While doing this, I also noticed something else. After doing an "exit"
and redoing rlogin -l gomer localhost, there is a very noticable delay of
4-5 seconds taken to actually login (this is on a totally idle PPro 200).
The very first rlogin is RAPID. The second rlogin is always delayed.

When I went to 2.0.29, rlogin was consistantly rapid *always*. And after
20 or more attempts at reproducing the incorrect row/cols stty values, I
gave up.

======= cut here =======

I have since tried the above with the pre 2.0.31-2 patches applied. The
same problems still occur so I guess I will be stuck with 2.0.29 again.

In addition to attempting to strace where rlogin was "pausing" temporarily
(see above), it appears to be on a connect( sometimes (but not
always) and on read() of /etc/motd. Again, with 2.0.29, things are very
rapid under all circumstances.

Version: 2.6.2
Comment: PGP Signed with PineSign 2.0


| Greg Patterson | EMAIL: |
| | |
| Move over Microsoft, | IRC: Wizird (#Linux) |
| Linux is here to stay. | WWW: |