Skip to contentSkip to conversation

Missing /dev/pts?

DISCUSSION Robert L. Harris 2 participants 2 messages Nov 19, 2003

lore
Share & tools
lore
#0 lore raw

Had to rebuild a machine due to a dead deathstar. It's 99% there at
this point I think except I can't log in via xterm, eterm or ssh. Poking around, the most obvious factor is that /dev/pts is empty.

I can see that /dev and /dev/pts mounted by using "df -k /dev /dev/pts" which returns:

Filesystem 1K-blocks Used Available Use% Mounted on
- 0 0 0 - /dev
devpts 0 0 0 - /dev/pts

whish is identicle to the machine I'm currently logged into and is working fine. /dev and /dev/pts are both compiled into the kernel and are reported by "cat /proc/filesystms". Devfsd is up and running with a command line of "/sbin/devfsd /dev".

Anyone have any idea why this is malfunctioning?

Robert

:wq!
---------------------------------------------------------------------------
Robert L. Harris | GPG Key ID: E344DA3B

                                         @ x-hkp://pgp.mit.edu

DISCLAIMER:
These are MY OPINIONS ALONE. I speak for no-one else.

Life is not a destination, it's a journey.
Microsoft produces 15 car pileups on the highway.
Don't stop traffic to stand and gawk at the tragedy.

#1 lore raw
Quoted 12 lines · Robert L. Harris

Had to rebuild a machine due to a dead deathstar. It's 99% there at
this point I think except I can't log in via xterm, eterm or ssh. Poking around, the most obvious factor is that /dev/pts is empty.

I can see that /dev and /dev/pts mounted by using "df -k /dev /dev/pts" which returns:

Filesystem 1K-blocks Used Available Use% Mounted on

(quoted diff) +0 -1
-		0		0	0		-	/dev

devpts 0 0 0 - /dev/pts

Quoted 6 lines

whish is identicle to the machine I'm currently logged into and is working fine. /dev and /dev/pts are both compiled into the kernel and are reported by "cat /proc/filesystms". Devfsd is up and running with a command line of "/sbin/devfsd /dev".

Anyone have any idea why this is malfunctioning?

Both devfs and devpts can be troublesome in several ways. First, the obvious: Make sure you didn't mount devpts on /dev/pts first and then devfs on /dev later. That will hide the devpts mount completely. (Assuming there is a pts directory in /dev before devfs gets mounted. Many people haven't actually removed anything from the oldfashioned /dev after installing devfs because it isn't necessary.)

You don't say what kernel version you're using. In 2.6, this problem exists when using devfs+devpts:

Devfs documentation claims there is a problem using devpts with devfs. The same docs also claim that you don't need devpts because devfs does the same job and does it without those problems. This used to be true. Devfs is becoming obsolete though, and someone did a "cleanup" patch that removed the pts support from devfs, forcing everybody to use devpts wether they use devfs or not. Unfortunately, the bad devfs/devpts interactions remained. I recently got rid of devfs for this reason, I got tired of xterms that
1. didn't appear, often crashing the x server instead
2. appeared after 20 -60 seconds or so. And then I got lots because I clicked the xterm menu over and over. :-(

This doesn't work, and won't likely be fixed ever now that devfs is being obsoleted. I recommend going back to oldfashioned /dev with plain device nodes while waiting for the udev stuff to become useable.

Udev will probably achieve many of the same goals as devfs, such as:
1. showing device nodes only for those devices that actually have a driver in the running kernel, avoiding clutter and wondering which cdrom driver is the "right" one.
2. showing a device node for every driver in the kernel, no need to remember to use MAKEDEV or even mknod just because you got some experimental hardware that your distro vendor hasn't heard about yet.

Udev will probably not get rid of the silly major/minor numbers though. The devfs approach didn't really need them, now they are being extended instead. At least we won't ever have to look them up and type them in if udev gets going. :-/

Helge Hafting