View Single Post
Old 12-05-2011, 02:39 PM   #12
rkomar
Wizard
rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.rkomar ought to be getting tired of karma fortunes by now.
 
Posts: 1,787
Karma: 4386796
Join Date: Oct 2010
Location: Vancouver, BC, Canada
Device: PRS-505, PB 902, PRS-T1, PB 623, PB 840
Quote:
Originally Posted by Seneca View Post
No, I think /dev/pts is uncritical. At least on my local Linux system it has the same modes as on the pocketbook. You don't need write access there, because the ptyX nodes are created by the kernel (actually the devpts pseudo filesystem) and not by the application creating a pseudo tty pair.
It seems to be important for sshd. If I leave the firmware defaults, I can log in via ssh but the shell has no pty and is missing many features. If I only change the permissions on /usr/dev/ptmx, then logging in via ssh fails completely (not even a broken shell). If I also add "gid=5" to the options for /dev/pts in /etc/fstab, then the ssh shell behaves perfectly when I log in.

It was when I was googling for similar ssh problems that I saw others suggesting the same changes for /dev/pty in /etc/fstab on other Unix systems. Perhaps this is only important for those applications that use the openpty() system call (like sshd does).

I noticed while googling that some sshd clones (dropbear?) use an openpty() clone instead of the actual system call. I think I'll take a look at that and see if it would work on a default PB reader.
rkomar is offline   Reply With Quote