Source-Changes-D archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: src/etc



On Sep 8,  9:31am, tsutsui%ceres.dti.ne.jp@localhost (Izumi Tsutsui) wrote:
-- Subject: Re: CVS commit: src/etc

| christos@ wrote:
| 
| > >> We can, however, use a modified MAKEDEV script for install cdroms (and 
then
| > >> use ptyfs).
| > >
| > >MAKEDEV is a generated file so I'm afraid it is a bit annoying
| > >to sync modified version with standard one.
| > 
| > but MAKEDEV can just run mkdir /dev/pts, the rest can be done in rc scripts.
| > Hell, the rc script can run mkdir -p /dev/pts before mounting.
| 
| I don't see your point.  Discussion here is that the target "init" in
| MAKEDEV (that is invoked by init(8)) should create compat pty nodes or not.
| 
http://cvsweb.NetBSD.org/bsdweb.cgi/src/etc/MAKEDEV.tmpl.diff?r1=1.157&r2=1.158&f=h

That would proliferate COMPAT_BSDPTY forever. The point being that MAKEDEV
does not need to do anything, it can all be done in the installation rc
scripts, since ptys are not needed that early.

| If tmpfs /dev created by init(8) is used only on installation or
| emergency environments, it's probably okay to create ipty.
| 
| If tmpfs /dev created by init(8) is also used by LiveCD etc.
| (i.e. root on (read only) cd9660 environments which might run
| on multiuser with ptyfs), it should not create compat pty nodes.
| 
| We can have special MAKEDEV which create "ipty" in "init" target,
| but "init" should also create "all" devices (or nodes defined in
| MD MAKEDEV.conf) that are defined in generated files.
| 
| I'm suggesting it's easier to fix etc.rc scripts in installcd
| to mount ptyfs and remove (revert) opty from the target init
| in src/etc/MAKEDEV.tmpl, rather than maintaining special MAKEDEV
| for installcd.

I am all for removing the ipty and opty targets (and COMPAT_BSDPTY) and
making everything use ptyfs.

christos


Home | Main Index | Thread Index | Old Index