I'm trying to understand how the path those functions take to be called.
>From a basic understanding, is that we register f_ops for tty's and
ptys. So we register tty_open for tty's which includes ptmx.
therefore when we open ptmx, tty_open gets called, we do the code for
PTMX, which calls devpts_pty_new() which creates the appropriate
enteries in /dev/pts. These f_ops are registered in chrdevs struct
(from fs/devices.c)
I'm a little more hazy on devpts_pty_kill. That seems to similiar
(except it's in pty_close, in pty.c) which is part of a much larger "tty
driver" structure and is the "close" member of that structure. This
structure is then registered. Then basically, it's also registered w/
tty_fops, but tty_release calls tty->driver.close.
Is this a correct understanding, or is there more going on behind the
scenes?
thanks,
shaya potter