On 12-07-12 02:15 PM, vivek poddar wrote:
Hi,
I want to know that what is the function of the /dev directory in
linux and what is the meaning of /dev/pts and /dev/tty and what they
do.
AFAIK, historically, unix systems ran on mainframes and allowed multiple
users to login simultaneously through teletype terminals (what I believe
we call as dumb-terminals these days). All this was before my time so
don't quote me on it. The teletype terminology is followed since then on
unix/linux systems.
You may very (Very VERY) loosely associate a tty with a monitor but the
correct way to think about it is thus (again my interpretation). Each
*nix system assumes you have several dumb-terminals (ttys) connected to
them (tty1-tty50). These are the multiple login prompts you can access
by pressing the Ctrl+Alt+F? key combinations. Since within each session,
a user may launch multiple terminal prompts, these terminal prompts are
associated with pseudo-terminals (pts). You can try this out by starting
a prompt (or logging in to a tty) and typing the command "tty".
If I remember correctly, linux has 4 kind of devices (or rather, device
interfaces). You have character type devices (ttys), block devices
(diskdrives,ramdisks), and two more that I can't remember (was fifos
another one?). The /dev directory provides access to all these devices
through the open/read/write etc system calls (sockets are also treated
as character streams, I believe). In fact, if you go to /dev and run
$ ls -l
you'll see entries that indicate the type of device (the first
character), e.g.,
crw-rw-rw- 1 root tty 5, 0 2012-07-13 13:31 tty
brw-rw---- 1 root disk 1, 0 2012-07-12 12:43 ram0
brw-rw---- 1 root disk 8, 0 2012-07-12 12:43 sda
If you want to see what's written on your harddrive block by block (or
character by character), you can open /dev/sda in a C (or perl or even
Java) program and start reading it. Of course, you will need root
permission and you run the risk of causing irreparable damage do your data.
Until several years back, /dev was populated with pointers to all
conceivable devices that can be created (or else, these were statically
created [2]). That's why, the "mount -o bind /dev /path/to/chroot/dev"
trick also works [3]. With udev, I noticed that /dev was not populated
with all device pointers but a plug-and-play feature was implemented [1].
Some references you may want to look at are:
The Linux Kernel Module Programming Guide (LKMPG) [4]
the chapter on Character devices [5]
In particular, I found what I was looking for. See section 3.1.6 of
LKMPG [6]. This explains what /dev is for.
My 2 cents..
Sharad
[1] http://en.wikipedia.org/wiki/Udev
[2] http://www.tldp.org/LDP/lfs/LFS-BOOK-6.1.1-HTML/chapter06/devices.html
[3] http://fermilinux.fnal.gov/documentation/tips/mount-bind-chroot/
[4] http://www.tldp.org/LDP/lkmpg/2.6/html/index.html
[5] http://www.tldp.org/LDP/lkmpg/2.6/html/x569.html
[6] http://www.tldp.org/LDP/lkmpg/2.6/html/x427.html
--
Mailing list guidelines and other related articles: http://lug-iitd.org/Footer
