Hi,

I am migrating a custom system (with no operating system per se) to run under Linux (it's actually a PC/104 architecture but it's basically a stripped down PC). It is a standard control application. The first question that struck me was, how can I protect against the user switching off the power supply without correctly shutting down? Ideally, since this is an embedded system with Linux running in the background only, there should be no concept of 'shutting down' to the user. They should be allowed to just switch it off and not worry about the filesystem (there is no filesystem use apart from loading the code). Is there any way I can get linux to read from disk initially (to load the applications) but then operate solely in RAM and never write to disk? Then the user would be able to switch off the unit without shutting down..

Looking forward to any replies.

Thanks
Michael

You can use a read-only file system. With an embedded system, you don't have that much to deal with. You can strip the kernel down to almost bare, and compile everything you need into the kernel itself; vs a modular kernel. Once you get it to boot, it will always boot, no matter how many times power is cut.

I'll tell you a little trick. I don't know how much disk you need, but linux will USB boot off up to a 4 GB flash drive. If you know what you're doing, you can go up to an 8 GB flash drive. That way you can configure the drive on a fully functional linux system, with every tool you could ever possibly want, and then plug it in the embedded device, and watch it go. It's cool.

There are embedded distros, that are made to just put on a flash drive, as read only. Read-only is a file system flag. You can set it with 'parted'.

If you look at some of the live cds (like Fedora) they have the option to run entirely from ram. You would have to reconfigure it a little bit (basically no options) but the basics setup is there.

Awesomemachine & lazlow: you have really answered my question and helped me out a great deal. thank you!!