Re: processing IOService in kext
site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com gc, I am implementing this over a disk image. Then I copy ANY application into the disk image in order to cache / optimize / re route it's requests. (say its looking for a certain file on the disk, which is not there, but I can get it from else where) What you're describing here is basically an HSM. You can do this in two different ways: = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... On Nov 17, 2004, at 11:26 AM, calcmach@actcom.net.il wrote: Why do you want to intercept file I/O calls? The system will generally do a better job of this, unless you know something that it doesn't. 1) Write a new filesystem, which implements sideways lookup. You will end up with a large synthetic volume that behaves like it's full of symbolic links. If the files are just somewhere else on the system, this will work fairly well. 2) Implement this by subverting library functions in userland. Here you are protected by the kernel's security model, since you're doing work as the same user that's running the program, and you can control things cleanly on a per-application basis. You also don't run the risk of running the kernel out of memory. This email sent to site_archiver@lists.apple.com
participants (1)
-
Mike Smith