Everything in the /proc directory is "imaginary." Every file that you see, every directory, is actually an output that is generated on-the-spot. You can therefore browse the Linux source-code to see for yourself exactly how /proc/cpuinfo is produced, because you will find the subroutine that generates every line that you see.
For example, if you do ls /proc, you will see a "directory" for every process-ID in the system at that particular instant in time. The kernel subroutine that's responding to your ls command has generated a "directory entry" for every process that it saw at this instant.
If you then did ls /proc/12345, another subroutine will determine if process #12345 exists at this instant, and if so it will produce a "directory listing" for the various bits of info that you could obtain; if not, it produces "file not found."
Last edited by sundialsvcs; 10-23-2012 at 11:35 AM.