A CPU can be either in kernel mode (fully privilege) or in user mode. The kernel requires kernel mode, while applications need to run in the user mode. But how can the CPU be in two modes at once?
0
votes
1 Answers
1
votes
Processors generally include a mode flag which indicates which mode the processor is in at a given time; that flag need not necessarily do a whole lot. In a simple implementation, the flag might only control whether the processor is allowed to change memory mappings; the processor would include an instruction which simply switches to user mode, and an instruction which simultaneously switches to kernel mode and jumps to a particular address.
If the kernel stores its own code at the aforementioned address and then switches the memory map so that the address in question is write-protected, then user code would be able to ask the kernel to do something by storing its request somewhere and making a call to a "switch to kernel mode and jump" instruction. The kernel code could then enable its private memory areas, examine the request stored by the user-mode code, act upon the request, disable its private memory areas, switch back to user mode, and return to executing user-mode code.
