Mail Archives: djgpp/1992/05/18/09:06:41
DJ,
>Some exceptions have been given their own TSS segments because I can't
>guarantee the availability of a stack for them to run in. For
>example, if a page fault occurs while extending the stack, handling it
>with a regular interrupt would cause a double fault, so the page fault
>handler has it's own stack. Same for timer, HD, and keyboard, since
>the page fault handler may need these interrupts to be in a valid
>state, I can't have these interrupts causing page faults.
>The remainder go through the common dispatch routines.
What if other H/W device (eg. serial port/network cards) interrupts while the
stack segment is not available? Should we give all H/W interrupts its own TSS?
thanks for your patience,
Michael.
- Raw text -