RE: [POC][RFC][PATCH 1/2] jump_function: Addition of new feature "jump_function"

[David Laight]

From: Masami Hiramatsu
> Sent: 09 October 2018 04:44
...
> I think we can replace the first 5 bytes of the default function
> to jmp instruction (to alternative function) instead of making
> this trampoline.

Or have a trampoline that is just a jump instruction and overwrite
the target address at run time to select the non-default code.
With care the target address can be aligned so that the write is atomic
and can be done while other cpu might be calling the function.

This will be lower impact that a 'jump indirect' - especially since
the latter would have to be implemented using a 'retpoline'.

It would also make it possible to re-instate the default function.
(By saving its address after the jump.)

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)