On Thu, 2 May 2019 13:49:29 -0700
Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> On Thu, May 2, 2019 at 1:22 PM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> >
> > Something like so; it boots; but I could've made some horrible mistake
> > (again).
>
> This actually looks much better to me.
>
> Maybe it's more lines (I didn't check), but it's a lot simpler in that
> now the magic of the int3 stack doesn't get exposed to anything else.
>
> We *could* also make this kernel-mode-only do_int3() be a special
> function, and do something like
>
> # args: pt_regs pointer (no error code for int3)
> movl %esp,%eax
> # allocate a bit of extra room on the stack, so that
> 'kernel_int3' can move the pt_regs
> subl $8,%esp
> call kernel_int3
> movl %eax,%esp
>
> and not do any stack switching magic in the asm code AT ALL. We'd do
>
> struct pt_regs *kernel_int3(struct pt_regs *regs)
> {
> ..
> return regs;
> }
>
> and now you the rule for call emulation ends up being that you need to
> "memmove()" the ptregs up and down properly, and return the new
> pt_regs pointer.
>
> Hmm? That would simplify the asm code further, but some people might
> find it objectionable?
>
The problem with this approach is that it would require doing the same
for x86_64, as the int3 C code is the same for both. And that may be a
bit more difficult on the x86_64 side because it's all done with a
simple flag in the idtentry macro to add the gap.
-- Steve
