Re: Kernel stack read with PTRACE_EVENT_EXIT and io_uring threads

From: Linus Torvalds
Date: Mon Jun 21 2021 - 11:38:35 EST

On Mon, Jun 21, 2021 at 6:55 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> On Tue, Jun 15, 2021 at 02:58:12PM -0700, Linus Torvalds wrote:
> > And I think our horrible "kernel threads return to user space when
> > done" is absolutely horrifically nasty. Maybe of the clever sort, but
> > mostly of the historical horror sort.
> How would you prefer to handle that, then? Separate magical path from
> kernel_execve() to switch to userland? We used to have something of
> that sort, and that had been a real horror...

Hmm. Maybe the alternatives would all be worse. The current thing is
clever, and shares the return path with the normal case. It's just
also a bit surprising, in that a kernel thread normally must not
return - with the magical exception of "if it had done a
kernel_execve() at some point, then returning is magically the way you
actually start user mode".

So it all feels very special, and there's not even a comment about it.

I think we only have two users of that thing (the very first 'init',
and user-mode-helpr), So I guess it doesn't really matter.