Q: user_enable_single_step() && update_debugctlmsr()

[Oleg Nesterov]

On 08/01, Oleg Nesterov wrote:
>
> And btw, this is offtopic, but the usage of update_debugctlmsr()
> doesn't look right to me (I can be easily wrong though). I'll write
> another email.

user_enable_single_step() does

	if (enable_single_step(child) && block) {
		unsigned long debugctl = get_debugctlmsr();

		debugctl |= DEBUGCTLMSR_BTF;
		update_debugctlmsr(debugctl);
		set_tsk_thread_flag(child, TIF_BLOCKSTEP);
	}

and I do not understand update_debugctlmsr() above (and other
callsites).

Lets ignore uprobes which needs the changes anyway. This is
only used by ptrace and the task is stopped. So, unless I missed
something obvious, this update_debugctlmsr() is simply unneeded,
__switch_to/__switch_to_xtra should notice _TIF_BLOCKSTEP and do
update_debugctlmsr(DEBUGCTLMSR_BTF).

But, worse, isn't it wrong? Suppose that debugger switches to
another TIF_SINGLESTEP && !TIF_BLOCKSTEP task, in this case
we "leak" DEBUGCTLMSR_BTF, no?

IOW, it seems to me we could safely remove update_debugctlmsr()
arch/x86/kernel/step.c. However, if we want to re-use this code
in uprobes, then we probably need to add "if (child == current)".

Or I am totally confused. Help!

Oleg.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/