Re: [PATCH 12/X] uprobes: x86: introduce abort_xol()

[Oleg Nesterov]

On 10/21, Srikar Dronamraju wrote:
>
> > If it is not clear, abort_xol() is needed when we should
> > re-execute the original insn (replaced with int3), see the
> > next patch.
>
> We should be removing the breakpoint in abort_xol().

Why? See also below.

> Otherwise if we just set the instruction pointer to int3 and signal a
> sigill, then the user may be confused why a breakpoint is generating
> SIGILL.

Which user?

gdb? Of course it can be confused. But it can be confused in any case.

> > +void abort_xol(struct pt_regs *regs)
> > +{
> > +	// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
> > +	// !!! Dear Srikar and Ananth, please implement me !!!
> > +	// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
> > +	struct uprobe_task *utask = current->utask;
> > +	regs->ip = utask->vaddr;
>
> nit:
> Shouldnt we be setting the ip to the next instruction after this
> instruction?

Not sure...

We should restart the same insn. Say, if the probed insn
was "*(int*)0 = 0", it should be executed again after SIGSEGV. Unless
the task was killed by this signal.

And in this case we should call uprobe_consumer()->handler() again,
we shouldn't remove "int3".

> I have applied all your patches and ran tests, the tests are all
> passing.
>
> I will fold them into my patches and send them out.

Great, thanks.

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/