This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: in stable 4.4.131 kprobes seems to be broken
On Tue, 10 Jul 2018 15:31:06 +0200
Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> On Tue, Jul 10, 2018 at 02:26:43PM +0200, Jack Wang wrote:
> > > That commit seems to be part of some x86 kprobes restructuring, I
> > > hesitated to try it. After your email tried to cherry-pick to 4.4.y
> > > and it does not go in smoothly at all, quite a bit of conflicts.
> > >
> > > I may not phrase it correctly: My point was that in 4.4 there
> > > are more places where synthesize_reljump function is called, which
> > > latter got removed. And this function modifies kprobes page, so
> > > lifting up read-only protection needs to be done in more places
> > > compared to latter kernels. Whether it can be done correctly and
> > > proper thing to do in do_debug/resume_execution function
> > > exception processing context I cannot tell.
> > >
> > > Thanks,
> > > Victor
> > >
> > > > thanks,
> > > >
> > > > greg k-h
> > > >
> > Hi Victor, Hi Greg,
> >
> > Alexey Makhalov (+cc) ported the fix to kernel 4.4 2 weeks ago.
> > https://www.spinics.net/lists/stable/msg245223.html
> >
> > It applies cleanly to 4.4.139.
>
> Ah, nice, that's already queued up and will be in the next 4.4.y release
> in a few days.
Thanks guys, I think Victor is correct. That bugfix patch should be picked up.
Thank you,
--
Masami Hiramatsu <mhiramat@kernel.org>