Date: Mon, 19 Jun 2017 18:56:37 +0200 (CEST)
From: Miroslav Benes <mbenes@suse.cz>
To: Josh Poimboeuf <jpoimboe@redhat.com>
cc: Petr Mladek <pmladek@suse.com>, Joe Lawrence <joe.lawrence@redhat.com>,
        live-patching@vger.kernel.org, linux-kernel@vger.kernel.org,
        Jessica Yu <jeyu@redhat.com>, Jiri Kosina <jikos@kernel.org>
Subject: Re: [PATCH 3/3] livepatch: add shadow variable sample program
In-Reply-To: <20170615134927.n53bn3ckizi6anhi@treble>
Message-ID: <alpine.LSU.2.20.1706191844100.5419@pobox.suse.cz>
References: <1496341526-19061-1-git-send-email-joe.lawrence@redhat.com> <1496341526-19061-4-git-send-email-joe.lawrence@redhat.com> <20170614142102.GA2583@pathway.suse.cz> <20170614145756.gom3zf6uv7ua423h@treble> <20170615105943.GE15013@pathway.suse.cz>
 <20170615134927.n53bn3ckizi6anhi@treble>
User-Agent: Alpine 2.20 (LSU 67 2015-01-07)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1460
Lines: 30


> > > I often wonder whether it's really a good idea to even allow the
> > > unloading of patch modules at all.  It adds complexity to the livepatch
> > > code.  Is it worth it?  I don't have an answer but I'd be interested in
> > > other people's opinion.
> > 
> > I could imagine a situation when a livepatch causes, for example,
> > performance, problems on a server because of the redirection
> > to the new code. Then it might be handy to disable the patch
> > and ftrace handlers completely.
> 
> Fair enough, though it sounds theoretical.  It would be good to know
> we're supporting actual real world use cases.

We distribute cumulative "replace_all" patches at SUSE. replace_all means 
that all previous patches are reverted in the process of application. All 
livepatch modules with zero refcount are removed. This keeps a number of 
loaded modules low and system's state well defined, which is always a good 
thing, because a customer might run into problems and we'd have to debug 
it.

It is true that it is a limitation too. Especially for state changes and 
data structure modifications. Sometimes it is easy to patch a system, but 
impossible to unpatch it. Because we don't have a consistency on a state 
level, only on a task/process level. But I perceive this also as an 
advantage. I have to always know what a livepatch does exactly and I 
discovered couple of problems just because I had to think about unloading 
of modules.

Miroslav