Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] Revert "kbuild: use -flive-patching when CONFIG_LIVEPATCH
 is enabled"
To:     Josh Poimboeuf <jpoimboe@redhat.com>, live-patching@vger.kernel.org
Cc:     linux-kernel@vger.kernel.org,
        Peter Zijlstra <peterz@infradead.org>,
        Randy Dunlap <rdunlap@infradead.org>
References: <696262e997359666afa053fe7d1a9fb2bb373964.1595010490.git.jpoimboe@redhat.com>
From:   Joe Lawrence <joe.lawrence@redhat.com>
Message-ID: <fc7d4932-a043-1adc-fd9b-96211c508f64@redhat.com>
Date:   Sun, 19 Jul 2020 23:35:20 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.5.0
MIME-Version: 1.0
In-Reply-To: <696262e997359666afa053fe7d1a9fb2bb373964.1595010490.git.jpoimboe@redhat.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 7/17/20 2:29 PM, Josh Poimboeuf wrote:
> Use of the new -flive-patching flag was introduced with the following
> commit:
> 
>    43bd3a95c98e ("kbuild: use -flive-patching when CONFIG_LIVEPATCH is enabled")
> 
> This flag has several drawbacks:
> 
> [ ... snip ... ]
> 
> - While there *is* a distro which relies on this flag for their distro
>    livepatch module builds, there's not a publicly documented way to
>    create safe livepatch modules with it.  Its use seems to be based on
>    tribal knowledge.  It serves no benefit to those who don't know how to
>    use it.
> 
>    (In fact, I believe the current livepatch documentation and samples
>    are misleading and dangerous, and should be corrected.  Or at least
>    amended with a disclaimer.  But I don't feel qualified to make such
>    changes.)

FWIW, I'm not exactly qualified to document source-based creation 
either, however I have written a few of the samples and obviously the 
kselftest modules.

The samples should certainly include a disclaimer (ie, they are only for 
API demonstration purposes!) and eventually it would be great if the 
kselftest modules could guarantee their safety as well.  I don't know 
quite yet how we can automate that, but perhaps some kind of post-build 
sanity check could verify that they are in fact patching what they 
intend to patch.

As for a more general, long-form warning about optimizations, I grabbed 
Miroslav's LPC slides from a few years back and poked around at some 
IPA-optimized disassembly... Here are my notes that attempt to capture 
some common cases:

http://file.bos.redhat.com/~jolawren/klp-compiler-notes/livepatch/compiler-considerations.html

It's not complete and I lost steam about 80% of the way through today. 
:)  But if it looks useful enough to add to Documentation/livepatch, we 
can work on it on-list and try to steer folks into using the automated 
kpatch-build, objtool (eventually) or a source-based safety checklist. 
The source-based steps have been posted on-list a few times, but I think 
it only needs to be formalized in a doc.

-- Joe