Re: [5.4.y] Found 27 commits that might missed

[Sasha Levin <sashal@kernel.org>]

On Sat, Sep 05, 2020 at 10:02:20AM +0200, SeongJae Park wrote:
>From: SeongJae Park <sjpark@amazon.com>
>
>On Sat, 5 Sep 2020 09:09:46 +0200 Greg KH <gregkh@linuxfoundation.org> wrote:
>
>> On Fri, Sep 04, 2020 at 04:17:48PM +0200, SeongJae Park wrote:
>> > > <snip>
>> > >
>> > > Stopping right here, if you have fixes that will not cleanly apply, and
>> > > you think they should be applied, please fix them and send the proper
>> > > backport.  I don't have the cycles to do these on my own.
>> > >
>> > > Same for anything else here that you think should be applied but does
>> > > not cleanly build/apply.
>> >
>> > Totally agreed.  Actually, I posted a similar report[1] before and received
>> > similar response.  I promised to back-port some of those by myself.  That's
>> > still in my TODO list, but I was unable to get a time to revisit it quite long
>> > time.  From this, I realized that it wouldn't be easy to review, test, and
>> > backport all of the such suspicious things by myself.  Scaling up to multiple
>> > stable series (the tool says there are 152 fixes and 147 mentions for 4.9.y)
>> > seems impossible.
>> >
>> > For the reason, I updated the tool to make the report to be sent to not only
>> > the stable maintainers but also the authors of the suspicious commits, because
>> > the review / test / backport of their own commits would be much easier that
>> > others.  As a result, we were able to find one suspended commit:
>> > https://lore.kernel.org/stable/CAKfTPtAkOes+HmVabRazhCBBUo0M+QW38q3Zzj_O3O+Ghvc1pA@mail.gmail.com/
>>
>> That work had already been done before your email was sent.
>>
>> I too can write a tool that sends out "this patch might be for stable,
>> will you do the work for it!" emails, but that's a bit rude to ask
>> others to do your work for you, don't you agree?  By asking me and
>> others to dig through this list, when you said you don't have the time
>> to do so, feels very odd to me.
>
>I thought the tool and this report are like a very simple form of the CI test
>bots like 0day, syzbot, or some kind of static analyzers.  Mine has quite large
>number of false positives, though.  Actually that was my only one concern.
>Therefore I thought asking the authors to check this could be a little bit
>annoying and therefore I asked them to let me know if they don't want this.
>I also thought making an explicit list of false-positive 'Fixes:' could help
>someone in the community.  Also, I didn't intend to make others do my work
>instead, but I just wanted to help the community finding missed patches.

And that's a good goal, but the help we need to accomplish that is in
the manual parts of the process which we can't automate: figuring out
whether a patch really needs to be backported, and doing the actual
backport.

I'd encourage you to pick a small subset of your results and try doing
just that - it's not "all of nothing" and even doing a few of these will
help.

-- 
Thanks,
Sasha