Re: [PATCH AUTOSEL 5.9 22/33] vhost scsi: add lun parser helper

[Sasha Levin]

On Fri, Dec 04, 2020 at 06:08:13PM +0100, Paolo Bonzini wrote:
> On 04/12/20 16:49, Sasha Levin wrote:
> > On Fri, Dec 04, 2020 at 09:27:28AM +0100, Paolo Bonzini wrote:
> > > On 01/12/20 00:59, Sasha Levin wrote:
> > > > It's quite easy to NAK a patch too, just reply saying "no" and it'll be
> > > > dropped (just like this patch was dropped right after your first reply)
> > > > so the burden on maintainers is minimal.
> > >
> > > The maintainers are _already_ marking patches with "Cc: stable".  
> > > That
> >
> > They're not, though. Some forget, some subsystems don't mark anything,
> > some don't mark it as it's not stable material when it lands in their
> > tree but then it turns out to be one if it sits there for too long.
>
> That means some subsystems will be worse as far as stable release 
> support goes.  That's not a problem:
>
> - some subsystems have people paid to do backports to LTS releases 
> when patches don't apply; others don't, if the patch doesn't apply the 
> bug is simply not fixed in LTS releases

Why not? A warning mail is originated and folks fix those up. I fixed a
whole bunch of these myself for subsystems I'm not "paid" to do so.

> - some subsystems are worse than others even in "normal" releases :)

Agree with that.

> > > (plus backports) is where the burden on maintainers should start 
> > > and end.  I don't see the need to second guess them.
> >
> > This is similar to describing our CI infrastructure as "second
> > guessing": why are we second guessing authors and maintainers who are
> > obviously doing the right thing by testing their patches and reporting
> > issues to them?
>
> No, it's not the same.  CI helps finding bugs before you have to waste 
> time spending bisecting regressions across thousands of commits.  The 
> lack of stable tags _can_ certainly be a problem, but it solves itself 
> sooner or later when people upgrade their kernel.

If just waiting with fixing issues is ok until a user might "eventually"
upgrade is acceptable then why bother with a stable tree to begin with?

--
Thanks,
Sasha