Re: [PATCH v1 5/5] sbm: SandBox Mode documentation

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 15 Feb 2024 10:11:05 +0100
Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:

> On Wed, Feb 14, 2024 at 08:42:54PM +0100, Petr Tesařík wrote:
> > On Wed, 14 Feb 2024 19:48:52 +0100
> > Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >   
> > > On Wed, Feb 14, 2024 at 05:31:12PM +0100, Petr Tesařík wrote:  
> > > > On Wed, 14 Feb 2024 16:11:05 +0100
> > > > Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > > >     
> > > > > On Wed, Feb 14, 2024 at 03:55:24PM +0100, Petr Tesařík wrote:    
> > > > > > OK, so why didn't I send the whole thing?
> > > > > > 
> > > > > > Decomposition of the kernel requires many more changes, e.g. in linker
> > > > > > scripts. Some of them depend on this patch series. Before I go and
> > > > > > clean up my code into something that can be submitted, I want to get
> > > > > > feedback from guys like you, to know if the whole idea would be even
> > > > > > considered, aka "Fail Fast".      
> > > > > 
> > > > > We can't honestly consider this portion without seeing how it would
> > > > > work, as we don't even see a working implementation that uses it to
> > > > > verify it at all.
> > > > > 
> > > > > The joy of adding new frameworks is that you need a user before anyone
> > > > > can spend the time to review it, sorry.    
> > > > 
> > > > Thank your for a quick assessment. Will it be sufficient if I send some
> > > > code for illustration (with some quick&dirty hacks to bridge the gaps),
> > > > or do you need clean and nice kernel code?    
> > > 
> > > We need a real user in the kernel, otherwise why would we even consider
> > > it?  Would you want to review a new subsystem that does nothing and has
> > > no real users?  If not, why would you want us to?  :)  
> > 
> > Greg, please enlighten me on the process. How is something like this
> > supposed to get in?  
> 
> If you were in our shoes, what would you want to see in order to be able
> to properly review and judge if a new subsystem was ok to accept?
> 
> > Subsystem maintainers will not review code that depends on core features
> > not yet reviewed by the respective maintainers. If I add only the API
> > and a stub implementation, then it brings no benefit and attempts to
> > introduce the API will be dismissed. I would certainly do just that if
> > I was a maintainer...  
> 
> Exactly, you need a real user.

Er, what I was trying to say was rather: You need a real implementation
of a core feature before a subsystem maintainer considers using it for
their subsystem.

But I get your point. I need *both*.

> > I could try to pack everything (base infrastructure, arch
> > implementations, API users) into one big patch with pretty much
> > everybody on the Cc list, but how is that ever going to get reviewed?  
> 
> How are we supposed to know if any of this even works at all if you
> don't show that it actually works and is useful?  Has any of that work
> even been done yet?  I'm guessing it has (otherwise you wouldn't have
> posted this), but you are expecting us to just "trust us, stuff in the
> future is going to use this and need it" here.

Understood.

> Again, we can not add new infrastructure for things that have no users,
> nor do you want us to.  Ideally you will have at least 3 different
> users, as that seems to be the "magic number" that shows that the
> api/interface will actually work well, and is flexible enough.  Just
> one user is great for proof-of-concept, but that usually isn't good
> enough to determine if it will work for others (and so it wouldn't need
> to be infrastructure at all, but rather just part of that one feature on
> its own.)
> 
> > Should I just go and maintain an out-of-tree repo for a few years,
> > hoping that it gets merged one day, like bcachefs? Is this the way?  
> 
> No, show us how this is going to be used.

OK, working on it.

> Again, think about what you would want if you had to review this.

Review, or merge? For a review, I would want enough information to
understand what it is *and* where it is going.

As a matter of fact, hpa does not like the x86 implementation. For
reasons that I do not fully understand (yet), but if the concept turns
out to be impractical, then my submission will serve a purpose, as I
can save myself (and anybody else with a similar idea) a lot of work by
failing fast.

Is this a valid way to get early feedback?

Thanks,
Petr T





[Index of Archives]     [Kernel Newbies]     [Security]     [Netfilter]     [Bugtraq]     [Linux FS]     [Yosemite Forum]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]     [Linux Resources]

  Powered by Linux