Re: [PATCH iwl-next 05/14] libeth: add control queue support

[Leon Romanovsky <leon@kernel.org>]

On Thu, Apr 10, 2025 at 03:59:23PM +0200, Larysa Zaremba wrote:
> On Thu, Apr 10, 2025 at 04:44:43PM +0300, Leon Romanovsky wrote:
> > On Thu, Apr 10, 2025 at 03:05:19PM +0200, Alexander Lobakin wrote:
> > > From: Leon Romanovsky <leon@kernel.org>
> > > Date: Thu, 10 Apr 2025 14:23:49 +0300
> > > 
> > > > On Thu, Apr 10, 2025 at 12:44:33PM +0200, Larysa Zaremba wrote:
> > > >> On Thu, Apr 10, 2025 at 11:21:37AM +0300, Leon Romanovsky wrote:
> > > >>> On Tue, Apr 08, 2025 at 02:47:51PM +0200, Larysa Zaremba wrote:
> > > >>>> From: Phani R Burra <phani.r.burra@intel.com>
> > > >>>>
> > > >>>> Libeth will now support control queue setup and configuration APIs.
> > > >>>> These are mainly used for mailbox communication between drivers and
> > > >>>> control plane.
> > > >>>>
> > > >>>> Make use of the page pool support for managing controlq buffers.
> > 
> > <...>
> > 
> > > >> Module dependencies are as follows:
> > > >>
> > > >> libeth_rx and libeth_pci do not depend on other modules.
> > > >> libeth_cp depends on both libeth_rx and libeth_pci.
> > > >> idpf directly uses libeth_pci, libeth_rx and libeth_cp.
> > > >> ixd directly uses libeth_cp and libeth_pci.
> > > > 
> > > > You can do whatever module architecture for netdev devices, but if you
> > > > plan to expose it to RDMA devices, I will vote against any deep layered
> > > > module architecture for the drivers.
> > > 
> > > No plans for RDMA there.
> > > 
> > > Maybe link the whole kernel to one vmlinux then?
> > 
> > It seems that you didn't understand at all about what we are talking
> > here. Please use the opportunity that you are working for the same
> > company with Larysa and ask her offline. She understood perfectly about
> > which modules we are talking.
> >
> 
> While I do understand what kind of module relationship you consider problematic,

Awesome, thanks.

> I still struggle to understand why stateless lib hierarchy can be problematic.

As I said already, I wrote my remark as a general comment. It is just
a matter of time when perfectly working system will be changed to less
working one. So when you and Alexander are focused to see what is wrong
now, I see what can be in the future.

To make it clear, even for people who sentimentally attached to libeth code:
 I didn't ask to change anything, just tried to understand why
 you did it like you did it.

> The fixes that you linked relate more to problematic resource sharing, of which 
> libeth has none, it does not have its own memory or its own threads, this is 
> just collection of data structures and functions.

It is just a matter of time and you will get same issues like I posted.

Thanks

> 
> > > 
> > > > 
> > > > BTW, please add some Intel prefix to the modules names, they shouldn't
> > > > be called in generic names like libeth, e.t.c
> > > 
> > > Two modules with the same name can't exist within the kernel. libeth was
> > > available and I haven't seen anyone wanting to take it. It's not common
> > > at all to name a module starting with "lib".
> > 
> > Again, please talk with Larysa. ETH part is problematic in libeth name
> > and not LIB.
> > 
> > Thanks