Re: [PATCH v15 00/35] Introduce SCMI Telemetry support

[Jonathan Cameron]

On Wed, 11 Nov 2026 10:15:16 +0000
Cristian Marussi <cristian.marussi@xxxxxxx> wrote:

> On Tue, Nov 10, 2026 at 04:38:09PM -0800, Jonathan Cameron wrote:
> > > The upcoming SCMI v4.0 specification [0] introduces a new SCMI protocol
> > > dedicated to System Telemetry.
> > >   
> > 
> > Hi Cristian,  
> 
> Hi Jonathan,
> 
> thanks for your time...I sincerely hope you're travelling abroad on some
> remote timezone given the time these reviews came in :D

I'm based in Pacific time these days so was only 4pm ;)

> 
> > 
> > So, the main issue here is pretty much summarized by:
> > 
> >  28 files changed, 12801 insertions(+), 99 deletions(-)
> > (obviously there are tests etc, but it's still > 10k lines)
> >  
> 
> Usually I try to implement the complete set of features of a new protocol
> but indeed this new Telemetry thing is huge by itself so it became quickly
> overwhelming to review...imagine to code and test it.
> 
> > There are relatively few reviewers who would even consider taking that on.
> > I foolishly thought it wouldn't be too bad (and I review a lot of code).
> > By about 3/4 of the way though, probably much earlier, I know the quality
> > of review went down considerably. Honestly it was probably a waste of
> > time continuing but I'm stubborn like that.
> > 
> > So, unless you have a path to bribing or sweet talking various folk to
> > deal with reviewing this with each having to set aside multiple days
> > to do it, likely repeating for a couple of versions yet, I think you've
> > got to figure out out how to make the first patch set more manageable.
> > Probably don't split it up as much as the mpam or CCA folk have had to,
> > but you have a lot of colleagues who are experiencing the same problem
> > at the moment. They may be able to advise.
> > 
> > So it would be nice if what was in that first patch set was 'functional'
> > but I don't think it needs to function on every systems or support more
> > than an absolutely minimal feature set.  Just fail to probe when you see
> > something that isn't supported by that minimal set if that means
> > the thing won't be functional.
> > 
> > Once that's in place, over a couple of additional patch sets you can
> > get to everything you have here.
> > 
> > So what if you choose one data path (shmti perhaps), no notifications,
> > events, maybe even simplify to one of the readback paths.
> > Maybe even drop event groups initially.   
> 
> Based on our real world initial customer of this I can certainly drop a
> few features that are not supported anyway by the real FW and move it to
> a later patchset, stuff like groups, notifications and single sample
> reads, but all of the rest is hardly removable while keeping the system
> functional and usable.

For purposes of upstream. I'm not sure we care if a particular implementation
is useable with the feature set initially supported.  We just care if
there is a path to test the subset supported (hacks / emulation, whatever is fine).

> 
> > Possibly even drop timestamps.  Definitely drop things like boot-on
> > support and debugfs (unless you are using that as a cheeky 'functional'
> > test ahead of the main userspace interface.  
> 
> Not sure what you mean by drop timestamps, definitelly cannot drop
> boot-on (it is how systems are configured indeed) neither debugfs used
> for bring-up and FW debug.

Timestamp variety is complex. Fairly sure you are allowed to report no
timestamps at all as one option. If you pretend they all look like that
initially it might bring some simplifications. Anyhow, you know this
code better than I do, so maybe that is harder than I'm thinking.

I'd still push boot-on out for a follow on patch set.  Sure can
aim to land that one quickly! Sure customers will
get irritated by lack of support for what they have. Taking equivalent
of MPAM.  The initial support for that covered pretty much nothing on
real platforms. Customers were (and still are) irritated. But that
allowed progress in the right direction.

> 
> > 
> > How much would that slim things down?  It doesn't even have to be
> > useful, just testable.  Realistically I think the hope of getting
> > enough review doubles if you can halve the size of what you are asking
> > people to look at.  
> 
> I am not sure how many lines I can shed with the above features drops but
> I'll give it a go anyway at slimming down the series after I have gone
> through this round of Sashiko pains.

If we have to fallback to breaking it into steps that don't do much
more than print a message to say 'stuff found' then do that. It's ugly
but MPAM did that, RME is doing that. No one likes dead code, but it
does allow progress to be made.  Ideal is end to end with subset of
features then add features but not always achievable.

Normal rule of thumb for landing a feature is sub 2k lines of code
because that is about what a reviewer can digest in one go.

Very rare anything above 5k makes progress. As Sudeep mentioned, a
lot of folk will open the cover letter, see loc and close it again unread.

Jonathan