Re: I3C Mastership RFC

[Boris Brezillon <boris.brezillon@collabora.com>]

On Mon, 25 Nov 2019 12:03:42 +0000
Vitor Soares <Vitor.Soares@synopsys.com> wrote:

> From: Przemyslaw Gaj <pgaj@cadence.com>
> Date: Mon, Nov 25, 2019 at 11:55:16
> 
> > The 11/25/2019 11:42, Vitor Soares wrote:  
> > > 
> > > Hi Boris,
> > > 
> > > From: Boris Brezillon <boris.brezillon@collabora.com>
> > > Date: Mon, Nov 25, 2019 at 11:34:52
> > >   
> > > > On Mon, 25 Nov 2019 11:19:44 +0000
> > > > Vitor Soares <Vitor.Soares@synopsys.com> wrote:
> > > >   
> > > > > > > > > 
> > > > > > > > > I based in all version and tried to pass everything to master.c file.    
> > > > > > > > 
> > > > > > > > I'm not sure what that means, but okay.
> > > > > > > >     
> > > > > > > > > Right now my challenge it to trigger mastership request when a device 
> > > > > > > > > driver want to access to the bus but I believe we can discuss that after.    
> > > > > > > > 
> > > > > > > > That's kind of a basic feature when talking about mastership handover,
> > > > > > > > but sure, we can discuss it after your RFC has been posted.    
> > > > > > > 
> > > > > > > I need to test if the time that device.c request the mastership and the 
> > > > > > > controller has effectively the ownership of the bus is short enough to 
> > > > > > > call i3c_dev_do_priv_xfers_locked(dev->desc, xfers, nxfers) before of all 
> > > > > > > housekeeping of bus takeover.
> > > > > > >     
> > > > > > > > 
> > > > > > > > Note that I'm not super happy to have to go back to square 1 and throw
> > > > > > > > away all of the work done by Przemek, especially since Przemek was the
> > > > > > > > first one to post a patchset and he never really said he didn't
> > > > > > > > want or didn't have time to continue working on this task (not even
> > > > > > > > mentioning the time I spent reviewing those patches...).
> > > > > > > > 
> > > > > > > > If Przemek is fine with this situation I'm okay making an exception,
> > > > > > > > but be aware that it's not how we usually do: the person that posts a
> > > > > > > > patchset first leads the thing (of course, it's even better if there's
> > > > > > > > some kind of coordination before the patch is posted).    
> > > > > > > 
> > > > > > > Honestly it looks like I'm competing on this which is not the case.
> > > > > > > I just pointed out my concerns about this adoption because I see several 
> > > > > > > issues on it. The point is, at the end you can pick some parts of my sec 
> > > > > > > master code and integrate in your solution.
> > > > > > >     
> > > > > > 
> > > > > > Can you point me to the version of the patch your changes are based on?
> > > > > > And also, can you tell me what issues you faced? I would like to check
> > > > > > if they are already adressed in my code.    
> > > > > 
> > > > > I used v3 and v4. From v5, I found useful the switch case (request, 
> > > > > deliver, handoff, takeover) in hc side.
> > > > > 
> > > > > I didn't hardly test how device.c request mastership but I suspect it 
> > > > > won't work properly. When you do i3c_dev_do_priv_xfers_locked() you might 
> > > > > not be the master yet.  
> > > > 
> > > > I'm pretty sure we solved that already (that's what
> > > > i3c_master_acquire_bus_ownership() calls are supposed to take care of).
> > > > Can you be a bit more specific? What makes you think the master might
> > > > not be in control of the bus when i3c_dev_do_priv_xfers_locked() is
> > > > called?  
> > > 
> > > You are assuming that after i3c_master_acquire_bus_ownership() return, 
> > > secondary master already owns the bus. Main master can ack the MR request 
> > > and not send the CETACCMST immediately.
> > >  
> > 
> > In Cadence HC driver, I'm waiting for GETACCMST longer, polling the
> > status and after I exit from ->request_mastership(), I'm the bus owner.
> > If not, error exit code is returned and we can't make the transfers.
> > Are you able to implement the same behavior?  
> 
> You can assume everyone will do in that way. What happen if you receive a 
> request or an information from current master?

We have this ->request_mastership() method so controllers that have
this logic (MR+wait(GETACCMST) automated can still interface with the
subsystem. If your controller handles the MR/GETACCMST separately, it
shouldn't be hard to implement, and we can even provide an helper if
people end up duplicating the code.

_______________________________________________
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c