Re: i3c application

[Boris Brezillon <bbrezillon@kernel.org>]

Hello Ryan,

On Sat, 2 Feb 2019 03:01:51 +0000
Ryan Chen <ryan_chen@aspeedtech.com> wrote:

> Hello Boris,
> 		Thanks a lot.
> 
> >+Vitor who started a discussion around i3c-tools and the associate
> > kernel <-> userspace interface.  
> 
> Does there have open source repo in internet? 

AFAICT there's nothing implemented yet, we just started discussing it.

> 
> 
> > Hi all,
> > 	I just work with I3C, does there have any application like i2c-tools  
> I can used?
> 
> >No, but I'll ask the same question I already asked Vitor: what would you  
> use i3c-tools for? What should it contain? I'm not against the idea, but I'd
> like to delimit the scope of the userspace >interface.
> 
> 
> Sure, it definitely a user space application.
> I will connect with some i3c device or have loop back interact. I3c master
> <-> i3c slave 

Sorry, I don't get it. I don't know what you mean by loop back. I also
don't see why you absolutely need a user-space app to interact with a
device when you can develop a simple driver running in kernel space,
and I see no answers to these questions in your reply.


> 
> > 	Or any Linux i3c application I can have i3c slave and master loop  
> test? 
> 
> >>Not sure I get this request correctly. Are you talking about a dummy slave  
> driver that would return any data it receives to the TX queue so that the
> next read request coming from the >>master returns what the master wrote in
> its previous write access?
> 
> 
> >>If that's what you have in mind, then it's definitely not supported.
> >>But before we even consider doing that, we should first introduce the  
> concept of I3C slave controller and then decide how we want to expose slave
> features. Note that Vitor and I disagree >>on the solution: I think we
> should mimic the USB gadget approach (where you can attach a generic profile
> to any USB device controller), and Vitor thinks slave IPs should have their
> >>profile/feature-set hardcoded in the driver (which works fine for  
> hardcoded slave IPs, but is not that great if the slave block is generic).
> 
> Thanks, if I can refer from with I3c-tools, I can start for i3c-master and
> i3c slave dummy test.

As I already said above, there's no i3c-tools yet, and there won't be
until we are clear on what this project should contain and what the
kernel <-> userspace interface should look like.

Can we have a more detailed description of your use case please?

Regards,

Boris

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