Date:	Sat, 23 Feb 2013 23:56:13 +0100
From:	Alessandro Rubini <rubini@...dd.com>
To:	rob@...dley.net
Cc:	linux-kernel@...r.kernel.org, greg@...ah.com, dcobas@...n.ch,
	cota@...ap.org, siglesias@...lia.com, linux-doc@...r.kernel.org
Subject: Re: [PATCH 4/8] FMC: add documentation for the core

> Acked-by: Rob Landley <rob@...dley.net>

Thanks.

> What is this *note thingy:: syntax? It recurs a lot. Some sort of  
> reference into the PDF you started out with a link to, maybe?

My documentation is texinfo and this is the ASCII output, manually
split into different files. I fixed some of the internal references
when I noted them.  But since it's manual work, I didn't do it
carefully.  I suspect I'll need to V2 for some reason anyways.

> A basic concept of this framework is that you have a 70x75 PCB? Is this  
> part of that ANSI-VITA standard?

Yes. the form-factor is fixed, like PC104 in the PC world.  Here's the
jpg of a carrier: http://www.ohwr.org/attachments/download/552 (from
http://www.ohwr.org/projects/spec/wiki). And here's the jpg of a
mezzanine:
http://www.ohwr.org/attachments/download/833/FmcAdc100M14b4cha_top_small.JPG

You find more on the net, also from commercial vendors.

>> +FMC, as such, is not a bus in the usual meaning of the term, because
>> +most carriers have only one connector, and carriers with several
>> +connectors have completely separate electrical connections to them.
>> +This package, however, implements a bus as a software abstraction.
> 
> USB is point to point connections with switches in between. It's still  
> got B in the acronym. I'm not sure what you're saying here.

FMC has no switches, no hotplug, not even shared wires like ISA.  I
took a while to explain some hardware guys that it ought to be a
"bus" in software. That's why I have the paragraph.

>> +What is SDB
>> +***********
>> +
>> +SDB (Self Describing Bus) is a set of data structures that we use for
>> +enumerating the internal structure of an FPGA image. We also use it  
>> as
>> +a filesystem inside the FMC EEPROM.
> 
> Are you trying to document infrastructure to implement a standard, or a  
> bespoke driver for a specific piece of hardware? How much of this is  
> generic? Are there other vendors who might someday want to use this  
> code?

SDB is already implemented and is generic.  We enumerate the devices
inside the FPGA already, and it only costs a few lines of VHDL (using
the support files by Wesley Terpstra) to describe your own FPGA
design. Wesley even discovers and enumerates devices over the
Ethernet, by means of Ethernet-driven device access and SDB-aware host
code -- all code is published.

The SDB specification is discussed in a mailing list and is public
(http://www.ohwr.org/attachments/download/1486/sdb-1.0.pdf). By having
64-bit vendor identifiers we encourage anyone who finds SDB useful to
become a vendor. There are a few vendors already, although I admit the
early adopters are involved in the same set of projects.

> Anyway, no serious objection, but I note that reading to this point I  
> didn't feel I had enough information to wrap my head around what it's  
> for. It's documentation by people who already know this stuff, for  
> people who already know this stuff. (There's a long tradition of that.  
> Oh well.)

I'm aware of this. But it's the same if you read USB documentation
without ever seeing one such device (I remember the feeling :).  FMC
is not common on the shelf, but it is already renown in high-profile
I/O environments.

Thanks
/alessandro
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists