Re: [PATCH 2/2] fpga: add FPGA manager debugfs

[Federico Vaga <federico.vaga@cern.ch>]

Hi Mortiz,

I'm not 100% into the problem to understand all cases. I'm putting on the 
table the point of view, mainly, of an user. If you say there are problems 
here or there I believe you. At the beginning, you did not say that this 
interface may introduce problems (and I'm interested in those problems since I 
implemented one and we are using it), but that you fear that it becomes the 
default (usually, being a default is a good thing).

Since you and Alan are working on this for a long time, you can read each 
other mind, but I need a more verbose email to understand ^_^'

Of course the interface must be safe, I totally agree. In order to make me 
understand what are the issues, can you list some of them? And expand your 
comment about  MMIO.

(I just did a dummy backport, and implemented that interface. So, I repeat 
myself: I do not have enough experience with this framework to understand all 
consequences, but I'm interested to know what are the risks behind this 
interface)

On Friday, August 17, 2018 12:00:34 AM CEST Moritz Fischer wrote:
> Hi Federico,
> 
> On Thu, Aug 16, 2018 at 11:21:32PM +0200, Federico Vaga wrote:
> > Hi,
> > 
> > On Thursday, August 16, 2018 10:04:51 PM CEST Alan Tull wrote:
> > > On Thu, Aug 16, 2018 at 1:59 PM, Moritz Fischer <mdf@kernel.org> wrote:
> > > > Hi Alan,
> > > 
> > > Hi Moritz,
> > > 
> > > > comments inline. While I see how this is useful, I have the
> > > > suspicion that from the moment this gets merged vendor kernels
> > > > will just default to use this ...
> > > 
> > > Yeah, I have that suspicion as well.  That's probably why I sat on
> > > this and didn't upstream it for 2 years.  But on the other hand, I
> > > keep hearing of lots of cases of people implementing this
> > > independently anyway.  At least if it is debugfs, it makes it clear
> > > that it's not intended for production use.
> > 
> > I'm one of those guys who implemented this independently.
> 
> We all have in one way or another ;) Most people on ARM run an out of tree
> patch using devicetree overlays these days I hope rather than /dev/mem
> and UIO ... or other vender solutions...
> 
> > @Mortiz
> > I do not see how this can be a bad thing (from what you wrote I guess you
> > prefer another interface). Which interface to use depends on the use case.
> > If you have this suspicion it's, I guess, because such interface it is
> > extremely easy to use.
> 
> What happens to a kernel driver doing MMIO with devices while you reload
> the entire FPGA from userland?
> 
> > @Alan
> > DebugFS can be a first step, but I would go for a normal device in /dev at
> > some point. I do not see why this should not be used in production
> 
> I'm not against having a userland interface to reprogram the FPGA, the
> Intel DFL code is a good example of a sensible one, doing so in a safe
> manner.
>
> Ideally we'll get around to have a more generic interface, as we get
> time to work on it.
> 
> - Moritz