Re: [PATCH] fpga: add simple userspace interface to trigger FPGA programming

[Moritz Fischer <mdf@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2670 bytes --]

On Sun, Dec 10, 2017 at 02:59:36PM -0800, Florian Fainelli wrote:
> 
> 
> On 12/10/2017 02:44 PM, Thomas Petazzoni wrote:
> > Hello,
> > 
> > On Sat, 9 Dec 2017 22:03:06 -0600, Alan Tull wrote:
> > 
> >>> We can actually use the Linux device driver model to do that though.
> >>> Assuming all peripherals behind the FPGA do have the FPGA manager's
> >>> struct device as a parent (which they should), when there is a bistream
> >>> load request, we ought to be able to teardown all of these child struct
> >>> device's and their corresponding drivers (unbind), load the bistream,
> >>> and then rebind the drivers accordingly.  
> >>
> >> Hi Florain,
> >>
> >> FPGA regions are a way of handling all that plus the bridges [1] or
> >> are you proposing something else?  An FPGA region knows what manager
> >> to use and what bridges (if any) needs to be disabled while
> >> programming is happening.  Applying a DT overlay targeting a FPGA
> >> region is used to program the FPGA.
> > 
> > I thought my initial patch made it clear: DT overlays are not an
> > applicable solution for my use case, for two reasons:
> > 
> >  1. My platform doesn't use Device Tree at all.
> > 
> >  2. The device implemented in the FPGA is self-discoverable because it
> >     sits on a PCI bus, so there is no point in doing a static
> >     description of the device in a DT overlay.
> 
> Agreed, sorry for side tracking the discussion a bit on that topic.

I do agree that wee need to find a solution for this. We should not
force people to use DT overlays in a system that is discoverable,
or out of tree hacks.

> 
> > 
> > So, solutions based on DT overlays are quite irrelevant for my use
> > case. Am I missing something here ?
> 
> I don't think you are, in fact, as it stands today, unless your FPGA has
> a bistream which is already loaded, in which case you don't really need
> the FPGA manager framework at all, this framework is completely useless
> in offering people the ability to load bistreams on non-DT enabled
> platforms without resorting to out of tree hacks like this one for instance:
> 
> https://github.com/ffainelli/linux/commit/78a4824165d7bf627c1b2dbb645ce013185331de
> 
> So what needs to be done so we can allow people to take full advantage
> of the fact that FPGAs are (re)programmable, even when the platform does
> not use DT?


I like your suggestion of tying into the device model, to tear down
everything underneath.

Would making the PCIe root complex a child to a FPGA region work? Seems
clumsy. Do we have other examples of devices reloading firmware
triggered by userland?

- Moritz

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]