Re: [PATCH v9 0/7] FPGA Manager Framework and Simple FPGA Bus

[atull <atull@opensource.altera.com>]

On Thu, 23 Jul 2015, Greg KH wrote:

> On Fri, Jul 17, 2015 at 10:51:10AM -0500, atull@opensource.altera.com wrote:
> > From: Alan Tull <atull@opensource.altera.com>
> > 
> > This patchset adds two chunks plus documentation:
> >  * fpga manager core: exports ABI functions that write an image to a FPGA
> >  * DT Overlay support: simple-fpga-bus to handle FPGA from a DT overlay
> > 
> > The core's ABI is minimal to start with: only 6 functions.  This gives a
> > common interface for programming various FPGA such that any higher level
> > interfaces such as the DT Overlays or anything else that is added can be
> > shared and not be manufacturor-specific.
> > 
> > The DT Overlays support exists for the usage where the FPGA will contain
> > some "hardware" that will need drivers.  Where that use model is not
> > appealing, the core ABI can be used to add a different use model such as
> > using an FPGA as acceleration as has been discussed.
> > 
> > This patchset gets rid of the sysfs controls that allowed direct
> > control of a FPGA from userspace.
> > 
> > This patchset is under drivers/staging as the interface could change.
> > 
> > The bindings for the socpfga fpga manager already are upstreamed as
> > 1b4e119 Alan Tull : doc: add bindings document for altera fpga manager
> > 
> > The DT Support is dependent on Pantelis's dtc overlay patches from
> > https://github.com/pantoniou/dtc.git
> > and his DT overlays configfs interface patches and fixes from
> > https://github.com/pantoniou/linux-beagle-track-mainline
> > 
> > efb0c04 Pantelis Antoniou : gcl: Fix resource linking
> > 85e785e Pantelis Antoniou : ARM: DT: Enable symbols when CONFIG_OF_OVERLAY is used
> > af0321f Pantelis Antoniou : OF: DT-Overlay configfs interface (v5)
> > 4c1c675 Pantelis Antoniou : configfs: Implement binary attributes (v4)
> > 
> > 
> > Alan Tull (7):
> >   staging: usage documentation for FPGA manager core
> >   staging: usage documentation for simple fpga bus
> >   staging: add bindings document for simple fpga bus
> >   staging: fpga manager: add sysfs interface document
> >   staging: fpga manager core
> >   staging: add simple-fpga-bus
> >   staging: fpga manager: add driver for socfpga fpga manager
> > 
> >  drivers/staging/Kconfig                            |    2 +
> >  drivers/staging/Makefile                           |    1 +
> >  .../Documentation/ABI/sysfs-class-fpga-manager     |   26 +
> >  .../Documentation/bindings/simple-fpga-bus.txt     |   61 ++
> >  drivers/staging/fpga/Documentation/fpga-mgr.txt    |  117 ++++
> >  .../staging/fpga/Documentation/simple-fpga-bus.txt |   48 ++
> >  drivers/staging/fpga/Kconfig                       |   31 +
> >  drivers/staging/fpga/Makefile                      |   10 +
> >  drivers/staging/fpga/fpga-mgr.c                    |  373 ++++++++++++
> >  drivers/staging/fpga/simple-fpga-bus.c             |  323 ++++++++++
> >  drivers/staging/fpga/socfpga.c                     |  616 ++++++++++++++++++++
> 
> All drivers/staging/*/ directories need a TODO file that lists what
> needs to be done to it in order to get the code out of staging.  Please
> redo the series and add that.
> 
> >  include/linux/fpga/fpga-mgr.h                      |  127 ++++
> 
> This should be within drivers/staging/ all staging code should be
> self-contained.
> 
> Why isn't this going into the "real" part of the kernel?  Why staging?
> 

Hi Greg,

For v10 next week, I will likely break this into two patchsets, one for the
real kernel (drivers/fpga) and one for staging.

fpga-mgr.c can go into drivers/fpga since both Xilinx and Altera have
already been using this code.  It's not likely to change much.

The part that should go into staging is whatever interface is
controversial, that may change.  That's simple-fpga-bus.c and any
other interfaces that get added that use the functions exported by
fpga-mgr.c.  Maybe this 2nd patch set should be a RFC since it is still
dependent on some of Pantelis' stuff that's not in yet.

Alan Tull

> thanks,
> 
> greg k-h
>