Re: [PATCH v12 1/3] fpga: fpga-mgr: support bitstream offset in image buffer

[Xu Yilun <yilun.xu@intel.com>]

On Sun, May 22, 2022 at 09:01:45PM +0800, Xu Yilun wrote:
> On Fri, May 13, 2022 at 07:27:53PM +0300, Ivan Bornyakov wrote:
> > At the moment FPGA manager core loads to the device entire image
> > provided to fpga_mgr_load(). But it is not always whole FPGA image
> > buffer meant to be written to the device. In particular, .dat formatted
> > image for Microchip MPF contains meta info in the header that is not
> > meant to be written to the device. This is issue for those low level
> > drivers that loads data to the device with write() fpga_manager_ops
> > callback, since write() can be called in iterator over scatter-gather
> > table, not only linear image buffer. On the other hand, write_sg()
> > callback is provided with whole image in scatter-gather form and can
> > decide itself which part should be sent to the device.
> > 
> > Add header_size and data_size to the fpga_image_info struct and adjust
> > fpga_mgr_write() callers with respect to them.
> > 
> >   * info->header_size indicates part at the beginning of image buffer
> >     that is *not* meant to be written to the device. It is optional and
> >     can be 0.
> > 
> >   * info->data_size is the size of actual bitstream data that *is* meant
> >     to be written to the device, starting at info->header_size from the
> >     beginning of image buffer. It is also optional and can be 0, which
> >     means bitstream data is up to the end of image buffer.
> > 
> > Also add parse_header() callback to fpga_manager_ops, which purpose is
> > to set info->header_size and info->data_size. At least
> > initial_header_size bytes of image buffer will be passed into
> > parse_header() first time. If it is not enough, parse_header() should
> > set desired size into info->header_size and return -EAGAIN, than it will
> > be called again with greater part of image buffer on the input.
> > 
> > Signed-off-by: Ivan Bornyakov <i.bornyakov@metrotek.ru>
> > ---
> >  drivers/fpga/fpga-mgr.c       | 150 ++++++++++++++++++++++++++--------
> >  include/linux/fpga/fpga-mgr.h |  13 ++-
> >  2 files changed, 127 insertions(+), 36 deletions(-)
> > 
> > diff --git a/drivers/fpga/fpga-mgr.c b/drivers/fpga/fpga-mgr.c
> > index a3595ecc3f79..c6ca395909a0 100644
> > --- a/drivers/fpga/fpga-mgr.c
> > +++ b/drivers/fpga/fpga-mgr.c
> > @@ -74,6 +74,15 @@ static inline int fpga_mgr_write_complete(struct fpga_manager *mgr,
> >  	return 0;
> >  }
> >  
> > +static inline int fpga_mgr_parse_header(struct fpga_manager *mgr,
> > +					struct fpga_image_info *info,
> > +					const char *buf, size_t count)
> > +{
> > +	if (buf && mgr->mops->parse_header)
> > +		return mgr->mops->parse_header(mgr, info, buf, count);
> > +	return 0;
> > +}
> > +
> >  static inline int fpga_mgr_write_init(struct fpga_manager *mgr,
> >  				      struct fpga_image_info *info,
> >  				      const char *buf, size_t count)
> > @@ -136,32 +145,61 @@ void fpga_image_info_free(struct fpga_image_info *info)
> >  EXPORT_SYMBOL_GPL(fpga_image_info_free);
> >  
> >  /*
> > - * Call the low level driver's write_init function.  This will do the
> > - * device-specific things to get the FPGA into the state where it is ready to
> > - * receive an FPGA image. The low level driver only gets to see the first
> > - * initial_header_size bytes in the buffer.
> > + * Call the low level driver's parse_header then write_init functions.
> > + * This will do the device-specific things to get the FPGA into the state
> > + * where it is ready to receive an FPGA image. If parse_header sets
> > + * info->header_size, the low level driver's write_init only gets to see the
> > + * first info->header_size bytes in the buffer, mgr->mops->initial_header_size
> > + * otherwise. If neither initial_header_size nor header_size are not set,
> > + * write_init will not get any bytes of image buffer.
> 
> Could we always initialize the info->header_size = initial_header_size
> at early stage, maybe in fpga_mgr_load(), and only query
> info->header_size afterward. This is to make the logic for header size
> simpler.

Sorry I remember some existing drivers are using initial_header_size but
still deal with no buffer offset in write(), so we cannot force
info->header_size be filled. Forget about the previous comment.

Thanks,
Yilun