On Fri, Jun 25, 2021 at 12:58:49PM -0700, trix@xxxxxxxxxx wrote:
> From: Tom Rix <trix@xxxxxxxxxx>
>
> If the fpga_image_info flags FPGA_MGR_REIMAGE bit is set
> swap out the reconfig ops for the reimage ops and do
> the load.
>
> Signed-off-by: Tom Rix <trix@xxxxxxxxxx>
> ---
> drivers/fpga/fpga-mgr.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/fpga/fpga-mgr.c b/drivers/fpga/fpga-mgr.c
> index c8a6bfa037933..5e53a0508087a 100644
> --- a/drivers/fpga/fpga-mgr.c
> +++ b/drivers/fpga/fpga-mgr.c
> @@ -419,6 +419,9 @@ int fpga_mgr_load(struct fpga_manager *mgr, struct fpga_image_info *info)
> {
> const struct fpga_manager_update_ops *uops = &mgr->mops->reconfig;
>
> + if (info->flags & FPGA_MGR_REIMAGE)
> + uops = &mgr->mops->reimage;
> +
So seems there is no big difference on processing 'reconfig' & 'reimage'
in FPGA mgr framework. We may not have to introduce 2 sets of ops in
fpga_mgr.
Could we just reuse the fpga_mgr_ops for reimage? The
fpga_mgr_ops.write_init() will pass the fpga_image_info to device driver,
and driver could be aware of the REIMAGE flag. Just like the handling of
other flags in fpga_image_info.
Thanks,
Yilun
> if (info->sgt)
> return fpga_mgr_buf_load_sg(mgr, info, info->sgt, uops);
> if (info->buf && info->count)
> --
> 2.26.3
