Re: [PATCHv1 3/4] dt-bindings: fpga: add authenticate-fpga-config property

[Xu Yilun]

On Tue, Nov 17, 2020 at 09:39:55AM -0600, Richard Gong wrote:
> 
> 
> On 11/16/20 8:24 PM, Xu Yilun wrote:
> >On Mon, Nov 16, 2020 at 08:14:52AM -0600, Richard Gong wrote:
> >>
> >>Hi Yilun,
> >>
> >>On 11/15/20 8:47 PM, Xu Yilun wrote:
> >>>On Sun, Nov 15, 2020 at 11:21:06AM -0800, Moritz Fischer wrote:
> >>>>Hi Richard,
> >>>>
> >>>>On Thu, Nov 12, 2020 at 12:06:42PM -0600, richard.gong@xxxxxxxxxxxxxxx wrote:
> >>>>>From: Richard Gong <richard.gong@xxxxxxxxx>
> >>>>>
> >>>>>Add authenticate-fpga-config property for FPGA bitstream authentication.
> >>>>>
> >>>>>Signed-off-by: Richard Gong <richard.gong@xxxxxxxxx>
> >>>>>---
> >>>>>  Documentation/devicetree/bindings/fpga/fpga-region.txt | 1 +
> >>>>>  1 file changed, 1 insertion(+)
> >>>>>
> >>>>>diff --git a/Documentation/devicetree/bindings/fpga/fpga-region.txt b/Documentation/devicetree/bindings/fpga/fpga-region.txt
> >>>>>index e811cf8..7a512bc 100644
> >>>>>--- a/Documentation/devicetree/bindings/fpga/fpga-region.txt
> >>>>>+++ b/Documentation/devicetree/bindings/fpga/fpga-region.txt
> >>>>>@@ -187,6 +187,7 @@ Optional properties:
> >>>>>  - external-fpga-config : boolean, set if the FPGA has already been configured
> >>>>>  	prior to OS boot up.
> >>>>>  - encrypted-fpga-config : boolean, set if the bitstream is encrypted
> >>>>>+- authenticate-fpga-config : boolean, set if do bitstream authentication
> >>>>It is unclear to me from the description whether this entails
> >>>>authentication + reconfiguration or just authentication.
> >>>>
> >>>>If the latter is the case this should probably be described as such.
> >>>
> >>>If it is just authentication, do we still need to disable bridges in
> >>>fpga_region_program_fpga?
> >>>
> >>
> >>Yes.
> >>
> >>Except for the actual configuration of the device, the authentication
> >>feature is the same as FPGA configuration.
> >
> >FPGA Bridges gate bus signals between a host and FPGA. So the FPGA
> >region could not be accessed by host when doing configuration. But for
> >this authentication, we are just writing the flash, we don't actually
> >touch the FPGA soft logic. The host should still be able to operate on
> >the old logic before reboot, is it?
> >
> Yes, it's feasible in theory but doesn't make much sense in practice. I
> prefer to keep fpga_region_program_fpga() unchanged.

I'm thinking of the case of inband reprograming, that the QSPI flash
controller itself is embedded in FPGA soft logic, then maybe host still
need to access FPGA on authentication.

Thanks,
Yilun

> >>
> >>>I'm wondering if the FPGA functionalities could still be working when
> >>>the authenticating is ongoing, or when the authenticating is failed.
> >>>
> >>
> >>
> >>
> >>>Thanks,
> >>>Yilun
> >>>
> >>>>
> >>>>>  - region-unfreeze-timeout-us : The maximum time in microseconds to wait for
> >>>>>  	bridges to successfully become enabled after the region has been
> >>>>>  	programmed.
> >>>>>-- 
> >>>>>2.7.4
> >>>>>
> >>>>
> >>>>Thanks