On Wed, 27 Dec 2023 at 17:43, Raymond Mao <raymond....@linaro.org> wrote: > > Hi Ilias and Simon, > > On Wed, 27 Dec 2023 at 04:32, Ilias Apalodimas <ilias.apalodi...@linaro.org> > wrote: >> >> On Tue, 26 Dec 2023 at 11:48, Simon Glass <s...@chromium.org> wrote: >> > >> > Hi, >> > >> > On Fri, Dec 22, 2023 at 9:32 PM Raymond Mao <raymond....@linaro.org> wrote: >> > > >> > > Get devicetree from a bloblist if it exists. >> > > If not, fallback to get FDT from the specified memory address. >> > > >> > > Signed-off-by: Raymond Mao <raymond....@linaro.org> >> > > --- >> > > Changes in v2 >> > > - Refactor of board_fdt_blob_setup(). >> > > >> > > board/emulation/qemu-arm/qemu-arm.c | 12 ++++++++++-- >> > > 1 file changed, 10 insertions(+), 2 deletions(-) >> > > >> > >> > I believe this patch will not be needed, since we have fdtdec_setup() >> > to do this. >> >> Yep, it's not. The board-specific portion should be limited to what >> happens if the bloblist is *not* detected by fdtdec_setup() (assuming >> the user wants to allow a fallback) >> > But I think whether we allow a fallback or not is really something that > should be > considered by a board vendor other than a common logic in fdtdec_setup().
It's not. The bloblist is coming from the standard register calling convention. Doesn't it make more sense to have the code - Scan for a bloblist if BLOBLIST is enabled - Fallback to board-specific methods if that register does not contain a bloblist, or contains one without a DT entry > And a board vendor can also decide whether to allow a FDT that comes from > bloblist > by this custom function. I am not sure I understand this one. That decision should be in u-boot proper, the vendor should only be allowed to provide his special DT if we dont find it in a transfer list Thanks /Ilias > > [...] > > Regards, > Raymond