Hi Alper, On Thu, 22 Oct 2020 at 14:37, Alper Nebi Yasak <[email protected]> wrote: > > It's possible to chainload U-Boot proper from the vendor firmware in > rk3399 chromebooks, but the way the vendor firmware sets up clocks is > somehow different than what U-Boot expects. This causes the display to > stay devoid of content even though vidconsole claims to work (with > patches in process of being upstreamed). > > This is meant to be a rk3399 version of commit d3cb46aa8c41 ("rockchip: > Init clocks again when chain-loading") which can detect the discrepancy, > but this patch can not so it always re-inits. > > Signed-off-by: Alper Nebi Yasak <[email protected]> > --- > The rk3288 version has rockchip_vop_set_clk in #ifndef CONFIG_BUILD_SPL, > and checks if that setup is already done before that's called. I think I > can do the #ifndef to avoid unnecessary re-inits for rk3399 as well, but > the vop clocks are set differently for each chip so I don't know how > correct that'd be. > > The pmucru setup is #ifndef CONFIG_BUILD_SPL on rk3399, so I think I can > also check for that, but that's technically in a different driver and I > don't know how best to do that. > > drivers/clk/rockchip/clk_rk3399.c | 7 +------ > 1 file changed, 1 insertion(+), 6 deletions(-)
Could we make this conditional on SPL not running? We could check that by enabling CONFIG_HANDOFF, for example. Regards, Simon
