On Mon, 2019-10-07 at 13:35 +1100, Jonathan Gray wrote: > On Sun, Oct 06, 2019 at 09:51:20PM -0400, Kurt Miller wrote: > > > > On Sun, 2019-10-06 at 18:47 +0200, Mark Kettenis wrote: > > > > > > > > > > > > > > > Date: Sun, 6 Oct 2019 11:42:03 +1100 > > > > From: Jonathan Gray <j...@jsg.id.au> > > > > > > > > On Sat, Oct 05, 2019 at 12:20:57PM -0400, k...@intricatesoftware.com > > > > wrote: > > > > > > > > > > > > > > > Various rockchip u-boot 2019.10rc4 aarch64 improvements: > > > > > * u-boot.itb is included in the all target for rockpro64 and > > > > > firefly-rk3399 so remove duplicate make for it. > > > > > * Build rock64-rk3328, except exclude idbloader.img which is broken. > > > > > * Install u-boot-spl-dtb.bin and mkimage for rock64, rockpro64 and > > > > > firefly-rk3399 so that idbloader.img can be created using rkbin TPL > > > > > layer. > > > > > * Provide a script to make idbloader.img from rkbin's TPL layer. > > > > > * For rockpro64 backport PCIe support from 5.3 dtb. > > > > > * For rock64 backport most 4.20 dtb components and additional bug > > > > > fixes. > > > > > > > > > > For rockpro64, I no longer need to have an updated dtb installed > > > > > in the i partition for PCIe to work. Although I am using the > > > > > rkbin TPL layer because the u-boot one doesn't recognize half > > > > > of my memory (other versions of this board dont have this > > > > > problem). > > > > > > > > > > For rock64 the dtb updates corrects the clock bindings so an > > > > > updated dtb is not needed in the i partition as well. However, > > > > > u-boot's TPL layer isn't working yet so the following steps > > > > > are needed to create a usable idbloader.img on a uSD for > > > > > install with miniroot66.fs: > > > > > > > > > > /usr/local/libexec/u-boot/mkidbloader-rkbin rock64-rk3328 > > > > > dd if=miniroot66.fs of=/dev/<uSD> bs=1m > > > > > dd if=idbloader.img of=/dev/<uSD> bs=512 seek=64 conv=sync > > > > > dd if=/usr/local/share/u-boot/rock64-rk3328/u-boot.itb \ > > > > > of=/dev/<uSD> bs=512 seek=16384 conv=sync > > > > > > > > > > One of the downsides of installing mkimage is the removal of > > > > > PKG_ARCH=*, but we get better support for both rock64 and > > > > > rockpro64 for where rkbin is still needed. I'm not sure how > > > > > well firefly works with the the u-boot TPL layer so it may > > > > > benefit from the ability to build idbloader.img with the > > > > > rkbin TPL layer as well. > > > > > > > > > > Thoughts? > > > > Dropping PKG_ARCH=* isn't something we should do in my opinion. For > > > > example arm package snapshots seem to be made every few months and with > > > > U-Boot 2019.10 final planned for Monday US time there wouldn't be a > > > > package available for arm in time for the final release build. > > > > > > > > I think for now you should drop the rk3328/mkimage parts and keep the > > > > rockpro64 dtb change. > > > Simon South just posted two patches to the U-Boot mailing list that > > > apparently fixes ro rock64 issues. That removes the need to use the > > > propriatary Rockchip loader in favour of U-Boot TPL. > > I've adjusted the u-boot changes to address dropping PKG_ARCH=* > > and incorporated Simon South's changes to see if the Rock64 > > will work with u-boot's TPL layer. Please see the attached diff > > for review. > This looks good but it isn't clear to me what the 3328 dts files > (rk3328.dtsi rk3328-rock64.dts) are synced against? > > There is quite a difference in rk3328.dtsi rk3328-rock64.dts and > linux 5.3. The other files have no or expected diffs to 5.3. >
For rk3328.dtsi I synced to 4.20, but the following headers were not yet in u-boot so I left out components related to them: #include <dt-bindings/power/rk3328-power.h> #include <dt-bindings/soc/rockchip,boot-mode.h> #include <dt-bindings/thermal/thermal.h> I needed to keep the mmc[0-2] aliases because rk3328-evb.dts referred to them. For rk3328-rock64.dts I synced to 4.20 but I noticed that the usb3 entry was later removed as well as some future bug fixes: https://github.com/torvalds/linux/commit/26e2d7b03ea7ff254bf78305aa44dda62e70b78e https://github.com/torvalds/linux/commit/9301d28dccd05bbcea0e69c382886d6a8b8e6bc2 I removed vcc-host1-5v-regulator when doing that since it wasn't referenced and not configured. Please let me know if you would like a different approach. Thanks, -Kurt
