Am Freitag, 24. Januar 2025, 11:21:00 CET schrieb Niklas Cassel: > On Thu, Jan 23, 2025 at 03:13:01PM +0100, Heiko Stübner wrote: > > Am Mittwoch, 22. Januar 2025, 17:12:26 CET schrieb Niklas Cassel: > > > On Wed, Jan 22, 2025 at 04:38:16PM +0100, Quentin Schulz wrote: > > > > So essentially, if SPL_ATF_NO_PLATFORM_PARAM is selected (the default for > > > > RK356x, RK3588, forced on on RK3308, enabled for the majority of RK3399 > > > > boards, enabled for all RK3328 boards) the DT won't be passed to TF-A so no > > > > issue in terms of size on that side. > > > > If it is not selected, for TF-A < 2.4 (released 20201117, 4 years ago), a > > > > DTB bigger than 64KiB will crash TF-A. > > > > If it is not selected, for TF-A >= 2.4, a DTB bigger than 128KiB will result > > > > in TF-A not being able to read the DTB (for Rockchip, that means not being > > > > able to derive the UART settings (controller and baudrate) to use, and will > > > > use the compile-time default instead). > > > > > > Not everyone is using binary blobs from Rockchip. > > > On my rock5b (rk3588), I'm building the bootloader using buildroot, > > > which is using upstream TrustedFirmware-A (v2.12). > > > > > > > > > > In short, I don't know where to go with that additional piece of > > > > information, but this is a bit bigger than simply moving things around and > > > > adding compile-time tests for overlay application. > > > > > > This is significant information indeed. > > > > I guess the question is, can this hurt existing devices? > > > > As Quentin mentioned, this only affects DTs that get handed over from > > U-Boot to TF-A (and maybe OP-TEE). > > > > So the whole range of things loading their DT from extlinux.conf or > > whatever are not really affected. > > > > > > DTs U-Boot can hand over are 2 types, > > (1) built from within u-boot and > > (2) stored somewhere centrally (SPI flash). > > > > > > Case (1) is again not affected, as U-Boot (and other bootloaders) may > > very well sync the DTS files, but generally not the build-system, so if > > U-Boot (or any other bootloader) creates DTBs with symbols is completely > > their own choice. > > > > > > And for case (2) I see the manufacturer being responsible. Having the DT > > in central storage makes it somewhat part of a "bios"-level in the hirarchy > > and the general guarantee is that new software _will work_ with older DTs, > > but the other way around is more a nice to have (old SW with new DTB). > > > > So if some manufacturer has a centrally located DTB this does not matter > > until they upgrade, and when that happens I do expect testing to happen > > at the manufacturers side, before rolling out a "bios update" > > Personally, I'm all for letting the kernel build the DTBs with symbols. > > (I have a patch that I keep rebasing on my tree only for that purpose. > Sure, I could modify my build scripts to build the DTB separately, > but with this patch, I do not need to do anything since the kernel > builds the DTBs already.) > > Other platforms, e.g. TI already build many boards with symbols: > https://github.com/torvalds/linux/blob/v6.13/arch/arm64/boot/dts/ti/Makefile#L242-L261 > > > You seems to have been against enabling symbols before: > https://lore.kernel.org/linux-rockchip/171941553475.921128.9467465539299233735.b4-ty@xxxxxxxxx/ > https://lore.kernel.org/linux-rockchip/1952472.6tgchFWduM@diego/ > > But if you have changed you mind, and you are no longer concerned about > doing so, then in my own self-interest I'm all for it :) I'm all for keeping compatibility as good as possible and that issue came on the table way too often already ;-) . In the past it was essentially easy to go with "just don't enable symbols" and not go down the nitty-gritty detail route - because that whole mesh of different firmware combinations gives me a headache ;-) [0] So finally going through those possible affected variants gave me those thoughts of "is there even an actual problem with existing boards?". Especially wrt forward<->backwards compatibility. Outcome is, I'm definitly not sure about myself, but also could not come up with an actual scenario. But that compile-time testing of applying DTBOs is way too great to pass up on :-) Heiko [0] If just some vendor would directly work on upstream TF-A from the beginning, instead of hacking up some half-decade old ATF ... ;-)
