Hi TI folks, I have put some thoughts in the github issue[1] that we are discussing FIT image spec, could you please take a look when you have time?
Best regards, Aristo [1] https://github.com/open-source-firmware/flat-image-tree/issues/32 Aristo Chen <[email protected]> 於 2026年2月10日週二 下午4:16寫道： > > Hi E Shattow and all, > > E Shattow <[email protected]> 於 2026年1月14日週三 上午1:29寫道： > > > > Hi Aristo, > > > > On 1/13/26 07:09, Aristo Chen wrote: > > > Hi E Shattow, > > > > > > Aristo Chen <[email protected]> 於 2025年11月5日週三 下午8:21寫道： > > >> > > >> Hi E Shattow, > > >> > > >> E Shattow <[email protected]> 於 2025年9月24日 週三 下午9:31寫道： > > >>> > > >>> Hi Aristo, > > >>> > > >>> On 9/24/25 04:43, Nishanth Menon wrote: > > >>>> On 06:37-20250924, Nishanth Menon wrote: > > >>>>> On 10:59-20250914, Aristo Chen wrote: > > >>>>>> This patch series enhances FIT image robustness by adding **memory > > >>>>>> region overlap detection** to `mkimage` and fixing existing overlaps > > >>>>>> in DTS files and `binman` tests. > > >>>> > > >>>> [...] > > >>>>>> > > >>>>> > > >>>>> Looks like i see a build regression in linux-next after this series. > > >>>> > > >>>> I fat fingered that one.. sorry, I meant u-boot next. > > >>>> > > >>>> > > >>>> Fails at commit 4d84fa1261eb, last pass was on commit d81c1118580f > > >>>> > > >>>>> > > >>>>> https://gist.github.com/nmenon/b2fc9e7680cc296062c7dced94105f76 > > >>>> > > >>>> I believe there are outstanding comments on V1 that have'nt been > > >>>> addressed either. Can we revert/drop this series for now while the > > >>>> comments are addressed? > > >>>> > > >>> > > >>> Similar to Nishanth, I am seeing a build regression, itb.map: > > >>> > > >>> ImagePos Offset Size Name > > >>> <none> 00000000 <none> itb > > >>> <none> 00000000 <none> fit > > >>> <none> 00000000 000c1698 uboot > > >>> <none> 00000000 000c1698 u-boot-nodtb > > >>> <none> 000c1698 00202128 opensbi > > >>> <none> 00000000 00202128 opensbi > > >>> > > >>> Yields this error: > > >>> ... > > >>> MKIMAGE fit-dtb.blob > > >>> Warning: not able to get `load` of node 'fdt-1' > > >>> Warning: not able to get `load` of node 'fdt-2' > > >>> Warning: not able to get `load` of node 'fdt-3' > > >>> Warning: not able to get `load` of node 'fdt-4' > > >>> Warning: not able to get `load` of node 'fdt-5' > > >>> Warning: not able to get `load` of node 'fdt-6' > > >>> Warning: not able to get `load` of node 'fdt-7' > > >>> CAT u-boot-fit-dtb.bin > > >>> ... > > >>> MKIMAGE u-boot.img > > >>> Warning: not able to get `load` of node 'fdt-1' > > >>> Warning: not able to get `load` of node 'fdt-2' > > >>> Warning: not able to get `load` of node 'fdt-3' > > >>> Warning: not able to get `load` of node 'fdt-4' > > >>> Warning: not able to get `load` of node 'fdt-5' > > >>> Warning: not able to get `load` of node 'fdt-6' > > >>> Warning: not able to get `load` of node 'fdt-7' > > >>> COPY u-boot.dtb > > >>> MKIMAGE u-boot-dtb.img > > >>> Warning: not able to get `load` of node 'fdt-1' > > >>> Warning: not able to get `load` of node 'fdt-2' > > >>> Warning: not able to get `load` of node 'fdt-3' > > >>> Warning: not able to get `load` of node 'fdt-4' > > >>> Warning: not able to get `load` of node 'fdt-5' > > >>> Warning: not able to get `load` of node 'fdt-6' > > >>> Warning: not able to get `load` of node 'fdt-7' > > >>> BINMAN .binman_stamp > > >>> Wrote map file './itb.map' to show errors > > >>> binman: Error 1 running 'mkimage -t -F ./itb.fit.fit': Warning: not able > > >>> to get `load` of node 'fdt-1' > > >>> [Config: conf-1] Error: Overlap detected: > > >>> - uboot: [0x40200000 - 0x402c1698] > > >>> - opensbi: [0x40000000 - 0x40202128] > > >>> > > >>> make[1]: *** [/tmp/u-boot.2.git/Makefile:1339: .binman_stamp] Error 1 > > >>> make[1]: Leaving directory '/home/es/build/u-boot' > > >>> make: *** [Makefile:198: sub-make] Error 2 > > >>> make: Leaving directory '/tmp/u-boot.2.git' > > >>> > > >>> If you would like to reproduce the issue, use config > > >>> starfive_visionfive2_defconfig and pass to OPENSBI env variable the path > > >>> of opensbi object 'fw_dynamic.bin' built using PLATFORM=generic, with > > >>> the ordinary gcc riscv64 toolchain. The starfive visionfive2 board > > >>> target in U-Boot supports multiple vendors and boards selected at > > >>> runtime and is an example of a more complex FIT usage that you may be > > >>> interested in for comparison and testing. > > >>> > > >>> Best regards, > > >>> > > >>> -E > > >> > > >> Sorry that I didn't noticed that my previous email sent to you was not > > >> including other people > > >> > > >> Long story short, I am now able to reproduce the overlap issue that > > >> you have encountered, and > > >> I spent some time trying to understand what is opensbi, please correct > > >> me if I am wrong > > >> > > >> AFAICT, there will be 2 types of firmware built, one is > > >> `fw_payload.bin`, and the other is > > >> `fw_dynamic.bin` > > >> - fw_payload.bin: contains the next-stage binary, so the file size > > >> will be bigger > > >> - fw_dynamic.bin: provides information about the next boot stage at > > >> runtime, so the size is smaller > > >> > > >> and here are the size of these 2 files built in my environment > > >> ``` > > >> -rw-rw-r-- 1 ubuntu ubuntu 273048 Oct 25 14:49 fw_dynamic.bin > > >> -rw-rw-r-- 1 ubuntu ubuntu 2105656 Oct 25 14:49 fw_payload.bin > > >> ``` > > >> > > >> When using my overlap checking mechanism, and passing the > > >> fw_dynamic.bin to build U-Boot, > > >> there is no overlap issue. However, when using fw_payload.bin, there > > >> will be an overlap issue. > > >> > > >> AFAICT, the SPL will load the opensbi firmware(assuming fw_payload.bin > > >> here) first, then load > > >> U-Boot into memory, and then jump to opensbi firmware, so I was a bit > > >> confused why the > > >> overlapped opensbi still works? > > >> > > >> I will find some time to figure out how the whole loading/booting > > >> process looks like, but it would > > >> be greatly appreciated if you can share some insight, thanks! > > > > > > I finally had some time to revisit this and managed to get a VisionFive2 > > > board for local testing. > > > > > > While trying to make it bootable I noticed the following docs: > > > - > > > https://docs.u-boot.org/en/latest/board/starfive/visionfive2.html#second-stage-bootloader-opensbi-fw-dynamic-bin-u-boot-main > > > - https://wiki.gentoo.org/wiki/StarFive_VisionFive_2#Building_OpenSBI > > > > > > Both appear to recommend using OpenSBI’s fw_dynamic.bin when building > > > U‑Boot. I couldn’t find any guidance on using fw_payload.bin together > > > with U‑Boot, which made me wonder why fw_payload.bin was used in your > > > earlier report. > > > > > > > Likely I mixed up the filenames when writing in conversation, as I did > > when writing on the opensbi mailing list about a dependency tracking bug > > in opensbi build system: > > > > https://lore.kernel.org/opensbi/[email protected]/ > > > > and > > > > https://lore.kernel.org/opensbi/[email protected]/ > > > > ...which is clearly a mistake to write "OPENSBI=/path/to/fw_payload.bin" > > when it should instead be "OPENSBI=/path/to/fw_dynamic.bin", > > acknowledged in the follow-up message: > > > > https://lore.kernel.org/opensbi/[email protected]/ > > > > > Could you clarify whether there is documentation (or a specific > > > requirement) that calls for fw_payload.bin in this flow? If the intended > > > path is fw_dynamic.bin, then the overlap issue disappears for me > > > > > > > For the VisionFive2 and derivative boards, I intended to refer to > > fw_dynamic.bin only. If I mention otherwise it was in error or for not > > knowing what I'm looking at and reporting more information than > > required. There are RISC-V boards not related to VisionFive2 or its > > StarFive JH-7110 SoC that do use the fw_payload.bin approach but I think > > I just didn't notice the difference when transcribing to e-mail. > > > > > Thanks for your help and guidance. > > >> > > >> Best regards, > > >> Aristo > > > > > > Best regards, > > > Aristo > > > > Thanks very much for the follow-up, > Thanks for letting me know that StarFive VisionFive2 is not affected > by my patch! > > > > > -E > > We’re currently discussing a new proposal[1] for handling TI platforms > in the FIT image spec. Ahmad Fatoum has a few questions about > TI-specific implementations—could the TI folks on this thread please > take a look when you have a moment? > > Best regards, > Aristo > > [1] https://github.com/open-source-firmware/flat-image-tree/issues/32
