02:22 <shadows> zen_desu: hi

02:22 <zen_desu> hi

02:23 <shadows> zen_desu: you may refer to the long explanation in cover leter for my series sent to LKML https://lore.kernel.org/lkml/20250502103101.957016-1-e@freeshell.de/

02:25 <shadows> short version is don't do that; boot SPI NOR flash mode of JH7110 into U-Boot and use the EFI support there to seek an EFI System Partition on your eMMC.... if possible.

02:27 <shadows> long version is that the method used by JH7110 eMMC boot mode "apparently" (I did not research any source or reverse-engineering effort for the zeroth-boot-loader in MaskROM) has some scheme which is incompatible to a GUID Partition Table data structure

02:28 <shadows> I spent about two solid days faffing around with it and decided that was as far as I could get without reverse engineering the MaskROM content

02:31 <zen_desu> yeah at the time being im getting my head wrapped around the sd booting stuff

02:31 <zen_desu> just because i don't want to trash the onboard spi

02:31 <shadows> you have serial UART cable? don't worry about bricking the vf2. It's robust, you can always recover it.

02:31 <zen_desu> i kinda don't see the point in using GPT for SPLs tho, i guess it works, i'd prefer just using some offset on the storage device

02:31 <zen_desu> ok cool, thanks for that

02:32 <zen_desu> https://wiki.gentoo.org/index.php?title=StarFive_VisionFive_2&action=history

02:32 <zen_desu> i've been updating this

02:32 <zen_desu> i'd appreciate a second look :D but i have not double checked this info on the reflash process

02:32 <zen_desu> https://wiki.gentoo.org/wiki/StarFive_VisionFive_2#SPI_issues_.2F_First_aid_kit i mvoed it here, and it seems simple enough, i;ve used xmodem before

02:33 <shadows> that's great! :) more documentation equals "more better" :D

02:33 <zen_desu> yeah i agree, i may make a uboot wiki account to change that bit on FW_TEXT_START=0x40000000 FW_OPTIONS=0

02:34 <zen_desu> it may be that you needed that on older versions? idk, setting that made it just not work for me

02:34 <shadows> ohhh... docs on wiki are generated from the source tree in ".rst" restructured-text format

02:35 <zen_desu> ah so it would just be done through a PR to the main source repo?

02:35 <shadows> I think Heinrich (xypron) was going to streamline the JH7110 docs a bit, or already has done so... it needs a little bit more tidying up to kind of unify everything and less duplication of common procedures

02:35 <shadows> yep all done via git

02:36 <zen_desu> i may look into it, you maybe pinged them so hopefully they can fix that bit or maybe explain it to me

02:37 <shadows> plus patches to the U-Boot Mailing list. There's some github"-ish" usage by the custodians for certain areas of code but at the end of it they're pulling from mail list patches and then submitting on mail list a request to U-Boot project leader to pull from those custodian branches

02:37 <zen_desu> I've gone this long without submitting patches via email but maybe eventually that will change lol

02:38 <shadows> oh... yeah, I set up one of the free linux public access shell accounts I'd had for some time just for this kind of developer mail list stuff. It was a couple of weeks for an old guy like me to wrap my mind around that.

02:39 <shadows> zen_desu: have a look at this table / identification guide I put together, as well https://freeshell.de/e/riscv64/vf2eeprom/

02:42 <zen_desu> ohhh that's nice

02:42 <zen_desu> i may add that to the see also bit

02:42 <zen_desu> what doe sa checkmark really mean?

02:44 <shadows> for example there was a hack at some point to extend Milk-V Mars devicetree at runtime and fix it up in-place in memory after loading so that the Mars CM and Mars CM Lite would function; but that all went away later when upstream Linux devicetree became de facto (and there's not yet any Mars CM or Mars CM Lite devicetree in upstream Linux)

02:46 <shadows> [nothing,checkmark,cross] to mean not present, present and functional, formerly present but now absent or not functional

02:52 <shadows> zen_desu: browse over https://github.com/geerlingguy/sbc-reviews/issues/22#issuecomment-1908676618 and pick out the parts you find interesting, it is some github issue comment tracking the Milk-V Mars CM for a YouTube content creator's list of SBC's and though now it's a bit stale I was updating the comment frequently

02:56 <shadows> for chat on topic of JH7110 that is not strictly about U-Boot also welcome to participate in the Pine64 Star64 room (which I access via Matrix; I think it's also there via other platforms) https://pine64.org/community/#chat-platforms

03:03 <zen_desu> thanks for that, i think i will try that in a bit

14:46 <Tartarus> Community call at the top of the hour

15:02 <Tartarus> ... now

15:22 <Tartarus> Hmmmm, I wonder if https://repost.aws/knowledge-center/free-tier-windows-instance would be enough to allow us to have a gitlab runner on Windows and in turn also do the Windows host tools build there and not only in Azure

15:23 <Tartarus> (The stm32 PR I just got breaks only on Windows hosts, which is harder to test due to being only in Azure)