On Mon, Oct 06, 2025 at 01:14:06PM -0600, Tom Rini wrote: > Hey all, > > So it's release day and so here is v2025.10. I will be merging the next > branch back to master shortly and pushing that out as well. > > As this is a full release, and not just a release candidate I'm once > again hoping for a few more people to read this than do my -rc emails or > just regular mailing list posts. As I mentioned with the v2025.07 > release, I was looking for a few people to step up and help with the > overall organization and management of the project. To that end, Peter > Robinson and Neil Armstrong have stepped up and have been helping me. > This has been part of the process for the project to join up under the > Software Freedom Conservancy's (SFC) umbrella and have a legal entity > that can help the project work with other legal entities on things like > donations. To that end, we'll have representatives from SFC on the next > call to answer any questions or concerns people might have. > > With that said, we'll have our regularly > scheduled community meeting tomorrow. The meeting details itself are: > https://meet.google.com/btj-wgcg-euw > October 7th, 2025. 9am (GMT -06:00) > > To join by phone: > https://meet.google.com/tel/btj-wgcg-euw?pin=1307528552322&hs=1
The meeting transcription from Gemini is at: https://docs.google.com/document/d/1ww2zFC1d9DG-Wybmo_Nz5u9IuvTO2UwJCAx2C-dZy9w/edit?tab=t.oyswh1neakco and I've only edited to show that MV is Marek, in later contexts. The summary, aside from a few errors (AP Boot Manager should be EFI Boot Manager) seems correct. The link is viewable in a private browser window, but here is the summary tab (the full transcript is also included) as markdown. Oct 7, 2025 ## U-Boot community meeting ### Summary Tom Rini announced that the U-Boot project is joining the Software Freedom Conservancy, with OSU OSL offering to host a GitLab instance, and the migration is aimed to be completed by the end of the year. Discussions focused on standardizing boot flows for embedded systems, specifically integrating TFA into a FIT image and proposing it as an extension of the FIT image boot flow rather than a strict UEFI standard. Key topics also included moving the AP boot manager to the end of the boot chain to improve flexibility, integrating mouse pointer support, and updates on Kbuild and MMU features. ### Details * **U-Boot Project Joining Software Freedom Conservancy** Tom Rini announced that the U-Boot project is set to join the Software Freedom Conservancy, with paperwork expected to be signed shortly. This move will facilitate resolving challenges such as migrating from Denk's infrastructure, with Oregon State Open Source Labs (OSU OSL) offering to host the project, including a GitLab instance ([00:01:28](?tab=t.ef7rmkft6mg6#heading=h.69uuqh2ce80q)). A representative from the Software Freedom Conservancy expressed their excitement and willingness to assist with any questions ([00:04:12](?tab=t.ef7rmkft6mg6#heading=h.hkg52to2w760)). * **Migration Timeline and GitLab Instance** Heinrich Schuchardt inquired about the migration schedule, with Tom Rini clarifying that while there is no immediate rush, it is desirable to complete the migration and data backup as soon as reliably possible, ideally by the end of the year. Tom Rini confirmed that OSU OSL is capable of running a GitLab instance, addressing concerns about the specialized skill set required for such services ([00:02:52](?tab=t.ef7rmkft6mg6#heading=h.qpi1k2wejoy0)). * **Boot Flow Standardization and TFA Integration** Heinrich Schuchardt initiated a discussion on supported boot flows, noting Marek's (MV) interest in booting via UEFI using a FIT image that includes TFA (Trusted Firmware-A) ([00:05:49](?tab=t.ef7rmkft6mg6#heading=h.s6w9d233u8po)). MV further elaborated on current issues with bundling U-Boot SPL, DRAM blobs, TF-A BL31, and U-Boot into a single flash.bin, citing licensing concerns and the difficulty of updating TF-A BL31 separately ([00:07:43](?tab=t.ef7rmkft6mg6#heading=h.el7xywl0griw)). They proposed a standardized boot flow where the bootloader loads the FIT image, including the kernel and TF-A BL31, to simplify updates and ensure ABI compatibility ([00:09:30](?tab=t.ef7rmkft6mg6#heading=h.o081i45tk1uh)). * **Debate on UEFI and Firmware Bundling** Andre Przywara expressed concerns about bundling the kernel with secure firmware like TF-A BL31, stating that secure firmware should ideally be part of the device's flash and in a separate security domain. MV clarified that starting TF-A BL31 from a FIT image via U-Boot running in EL3 is already functional upstream and their goal is to standardize this flow with UEFI ([00:11:03](?tab=t.ef7rmkft6mg6#heading=h.xx2r3tv04fud)) ([00:23:07](?tab=t.ef7rmkft6mg6#heading=h.tk7jgdwuir6l)). Tom Rini noted that while open-embedded distributions might support such bundling, other distributions like Ubuntu or Red Hat would likely not ([00:12:13](?tab=t.ef7rmkft6mg6#heading=h.671m41uhflpb)). * **The Role of UEFI in Embedded Systems** MV articulated their interest in UEFI for embedded systems, believing that larger embedded systems will increasingly adopt standardized boot flows, which they see UEFI providing. Andre Przywara questioned the necessity of UEFI for MV's embedded use case, suggesting that the described flow is a pragmatic workaround rather than a standard UEFI approach ([00:14:01](?tab=t.ef7rmkft6mg6#heading=h.ua0wnfhtpog)). Tom Rini acknowledged that ARM's perspective is to move towards UEFI for all platforms and that MV's problem is a recurring issue that needs a solution ([00:15:20](?tab=t.ef7rmkft6mg6#heading=h.gp5usmy34p5y)). * **Challenges with Standardizing Boot Flows** Simon Glass questioned how Grub would handle a FIT image if UEFI were used, suggesting that Grub might not inherently support FIT images. Heinrich Schuchardt clarified that Grub can be placed within a FIT image and started this way ([00:20:19](?tab=t.ef7rmkft6mg6#heading=h.et19coo8jdd)). Ilias Apalodimas corrected the misconception that U-Boot functions are inaccessible after running Grub under UEFI, explaining that they remain accessible until \`exit boot services\` is called by the OS ([00:19:23](?tab=t.ef7rmkft6mg6#heading=h.kyn38w6xqllt)). Ilias Apalodimas raised security concerns about loading BL31 from U-Boot, as BL31 is typically authenticated by silicon keys and should run in EL3, while U-Boot usually runs in EL2 ([00:21:20](?tab=t.ef7rmkft6mg6#heading=h.u58myncx66h4)). * **Proposed Solution for Boot Flow Standardization** Tom Rini suggested documenting MV's proposed boot flow on multiple non-IMX platforms, such as Renaissance boards, to establish an "ad hoc" standard for 64-bit ARM platforms dealing with potential kernel breakages. Andre Przywara proposed treating this as an extension of the FIT image boot flow, making it a U-Boot standard rather than trying to fit it into the strict UEFI standard. MV concurred with this approach, agreeing to not focus on UEFI for this specific boot flow ([00:24:21](?tab=t.ef7rmkft6mg6#heading=h.1oi3q6x0ziml)) ([00:26:49](?tab=t.ef7rmkft6mg6#heading=h.1gmyx6dnamg)). * **Moving AP Boot Manager to the End of the Boot Chain** Heinrich Schuchardt discussed Simon Glass's 17 patches aimed at moving the AP boot manager to the end of the boot chain, just before network booting ([00:31:28](?tab=t.ef7rmkft6mg6#heading=h.g4y66tc6cx8p)). Tom Rini expressed optimism about resolving outstanding points and stated that the changes are moving in the right direction. Simon Glass explained that this change would make global boot methods more flexible, albeit more complex, by introducing an order to their execution ([00:33:02](?tab=t.ef7rmkft6mg6#heading=h.jbkt0hflfk5s)). Andre Przywara supported this change, noting that the global boot method was previously unpredictable and an established order would solve existing problems ([00:34:29](?tab=t.ef7rmkft6mg6#heading=h.ufyp2uxkjl3f)). * **EFI Boot Manager Improvements and SunXi Conversions** Heinrich Schuchardt brought up the possibility of integrating mouse pointer support from EDK2 into U-Boot for graphical boot managers, a feature Simon Glass had been working on ([00:34:29](?tab=t.ef7rmkft6mg6#heading=h.ufyp2uxkjl3f)). Tom Rini requested prioritizing the resolution of the global boot methods issue before tackling additional large series of patches ([00:35:39](?tab=t.ef7rmkft6mg6#heading=h.3id2x16rmft0)). Andre Przywara highlighted that the SunXi conversion to standard boot is currently blocked because the global boot method takes precedence over existing boot methods, causing breakages for users relying on traditional boot scripts ([00:38:52](?tab=t.ef7rmkft6mg6#heading=h.x428xffwzdy8)). Heinrich Schuchardt encouraged Andre Przywara to test Simon Glass's latest series to address this issue, which Simon Glass intended to resend ([00:41:15](?tab=t.ef7rmkft6mg6#heading=h.8xnc1coo6x1d)). * **Kbuild and MMU Updates** Tom Rini reminded Ilias Apalodimas about outstanding Kbuild updates, which Ilias Apalodimas confirmed Sugus was handling and would post soon ([00:43:47](?tab=t.ef7rmkft6mg6#heading=h.u8luwdvjpl9n)). Ilias Apalodimas noted that the most challenging part of Kbuild, shared library linking, had been completed without issues ([00:44:44](?tab=t.ef7rmkft6mg6#heading=h.735mexbsad5k)). Heinrich Schuchardt inquired about the status of setting the U-Boot binary to non-writable in the MMU, to which Ilias Apalodimas responded that memory permissions have landed in master as an optional feature, but configuring EFI runtime services as read-execute is still a complex ongoing task ([00:45:47](?tab=t.ef7rmkft6mg6#heading=h.mn59kqyfilfq)). * **Conclusion of Meeting** Tom Rini concluded the meeting by thanking everyone for their time and scheduled the next meeting in three weeks, following the normal cadence. Tom Rini encouraged anyone with concerns about the U-Boot project joining the Software Freedom Conservancy to reach out ([00:48:03](?tab=t.ef7rmkft6mg6#heading=h.lrh9q6iwa0a9)). -- Tom
signature.asc
Description: PGP signature
