On Mon, Apr 28, 2025 at 03:40:32PM -0600, Tom Rini wrote: > Hey all, > > So it's release day and I have tagged and pushed things out. Looking at > my own TODO list, I think it's in reasonable shape. I do think there's a > few other pull requests that need to happen still and I am optimistic > will come shortly (but I've not reached out to anyone in particular). If > there's anything big that people see missing, please speak up. > > We're continuing with a community meeting following the release and the > calendar link is > https://calendar.google.com/calendar/event?action=TEMPLATE&tmeid=N280N2tlcXE3aDVtbjlicnNkcm82YnA1bDAgODliZTdiODEzMTM2YmVjMDZhODNkZTRkYTU5NzQ1ZjBhYmQxMWMxYjgzNjA2MWFlMDZjMWM3ZGJjZDE4ZGY0MUBn&tmsrc=89be7b813136bec06a83de4da59745f0abd11c1b836061ae06c1c7dbcd18df41%40group.calendar.google.com > and once again this is the same time as the previous meeting. The > meeting details itself are: > https://meet.google.com/btj-wgcg-euw > April 29th, 2025. 9am (GMT -06:00) > > To join by phone: > https://meet.google.com/tel/btj-wgcg-euw?pin=1307528552322&hs=1
And here are the notes I took on the call. - I noted that rc1 is out, I think my queue looks good and while most of the PRs I expected have come in, a few haven't yet. And I may need to look at the SPI/NAND/NOR queue myself. - Heinrich asked about updating QEMU in CI and the two platform failures that updating has shown. I've not had time to look at the vexpress_ca9x4 failure to see if this is a regression in QEMU or exposing a U-Boot bug. Heinrich has not had time to look at the RISC-V failure of sifive_unleashed. He will try and boot -rc1 on an Icicle platform which is the same cores at least as the unleashed. As part of this we had a short general discussion of some of the RISC-V platforms and cores they utilize. - Heinrich noted E Shattow's issue that was brought up on IRC, of having RISC-V platforms that read the EEPROM in SPL to determine what the platform is and then load the proper device tree, but that they need to do this again in full U-Boot in order to set fdtfile in the environment. Is there a better way of doing this to avoid the double-read? Heinrich thinks there might be some mechanism to copy some properties today, but not everything. Simon noted bloblist could be an option. Further discussion about fdtfile was had. I suggested looking at board/ti/common/ as they also have this problem and have been doing something at least for a number of years. Simon noted SPL_HANDOFFF may be another option here. - Heinrich noted there's many bootm stages and states are not well documented. Wondered who has a good understanding these days of the states and stages and who could describe it. Simon knows much of it, but it's not comprehensive. Simon has been trying to make it so you can work without CMDLINE, which means having function calls to go through the states. Simon can start by describing the constants better at least. - Heinrich brought up his RFC for efi boot manager + bootstd he posted. Wanted to know if Simon had looked in to the rest of the patches. Simon is OK enough with the approach. Simon thinks that boot manager should be more iterative, but a discussion for later. Heinrich noted that for background at least, the spec requires certain behaviors to be done. - Andre noted that Heinrich said Ubuntu uses fdtfile, does that mean it always ignores the U-Boot tree? sunxi takes the running tree and modifies it in tf-a/u-boot and that's the right one for the OS. This lead to a fairly long discussion of how Ubuntu does and doesn't work today, but at the end of the day U-Boot can and does support the various contrasting approaches used here. - Andre asked who pays for CI (Gitlba), I noted it's all donated and use as much as you like as the jobs are split such that aside from bottlenecks on the high end machines no one else should get overly blocked. Simon asked if Andre can setup a lab of his HW visible to Gitlab. Andre will see what if anything can be done, but the sunxi stuff is his personal project. - Heinrich asked of sunxi has newer hardware for example and Andre gave a short general update. - Simon wanted to talk about standard passage. Heinrich asked for a brief overview. I asked that we not go over it without other people that use the code on the call. I noted the vexpress_fvp option and will enable it in CI later today. - Simon asked about the Python patches he posted earlier in the day and I reiterated that all patches for mainline need to be against mainline and not some other tree. -- Tom
signature.asc
Description: PGP signature
