@linear Yeah HDMI getting mainline would be really nice but typically for computers I'm fine with just a serial port and an ethernet port (in fact hate the lack of serial in modern x86-pc computers).
In a way perfection would be more PCIe slots so you could just slap a normal GPU.
that seems to have fallen by the wayside, and after several revisions the patch now seems to have been abandoned for a few months and does not cleanly build
everything else i care about seems to work fine, though
the patches don't work in 6.12 because they depend on functionality not yet introduced, but that is present in 6.13
and they don't work in 6.13 because they rely on an api that was present in 6.12 but removed in 6.13 :P
i managed to modify them to get them building but then they proceed to not actually work
i'm half tempted to keep poking at this now that i found my jtag adapter and can debug if things break, and send an updated patchset on behalf of the original author. and in the meantime i'm content to just use this over ssh / serial
@lanodan@queer.hacktivis.me yeah the single pcie 2.0 lane is really limiting. this is why i am interested in a board based on the SpacemiT K1, like the framework-compatible board from DeepComputing or the one BananaPi has
@RedTechEngineer@linear It's horribly bad in terms of actual information, sometimes it even feels like there's somehow a telephone game going on because Michael Larabel will just hallucinate bits of things.
Case in point the article you linked mentions the SpacemiT K1, which is just an SoC (so no actual boards…) and absolutely nothing about what actually is getting merged, like nothing if it's initial patches for some subsystems, or the final patches for full support of the SoC.
Compare with articles from lwn.net or like https://rvspace.org/en/project/JH7110_Upstream_Plan to get an actual idea of the number of bits needed to actually support a board (VisionFive2 with the JH7110 SoC).