<Esmil>
yeah, and why are you not adding sdio1 while you're at it
<Esmil>
..at least add the node to the th1520.dtsi
clemens3 has quit [Quit: WeeChat 2.7]
<Esmil>
know the reference names are a bit bikeshedding, but the other names seem to line up with the documentation, so it'd be a shame to ruin a good system ;)
Andre_Z has joined #riscv
<drewfustini>
Esmil: my original thought was mmc was generic name for the controller ip, but what you wrote make sense too. I avoided the 3rd controller instance which is used for SDIO as I didn't want to get into testing SDIO and the WiFi out until the eMMC and MicroSD had landed. But I can see your point about adding it to the dtsi and then enabling it in dts in the future.
shamoe has quit [Quit: Connection closed for inactivity]
<Esmil>
yeah, since emmc and sdio0 seem to work with the same driver, i'd be very surprised if sdio1 will need a different compatible/number of clocks
<Esmil>
drewfustini: ^
<sevan>
maybe Ubuntu could ship zfs kmod's for the risc-v build now?
shamoe has joined #riscv
<geertu>
drewfustini: mmc[01] are the mmc instances in the SoC.dtsi, which are SoC-specific
<geertu>
How they are wired up and used (e.g. eMMC or µSD) is board-specific.
<geertu>
Esmil: ^
<Esmil>
yes, i know. but the documenation calls the peripherals EMMC, SDIO0 and SDIO1, so i think it makes sense to keep those names as the reference handles
<Esmil>
(like the other peripherals in th1520.dtsi does)
<Esmil>
*do
davidpaul has joined #riscv
<geertu>
Esmil: Ouch, so you can make a board that has eMMC on the instance that's officially called µSD?
<Esmil>
i guess you could, yes
<Esmil>
..but luckily that's now how the Lichee Pi 4A or BeagleV Ahead are designed
<Esmil>
*not
<drewfustini>
I think Esmil is refering to the T-Head Address Map PDF which has
<drewfustini>
so I can see the logic in me changing to those labels in the dtsi
<geertu>
IC
<drewfustini>
the only two boards that are shipping currently (lpi4a and ahead) are using SDIO0 controller for microSD slot
<drewfustini>
I think in terms of IP, there are 3 versions of the DWC Mobile Storage Host Controller and they to call them in the address map: EMMC, SDIO0, SDIO1
<drewfustini>
s/3 versions/3 instances/
<drewfustini>
I suppose the dts files can also have an alias for to map to a more logical name to make it clearer what is connected to what on the actual boards. that is a normal approach, right?
<Esmil>
yeah, and probably EMMC is instantiated a little different, or maybe it's the only one that can do bus_width = <8> or something like that
foton has joined #riscv
<drewfustini>
The vendor SDK has this in the dtsi for SoC. So you are right there is not really much difference as the dtsi level. https://www.irccloud.com/pastebin/7jsF5HMU/
<mps>
all boards (arm32, arm64) with emmc and mmc have device named mmc0 and mmc1
<geertu>
drewfustini: Esmil: Aliases belong in the board .dts
<drewfustini>
Okay, that makes sense.
<geertu>
And they should indeed be of the form mmc%u
davidpaul has quit [Quit: Client closed]
foton has quit [Client Quit]
<drewfustini>
Okay, thanks. I'll work up a new dts series with these changes.
foton has joined #riscv
<courmisch>
mps: U-boot seems to sometimes call them mmc1 and mmc2, which got me quite confused
<courmisch>
it seems that AnalogLang's CRM glitched and resent me an old invoice
<courmisch>
but I guess it means that they are now shipping the second batch of K230 boards
<mps>
courmisch: yes, but looks like they work to consolidate this
<mps>
one of my arm32 had wifi card as mmc1 (actually SDIO) but now changed to mmc2
jacklsw has quit [Ping timeout: 245 seconds]
<mps>
few years ago I argued with some developers about this issues
<mps>
and looks like they started to 'fix' this
heat has quit [Read error: Connection reset by peer]
heat has joined #riscv
cow321 has quit [Remote host closed the connection]
cow321 has joined #riscv
aburgess has quit [Ping timeout: 264 seconds]
cow321 has quit [Remote host closed the connection]
cow321 has joined #riscv
khem has joined #riscv
sakman has quit [Ping timeout: 264 seconds]
solol has quit [Quit: Client closed]
<palmer>
so I'm back to work this week, but I'm super buried so just poke me on stuff if I'm missing anything. I'm going to go scrub through patchwork for fixes...
vagrantc has joined #riscv
sakman has joined #riscv
Tenkawa has joined #riscv
<Tenkawa>
Esmil: Greetings... before I pulled it and tried I wanted to see if your 6.7-rc4 was in a useable state on the JH7110 at the moment
<Esmil>
you can try my jh7110 branch, but the most important patches are the two fixing module loading
<Tenkawa>
Esmil: It said those patches from 20231127-module_linking_freeing-v4-0-a2ca1d7027d0@rivosinc.com/ were already applied to your jh7110 source...is this correct?
<Tenkawa>
(Just wanted to make sure patch wasn't misleading me)
<Tenkawa>
everything in arch/riscv/kernel/module.c was applied here
aburgess has joined #riscv
<palmer>
Esmil: I'm just putting those in the queue for fixes
aburgess_ has joined #riscv
aburgess has quit [Ping timeout: 276 seconds]
shamoe has quit [Quit: Connection closed for inactivity]