01:50 <tlwoerner> marex: i thought it would be nice to add "gpio-line-names" for the olinuxino-maxi, but then i realized that in order to do so i'd have to create 4 u-boot defconfigs and 4 linux kernel device trees

01:51 <marex> tlwoerner: how so ?

01:51 <tlwoerner> unless i can figure out how to programmatically tell the difference between the 4 imx233-olinuxino boards

01:52 <tlwoerner> because the u-boot defconfig includes the name of the linux device tree

01:52 <marex> oh, because of the variants

01:52 <marex> there is the maxi, mini, nano and something ?

01:52 <tlwoerner> micro

01:53 <marex> good thing they stopped just short of yocto

01:53 <tlwoerner> lol!

01:54 <marex> tlwoerner: hmmm, does the same U-Boot work on all of them ?

01:54 <marex> maybe you can at least shift the decision to U-Boot shell

01:54 <marex> and there you can script it

01:54 <tlwoerner> apparently (i only have the maxi...i think)

01:55 <tlwoerner> a script in u-boot would have to "look" for some difference? e.g. check some gpios or something?

01:56 <marex> yeah, but at least you dont have to encode four DTs into U-Boot itself

01:58 <tlwoerner> okay. so use the same u-boot on all 4, write a script to figure out the variant, which then chooses the correct DT for linux?

01:58 <marex> tlwoerner: that might be easiest , if possible , yes

02:00 <tlwoerner> would i have to rely on the manufacturer doing something explicit in hardware, or are there other tricks? e.g. probing peripherals?

02:00 <tlwoerner> i guess looking through the schematics or manuals might show peripherals at different locations

02:02 <marex> tlwoerner: there are differences

02:03 <marex> the reduced versions might not have all the I2C peripherals , although , probing i2c is a bit iffy practice, it might lock up ancient I2C devices

02:06 <tlwoerner> digikey doesn't have any in stock

02:07 <tlwoerner> mouser has a couple

15:31 <Tartarus> For platforms where there's a run-time way to check what the variant is, if U-Boot doesn't have to care which one it's on to function, 1 U-Boot picking the right dtb for Linux is best. If U-Boot itself has to care (some difference that can't be run-time safely failed) then SPL picking the right dtb to pass to full U-Boot is best, assuming we have space for that.

16:56 <marex> tlwoerner: there is also DTB_RESELECT

16:56 <marex> Tartarus: ^

16:56 <marex> that's u-boot picking the right dtb for u-boot

17:03 <Tartarus> Right, if we need it

17:04 <Tartarus> Esp given the "fun" of some of the TI evms, if we can just punt to Linux, with the full DTB and not have to worry about things in U-Boot, that's the best

17:04 <Tartarus> But if we need it in U-Boot, DTB_RESELECT (or the SPL option to find the right FIT name match?)

18:48 <marex> Tartarus: if we could just apply DTOs in Linux ...