20:17 <javier__> steev: I had some time to do a git bisect today and found that the offending commit is 965278dcb8ab ("ARM: 8356/1: mm: handle non-pmd-aligned end of RAM")

20:17 <javier__> steev: v4.1 + the mentioned commit reverted makes the Peach Pit/Pi to boot using a FIT image

20:18 <si1v3r> Did it make wifi work as a builtin?

20:18 <si1v3r> a fit image with modules sounds kinda silly.

20:19 <steev> javier__: oh wow, thanks - i'm sorry, i ended up getting my minnie early (i guess amazon screwed up and shipped some early) and i was playing with it all weekend

20:19 <javier__> si1v3r: we are talking about a completely unrelated issue

20:19 <steev> si1v3r: uh, a fit image is the kernel binary and dtb files

20:20 <javier__> si1v3r: the WiFi not working when built-in is related to how the in-kernel firmware loader works

20:20 <javier__> it tries to load the firmware before the rootfs is mounted, that fails and doesn't retry

20:20 <steev> damn it

20:21 <steev> javier__: happen to know how to make a "--depth=1" pull in the rest of history

20:21 <steev> that wasn't particularly clear - i cloned with --depth=1 initially i guess

20:22 <steev> wondering how to pull in previous history, but i'm trying a git pull --depth 30

20:22 <steev> it seems to be doing something

20:22 <javier__> steev: don't worry, I was courious. I'm not that familiar with the ARM MMU code though to understand what's going on

20:22 <Wizzup> git fetch

20:23 <Wizzup> ?

20:23 <steev> Wizzup: if it's depth 1 it doesn't update the old history, you have to pass it a new --depth

20:23 <javier__> I'll take a look to the patch tomorrow and see if I can report to the list

20:23 <javier__> steev: sorry, I don't get the question

20:23 <steev> huh, that's definitely interesting, i'm gonna mention it in #armlinux

20:25 <javier__> steev: cool, I gotta go now

20:25 <javier__> bye!

20:25 <steev> have fun! thanks for the update!

20:25 <javier__> steev: you are welcome

20:46 <steev> Wizzup: i'mma test with javier__'s claimed fix

20:46 <Wizzup> :L)

20:46 <Wizzup> :)

23:06 <steev> javier__: confirmed, reverted 965278dcb8ab here, built image and peach_pi boots and so far all i've tested works

23:07 <javier__> steev: yep, I'll look tomorrow at the patch and try to figure out what is doing differently mainline u-boot than the stock RO u-boot

23:10 <steev> javier__: i thought peach used coreboot, no?

23:11 <javier__> steev: no, it uses u-boot as its RO bootloader

23:11 <javier__> steev: there is a coreboot + depthcharge port for Snow and Peach Pit/Pi but it was not finished / stable enough to ship it

23:11 <steev> ahh

23:12 <steev> it would definitely be nice to have them, but yeah, it would be nice to know why that commit breaks things

23:13 <javier__> yeah, althought I thought the MMU was enabled by the kernel after is decompressed so I don't understand why is working with mainline u-boot but no with the stock one

23:13 <javier__> anyway, I'll try to look at it and try to answer Mark Rutland's patch thread with a sensible report

23:14 <javier__> since is something that has be handled during the v4.1 -rc cycle because is a regression

23:15 <javier__> I'll also ask sjoerd tomorrow who has a better understanding than me about the u-boot and kernel interactions and how the MMU is setup

23:25 <steev> ah

23:25 <steev> good idea, and yeah, not sure how that was missed

23:25 <steev> aside from people not using vendor supplied :)