<Ashy>
need to actually test it on the rockpro64...
ryantrinkle has quit [Ping timeout: 268 seconds]
<Ashy>
it boots \o/
samrose has joined #nixos-aarch64
ryantrinkle has joined #nixos-aarch64
orivej has quit [Ping timeout: 245 seconds]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 240 seconds]
zupo has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 240 seconds]
zupo has joined #nixos-aarch64
<samueldr>
maybe someone has a pointer (no, not those) to share, I'm having issue with a specific phone. Any kernels I build through nix and its cross-compilers will not boot on-device, though any kernel built using random cross-compilers, either the official one from the OEM, or random ROM kernels will
<gchristensen>
wow
<gchristensen>
does diffoscope have anything interesting to sayV
<samueldr>
and I'm 99.9% sure it's the kernel build itself, and not the packaging, as unpacking and repacking every files, with a script, and changing only the kernel shows the same results
<samueldr>
hmm, I don't know, I don't know if it has fancy heuristics for kernel images
<sphalerite>
samueldr: same config?
<samueldr>
"yes" and "no"
<samueldr>
a multitude of revisions
<samueldr>
but the phone directly goes to fastboot
<samueldr>
so I guess it fails early enough that the bootloader can still fall back
<samueldr>
I don't have any kind of serial :/
<samueldr>
and I am having issues with the build with "the same configuration"... that's my next step
<samueldr>
make oldconfig and make prepare are used
<samueldr>
and this changes the configuration
<samueldr>
though I *did* extract the config from the actual built image (multiple images)
<samueldr>
and don't see anything really, but at the same time, I don't know what could be causing this
<samueldr>
I even thought "maybe it's ABL that can't ungzip the kernel right, let's repack the Image.gz!"
<samueldr>
nah, unpacked/repacked the same way, mine fails, oem works
<samueldr>
I looked at the start through ghidra, don't see anything silly, but don't know what to look for
<samueldr>
code1 is an unconditional jump to a (different) location
<samueldr>
and different too for different working kernels
<sphalerite>
can you compare the vmlinux.o's with diffoscope maybe?
<samueldr>
yeah, that's the next step, though does it have fancy heuristics there?
<gchristensen>
I don't think so
<gchristensen>
but diffoscope is just about the only tool in my toolbox, so thusly is my trusty hammer
<samueldr>
hmm, the html view is not that useful
<samueldr>
Max diff block lines reached; 378187084/378211205 bytes (99.99%) of diff not shown.
<samueldr>
(tbf, right now it's not from the same revision, I have to get another build going)
<gchristensen>
ouch
ris has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos-aarch64
fooker has joined #nixos-aarch64
<fooker>
Hi, I'm trying to boot a RPi 3B+ with the latest 19.03 image (19.03.173549.ebdc4419310) using a serial adapter. I'm seeing uBoot starting the kernel with 'Starting kernel ...' but no further output. Of course I've added the 'console=ttyS1,115200n8' and I can see it in uBoot. Any hints?
<samueldr>
hi, the u-boot from 19.03, with the latest kernel, will not boot fine on the 3B+
<samueldr>
you may want, fooker, to use 19.09 (soon to be the current stable) or nixos-unstable
<samueldr>
where the u-boot version has been upgraded and knows about the 3B+
<fooker>
ok. thanks. I will give it a try
<samueldr>
going back to diffoscoping, got back to what is supposed to be the same revision of the kernel (though won't build without changes to the config)
<samueldr>
still, I guess it would need heuristics to make sense of the output to make it useful in that case
<samueldr>
though both were built with gcc 4.9
<samueldr>
but not the same revision obviously :/
<samueldr>
and here even code1 is the same, so if it was an issue (it wasn't, since other booting kernel differ) it would be good now
<samueldr>
hmm
<samueldr>
I have re-done the steps to make it build, from the original config file
<samueldr>
most of the changes shouldn't really matter, hopefully
<samueldr>
though it definitely points towards the source used not being what was released
orivej has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]