| 08:20 | Ke> | noboot with rmk/mcbin |
| 08:21 | Ke> | again the usual nothing in serial console, maybe a bad toolchain again |
| 08:21 | Ke> | or incompatible device tree |
| 08:25 | Ke> | was actually good to have reboot once in a while systemd had semantics changes and boot failed due to that |
| 08:27 | Ke> | hmm or perhaps I just had not rebooted since last config changes |
| 08:44 | jnettlet[m]> | Ke: this is on Debian Stretch? |
| 08:44 | Ke> | buster |
| 08:45 | Ke> | the device tree is from linux-4.16ish with gpio pin uarts enabled |
| 08:45 | jnettlet[m]> | that shouldn't make a difference |
| 08:46 | Ke> | something definitely is |
| 08:46 | Ke> | there were also some oldconfig questions, it's always possible I disabled some driver |
| 08:46 | jnettlet[m]> | although I believe rmk's patches require some dts changes....but generally that would just cause devices not to work....not to not boot |
| 08:46 | Ke> | I can retry in a while |
| 08:47 | Ke> | eg. after I have tested debian linux-4.16 etc. |
| 10:11 | jnettlet[m] | 10:11 * jnettlet[m] goes to download Debian Buster |
| 10:12 | Ke> | might not be worth it |
| 10:12 | Ke> | but would be nice if debian stable toolchains were in the testing matrix |
| 10:12 | Ke> | at least as soon as buster gets to be stable |
| 10:13 | Ke> | right now there is a new toolchain every other week |
| 10:13 | Ke> | in general should be more linaro kind of thing to watch those |
| 10:13 | suihkulokki> | ? |
| 10:14 | suihkulokki> | what's the topic? |
| 10:14 | Ke> | bad toolchains for aarch64 |
| 10:15 | jnettlet[m]> | Ke: are you compiling the kernel on the mcbin? |
| 10:15 | Ke> | yes |
| 10:16 | Ke> | native compile with make deb-binpkg or whatever it was |
| 10:16 | Ke> | though on a vm, but still native buster compile, I don't think the vm matters |
| 10:17 | suihkulokki> | and there is some issues on the native compile? |
| 10:17 | Ke> | well very early boot fails for me |
| 10:17 | Ke> | suihkulokki: have you tested linux-4.16 on mcbin? |
| 10:18 | suihkulokki> | running it right now |
| 10:18 | suihkulokki> | from experimental |
| 10:18 | Ke> | whatever I tried was linux-4.16 with rmks mcbin branch on top |
| 10:18 | Ke> | but the oldconfig I did was moderately lazy |
| 10:19 | Ke> | I guess debian does native compiles using approriate debian toolchain? |
| 10:19 | suihkulokki> | yes |
| 10:19 | suihkulokki> | I realllyyy doubt it's the toolchain |
| 10:20 | suihkulokki> | what does file /boot/vmlinuz* say |
| 10:20 | Ke> | in my experience if very early boot fails, it's often build environment, if something else, it's never the build env |
| 10:21 | Ke> | /boot/vmlinuz-4.16.0: gzip compressed data, max compression, from Unix |
| 10:21 | Ke> | suihkulokki: wow, there is something suspicious there |
| 10:21 | suihkulokki> | grub should be able uncompress that, but chances are it didn't |
| 10:22 | Ke> | suihkulokki: well the buster kernel is /boot/vmlinuz-4.15.0-2-arm64: MS-DOS executable |
| 10:22 | Ke> | suihkulokki: so there is a trivial difference |
| 10:23 | Ke> | make -j$(nproc) bindeb-pkg was what I used, wonder what controls, what is selected as vmlinux |
| 10:23 | suihkulokki> | you can try gunzip the file and see if that works |
| 10:24 | Ke> | zcat /boot/vmlinuz-4.16.0 | file - |
| 10:24 | Ke> | /dev/stdin: MS-DOS executable |
| 10:28 | Ke> | diff between .configs does not seem to show anything compression related |
| 10:28 | Ke> | did the bindeb-pkg make target change |
| 10:30 | Ke> | let's see if that did the trick |
| 10:43 | Ke> | nope |
| 10:43 | Ke> | still fails |
| 10:47 | Ke> | suihkulokki: is linaro testing debian toochains somehow? |
| 10:48 | Ke> | I guess they receive more and more testing, as people start actually running distro kernels on devices |
| 10:50 | suihkulokki> | Yes, the toolchains are intensly tested |
| 10:51 | suihkulokki> | Ke: did you try with earlycon kernel parameter? |
| 10:51 | Ke> | nope, I have no idea what low level parameters should I give |
| 10:51 | Ke> | I have to specify the uart somehow, as I understand |
| 10:52 | suihkulokki> | just add earlycon, the uart parameters are already set in the device tree |
| 10:52 | Ke> | should the debian default .config support that? |
| 10:53 | Ke> | or do ICONFIG_SERIAL_EARLYCON=y |
| 10:53 | Ke> | I guess that's it |
| 10:53 | suihkulokki> | try and find out? :) |
| 10:59 | Ke> | suihkulokki: thanks, but nothing |
| 10:59 | Ke> | either I need the low level params or something else fails |
| 11:03 | suihkulokki> | if earlycon doesn't help, it often points to device tree issues |
| 11:05 | Ke> | possibly, this device tree is though throughly tested on latest kernel on debian buster |
| 11:05 | Ke> | linux-4.15.0-2 or something |
| 11:06 | Ke> | I could get a new dtc from linux-4.17-rc1 |
| 11:06 | suihkulokki> | yes, but it's possible rmk's kernel doesnt like it |
| 11:06 | Ke> | or dtb |
| 11:06 | Ke> | yup |
| 11:07 | suihkulokki> | just point to the dtb in the package you built |
| 11:14 | Ke> | still absolutely nothing |