Saturday, 2023-03-18

*** poetaster is now known as Guest813008:36
*** ggabriel is now known as Guest814212:35
*** ggabriel is now known as Guest815516:10
T42<b100dian> Anyone has experiece with enabling ramoops on a kernel that doesn't seem to write to /sys/fs/pstore?17:46
T42<b100dian> I'm trying to follow but I don't know if any address will do17:47
malwhich device and kernel version?17:51
T42<elros34> be aware this may not work if your kernel crash very early. That was the case for CONFIG_VT long time ago, enabling it caused crash but there was nothing. Only connecting to uart reveals crash and faulty commit.17:53
T42<elros34> Have you even tried to add something to lineages's ramdisk content to confirm that your issue is not related to some checksum/security?17:55
T42<b100dian> mal: I'm trying to boot asus sake, base is lineage 18.1, is the device with boot.img and vendor_boot.img (header version 3). Kernel is 5.418:00
malhave you correctly assembled the kernel?18:01
T42<b100dian> @elros34 I did try to mkbootimg from unpacked lineage images and it worked. I did not yet try to make a change to lineage's ramdisk, that can be arranged18:01
malhmm, but if you do regular hybris-hal build I think it should get needed parameters from android side build system18:01
T42<b100dian> mal: I tried three kernels with lineage boot images which correspond to these last 3 commits18:02
malI should try to figure out why highlighting is not working when the person using my nick is in telegram18:03
T42<elros34> do you still mess with hybris-boot init-script permission and and create cpio manually to make everything even harder to debug?18:04
T42<b100dian> They all booted lineage, except the last one that didn't so fully, flickered and didn't have touchscreen and rebooted after some time (but it did have lineage display)18:04
T42<b100dian> @elros34 I reverted to root:root in cpio :)18:04
T42<b100dian> (I did not "manually" create it, but throuh make hybris-recovery (IIRC), but I did that root:root change one and is now backed out)18:05
T42<b100dian> Also @elros34 since then I tried an init that just "sleep 30; reboot -f" (shortly put) and the device still reboots faster - that's why I believe kernel has some issues with this busybox init sh script. But I don't have pstore ramoops logs..18:07
T42<elros34> and did you add all needed lines first like busybox install * ?18:07
T42<b100dian> Umm good point, I didn't notice that!18:09
T42<b100dian> I _thought_ sleep and reboot were internal commands..18:11
T42<elros34> You should do small steps, you have working lineage boot.img then make small changes to it, compile your kernel, disabled verity/selinux, add minimal hybris config, enable rndis, add small change to ramdisk and test every time boot.img. It shouldn't be hard to figure out everything with such a steps, it's only time consuming18:13
T42<elros34> the only "internal" commans are 'sh' and 'mdev' but you still probably need PATH set so it's not worth to do it like that18:14
T42<b100dian> Thanks for the advice. Just checking. verity is blank vbmeta and selinux is setting it androidboot-permissive?18:21
T42<b100dian> Also, meanwhile I generated a panic like exemplified here, but under lineage, and I got a ramdump screen on next reboot (and no sys/fs/pstore/ramoops in the next one) - that's why my initial question today18:22
T42<elros34> maybe you can use CONFIG_PANIC_TIMEOUT to figure out whether you have panic at all18:53
T42<elros34> or printk.always_kmsg_dump=1 to just test pstore18:54
T42<b100dian> These are great ideas, thanks!18:59
T42<b100dian> Hmm.. appending one file to lineage boot ramdisk makes it not boot22:35
T42<b100dian> I've unpacked the lz4 to ramdisk-appendone.cpio, then used `find ./testonefile | cpio -oA -H newc -F ramdisk-appendone.cpio`, then lz4 it back and mkbootimg 'as always' (e.g. as when I changed the kernel and it worked)22:37

Generated by 2.17.1 by Marius Gedminas - find it at!