Sunday, 2020-06-07

*** ahjolinna_ is now known as ahjolinna02:45
r0kk3rzdeathmist: starting over just wastes time02:47
T42<DrRamm> Hello guys, I got question: according  `11.1  Extracting the rootfs via adb` and `11.2  Flashing the boot image via adb` I've unpacked and flashed. But after reboot I booted in lineage (which shows me internal error caused vbmeta coz boot.img has different hash).08:24
T42<DrRamm> @DrRamm [Hello guys, I got question: according  11.1  E …], And journalctl is empty08:57
T42<DrRamm> ,---08:57
T42<DrRamm> | Sailfish OS (Rokua)08:57
T42<DrRamm> '---08:57
T42<DrRamm> [root@localhost ~]# journalctl -b08:57
T42<DrRamm> No journal files were found.08:57
T42<DrRamm> -- No entries --08:57
ggabrielDrRamm, at risk of pointing out the obvious... is the journal service started?08:58
T42<DrRamm> @ggabriel [DrRamm, at risk of pointing out the obvious... …], ggabriel, not sure. I rechecked my kernel via mer-config tool and I not sure about these warnings:09:14
T42<DrRamm> Do Not Exist DNE09:14
T42<DrRamm> CONFIG_RTC_DRV_CMOS -- optional09:14
T42<DrRamm> CONFIG_LBDAF -- for 32bits, I have 64bitd09:14
ggabrielDrRamm, I meant journalctl status systemd-journal - is that started? I would also check dmesg. I'm not sure how that kernel configuration is related to not having a journal09:16
deathmistggabriel: I've seen that "root@localhost" prompt before, he chrooted into the Sailfish OS rootfs from TWRP and expected to get journalctl logs from a system that never booted to the rootfs in the first place, he probably also needs to revert a commit that skips the ramdisk entirely09:33
deathmistafaik he can patch init/initramfs.c in his kernel to only return 0 from the skip_initramfs_param() method09:35
ggabrieldeathmist: thanks, so it wasn't all that obvious ;)09:37
deathmist@DrRamm want to link your kernel?09:42
T42<DrRamm> @deathmist [afaik he can patch init/initramfs.c in his ker …], [hybris-16.0 0bc02edd52ab] init: disable skip_initramfs thx deathmist09:43
T42<DrRamm>  1 file changed, 4 insertions(+), 1 deletion(-)09:43
T42<DrRamm> deathmist, thank you.09:44
deathmistthere was a better fix I was thinking about, e.g. reverting a commit in your kernel with a similar message as
T42<DrRamm> @deathmist [there was a better fix I was thinking about, e …], Well revert is fine, yeah "return 0" is a little bit dirty, but it's work09:45
T42<DrRamm> Okay, I've tried to solve problem by myself, but seems like I need help again:10:28
T42<DrRamm> I faced with "```[    5.206086] init: Couldn't open /sys/fs/selinux/null: No such file or directory```"10:28
T42<DrRamm> So seems like disabled AUDIT cause that and my device need keep it enabled. Okay. mer-config tool says10:28
T42<DrRamm> "```if your device needs its support in kernel, set AUDIT=y and SELINUX_BOOTPARAM=y ```"10:28
T42<DrRamm> +CONFIG_AUDIT=y10:28
T42<DrRamm> +# CONFIG_SECURITY_SELINUX_DISABLE is not set10:28
T42<DrRamm> +# CONFIG_DEFAULT_SECURITY_SELINUX is not set10:28
T42<DrRamm> my cmdline is ```10:28
T42<DrRamm> [    0.000000] Kernel command line: console=ttymxc1,115200 init=/init selinux=0 audit=0```10:28
deathmist@DrRamm send the full log of journal/dmesg, don't just pick random parts and hope that's the issue, in which telnet port is this, 23 or 2323?10:51
deathmistand please upload it to pastebin, don't use telegram specific features as most of us use IRC/Matrix10:52
T42<DrRamm> Dear deathmist, defconf + dmesg
T42<DrRamm> >in which telnet port is this, 23 or 2323?10:57
T42<DrRamm> I use uart10:57
deathmistah, lucky you ;) have you even tried USB though?10:58
deathmistit definitely did not even load the hybris initramfs, you need that if you want to boot Sailfish OS btw10:58
T42<DrRamm> deathmist, I do not see sense in USB if I have console via UART10:59
T42<DrRamm> >it definitely did not even load the hybris initramfs10:59
T42<DrRamm> Yes, because missed "`/sys/fs/selinux/null`"10:59
T42<DrRamm> deathmist, there is full defconfig
deathmistthat's not the issue, even if SELinux was enabled it should get to the hybris initramfs and print something more than that dmesg and reboot immediately to bootloader, what even is your device?11:05
deathmistSELinux should be disabled either way (SECURITY_SELINUX_BOOTPARAM_VALUE=0), does this kernel even boot LineageOS? I don't really see that as DEFAULT_SECURITY_TOMOYO was set by default for you11:05
T42<DrRamm> deathmist, my mistake, DEFAULT_SECURITY_TOMOYO enabled by default after enable back AUDIT11:21
deathmistI still have no clue what device, which kernel source or what hybris version you are trying to use11:22
T42<DrRamm> deathmist, device based on imx8mm (evk), kernel sources, hybris 16.011:31
deathmist@DrRamm maybe try reverting 6a45470bc36a2e59c8f131ffc90bf38f392525ed ("ANDROID: initramfs: Add skip_initramfs command line option") from the kernel instead of doing the ugly return 0 hack I suggested earlier, maybe that is the issue because I don't really see the initramfs coming up11:43
deathmistalso set SECURITY_SELINUX_BOOTPARAM_VALUE=0 in your defconfig as I suggested earlier11:44
T42<elros34> Did you flash hybris-boot.img or boot.img (android one)?11:50
T42<DrRamm> Elros,11:57
T42<DrRamm> >Did you flash hybris-boot.img or boot.img11:57
T42<DrRamm>  hybris-boot.img11:57
T42<DrRamm> deathmist,11:57
T42<DrRamm> >try reverting and set SECURITY_SELINUX_BOOTPARAM_VALUE=011:57
T42<DrRamm> ok11:57
T42<DrRamm> Elros, deathmist can you help me I make kernel changes from scratch using mer-config tool, reverted skip_initramfs and enable CONFIG_SECURITY_SELINUX_BOOTPARAM (with cmdline in as described in "`9.3.1  Bootloops`" ) but it causes bootloop. Without CONFIG_SECURITY_SELINUX_BOOTPARAM I can bo15:04
T42ot into lineage. Why I can boot in lineage OS if I flashed hybris-boot.img?15:04
T42<DrRamm> btw in SailfishOS-HardwareAdaptationDevelopmentKit- at chapter `11.1  Extracting the rootfs via adb` we can see tar arguments ```tar -xvzf``` but it's not correct: according there should be used ```tar -xvjf```15:13
T42<elros34> as mentioned by deathmist earlier there is no indication of initramfs in your dmesg. It looks like it init android. I still see in your bootparams skip_initramfs. Are you sure you reverted commit which add support for it and then rebuild hybris-15:21
T42<DrRamm> @elros34 [as mentioned by deathmist earlier there is no …], Elros, this link from official sources....
T42<elros34> You can see that in populate_rootfs() there is printk about unpacking rootfs which should be in your dmesg. I am not 100% sure (somebody need to confirm it) but I think you should not have that skip_initramfs in boot params now15:51
T42<elros34> You never said what kind of device is that so does this device even use abootimg by default or maybe u-boot bootloader + kernel?15:59
T42<DrRamm> @elros34 [You never said what kind of device is that so …], imx8mm, uboot + kernel16:38
T42<DrRamm> @elros34 [You can see that in populate_rootfs() there is …], Dear Elros, I've removed skip_initramfs in u-boot and now cmdline without that (but one more space before rootwait, but it doesnt matter). make installclean and make hybris-hal after that and flashed new uboot and hybris-boot.img. As result I booted into LineageOS and do17:07
T42not see any "Mer Debug" lines in dmesg as described in `9.2  Operating Blind on an Existing Device`. And yes, adb via usb works fine (you can check in my dmesg USB_STATE=CONFIGURED)
T42<elros34> @DrRamm so if that device don't use abootimg but regular zImage then hybris-boot.img will not work. What file returns when you run it against your regular android kernel image?18:12
T42<elros34> on mainline (not android) I set: CONFIG_INITRAMFS_SOURCE to out/target/product/$DEVICE/obj/ROOT/hybris-boot_intermediates/initramfs/ and then I put zImage to /boot so u-boot can find it18:14
T42<DrRamm> @elros34 [on mainline (not android) I set: CONFIG_INITRA …], @elros34 Correct me, how I should change lineage_evk_8mm_defconfig:CONFIG_INITRAMFS_SOURCE="" ?18:19
T42<DrRamm> I found  example like that CONFIG_INITRAMFS_SOURCE="$$KERNEL_INITRAMFS_SOURCE"18:20
T42<elros34> $ANDROID_ROOT/out/target/product/$DEVICE/obj/ROOT/hybris-boot_intermediates/initramfs/. You can use absolute path if you want18:21
T42Devotee_Of_My_GOD was added by: Devotee_Of_My_GOD18:23
T42<DrRamm> @elros34 [$ANDROID_ROOT/out/target/product/$DEVICE/obj/R …], seems working18:26
T42<DrRamm> ```[ 87% 750/857] Making initramfs : /home/kotsarev/mer/android/droid/out/target/product/evk_8mm/obj/ROOT/hybris-boot_intermediates/boot-initramfs.gz```18:26
T42<DrRamm> ```CONFIG_INITRAMFS_SOURCE="$$ANDROID_ROOT/out/target/product/$$DEVICE/obj/ROOT/hybris-boot_intermediates/initramfs/"```18:26
T42<DrRamm> @elros34  can you check my unpacked kernel? I use mkboot for that
T42<elros34> you sitll did not tell what 'file android_image' returns so I don't know what to expect in that paste18:40
T42<DrRamm> @elros34 [you sitll did not tell what 'file android_imag …], Sorry, didn't get you: are we talking about hybris-boot.img? And Image instead of zImage. At Lineage 16.0 I use boot.img as we get used to.18:44
T42<elros34> ok so take android boot.img and run 'file boot.img'.18:46
T42<elros34> There is nothing sailfish specifig so far, either your device need initramfs packged together with kernel in android bootimg (abootimg) or you use config_initramfs_source. That should be easy to figure out.18:48
T42<DrRamm> @elros34 [ok so take android boot.img and run 'file boot …], boot.img from Lineage 16.018:51
T42<DrRamm> $ file boot.img18:51
T42<DrRamm> boot.img: Android bootimg, kernel (0x40480000), ramdisk (0x43600000), page size: 2048, cmdline (init=/init androidboot.console=ttymxc1 androidboot.hardware=freescale cma=800M@0x400M-0xb80M an)18:51
T42<DrRamm> $ file hybris-boot.img18:51
T42<DrRamm> hybris-boot.img: Android bootimg, kernel (0x40480000), ramdisk (0x43600000), page size: 2048, cmdline (init=/init androidboot.console=ttymxc1 androidboot.hardware=freescale cma=800M@0x400M-0xb80M an)18:51
T42<elros34> then I don't know18:57
T42<elros34> I thought CONFIG_INITRAMFS_SOURCE will force kernel to use initramfs. Clearly from dmesg mmcblk2p5 is mounted (which I assume is your /system) and then some init (probably android) is started.18:57
T42<DrRamm> @elros34 [There is nothing sailfish specifig so far, eit …], If I see something like this18:57
T42<DrRamm> ```[    7.708554] usb0: HOST MAC18:57
T42<DrRamm> [    7.712917] usb0: MAC18:57
T42<DrRamm> [    7.722714] IPv6: ADDRCONF(NETDEV_UP): usb0: link is not ready18:57
T42<DrRamm> [    8.229095] android_work: sent uevent USB_STATE=CONNECTED```18:57
T42<DrRamm> Is that mean rootfs is working? `lsusb -v | grep iSerial` don't show me Mer,18:57
T42<DrRamm> @elros34 [I thought CONFIG_INITRAMFS_SOURCE will force k …], Yes, system_a is mmcblk2p518:58
T42<elros34> I wouldn't care about rootfs at all because your kernel still doesn't boot initramfs but instead boot from /system18:58
T42<DrRamm> @elros34 [I wouldn't care about rootfs at all because yo …], can you check dmesg? I have blank screen (at least it doesn't boot into lineage)
T42<maffeen> Can I use AOSP instead of Lineage OS?19:03
T42<maffeen> With Pie19:03
T42<DrRamm> @maffeen [Can I use AOSP instead of Lineage OS?], You can switch to Lineage during one day...19:04
T42<maffeen> @DrRamm [You can switch to Lineage during one day...], My device hasn't a Lineage OS and I'm having problems compiling. The only device tree on my device for Pie is experimental19:05
T42<maffeen> All the ROMs are based on Q, there is just BeastROM that is Pie19:05
T42<elros34> @DrRamm looks better but hybris-boot script doesn't print anything to dmesg so it's hard to guess. Do you have /init.log?19:05
T42<DrRamm> @elros34 [@DrRamm looks better but hybris-boot script do …], Wew my host can see that19:06
T42<DrRamm> [344279.827504] usb 1-1.2: Manufacturer: Mer Boat Loader19:06
T42<DrRamm> [344325.390962] usb 1-1.2: Manufacturer: Mer Boat Loader19:06
T42<DrRamm> [344365.839468] usb 1-1.2: Manufacturer: Mer Boat Loader19:06
T42<elros34> so it's good now. What helped?19:06
T42<DrRamm> @elros34 [so it's good now. What helped?], CONFIG_INITRAMFS_SOURCE....19:07
T42<DrRamm> @elros34 [@DrRamm looks better but hybris-boot script do …], yes telnet works. And I have log. Thank you!19:10
T42Ced %lastname% was added by: Ced %lastname%19:50
kalubedummy_netd is failing to build with21:31
kalube/srv/mer/toolings/SailfishOS- /tmp/ccW1TktK.ltrans0.ltrans.o: undefined reference to symbol 'g_main_loop_unref'21:31
kalube/srv/mer/toolings/SailfishOS- /srv/mer/targets/oneplus-enchilada-armv7hl/usr/lib/ error adding symbols: DSO missing from command line21:31
kalubeAny ideas?21:31
T42<elros34> maybe clean leftovers from previous build21:34
kalube@eros34 it's beena  while since I've built sailfish - that was on my first build on a fresh build tree and tooling21:35
kalubeForgive me - what should I run to clean up?21:35
T42<elros34> just an idea if there is anything in sources dir from previous build. IIRC: git checkout -- .; git clean -xdf  or remove directory and clone it again21:38
kalubeI ran21:39
kaluberm -rf hybris/mw/dummy_netd21:39
kalubebp --mw=
kalubegot the same error21:39
T42<elros34> Are you sure you use target and tooling? Latest still points to 3.2.1.x21:40
kalubesdk-assistant list says
kalubeYeah I used the version specifically for tooling not latest21:42
T42<elros34> check what sb2 ssu re returns21:42
kalubeyeh 3.321:42
kalubeI grep'd logs but couldn't find much21:43
kalubeWonder why it won't build21:43
T42<elros34> there wasn't any changes in dummy_netd so it must be something in your build env21:44
kalubeYeah that wouldn't surprise me to be honest...21:47
T42<elros34> unless it's too new libgbinder21:47
kalubeI have built libgbinder21:47
kalubethat could be it?21:47
kalubelibgbinder version : 1.0.43-021:48
kalubenot sure what version it expects21:48
T42<elros34> >= 1.0.721:49
kalubecould be too new then21:49
kalubeI grep21:49
kalubegrep'd libgbinder src for the function it can't find and got no results21:49
kalubewill try and older version of libgbinder21:49
kalubeoh wait nevermind the function is in a different subdir21:50
kalubelibgbinder probably isn't the issue21:50
kalubeg_main_loop_unref is in glib21:52
T42<elros34> still it's worth to try as there is no other dependencies22:10
T42<kalube> I'll try 1.0.35 - sorry for the late reply matrix decided not to pick up that message for some reason22:16
T42<kalube> huh, older libgbinder fixed it. Cheers22:23
T42<kalube> Will need to figure out what commit breaks it...22:23
T42<elros34> I would test this one: Especially pkgconfig change22:24
T42<kalube> think line 8 might be it
T42<kalube> Need to actually get stuff built as I was hoping to mess about with fingerprint today22:26
T42<kalube> I'll try changing that line back and see if that solves it22:26
T42<kalube> > if only there was CI to check this stuff22:26
T42<erfanoabdi> I made another package to build in Android side only, called stub netd and doing same thing as dummy netd but only returning valid values on output
T42<erfanoabdi> Maybe this can be useful...22:27
T42<kalube> @erfanoabdi cool! What does that do differently exactly? Sorry my understanding of android HAL is pretty... shit22:28
T42<erfanoabdi> @kalube [@erfanoabdi cool! What does that do differentl …], Dummy netd is hal for netd on gnu/linux side and just opening a binder interface and returning not valid values on called functions (netmgrd errors out about this on logcat but bot fetal)22:32
T42<erfanoabdi> But my hal is on Android side no need to deal with libgbinder and opening same interface but returning "OK" on any called functions so netmgrd likes it :)22:32
T42<kalube> haha, right got ya22:33
T42<kalube> Failing on droid-hal-version now... Error: Subprocess failed. Error: RPM failed: error: unpacking of archive failed on file /boot/boot-initramfs.gz;5edd6faa: cpio: open failed - File exists22:53

Generated by 2.17.1 by Marius Gedminas - find it at!