| T42 | <edp_17> @elros34 : I am on 4.3 now and it reboots continuously. I've masked lipstick but the issue still persists, so that is not the cause. | 08:28 |
|---|---|---|
| T42 | <edp_17> Is it normal that I do not have user@100000.service but I do have user@.service? | 08:28 |
| T42 | <elros34> yeah | 08:29 |
| T42 | <elros34> but still you mask user@100000.service | 08:29 |
| T42 | <elros34> I assume you were talking about files in /usr/lib/systemd/system/ | 08:32 |
| T42 | <edp_17> Yes, I was talking about /usr/lib/systemd/system/ | 08:36 |
| T42 | <edp_17> I'll try to mask user@100000.service | 08:37 |
| T42 | <edp_17> should I put this into /etc/systemd/user/ or /etc/systemd/system/ ? | 08:37 |
| T42 | <elros34> original service is in /usr/lib/systemd/system so /etc/systemd/system/ | 08:40 |
| T42 | <edp_17> I've masked it ther and the reboot has stopped. | 08:43 |
| T42 | <edp_17> Now, I can add persistent journaling and let it reboot to get some logs. /sys/fs/pstore is currently empty. | 08:45 |
| T42 | <edp_17> journal: https://paste.ubuntu.com/p/WD3Kkjngmn/ | 08:54 |
| T42 | <edp_17> logcat: https://paste.ubuntu.com/p/FHrV6vDK7X/ | 08:54 |
| T42 | <edp_17> dmesg: https://paste.ubuntu.com/p/cdCCGXztdz/ | 08:54 |
| T42 | <edp_17> /sys/fs/pstore is still empty | 08:54 |
| T42 | <edp_17> /proc/last_kmsg: https://paste.ubuntu.com/p/rZh4pbYvGn/ | 08:56 |
| T42 | <elros34> what do you have in /proc/cmdline? | 09:10 |
| T42 | <edp_17> /proc/cmdline: https://paste.ubuntu.com/p/7j5HP6NWyQ/ | 09:12 |
| T42 | <elros34> "androidboot.selinux=enforcing" > selinux=0 audit=0 | 09:21 |
| Thaodan | androidboot.selinux=permissive that's what I have on xperia 1 | 09:22 |
| Thaodan | To be exact I have: "rcupdate.rcu_expedited=1 rcu_nocbs=0-7 cgroup.memory=nokmem,nosocket lpm_levels.sleep_disabled=1 service_locator.enable=1 msm_drm.blhack_dsi_display0=dsi_panel_somc_kumano_cmd:config0 androidboot.selinux=permissive androidboot.memcg=1 msm_rtb.filter=0x3F ehci-hcd.park=3 coherent_pool=8M sched_enable_power_aware=1 user_debug=31 printk.devkmsg=on loop.max_part=16 kpti=0 androidboot.hardware=griffin buildvariant=user | 09:22 |
| Thaodan | audit=0 loglevel=3 androidboot.verifiedbootstate=orange androidboot.keymaster=1 root=PARTUUID=f9cdf7ba-b834-a72a-f1c9-d6e0c0983896 androidboot.bootdevice=1d84000.ufshc androidboot.baseband=msm msm_drm.dsi_display0=2: androidboot.slot_suffix=_a skip_initramfs rootwait ro init=/init androidboot.dtbo_idx=0 androidboot.dtb_idx=0 androidboot.bootloader=xboot oemandroidboot.xboot=1315-2591_X_Boot_SM8150_LA2.0_Q_115 | 09:22 |
| Thaodan | androidboot.serialno=QV711BUJ1T oemandroidboot.babe08a4=598 startup=0x00008004 warmboot=0x77665501 oemandroidboot.babe09a9=00 oemandroidboot.babe137e=51 oemandroidboot.imei=35498510150428 oemandroidboot.securityflags=0x00000003 | 09:22 |
| Thaodan | " | 09:22 |
| T42 | <edp_17> I have selinux=0 but not audit=0 in BoardConfigCommon.mk: https://paste.ubuntu.com/p/FQG6PfrcmT/ (re @elros34: "androidboot.selinux...) | 11:17 |
| T42 | <edp_17> I'll add audit=0 and retry. | 11:18 |
| T42 | <edp_17> It didn't help. I've added audit=0 to the end of BOARD_KERNEL_CMDLINE, rebuilt hybris-hal and flashed the new hybris-boot.img on device. The /proc/cmdline is the same: https://paste.ubuntu.com/p/8Gk7j3VJhH/ | 11:44 |
| T42 | <elros34> obviously, if selinux=0 was ignored then why audit=0 wouldnt try CONFIG_CMDLINE_EXTEND | 11:54 |
| T42 | <edp_17> @elros34 : Is installing the kernel on device the only way to check whether the selinux=0 audit=0 are in there? | 12:41 |
| T42 | <elros34> I guess grepping zImage might also works | 12:41 |
| T42 | <edp_17> How, zcat doesn't work on it. | 12:41 |
| T42 | <edp_17> ? | 12:41 |
| T42 | <elros34> I didn't try, just guessing. just "grep audit <path/to/kernel zImage>" | 12:42 |
| T42 | <TheVancedGamer> no? | 12:42 |
| T42 | <TheVancedGamer> file zImage | 12:42 |
| T42 | <TheVancedGamer> wait, that doesn't work | 12:43 |
| T42 | <edp_17> No, it didn't. :) | 12:43 |
| T42 | <edp_17> I'll flash it and try. No worries. | 12:44 |
| T42 | <TheVancedGamer> there used to be a way to read cmdline inside kernel | 12:44 |
| T42 | <TheVancedGamer> I can't remember it | 12:44 |
| T42 | <edp_17> Nope, /proc/cmdline is still: androidboot.selinux=enforcing | 12:47 |
| T42 | <edp_17> And hybris-boot.img has the same size as before. | 12:47 |
| T42 | <elros34> have you checked .config? | 12:47 |
| T42 | <edp_17> Here: https://paste.ubuntu.com/p/b69pz5NcxJ/ | 12:51 |
| T42 | <edp_17> Grepping for SELINUX | 12:52 |
| T42 | <elros34> grep for cmdline which we are talking about. Good that you have CONFIG_SECURITY_SELINUX_BOOTPARAM_VALUE=0 but your dmesg is not full so still it's not clear | 12:55 |
| T42 | <edp_17> How can I get a full dmesg? (user@100000 and lipstick is currently masked, so the device doesn't reboot) | 13:02 |
| T42 | <elros34> like in hot-hadk: kernel buffer is small so you need to be fast | 13:03 |
| T42 | <edp_17> grep CMDLINE in .config: CONFIG_CMDLINE="" | 13:03 |
| T42 | <edp_17> This is the issue then!? | 13:03 |
| T42 | <elros34> so you were supposed to add CONFIG_CMDLINE_EXTEND=y | 13:03 |
| T42 | <elros34> did you add it? | 13:04 |
| T42 | <edp_17> Yes, I did. In the kernel config there is a CONFIG_CMDLINE="" | 13:06 |
| T42 | <edp_17> I add the selinux=0 audit=0 into it and re-try | 13:06 |
| T42 | <edp_17> Now, I have CONFIG_CMDLINE="selinux=0 audit=0" in .config. And hybris-boot.img size is different. (Actually, the new one is smaller.) | 13:09 |
| T42 | <edp_17> I flash it on device and try. | 13:10 |
| T42 | <adampigg> there are tools to extract a bootimg and show you the contents | 13:10 |
| T42 | <edp_17> /proc/cmdline still has: androidboot.selinux=enforcing | 13:13 |
| T42 | <elros34> obviously it will have it because you did not remove it. The point is about whether there is something new after enabling config_cmdline_extend | 13:15 |
| T42 | <edp_17> I see. Here is the full contect: https://paste.ubuntu.com/p/X8dvwhzj49/ | 13:18 |
| T42 | <elros34> so if you are sure that extend is enabled then maybe only editing dtb will help | 13:25 |
| T42 | <edp_17> Well, I've added that into BoardConfigCommon.mk. How can I debug dtb? (What is dtd by the way?) | 13:41 |
| T42 | <edp_17> I meant how can I edit dtb? | 13:42 |
| T42 | <elros34> you added what to BoardConfigCommon.mk? | 13:43 |
| T42 | <edp_17> CONFIG_CMDLINE_EXTEND := selinux=0 audit=0 | 13:45 |
| T42 | <edp_17> I though, this is what I had to add. | 13:45 |
| T42 | <elros34> https://cateee.net/lkddb/web-lkddb/CMDLINE_EXTEND.html | 13:46 |
| T42 | <edp_17> Ops, Sorry. I remove and add into defconfig. | 13:46 |
| T42 | <edp_17> I've added CONFIG_CMDLINE_EXTEND=y into defconfig but it has stripped out. .config doesn't have it. | 13:51 |
| T42 | <edp_17> Would this work? https://cateee.net/lkddb/web-lkddb/CMDLINE.html | 13:56 |
| T42 | <edp_17> If I set CONFIG_CMDLINE="selinux=0 audit=0" ? | 13:56 |
| T42 | <elros34> I though you already set it together with extend? Did you even had CONFIG_CMDLINE_EXTEND is not set or =n in .config? If not then it's not supported | 13:59 |
| T42 | <edp_17> No, because when I set CONFIG_CMDLINE="selinux=0 audit=0" in defconfig, the CONFIG_CMDLINE_EXTEND was added into BoardConfigCommon.mk. | 14:02 |
| T42 | <edp_17> Then I removed it from boardconfigcommon and added into defconfig. | 14:02 |
| T42 | <edp_17> I didn't have CONFIG_CMDLINE_EXTEND in neither defconfig nor .config by default. And no, it wasn't 'n' or 'not set' in .config either. | 14:03 |
| T42 | <edp_17> Interesting because CMDLINE_EXTEND is mentioned in arch/arm/Kconfig. | 14:06 |
| T42 | <edp_17> @elros34 : what would you like to see in /proc/cmdline ? | 14:15 |
| T42 | <elros34> selinux=0 audit=0 | 14:22 |
| T42 | <edp_17> I see. | 14:24 |
| T42 | <edp_17> @elros34 : There is something I don't understand. If I flash the kernel that works well with 4.2, that doesn't have the selinux=0 in /proc/cmdline. | 15:20 |
| T42 | <edp_17> So, why do you expect to see selinux=0 in there? | 15:22 |
| T42 | <edp_17> and the device keeps rebooting the same way with that kernel built for 4.2. :( | 15:24 |
| T42 | <elros34> it should be there for every release: https://github.com/mer-hybris/mer-kernel-check/pull/27/files for hybris-14. I am not claiming this is your main issue because I do not see anything obvious in logs nor selinux state in partial dmesg. Also at some point selinux was enabled in systemd so thisi could also play a role here | 15:27 |
| T42 | <elros34> could be also something else, dsme wdd: Error opening WD /dev/watchdog0: Device or resource busy, maybe it triggers reboot | 15:31 |
| T42 | <edp_17> I probably have an old mer-kernel-check because line 290 and 292 are not in my localmer_verify_kernel_config | 15:31 |
| T42 | <elros34> they are not there because this commit is not merged since forever | 15:32 |
| T42 | <edp_17> Yeah, I am on this commit 9baf0a05d4872bd7b636dc85fad276dcfb5bcb88. There are few new ones since. | 15:32 |
| T42 | <edp_17> I've checked journal and this message is there but the device didn't get rebooted (user@100000.service is masked) (re @elros34: could be also someth...) | 15:37 |
| T42 | <edp_17> If this was the trigger, the device should have rebooted now too. No? | 15:37 |
| T42 | <elros34> probably | 15:38 |
| T42 | <edp_17> dsme is a good catch because if I try to OTA the device with a kernel that supports waydroid installed, the device reboots in middle the upgrade process, right at when it is installing dsme. | 15:39 |
| T42 | <edp_17> @elros34 : I also don't understand why the device didn't get reboot when 4.3 is flashed (no OTA updated) and booted up at first time. If the kernel or something else at the sfos side is wrong the device should keep rebooting right after the initial setup, shouldn't it? This behaviour just doesn't make sense. | 16:47 |
| T42 | <edp_17> Is there any process that isn't running when the device is going through the initial setup and right after but kicks in after the reboot? | 16:48 |
| T42 | <elros34> according to your logs there is one: waydroid*.service | 16:49 |
| T42 | <edp_17> Yes it is on my OTA updated device. I already thought about that but when I flash the image, that is not there. And the device keeps rebooting. So I don't think that would be the one. | 16:50 |
| T42 | <edp_17> Tomorrow, I'll try again without waydroid*.service to see whether that solves the issue. | 16:52 |
| T42 | <elros34> no idea what is wrong if there is no reboot reason in log | 16:55 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!