This bug started in redhat bugzilla because I thought it's special for Fedora. https://bugzilla.redhat.com/show_bug.cgi?id=1492349 It reproduced on my Asus GL553VE, and tested on kernels 4.8.9 - 4.15.6 System loads correct with grub option acpi=off, but without it fails with error: tmp_crb MSFT010:00: [Firmware Bug] ACPI region does not cover the entire command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x200] vs fed40080 f80 NMI: watchdog: Watchdog detected hard lookup on cpu1 Watchdog: BUG: Soft lookup - CPU#2 stuck for 23s! Logs for fedora 26, 4.15.6: https://paste.fedoraproject.org/paste/tz0-1GBgWNw6dIKIAsqn6Q As I found it isn't only fedoras bug. And reproduced on other ASUS notebooks. This problem looks like a BIOS bug. But ASUS doesn't support Linux and they wouldn't fix it.
I can not open any of the log you attached. can you please attach the logs here?
Created attachment 275903 [details] boot log
Created attachment 275905 [details] boot log verbose
(In reply to Zhang Rui from comment #1) > I can not open any of the log you attached. > can you please attach the logs here? I have reproduced it on kernel 4.16.6, and collected logs with "journalctl -b -1 --no-pager --utc --no-hostname" (and with -o verbose). But there aren't strings from system booting log. It stacks on: > tmp_crb MSFT010:00: [Firmware Bug] ACPI region does not cover the entire > command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x200] vs fed40080 > f80 > NMI: watchdog: Watchdog detected hard lookup on cpu1 > Watchdog: BUG: Soft lookup - CPU#2 stuck for 23s! If you need other information just say how to get it. I'm not very good with this new journalctl.
(In reply to Fedor Koshel from comment #4) > (In reply to Zhang Rui from comment #1) > > I can not open any of the log you attached. > > can you please attach the logs here? > > I have reproduced it on kernel 4.16.6, and collected logs with "journalctl > -b -1 --no-pager --utc --no-hostname" (and with -o verbose). But there > aren't strings from system booting log. It stacks on: what do you mean? why we can not get the strings you mentioned?
@Fedor, do you have the full log after NMI: watchdog: Watchdog detected hard lookup on cpu1
(In reply to Zhang Rui from comment #5) > (In reply to Fedor Koshel from comment #4) > > (In reply to Zhang Rui from comment #1) > > > I can not open any of the log you attached. > > > can you please attach the logs here? > > > > I have reproduced it on kernel 4.16.6, and collected logs with "journalctl > > -b -1 --no-pager --utc --no-hostname" (and with -o verbose). But there > > aren't strings from system booting log. It stacks on: > > what do you mean? why we can not get the strings you mentioned? I see this strings on the screen when I'm trying to boot the system. But in this moment system is totally stuck and I can't change tty and collect logs. And I can't find them, in journalctl log, when booting with the previous kernel. There isn't boot.log anymore, maybe I should use some special journalctl keys, but I didn't find correct combination. I can only rewrite this from the screen, or make a photo. Or I can try any commands if you recommend me.
Created attachment 276343 [details] Boot screen photo In text form: NMI watchdog: Watchdog detected hard LOOKUP on cpu 2 INFO: rcu_shed detected stalls on CPUs/tasks: 2-...0: (8 GPs behind) idle=44e/1/4611686018427387904 softirq=3072/3072 fqs=15000 (detected by 6, t=60002 jiffies, g=1882, c=18811=209)
Looks like there's an tpm_crb issue before this hang, how about disabling the Trusted Platform Module in BIOS and try again?
(In reply to Chen Yu from comment #9) > Looks like there's an tpm_crb issue before this hang, how about disabling > the Trusted Platform Module in BIOS and try again? There isn't TPM entry in BIOS. I have already tried to disable secure boot, fast boot, and other boot params but it didn't help.
First of all, for all the kernel versions that you have tries, acpi=off is mandatory for the kernel to boot, right? please check the following command lines separately and see if kernel boots 1. idle=poll 2. nox2apic 3. maxcpus=1
Created attachment 276841 [details] idle=poll boot log
Created attachment 276843 [details] idle=poll verbose log
Created attachment 276845 [details] nox2apic boot log
Created attachment 276847 [details] nox2apic verbose log
Created attachment 276849 [details] maxcpus=1 boot log
Created attachment 276851 [details] maxcpus=1 verbose log
(In reply to Zhang Rui from comment #11) > First of all, for all the kernel versions that you have tries, acpi=off is > mandatory for the kernel to boot, right? > > please check the following command lines separately and see if kernel boots > 1. idle=poll > 2. nox2apic > 3. maxcpus=1 Yes all of them worked without acpi. I've tried all of these params and no one works. The logs are in attachments.
It seems there's a issue in nouveau which hang at ioread32 in nv04_timer_read() according to the idle=poll log. Please blacklist the nouveau driver in grub and trt again.
(In reply to Chen Yu from comment #19) > It seems there's a issue in nouveau which hang at ioread32 in > nv04_timer_read() > according to the idle=poll log. Please blacklist the nouveau driver in grub > and trt again. Yes, it helps, thank you. There are still a lot of problems with the proprietary NVIDIA driver, but I can boot system.