[Bug 765378] New: Boot without "acpi=off" hangs. Machine doesn't power off with "acpi=off" used.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c0 Summary: Boot without "acpi=off" hangs. Machine doesn't power off with "acpi=off" used. Classification: openSUSE Product: openSUSE 12.1 Version: Final Platform: x86-64 OS/Version: openSUSE 12.1 Status: NEW Severity: Normal Priority: P5 - None Component: Kernel AssignedTo: kernel-maintainers@forge.provo.novell.com ReportedBy: paternot@snap.com.br QAContact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0) Gecko/20100101 Firefox/12.0 OpenSuse 12.1 cannot be installed without passing "acpi=off" to the installer. After installed, and fully updated, it can't be used without this parameter either. Any attempt to do so results in a locked machine. Passing "acpi=off" solves this problem - but creates another. I can't shutdown the computer. It goes all the way, until showing "the system will be halted intermediately". And then... nothing. It doesn't turns itself off. I got the same response with systemd and systemvinit. Also, "halt", "halt -p", "init 0" and "shutdown" gave me the same result. Reproducible: Always Steps to Reproduce: 1. Install OpenSuse 12.1 (x86_64) 2. Do a full update 3. Shut down the system. Actual Results: The system is halted, but not turned off. Expected Results: The system should turn off at the end of the process. I did a minimal install, text mode. The drive is a SSD Corsair, but I can reproduce this with a Seagate, 500GB. The motherboard is an Intel, model DP67BA, step 3. This is the fourth motherboard of this model, and the only one with this problem. The only difference is the BIOS. This one uses the version "BAP6710H.86A.0077.2012.0316.1103" - which is the newest version from Intel. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c1 Jeff Mahoney <jeffm@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jeffm@suse.com --- Comment #1 from Jeff Mahoney <jeffm@suse.com> 2012-06-04 22:04:08 EDT --- The bug isn't that the machine can't shutdown. It is common for modern hardware to not have the ability to power off via software without ACPI enabled. The real bug is why the install can't proceed without acpi=off. Where does it lock up? Does it crash? Can you capture any sort of log? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c2 --- Comment #2 from Marcelo Rodrigues <paternot@snap.com.br> 2012-06-05 15:26:56 UTC --- Created an attachment (id=493684) --> (http://bugzilla.novell.com/attachment.cgi?id=493684) Picture of the machine locked up. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c3 --- Comment #3 from Marcelo Rodrigues <paternot@snap.com.br> 2012-06-05 15:27:34 UTC --- I can't send a log, since the machine locks up even before it finds the mouse (it's optic, and the led doesn't turns on again). I took an picture, using the text based installation. The machine was already locked up at this point. Shortly after I took this picture, it restarted by itself. I used a DVD, final release. OpenSuse 12.1 x86_64. Language: english. Install mode: text based. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c Jeff Mahoney <jeffm@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|kernel-maintainers@forge.pr |trenn@suse.com |ovo.novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c4 --- Comment #4 from Marcelo Rodrigues <paternot@snap.com.br> 2012-06-08 15:57:06 UTC --- Update: I was in a hurry to put this machine in production, so I swapped the motherboards. I used another one, identical but for the BIOS version - which is older. The CPU was a different one too: The machine with problems was using an i3 2100, and the machine running fine was using an i5 2500. Both CPUs are supported by both the motherboards, according to the Intel website. The problem with ACPI follows the i3 around: now the older motherboard cannot use ACPI, and the new one (with the i5) can. I disabled execution bit, VT technology and hyperthreading on the machine with the i3. No use, the behavior is still the same. The weird machine is using a single memory, and the good one is using four of them. But, according to the manual, this configuration is supported. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
I was in a hurry to put this machine in production, so I swapped the motherboards. I used another one, identical but for the BIOS version - which is older. That's a great hint. Is it possible to retrieve output of the "acpidump" command on the machine with
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c5 Thomas Renninger <trenn@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P3 - Medium Status|NEW |ASSIGNED --- Comment #5 from Thomas Renninger <trenn@suse.com> 2012-06-11 08:30:26 UTC --- This looks like a hard one: No oops/backtrace. The NMI watchdog should be enabled by default and kick in, but it does not (should kick in after 10s iirc, but I expect you waited that long...). the old (working) and the new (not booting when acpi is enabled) BIOSes. Just pass acpi=off on the affected platform, the tool should still be able to retrieve the ACPI BIOS tables. Hm, best also attach dmidecode output of both BIOSes. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c Thomas Renninger <trenn@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO InfoProvider| |paternot@snap.com.br -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c6 --- Comment #6 from Marcelo Rodrigues <paternot@snap.com.br> 2012-06-11 14:56:02 UTC --- Created an attachment (id=494335) --> (http://bugzilla.novell.com/attachment.cgi?id=494335) BIOS dump I'm sending the BIOS dump of both motherboards. The file "dp67ba-i3" is of the OLD board, with an i3 now. The file "dp67ba-i5" is of the NEW board, with an i5 now. It is important to remember that the ACPI problem follows the i3 around. Wichever board I put it in, the problem arises. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c7 --- Comment #7 from Marcelo Rodrigues <paternot@snap.com.br> 2012-06-11 15:01:02 UTC --- Update: I got some time with these machines, in the weekend. Testing with the ACPI parameters got me a functioning setup: acpi=strict nohz=off If I use onle "acpi=strict", the machine locks up. If I use only "nohz=off", the machines boots ok - but no frequency scaling nor the power off works. With both the system works great. At least until now. The clock changes, the heat is about right and the machine turns itself off - as it should be. I couldn't test the installation with this parameters - but it runs great. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c8 Thomas Renninger <trenn@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED CC| |youquan.song@intel.com InfoProvider|paternot@snap.com.br | --- Comment #8 from Thomas Renninger <trenn@suse.com> 2012-06-15 11:50:57 UTC --- Sorry for the late response. Quick summary: The machine is running fine with the i5 2500 CPU, but if you put in a i3 2100 CPU the machine does not boot unless acpi=off is added. A very nice hint comes from comment #7 where you describe the acpi=strict nohz=off boot parameter workarounds. Hm, that would mean it's not the intel idle driver as this one should still kick in in 12.1, even with acpi=off added. Not 100% sure whether nohz=off prevents cpuidle subsystem to initialize, but I guess not. Adding Intel, this probably still is a mainline issue. It could get verified by trying the latest -vanilla flavor kernel of the day from here: http://download.opensuse.org/repositories/Kernel:/HEAD/standard/x86_64/ You might want to try to blacklist acpi-cpufreq driver. Depending on how intrusive it is for you to provide the data, this might help for further investigations: /proc/cpuinfo # for both CPUs dmesg with working "acpi=strict nohz=off" boot params. Hopefully Intel has an idea. Other related params to further pin things down (for me it looks like a timer issue): clocksource=tsc or clocksource=hpet (whatever is not used by default) hpet=disable or tsc=off (switch off whatever is used by default) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=765378 https://bugzilla.novell.com/show_bug.cgi?id=765378#c9 Thomas Renninger <trenn@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #9 from Thomas Renninger <trenn@suse.com> 2013-04-18 13:59:43 UTC --- This bug is rather old, I should have set to needinfo. Please re-open if this still happens with latest code base. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com