Project

General

Profile

Bug #2736

kernel panics on acpi_timer_probe function

Added by cnb almost 3 years ago. Updated almost 3 years ago.

Status:
New
Priority:
High
Assignee:
-
Category:
ACPI
Target version:
Start date:
11/17/2014
Due date:
% Done:

0%


Description

Booting with acpi module enable and hpet disabled in loader.conf results in a kernel panic, I'm unable to type anything in ddb as the the system is frozen
The hardware is a gobook vr-2 general dynamics itronix IX605 bios version 124. I'll update to bios version 125 and update the ticket if needed.

20141116_201206.jpg View - kernel panic (3.46 MB) cnb, 11/17/2014 11:05 AM

20141118_114403.jpg View (3.64 MB) cnb, 11/18/2014 09:21 AM

bios.tar.gz (53.7 KB) cnb, 11/18/2014 10:30 AM

20141118_114403.jpg View (3.64 MB) cnb, 11/19/2014 04:17 AM

DSC_0187.JPG View - new kernel panic (1.38 MB) cnb, 11/19/2014 04:19 AM

20141119_130057.jpg View (2.98 MB) cnb, 11/19/2014 08:05 AM

20141119_130838.jpg View (2.94 MB) cnb, 11/19/2014 08:11 AM

dmesg.boot (18.1 KB) cnb, 11/19/2014 03:14 PM

dmesg.boot (81.5 KB) cnb, 11/20/2014 05:13 AM

mptable (2.78 KB) cnb, 11/20/2014 04:32 PM

mptable.txt View (3.03 KB) cnb, 11/20/2014 04:32 PM

dmesg.boot (97.4 KB) cnb, 11/21/2014 09:52 AM

dmesg.boot (120 KB) cnb, 11/21/2014 10:30 AM

History

#1 Updated by swildner almost 3 years ago

On Mon, 17 Nov 2014 20:14:54 +0100,
<> wrote:

> Booting with acpi module enable and hpet disabled in loader.conf results
> in a kernel panic, I'm unable to type anything in ddb as the the system
> is frozen
> The hardware is a gobook vr-2 general dynamics itronix IX605 bios
> version 124. I'll update to bios version 125 and update the ticket if
> needed.

Have you tried disabling acpi_timer too like proposed on the IRC channel?

Please try with 'timer' added to debug.acpi.disabled, e.g.
debug.acpi.disabled="hpet timer"

Sascha

#2 Updated by swildner almost 3 years ago

Also, is it possible to see a picture/screenshot of the panic?

Best regards,
Sascha

#3 Updated by cnb almost 3 years ago

Hello,
I have attached a screenshot of the panic in the bug description.

Bests

#4 Updated by swildner almost 3 years ago

Sorry, I overlooked the attachment.

I have some fixes for acpi_timer.c

Could I build you an ISO/IMG with them and you try booting with it?

Sascha

#5 Updated by cnb almost 3 years ago

Adding the debug.acpi.disabled="hpet timer" fixes the timer issue and boots with acpi enabled, there are some errors about ACPI in the log.
I have attached a screenshot.

#6 Updated by swildner almost 3 years ago

Hi,

can you rebuild the kernel (actually, just the acpi module should be enough) with this patch?

http://leaf.dragonflybsd.org/~swildner/0001-kernel-acpi-Convert-acpi_timer-to-using-ACPICA-funct.patch

Please test if this helps with the acpi_timer panic you were seeing. If it still panics, another screenshot would be nice.

Thanks,
Sascha

#7 Updated by swildner almost 3 years ago

Also, could you tar the output of acpidump -b (it creates several .dat files) and attach it here?

Thanks,
Sascha

#8 Updated by cnb almost 3 years ago

Thanks Sacha, I'll try the patch and let you know.
I have attached the acpidump -b contents, I'm trying to recompile dsdt.dsl using asl but there are several errors.
I was replacing things like CMSI = Arg0 to Store(Arg0,CMSI) maybe I'm mistaken to fix the errors.

Bests

#9 Updated by swildner almost 3 years ago

This patch:

http://leaf.dragonflybsd.org/~swildner/gobook_dsdt.diff

should make the DSDT compile. Beware, while i'm pretty confident about the CAPB change, the Range Minimum/Length one is nothing more than an educated guess.

Good luck,
Sascha

#10 Updated by swildner almost 3 years ago

Also, beware that the patch is against a dsdt.dsl which was disassembled using master's iasl(8) program which recently introduced C-style syntax for operators. If yours is disassembled using an older iasl, that CAPB change should probably be "And (CAPB, 0xFFFFFFFC, CAPB)".

Regards

#11 Updated by cnb almost 3 years ago

Thanks I'll try and let you know.

Bests

#12 Updated by cnb almost 3 years ago

trying to use the fixed dsdt.aml failed with the message cannot load /boot/kernel/acpi_dsdt.aml.
I have applied your patch and commented the disable hpet timer, but it still panics I have attached a screenshot of this new kernel panic.

#13 Updated by cnb almost 3 years ago

Here is the screenshot

#14 Updated by swildner almost 3 years ago

* Have you copied to dsdt.aml to /boot/kernel/acpi_dsdt.aml? (Note: different source and target names)

* Regarding the panic: Could you try with my patch and with _only_ hpet disabled? It was meant to fix the acpi_timer issue for which you previously had to also add "timer" to the disabled ACPI parts.

Regards,
Sascha

#15 Updated by cnb almost 3 years ago

Yes, I copied to that path and modified loader.conf to use that file.
Regarding timer if I boot without acpi the timer goes twice as fast. I have updated to master, I'll patch and recompile.

#16 Updated by cnb almost 3 years ago

The kernel panic is produced when I remove the disable hpet timer line from loader.conf.

#17 Updated by cnb almost 3 years ago

booted with acpi.debug.disable="hpet" with patch applied causes kernel panic. screenshot attached

#18 Updated by cnb almost 3 years ago

Sometimes booting with acpi enabled produces the following system freeze, but I do a cold start it will go away

#19 Updated by cnb almost 3 years ago

  • File dmesg.boot added
  • Target version changed from 4.0.x to 3.8.0

I have attached boot messages

#20 Updated by cnb almost 3 years ago

boot verbose attached, I was unable to boot with your patch the acpi_ec0 is loaded with your patch but after that the system freezes

#21 Updated by swildner almost 3 years ago

  • Target version changed from 3.8.0 to 4.0.x

Could you give me a picture of that freeze with the patch?

Regards,
Sascha

#22 Updated by swildner almost 3 years ago

Can you give me 2 more things:

* Do "mptable -raw" (it will create /tmp/mptable)

* Do "mptable -verbose >/tmp/mptable.txt"

Please attach both /tmp/mptable and /tmp/mptable.txt to this issue.

Thanks,
Sascha

#23 Updated by cnb almost 3 years ago

Here is the output from mptable.

#24 Updated by cnb almost 3 years ago

swildner wrote:
> Could you give me a picture of that freeze with the patch?
>
> Regards,
> Sascha

Hi, the only thing that is different from previous boot is that it freezes in this point acpi_ec0: <Embedded Controller: GPE 0x17> port 0x66,0x62 on acpi0

#25 Updated by cnb almost 3 years ago

Dmesg boot verbose with the following loader.conf

vfs.root.mountfrom="hammer:serno/SB228DGGKDZL1H.s1d"
if_iwn_load="YES"
iwn4965fw_load="YES"
if_iwi_load="YES"
legal.intel_iwi.license_ack=1
#hw.ioapic_enable=0
hw.psm.synaptics_support=1
#hw.acpi.osname="Windows 2001"
debug.acpi.disabled="hpet timer"
#acpi_dsdt_load=yes
hw.ata.atapi_dma=0
sound_load="YES"
snd_hda_load="YES"
#hint.apic.0.disabled=1
boot_verbose="YES"

#26 Updated by cnb almost 3 years ago

acpi.ko debug build
debug.acpi.layer="ACPI_EC"
debug.acpi.level="ACPI_LV_ALL"

Also available in: Atom PDF