Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 20 Feb 2010 04:40:48 +0100 (CET)
From:      kama <kama@pvp.se>
To:        Doug Poland <doug@polands.org>
Cc:        freebsd-proliant@freebsd.org, freebsd-acpi@freebsd.org
Subject:   Re: ProLiant DL385 (G1) hangs on boot w > 4GB RAM
Message-ID:  <20100220021842.U94967@ns1.as.pvp.se>
In-Reply-To: <4B57D402.1070502@polands.org>
References:  <20100120230733.GA19319@polands.org> <4B57B67A.5080401@gmail.com> <4B57D402.1070502@polands.org>

next in thread | previous in thread | raw e-mail | index | archive | help

On Wed, 20 Jan 2010, Doug Poland wrote:

> On 2010-01-20 20:05, Mark Atkinson wrote:
> > On 1/20/2010 3:07 PM, Doug Poland wrote:
> >> Hello,
> >>
> >> I've got an HP DL386(G1) with 4GB RAM.  The box boots normally until I
> >> added 4GB more of RAM.  In this case the RAM came from another working
> >> DL385.  I've tried 6GB and 8GB RAM and both scenarios hangs at the same
> >> spot.
> >>
> >> Any suggestions?
> >>
> >> Copyright (c) 1992-2010 The FreeBSD Project.
> >> Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
> >> 	The Regents of the University of California. All rights reserved.
> >> FreeBSD is a registered trademark of The FreeBSD Foundation.
> >> FreeBSD 8.0-STABLE #0: Tue Jan 19 12:46:14 CST 2010
> >>      root@ararat.polands.org:/usr/obj/usr/src/sys/GENERIC amd64
> >> Timecounter "i8254" frequency 1193182 Hz quality 0
> >> CPU: AMD Opteron(tm) Processor 252 (2605.92-MHz K8-class CPU)
> >>    Origin = "AuthenticAMD"  Id = 0x20f51  Stepping = 1
> >>    Features=0x78bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2>
> >>    Features2=0x1<SSE3>
> >>    AMD Features=0xe2500800<SYSCALL,NX,MMX+,FFXSR,LM,3DNow!+,3DNow!>
> >> real memory  = 8589934592 (8192 MB)
> >> avail memory = 8104611840 (7729 MB)
> >> ACPI APIC Table:<HP     00000083>
> >> FreeBSD/SMP: Multiprocessor System Detected: 2 CPUs
> >> FreeBSD/SMP: 2 package(s) x 1 core(s)
> >>   cpu0 (BSP): APIC ID:  0
> >>   cpu1 (AP): APIC ID:  1
> >> ACPI Warning: Invalid length for Pm1aControlBlock: 32, using default 16 20090521 tbfadt-707
> >> ACPI Warning: Invalid length for Pm1bControlBlock: 32, using default 16 20090521 tbfadt-707
> >> MADT: Forcing active-low polarity and level trigger for SCI
> >> ioapic0<Version 1.1>  irqs 0-23 on motherboard
> >> ioapic1<Version 1.1>  irqs 24-27 on motherboard
> >> ioapic2<Version 1.1>  irqs 28-31 on motherboard
> >> ioapic3<Version 1.1>  irqs 32-35 on motherboard
> >> ioapic4<Version 1.1>  irqs 36-39 on motherboard
> >> kbd1 at kbdmux0
> >> acpi0:<HP A05>  on motherboard
> >> acpi0: [ITHREAD]
> >> acpi0: Power Button (fixed)
> >> Timecounter "ACPI-fast" frequency 3579545 Hz quality 1000
> >> acpi_timer0:<32-bit timer at 3.579545MHz>  port 0x908-0x90b on acpi0
> >> pcib0:<ACPI Host-PCI bridge>  on acpi0
> >> pci0:<ACPI PCI bus>  on pcib0
> >>
> >>
> >> and here it hangs.
> >>
>  >
>  > You could probably safely post this on the acpi list.   First thing
>  > to do is test a -current snapshot and see if it boots.   Also, my
>  > DL380-G3s had about 4 different ACPI options in the BIOS, only one
>  > would let it boot without hanging.
>  >
> Thanks, I am posting this to -acpi as well.  As an experiment, I put all
> 8GB RAM in the other server, and it won't boot either.  It hangs at the
> same point.  While I await a response, I think I'll make sure all the
> latest HP patches for BIOS and such are loaded.
>

On Wed, Jan 27, 2010 at 05:33:30PM +0100, Eirik ?verby wrote:
>> Hi,
>>
>> I understand you're facing problems booting your DL385s on FreeBSD 8.
>> We hit the same problem today, and found that the solution (or rather,
>> workaround) is to boot in verbose mode. I'm not sure if this causes
>> any performance regressions (it seems not), but to make it permanent,
>> add -v to /boot.config (that is in the root directory).
>>
>> Feel free to forward this to the -acpi list and start an investigation
>> - it baffles me that this should magically make FreeBSD boot.
>>
>> FYI, 7.2 works fine on the same hardware. And with this hack, 8.0
>> works equally well.
>>
> Hi Eirik,
>
> Wow, that workaround allows these boxes to boot now, with more than 4GB
> RAM.  Thanks for the tip!
>
> Interestingly enough, the boxes HANG with < 4GB RAM and boot verbose
> selected.  Go figure?

Hi!

I have noticed this too. (And reported it in STABLE a couple of month
ago, which unfortunately gave me next to nothing in help)

But I have been diging in to it a bit.

in /boot/loader.conf I have now added this.
----- 8< -------
# Disable asf on the network card (bge) (causes freezes)
hw.bge.allow_asf="0"

hw.acpi.verbose="1"
debug.acpi.max_threads="1"
debug.acpi.disabled="timer"

debug.acpi.layer="ACPI_ALL_COMPONENTS"
debug.acpi.level="ACPI_LV_VERBOSE,ACPI_LV_VERBOSITY3,ACPI_LV_VERBOSITY1,ACPI_LV_ALL_EXCEPTIONS"
-------- >8 -----

and then choose to boot with verbose logging will make my system always to
boot up. If I leave out these option it will freeze with verbose boot.
booting without verbose it will hang it whatever I choose.

If I reboot the system quite alot it will eventually boot up without any
of these options set. But thats maybe once every 50 times.

I dont know how to proceed and give any more info. Since it will freeze
heavily just in the beginning and I cant get it to debugging mode.

I have tried to add more printf's where I think it makes it to hang. But I
have currenly only access to ILO through ssh. So I cant get a proper
printout when it is actually stopping. But it seems to be within the timer
probing/calculation/init part, but cant confirm it for sure.

And if I choose boot without ACPI it will also always boot. So I presume
it is within the ACPI code.

This happens for me on both 7.2 and 8.0. And its only DL385 where I have
noticed this behavour.

My system is using 3G of RAM and running i386 build of the OS.

I get a warning in the logs when booting the system up:

ACPI Warning: Invalid length for Pm1aControlBlock: 32, using default 16
20090521 tbfadt-707
ACPI Warning: Invalid length for Pm1bControlBlock: 32, using default 16
20090521 tbfadt-707

Dont know if that has anything todo with it. I had the same messages on
the DL380 G3. Those servers where coredumping and in a constant reboot
loop after an upgrade from 7.2 to 8.0. But changing the "OS" in the BIOS
from 'Linux' to 'Other OS' made the system to boot. Still getting those
warning messages.

ACPI Warning: Invalid length for Pm1aControlBlock: 32, using default 16
(20100121/tbfadt-707)
ACPI Warning: Invalid length for Pm1bControlBlock: 32, using default 16
(20100121/tbfadt-707)

On DL385 you cant choose the "OS" in BIOS.

/Bjorn




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100220021842.U94967>