Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
googling around I already find some (scary...) info, and tried w/o luck booting with acpi=off (boot, but it seems like hanging on every screen, then eventually go on with the next after some minutes!), noapic and nolapic.
any help appreciated
Cheers
Gianluca
On Thu, 2006-11-23 at 19:06 +0100, Gianluca Sforna wrote:
Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
try nmi_watchdog=0
another thing to try is disabling USB PS/2 emulation in the bios...
On 11/23/06, Arjan van de Ven arjan@fenrus.demon.nl wrote:
On Thu, 2006-11-23 at 19:06 +0100, Gianluca Sforna wrote:
Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
try nmi_watchdog=0
another thing to try is disabling USB PS/2 emulation in the bios...
Thanks Arjan, but this voodoo spell did not work...
What I determined so far is:
* boot stops earlier if I do not pass acpi=off * wioth acpi=off boot stops after PS/2 mouse and KB recognition * I can proceed further by either adding noapic or by switching to USB mouse _and_ keyboard ( can not find any setting in the bios for disabling PS/2 emulation).
I could do a more detailed report on monday, when I will get back at work.
FWIW, I could not boot also with the Linux Firmware Kit cdrom I have around here, not even with the "safe" mode.
I guess this is going to be tricky installation...
Gianluca Sforna <giallu <at> gmail.com> writes:
On 11/23/06, Arjan van de Ven <arjan <at> fenrus.demon.nl> wrote:
On Thu, 2006-11-23 at 19:06 +0100, Gianluca Sforna wrote:
Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
Hi,
We've bought 5 of these desktops and and have exactly the same problems.. It seems that it's not possible to boot into any Linux System at all. Besides FC, we've tried Knoppix 4.02, 5.01, Ubuntu Breezy and Edgy 6.10.
So far I've tried passing: "nolapic noapic acpi=off fb=false pci=bios" to the kernel but none of these in any combination worked.
Did you have any success?
I guess this is going to be tricky installation...
I hope there will be an installation at all :-)
greetings, dominik.
On 11/27/06, Dominink Schnitzer dominik-dated-1167177383.0e3370@schnitzer.at
We've bought 5 of these desktops and and have exactly the same problems.. It seems that it's not possible to boot into any Linux System at all. Besides FC, we've tried Knoppix 4.02, 5.01, Ubuntu Breezy and Edgy 6.10.
I have 3 of them myself... I probably had to check:
http://www.hp.com/wwsolutions/linux/certifications.html
_before_ buying them... fortunately enough the Proliant server I bought along with these clients IS listed as compatible.
I guess this is going to be tricky installation...
I hope there will be an installation at all :-)
me too...
On 11/26/06, Gianluca Sforna giallu@gmail.com wrote:
I could do a more detailed report on monday, when I will get back at work.
As promised here is a more detailed report:
right now I have attached only a USB keyboard to it; during the (PXE) boot without any kernel parameters, everything seems normal until the lines:
- Freeing initrd memory: 5802 freed - HP Compaq Laptop seriers board detected. Selecting BIOS-methods for reboots.
which should be harmless, but wrong nonetheless, since this is everithing but a laptop. Then it starts ACPI stuff:
- ACPI: bus type pci registered - PCI: Using MMCONFIG - Setting up standard PCI resources - ACPI: interpreter enabled - ACPI: using IOAPIC for interrupt routing - ACPI: PCI Root Bridge [PCI0] (0000:00) - ACPI: Assume root bridge [_SB_.PCI0] bus is 0
then hangs.
The same happens adding noapic; the only difference is a a couple of lines disappeared: ENABLING IO-APIC IRQs .. TIMER: vector=0x31 apic1=0 pin1=2 apic2=-1 pin2=-1
with nolapic still hangs at the same line, but I have more scary messages: ENABLING IO-APIC IRQs BIOS bug, IO-APIC#0 ID 1 is already used!... ... fixing up to 2. (tell your hw vendor) ..TIMER: vector=0x31 apic1=-1 pin1=-1 apic2=-1 pin2=-1 ...trying to set up timer (IRQ0) through the 8259A ... failed. ...trying to set up timer as Virtual Wire IRQ... works
with acpi=off I can reach anaconda, but the slowness is there despite I'm not using anymore PS/2 peripherals
I hope this is somewhat useful...
Gianluca Sforna <giallu <at> gmail.com> writes:
On 11/26/06, Gianluca Sforna <giallu <at> gmail.com> wrote:
with acpi=off I can reach anaconda, but the slowness is there despite I'm not using anymore PS/2 peripherals
I hope this is somewhat useful...
After a BIOS flash, my dc7700 will run w/ or w/o acpi enabled, but it feels like it is running at 300MHz or less.
Starting udev takes almost 5 minutes.
T Swan wrote: ...
After a BIOS flash, my dc7700 will run w/ or w/o acpi enabled, but it feels like it is running at 300MHz or less.
Starting udev takes almost 5 minutes.
My dc7700 works ok.
I boot with acpi=off hda=noprobe hdc=noprobe
Mogens
At 12:31 AM +0000 1/3/07, T Swan wrote:
Gianluca Sforna <giallu <at> gmail.com> writes:
On 11/26/06, Gianluca Sforna <giallu <at> gmail.com> wrote:
with acpi=off I can reach anaconda, but the slowness is there despite I'm not using anymore PS/2 peripherals
I hope this is somewhat useful...
After a BIOS flash, my dc7700 will run w/ or w/o acpi enabled, but it feels like it is running at 300MHz or less.
Starting udev takes almost 5 minutes.
Review the BIOS settings? Mine can get into a "safe mode" if it thinks booting failed, and that includes lowering the CPU clock speed as well as making the memory slower.
Gianluca Sforna wrote:
Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
googling around I already find some (scary...) info, and tried w/o luck booting with acpi=off (boot, but it seems like hanging on every screen, then eventually go on with the next after some minutes!), noapic and nolapic.
Yes, we have ordered several of these, and Fedora is painfully slow (it runs so much better on a 5 year old laptop). Originally, we thought it might be broken ACPI BIOS implementation and Linux just interacts poorly with it, but we aren't sure. With the v1.06 BIOS update the system will install without disabling ACPI, but it runs about 1/300th the speed it should. One would think with HP's commitment to Linux they wouldn't drop a %#@$!@ like this out on the market. According to a few people in HP's forums, they do not want to hear anything about Linux not running on the DC7700.
On FC6 2.16.18-2869: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m8.631s user 0m8.385s sys 0m0.203s
On FreeBSD 6.2RC2: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m0.025s user 0m0.025s sys 0m0.000s
I have tried almost every combination of clock= acpi= acpi_os_name="Spoo!" noapictimer nolapic noapic acpi=irq ... that you could imagine with no success in improving the performance. An experimental 2.6.20 (rt) kernel seemed to run about 10 times faster, but it was still to slow to be used. (Screen repaints on login are about 5 seconds)
Once, during a reinstall the rescue shell (using x86_64) on the installation CD ran at the same speed as BSD. I have not been able to see reproduce it, ever. I have installed both i386 and x86_64 versions with no real discernible difference in performance. I will assist in anyway I can, if someone wants to help track this down.
Thomas Swan wrote: ...
On FC6 2.16.18-2869: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m8.631s user 0m8.385s sys 0m0.203s
This is not my experience:
On FC6 x86_64 2.6.18-1.2869.fc6:
real 0m0.014s user 0m0.013s sys 0m0.000s
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
Mogens
Mogens Kjaer wrote:
Thomas Swan wrote: ...
On FC6 2.16.18-2869: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m8.631s user 0m8.385s sys 0m0.203s
This is not my experience:
On FC6 x86_64 2.6.18-1.2869.fc6:
real 0m0.014s user 0m0.013s sys 0m0.000s
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
I tried installing FC6 i386 on this box:
2.6.18-1.2869.fc6:
real 0m0.017s user 0m0.014s sys 0m0.003s
It also needs acpi=off in order to boot. BIOS is still v1.06.
Mogens
Mogens Kjaer wrote:
Mogens Kjaer wrote:
Thomas Swan wrote: ...
On FC6 2.16.18-2869: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m8.631s user 0m8.385s sys 0m0.203s
This is not my experience:
On FC6 x86_64 2.6.18-1.2869.fc6:
real 0m0.014s user 0m0.013s sys 0m0.000s
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
I tried installing FC6 i386 on this box:
2.6.18-1.2869.fc6:
real 0m0.017s user 0m0.014s sys 0m0.003s
It also needs acpi=off in order to boot. BIOS is still v1.06.
Mogens
All of our HP DC7700 (1.86GHz processors) are having this problem. It will boot with no options to the kernel and with acpi=off. Defaults are turned on in the bios. After about 48 hours of uptime the speed improves by a little. 7 seconds instead of 8-10 to run the little test. Both on 64bit and 32bit FC6 installations.
On 1/11/07, Thomas Swan tswan@idigx.com wrote:
Mogens Kjaer wrote:
Mogens Kjaer wrote:
Thomas Swan wrote: ...
On FC6 2.16.18-2869: # time ( let X=0; while [ $X -lt 1000 ] ; do let X=$X+1 ; done ) real 0m8.631s user 0m8.385s sys 0m0.203s
This is not my experience:
On FC6 x86_64 2.6.18-1.2869.fc6:
real 0m0.014s user 0m0.013s sys 0m0.000s
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
I tried installing FC6 i386 on this box:
2.6.18-1.2869.fc6:
real 0m0.017s user 0m0.014s sys 0m0.003s
It also needs acpi=off in order to boot. BIOS is still v1.06.
Mogens
All of our HP DC7700 (1.86GHz processors) are having this problem. It will boot with no options to the kernel and with acpi=off. Defaults are turned on in the bios. After about 48 hours of uptime the speed improves by a little. 7 seconds instead of 8-10 to run the little test. Both on 64bit and 32bit FC6 installations.
Note for myself: do not buy anything from HP in the future, unless it is not listed in http://www.hp.com/wwsolutions/linux/certifications.html
Thomas Swan wrote: ...
All of our HP DC7700 (1.86GHz processors) are having this problem. It will boot with no options to the kernel and with acpi=off. Defaults are turned on in the bios. After about 48 hours of uptime the speed improves by a little. 7 seconds instead of 8-10 to run the little test. Both on 64bit and 32bit FC6 installations.
Mine is a DC7700 with 2.13GHz processors:
Intel(R) Core(TM)2 CPU 6400 @ 2.13GHz
Could it be a cpuspeed problem? What is the "cpu MHz" line in /proc/cpuinfo say?
Mogens
On Thu, 2007-01-11 at 08:42 +0100, Mogens Kjaer wrote:
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
Mogens
I have an HP laptop that has problems. It is a dual core, not sure if yours is. Can you try it with acpi on and the noapic command?
Trever
Trever L. Adams wrote:
On Thu, 2007-01-11 at 08:42 +0100, Mogens Kjaer wrote:
Note, this is the 64 bit FC6, acpi=off.
It won't boot without acpi=off, BIOS is v1.06
Mogens
I have an HP laptop that has problems. It is a dual core, not sure if yours is. Can you try it with acpi on and the noapic command?
Trever
This is a dc7000 with a "core 2 duo" processor.
Trever L. Adams wrote: ...
Can you try it with acpi on and the noapic command?
No, it still doesn't boot.
The last line written is:
ACPI: Assume root bridge [_SB_.PCI0] bus is 0
Mogens
Mogens Kjaer wrote:
Trever L. Adams wrote: ...
Can you try it with acpi on and the noapic command?
No, it still doesn't boot.
The last line written is:
ACPI: Assume root bridge [_SB_.PCI0] bus is 0
Mogens
I've tried with the 2.6.19-1.2895 x86_64 kernel, and now it requires both acpi=off and noapic.
Without noapic it panics immediately.
Mogens
Mogens Kjaer wrote:
Mogens Kjaer wrote:
Trever L. Adams wrote: ...
Can you try it with acpi on and the noapic command?
No, it still doesn't boot.
The last line written is:
ACPI: Assume root bridge [_SB_.PCI0] bus is 0
Mogens
I've tried with the 2.6.19-1.2895 x86_64 kernel, and now it requires both acpi=off and noapic.
Without noapic it panics immediately.
upgraded the BIOS to at least 1.09 and the system requires no special parameters to install and run. However, adding hda=noprobe and hdc=noprobe significantly improves disk i/o (from 4MB/s to 65MB/s). Problem solved.
Thomas Swan wrote: ...
upgraded the BIOS to at least 1.09 and the system requires no special parameters to install and run. However, adding hda=noprobe and hdc=noprobe significantly improves disk i/o (from 4MB/s to 65MB/s). Problem solved.
Not for me. I've flashed 1.09A, the BIOS lists it as:
786E1 v01.09
and no difference for me.
Without acpi=off it stops with:
ACPI:Assume root bridge [_SB_.PCI0] bus is 0
Even with acpi=off it won't run kernel 2.6.19-1.2895.fc6 (x86_64).
Mogens
Hi,
In the school where I am responsible for Linux on the desktop we just received a number of HP DC7700's. When trying to boot it the normal way we see a message from the kernel that there is some IO-APIC IRQ problem which should be fixed in the vendor BIOS and during the ACPI initialisation the machine fully hangs. I'm using the 1.05 bios release.
After booting the computer with the kernel boot option 'acpi=off' we succeed to start anaconda. However, the I/O with the hard drive (for example formatting) was very slow. After the installation was completed I tried installing the latest rawhide kernel (2.6.20rcX).
When using this kernel the I/O speed was improved a lot! With the latest FC6 kernel (2.6.18.something) the program hdparm reported a speed of 3,5MB/sec while with 2.6.20rcX the speed was around 55MB/sec. With this setup the computer responded a lot more smoothly however I think there is still some more room for improvement.
If you need more info, just ask.
Regards,
Erik van Pienbroek
Op donderdag 23-11-2006 om 19:06 uur [tijdzone +0100], schreef Gianluca Sforna:
Hi all, I just put my hands on a shiny HP DC7700, a Core 2 Duo based PC, but I can't boot the installer because it hangs right after detecting keyboard and mouse (PS/2)
Does anyone experienced similar issues?
googling around I already find some (scary...) info, and tried w/o luck booting with acpi=off (boot, but it seems like hanging on every screen, then eventually go on with the next after some minutes!), noapic and nolapic.
any help appreciated
Cheers
Gianluca
Erik van Pienbroek wrote: ...
After booting the computer with the kernel boot option 'acpi=off' we succeed to start anaconda. However, the I/O with the hard drive (for example formatting) was very slow.
I have set SATA emulation to IDE, and boot with
acpi=off hda=noprobe hdc=noprobe
then the IO speed is OK:
# hdparm -t -T /dev/sda
/dev/sda: Timing cached reads: 4460 MB in 2.00 seconds = 2230.11 MB/sec Timing buffered disk reads: 208 MB in 3.02 seconds = 68.89 MB/sec
The "hda=noprobe hdc=noprobe" needs to be added manually to grub.conf after installation.
Mogens
Op vrijdag 12-01-2007 om 11:54 uur [tijdzone +0100], schreef Mogens Kjaer:
Erik van Pienbroek wrote: ...
After booting the computer with the kernel boot option 'acpi=off' we succeed to start anaconda. However, the I/O with the hard drive (for example formatting) was very slow.
I have set SATA emulation to IDE, and boot with
acpi=off hda=noprobe hdc=noprobe
then the IO speed is OK:
Hi,
I just tested out your tip and this works great with FC6!
Regards,
Erik van Pienbroek
devel@lists.stg.fedoraproject.org
-
Arjan van de Ven -
Dominink Schnitzer -
Erik van Pienbroek -
Gianluca Sforna -
Mogens Kjaer -
T Swan -
Thomas Swan -
Tony Nelson -
Trever L. Adams