slow boot on intel DG45ID

[amir]

Binary package hint: linux-image-2.6.27-4-generic

I have an intel DG45ID board with latest intrepid kernel 2.6.27 and I see about 17 seconds lost on boot. here is the part of the dmesg:

[ 0.977380] pci 0000:00:02.0: Boot video device
[ 0.980826] Switched to high resolution mode on CPU 1
[ 0.984004] Switched to high resolution mode on CPU 0
[ 8.976007] pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
[ 16.976007] pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
[ 16.976351] isapnp: Scanning for PnP cards...
[ 17.329570] isapnp: No Plug & Play device found

I've attached complete dmesg.

thank you,
amir

[Chris Dumont (chrisdumont)]

Hi,

I also have the DG45ID motherboard. I found that if I disabled "Legacy support" (I can't remember the exact name and I don't want to interrupt my OS install running at the moment) in the USB devices section of the BIOS then the kernel would boot normally.

Hope this helps,
Chris D

[amir (amir-in)]

Hi Chris,

thanks for the help, it worked great, the entire boot takes 15 seconds!!! this is amazing. there is one more problem I have with this board is that about 1 in 10 boots the machine halts in the middle of the boot and I need to do hard restart. I suspect that it is something to do with ipv6 from this:

Sep 28 16:21:35 strangelove kernel: [36409.656916] ip6_tables: (C) 2000-2006 Netfilter Core Team
Sep 28 16:21:37 strangelove exiting on signal 15

(maybe I need to open a new bug)
I'm not closing this bug beacuse I think it still needs to be addressed in the kernel level and not in the user level.

thank you,
Amir

[Mikko Rantalainen (mira)]

I, too, have DG45ID motherboard and disabling USB Legacy support in the BIOS is not a great choice because then you cannot use keyboard in the boot loader.

Whatever causes about 8 second timeout before kernel message "pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001" needs to be fixed (note that that timeout is executed twice in row because of the two EHCI controllers). Everything seems to work okay regardless of this error so perhaps the whole timeout could be skipped or reduced to nearly zero on DG45ID (and probably on all G45/P45 based intel motherboards?).

[Mikko Rantalainen (mira)]

With the same DG45ID the grub seems to have a long delay in the boot before displaying anything at all. I haven't figured out if this is because of USB legacy support, too.

[Mikko Rantalainen (mira)]

Tested with latest intrepid ibex last weekend: After the BIOS ends, it takes about 8-15 seconds before "grub loading stage..." text is displayed on the screen. The "USB legacy support" setting in the BIOS does not make any difference. There's no such delay when using NTLDR (Windows XP boot loader) only and starting Windows XP so grub is doing something stupid which causes additional delays to boot process.

[Alan (mrintegrity)]

I am considering buying this board. Perhaps someone could test this with different boot loaders like lilo?

[Chris Dumont (chrisdumont)]

On Tue, 2008-10-14 at 14:20 +0000, Mikko Rantalainen wrote:
> I, too, have DG45ID motherboard and disabling USB Legacy support in the
> BIOS is not a great choice because then you cannot use keyboard in the
> boot loader.
>
True, but I find the tradeoff worth it. I haven't yet had a need to
interrupt GRUB. If it was essential I would just adjust the BIOS to let
me do that.

> Whatever causes about 8 second timeout before kernel message "pci
> 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001" needs to
> be fixed (note that that timeout is executed twice in row because of the
> two EHCI controllers). Everything seems to work okay regardless of this
> error so perhaps the whole timeout could be skipped or reduced to nearly
> zero on DG45ID (and probably on all G45/P45 based intel motherboards?).
>
The delay is purposely coded in the kernel to draw your attention to the
"BIOS bug". You could fairly easily strip that out if you're interested
in making yourself a custom kernel. Or, if you petition the right people
that delay could be removed (for every motherboard or just for
boards/chipsets known to work okay even with the "bug").

[Chris Dumont (chrisdumont)]

On Wed, 2008-10-22 at 14:00 +0000, Mikko Rantalainen wrote:
> Tested with latest intrepid ibex last weekend: After the BIOS ends, it
> takes about 8-15 seconds before "grub loading stage..." text is
> displayed on the screen. The "USB legacy support" setting in the BIOS
> does not make any difference. There's no such delay when using NTLDR
> (Windows XP boot loader) only and starting Windows XP so grub is doing
> something stupid which causes additional delays to boot process.
>
I experience no such delay. I'm using Debian with 2.6.26. On other Intel
boards I've noticed strange behaviour if I didn't use the first SATA
port.

[Mikko Rantalainen (mira)]

I'm running DG45ID with BIOS version 0079 (flashed it about two weeks ago, barely days before version 0081 came out). I'll test with BIOS version 0081 (perhaps later today) to see if it fixes the grub delay issue. The version 0081 changelog says "Added patch for the issue of GATE A20 can not open." [sic] which sounds like it could make a difference for grub.

If BIOS version 0081 does not fix the delay, I think the best choice would be to request support from Intel. If the "BIOS bug" reported by the linux kernel is real, then Intel should fix the issue!

If I remember correctly I have two Samsung 750 GB SATA disks connected to SATA1 and SATA2 and Samsung DVD+RW drive connected to SATA3. The BIOS seems to always wait for both HDDs to spin up at the start even though I have no hard drive delay set in the BIOS.

The board also supports UEFI boot and I think ELILO should work with it. I have no previous experience with ELILO and I need to dual boot to Windows so I have to figure out if that's possible.

[Offtopic notes: if you plan to get DG45ID and put Windows XP in it, you'll need to slipstream SP3 and install SATA drivers during the install from the floppy drive or slipstreamed on the install CD if you plan to use AHCI or RAID support with it. In the end, I think that DG45ID is a nice board if you can live without the legacy features such as IDE, PS/2 and the friends]

[Mikko Rantalainen (mira)]

Updates to my last comment:
I'm now running BIOS version 0085 (never could successfully install 0081) and the grub causes no longer delays during the boot plus the BIOS itself boots much faster. I have two Samsung 750GB SATA drives connected as SATA0 and SATA1 and SATA2 is a DVD-RW drive.

The board works otherwise fine but there's long pause in the boot process with the same messages:
[ 0.937109] pci 0000:00:02.0: Boot video device
[ 8.936007] pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
[ 16.936007] pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001

I tried to report this issue to intel but I could not find any official way to report bugs. I submitted the bug report to visitor survey, and I can only hope that it will be forwarded to the correct people.

[Mikko Rantalainen (mira)]

I've contacted intel about the BIOS handoff issue.

[Mikko Rantalainen (mira)]

I have done some correspondence with Intel and it seems that the official response is that DG45ID is not (at least currently) supported on Linux and Intel is not going to fix this issue. The fact that this board works against the EHCI specification published by Intel (http://www.intel.com/technology/usb/ehcispec.htm page 121 and page 122) does not matter and all I was able to extract from Intel about this issue was "The link you provided does not relate specifically to this board and it is just general information for EHCI Controller." The fact that linux kernel tries very hard to follow that specification did not matter.

So the end result is that either you compile your own kernel with smaller timeout or try to get workaround for this issue in linux kernel some other way. Do not expect Intel to help in any way. The other response I got was "We cannot provide further information on the BIOS other than the information that is publicly available." I think this means that Intel is not going to say how the EHCI hand-off is even supposed to work!

[comcute (comcute)]

Intel DG45FC board has the same problem, it's really frustating if it's true, what Mikko Rantalainen previously posted.

[Mikko Rantalainen (mira)]

I haven't got any evidence to suggest that Intel is going to fix this issue. Notice that both DG45ID and DG45FC are officially unsupported with Linux according to Intel (http://www.intel.com/support/motherboards/desktop/sb/CS-008326.htm).

[Rubblebuster (erik-fornoff)]

It's not only a problem of these two motherboards!
In fact it seems to be a more generic problem, since on my Toshiba Satellite L300-1CM Notebook ( http://de.computers.toshiba-europe.com/innovation/product/Satellite-L300-1CM/1060892/toshibaShop/false/ ) exactly the same problem occurs...

[Francis Liu (fishzle)]

Indeed it's not just Toshiba and Intel. I've got Intrepid 8.10, kernel 2.6.27-11-generic running on a ASUS P5B Deluxe and I get an 8 second delay

[ 1.252994] io scheduler cfq registered (default)
[ 9.252214] pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001

[Mikko Rantalainen (mira)]

Would the patch in http://bugzilla.kernel.org/show_bug.cgi?id=12148 be acceptable in Ubuntu kernel? It adds a new kernel command line parameter ehci_ho_to (EHCI hand-off timeout) which defaults to 5000 ms in the patch. The patch can be seen at http://bugzilla.kernel.org/attachment.cgi?id=20442&action=view

The ehci_ho_to kernel parameter could be used with the motherboards that suffer from this bug to improve the speed of the boot process. It should be noted that the timeout is not defined in any specification so there's no safe default. The timeout of 5000 ms in the current kernel just seems to work for all hardware. Some (most?) hardware could use much smaller timeout.

[ieL3kudohg (iel3kudohg-deactivatedaccount)]

A new BIOS for this board was released on 09/03/2009. Does the problem still exist with that release?

[AmenophisIII (amenophisiii)]

i didnt look especially for it when i updated, but im pretty sure, that it did not change anything in respect to this bug

[comcute (comcute)]

DG45FC BIOS version 0093 with date 4/10/2009 didn't fix this problem, but POST process is much quicker now.

[Mikko Rantalainen (mira)]

Has anybody tried new BIOS version 0095? The release notes say "Fixed issue with USB emulation type." - whatever that means... Be warned though, I had issues with 0093 BIOS update failing and I had to resort to BIOS recovery (which did work) to get the system to boot again. I'll try upgrading to 0095 when I have time reserved to do the BIOS recovery again, if needed.

[Oleksij Rempel (olerem)]

BIOS version 0095 do not solve this problem

[tamil (tamilpuyal-28)]

really i am very disappointed of this board...............WORST LATEST TECHNOLOGY INTEL BOARD DG45ID ..........very very worst mouse and keyboard does not work ...................

[Oleksij Rempel (olerem)]

BIOS version 0101 do not solve this problem too. I guess, there will no
fix from Intel. Kernel people will not give up too...

[pcpbslack (pcpbslack)]

I can confirm that this bug affects Ubuntu 9.04 running on Intel DP43TF motherboard (with the latest BIOS driver installed).

kernel: [ 0.909820] io scheduler cfq registered (default)
kernel: [ 8.908008] pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
kernel: [ 16.908008] pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
kernel: [ 16.908140] pci 0000:01:00.0: Boot video device

Hope this bug is solved soon.

[caduqued (caduqued)]

Hi There,

I also can confirm that the bug is present in the Intel DQ45CB motherboard. I am using Slamd64, and the message is exactly as posted for so many people here:

pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001

My boot process hags up for 2-3 sec, and after that it is normal, but with those two lines in dmesg.

A bit frustation with this motherboard and so much trouble to get it bug-free, Any ideas so far?

[Richard Huddleston (rhuddusa)]

i too have the DQ45CB and can say that disabling USB legacy support solved the problem for me, with the pre 85 bios, that one bricked my motherboard :(

[A. Campbell (allenc28)]

I confirm this problem with the Intel DP45SG motherboard. Using latest BIOS version 0118.

pci 0000:00:1a.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001
pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001

Intel should correct the non-compliant ECHI behavior of this line of products.

[baeza (baeza)]

Kernel 2.6.30 (stable is 2.6.30.5) has reduced the waiting time for BIOS response from present 5000 ms to 1000 ms, so BIOS EHCI non-compliant only delay boot in less than 2 seconds. Please update pci-quirks.c as soon as properly tested.

[Eric R Drayer (ericdrayer)]

I just bought the intel DP43TF.
I am using ubuntu 9.04 amd64 and radeon 9000 video card.
I was using an abit ip35v mother board with the radeon 9000 with no Blender or mouse problems.
I have the ehci bug as above now.
I have other problems as well.
Could this board be interfering with X11 and or the open source radeon driver?
The mouse pointer icon is a 1 inch square and blender has no text (the entire interface has all the text removed).
Curiously the screen shot actually shows the correct icon.
I have tried other mice.
I have the latest updates.
I have a debian install (same hard drive) with the same problem.
I have run the 9.04 install dvd so that ubuntu is runnning of the dvd and had the same problems.
I tried to use karmic on the last board but after the install and grub there was no way to get video or use terminal(recovery).
Even if I had been able to get a recovery prompt I do not know how to change the configuration to use a different driver as xorg.conf has been depreciated.
The abit board has died... bad bios maby.

[Eric R Drayer (ericdrayer)]

I just bought the intel DP43TF.
I am using ubuntu 9.04 amd64 and radeon 9000 video card.
I was using an abit ip35v mother board with the radeon 9000 with no Blender or mouse problems.
I have the ehci bug as above now.
I have other problems as well.
Could this board be interfering with X11 and or the open source radeon driver?
The mouse pointer icon is a 1 inch square and blender has no text (the entire interface has all the text removed).
Curiously the screen shot actually shows the correct icon.
I have tried other mice.
I have the latest updates.
I have a debian install (same hard drive) with the same problem.
I have run the 9.04 install dvd so that ubuntu is runnning of the dvd and had the same problems.
I tried to use karmic on the last board but after the install and grub there was no way to get video or use terminal(recovery).
Even if I had been able to get a recovery prompt I do not know how to change the configuration to use a different driver as xorg.conf has been depreciated.
The abit board has died... bad bios maby.

[executorvs (executorvs)]

I'm encountering the same error on my acer timeline 5810TZ in lucid. it adds about 5 seconds to my boot.

[Remy Damour (remy-damour)]

I'm using an acer 2930 and this bug is still there (and really annoying) => +7 secs every time I turn on the computer
it makes linux look like an old os....

[Rubblebuster (erik-fornoff)]

@remy
As already mentioned in comment #30 a workaround has been added to kernel 2.6.30 by reducing the delay-time from ~ 8.000 ms per EHCI-controller to 1.000 ms per EHCI-controller for such Boards behaving as reported. Which Ubuntu-version do you use and which kernel? At least for my Toshiba Satellite laptop in combination with Lucid I now see a 2 seconds delay (which is OK for me) - instead of 16 seconds in previous Ubuntu version.

Additionally if you followed the correspondence in this bug-report (especially comment #13) you might have recognized that this problem is nothing about "Linux looking like an old OS" but in fact once more about "Linux tries to relate on specifications" - but if hardware-vendors doesn't take care about them - or if specific functionality is not described as accurate as necessary then we have a problem.

Anyhow it's important to understand, that this is not a Linux bug => there's a problem with your hardware in combination with the BIOS and the "old OS" is trying to deal with that... :-)

[UnableToLoad (unabletoload)]

also my notebook has the same issue, sony vaio VPCEB3J1E attached dmesg and lspci -vv,
kernel 2.6.37-020637rc2-generic

[UnableToLoad (unabletoload)]

PS i'm using maverick

[Brad Figg (brad-figg)]

This bug was filed against a series that is no longer supported and so is being marked as Won't Fix. If this issue still exists in a supported series, please file a new bug.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[Thanasis Naskos (naskoos)]

I'm using Ubuntu 14.04 beta 2 in a lenovo g510 laptop. This bug affects me!

Here is a snapshot from my dmesg:
[ 0.553488] pci 0000:00:02.0: Boot video device
[ 2.155741] pci 0000:00:1a.0: EHCI: BIOS handoff failed (BIOS bug?) 01010001
[ 3.757032] pci 0000:00:1d.0: EHCI: BIOS handoff failed (BIOS bug?) 01010001
[ 3.757214] pci 0000:03:00.0: set MSI_INTX_DISABLE_BUG flag
[ 3.757222] PCI: CLS 64 bytes, default 64

I'm attaching the output of lspci -vv.