Jaunty does not boot from USB on SB600/RS690

[Dan Dart]

{ begin edit section by Paul}
SB600 and SB700 pre revision A21 are problematic on ATI RS690 host bridge:
Workaround "acpi=off pci=nomsi" did not work.
http://lkml.org/lkml/2008/1/24/431
Looks like some patch was given on Linux mailing list but unclear to me if it was finally accepted:
http://lkml.org/lkml/2008/1/24/484
{close edit section }

Ubuntu 8.10 kernel (2.6.27) succeeds booting but only on hpet=disable

Self-compiled kernels up to 2.6.28.2 succeed to boot on ubuntu 8.04 (no initrd)

Anything higher (2.6.28.3+ on Ubuntu 8.10) fails to boot with messages about not being able to find the hard disk, and drops to a prompt (evidently the hard disk was found but appears in /sys as an error instead of /dev)

2.6.29 on Ubuntu 8.10 fails to boot, and lists the disks that the ATA system found (includes hard disk) but gets to a kernel panic because it cannot find (0,0) or the sda device node.

Ubuntu 9.04 kernel (2.6.28.8 based) fails to boot, dropping to busybox prompt, even with acpi=off, nolapic, noapic, hpet=disable, edd=on, "verbose" replacing "splash silent" options. and on expert mode, the messages when finding an ext3 filesystem loop over and over before it gets to busybox.

Motherboard: Gigabyte GA-MA69VM-S2, SATA300 jumper on Maxtor disk removed.

AMD Athlon X2 4800+ 2.5GHz virtualisation enabled
2 GB DDR800 RAM

Cheers!

[Jon Charge (seropith)]

Hello and thank you for your report.

Have you attempted booting with the SATA jumper in place?

Thank you for your interest!

Jon

[Dan Dart (dandart)]

> Have you attempted booting with the SATA jumper in place?
I have not, simply for the fact that I cannot find it. I will however
try with the BIOS setting changed to a different type of hard disk.

[Paul Dufresne (paulduf)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. Unfortunately we can't fix it without more information.

Please include the following additional information, if you have not already done so (pay attention to lspci's additional options), as required by the Ubuntu Kernel Team:
1. Please include the output of the command "uname -a" in your next response. It should be one, long line of text which includes the exact kernel version you're running, as well as the CPU architecture.
2. Please run the command "dmesg > dmesg.log" after a fresh boot and attach the resulting file "dmesg.log" to this bug report.
3. Please run the command "sudo lspci -vvnn > lspci-vvnn.log" and attach the resulting file "lspci-vvnn.log" to this bug report.

For your reference, the full description of procedures for kernel-related bug reports is available at https://wiki.ubuntu.com/KernelTeamBugPolicies Thanks in advance!

[Paul Dufresne (paulduf)]

Symptoms similar on https://bugs.launchpad.net/bugs/350672

[Dan Dart (dandart)]

uname -a:
Linux dominator 2.6.27-7-generic #1 SMP Tue Nov 4 19:33:06 UTC 2008
x86_64 GNU/Linux

[Dan Dart (dandart)]

uname -a:
Linux dominator 2.6.27-7-generic #1 SMP Tue Nov 4 19:33:06 UTC 2008 x86_64 GNU/Linux

[Dan Dart (dandart)]

Tried booting Ubuntu Jaunty beta (2.6.28.8) on legacy IDE, native IDE and AHCI SATA modes and still got a busybox.

[Paul Dufresne (paulduf)]

We should probably call it a SATA-1.5Gb/s jumper, because when it is there, the speed is limited.
When it is not there, auto-negotiation is supposed to be made with motherboard (if it supports it).

Anyway, in your case, the motherboards supports it: 4 SATA 3Gb/s connector
I would give the link normally, but there seems to be a bad bug with my mozilla.

Your dmesg log suggest it has some problem to reset the hard disk:
[ 2.008723] scsi3 : ahci
[ 2.008823] ata1: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f100 irq 22
[ 2.008827] ata2: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f180 irq 22
[ 2.008830] ata3: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f200 irq 22
[ 2.008833] ata4: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f280 irq 22
[ 2.492100] ata1: softreset failed (device not ready)
[ 2.492106] ata1: failed due to HW bug, retry pmp=0
[ 2.656036] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[ 2.657264] ata1.00: HPA unlocked: 488395055 -> 488397168, native 488397168
[ 2.657269] ata1.00: ATA-7: MAXTOR STM3250310AS, 3.AAF, max UDMA/133
[ 2.657271] ata1.00: 488397168 sectors, multi 16: LBA48 NCQ (depth 31/32)
[ 2.657277] ata1.00: SB600 AHCI: limiting to 255 sectors per cmd
[ 2.658015] ata1.00: SB600 AHCI: limiting to 255 sectors per cmd
[ 2.658017] ata1.00: configured for UDMA/133
[ 2.976035] ata2: SATA link down (SStatus 0 SControl 300)
[ 3.296033] ata3: SATA link down (SStatus 0 SControl 300)
[ 3.616034] ata4: SATA link down (SStatus 0 SControl 300)
[ 3.616080] isa bounce pool size: 16 pages

I think this bug have enough info for a member of the kernel team to begin investigating.
So I am confirming it.

[Dan Dart (dandart)]

Didn't suspect that.

The messages have been doing

ata1: softreset failed (device not ready)

for ages, since 2.6.27 I think.

[Dan Dart (dandart)]

Just discovered that the reason that Ubuntu would go to busybox was
that it could not find the squashfs image because I installed it onto
a USB disk.

Shall try the 9.04 beta from cdrom and post results.

[Dan Dart (dandart)]

8.10 CD: Boots
8.10 USB: Boots

While 8.10 boots, hpet=disable or acpi=off is still required to boot
else it hangs and even then, I get messages such as:
Clocksource tsc unstable (delta = -211219077 ns)
which have been excessive in the past when trying to boot a Xen kernel.
which may or may not be related to this bug.
But I've had that one since Ubuntu 8.04 (kernel 2.6.24)

9.04 USB: Does not boot. 80 odd dmesg's about initialising the HD and
finding ext3-FS:

kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.

Then drops to busybox.

[Paul Dufresne (paulduf)]

You seems to both (you and newly marked duplicate bug #350672,
RS690 as host bridge in lspci.
Searching on Google I found this that seems could be related:
http://lkml.org/lkml/2008/8/30/208 (check all the thread because patch correctness was discussed).

My problem is that I don't understand the quarter of the thread.

[Dan Dart (dandart)]

This is what may be related to a bug in the kernel regarding to the
host bridge, or SATA chipset, preventing boot.

[Paul Dufresne (paulduf)]

Sorry, seems to be bad link.
I wanted to speak of the following:
http://lkml.org/lkml/2008/1/24/78

About disabling the MSI of AMD RS690.
Except I don't know what is MSI.
Googling MSI+linux+kernel:
http://www.mjmwired.net/kernel/Documentation/MSI-HOWTO.txt

[Paul Dufresne (paulduf)]

Dan, can you try to boot with kernel parameter:
pci=nomsi

Extract of http://www.mjmwired.net/kernel/Documentation/MSI-HOWTO.txt :
"For debugging purpose, the user may also pass pci=nomsi on the kernel command-line to explicitly disable MSI globally. But, once the appropriate quirks are added to the kernel, this option should not be required anymore."

[Dan Dart (dandart)]

I will try compiling kernel without MSI and see if that works.

[Dan Dart (dandart)]

> pci=nomsi

sure.
 will try that now

[Dan Dart (dandart)]

>pci=nomsi
No luck with Jaunty beta (2.6.28.8) I will try to compile a 2.6.29
without msi support at all.
Jaunty kept trying to initialise my hard disk even though it wasn't
meant to (usb boot) and made 3 of those messages every second:

kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.

Then stops and goes into busybox.

[Dan Dart (dandart)]

No luck with 2.6.29 either.

I /am/ supposed to disable ATA/ATAPI/MFM/RLL and enable Serial ATA
(prod) and Parallel ATA (experimental) aren't I?
(I have disabled all experimental modules in there)

(I have a SATA HD and a PATA CD)

Just thought, something in there could be the problem.. I doubt it,
I'm still getting the "HW Bug" message, whether I have MSI or not...

[Paul Dufresne (paulduf)]

>>pci=nomsi
>No luck with Jaunty beta (2.6.28.8) I will try to compile a 2.6.29 without msi support at all.
From what I did read, that option should have disable it totally, so I
don't expect a different result from recompiling.

While in busybox, is there some logs that could give some more info?
Like, does 'dmesg' give any results?

2009/3/31, Dan Dart <email address hidden>:
>>pci=nomsi
> No luck with Jaunty beta (2.6.28.8) I will try to compile a 2.6.29
> without msi support at all.
> Jaunty kept trying to initialise my hard disk even though it wasn't
> meant to (usb boot) and made 3 of those messages every second:
>
> kjournald starting. Commit interval 5 seconds
> EXT3-fs: mounted filesystem with ordered data mode.
>
> Then stops and goes into busybox.
>
> --
> Gigabyte GA-MA69VM-S2 does not boot Jaunty
> https://bugs.launchpad.net/bugs/350531
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Dan Dart (dandart)]

Dmesg is available I think... but I don't know how to save it because
I can't mount the disk!

I'll try though.

/var/log isn't available as it's only at the ramdisk stage at that moment...

[Dan Dart (dandart)]

I could mount /dev/sda1, the hard disk.... but not /dev/sdb1, the usb
disk which it was booting from!!!

In the messages I could see it couldn't find modules.dep in
/lib/modules/2.6.28..etc/
And it said ata1: softreset failed again.
Will attach dmesg output.

[Paul Dufresne (paulduf)]

According to OpenSuse bug, both options are needed: "acpi=off pci=nomsi"

[Paul Dufresne (paulduf)]

Usefull to read for kernel team:
http://lkml.org/lkml/2008/1/25/307

http://lkml.org/lkml/2008/1/24/431 (will be fixed in hardware in SB700 revion A21)

http://www.google.ca/url?sa=U&start=7&q=http://www.linuxforums.org/forum/wireless-internet/141168-ubuntu-a215-s4757-wireless-card.html&ei=GAjTSdDgMZu0MaqV0ekC&sig2=Q3JGY3G7aOPAmojRIIjuiw&usg=AFQjCNHCWr_ZGfOohDRVXl82ogUzp3h3Gw
seems to suggest it does affect wireless cards to, so it seems to be a general hardware bug with MSI rather than just to a specific driver.

just a bug of links with SB600 and RS690 terms:
http://webui.sourcelabs.com/api/Filter_hardware%253Aati-sb600

[Paul Dufresne (paulduf)]

Confirming that duplicate bug #350672 is also RS690 with SB600.
00:00.0 Host bridge [0600]: ATI Technologies Inc RS690 Host Bridge [1002:7910]
[ 20.496564] SB600_PATA: IDE controller (0x1002:0x438c rev 0x00) at PCI slot 0000:00:14.1

[jusko (jusko)]

Is there any chance to fix this problem until official realse ? I would like to use Ubuntu 9.04 :-( Hardy Heron is ok but... it's still 8.04.

[Dan Dart (dandart)]

acpi=off pci=nomsi doesn't work either.

[Dan Dart (dandart)]

Anything else I should test?

[Dan Dart (dandart)]

Intrepid works with kernel 2.6.29.1 from the Kernel PPAs,

So, either, our problem was fixed from the time of Jaunty kernel to the time of 2.6.29.1...
... or the problem is Jaunty's other software.

[Dan Dart (dandart)]

Tried booting again to Jaunty Beta from USB. Tried disabling SATA, network, PATA in bios, bootup pauses with no dmesg output for as long as it did before with the many ext3 messages in a loop and drops to busybox. Tried various "acpi=off", "hpet=disable", "pci=nomsi" options all to no avail. After message "Mounting root image" appears it just stops for a minute. All options drop to busybox.

This is serious! I can't use Ubuntu!
Please mark as critical, because it still doesn't boot AT ALL!

Note: It still does not find modules.dep, which is probably involved.

Proposed openSUSE solution of "acpi=off pci=nomsi" fails completely as before.

What else can I try?

[Dan Dart (dandart)]

Did I tell you? This uses amd64 kernel on and64 machine. Just checking you knew that.

[Dan Dart (dandart)]

This link is not to my chipset or kernel.

[Dusan Maliarik (dusan-maliarik)]

Still no luck, I don't want to waste CD just for another release, not because my cd drive is not working, it's just stupid that there is USB installer and it won't work :(

[Dan Dart (dandart)]

openSUSE bug isn't the same. Not the same version of kernel, not the same dmesg output. Plus the fixes described there do not work.

[jusko (jusko)]

And we have official relase of 9.04 Jaunty Jackalope... and still i have this bug(?) (i have BusyBox even on 8.10) :-( Only chance to use newest version of Ubuntu is waiting to 9.10... if kernel team fix this problem. Maybe it's good time to change distro... :-(

[Dan Dart (dandart)]

hpet=disable or acpi=off work in 8.10 for me...

Happy release without us, guys...

[acfrazier (acfrazier)]

Can also confirm this bug on my Gateway T-1616 laptop, ATI chipset, AMD turion 64 x2, upgraded to 4 gb ram and 500 gb hdd,
Oddly enough, I upgraded to Ubuntu 9.04 and the actual system boots fine, just not the LiveCD.

[Dan Dart (dandart)]

@acfrazier Can you post your dmesg and lspci -vv so we know it's
exactly the same? Maybe we'll be able to get something different so we
know which hardware it is.

[acfrazier (acfrazier)]

@Dan Dart I would but I can't figure out how to output those to a file from within BusyBox. I booted the LiveCD from my USB hard drive, since it goes so much faster.

[Dan Dart (dandart)]

you do:
dmesg > dmesg.out
mkdir /bob
mount /dev/xxx (your HD) /bob
cp dmesg.out /bob

and it's on your disk so you can upload

also give us a lshw (from any system)

[acfrazier (acfrazier)]

Attached are the requested dmesg output from the 9.04 livecd boot, and lshw output from a working Ubuntu 8.10 LiveCD.

I can't put more than one attachment, so I'll do two separate posts.

[acfrazier (acfrazier)]

Here is the lshw output also as requested.

[Dan Dart (dandart)]

That concludes it. We all have the RS690 / SB600 thing.. so that must
be what's broken. Did you have to hpet=disable on 8.10?

[acfrazier (acfrazier)]

No, 8.10 runs beautifully.
It does spam the EXT3-fs: mounted filesystem with ordered data mode.
for a little while, then snaps out of it and boots otherwise normally.

[Dan Dart (dandart)]

The CD for Jaunty DOES work for me! I haven't had a CD of it till now. But the USB version does not boot.
I am confused why.

[Dan Dart (dandart)]

dmesg attached. Spams only 3 of each message:

kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.

Then initialises the CD. So it could be something about recognising ISO but not vfat filesystems at boot.

[Dan Dart (dandart)]

dmesg attached. Spams only 3 of each message:

kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.

Then initialises the CD. So it could be something about recognising ISO but not vfat filesystems at boot.

[acfrazier (acfrazier)]

I am going to test the CD on my hardware right now.

[acfrazier (acfrazier)]

That is very odd. The CD boots, just not the USB.

[jusko (jusko)]

Boot from CD - won't work.
Boot from USB - won't work :-(

Someone fix this problem someday or only way is waiting for 9.10 ? Only way to run 9.04 is upgrading from my 8.04. 2 hours of upgrade and Jaunty won't see my CD-ROM drive - I don't have time and knowledge for fix this. At this moment for me only way to using Ubuntu is Hardy Heron. Anyway - i've spend 15 minutes with Jaunty - i don't like it :-P I have this problem "kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode." only on Ubuntu. 8.10 won't work - 9.04 too. I hope 9.10 will be working... at least.

[Sergey Zelenev (zelenev)]

This bug also affects HP T5730 and T5735 (both RS690). Ubuntu 9.04 does not boot at all when trying to boot from either an external USB hard drive, or a USB CD drive. Ubuntu 8.10 boots just fine. Will post lspci once I get it to work from internal ATA flash drive.

[Dan Dart (dandart)]

Sergey: Tried from CD? That fixed it for me.

[Sergey Zelenev (zelenev)]

LiveCD does not boot since the only CD drive the T5730 has is a USB drive - there are no internal IDE connectors.
However the text mode installer kernel from Ubuntu 9.04 works fine and detects all the drives - for example, in rescue mode.
I got Ununtu 9.04 installed on the built-in ATA flash.
The built-in ATA flash is not large enough to hold both /boot, /lib and /usr, so I have to store /usr on a USB drive, but the system still does not boot all the way.

[Sergey Zelenev (zelenev)]

Got my system to boot, but it is still not booting properly.
I had to add usb_storage to /etc/modules
then I had to update initramfs

kernel still does not find my USB filesystems at boot and exits to a shell, however couple seconds later, at 20 seconds since boot it finds my USB filesystems
after that I do mount -a and it mounts everyting I need
I then exit from shell and it continues to boot and eventually boots into Xorg

[Sergey Zelenev (zelenev)]

lspci -vvnn

[Sergey Zelenev (zelenev)]

uname -a:
Linux Bedroom 2.6.28-11-generic #42-Ubuntu SMP Fri Apr 17 01:58:03 UTC 2009 x86_64 GNU/Linux

attached is dmidecode output

[Daniel Kulesz (kuleszdl)]

I found another "manual" workaround for this problem, which works also on a HP t5735. When booting Ubuntu 9.04/i386 you get dropped to the initram shell since it cannot find the USB device. Plugging out and re-plugging in the USB device makes it magically appear, therefore typing "exit" at the initram shell prompt continues normal booting and succeeds. After installing Debian 5.02 on the same machine the USB boot problems are gone. I haven't tried earlier Ubuntu versions though.

I have similiar problems on completely different machines: On a very old 266MHz ThinTune thin client, when running Ubuntu 9.04 Server, the machine first boots the kernel fine off an internal CF card and is then supposed to continue booting userland and friends from USB. The last step fails though - unless you (unplug and re-)plug in the USB device after getting dropped to the initram shell. The same happens for a "UD2" thin client from Igel (http://www.igel.de/igel/live.php,navigation_id,3561,_psmand,1.html) which is equipped with a VIA chipset, CPU etc.

To me it seems that the USB boot problems mentioned here are *NOT* related to the ATI RS600/690 chipset, but affect a lot of other machines as well. I attached the lspci output of the thintune machine.

I suggest, we should change the topic to reflect the generality of the problem.

[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.