2.6.22-14-rt halts during boot on some computers, without the nosmp noapic nolapic kernel options

Bug #175227 reported by LA
10
Affects Status Importance Assigned to Milestone
Ubuntu Studio
Invalid
Undecided
Unassigned
linux (Ubuntu)
Fix Released
Undecided
Alessio Igor Bogani
linux-source-2.6.22 (Ubuntu)
Won't Fix
Low
Alessio Igor Bogani

Bug Description

Binary package hint: linux-image-2.6.22-14-rt

Fresh installation of Ubuntustudio 7.10 on an ACER Aspire 5102 .

The system boots, the Ubuntustudio splash appears. The halts at the same point (even for hours), unless I keep pressing ANY key (even alt or ctrl) until it gets into the graphical interface.
Switching to the boot terminal ( Alt+Ctrl+F1 ) I noticed that it waits for me to press any key, in order to proceed one step further in the boot process.
The same happens if I shutdown the system. It doesn't shutdown unless I press any key for 2-3 times.

I finally found sth similar reported in ubuntuforums ( http://ubuntuforums.org/showthread.php?t=448792 ), by another acer 5100 user, and the proposed solution worked.
They proposed adding nosmp noapic nolapic as kernel boot parameters in the /boot/grub/menu.lst
Of course this solution is not acceptable because nosmp takes advantage of only half of my 64 x2 CPU processor.

The problem didn't occur
- with the same ubuntustudio version, freshly installed, on my desktop PC.
- after downgrading my laptop to the "-generic" kernel ,without any kernel boot option
- on an older installation of ubuntu Feisty, on the same laptop.
therefore I consider it as a potential bug.

Revision history for this message
LA (radiobuzzer) wrote :
description: updated
Revision history for this message
LA (radiobuzzer) wrote :
Revision history for this message
LA (radiobuzzer) wrote :
Revision history for this message
LA (radiobuzzer) wrote :
description: updated
LA (radiobuzzer)
description: updated
Revision history for this message
Alessio Igor Bogani (abogani) wrote :

Could you test the three options (nosmp noapic nolapic) separately?
Please remember to comment out quiet and splash options.

Thank you!

Revision history for this message
LA (radiobuzzer) wrote :

Yes. Using just the NOAPIC option doesn't solve anything, computer halts as described above, unless keys are pressed. I forgot to keep a log of that, if you need it let me do it.

Using just the nosmp option, causes a lot of errors, mainly "Unable to locate IOAPIC for GSI 10" also for GSI 11, 16 . Then "hda lost interupt" and then it doesn't mount the root file system and gives me a commandline which I can't do much within.

Using just the NOLAPIC option, boots quiet fast, but when the X boots, it gives an error message "Failed to initialize HAL" and then no ethernet, no wireless working etc.

I attach the logs concerning the nolapic boot.

Revision history for this message
LA (radiobuzzer) wrote :
Revision history for this message
dragon76 (tfishwall) wrote :

I can confirm this bug on Everex StepNote ST5340T (same as Averatec 2300 series).

AMD x2 cpu
SATA HD
Nvidia chipset

Revision history for this message
dragon76 (tfishwall) wrote :

I have confirmed the bug

Changed in linux-source-2.6.22:
status: New → Confirmed
Changed in ubuntustudio:
status: New → Confirmed
Revision history for this message
dragon76 (tfishwall) wrote :

Almost forgot... I'm running a Kubuntu Gutsy install, not Ubuntu Studio. Kernel with problem is 2.6.22-14-rt amd64.
32bit generic and rt run fine. 64bit generic runs fine.

Revision history for this message
Lorenzo (deodato-art) wrote :

I have a similar - but more radical - problem on an Acer Travelmate 4152 Lmi (a Centrino-based notebook with a 32-bit Pentium M 1.73 GHz). A few instants after booting, the 2.6.22-14-rt kernel hangs with a long stack trace, of which I can only see the last part. I will post the trace as soon I have access to the notebook again (currently I am away from home).

Two notes:
- The problem occurred also with the previous version of the 2.6.22 "rt" kernel (the one initially shipped with Gutsy); it did not occur with the 2.6.20-lowlatency shipped with feisty, so I think it is a bug inserted in the .22 series.
- Starting the kernel with the "nosmp noapic nolapic" options does not change anything - the problem happens anyway.
- I am running Ubuntu Gutsy upgraded from a Feisty clean installation.

Revision history for this message
Lorenzo (deodato-art) wrote :

As promised, I "dumped" the error messages printed by the kernel on start. The log follows:

[<c0240b51>] pnp_assign_resources+0x281/0x2d0
[<c0240c1c>] pnp_auto_config_dev+0x7c/0xc0
[<c023eb0b>] pnp_alloc+0x1b/0x60
[<c0240c83>] pnp_activate_dev+0x23/0x50
[<c0241e7a>] quirk_smc_enable+0x1a/0x300
[<c0242cff>] pnpacpi_option_resource+0x11f/0x460
[<0x204d6b>] vsnprintf+0x54b/0x630
[<c0241df1>] pnp_fixup_device+0x41/0x50
[<c023e979>] __pnp_add_device+0x9/0xd0
[<c0418097>] pnpacpi_add_device_handler+0x217/0x2e0
[<c023099e>] acpi_ns_get_device_callback+0x131/0x14f
[<c023205a>] acpi_ns_walk_namespace+0x102/0x11c
[<c02307e6>] acpi_get_devices+0x58/0x6b
[<c023086d>] acpi_ns_get_device_callback+0x0/0x14f
[<c0417e80>] pnpacpi_add_device_handler+0x0/0x2e0
[<c0417dbd>] pnpacpi_init+0x5d/0x90
[<c03fb591>] kernel_init+0x141/0x3b0
[<c010300a>] __switch_to+0xaa/0x1d0
[<c01271a3>] schedule_tail+0x53/0x120
[<c0104216>] ret_from_fork+0x6/0x20
[<c03fb450>] kernel_init+0x0/0x3b0
[<c03fb450>] kernel_init+0x0/0x3b0
[<c0105627>] kernel_thread_helper+0x7/0x10
===============

After the last message, the machine hangs and I have to do an hard reboot.

Revision history for this message
jwmislan (jwmislan) wrote :

Micro Star International - AMD-64 Turion - TL-56 - MCP-51 High Definition Audio - pci vendor - nVidia Corporation - pci product - G70 [ GeForce Go 7600
Ubuntu Studio installed over standard Ubuntu Gutsy Desktop.
Running kernels - 2.6.22-14-generic - and - 2.6.22-14-rt

Generic kernel always boots flawlessly -
rt kernel, always stalls - waiting for any key presses.
If I press the (Blue-Power-switch)-(acts as resume) , the boot goes through to just before the Nvidia splash screen, and then stalles again until I press enter 2, or 3 times - then proceeds to gdm login.

I tried noacpci in the grub boot line - that works and allows boot without any stalls, problem is that the realtime module does not load without acpic activated.

We want both, acpic and realtime functions running as should be standard. (Ubuntu Studio)

We don't want to disable something in order to accommodate some buggy code in the rt-kernel when the buggy lines could easily be modified/corrected, and an update offered to resolve this matter.

I'm surprised that we have to endure this inconvenience throughout the entire 7.04 release, and with no indication that some solution is being worked, or that there is any acknowledgment at all for this system flaw, situation.

Any way, I'm acknowledging that I, and many others are still enduring this inconvenient system flaw.

What Next ? - May be it will get some attention, and be fixed !!

JWM

Changed in ubuntustudio:
status: Confirmed → Invalid
Revision history for this message
Schmallon (matthias-kleine) wrote :

I have more or less the same problem as Lorenzo. If I try to boot the linux-rt kernel the system freezes after a second and I end up with a stack trace. I've tried several boot options without success. I'm working on a HP nx8220 (also a centrino based notebook). Is there any information I can provide that would help determining the source of the problem?

Revision history for this message
Alessio Igor Bogani (abogani) wrote :

@Schmallon:

Could you test linux-rt on Hardy Beta?

Thanks!

Revision history for this message
Lorenzo (deodato-art) wrote :

@Alessio:

I tried the hardy 2.6.24 realtime kernel about three months ago and it did not show the problem, so I think the bug is specific to 2.6.22.

Changed in linux:
assignee: nobody → abogani
status: New → Fix Released
Changed in linux-source-2.6.22:
assignee: nobody → abogani
Changed in linux-source-2.6.22:
importance: Undecided → Low
status: Confirmed → Triaged
Revision history for this message
Alessio Igor Bogani (abogani) wrote :

The fix is available in the Hardy release. The backporting isn't possible so i mark this bus as "Won't Fix" for Gutsy.

Changed in linux-source-2.6.22:
status: Triaged → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.