[DragonFlyBSD - Bug #2484] virtio hangs boot of i386 Ideapad with acpi disabled

[Sascha Wildner via Redmine]

Issue #2484 has been updated by swildner.


On Thu, 10 Jan 2013 00:47:30 +0100, David Shao via Redmine  
<bugtracker-admin at leaf.dragonflybsd.org> wrote:

>
> Issue #2484 has been reported by David Shao.
>
> ----------------------------------------
> Bug #2484: virtio hangs boot of i386 Ideapad with acpi disabled
> http://bugs.dragonflybsd.org/issues/2484
>
> Author: David Shao
> Status: New
> Priority: Normal
> Assignee:
> Category:
> Target version:
>
>
> For a Lenovo Ideapad netbook (i386) with acpi disabled using master  
> commits through
> commit 1d0de3d37b9efb6301b10515fd12d922c8602093
> Date:   Tue Jan 8 19:21:35 2013 +0100
>
>     kernel/vfs: Remove some unused variables.
>
> boot hangs after devices:
>
> md0: invalid primary partition table: no magic
> Math emulator present
> hptrr: RocketRAID 17xx/2xxx SATA controller driver v1.2
> hpt27xx: RocketRAID 27xx controller driver v1.0 (Jan  9 2013 12:15:05)
>
> and before
>
> IOAPIC: rman cpu0 0 - 1
> IOAPIC: rman cpu0 3 - 95
> IOAPIC: rman cpu0 97 - 191
> IOAPIC: rman cpu1 24 - 95
> IOAPIC: rman cpu1 97 - 191
> npx0.nexus0.root0
> npx0: <math processor> [tentative] on motherboard
> npx0: INT 16 interface
>
> The boot problem was bisected to be after:

Here's what I know about this bug:

* It seems to occur on i386 only.

* The boot hanging occurs randomly. The box I have that exhibits the bug  
boots in 70-80% of the cases and hangs in 20-30%. So I think you had a  
wrong impression during bisecting when it booted. Just go and try booting  
a few times with a state that hung, and you will see that eventually it  
will boot. Matt had the same issue once on an Atom box and committed a fix  
to alc(4) which he thought was related to the issue, when in fact it was  
not, just because it happened to boot after that. The fix was OK  
nevertheless (just not related). :)

* The bug is not related to hptrr or hpt27xx drivers. The messages those  
drivers give are informational and don't indicate that any such adapter  
was found. In a booting dmesg you will notice that later on they both say  
that no matching adapter was found. Also, if hpt* drivers are removed from  
the kernel config, the boot hanging will continue (just with the "Math  
emulator present" as the last message). If _any_ messages are related to  
it, it is the IOAPIC: rman stuff (which it doesn't get to when not  
booting), but I'm not sure if something in ioapic is related to the issue.

* The bug has nothing to do with virtio either. It started before we  
imported that.

Best regards,
Sascha
----------------------------------------
Bug #2484: virtio hangs boot of i386 Ideapad with acpi disabled
http://bugs.dragonflybsd.org/issues/2484

Author: davshao
Status: New
Priority: Normal
Assignee: vsrinivas
Category: 
Target version: 


For a Lenovo Ideapad netbook (i386) with acpi disabled using master commits through
commit 1d0de3d37b9efb6301b10515fd12d922c8602093
Date:   Tue Jan 8 19:21:35 2013 +0100

    kernel/vfs: Remove some unused variables.

boot hangs after devices:

md0: invalid primary partition table: no magic
Math emulator present
hptrr: RocketRAID 17xx/2xxx SATA controller driver v1.2
hpt27xx: RocketRAID 27xx controller driver v1.0 (Jan  9 2013 12:15:05)

and before

IOAPIC: rman cpu0 0 - 1
IOAPIC: rman cpu0 3 - 95
IOAPIC: rman cpu0 97 - 191
IOAPIC: rman cpu1 24 - 95
IOAPIC: rman cpu1 97 - 191
npx0.nexus0.root0
npx0: <math processor> [tentative] on motherboard
npx0: INT 16 interface

The boot problem was bisected to be after:

commit 42fdf81ee882d38808efe2772a68a9c4991a5f76
Date:   Sun Jan 6 20:01:27 2013 +0800

and before:

commit f31f6d84db329735c062a41c0f729843cd4d0b6f
Date:   Mon Jan 7 09:48:18 2013 +0100

    kernel/hammer: Remove unused variables and add __debugvar.

In addition a workaround is to disable kernel options for virtio for a custom kernel config:

# VirtIO support
# device		virtio		# VirtIO core
# device		virtio_blk	# VirtIO disk driver
# device		virtio_pci	# VirtIO transport over PCI bus

Attached is a dmesg for the booting configuration with acpi and virtio disabled.



-- 
You have received this notification because you have either subscribed to it, or are involved in it.
To change your notification preferences, please click here: http://bugs.dragonflybsd.org/my/account