Opened 13 years ago
Closed 13 years ago
#8033 closed defect (fixed)
Since 4.0.0 Linux guest fails to complete booting with a PIT timer => Fixed in SVN
| Reported by: | rocks | Owned by: | |
|---|---|---|---|
| Component: | other | Version: | VirtualBox 4.0.0 |
| Keywords: | Cc: | ||
| Guest type: | other | Host type: | other |
Description
Since upgrading to VirtualBox 4.0.0, my Gentoo guest fails to complete booting with a tickless kernel. Passing nohz=off works around the problem.
Tested with 2.6.31 and 2.6.36.
Attachments (2)
Change History (8)
by , 13 years ago
comment:1 by , 13 years ago
comment:2 by , 13 years ago
While the boot process doesn't always freeze in the exact same position, it is always very close, and it definitely never reaches init. When it stops, VirtualBox drops from ~50% CPU to ~1%.
comment:3 by , 13 years ago
Any chance to get the complete guest kernel output? You could setup a serial interface with a raw file backend and redirect the kernel console to the serial port (append console=ttyS0 to the kernel command line). The interesting question is why the kernel chose the PIT as timer and not the Local APIC.
comment:4 by , 13 years ago
| priority: | major → critical |
|---|
Actually reproducible if the guest kernel is booted with nolapic.
comment:5 by , 13 years ago
| Summary: | Since 4.0.0 Linux guest fails to complete booting with tickless kernel → Since 4.0.0 Linux guest fails to complete booting with a PIT timer => Fixed in SVN |
|---|
This is not related to a tickless kernel. Booting with nolapic nohz=off shows the same problem. That problem will be fixed in the next maintenance release.
Can you explain this a bit more? The guest boots fine until a certain phase and then stops booting? Does it stop always at the same location or is that different? What happens if it stops: Does the VM process consume any CPU cycles?