Ticket #8018 (closed defect: duplicate)
upgrade to 4.0.0 causes host panic
Reported by: | helix84 | Owned by: | |
---|---|---|---|
Component: | other | Version: | VirtualBox 4.0.0 |
Keywords: | Cc: | ||
Guest type: | Linux | Host type: | Linux |
Description
Host: Debian Linux Lenny Guest: Debian Linux Lenny
After upgrading from 3.2.12 to 4.0.0+extpack and starting one of the guests, I get a host panic. The only thing with relevant info I have is a screenshot of the host panic screen. I'll also attach VBox.log of the guest that caused the crash. I can provide any log files on request.
Interestingly, another guest with the same OS runs fine. It crashed both when I started the guest both from VBoxHeadless and VirtualBox GUI, in the latter case I noticed the last thing the guest was doing was "Setting up network..." Both guests use the Am79C973 NIC. The host interfaces setup is as follows: Working guest -> tap -> br -> vlan Crashing guest -> vlanXY
Luckily the downgrade to 3.2.12 works flawlessly.
Attachments
Change History
Changed 12 years ago by helix84
-
attachment
vbox4panic.jpg
added
Changed 12 years ago by helix84
-
attachment
VBox.log.1
added
vbox.log of the guest responsible for the crash
comment:1 Changed 12 years ago by helix84
This may be unrelated but after the upgrade I also noticed a significant slowdown in the command-line tools (their speed matters to me because I run them in batch from a few scripts), in particular VBoxManage showvminfo. I didn't measure the speed of guests themselves.
comment:2 Changed 12 years ago by frank
There is a known problem with bridged networking, see #7908. Could you check if the patch posted there works?
comment:3 Changed 12 years ago by helix84
It's a production server, but I can experiment at night. Should I apply both patches? What does each one do, what should I watch for and which log files will you need?
Another experiment I'll do is to change host networking to guest -> tap -> br -> vlan and see if it crashes.
Btw tc qdisc of all my interfaces looks like this: qdisc pfifo_fast 0: dev eth1 root bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
comment:4 Changed 12 years ago by frank
Only the noqueue patch. This patch ensures that a kernel function is not used if it was not initialized. It should be uncritical. This happens with some network cards. This fix will be included in the next maintenance release. Unfortunately we didn't see this problem during the Beta phase.
host panic screen