[vbox-dev] I have a trouble when using USB2.0

[druid0523]

Hello, all vbox developers:
My company want to use VMs to run our android auto-testcases. So I do some tests to determine whether Virtualbox works.
Virtualbox 4.2.24 host on Ubuntu 12.04 server. There are 8 VMs, each of which has 2 cpu, 2048mb Memory, and 2 android mobile phones passthrough by USB2.0, OS is WIN7. I start the VMs using headless mode. I run a simple test on each VM: pushing a file into mobile phone thousands of times with adb. I hope the tests run a day without any VM down.
But I have a trouble. I do not know why but some VMs seem to be down when the tests have ran about an hour. I can not ping them, can not connect them with Remote Desktop. If I turn on VRDE, it just show a black screen. But The VBoxHeadLess processes are still there. I can not do anything with vboxmanage, it seems that VMs are locked. For example, I shutdown it with vboxmanage, wait a long time(about an hour), but vboxmanage has no response.
If I kill the VBoxHeadLess processes, the states of the VMs show that it is "Aborted". If I turn on the VMs again, Windows tell me I shutdown the system in abnormal way. Windows log has no errors before the VMs down, expect the abnormal shutdown warning.
I use gdb to attach the VM process. And then I found that the thread "EMT-1" may have a dead loop. The thread is always running between line 3093 to 3112 in file /mnt/tinderbox/extpacks-4.2/src/VBox/Devices/USB/DevEHCI.cpp. 
There may be a problem in the extpack, but I have no source, I can not find and fix it. Does any one know why and how to fix it?

Thank you very much! Hope a reply quickly!




druid0523
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.virtualbox.org/pipermail/vbox-dev/attachments/20140527/11949ed5/attachment.html>