OSX host crasher repro

[swork]

I have a simple-case repro for a host crash problem possibly related to several open or "closed - works for me" bug records here. It's quite consistent on my host (MacBookPro6,1, 8GB RAM). The VM config simplifications described below don't seem to be necessary but might help rule out culprits.

1. Create two VMs, all defaults. Simplify the situation by eliminating disks (no devices attached to default IDE bus), disable audio, disable network cards.

2. Run a shell-script loop starting both VMs then stopping both VMs. ("for i in 1 2 3 4 5 6 7 8 9 ; vboxmanage startvm empty1 ; vboxmanage startvm empty2 ; vboxmanage controlvm empty1 poweroff ; vboxmanage controlvm empty2 poweroff ; done")

3. See host crash after a few cycles.

Notes:

1. This is not a realistic use case, but I hit this (or a similar-symptom) crasher in actual use with two Ubuntu guests, with a script automatically stopping them at about the same time (one shell process issues "vboxmanage controlvm name poweroff" for both VMs in succession). I marked this "critical," hope you agree.

2. Crash does not reproduce here if the IDE bus itself is removed from both VMs; could be the bus is implicated or just that the timing changes.

3. Host OS is OSX 10.7.3

I'll attach a system crash report.

[swork]

Post the bug and symptoms change. Host crash hasn't reproduced, though startup hangs the script at the same point crash had been occurring:

Waiting for VM "empty1" to power on...
VM "empty1" has been successfully started.
Waiting for VM "empty2" to power on...
VM "empty2" has been successfully started.
0%...10%...20%...30%...40%...50%...60%...70%...80%...90%...100%
0%...10%...20%...30%...40%...50%...

I haven't yet identified what's different now from the trials that led me to post the bug in the first place. (I also have no system crash report to post yet.)

[swork]

OSX system crash report following repro

[swork]

Succeeded in reproducing; just took more attempts than I experienced initially (about a dozen runs of the script fragment; also note missing "do" before first "vboxmanage"). I've attached the resulting system crash report. Happy to supply more information and to retest as needed.

[Frank Mehnert]

Most likely fixed with VBox 4.1.22.