Context Navigation

Changes between Initial Version and Version 1 of Ticket #20468, comment 1

initial	v1
3	3	1. In the VirtualBox forum thread, I've requested and analyzed several VBox.log files, created by VirtualBox 6.1.22, especially with 2\|3\|4 vCPUs. At the time of the Guru Meditation 1155 (VINF_EM_TRIPLE_FAULT), all vCPUs except vCPU0 were in a similar state, either in real mode or 32-bit protected mode, with EIP not pointing to a meaningful instruction. The "crashing" vCPU<n> seemed to be one step ahead of the others each time (EDI seemed to contain an offset or counter), for <n> not necessarily being 1, and the last VM-exit often was a regularly occurring "MMIO access".
4	4
5		2. In the VirtualBox 6.1.24 VBox.log file above, the log messages are very similar to those in the VBox.log files in the VirtualBox forum thread. One notable difference are the RIP values in the VM-exit list, which contain 20-bit and 32-bit addresses, whereas in the older VBox.log files I've only seen 20-bit addresses.
	5	2. In the VirtualBox 6.1.24 VBox.log file above, the log messages are very similar to those in the VBox.log files in the VirtualBox forum thread. One notable difference are the RIP values in the VM-exit list of vCPU1, which contain 20-bit and 32-bit addresses, whereas in the older VBox.log files I've only seen 20-bit addresses.
6	6
7	7	3. In the VirtualBox 6.1.24 VBox.log file above, the VirtualBox GA were updated just before the reboot, which could have influenced the problem. In the older VBox.log files, this wasn't the case.