I think I've found a critical bug in latest VBox 5.2.8. It causes a *host* computer BSOD when booting a VM using a physical E2B drive attached to host computer using the raw VMDK/Virtual Machine USB Boot tool trick.
Used E2B is updated to latest v1.98 and it boots the host physical computer as well as the same VM (if previous VBox versions up to 5.2.6 are used) without issuess.
I've narrowed the problem down to triggers being 2 lines (if commented out with # then no BSOD) in \_ISO\e2b\grub\menu.lst:
dd if=(md)0x3000+0xA0 of=(md)0xa000+0xA0 > nul
and line 575
cat --locate="\ntimeout " --replace=\n# (md)0xa000+0xA0
Both lines try to access memory at 0xA000 and seem related to the FASTLOAD feature. Other 3 lines access memory at that offset previously (443, 444 and 451) but I've not tested their impact because they are not executed if FASTLOAD is not enabled by user (I don't use it). However lines 548 and 575 are unconditionally executed.
This is clearly a bug in latest VBox (current and previous E2B works like a charm on real hardware and previous VBox versions), but I would like to get bug confirmation from you if possible since opening a ticket at VBox bug tracker is going to requiere some cumbersome account creation.
Thanks in advance.
Subject Possible bug in latest VBox 5.2.8 triggered by E2B menu.lst