Update config.strip for amd64 boot failure in VirtualBox #32
Conversation
This would most likely fatten up the kernel too much to be accepted, but I wanted to put it out for posterity. I spent the time going through line-by-line and group-by-group (ugh) to see which options from radeondrm onward cause a crash during kernel load in VirtualBox 4.3.26. I also opened a ticket in the hopes that some of the problem is on their end, with a testable disk image: https://www.virtualbox.org/ticket/14035 Of note: None of these changes are needed to boot i386 in the same VirtualBox configuration. Weird.
|
It seems apparent the real problem is somewhere else in the kernel. Also, funny enough, these changes might break amd64 on real hardware. I don't know if you've tried that? |
|
Well the problem is still a problem. The diagnostics from Frank give us a way to fix the problem. I think the IDT setup needs to be addressed. Can you peek at the kernel some more in this regard? The FLASHRD kernel configs continually expose this problem. I think it's a generic problem that is layout dependent???? |
|
This is mostly getting above my head now, but I'm happy to try. Will have some free time next week. |
|
We are not alone. Doo doo dooo. https://www.mail-archive.com/misc@openbsd.org/msg138735.html |
|
I believe the issue you faced here was fixed here in the 5.8 release: However, this guy also says that he no longer saw the issue as of Jul 10th and reiterated it was fixed for "all sizes" of ramdisks as of Jul 14th, a month before these fixes were added to the tree. I suspect your specific issue is fixed by Mike Larkin in these commits. But i'm also going to merge your patch because it fits in with my original intent; to make a largely GENERIC kernel available to users, because Mike's patch actually fixes the original problem that I was trying to solve the wrong way. |
This would most likely fatten up the kernel too much to be accepted, but I wanted to put it out for posterity. I spent the time going through line-by-line and group-by-group (ugh) to see which options from radeondrm onward cause a crash during kernel load in VirtualBox 4.3.26.
I also opened a ticket in the hopes that some of the problem is on their end, with a testable disk image: https://www.virtualbox.org/ticket/14035
Of note: None of these changes are needed to boot i386 in the same VirtualBox configuration. Weird.