Skip to contentSkip to conversation

Paravirt_ops for x86_64

PATCH In review Steven Rostedt 2 participants 3 messages Mar 8, 2007 virtio core

lore
Share & tools
lore
#0 lore raw

cover Paravirt_ops for x86_64

Hi all,

Glauber and I have been playing around with getting paravirt ops working on the x86_64. We have our own version of lguest running there too. We have lguest64 running up to console_init, and then it takes an interrupt storm. I'll be posting those patches tomorrow.

Glauber and I have been sharing a git repo that isn't open to the public. (we don't want people to see how crappy our stuff is yet ;) Well, it's still kind of crappy, but it's starting to work. So I spent today, pulling out patches related only to paravirt ops from our git repo, and separating them to be readable. Well, here they are.

Tomorrow, I'll post what we got for lguest64 running.

Thanks!

-- Steve

--

#1 lore raw
Quoted 3 lines · * Steven Rostedt ([email protected]) wrote:

Well, it's still kind of crappy, but it's starting to work. So I spent today, pulling out patches related only to paravirt ops from our git repo, and separating them to be readable. Well, here they are.

Cool. I had split these up slightly differently, with core changes coming first (like places where raw asm is still left behind), followed by adding in the paravirt bits. I think it would be useful to get those simple basic changes to the front of the queue, because they can largely be merged right away w/out issue. I'll look through the lot tomorrow.

thanks,
-chris

#2 lore raw
Quoted 11 lines · Chris Wright
Quoted 3 lines · * Steven Rostedt ([email protected]) wrote:

Well, it's still kind of crappy, but it's starting to work. So I spent today, pulling out patches related only to paravirt ops from our git repo, and separating them to be readable. Well, here they are.

Cool. I had split these up slightly differently, with core changes coming first (like places where raw asm is still left behind), followed by adding in the paravirt bits. I think it would be useful to get those simple basic changes to the front of the queue, because they can largely be merged right away w/out issue. I'll look through the lot tomorrow.

The way I split it, and what took me so long to split it, was that I wanted each patch to be able to compile. There was one two-way dependency, I think that was between patches 7 and 8. But the most part, each addition of a patch allowed for a successful compile. I didn't test running those though ;)

I can work on splitting it better. After pushing out the lguest64 patches (coming soon).

-- Steve