Messages in this thread | | | Date | Sun, 23 Dec 2007 00:04:38 -0500 | From | Jeff Garzik <> | Subject | Re: [patch] Make MMCONFIG space (extended PCI config space) a driver opt-in issue |
| |
Linus Torvalds wrote: > I want to limit that downside. Right now, the easiest way to limit it > seems to be to say that those (very very few) drivers that actually care > could enable it. That way, we automatically limit it to only those > machines that have hardware that cares.
Then let's do it right: disable mmconfig by default on x86, and enable it when passed "pci=mmconfig".
For the rare -- you and I agree its very rare -- case where it is REQUIRED, the user can pass pci=mmconfig as instructed by driver documentation somewhere.
Let's not bend over backwards and introduce an API for these presently-theoretical cases. Given the complete lack of hw vendor testing and potential to confuse userspace, the two choices for a computer should be "mmconfig off" or "mmconfig on."
Kernel hackers developing drivers and code for new machines will know enough to pass pci=mmconfig if they NEED it. That practice will only become annoying when x86 hardware actually starts to NEED extended config space -- at which future time we can revisit, as you describe.
> And yes, if you want the capability following to notice automatically when > capabilities really do go into the 0x100+ range, that's fine. I suspect
Yes, we /must/ do this checking, if we don't already.
Jeff
| |