Linux kernel compatibility support
Details
Today
Abandoned, and moved to; https://reviews.freebsd.org/D61901
Yesterday
folio_page() and folio_file_page() is already committed in D61217.
folio_mapped() and folio_unlock() looks good.
folio_lock() have to be changed. IMO to vm_page_busy_acquire(&folio->page, 0); acording to my previous comment.
Fri, Nov 20
Mon, Nov 9
Thu, Nov 5
Revision (and commit ?) title is wrong :)
Wed, Nov 4
I tested the patch with my amdgpu GPU and it works fine.
I should add that I am using this on my daily driver which also runs amdgpu and have not noticed anything wrong.
Tue, Nov 3
Update following bz@'s comment.
Mon, Nov 2
I'd suggest to reduce it to this.
The ath11k explanation in the commit message is for ath11k but not generic LinuxKPI. There's a further issue if the commit message is correct to be dealt with as the native PCI stack, I believe tracks MSI state (also for resume) and may not be able to do so?
Adding @dumbbell as msix will change behaviour for amdgpu I believe; msi is indeed ath11k only currently. I'll deal with this as soon as I have stable internet again (as one notices as the submit didn't work the first time).
nice catch! bz, what do you think?
Oct 20 2026
Seems to fix the problem for me, thanks. I tested 6.6-lts, 6.12-lts, each with wayland and with both the modesetting and legacy intel Xorg drivers.
Thanks for the feedback. I'll wait for @markj
It restored functionality of i915kms recently broken on CURRENT
Oct 19 2026
Oct 17 2026
Oct 16 2026
+Ka Ho (mentor)
