I didn't work on the implementation of dma-buf for FreeBSD. We have this BSD-licensed implementation in drm-kmod right now and we can surely move that to base. I find it receives changes infrequently in Linux, so hopefully, that shouldn't be something that slows down or interferes with making a given version of drm-kmod compatible with multiple versions of FreeBSD.

In D58320#1341246, @bz wrote:

Note: I am still confused as to why we do all this checking manually there instead of simply call vm_page_unwire() of which we seem to duplicate the logic?

this is 6.14? So there's a new case that we didn't know about during the time we fixed master (before/after 6.12?) or had I missed this before?

Oops, I forgot to mark the review as needing changes.

I made an inline comment in linux_free_page(): there is a refcounting issue vs. the call to lkpi_page_unset_vm_page().

The patch looks good to me.

Indeed, I kind of remembered something similar but a quick search didn’t return this review. Thank you!

Address comment from @emaste by uncommenting the include of <linux/plist.h> and adding it as a dummy header.

This patch looks good to me. I just noted one comment which is unclear to me.

Fix typo, reported by @emaste

Thanks for the explanation!

I agree with @bz about adding printed TODO when called. I’m still accepting this patch because it should be part of another review probably.

What are the use cases for this macro?

Instead of this hack around <linux/radix-tree.h> and <linux/xarray.h>, I implemented the following solution in drm-kmod:
https://github.com/freebsd/drm-kmod/pull/477