In D11323#234611, @trasz wrote:

I don't have a slightest idea why the special case is required. But it might be, and removing it altogether might break somebody's system. Adding additional condition is much less intrusive, and thus safer.

More errors explanations.
Fix unrelated typo.

Honestly, I dislike the change. IMO, the msdosfs code should be copied to the userspace utility instead of making the hard-to-deal knot between kernel and userspace code.

Can you explain why this special case is required ?

Edit the description of an error from MAP_GUARD.

In D11305#234511, @alc wrote:

In D11305#234137, @kib wrote:

The rewrite of the swap pager to use radix trie for swblock tracking is still somewhere in my repo.

I'd like to find a way to reduce the amount of physical memory that will be consumed by the leaves. Otherwise, I like it.

Move note about MAP_STACK use of MAP_GUARD to MAP_STACK block.
Document EINVAL on invalid MAP_GUARD call, explicitely listing the requirements.

Ed' note.

The rewrite of the swap pager to use radix trie for swblock tracking is still somewhere in my repo.

I combined the feedback, selecting the parts from each. I also added half-sentence mentioning that stacks are not mapped fully in advance.

This looks fine to me, with the note I did in the previous accept action.

Overall, I have only one big issue with this patch, ignoring the small nits I noted inline.

In D11270#233386, @markj wrote:

Are the talk(s) you referenced the ones from cppcon?

edit: looks like this is covered in the last ~15 minutes of https://www.youtube.com/watch?v=M15UKpNlpeM

Provide the release fence in refcount_release(), instead of relying on user handling it correctly.

In fact, we can add atomic_fence_rel() before the atomic_fetchadd_int() line for the same effect.

In D11025#232133, @sobomax wrote:

Having absolute timer that expires relative to the CLOCK_MONOTONIC solves both issues cleanly and elegantly.

In D11025#232110, @sobomax wrote:

@kib how difficult is it to also implement real NOTE_MONOTONIC, i.e. event that expires in monotonic time from the getgo? I think this would be very useful for real-world applications that care about precise timekeeping. We use CLOCK_MONOTONIC pretty much all across our code here. Yes, it can be added later on, but without it the API is somewhat incomplete IMHO.

Update patch to the current state, including fixes to PT_GET_SC_ARGS (it passes tests) and to scescx.c test itself.

Fix make universe. Changes are mostly in tests, in particular, NetBSD tests customizations are no longer needed because event.data becomes 64bit.

In D11146#230856, @alc wrote:

I'd like to add something else to this patch. Specifically, I'd like to remove the field "bl_free" from struct blmeta. While it is correctly maintained, it is never used. Moreover, unless the entire tree is a single leaf, the field "u.bmu_avail" at the root provides the count of free blocks in the tree.

This is not a review, I did not read the code for real.
Still, I noticed several somewhat big issues, with the object relock in page fault handler, with unused newbus attachment, and with headers.
I also noticed a lot of style bugs, for which I mostly pointed out one occurrence, to not make notes unreadable. Please fix them all.

Document PT_GET_SC_ARGS.