• Move the sleep into zone_alloc_item_locked() and put it into a loop.

As said, I'm not seeking for a review & approval so that I commit it. I'm not a ports committer. I'm seeking for a person who can finalize this ugly patch and commit. Thanks!

• Rewrite uma_zalloc_arg() so that there is a single sleeping point.

• Improve comment on why uk_lock needs to be first.

Address Mark's review.

• Move assertion to beginning of the function.
• Remove unlock avoidance in uma_zfree_arg(). Can't remember why it was

P.S. In pf itself, the dtrace probing should just substitute existing printf(9) debugging.

Committed years ago.

Merge remote-tracking branch 'FreeBSD/master' into pbuf to take
in and resolve with recent D16666 check-in.

My current patchset for UMA has something similar. I also have zone_put_bucket() although named differently, to keep accounting of all items for a zone in buckets. So this change goes in line with my changes. At Netflix we also run the vm_page_cache zone with certain bucket limit, instead of no buckets at all.

I'm not absolutely sure that all possible races are fixed. There still could be dangling ifnet pointers. But that's up to your justification. If you are sure everything is covered, feel free to remove.

Doesn't make sense to remove it from services(5) database. It isn't a database of protocols FreeBSD supports. It is just database of port numbers that are well known.

This is extremely hacky. The only proper fix is to find situations when rcvif isn't cleared for a packet that is reused for transmit and fix them.

syncache_chkrst() isn't a public function, so no worries on KBI.

Since struct inpcb is no longer exported to the userland, the spare field can be removed.

Update patch wrt newer head.

It is on my queue. The plan is to use so_error for this mark, instead of SS_ISDISCONNECTING.

Hey guys, I know that ifdefs suck, but still going to forward with checking this in. If you can suggest some change to ifdefs, I am open to that. E.g. keep them minimal, so there is no duplication between debug and !debug case. Or you may prefer to keep the maximal, so that code is more readable. I can do either, but I'd like to get forward with this change.

In D15199#320705, @markj wrote:

I think all of these inline ifdefs make the code a lot harder to read. Could we instead introduce wrapper functions for the ctor and dtor and hide the logic there?

I did not do this in the past because it was not possible to add or increment 64-bit counters atomically. Did this change? Instead of providing bad values I decided to follow the words of the RFC.

In D14980#316840, @mav wrote:

Is this API and device classification exist somewhere else, or being just designed? Differentiation between STRIPE, MIRROR and RAID sound somewhat odd to me, same as between SPINNING and CDROM. Besides it duplicates "GEOM::rotation_rate" attribute, we already have and use in some places (though obviously it is not passed through UFS).

• Support vnode backed md(4).
• Bump disk version.

Instead of using strcmp() augment struct disk with media info.