Whoops, this should be committed already according to https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=21519#c18 .

Properly initialize zstream structure.

+markj for DTrace.

Looks mostly fine to me. I think you don't really need to set zalloc to zcalloc_nowait as the default zcalloc is zcalloc_nowait.

Rebase over D21156.

Thanks! I think I'll commit this at some point this week (doing a tinderbox build now).

Fix compile options for zlib crc32.c.

Fix typo.

Good catch!

Please take another look.

Address reviewer comments.

Will commit after a tinderbox build.

Submitted as rS350496.

Update license for zcalloc;

Modified zlib_mod.c to use standard FreeBSD Foundation license.

+@mav @avg @mmacy for ZFS portion of change as a FYI (sys/cddl/contrib/opensolaris/uts/common/zmod). TL;DR: the bundled zlib (1.2.3) was removed in favor of new zlib; alloc/free functions merged from zmod_subr.c into zmod.c.

+@markm (Please confirm if the license is Okay for sys/dev/zlib/zlib_mod.c, the file is largely based on your change rS130799 in 2004 for sys/net/zlib.c).

Add license plate for sys/dev/zlib/zlib_mod.c.

Rebase.

+kib@

In D21033#456475, @markj wrote:

In D21033#456428, @delphij wrote:

What would happen for existing binaries that uses this?

They would fail to compile. But they are trivial to fix, since the users of the cap_random capability can be directly converted to use arc4random_buf().

What would happen for existing binaries that uses this?

LGTM, thanks!

Thanks! Please see my comment inline.

Thanks! This looks mostly good to me. I intend to commit this version with some minor tweaks after some testing in a day or two, so please yell if you don't agree with some change proposals (commented inline).

I think I have accepted back in April but let's make it an explicit one. Be sure to update the dates though.

In D20698#447294, @mat wrote:

I don't know about this, I think it is better to send an email that is ignored than silently doing things.

Committed as rS349151

For the microoptimization: Note that it's possible to avoid the additional branching by doing unlock in both cases, e.g.:

Actually, the delay/sleep is probably not necessary (releasing the lock is sufficient for another thread to race in).

LGTM but please amend the wmsg message to not exceed 6 characters (see comment inline) prior to commit.