Add a lot of missing END directives. Tinderbox passes on arm64 and i386.
PowerPC mostly passes, except for all the PS3 HV calls which still need END
directives (and there are hundreds of these).

Move common ossl files to conf/files.

Yeah, seems fine to me.

the env fix was committed with https://github.com/freebsd/freebsd/commit/5afe096970b19e2d3bc3b44a1a09ae788e33795c

How about something like:

<para>Spell-checking with fully-featured dictionaries is encouraged, and can
  be accomplished by using <package>textproc/hunspell</package> or
  <package>textproc/aspell</package> combined with
  <package>textproc/en-hunspell</package> or
  <package>textproc/en-aspell</package>, respectively.</para>

In D27385#611796, @0mp wrote:

Looks good to me, although I'd probably replace ISO-8859-1 with UTF-8 to keep the example a bit more modern. ;)

• Move cfi_startproc in ELFv1 prologue after DOT_LABEL

In D26011#611502, @mjg wrote:

In D26011#611477, @markj wrote:

In D26011#611350, @kib wrote:

That said, IMO adding a new msleep flag to indicate that the PCATCH-ed sleep is billed to kernel, and use it to mark exceptions in pipelock() and perhaps NFS seems to cause less churn.

The current approach is a lot of churn but seems easier to maintain:

• It's rare to add new USERWAIT points, I think, and more common to add ad-hoc synchronization. It's not just pipelock() that needs to be modified. The sleep in callout_drain() for example needs annotation because the caller is blocking while waiting for another thread to do work.
• If a new USERWAIT point is added and the developer forgets to annotate, it will show up in profiles and can therefore easily be fixed.

So, regardless of pipelock, I consider PCATCH as too much of a hammer to implicate waiting on user.

The churn in the patch is not significant and is mostly mechanical, more importantly to expand your second point, it does not accidentally hide any sleeping. With manual annotation at worst some points will get it when they should not have, but none will be hidden from the get go.

Looks good to me, although I'd probably replace ISO-8859-1 with UTF-8 to keep the example a bit more modern. ;)

In D27386#611787, @debdrup wrote:

Would it make sense to mention hunspell/aspell too? While igor does do some spellchecking, it does so on a rather limited set of words defined in the init_spellingerrors function.

Not familiar with the code, but it looks ok to me.

Would it make sense to mention hunspell/aspell too? While igor does do some spellchecking, it does so on a rather limited set of words defined in the init_spellingerrors function.

I just posted https://reviews.freebsd.org/D27387 for i386/ppc/arm64 as well.

Fix space between filename and ':'

In D27177#611740, @cem wrote:

We should probably do something similar for arm/mips/i386/ppc. (Of course, I am a hypocrite here; I did the same for only amd64 in rS353959.)

We should probably do something similar for arm/mips/i386/ppc. (Of course, I am a hypocrite here; I did the same for only amd64 in rS353959.)

This patch seems to complicate things at the first glance.
But in the end, the sowakeup() call itself is the problem.
It requires to be called while holding the lock and releases it.

Hmpf, this ran into the same issue as last time ...

In D5986#611427, @dsl_mcusim.org wrote:

There's some information here if you haven't found it already: https://wiki.freebsd.org/Phabricator

No, I haven't seen it yet - thanks for sharing! What's a preferred way to create a revision: via command line or web interface?

In D27376#611605, @kib wrote:

I mean, either MSDOS_EXFAT option should go out, or for testing period I propose to build two modules from the same code base. I do not propose to copy sources. Second module would have -DMSDOS_EXFAT added to CFLAGS, and perhaps should rename VFS_SET() so that resulting vfsconf is named differently and can be used simultaneously with msdosfs.ko.

Ignoring the test case (but thanks for adding it) this looks good to me.
Thank you for bringing it back so quickly.

LGTM
Thanks