Separate lastrevision into multiple, more structured fields in Makefile.
Remove perl from the mix.
Add tail command to doc.commands.mk.

More updates based on @wblock's input.

Changes based on review.

Add two missing environment variables.

Also document the fact that SSLv3 is disabled by default now.

@jhb: Please see D3351

@jhb: Please see D3351

Yep, that did the trick.

I see what happened - the patch in question uses MSDOS CR-LF, it seems like those got stripped on commit. The easiest fix is probably to strip \r from all sources before patching.

Will take a look.

Yes, please see link in "Test Plan"

Add missing dependency on php5-zlib

Fix pkg-plist

Not sure if using %%WWWOWN%% in the path name is a great idea (just realized this, was created using make makeplist).

Reduce pause in init (spec sheet asks for 1ms or more)

Merge latest changes, add BUGS section to man page, minor clean-ups

In D2811#64102, @eadler wrote:

In D2811#64067, @grembo wrote:

In case anyone wonders why I committed the same patch 10 times: Arcanist is showing this on arc diff --update:

/tmp//edit.5nq6p1rhpvs44s4s/differential-update-comments: 10 lines, 359
characters. Linting...
 LINT OKAY  No lint problems.
Running unit tests...
No unit test engine is configured for this project.
 SKIP STAGING  Phabricator does not support staging areas for this
repository. Exception
ERR-CONDUIT-CORE: Include of
'/usr/local/www/phabricator/phabricator/src/extensions/herald/customaction/SetTaskViewPolicyHeraldCustomAction.php'
failed! (Run with `--trace` for a full exception trace.)

this was fixed and shouldn't happen again. We do get reports when this happens, but someone needs to be awake to deal with them :-)

In case anyone wonders why I committed the same patch 10 times: Arcanist is showing this on arc diff --update:

This is committed now, thanks again to @kib and @wblock for their most helpful reviews, I hope it didn't consume too much of your time.

In D3068#63930, @kib wrote:

In D3068#63711, @imp wrote:

Is there some reason not to compile it on other architectures? It's a generic i2c device, from all appearances. The fewer crazy if statements like this that we have the better. We already have *WAY* too many of them.

Note that the driver assumes that whatever device is found on the given iic bus at the address 0x67, it is his own. This might have quite unforeseen consequences.

In D3068#63711, @imp wrote:

In D3068#63620, @grembo wrote:

Also, I did not verified the build. Right now the driver is built on all arches, is it reasonable ?

Based on a quick check on google that trackpad is also found in some ARM devices (there's some Samsung ARM Chromebook using) , not sure if thats arm or arm64 though. I can't test it on any of those, but leaving it in would at least allow users of these machines to test and file a bug report.

That said, would this look reasonable to you:

# $FreeBSD: head/sys/modules/i2c/Makefile 93023 2002-03-23 15:49:15Z nsouch $

SUBDIR =
SUBDIR += controllers if_ic smbus iicbus iicbb iicsmb iic smb
.if ${MACHINE_CPUARCH} == "i386" || ${MACHINE_CPUARCH} == "amd64" ||
    ${MACHINE_CPUARCH} == "arm" || ${MACHINE_CPUARCH} == "arm64"
SUBDIR += cyapa
.endif

(the same would be done to man/man4/Makefile of course)

Is there some reason not to compile it on other architectures? It's a generic i2c device, from all appearances. The fewer crazy if statements like this that we have the better. We already have *WAY* too many of them.

In D3068#63643, @kib wrote:

Is the same smbus address used on other platforms ? Apparently the question of the arches and device_attach() could be related.

Also, I did not verified the build. Right now the driver is built on all arches, is it reasonable ?

Change locking in for (now while) loop

More changes based on @kib's review.

More lessons learned from cyapa review

Man page update: As @wblock pointed out that "modern" is a pretty relative
term that won't age well, I replaced it with "supported". This is more
accurate anyway and anyone dealing with this driver most likely has
a laptop that's modern by today's standards anyway.

Apply lessons learned from @kib's review of cyapa.

More changes based on @kib's code review.

Changes based on @wblock's review of the man page and
@kib's review of the driver code.

Changed the format so it looks a bit more like the original plus
gives an indication of the origin

Stop wasting inodes.

Manpage changes as requested by @wblock.

Fix use of bzero and remove unnecessary cast.

poudriere build logs are available here:

Further improvements to the port, some of which will be reported
to upstream.

Fix static initialization order fiasco in icegridnode and icegridregistry.

Changes based on @bapt's comments

https://bugs.freebsd.org/bugzilla/attachment.cgi?id=158125
https://bugs.freebsd.org/bugzilla/attachment.cgi?id=158126
https://bugs.freebsd.org/bugzilla/attachment.cgi?id=158127

There's a link to the PR under "Test plan", which has the build logs.