Perhaps time to make sure entire kernel & all modules are built from scratch, without reusing old obj files, and without ccache -- just to make sure it's a clean slate?

Patch has been widthdrawn as lang/phantomjs doesn't build on 12/13, and phantomjs is no longer maintained (see: https://github.com/ariya/phantomjs/issues/15344). The fork (https://github.com/vitallium/phantomjs/) used by this port seems abandoned as well.

Missed some changes in the files/ directory. Testing the created patch before updating.

Made code a bit more readable by storing jid as a variable, instead of double ?: syntax.
Fixed trailing whitespace.

In D18618#397024, @mjg wrote:

See this review https://reviews.freebsd.org/D27

Added jid to log message for sigexit as well.
Made formatting more consistent (removed a colon)

Fix whitespace

Reminder to whomever commits this: PORTREVISION is a moving target and should be updated properly when committing.

Looks good to me.
I added port maintainer (Marius Halden) to reviewers list as well.

Some minor corrections. Turns out it helps if I run tests on the right tree. :)

More trailing whitespace

Update according to feedback

I've not cycled back to this diff, and probably won't get to do the research necessary to document the remaining MIBs. Can this revision be committed as-is?

Closing this revision, apparently one already exists from a year ago. D12476

This was superseded by/included in rS329853

Was committed in rS323390. (This comment is only to close the revision)

Just adding that I've run the test procedure by creating a custom installation media and installing to a VM. Ticking the PID randomization checkbox works as intended.

In D12433#257843, @kristof wrote:

This builds on r323390, right?

It builds on the previous work committed in rS323390, yes.

Since the upper bound of modulus was picked somewhat randomly, change it so the math is simpler. (suggested by des@)

Fixed bug introduced in previous revision: setting kern.randompid=0 sets it to 100
Restructured if-then-else tree to make slightly more sense while here. (feedback from des@)

Implement random modulus with kern.randompid=1
Fixed some incorrect assumptions (casting+abs()) and using cross-cast instead of modulus for more randomized results.

In D12238#254056, @des wrote:

BTW, 100 is a reasonable lower bound, but perhaps pid_max - 100 would be a better upper bound than 999, since it is already the maximum for an explicitly selected value.

This patch doesn't implement PID randomization; that's already in. This patch just makes the sysctl value "1" generate a random modulus, instead of leaving the feature disabled.

Want feedback on: Is the number range 100-999 sane?

This diff now also includes changes to examples, and new examples.

Fixed missing line, and ordered includes properly in libifconfig_internal.c.

Fixed another nit that got broken in previous fix. :)

Fixed minor nit, and fixed up examples as well.

danfe: Thanks for the feedback on style(9). I will address it in the upcoming style(9) fix, no need to submit the feedback again. :)

Patch now applies against lib/libifc and share/examples/libifc instead of attempting to move the files.
Before commit, please use 'svn mv' as follows to rename things properly:

Added missing $FreeBSD$ tags

Now tells lib/Makefile to build libifc, so it actually gets built with 'buildworld'
Disabled building of shared library for now, to further state this is not production ready.

Removed libifc_ioctlwrap_caddrt as it was not necessary.
'struct ifreq ifr' is now always initialized by overwriting it with 0's.
Makefile now defines WARN?=6 to fault on all warnings. (Warnings are future "gifts" we don't want)

Fixed a few missed renames.

The only change in this revision is the library has been renamed from "libifconfig" to "libifc", to be in sync with the prefixes and what people are actually calling it. Uploading it so it will be easier to track changes between revisions here.

Addressed comments.

Addressed feedback.

Addressed feedback

Removed things that are not related to FreeBSD base.

Simplified language as suggested by reviewers

sysctl MIB descriptions now end without punctuation, to be in line with how this is generally done. Also removed ending period of some existing descriptions.

Please do not commit this yet. Out of a couple thousand sysctl's, only about 130 had a description ending with a period. Although I did not filter out the ones without a description, I still think this is telling.

Addressed inline comments.

in ifconfig.c, the function setifmetric(...) use "sizeof (ifr.ifr_name)", while the function ifconfig(...) line795 and 812 use "sizeof ifr.ifr_name".

Updated revision addressing inline comments.
Moved Warren Block and Ed Maste from Subscribers to Reviewers.

Updated revision to have proper context

Renamed 'echoifname' to 'printifname' to be more consistent.
New behaviour: The 'name' command will now trigger printing of new interface name if the rename is successful. Example: "ifconfig tap0 name blah" will echo "blah" on success, and not echo anything on failure.

I'm submitting this revision to have the current state of the work on this in the review. I've researched and documented more MIBs, grouped under the 'Congestion Control' sub-header. The MIBs listed under the Misc sub-header have not been updated yet. MIB names will be formatted with underline to match the TCP(4) man page. MIBs will gain the underline formatting as their descriptions are updated or considered OK.

I'm sorry about not including context; will do that with next revision.

The SCTP subsection felt awkward, so I've reworded it to flow better for the reader.

Corrected synopsis to state the sctp.blackhole MIB has valid values of 0, 1 and 2.

Fixed a stray space in synopsis Brq parameter.

Improved formatting of synopsis according to wblock's feedback.
Added subsections to the DESCRIPTION section for SCTP, TCP, and UDP
Moved the SCTP subsection to the top, so that the subsections are sorted alphabetically
Phrases which list SCTP, TCP, and UDP together are now consistent in listing them alphabetically
Moved the 'summary' paragraph from the bottom of DESCRIPTION to directly before the first subsection.

Updated to address feedback from wblock.