Paths

Table of Contentst

* scripting for random(4)
ClosedPublic
Actions

Authored by markm on Jun 27 2015, 9:43 AM.

Details

Reviewers

emaste
jilles
delphij
imp

Commits

rS285288: Address review.

Summary

The new random(4) needs to have its etc/rc.d/* scripting tightened up.

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

markm updated this revision to Diff 6496.Jun 27 2015, 9:43 AM

markm retitled this revision from to Fix up the startup/shutdown etc/rc.d/* scripting for random(4).

markm updated this object.

markm edited the test plan for this revision. (Show Details)

markm added reviewers: delphij, emaste, imp.

markm added a parent revision: D2025: Tidy up random(4) driver after developer feedback and code audit..

The idea is to commit this when I do D2025 on July 30, 2015.

jilles added a subscriber: jilles.Jun 28 2015, 8:31 PM

jilles added inline comments.

etc/rc.d/random
48 ↗	(On Diff #6496)	This expands to twice `sysctl -n`.
share/man/man5/rc.conf.5
4057 ↗	(On Diff #6496)	If `/boot` is a separate filesystem, then this may well be the only thing writing to it regularly. This might affect reliability negatively.

markm added a reviewer: jilles.Jun 28 2015, 9:45 PM

markm added inline comments.

etc/rc.d/random
48 ↗	(On Diff #6496)	Oops! Will fix!
share/man/man5/rc.conf.5
4057 ↗	(On Diff #6496)	I'm not sure I understand you completely here. Do you mean this may be the only thing stashing entropy regularly, and this will therefore affect random(4) reliability? If so, I'm not sure you are right; there is a cron job adding stuff to /var/db/entropy/ which I intend to extend. Do you mean that this may be wearing out /boot by excessive writing? If so, is one file write per reboot really that onerous? If so (again), could it go somewhere else?

jilles added inline comments.Jun 28 2015, 10:15 PM

share/man/man5/rc.conf.5
4057 ↗	(On Diff #6496)	I mean that this very slightly increases the risk of /boot getting corrupted. A rather minor issue given the importance of /. Also, /boot might be read-only, especially in systems that are more like embedded systems.

markm added inline comments.Jun 28 2015, 10:41 PM

share/man/man5/rc.conf.5
4057 ↗	(On Diff #6496)	Erm, OK, the system owner does have override powers, including the ability to disable its altogether if they wish. Is this acceptable? This reminds me - I need to ensure the loader scripting is OK too.

Fix script error.

Update default loader.conf.

Update man pages.

markm marked 2 inline comments as done.Jun 29 2015, 8:01 AM

markm added inline comments.

share/man/man5/rc.conf.5
4057 ↗	(On Diff #6532)	Can we leave this a system-owner configurable?

markm mentioned this in rS284959: Huge cleanup of random(4) code..Jun 30 2015, 5:01 PM

jilles added inline comments.Jul 5 2015, 4:16 PM

etc/rc.d/random
127 ↗	(On Diff #6532)	The `''` part can't happen because of the ${var:=value} expansion. Either remove it or use ${var=value} instead of ${var:=value}.
share/man/man5/rc.conf.5
4057 ↗	(On Diff #6532)	Yes.

Address review comments.

etc/rc.d/random
127 ↗	(On Diff #6532)	Fixing similar elsewhere.

Closed by commit rS285288: Address review. (authored by markm). · Explain WhyJul 8 2015, 6:46 PM

This revision was automatically updated to reflect the committed changes.

Revision Contents
Changeset List

Path

Size

head/

etc/

rc.d/

random

8 lines

Diff 6789

View Options

Fix up the startup/shutdown etc/rc.d/* scripting for random(4)ClosedPublicActions