Bug 231181

Summary:	random(4): Error transitioning from unseeded to seeded state
Product:	Base System	Reporter:	Conrad Meyer <cem>
Component:	kern	Assignee:	Xin LI <delphij>
Status:	Closed FIXED
Severity:	Affects Only Me	CC:	rgrimes
Priority:	---	Flags:	rgrimes: mfc-stable11+ delphij: mfc-stable10-
Version:	CURRENT
Hardware:	Any
OS:	Any
Bug Depends on:
Bug Blocks:	228911

Description Conrad Meyer freebsd_committer

2018-09-05 16:56:31 UTC

Reported here:

https://lists.freebsd.org/pipermail/freebsd-current/2018-September/071067.html

Review out here:

https://reviews.freebsd.org/D17049

Comment 1 Conrad Meyer freebsd_committer

2018-09-05 16:58:07 UTC

Suggest blocker status for 12.0 release, but I am not tied to it (relatively uncommon to be in an unseeded state and such configurations *could* work around it).

Comment 2 Conrad Meyer freebsd_committer

2018-09-05 17:00:35 UTC

Formally approved by secteam (delphij@), vangyzen@ in phab.  Tagged re@ for approval in freeze.

Comment 3 Rodney W. Grimes freebsd_committer

2018-09-05 19:39:16 UTC

(In reply to Conrad Meyer from comment #2)
You need to send an official RFA to re@ if you want to commit this, tagging in phabricator does not get you there.

Thanks,
Rod

Comment 4 Conrad Meyer freebsd_committer

2018-09-05 19:55:06 UTC

(In reply to Rodney W. Grimes from comment #3)
I am happy to commit before thaw or after thaw -- choice is re's.

Comment 5 commit-hook freebsd_committer

2018-09-09 17:12:45 UTC

A commit references this bug:

Author: delphij
Date: Sun Sep  9 17:12:31 UTC 2018
New revision: 338542
URL: https://svnweb.freebsd.org/changeset/base/338542

Log:
  random(4): Squash non-error timeout codes from tsleep(9).

  In both scenarios a timeout (EWOULDBLOCK) is considered as a
  normal condition and the error should not pop up to upper layers.

  PR:		231181
  Submitted by:	cem
  Reported by:	lev
  Reviewed by:	vangyzen, markm, delphij
  Approved by:	re (kib)
  Approved by:	secteam (delphij)
  Differential Revision:	https://reviews.freebsd.org/D17049

Changes:
  head/sys/dev/random/randomdev.c

Comment 6 Conrad Meyer freebsd_committer

2018-09-20 17:31:04 UTC

It is no longer blocking 12.0.  Should the blocking relationship be removed, or the bug closed?

Comment 7 Rodney W. Grimes freebsd_committer

2018-09-20 19:36:18 UTC

(In reply to Conrad Meyer from comment #6)
If the bug is fixed the PR should be closed.

Comment 8 Conrad Meyer freebsd_committer

2018-09-20 20:09:46 UTC

(In reply to Rodney W. Grimes from comment #7)
It's fixed on CURRENT, but not on stable.

Comment 9 Rodney W. Grimes freebsd_committer

2018-09-20 20:29:41 UTC

@delphij is there some reason you did not mark your commit as MFC:?

Comment 10 Conrad Meyer freebsd_committer

2018-09-20 20:55:47 UTC

The primary bug has lower impact on 11 due to the specifics of the previous arc4random(3).  And the second tsleep was not introduced until after 11, unless someone MFC'ed it.  I don't know if this bug is relevant for 10 at all; random's structure changed heavily in ~2015.  (Also, are there going to be any more 10.x releases anyway?)

Comment 11 Xin LI freebsd_committer

2018-09-24 21:48:46 UTC

(In reply to Rodney W. Grimes from comment #9)
I'll merge the applicable portion to stable/11.

The issue was introduced in r284959 which was not merged to stable/10.

Comment 12 commit-hook freebsd_committer

2018-09-25 05:18:30 UTC

A commit references this bug:

Author: delphij
Date: Tue Sep 25 05:18:20 UTC 2018
New revision: 338923
URL: https://svnweb.freebsd.org/changeset/base/338923

Log:
  Partial MFC of r338542:

  random(4): Squash non-error timeout code from tsleep(9).

  PR:		231181
  Submitted by:	cem
  Reported by:	lev
  Reviewed by:	vangyzen, markm, delphij
  Approved by:	secteam (delphij)
  Differential Revision:	https://reviews.freebsd.org/D17049

Changes:
_U  stable/11/
  stable/11/sys/dev/random/randomdev.c