Bug 1412611

Summary:	NSS should use P-256 for TLS by default, not new curves
Product:	Red Hat Enterprise Linux 7	Reporter:	Kai Engert (:kaie) (inactive account) <kengert>
Component:	nss	Assignee:	Daiki Ueno <dueno>
Status:	CLOSED ERRATA	QA Contact:	Hubert Kario <hkario>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	7.4	CC:	dueno, hkario, rrelyea, szidek, tmraz
Target Milestone:	rc
Target Release:	7.4
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	nss-3.28.3-4.el7	Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2017-08-01 16:50:07 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Kai Engert (:kaie) (inactive account) 2017-01-12 12:40:53 UTC

Hubert reported, the current behavior of NSS 3.28.x isn't ideal for our compatibility needs, and has suggested a change to use curve P-256 by default.

We'll need that fix (see upstream bug) for RHEL 7.4.0 which will pick up a new softokn that supports additional curves.

Comment 1 Tomas Mraz 2017-01-12 13:07:23 UTC

The other option is to change the ciphersuite selection so when client does not send a group list the server chooses non-ecc ciphersuite. That's what openssl will do in future releases.

Comment 2 Kai Engert (:kaie) (inactive account) 2017-01-12 13:27:11 UTC

Tomas, thanks for your suggestion. The implementation discussion is happening upstream. We either have to forward your suggestion, or you could participate directly in the upstream NSS bug. Preference?

Comment 3 Tomas Mraz 2017-01-12 14:21:05 UTC

I'll add a comment to the upstream bug.

Comment 4 Tomas Mraz 2017-01-12 14:23:49 UTC

Hmm, I actually see that on the upstream bug the option to treat missing supported groups extension as no support for ecc is already discussed.

Comment 5 Kai Engert (:kaie) (inactive account) 2017-01-13 11:51:04 UTC

The upstream bug has received a patch, so I think it's possible to backport it to the version we'll use in RHEL 7.4.0, adding devel-ack+

Comment 6 Hubert Kario 2017-01-13 11:57:37 UTC

tlsfuzzer test cases implicitly test this, since current version doesn't support X25519, but writing a test case that expects P-256 in case no supported_groups are sent is trivial

Comment 7 Kai Engert (:kaie) (inactive account) 2017-02-24 14:14:22 UTC

Daiki, could you please backport the upstream fix for 7.4.0 ?

Hubert, can you please help with qa-ack?

Comment 9 Daiki Ueno 2017-03-15 09:54:27 UTC

The upstream patch has already been backported to the 3.28 branch and included in the 3.28.2:
https://hg.mozilla.org/projects/nss/rev/970f503f579e

So rebasing to 3.28.3 should fix the issue.

Comment 12 errata-xmlrpc 2017-08-01 16:50:07 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2017:1977