Bug 15174

Summary:	kernel 3.14.31 breaks adjtimex which breaks chrony
Product:	Mageia	Reporter:	David Walser <luigiwalser>
Component:	RPM Packages	Assignee:	Thomas Backlund <tmb>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	Normal
Version:	4
Target Milestone:	---
Hardware:	i586
OS:	Linux
Whiteboard:
Source RPM:	kernel-3.14.31-1.mga4.src.rpm	CVE:
Status comment:

Description David Walser 2015-01-31 21:12:38 CET

Confirmed this problem on 2 laptops so far that I've installed 3.14.31 on (desktop i586 in both cases).  I get an error like this and chronyd won't run:
systemd[1]: Starting NTP client/server...
chronyd[2869]: chronyd version 1.29.1 starting
chronyd[2869]: Linux kernel major=3 minor=14 patch=31
chronyd[2867]: adjtimex failed for set_frequency, freq_ppm=9.8644e+00 scaled_freq=-9.8644e+00 required_tick=10000
systemd[1]: chronyd.service: control process exited, code=exited status=1
systemd[1]: Failed to start NTP client/server.
systemd[1]: Unit chronyd.service entered failed state.


chrony works fine on the previous 3.14.29 build.

Reproducible: 

Steps to Reproduce:

Comment 1 David Walser 2015-01-31 23:03:58 CET

Also confirmed on my PogoLinux workstation with the server kernel.

Comment 2 Thomas Backlund 2015-02-01 10:47:43 CET

Ok, this seems to trigger the security hardening fix that went into 3.14.31 (and upcoming mga5) that now validates userspace inputs to protect from buffer overflows.

seems chrony has also been using a private copy of timer headers, so any changes to kernel headers have been missed.

http://git.tuxfamily.org/chrony/chrony.git/commit/?id=fc73accfe5bc0f8ffaf78730da2c46a7f1de7acc

If you feel up for it you could try to rebuild crony with this and see if it changes anything...

I will check it out better when I get back home...

Comment 3 David Walser 2015-02-01 16:01:50 CET

Thanks, I rebuilt chrony with that patch and it didn't change anything; same error.

Comment 4 Thomas Backlund 2015-02-02 15:56:24 CET

Ok, it's currently being debated upstream to what the proper fix is ... I think we'll have a solution soon... I will hold back this update some days and if no solution I will revert the offending fix temporarily...

Comment 5 Thomas Backlund 2015-02-03 01:07:38 CET

A kernel-3.14.31-2.mga4 with a potential fix is currently building...

Comment 6 David Walser 2015-02-03 14:13:21 CET

Much better.  Thanks!

Status: NEW => RESOLVED
Resolution: (none) => FIXED