Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 528139

Summary: FTRACE: check for failure for all conversions, tracing: correct module boundaries for ftrace_release
Product: Red Hat Enterprise MRG Reporter: John Kacur <jkacur>
Component: realtime-kernelAssignee: Luis Claudio R. Goncalves <lgoncalv>
Status: CLOSED ERRATA QA Contact: David Sommerseth <davids>
Severity: medium Docs Contact:
Priority: high    
Version: 1.1.7CC: bhu, lgoncalv, ovasik, srostedt, williams
Target Milestone: 1.1.9   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: 528136 Environment:
Last Closed: 2009-11-03 18:22:07 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 528136    
Bug Blocks:    
Attachments:
Description Flags
check for failures on all modifications
none
Actually skip freed records none

Description John Kacur 2009-10-09 10:43:56 UTC 
+++ This bug was initially created as a clone of Bug #528136 +++

Description of problem:
commit ids:

3279ba37db5d65c4ab0dcdee3b211ccb85bb563f

and

e7247a15ff3bbdab0a8b402dffa1171e5c05a8e0

Are critical patches than need to be pulled in. GregKH will be pulling
them into stable releases too.

-- Steve

--- Additional comment from jkacur on 2009-10-09 06:37:23 EDT ---

Created an attachment (id=364233)
 ftrace: check for failure for all conversions

--- Additional comment from jkacur on 2009-10-09 06:38:13 EDT ---

Created an attachment (id=364234)
tracing: correct module boundaries for ftrace_release

This cloned bugzilla was created to handle backporting this for 2.6.24.7-rtx-mrgN
Comment 1 Steven Rostedt 2009-10-21 14:54:32 UTC 
Seems this code is really broken for 2.6.24-rt. Doing the following will hang the box:

# cd /debug/tracing
# echo ftrace > current_tracer
# rmmod e1000e (or any module)
# echo none > current_tracer

Then the system locks up hard in stop_machine.

Looking into it.
Comment 2 Steven Rostedt 2009-10-22 01:45:32 UTC 
Created attachment 365638 [details]
check for failures on all modifications

Looking at the old 2.6.24-rt kernel, the same code is slightly buggy. This patch adds a check on all modifications to the kernel text. It will disable the function tracer if it ever detects an issue.
Comment 3 Steven Rostedt 2009-10-22 01:47:04 UTC 
Created attachment 365639 [details]
Actually skip freed records

The module release code correctly freeded the records (unlike the bug in 2.6.31). The bug here is that that modification code ignored the free flag and processed them anyway!
Comment 4 Luis Claudio R. Goncalves 2009-10-22 20:53:38 UTC 
Patches added to kernel -135 patch queue
Comment 5 David Sommerseth 2009-10-27 18:21:22 UTC 
Verified by code review against kernel-rt-2.6.24.7-136.  Attach patches found as bz528139-001-ftrace-check-for-failure-for-all-conversions.patch and bz528139-002-tracing-correct-module-boundaries-for-ftrace_release.patch.
Comment 7 errata-xmlrpc 2009-11-03 18:22:07 UTC 
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2009-1540.html