Bug 1326944

Summary: The oprofile linux kernel driver is no longer needed
Product: [Fedora] Fedora Reporter: William Cohen <wcohen>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: unspecified    
Version: rawhideCC: gansalmon, itamar, jonathan, kernel-maint, madhu.chinakonda, mchehab
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2016-04-27 11:45:01 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:

Description William Cohen 2016-04-13 20:23:43 UTC
OProfile 1.0.0 released in September 2014 switched to exclusively using the linux perf interface and removed the user-space code that used the linux kernel oprofile driver. There is no plans to reintroduce the need for the oprofile kernel module. Thus, the rawhide kernel configuration can changed from:

CONFIG_OPROFILE=m

to 

# CONFIG_OPROFILE is not set

Comment 1 Josh Boyer 2016-04-13 20:25:49 UTC
(In reply to William Cohen from comment #0)
> OProfile 1.0.0 released in September 2014 switched to exclusively using the
> linux perf interface and removed the user-space code that used the linux
> kernel oprofile driver. There is no plans to reintroduce the need for the
> oprofile kernel module. Thus, the rawhide kernel configuration can changed
> from:
> 
> CONFIG_OPROFILE=m
> 
> to 
> 
> # CONFIG_OPROFILE is not set

Should the upstream oprofile driver be sent to staging so it can be removed in a future kernel release?

Comment 2 William Cohen 2016-04-13 20:49:37 UTC
The request to turn off the oprofile module was mainly to save some space (21KB for oprofile.ko.xz about 1.27MB debuginfo on x86_64).

Isn't the staging just for the addition of drivers in development not ready to be merged into the kernel? That seems to be the explanation in https://lwn.net/Articles/324279/ .  Is there a pointer that mentions moving driver to staging before removing?  There are architecture specific bits for the various processors, so it probably is not going to be that clean a process to move the oprofile driver code to staging.

Comment 3 Josh Boyer 2016-04-14 11:33:54 UTC
(In reply to William Cohen from comment #2)
> The request to turn off the oprofile module was mainly to save some space
> (21KB for oprofile.ko.xz about 1.27MB debuginfo on x86_64).
> 
> Isn't the staging just for the addition of drivers in development not ready
> to be merged into the kernel? That seems to be the explanation in
> https://lwn.net/Articles/324279/ .  

It started that way, but the document you point to is 7 years old.  In the meantime, it can be used for new drivers coming in to the kernel or to remove old and crufty drivers nobody maintains.

> Is there a pointer that mentions moving driver to staging before removing?  

I can't find a document, but if you look in the git history you'll see more than a dozen drivers have been removed this way.

> There are architecture specific bits for
> the various processors, so it probably is not going to be that clean a
> process to move the oprofile driver code to staging.

Hm, yeah that is probably a deal breaker.  On the other hand, getting the arch bits removed upstream if they aren't useful would probably be a good idea anyway.

I'll look at turning off the config option later today.

Comment 4 Josh Boyer 2016-04-14 18:30:12 UTC
Driver disabled in git.  Will be in a build soon.