Bug 698256

Summary:	CONFIG_TUNE_CELL should be disabled on PPC/PPC64
Product:	[Fedora] Fedora	Reporter:	Karsten Hopp <karsten>
Component:	kernel	Assignee:	Kyle McMartin <kmcmartin>
Status:	CLOSED RAWHIDE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	15	CC:	davej, gansalmon, hbabu, itamar, jonathan, kernel-maint, madhu.chinakonda, peterm
Target Milestone:	---
Target Release:	---
Hardware:	powerpc
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-06-23 12:29:35 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Karsten Hopp 2011-04-20 14:21:06 UTC

Description of problem:
CONFIG_TUNE_CELL causes problems on non-Power7 machinesand needs to be disabled.
Kernels build with this option won't boot on older PPC machines:
'Illegal Instruction'
The offending instruction is ldbrx which is an ISA 2.06 instruction found on power7 in 64-bit mode only.
Josh Boyer mentioned on IRC that he's 'pretty sure CONFIG_TUNE_CELL=y is set in the configs.  that might be allowing gcc/binutils to generate that instruction'

I've done a scratch build of a kernel with TUNE_CELL disabled and it boots where the kernel with TUNE_CELL enabled doesn't boot.
Due to the huge initrd we most likely won't be able to install on PS3 with Cell anyway so disabling this should be ok.

Comment 1 Chuck Ebbert 2011-04-25 01:46:46 UTC

config TUNE_CELL
        bool "Optimize for Cell Broadband Engine"
        depends on PPC64 && PPC_BOOK3S
        help
          Cause the compiler to optimize for the PPE of the Cell Broadband
          Engine. This will make the code run considerably faster on Cell
===>      but somewhat slower on other machines. This option only changes
===>      the scheduling of instructions, not the selection of instructions
===>      itself, so the resulting kernel will keep running on all other
===>      machines. When building a kernel that is supposed to run only
          on Cell, you should also select the POWER4_ONLY option.

So, maybe this is a binutils bug?

Comment 2 Kyle McMartin 2011-04-25 02:19:55 UTC

Probably -- it's similar to the pentium-versus-586-versus-686 catastrophe on i386. Safest just to turn off the -mtune and not worry about it in the future.

Comment 3 Chuck Ebbert 2011-06-23 12:29:35 UTC

This change was made in f15 and rawhide on Jun 24.