Skip to contentSkip to conversation

ARM: tlb: Compile out MULTI_TLB specific code when MULTI_TLB is undefined

PATCH In review Tzu-Jung Lee 2 participants 4 messages Sep 16, 2011 – Sep 17, 2011 Arm (32-bit)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ARM: tlb: Compile out MULTI_TLB specific code when MULTI_TLB is undefined

To eliminate the 'undefined cpu_tlb' errors when the kernel is built with -O0

Change-Id Ie5afea022b30dbe23a7ee3f3fc6af0dd6479b980
arch/arm/include/asm/tlbflush.h +3 -4
index d2005de..a3f7aea 100644
@@ -230,23 +230,22 @@ struct cpu_tlb_fns {
  */
 #ifdef MULTI_TLB
 
+extern struct cpu_tlb_fns cpu_tlb;
 #define __cpu_flush_user_tlb_range	cpu_tlb.flush_user_range
 #define __cpu_flush_kern_tlb_range	cpu_tlb.flush_kern_range
+#define __cpu_tlb_flags			cpu_tlb.tlb_flags
 
 #else
 
 #define __cpu_flush_user_tlb_range	__glue(_TLB,_flush_user_tlb_range)
 #define __cpu_flush_kern_tlb_range	__glue(_TLB,_flush_kern_tlb_range)
+#define __cpu_tlb_flags			0
 
 extern void __cpu_flush_user_tlb_range(unsigned long, unsigned long, struct vm_area_struct *);
 extern void __cpu_flush_kern_tlb_range(unsigned long, unsigned long);
 
 #endif
 
-extern struct cpu_tlb_fns cpu_tlb;
-
-#define __cpu_tlb_flags			cpu_tlb.tlb_flags
-
 /*
  *	TLB Management
  *	==============
Signature · 1 line
1.7.6.1
#1 lore raw
Quoted 1 line · Tzu-Jung Lee

To eliminate the 'undefined cpu_tlb' errors when the kernel is built with -O0

The kernel requires to be built with optimization in many places, and so this is an option we don't support.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:
#2 lore raw
Quoted 11 lines · Russell King
Quoted 1 line · Tzu-Jung Lee

To eliminate the 'undefined cpu_tlb' errors when the kernel is built with -O0

The kernel requires to be built with optimization in many places, and so this is an option we don't support.

--
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Hi Russell,

Indeed, this is only one of the places where we used to particularly build with higher optimization level. More than often, we only disable the optimization for interested object files with CFLAGS_foo.o += -g -O0. While porting kernel to different platforms, however, failures at the early booting stage usually makes low level initialization(mm/tlb/cache/cpu...) become the objects that we want to disable the optimization for the ease of diagnostic with source level debugger. In this cases, patches to the source code become required to get the -O0 work for them. Would you consider to have these patches merged if they doesn't have side effect.

Thanks,
Roy

#3 lore raw
Quoted 10 lines · Tzu-Jung Lee

Indeed, this is only one of the places where we used to particularly build with higher optimization level. More than often, we only disable the optimization for interested object files with CFLAGS_foo.o += -g -O0. While porting kernel to different platforms, however, failures at the early booting stage usually makes low level initialization(mm/tlb/cache/cpu...) become the objects that we want to disable the optimization for the ease of diagnostic with source level debugger. In this cases, patches to the source code become required to get the -O0 work for them. Would you consider to have these patches merged if they doesn't have side effect.

We've gone some 17 or so years without needing -O0 to debug the kernel. Why has it now become necessary?

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of: