RFC: [2.6 patch] i386: always enable REGPARM
Is there still a known reason to disable CONFIG_REGPARM on i386 with gcc 3?
If not, I'd suggest the patch below for -mm to find all remaining bugs (if there are any).
diffstat output:
arch/i386/Kconfig | 13 ------------- arch/i386/Makefile | 2 +- include/asm-i386/linkage.h | 5 +---- include/asm-i386/module.h | 8 +------- 4 files changed, 3 insertions(+), 25 deletions(-)
Signed-off-by Adrian Bunk <[email protected]>
arch/i386/Kconfig renamed from arch/i386/Kconfig.old +0 -13
@@ -876,19 +876,6 @@ depends on (((X86_SUMMIT || X86_GENERICARCH) && NUMA) || (X86 && EFI)) default y -config REGPARM - bool "Use register arguments (EXPERIMENTAL)" - depends on EXPERIMENTAL - default n - help - Compile the kernel with -mregparm=3. This uses an different ABI - and passes the first three arguments of a function call in registers. - This will probably break binary only modules. - - This feature is only enabled for gcc-3.0 and later - earlier compilers - generate incorrect output with certain kernel constructs when - -mregparm=3 is used. - source "drivers/perfctr/Kconfig" config KERN_PHYS_OFFSET
arch/i386/Makefile renamed from arch/i386/Makefile.old +1 -1
@@ -55,7 +55,7 @@ # -mregparm=3 works ok on gcc-3.0 and later # GCC_VERSION := $(call cc-version) -cflags-$(CONFIG_REGPARM) += $(shell if [ $(GCC_VERSION) -ge 0300 ] ; then echo "-mregparm=3"; fi ;) +cflags-y += $(shell if [ $(GCC_VERSION) -ge 0300 ] ; then echo "-mregparm=3"; fi ;) # Disable unit-at-a-time mode, it makes gcc use a lot more stack # due to the lack of sharing of stacklots.
include/asm-i386/linkage.h renamed from include/asm-i386/linkage.h.old +1 -4
@@ -4,10 +4,7 @@#define asmlinkage CPP_ASMLINKAGE __attribute__((regparm(0)))#define FASTCALL (x) x __attribute__((regparm(3)))#define fastcall __attribute__((regparm(3))) - -#ifdef CONFIG_REGPARM -# define prevent_tail_call (ret) __asm__ ("" : "=r" (ret) : "0" (ret)) -#endif +#define prevent_tail_call (ret) __asm__ ("" : "=r" (ret) : "0" (ret))#ifdef CONFIG_X86_ALIGNMENT_16 #define __ALIGN .align 16,0x90
include/asm-i386/module.h renamed from include/asm-i386/module.h.old +1 -7
@@ -56,18 +56,12 @@#error unknown processor family#endif -#ifdef CONFIG_REGPARM -#define MODULE_REGPARM "REGPARM " -#else -#define MODULE_REGPARM "" -#endif -#ifdef CONFIG_4KSTACKS #define MODULE_STACKSIZE "4KSTACKS "#else #define MODULE_STACKSIZE ""#endif -#define MODULE_ARCH_VERMAGIC MODULE_PROC_FAMILY MODULE_REGPARM MODULE_STACKSIZE +#define MODULE_ARCH_VERMAGIC MODULE_PROC_FAMILY MODULE_STACKSIZE#endif /* _ASM_I386_MODULE_H */
Quoted 3 lines · Adrian Bunk
Is there still a known reason to disable
CONFIG_REGPARMon i386 with
gcc 3?
Well it lets us turn it off again if we suspect a regparm-related bug. That's happened a couple of times.
Quoted 3 lines · Adrian Bunk
Is there still a known reason to disable
CONFIG_REGPARMon i386 with gcc 3?
I wouldn't want to remove the config option, but we _can_ turn it around, and just force it to 'y'. And if nobody reports any strange problems in a few months, _then_ we can remove it.
I just make the diff be something like the appended.
Linus
----
===== arch/i386/Kconfig 1.134 vs edited =====
arch/i386/Kconfig +2 -11
@@ -862,17 +862,8 @@ default y config REGPARM - bool "Use register arguments (EXPERIMENTAL)" - depends on EXPERIMENTAL - default n - help - Compile the kernel with -mregparm=3. This uses an different ABI - and passes the first three arguments of a function call in registers. - This will probably break binary only modules. - - This feature is only enabled for gcc-3.0 and later - earlier compilers - generate incorrect output with certain kernel constructs when - -mregparm=3 is used. + bool + default y endmenu