[PATCH] powerpc: Fix runlatch performance issues

The runlatch SPR can take a lot of time to write. My original runlatch
code would set it on every exception entry even though most of the time
this was not required. It would also continually set it in the idle
loop, which is an issue on an SMT capable processor.

Now we cache the runlatch value in a threadinfo bit, and only check for
it in decrementer and hardware interrupt exceptions as well as the idle
loop. Boot on POWER3, POWER5 and iseries, and compile tested on pmac32.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

arch/powerpc/kernel/head_64.S

@@ -319,7 +319,6 @@ exception_marker:
 label##_pSeries:					\
 	HMT_MEDIUM;					\
 	mtspr	SPRN_SPRG1,r13;		/* save r13 */	\
-	RUNLATCH_ON(r13);				\
 	EXCEPTION_PROLOG_PSERIES(PACA_EXGEN, label##_common)
 
 #define STD_EXCEPTION_ISERIES(n, label, area)		\
@@ -327,7 +326,6 @@ label##_pSeries:					\
 label##_iSeries:					\
 	HMT_MEDIUM;					\
 	mtspr	SPRN_SPRG1,r13;		/* save r13 */	\
-	RUNLATCH_ON(r13);				\
 	EXCEPTION_PROLOG_ISERIES_1(area);		\
 	EXCEPTION_PROLOG_ISERIES_2;			\
 	b	label##_common
@@ -337,7 +335,6 @@ label##_iSeries:					\
 label##_iSeries:							\
 	HMT_MEDIUM;							\
 	mtspr	SPRN_SPRG1,r13;		/* save r13 */			\
-	RUNLATCH_ON(r13);						\
 	EXCEPTION_PROLOG_ISERIES_1(PACA_EXGEN);				\
 	lbz	r10,PACAPROCENABLED(r13);				\
 	cmpwi	0,r10,0;						\
@@ -390,6 +387,7 @@ label##_common:						\
 label##_common:						\
 	EXCEPTION_PROLOG_COMMON(trap, PACA_EXGEN);	\
 	DISABLE_INTS;					\
+	bl	.ppc64_runlatch_on;			\
 	addi	r3,r1,STACK_FRAME_OVERHEAD;		\
 	bl	hdlr;					\
 	b	.ret_from_except_lite
@@ -407,7 +405,6 @@ __start_interrupts:
 _machine_check_pSeries:
 	HMT_MEDIUM
 	mtspr	SPRN_SPRG1,r13		/* save r13 */
-	RUNLATCH_ON(r13)
 	EXCEPTION_PROLOG_PSERIES(PACA_EXMC, machine_check_common)
 
 	. = 0x300
@@ -434,7 +431,6 @@ END_FTR_SECTION_IFCLR(CPU_FTR_SLB)
 data_access_slb_pSeries:
 	HMT_MEDIUM
 	mtspr	SPRN_SPRG1,r13
-	RUNLATCH_ON(r13)
 	mfspr	r13,SPRN_SPRG3		/* get paca address into r13 */
 	std	r3,PACA_EXSLB+EX_R3(r13)
 	mfspr	r3,SPRN_DAR
@@ -460,7 +456,6 @@ data_access_slb_pSeries:
 instruction_access_slb_pSeries:
 	HMT_MEDIUM
 	mtspr	SPRN_SPRG1,r13
-	RUNLATCH_ON(r13)
 	mfspr	r13,SPRN_SPRG3		/* get paca address into r13 */
 	std	r3,PACA_EXSLB+EX_R3(r13)
 	mfspr	r3,SPRN_SRR0		/* SRR0 is faulting address */
@@ -491,7 +486,6 @@ instruction_access_slb_pSeries:
 	.globl	system_call_pSeries
 system_call_pSeries:
 	HMT_MEDIUM
-	RUNLATCH_ON(r9)
 	mr	r9,r13
 	mfmsr	r10
 	mfspr	r13,SPRN_SPRG3
@@ -575,7 +569,6 @@ slb_miss_user_pseries:
 system_reset_fwnmi:
 	HMT_MEDIUM
 	mtspr	SPRN_SPRG1,r13		/* save r13 */
-	RUNLATCH_ON(r13)
 	EXCEPTION_PROLOG_PSERIES(PACA_EXGEN, system_reset_common)
 
 	.globl machine_check_fwnmi
@@ -583,7 +576,6 @@ system_reset_fwnmi:
 machine_check_fwnmi:
 	HMT_MEDIUM
 	mtspr	SPRN_SPRG1,r13		/* save r13 */
-	RUNLATCH_ON(r13)
 	EXCEPTION_PROLOG_PSERIES(PACA_EXMC, machine_check_common)
 
 #ifdef CONFIG_PPC_ISERIES
@@ -894,7 +886,6 @@ unrecov_fer:
 	.align	7
 	.globl data_access_common
 data_access_common:
-	RUNLATCH_ON(r10)		/* It wont fit in the 0x300 handler */
 	mfspr	r10,SPRN_DAR
 	std	r10,PACA_EXGEN+EX_DAR(r13)
 	mfspr	r10,SPRN_DSISR
@@ -1042,6 +1033,7 @@ hardware_interrupt_common:
 	EXCEPTION_PROLOG_COMMON(0x500, PACA_EXGEN)
 hardware_interrupt_entry:
 	DISABLE_INTS
+	bl	.ppc64_runlatch_on
 	addi	r3,r1,STACK_FRAME_OVERHEAD
 	bl	.do_IRQ
 	b	.ret_from_except_lite

arch/powerpc/kernel/process.c

@@ -888,3 +888,35 @@ void dump_stack(void)
 	show_stack(current, NULL);
 }
 EXPORT_SYMBOL(dump_stack);
+
+#ifdef CONFIG_PPC64
+void ppc64_runlatch_on(void)
+{
+	unsigned long ctrl;
+
+	if (cpu_has_feature(CPU_FTR_CTRL) && !test_thread_flag(TIF_RUNLATCH)) {
+		HMT_medium();
+
+		ctrl = mfspr(SPRN_CTRLF);
+		ctrl |= CTRL_RUNLATCH;
+		mtspr(SPRN_CTRLT, ctrl);
+
+		set_thread_flag(TIF_RUNLATCH);
+	}
+}
+
+void ppc64_runlatch_off(void)
+{
+	unsigned long ctrl;
+
+	if (cpu_has_feature(CPU_FTR_CTRL) && test_thread_flag(TIF_RUNLATCH)) {
+		HMT_medium();
+
+		clear_thread_flag(TIF_RUNLATCH);
+
+		ctrl = mfspr(SPRN_CTRLF);
+		ctrl &= ~CTRL_RUNLATCH;
+		mtspr(SPRN_CTRLT, ctrl);
+	}
+}
+#endif

arch/powerpc/platforms/iseries/setup.c

@@ -648,6 +648,7 @@ static void yield_shared_processor(void)
 	 * here and let the timer_interrupt code sort out the actual time.
 	 */
 	get_lppaca()->int_dword.fields.decr_int = 1;
+	ppc64_runlatch_on();
 	process_iSeries_events();
 }
 

include/asm-powerpc/reg.h

@@ -615,27 +615,9 @@
 #define proc_trap()	asm volatile("trap")
 
 #ifdef CONFIG_PPC64
-static inline void ppc64_runlatch_on(void)
-{
-	unsigned long ctrl;
 
-	if (cpu_has_feature(CPU_FTR_CTRL)) {
+extern void ppc64_runlatch_on(void);
-		ctrl = mfspr(SPRN_CTRLF);
+extern void ppc64_runlatch_off(void);
-		ctrl |= CTRL_RUNLATCH;
-		mtspr(SPRN_CTRLT, ctrl);
-	}
-}
-
-static inline void ppc64_runlatch_off(void)
-{
-	unsigned long ctrl;
-
-	if (cpu_has_feature(CPU_FTR_CTRL)) {
-		ctrl = mfspr(SPRN_CTRLF);
-		ctrl &= ~CTRL_RUNLATCH;
-		mtspr(SPRN_CTRLT, ctrl);
-	}
-}
 
 extern unsigned long scom970_read(unsigned int address);
 extern void scom970_write(unsigned int address, unsigned long value);
@@ -645,15 +627,6 @@ extern void scom970_write(unsigned int address, unsigned long value);
 #define __get_SP()	({unsigned long sp; \
 			asm volatile("mr %0,1": "=r" (sp)); sp;})
 
-#else /* __ASSEMBLY__ */
-
-#define RUNLATCH_ON(REG)			\
-BEGIN_FTR_SECTION				\
-	mfspr	(REG),SPRN_CTRLF;		\
-	ori	(REG),(REG),CTRL_RUNLATCH;	\
-	mtspr	SPRN_CTRLT,(REG);		\
-END_FTR_SECTION_IFSET(CPU_FTR_CTRL)
-
 #endif /* __ASSEMBLY__ */
 #endif /* __KERNEL__ */
 #endif /* _ASM_POWERPC_REG_H */

include/asm-powerpc/thread_info.h

@@ -113,7 +113,7 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_POLLING_NRFLAG	4	/* true if poll_idle() is polling
 					   TIF_NEED_RESCHED */
 #define TIF_32BIT		5	/* 32 bit binary */
-/* #define SPARE		6 */
+#define TIF_RUNLATCH		6	/* Is the runlatch enabled? */
 #define TIF_ABI_PENDING		7	/* 32/64 bit switch needed */
 #define TIF_SYSCALL_AUDIT	8	/* syscall auditing active */
 #define TIF_SINGLESTEP		9	/* singlestepping active */
@@ -131,7 +131,7 @@ static inline struct thread_info *current_thread_info(void)
 #define _TIF_NEED_RESCHED	(1<<TIF_NEED_RESCHED)
 #define _TIF_POLLING_NRFLAG	(1<<TIF_POLLING_NRFLAG)
 #define _TIF_32BIT		(1<<TIF_32BIT)
-/* #define _SPARE		(1<<SPARE) */
+#define _TIF_RUNLATCH		(1<<TIF_RUNLATCH)
 #define _TIF_ABI_PENDING	(1<<TIF_ABI_PENDING)
 #define _TIF_SYSCALL_AUDIT	(1<<TIF_SYSCALL_AUDIT)
 #define _TIF_SINGLESTEP		(1<<TIF_SINGLESTEP)