Bring back r313037, with fixes for mips:

Implement get_pcpu() for amd64/sparc64/mips/powerpc, and use it to
replace pcpu_find(curcpu) in MI code.

Reviewed by:	andreast, kan, lidl
Tested by:	lidl(mips, sparc64), andreast(powerpc)
Differential Revision:	https://reviews.freebsd.org/D9587

8 files changed, 34 insertions, 15 deletions

diff --git a/sys/amd64/include/pcpu.h b/sys/amd64/include/pcpu.h
index a4f4e1dc731f..7e3c1b0bbb27 100644
--- a/sys/amd64/include/pcpu.h
+++ b/sys/amd64/include/pcpu.h
@@ -78,6 +78,7 @@
 
 extern struct pcpu *pcpup;
 
+#define	get_pcpu()		(pcpup)
 #define	PCPU_GET(member)	(pcpup->pc_ ## member)
 #define	PCPU_ADD(member, val)	(pcpup->pc_ ## member += (val))
 #define	PCPU_INC(member)	PCPU_ADD(member, 1)
@@ -203,6 +204,15 @@ extern struct pcpu *pcpup;
 	}								\
 }
 
+#define	get_pcpu() __extension__ ({					\
+	struct pcpu *__pc;						\
+									\
+	__asm __volatile("movq %%gs:%1,%0"				\
+	    : "=r" (__pc)						\
+	    : "m" (*(struct pcpu *)(__pcpu_offset(pc_prvspace))));	\
+	__pc;								\
+})
+
 #define	PCPU_GET(member)	__PCPU_GET(pc_ ## member)
 #define	PCPU_ADD(member, val)	__PCPU_ADD(pc_ ## member, val)
 #define	PCPU_INC(member)	__PCPU_INC(pc_ ## member)
diff --git a/sys/kern/kern_rmlock.c b/sys/kern/kern_rmlock.c
index b74253f8a7ab..70f11413ad2b 100644
--- a/sys/kern/kern_rmlock.c
+++ b/sys/kern/kern_rmlock.c
@@ -156,7 +156,7 @@ unlock_rm(struct lock_object *lock)
 		 */
 		critical_enter();
 		td = curthread;
-		pc = pcpu_find(curcpu);
+		pc = get_pcpu();
 		for (queue = pc->pc_rm_queue.rmq_next;
 		    queue != &pc->pc_rm_queue; queue = queue->rmq_next) {
 			tracker = (struct rm_priotracker *)queue;
@@ -258,7 +258,7 @@ rm_cleanIPI(void *arg)
 	struct rmlock *rm = arg;
 	struct rm_priotracker *tracker;
 	struct rm_queue *queue;
-	pc = pcpu_find(curcpu);
+	pc = get_pcpu();
 
 	for (queue = pc->pc_rm_queue.rmq_next; queue != &pc->pc_rm_queue;
 	    queue = queue->rmq_next) {
@@ -355,7 +355,7 @@ _rm_rlock_hard(struct rmlock *rm, struct rm_priotracker *tracker, int trylock)
 	struct pcpu *pc;
 
 	critical_enter();
-	pc = pcpu_find(curcpu);
+	pc = get_pcpu();
 
 	/* Check if we just need to do a proper critical_exit. */
 	if (!CPU_ISSET(pc->pc_cpuid, &rm->rm_writecpus)) {
@@ -416,7 +416,7 @@ _rm_rlock_hard(struct rmlock *rm, struct rm_priotracker *tracker, int trylock)
 	}
 
 	critical_enter();
-	pc = pcpu_find(curcpu);
+	pc = get_pcpu();
 	CPU_CLR(pc->pc_cpuid, &rm->rm_writecpus);
 	rm_tracker_add(pc, tracker);
 	sched_pin();
@@ -641,7 +641,7 @@ _rm_rlock_debug(struct rmlock *rm, struct rm_priotracker *tracker,
 #ifdef INVARIANTS
 	if (!(rm->lock_object.lo_flags & LO_RECURSABLE) && !trylock) {
 		critical_enter();
-		KASSERT(rm_trackers_present(pcpu_find(curcpu), rm,
+		KASSERT(rm_trackers_present(get_pcpu(), rm,
 		    curthread) == 0,
 		    ("rm_rlock: recursed on non-recursive rmlock %s @ %s:%d\n",
 		    rm->lock_object.lo_name, file, line));
@@ -771,7 +771,7 @@ _rm_assert(const struct rmlock *rm, int what, const char *file, int line)
 		}
 
 		critical_enter();
-		count = rm_trackers_present(pcpu_find(curcpu), rm, curthread);
+		count = rm_trackers_present(get_pcpu(), rm, curthread);
 		critical_exit();
 
 		if (count == 0)
@@ -797,7 +797,7 @@ _rm_assert(const struct rmlock *rm, int what, const char *file, int line)
 			    rm->lock_object.lo_name, file, line);
 
 		critical_enter();
-		count = rm_trackers_present(pcpu_find(curcpu), rm, curthread);
+		count = rm_trackers_present(get_pcpu(), rm, curthread);
 		critical_exit();
 
 		if (count != 0)
diff --git a/sys/mips/include/pcpu.h b/sys/mips/include/pcpu.h
index d4f240115fce..77a3983c8872 100644
--- a/sys/mips/include/pcpu.h
+++ b/sys/mips/include/pcpu.h
@@ -39,16 +39,17 @@
 	struct	pmap	*pc_curpmap;		/* pmap of curthread */	\
 	u_int32_t	pc_next_asid;		/* next ASID to alloc */ \
 	u_int32_t	pc_asid_generation;	/* current ASID generation */ \
-	u_int		pc_pending_ipis;	/* IPIs pending to this CPU */
+	u_int		pc_pending_ipis;	/* IPIs pending to this CPU */ \
+	struct	pcpu	*pc_self;		/* globally-uniqe self pointer */
 
 #ifdef	__mips_n64
 #define	PCPU_MD_MIPS64_FIELDS						\
 	PCPU_MD_COMMON_FIELDS						\
-	char		__pad[61]
+	char		__pad[53]
 #else
 #define	PCPU_MD_MIPS32_FIELDS						\
 	PCPU_MD_COMMON_FIELDS						\
-	char		__pad[193]
+	char		__pad[189]
 #endif
 
 #ifdef	__mips_n64
@@ -65,6 +66,13 @@ extern char pcpu_space[MAXCPU][PAGE_SIZE * 2];
 extern struct pcpu *pcpup;
 #define	PCPUP	pcpup
 
+/*
+ * Since we use a wired TLB entry to map the same VA to a different
+ * physical page for each CPU, get_pcpu() must use the pc_self
+ * field to obtain a globally-unique pointer.
+ */
+#define	get_pcpu()		(PCPUP->pc_self)
+
 #define	PCPU_ADD(member, value)	(PCPUP->pc_ ## member += (value))
 #define	PCPU_GET(member)	(PCPUP->pc_ ## member)
 #define	PCPU_INC(member)	PCPU_ADD(member, 1)
diff --git a/sys/mips/mips/machdep.c b/sys/mips/mips/machdep.c
index 3d44cdad0e5f..9b375e0e55f6 100644
--- a/sys/mips/mips/machdep.c
+++ b/sys/mips/mips/machdep.c
@@ -475,6 +475,7 @@ cpu_pcpu_init(struct pcpu *pcpu, int cpuid, size_t size)
 
 	pcpu->pc_next_asid = 1;
 	pcpu->pc_asid_generation = 1;
+	pcpu->pc_self = pcpu;
 #ifdef SMP
 	if ((vm_offset_t)pcpup >= VM_MIN_KERNEL_ADDRESS &&
 	    (vm_offset_t)pcpup <= VM_MAX_KERNEL_ADDRESS) {
diff --git a/sys/net/netisr.c b/sys/net/netisr.c
index 755c288d2269..b96ff1c070a7 100644
--- a/sys/net/netisr.c
+++ b/sys/net/netisr.c
@@ -1268,9 +1268,7 @@ netisr_start_swi(u_int cpuid, struct pcpu *pc)
 static void
 netisr_init(void *arg)
 {
-#ifdef EARLY_AP_STARTUP
 	struct pcpu *pc;
-#endif
 
 	NETISR_LOCK_INIT();
 	if (netisr_maxthreads == 0 || netisr_maxthreads < -1 )
@@ -1308,7 +1306,8 @@ netisr_init(void *arg)
 		netisr_start_swi(pc->pc_cpuid, pc);
 	}
 #else
-	netisr_start_swi(curcpu, pcpu_find(curcpu));
+	pc = get_pcpu();
+	netisr_start_swi(pc->pc_cpuid, pc);
 #endif
 }
 SYSINIT(netisr_init, SI_SUB_SOFTINTR, SI_ORDER_FIRST, netisr_init, NULL);
diff --git a/sys/powerpc/include/cpufunc.h b/sys/powerpc/include/cpufunc.h
index 378274b67e36..c70a94d31b6c 100644
--- a/sys/powerpc/include/cpufunc.h
+++ b/sys/powerpc/include/cpufunc.h
@@ -201,7 +201,7 @@ intr_restore(register_t msr)
 }
 
 static __inline struct pcpu *
-powerpc_get_pcpup(void)
+get_pcpu(void)
 {
 	struct pcpu *ret;
 
diff --git a/sys/powerpc/include/pcpu.h b/sys/powerpc/include/pcpu.h
index 79cdd3039fff..b6949cf7bdc5 100644
--- a/sys/powerpc/include/pcpu.h
+++ b/sys/powerpc/include/pcpu.h
@@ -142,7 +142,7 @@ struct pvo_entry;
 
 #ifdef _KERNEL
 
-#define pcpup	((struct pcpu *) powerpc_get_pcpup())
+#define pcpup	(get_pcpu())
 
 static __inline __pure2 struct thread *
 __curthread(void)
diff --git a/sys/sparc64/include/pcpu.h b/sys/sparc64/include/pcpu.h
index df43810ef0b5..51524fba4a0a 100644
--- a/sys/sparc64/include/pcpu.h
+++ b/sys/sparc64/include/pcpu.h
@@ -74,6 +74,7 @@ struct pcpu;
 register struct pcb *curpcb __asm__(__XSTRING(PCB_REG));
 register struct pcpu *pcpup __asm__(__XSTRING(PCPU_REG));
 
+#define	get_pcpu()		(pcpup)
 #define	PCPU_GET(member)	(pcpup->pc_ ## member)
 
 static __inline __pure2 struct thread *