ia64: Fix SMP fph-handling. Patch by Asit Mallick with some additional

      changes by yours truly.

arch/ia64/kernel/process.c

@@ -378,7 +378,7 @@ copy_thread (int nr, unsigned long clone_flags,
 #	define THREAD_FLAGS_TO_SET	0
 	p->thread.flags = ((current->thread.flags & ~THREAD_FLAGS_TO_CLEAR)
 			   | THREAD_FLAGS_TO_SET);
-	p->thread.last_fph_cpu = -1;
+	ia64_drop_fpu(p);	/* don't pick up stale state from a CPU's fph */
 #ifdef CONFIG_IA32_SUPPORT
 	/*
 	 * If we're cloning an IA32 task then save the IA32 extra
@@ -606,11 +606,7 @@ flush_thread (void)
 {
 	/* drop floating-point and debug-register state if it exists: */
 	current->thread.flags &= ~(IA64_THREAD_FPH_VALID | IA64_THREAD_DBG_VALID);
-
-#ifndef CONFIG_SMP
-	if (ia64_get_fpu_owner() == current)
-		ia64_set_fpu_owner(0);
-#endif
+	ia64_drop_fpu(current);
 }
 
 #ifdef CONFIG_PERFMON
@@ -648,13 +644,10 @@ release_thread (struct task_struct *task)
 void
 exit_thread (void)
 {
-#ifndef CONFIG_SMP
-	if (ia64_get_fpu_owner() == current)
-		ia64_set_fpu_owner(0);
-#endif
+	ia64_drop_fpu(current);
 #ifdef CONFIG_PERFMON
        /* if needed, stop monitoring and flush state to perfmon context */
-	if (current->thread.pfm_context) 
+	if (current->thread.pfm_context)
 		pfm_flush_regs(current);
 
 	/* free debug register resources */

arch/ia64/kernel/ptrace.c

@@ -598,18 +598,13 @@ inline void
 ia64_flush_fph (struct task_struct *task)
 {
 	struct ia64_psr *psr = ia64_psr(ia64_task_regs(task));
-#ifdef CONFIG_SMP
-	struct task_struct *fpu_owner = current;
-#else
-	struct task_struct *fpu_owner = ia64_get_fpu_owner();
-#endif
 
-	if (task == fpu_owner && psr->mfh) {
+	if (ia64_is_local_fpu_owner(task) && psr->mfh) {
 		psr->mfh = 0;
-		ia64_save_fpu(&task->thread.fph[0]);
 		task->thread.flags |= IA64_THREAD_FPH_VALID;
-		task->thread.last_fph_cpu = smp_processor_id();
+		ia64_save_fpu(&task->thread.fph[0]);
 	}
+	ia64_drop_fpu(task);
 }
 
 /*
@@ -628,11 +623,8 @@ ia64_sync_fph (struct task_struct *task)
 	ia64_flush_fph(task);
 	if (!(task->thread.flags & IA64_THREAD_FPH_VALID)) {
 		task->thread.flags |= IA64_THREAD_FPH_VALID;
-		task->thread.last_fph_cpu = -1;		/* force reload */
 		memset(&task->thread.fph, 0, sizeof(task->thread.fph));
 	}
-	if (ia64_get_fpu_owner() == task)
-		ia64_set_fpu_owner(0);
 	psr->dfh = 1;
 }
 

arch/ia64/kernel/signal.c

@@ -143,11 +143,11 @@ restore_sigcontext (struct sigcontext *sc, struct sigscratch *scr)
 		__copy_from_user(current->thread.fph, &sc->sc_fr[32], 96*16);
 		psr->mfh = 0;	/* drop signal handler's fph contents... */
 		if (psr->dfh)
-			current->thread.last_fph_cpu = -1;
+			ia64_drop_fpu(current);
 		else {
+			/* We already own the local fph, otherwise psr->dfh wouldn't be 0.  */
 			__ia64_load_fpu(current->thread.fph);
-			ia64_set_fpu_owner(current);
-			current->thread.last_fph_cpu = smp_processor_id();
+			ia64_set_local_fpu_owner(current);
 		}
 	}
 	return err;

arch/ia64/kernel/traps.c

@@ -247,7 +247,8 @@ disabled_fph_fault (struct pt_regs *regs)
 	psr->dfh = 0;
 #ifndef CONFIG_SMP
 	{
-		struct task_struct *fpu_owner = ia64_get_fpu_owner();
+		struct task_struct *fpu_owner
+			= (struct task_struct *)ia64_get_kr(IA64_KR_FPU_OWNER);
 
 		if (fpu_owner == current)
 			return;
@@ -256,7 +257,7 @@ disabled_fph_fault (struct pt_regs *regs)
 			ia64_flush_fph(fpu_owner);
 	}
 #endif /* !CONFIG_SMP */
-	ia64_set_fpu_owner(current);
+	ia64_set_local_fpu_owner(current);
 	if ((current->thread.flags & IA64_THREAD_FPH_VALID) != 0) {
 		__ia64_load_fpu(current->thread.fph);
 		psr->mfh = 0;

include/asm-ia64/processor.h

@@ -417,17 +417,28 @@ ia64_set_kr (unsigned long regnum, unsigned long r)
 	}
 }
 
-static inline struct task_struct *
-ia64_get_fpu_owner (void)
-{
-	return (struct task_struct *) ia64_get_kr(IA64_KR_FPU_OWNER);
-}
+/*
+ * The following three macros can't be inline functions because we don't have struct
+ * task_struct at this point.
+ */
 
-static inline void
-ia64_set_fpu_owner (struct task_struct *t)
-{
-	ia64_set_kr(IA64_KR_FPU_OWNER, (unsigned long) t);
-}
+/* Return TRUE if task T owns the fph partition of the CPU we're running on. */
+#define ia64_is_local_fpu_owner(t)								\
+({												\
+	struct task_struct *__ia64_islfo_task = (t);						\
+	(__ia64_islfo_task->thread.last_fph_cpu == smp_processor_id()				\
+	 && __ia64_islfo_task == (struct task_struct *) ia64_get_kr(IA64_KR_FPU_OWNER));	\
+})
+
+/* Mark task T as owning the fph partition of the CPU we're running on. */
+#define ia64_set_local_fpu_owner(t) do {						\
+	struct task_struct *__ia64_slfo_task = (t);					\
+	__ia64_slfo_task->thread.last_fph_cpu = smp_processor_id();			\
+	ia64_set_kr(IA64_KR_FPU_OWNER, (unsigned long) __ia64_slfo_task);		\
+} while (0)
+
+/* Mark the fph partition of task T as being invalid on all CPUs.  */
+#define ia64_drop_fpu(t)	((t)->thread.last_fph_cpu = -1)
 
 extern void __ia64_init_fpu (void);
 extern void __ia64_save_fpu (struct ia64_fpreg *fph);

include/asm-ia64/system.h

@@ -217,13 +217,11 @@ extern void ia64_load_extra (struct task_struct *task);
 	 || IS_IA32_PROCESS(ia64_task_regs(t)) || PERFMON_IS_SYSWIDE())
 
 #define __switch_to(prev,next,last) do {							 \
-	struct task_struct *__fpu_owner = ia64_get_fpu_owner();					 \
 	if (IA64_HAS_EXTRA_STATE(prev))								 \
 		ia64_save_extra(prev);								 \
 	if (IA64_HAS_EXTRA_STATE(next))								 \
 		ia64_load_extra(next);								 \
-	ia64_psr(ia64_task_regs(next))->dfh =							 \
-		!(__fpu_owner == (next) && ((next)->thread.last_fph_cpu == smp_processor_id())); \
+	ia64_psr(ia64_task_regs(next))->dfh = !ia64_is_local_fpu_owner(next);			 \
 	(last) = ia64_switch_to((next));							 \
 } while (0)
 
@@ -239,7 +237,6 @@ extern void ia64_load_extra (struct task_struct *task);
 		ia64_psr(ia64_task_regs(prev))->mfh = 0;			\
 		(prev)->thread.flags |= IA64_THREAD_FPH_VALID;			\
 		__ia64_save_fpu((prev)->thread.fph);				\
-		(prev)->thread.last_fph_cpu = smp_processor_id();		\
 	}									\
 	__switch_to(prev, next, last);						\
 } while (0)