x86/fpu: Split old_fpu & new_fpu handling into separate functions

By moving all of the new_fpu state handling into switch_fpu_finish(),
the code can be simplified some more.

This gets rid of the prefetch, but given the size of the FPU register
state on modern CPUs, and the amount of work done by __switch_to()
inbetween both functions, the value of a single cache line prefetch
seems somewhat dubious anyway.

Signed-off-by: Rik van Riel <riel@redhat.com>
Acked-by: Dave Hansen <dave.hansen@intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1476447331-21566-3-git-send-email-riel@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

arch/x86/include/asm/fpu/internal.h

@@ -552,27 +552,15 @@ static inline int fpregs_active(void)
  *
  * This is a two-stage process:
  *
- *  - switch_fpu_prepare() saves the old state and
- *    sets the new state of the CR0.TS bit. This is
- *    done within the context of the old process.
+ *  - switch_fpu_prepare() saves the old state.
+ *    This is done within the context of the old process.
  *
  *  - switch_fpu_finish() restores the new state as
  *    necessary.
  */
-typedef struct { int preload; } fpu_switch_t;
-
-static inline fpu_switch_t
-switch_fpu_prepare(struct fpu *old_fpu, struct fpu *new_fpu, int cpu)
+static inline void
+switch_fpu_prepare(struct fpu *old_fpu, int cpu)
 {
-	fpu_switch_t fpu;
-
-	/*
-	 * If the task has used the math, pre-load the FPU on xsave processors
-	 * or if the past 5 consecutive context-switches used math.
-	 */
-	fpu.preload = static_cpu_has(X86_FEATURE_FPU) &&
-		      new_fpu->fpstate_active;
-
 	if (old_fpu->fpregs_active) {
 		if (!copy_fpregs_to_fpstate(old_fpu))
 			old_fpu->last_cpu = -1;
@@ -584,16 +572,6 @@ switch_fpu_prepare(struct fpu *old_fpu, struct fpu *new_fpu, int cpu)
 		trace_x86_fpu_regs_deactivated(old_fpu);
 	} else
 		old_fpu->last_cpu = -1;
-
-	if (fpu.preload) {
-		if (fpregs_state_valid(new_fpu, cpu))
-			fpu.preload = 0;
-		else
-			prefetch(&new_fpu->state);
-		fpregs_activate(new_fpu);
-	}
-
-	return fpu;
 }
 
 /*
@@ -601,15 +579,19 @@ switch_fpu_prepare(struct fpu *old_fpu, struct fpu *new_fpu, int cpu)
  */
 
 /*
- * By the time this gets called, we've already cleared CR0.TS and
- * given the process the FPU if we are going to preload the FPU
- * state - all we need to do is to conditionally restore the register
- * state itself.
+ * Set up the userspace FPU context for the new task, if the task
+ * has used the FPU.
  */
-static inline void switch_fpu_finish(struct fpu *new_fpu, fpu_switch_t fpu_switch)
+static inline void switch_fpu_finish(struct fpu *new_fpu, int cpu)
 {
-	if (fpu_switch.preload)
-		copy_kernel_to_fpregs(&new_fpu->state);
+	bool preload = static_cpu_has(X86_FEATURE_FPU) &&
+		       new_fpu->fpstate_active;
+
+	if (preload) {
+		if (!fpregs_state_valid(new_fpu, cpu))
+			copy_kernel_to_fpregs(&new_fpu->state);
+		fpregs_activate(new_fpu);
+	}
 }
 
 /*

arch/x86/kernel/process_32.c

@@ -232,11 +232,10 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 	struct fpu *next_fpu = &next->fpu;
 	int cpu = smp_processor_id();
 	struct tss_struct *tss = &per_cpu(cpu_tss, cpu);
-	fpu_switch_t fpu_switch;
 
 	/* never put a printk in __switch_to... printk() calls wake_up*() indirectly */
 
-	fpu_switch = switch_fpu_prepare(prev_fpu, next_fpu, cpu);
+	switch_fpu_prepare(prev_fpu, cpu);
 
 	/*
 	 * Save away %gs. No need to save %fs, as it was saved on the
@@ -295,7 +294,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 	if (prev->gs | next->gs)
 		lazy_load_gs(next->gs);
 
-	switch_fpu_finish(next_fpu, fpu_switch);
+	switch_fpu_finish(next_fpu, cpu);
 
 	this_cpu_write(current_task, next_p);
 

arch/x86/kernel/process_64.c

@@ -264,9 +264,8 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 	int cpu = smp_processor_id();
 	struct tss_struct *tss = &per_cpu(cpu_tss, cpu);
 	unsigned prev_fsindex, prev_gsindex;
-	fpu_switch_t fpu_switch;
 
-	fpu_switch = switch_fpu_prepare(prev_fpu, next_fpu, cpu);
+	switch_fpu_prepare(prev_fpu, cpu);
 
 	/* We must save %fs and %gs before load_TLS() because
 	 * %fs and %gs may be cleared by load_TLS().
@@ -416,7 +415,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 		prev->gsbase = 0;
 	prev->gsindex = prev_gsindex;
 
-	switch_fpu_finish(next_fpu, fpu_switch);
+	switch_fpu_finish(next_fpu, cpu);
 
 	/*
 	 * Switch the PDA and FPU contexts.