s390/kernel: lazy restore fpu registers

Improve the save and restore behavior of FPU register contents to use the
vector extension within the kernel.

The kernel does not use floating-point or vector registers and, therefore,
saving and restoring the FPU register contents are performed for handling
signals or switching processes only.  To prepare for using vector
instructions and vector registers within the kernel, enhance the save
behavior and implement a lazy restore at return to user space from a
system call or interrupt.

To implement the lazy restore, the save_fpu_regs() sets a CPU information
flag, CIF_FPU, to indicate that the FPU registers must be restored.
Saving and setting CIF_FPU is performed in an atomic fashion to be
interrupt-safe.  When the kernel wants to use the vector extension or
wants to change the FPU register state for a task during signal handling,
the save_fpu_regs() must be called first.  The CIF_FPU flag is also set at
process switch.  At return to user space, the FPU state is restored.  In
particular, the FPU state includes the floating-point or vector register
contents, as well as, vector-enablement and floating-point control.  The
FPU state restore and clearing CIF_FPU is also performed in an atomic
fashion.

For KVM, the restore of the FPU register state is performed when restoring
the general-purpose guest registers before the SIE instructions is started.
Because the path towards the SIE instruction is interruptible, the CIF_FPU
flag must be checked again right before going into SIE.  If set, the guest
registers must be reloaded again by re-entering the outer SIE loop.  This
is the same behavior as if the SIE critical section is interrupted.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

5 files changed, 18 insertions, 104 deletions

diff --git a/arch/s390/include/asm/ctl_reg.h b/arch/s390/include/asm/ctl_reg.h
index d7697ab802f6..17a373576868 100644
--- a/arch/s390/include/asm/ctl_reg.h
+++ b/arch/s390/include/asm/ctl_reg.h
@@ -46,6 +46,8 @@ static inline void __ctl_clear_bit(unsigned int cr, unsigned int bit)
 	__ctl_load(reg, cr, cr);
 }
 
+void __ctl_set_vx(void);
+
 void smp_ctl_set_bit(int cr, int bit);
 void smp_ctl_clear_bit(int cr, int bit);
 
diff --git a/arch/s390/include/asm/fpu-internal.h b/arch/s390/include/asm/fpu-internal.h
index cc44c75fc4f7..237f8fcbe46b 100644
--- a/arch/s390/include/asm/fpu-internal.h
+++ b/arch/s390/include/asm/fpu-internal.h
@@ -28,9 +28,14 @@ struct fpu {
 	};
 };
 
+void save_fpu_regs(struct fpu *fpu);
+
 #define is_vx_fpu(fpu) (!!((fpu)->flags & FPU_USE_VX))
 #define is_vx_task(tsk) (!!((tsk)->thread.fpu.flags & FPU_USE_VX))
 
+/* VX array structure for address operand constraints in inline assemblies */
+struct vx_array { __vector128 _[__NUM_VXRS]; };
+
 static inline int test_fp_ctl(u32 fpc)
 {
 	u32 orig_fpc;
@@ -48,76 +53,6 @@ static inline int test_fp_ctl(u32 fpc)
 	return rc;
 }
 
-static inline void save_fp_ctl(u32 *fpc)
-{
-	asm volatile(
-		"       stfpc   %0\n"
-		: "+Q" (*fpc));
-}
-
-static inline int restore_fp_ctl(u32 *fpc)
-{
-	int rc;
-
-	asm volatile(
-		"	lfpc    %1\n"
-		"0:	la	%0,0\n"
-		"1:\n"
-		: "=d" (rc) : "Q" (*fpc), "0" (-EINVAL));
-	return rc;
-}
-
-static inline void save_fp_regs(freg_t *fprs)
-{
-	asm volatile("std 0,%0" : "=Q" (fprs[0]));
-	asm volatile("std 2,%0" : "=Q" (fprs[2]));
-	asm volatile("std 4,%0" : "=Q" (fprs[4]));
-	asm volatile("std 6,%0" : "=Q" (fprs[6]));
-	asm volatile("std 1,%0" : "=Q" (fprs[1]));
-	asm volatile("std 3,%0" : "=Q" (fprs[3]));
-	asm volatile("std 5,%0" : "=Q" (fprs[5]));
-	asm volatile("std 7,%0" : "=Q" (fprs[7]));
-	asm volatile("std 8,%0" : "=Q" (fprs[8]));
-	asm volatile("std 9,%0" : "=Q" (fprs[9]));
-	asm volatile("std 10,%0" : "=Q" (fprs[10]));
-	asm volatile("std 11,%0" : "=Q" (fprs[11]));
-	asm volatile("std 12,%0" : "=Q" (fprs[12]));
-	asm volatile("std 13,%0" : "=Q" (fprs[13]));
-	asm volatile("std 14,%0" : "=Q" (fprs[14]));
-	asm volatile("std 15,%0" : "=Q" (fprs[15]));
-}
-
-static inline void restore_fp_regs(freg_t *fprs)
-{
-	asm volatile("ld 0,%0" : : "Q" (fprs[0]));
-	asm volatile("ld 2,%0" : : "Q" (fprs[2]));
-	asm volatile("ld 4,%0" : : "Q" (fprs[4]));
-	asm volatile("ld 6,%0" : : "Q" (fprs[6]));
-	asm volatile("ld 1,%0" : : "Q" (fprs[1]));
-	asm volatile("ld 3,%0" : : "Q" (fprs[3]));
-	asm volatile("ld 5,%0" : : "Q" (fprs[5]));
-	asm volatile("ld 7,%0" : : "Q" (fprs[7]));
-	asm volatile("ld 8,%0" : : "Q" (fprs[8]));
-	asm volatile("ld 9,%0" : : "Q" (fprs[9]));
-	asm volatile("ld 10,%0" : : "Q" (fprs[10]));
-	asm volatile("ld 11,%0" : : "Q" (fprs[11]));
-	asm volatile("ld 12,%0" : : "Q" (fprs[12]));
-	asm volatile("ld 13,%0" : : "Q" (fprs[13]));
-	asm volatile("ld 14,%0" : : "Q" (fprs[14]));
-	asm volatile("ld 15,%0" : : "Q" (fprs[15]));
-}
-
-static inline void save_vx_regs(__vector128 *vxrs)
-{
-	typedef struct { __vector128 _[__NUM_VXRS]; } addrtype;
-
-	asm volatile(
-		"	la	1,%0\n"
-		"	.word	0xe70f,0x1000,0x003e\n"	/* vstm 0,15,0(1) */
-		"	.word	0xe70f,0x1100,0x0c3e\n"	/* vstm 16,31,256(1) */
-		: "=Q" (*(addrtype *) vxrs) : : "1");
-}
-
 static inline void save_vx_regs_safe(__vector128 *vxrs)
 {
 	unsigned long cr0, flags;
@@ -126,20 +61,13 @@ static inline void save_vx_regs_safe(__vector128 *vxrs)
 	__ctl_store(cr0, 0, 0);
 	__ctl_set_bit(0, 17);
 	__ctl_set_bit(0, 18);
-	save_vx_regs(vxrs);
-	__ctl_load(cr0, 0, 0);
-	arch_local_irq_restore(flags);
-}
-
-static inline void restore_vx_regs(__vector128 *vxrs)
-{
-	typedef struct { __vector128 _[__NUM_VXRS]; } addrtype;
-
 	asm volatile(
 		"	la	1,%0\n"
-		"	.word	0xe70f,0x1000,0x0036\n"	/* vlm 0,15,0(1) */
-		"	.word	0xe70f,0x1100,0x0c36\n"	/* vlm 16,31,256(1) */
-		: : "Q" (*(addrtype *) vxrs) : "1");
+		"	.word	0xe70f,0x1000,0x003e\n"	/* vstm 0,15,0(1) */
+		"	.word	0xe70f,0x1100,0x0c3e\n"	/* vstm 16,31,256(1) */
+		: "=Q" (*(struct vx_array *) vxrs) : : "1");
+	__ctl_load(cr0, 0, 0);
+	arch_local_irq_restore(flags);
 }
 
 static inline void convert_vx_to_fp(freg_t *fprs, __vector128 *vxrs)
@@ -177,24 +105,6 @@ static inline void fpregs_load(_s390_fp_regs *fpregs, struct fpu *fpu)
 		       sizeof(fpregs->fprs));
 }
 
-static inline void save_fpu_regs(struct fpu *fpu)
-{
-	save_fp_ctl(&fpu->fpc);
-	if (is_vx_fpu(fpu))
-		save_vx_regs(fpu->vxrs);
-	else
-		save_fp_regs(fpu->fprs);
-}
-
-static inline void restore_fpu_regs(struct fpu *fpu)
-{
-	restore_fp_ctl(&fpu->fpc);
-	if (is_vx_fpu(fpu))
-		restore_vx_regs(fpu->vxrs);
-	else
-		restore_fp_regs(fpu->fprs);
-}
-
 #endif
 
 #endif /* _ASM_S390_FPU_INTERNAL_H */
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
index 3024acbe1f9d..c4f4c52aaa23 100644
--- a/arch/s390/include/asm/kvm_host.h
+++ b/arch/s390/include/asm/kvm_host.h
@@ -22,6 +22,7 @@
 #include <linux/kvm.h>
 #include <asm/debug.h>
 #include <asm/cpu.h>
+#include <asm/fpu-internal.h>
 #include <asm/isc.h>
 
 #define KVM_MAX_VCPUS 64
@@ -498,10 +499,9 @@ struct kvm_guestdbg_info_arch {
 
 struct kvm_vcpu_arch {
 	struct kvm_s390_sie_block *sie_block;
-	s390_fp_regs      host_fpregs;
 	unsigned int      host_acrs[NUM_ACRS];
-	s390_fp_regs      guest_fpregs;
-	struct kvm_s390_vregs	*host_vregs;
+	struct fpu	  host_fpregs;
+	struct fpu	  guest_fpregs;
 	struct kvm_s390_local_interrupt local_int;
 	struct hrtimer    ckc_timer;
 	struct kvm_s390_pgm_info pgm;
diff --git a/arch/s390/include/asm/processor.h b/arch/s390/include/asm/processor.h
index 19f51db7c5e6..c417015c5304 100644
--- a/arch/s390/include/asm/processor.h
+++ b/arch/s390/include/asm/processor.h
@@ -14,10 +14,12 @@
 #define CIF_MCCK_PENDING	0	/* machine check handling is pending */
 #define CIF_ASCE		1	/* user asce needs fixup / uaccess */
 #define CIF_NOHZ_DELAY		2	/* delay HZ disable for a tick */
+#define CIF_FPU			3	/* restore vector registers */
 
 #define _CIF_MCCK_PENDING	(1<<CIF_MCCK_PENDING)
 #define _CIF_ASCE		(1<<CIF_ASCE)
 #define _CIF_NOHZ_DELAY		(1<<CIF_NOHZ_DELAY)
+#define _CIF_FPU		(1<<CIF_FPU)
 
 #ifndef __ASSEMBLY__
 
diff --git a/arch/s390/include/asm/switch_to.h b/arch/s390/include/asm/switch_to.h
index caf4f23462b0..0a4a3150b7d7 100644
--- a/arch/s390/include/asm/switch_to.h
+++ b/arch/s390/include/asm/switch_to.h
@@ -36,7 +36,7 @@ static inline void restore_access_regs(unsigned int *acrs)
 	}								\
 	if (next->mm) {							\
 		update_cr_regs(next);					\
-		restore_fpu_regs(&next->thread.fpu);			\
+		set_cpu_flag(CIF_FPU);					\
 		restore_access_regs(&next->thread.acrs[0]);		\
 		restore_ri_cb(next->thread.ri_cb, prev->thread.ri_cb);	\
 	}								\