s390/kvm: convert to regular kernel fpu user

KVM modifies the kernel fpu's regs pointer to its own area to implement its
custom version of preemtible kernel fpu context. With general support for
preemptible kernel fpu context there is no need for the extra complexity in
KVM code anymore.

Therefore convert KVM to a regular kernel fpu user. In particular this
means that all TIF_FPU checks can be removed, since the fpu register
context will never be changed by other kernel fpu users, and also the fpu
register context will be restored if a thread is preempted.
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

arch/s390/include/asm/kvm_host.h

@@ -743,7 +743,6 @@ struct kvm_vcpu_arch {
 	struct kvm_s390_sie_block *vsie_block;
 	unsigned int      host_acrs[NUM_ACRS];
 	struct gs_cb      *host_gscb;
-	struct fpu	  host_fpregs;
 	struct kvm_s390_local_interrupt local_int;
 	struct hrtimer    ckc_timer;
 	struct kvm_s390_pgm_info pgm;

arch/s390/kernel/entry.S

@@ -220,8 +220,6 @@ SYM_FUNC_START(__sie64a)
 	oi	__SIE_PROG0C+3(%r14),1		# we are going into SIE now
 	tm	__SIE_PROG20+3(%r14),3		# last exit...
 	jnz	.Lsie_skip
-	TSTMSK	__SF_SIE_FLAGS(%r15),_TIF_FPU
-	jo	.Lsie_skip			# exit if fp/vx regs changed
 	lg	%r14,__SF_SIE_CONTROL_PHYS(%r15)	# get sie block phys addr
 	BPEXIT	__SF_SIE_FLAGS(%r15),_TIF_ISOLATE_BP_GUEST
 .Lsie_entry:

arch/s390/kvm/interrupt.c

@@ -584,7 +584,11 @@ static int __write_machine_check(struct kvm_vcpu *vcpu,
 
 	mci.val = mchk->mcic;
 	/* take care of lazy register loading */
-	save_user_fpu_regs();
+	fpu_stfpc(&vcpu->run->s.regs.fpc);
+	if (cpu_has_vx())
+		save_vx_regs((__vector128 *)&vcpu->run->s.regs.vrs);
+	else
+		save_fp_regs((freg_t *)&vcpu->run->s.regs.fprs);
 	save_access_regs(vcpu->run->s.regs.acrs);
 	if (MACHINE_HAS_GS && vcpu->arch.gs_enabled)
 		save_gs_cb(current->thread.gs_cb);
@@ -648,7 +652,7 @@ static int __write_machine_check(struct kvm_vcpu *vcpu,
 	}
 	rc |= write_guest_lc(vcpu, __LC_GPREGS_SAVE_AREA,
 			     vcpu->run->s.regs.gprs, 128);
-	rc |= put_guest_lc(vcpu, current->thread.ufpu.fpc,
+	rc |= put_guest_lc(vcpu, vcpu->run->s.regs.fpc,
 			   (u32 __user *) __LC_FP_CREG_SAVE_AREA);
 	rc |= put_guest_lc(vcpu, vcpu->arch.sie_block->todpr,
 			   (u32 __user *) __LC_TOD_PROGREG_SAVE_AREA);

arch/s390/kvm/kvm-s390.c

@@ -4829,8 +4829,6 @@ static int __vcpu_run(struct kvm_vcpu *vcpu)
 			       vcpu->run->s.regs.gprs,
 			       sizeof(sie_page->pv_grregs));
 		}
-		if (test_thread_flag(TIF_FPU))
-			load_user_fpu_regs();
 		exit_reason = sie64a(vcpu->arch.sie_block,
 				     vcpu->run->s.regs.gprs);
 		if (kvm_s390_pv_cpu_is_protected(vcpu)) {
@@ -4951,16 +4949,11 @@ static void sync_regs(struct kvm_vcpu *vcpu)
 	}
 	save_access_regs(vcpu->arch.host_acrs);
 	restore_access_regs(vcpu->run->s.regs.acrs);
-	/* save host (userspace) fprs/vrs */
-	save_user_fpu_regs();
-	vcpu->arch.host_fpregs.fpc = current->thread.ufpu.fpc;
-	vcpu->arch.host_fpregs.regs = current->thread.ufpu.regs;
+	fpu_lfpc_safe(&vcpu->run->s.regs.fpc);
 	if (cpu_has_vx())
-		current->thread.ufpu.regs = vcpu->run->s.regs.vrs;
+		load_vx_regs((__vector128 *)&vcpu->run->s.regs.vrs);
 	else
-		current->thread.ufpu.regs = vcpu->run->s.regs.fprs;
-	current->thread.ufpu.fpc = vcpu->run->s.regs.fpc;
-
+		load_fp_regs((freg_t *)&vcpu->run->s.regs.fprs);
 	/* Sync fmt2 only data */
 	if (likely(!kvm_s390_pv_cpu_is_protected(vcpu))) {
 		sync_regs_fmt2(vcpu);
@@ -5021,12 +5014,11 @@ static void store_regs(struct kvm_vcpu *vcpu)
 	kvm_run->s.regs.pfc = vcpu->arch.pfault_compare;
 	save_access_regs(vcpu->run->s.regs.acrs);
 	restore_access_regs(vcpu->arch.host_acrs);
-	/* Save guest register state */
-	save_user_fpu_regs();
-	vcpu->run->s.regs.fpc = current->thread.ufpu.fpc;
-	/* Restore will be done lazily at return */
-	current->thread.ufpu.fpc = vcpu->arch.host_fpregs.fpc;
-	current->thread.ufpu.regs = vcpu->arch.host_fpregs.regs;
+	fpu_stfpc(&vcpu->run->s.regs.fpc);
+	if (cpu_has_vx())
+		save_vx_regs((__vector128 *)&vcpu->run->s.regs.vrs);
+	else
+		save_fp_regs((freg_t *)&vcpu->run->s.regs.fprs);
 	if (likely(!kvm_s390_pv_cpu_is_protected(vcpu)))
 		store_regs_fmt2(vcpu);
 }
@@ -5034,6 +5026,7 @@ static void store_regs(struct kvm_vcpu *vcpu)
 int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 {
 	struct kvm_run *kvm_run = vcpu->run;
+	DECLARE_KERNEL_FPU_ONSTACK(fpu);
 	int rc;
 
 	/*
@@ -5075,6 +5068,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		goto out;
 	}
 
+	kernel_fpu_begin(&fpu, KERNEL_FPC | KERNEL_VXR);
 	sync_regs(vcpu);
 	enable_cpu_timer_accounting(vcpu);
 
@@ -5098,6 +5092,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 
 	disable_cpu_timer_accounting(vcpu);
 	store_regs(vcpu);
+	kernel_fpu_end(&fpu, KERNEL_FPC | KERNEL_VXR);
 
 	kvm_sigset_deactivate(vcpu);
 
@@ -5172,8 +5167,11 @@ int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr)
 	 * switch in the run ioctl. Let's update our copies before we save
 	 * it into the save area
 	 */
-	save_user_fpu_regs();
-	vcpu->run->s.regs.fpc = current->thread.ufpu.fpc;
+	fpu_stfpc(&vcpu->run->s.regs.fpc);
+	if (cpu_has_vx())
+		save_vx_regs((__vector128 *)&vcpu->run->s.regs.vrs);
+	else
+		save_fp_regs((freg_t *)&vcpu->run->s.regs.fprs);
 	save_access_regs(vcpu->run->s.regs.acrs);
 
 	return kvm_s390_store_status_unloaded(vcpu, addr);

arch/s390/kvm/vsie.c

@@ -18,7 +18,6 @@
 #include <asm/sclp.h>
 #include <asm/nmi.h>
 #include <asm/dis.h>
-#include <asm/fpu.h>
 #include <asm/facility.h>
 #include "kvm-s390.h"
 #include "gaccess.h"
@@ -1149,8 +1148,6 @@ static int do_vsie_run(struct kvm_vcpu *vcpu, struct vsie_page *vsie_page)
 	 */
 	vcpu->arch.sie_block->prog0c |= PROG_IN_SIE;
 	barrier();
-	if (test_thread_flag(TIF_FPU))
-		load_user_fpu_regs();
 	if (!kvm_s390_vcpu_sie_inhibited(vcpu))
 		rc = sie64a(scb_s, vcpu->run->s.regs.gprs);
 	barrier();