Merge tag 's390-6.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull s390 updates from Alexander Gordeev:

 - Add machine variable capacity information to /proc/sysinfo.

 - Limit the waste of page tables and always align vmalloc area size and
   base address on segment boundary.

 - Fix a memory leak when an attempt to register interruption sub class
   (ISC) for the adjunct-processor (AP) guest failed.

 - Reset response code AP_RESPONSE_INVALID_GISA to understandable by
   guest AP_RESPONSE_INVALID_ADDRESS in response to a failed
   interruption sub class (ISC) registration attempt.

 - Improve reaction to adjunct-processor (AP)
   AP_RESPONSE_OTHERWISE_CHANGED response code when enabling interrupts
   on behalf of a guest.

 - Fix incorrect sysfs 'status' attribute of adjunct-processor (AP)
   queue device bound to the vfio_ap device driver when the mediated
   device is attached to a guest, but the queue device is not passed
   through.

 - Rework struct ap_card to hold the whole adjunct-processor (AP) card
   hardware information. As result, all the ugly bit checks are replaced
   by simple evaluations of the required bit fields.

 - Improve handling of some weird scenarios between service element (SE)
   host and SE guest with adjunct-processor (AP) pass-through support.

 - Change local_ctl_set_bit() and local_ctl_clear_bit() so they return
   the previous value of the to be changed control register. This is
   useful if a bit is only changed temporarily and the previous content
   needs to be restored.

 - The kernel starts with machine checks disabled and is expected to
   enable it once trap_init() is called. However the implementation
   allows machine checks early. Consistently enable it in trap_init()
   only.

 - local_mcck_disable() and local_mcck_enable() assume that machine
   checks are always enabled. Instead implement and use
   local_mcck_save() and local_mcck_restore() to disable machine checks
   and restore the previous state.

 - Modification of floating point control (FPC) register of a traced
   process using ptrace interface may lead to corruption of the FPC
   register of the tracing process. Fix this.

 - kvm_arch_vcpu_ioctl_set_fpu() allows to set the floating point
   control (FPC) register in vCPU, but may lead to corruption of the FPC
   register of the host process. Fix this.

 - Use READ_ONCE() to read a vCPU floating point register value from the
   memory mapped area. This avoids that, depending on code generation, a
   different value is tested for validity than the one that is used.

 - Get rid of test_fp_ctl(), since it is quite subtle to use it
   correctly. Instead copy a new floating point control register value
   into its save area and test the validity of the new value when
   loading it.

 - Remove superfluous save_fpu_regs() call.

 - Remove s390 support for ARCH_WANTS_DYNAMIC_TASK_STRUCT. All machines
   provide the vector facility since many years and the need to make the
   task structure size dependent on the vector facility does not exist.

 - Remove the "novx" kernel command line option, as the vector code runs
   without any problems since many years.

 - Add the vector facility to the z13 architecture level set (ALS). All
   hypervisors support the vector facility since many years. This allows
   compile time optimizations of the kernel.

 - Get rid of MACHINE_HAS_VX and replace it with cpu_has_vx(). As
   result, the compiled code will have less runtime checks and less
   code.

 - Convert pgste_get_lock() and pgste_set_unlock() ASM inlines to C.

 - Convert the struct subchannel spinlock from pointer to member.

* tag 's390-6.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (24 commits)
  Revert "s390: update defconfigs"
  s390/cio: make sch->lock spinlock pointer a member
  s390: update defconfigs
  s390/mm: convert pgste locking functions to C
  s390/fpu: get rid of MACHINE_HAS_VX
  s390/als: add vector facility to z13 architecture level set
  s390/fpu: remove "novx" option
  s390/fpu: remove ARCH_WANTS_DYNAMIC_TASK_STRUCT support
  KVM: s390: remove superfluous save_fpu_regs() call
  s390/fpu: get rid of test_fp_ctl()
  KVM: s390: use READ_ONCE() to read fpc register value
  KVM: s390: fix setting of fpc register
  s390/ptrace: handle setting of fpc register correctly
  s390/nmi: implement and use local_mcck_save() / local_mcck_restore()
  s390/nmi: consistently enable machine checks in trap_init()
  s390/ctlreg: return old register contents when changing bits
  s390/ap: handle outband SE bind state change
  s390/ap: store TAPQ hwinfo in struct ap_card
  s390/vfio-ap: fix sysfs status attribute for AP queue devices
  s390/vfio-ap: improve reaction to response code 07 from PQAP(AQIC) command
  ...

32 files changed, 190 insertions, 179 deletions

diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 959adae10f80..8f39f0424796 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -123,7 +123,6 @@ config S390
 	select ARCH_USE_BUILTIN_BSWAP
 	select ARCH_USE_CMPXCHG_LOCKREF
 	select ARCH_USE_SYM_ANNOTATIONS
-	select ARCH_WANTS_DYNAMIC_TASK_STRUCT
 	select ARCH_WANTS_NO_INSTR
 	select ARCH_WANT_DEFAULT_BPF_JIT
 	select ARCH_WANT_IPC_PARSE_VERSION
diff --git a/arch/s390/boot/ipl_parm.c b/arch/s390/boot/ipl_parm.c
index 2ab4872fbee1..b24de9aabf7d 100644
--- a/arch/s390/boot/ipl_parm.c
+++ b/arch/s390/boot/ipl_parm.c
@@ -274,7 +274,7 @@ void parse_boot_command_line(void)
 			memory_limit = round_down(memparse(val, NULL), PAGE_SIZE);
 
 		if (!strcmp(param, "vmalloc") && val) {
-			vmalloc_size = round_up(memparse(val, NULL), PAGE_SIZE);
+			vmalloc_size = round_up(memparse(val, NULL), _SEGMENT_SIZE);
 			vmalloc_size_set = 1;
 		}
 
diff --git a/arch/s390/boot/startup.c b/arch/s390/boot/startup.c
index 8104e0e3d188..9cc76e631759 100644
--- a/arch/s390/boot/startup.c
+++ b/arch/s390/boot/startup.c
@@ -255,7 +255,8 @@ static unsigned long setup_kernel_memory_layout(void)
 	VMALLOC_END = MODULES_VADDR;
 
 	/* allow vmalloc area to occupy up to about 1/2 of the rest virtual space left */
-	vmalloc_size = min(vmalloc_size, round_down(VMALLOC_END / 2, _REGION3_SIZE));
+	vsize = round_down(VMALLOC_END / 2, _SEGMENT_SIZE);
+	vmalloc_size = min(vmalloc_size, vsize);
 	VMALLOC_START = VMALLOC_END - vmalloc_size;
 
 	/* split remaining virtual space between 1:1 mapping & vmemmap array */
diff --git a/arch/s390/crypto/chacha-glue.c b/arch/s390/crypto/chacha-glue.c
index 5fae187f947a..ed9959e6f714 100644
--- a/arch/s390/crypto/chacha-glue.c
+++ b/arch/s390/crypto/chacha-glue.c
@@ -82,7 +82,7 @@ void chacha_crypt_arch(u32 *state, u8 *dst, const u8 *src,
 	 * it cannot handle a block of data or less, but otherwise
 	 * it can handle data of arbitrary size
 	 */
-	if (bytes <= CHACHA_BLOCK_SIZE || nrounds != 20 || !MACHINE_HAS_VX)
+	if (bytes <= CHACHA_BLOCK_SIZE || nrounds != 20 || !cpu_has_vx())
 		chacha_crypt_generic(state, dst, src, bytes, nrounds);
 	else
 		chacha20_crypt_s390(state, dst, src, bytes,
diff --git a/arch/s390/include/asm/ap.h b/arch/s390/include/asm/ap.h
index 40c2b82f083a..43ac4a64f49b 100644
--- a/arch/s390/include/asm/ap.h
+++ b/arch/s390/include/asm/ap.h
@@ -88,7 +88,7 @@ static inline bool ap_instructions_available(void)
 }
 
 /* TAPQ register GR2 response struct */
-struct ap_tapq_gr2 {
+struct ap_tapq_hwinfo {
 	union {
 		unsigned long value;
 		struct {
@@ -96,11 +96,13 @@ struct ap_tapq_gr2 {
 			unsigned int apinfo : 32; /* ap type, ... */
 		};
 		struct {
-			unsigned int s	   :  1; /* APSC */
-			unsigned int m	   :  1; /* AP4KM */
-			unsigned int c	   :  1; /* AP4KC */
-			unsigned int mode  :  3;
-			unsigned int n	   :  1; /* APXA */
+			unsigned int apsc  :  1; /* APSC */
+			unsigned int mex4k :  1; /* AP4KM */
+			unsigned int crt4k :  1; /* AP4KC */
+			unsigned int cca   :  1; /* D */
+			unsigned int accel :  1; /* A */
+			unsigned int ep11  :  1; /* X */
+			unsigned int apxa  :  1; /* APXA */
 			unsigned int	   :  1;
 			unsigned int class :  8;
 			unsigned int bs	   :  2; /* SE bind/assoc */
@@ -126,11 +128,12 @@ struct ap_tapq_gr2 {
 /**
  * ap_tapq(): Test adjunct processor queue.
  * @qid: The AP queue number
- * @info: Pointer to queue descriptor
+ * @info: Pointer to tapq hwinfo struct
  *
  * Returns AP queue status structure.
  */
-static inline struct ap_queue_status ap_tapq(ap_qid_t qid, struct ap_tapq_gr2 *info)
+static inline struct ap_queue_status ap_tapq(ap_qid_t qid,
+					     struct ap_tapq_hwinfo *info)
 {
 	union ap_queue_status_reg reg1;
 	unsigned long reg2;
@@ -158,7 +161,7 @@ static inline struct ap_queue_status ap_tapq(ap_qid_t qid, struct ap_tapq_gr2 *i
  * Returns AP queue status structure.
  */
 static inline struct ap_queue_status ap_test_queue(ap_qid_t qid, int tbit,
-						   struct ap_tapq_gr2 *info)
+						   struct ap_tapq_hwinfo *info)
 {
 	if (tbit)
 		qid |= 1UL << 23; /* set T bit*/
diff --git a/arch/s390/include/asm/ctlreg.h b/arch/s390/include/asm/ctlreg.h
index 6d4b85f2b541..72a9556d04f3 100644
--- a/arch/s390/include/asm/ctlreg.h
+++ b/arch/s390/include/asm/ctlreg.h
@@ -141,22 +141,26 @@ static __always_inline void local_ctl_store(unsigned int cr, struct ctlreg *reg)
 		: [cr] "i" (cr));
 }
 
-static __always_inline void local_ctl_set_bit(unsigned int cr, unsigned int bit)
+static __always_inline struct ctlreg local_ctl_set_bit(unsigned int cr, unsigned int bit)
 {
-	struct ctlreg reg;
+	struct ctlreg new, old;
 
-	local_ctl_store(cr, &reg);
-	reg.val |= 1UL << bit;
-	local_ctl_load(cr, &reg);
+	local_ctl_store(cr, &old);
+	new = old;
+	new.val |= 1UL << bit;
+	local_ctl_load(cr, &new);
+	return old;
 }
 
-static __always_inline void local_ctl_clear_bit(unsigned int cr, unsigned int bit)
+static __always_inline struct ctlreg local_ctl_clear_bit(unsigned int cr, unsigned int bit)
 {
-	struct ctlreg reg;
+	struct ctlreg new, old;
 
-	local_ctl_store(cr, &reg);
-	reg.val &= ~(1UL << bit);
-	local_ctl_load(cr, &reg);
+	local_ctl_store(cr, &old);
+	new = old;
+	new.val &= ~(1UL << bit);
+	local_ctl_load(cr, &new);
+	return old;
 }
 
 struct lowcore;
diff --git a/arch/s390/include/asm/fpu/api.h b/arch/s390/include/asm/fpu/api.h
index 9acf48e53a87..d6ca8bc6ca68 100644
--- a/arch/s390/include/asm/fpu/api.h
+++ b/arch/s390/include/asm/fpu/api.h
@@ -46,26 +46,33 @@
 
 #include <linux/preempt.h>
 #include <asm/asm-extable.h>
+#include <asm/fpu/internal.h>
 
 void save_fpu_regs(void);
 void load_fpu_regs(void);
 void __load_fpu_regs(void);
 
-static inline int test_fp_ctl(u32 fpc)
+/**
+ * sfpc_safe - Set floating point control register safely.
+ * @fpc: new value for floating point control register
+ *
+ * Set floating point control register. This may lead to an exception,
+ * since a saved value may have been modified by user space (ptrace,
+ * signal return, kvm registers) to an invalid value. In such a case
+ * set the floating point control register to zero.
+ */
+static inline void sfpc_safe(u32 fpc)
 {
-	u32 orig_fpc;
-	int rc;
-
-	asm volatile(
-		"	efpc    %1\n"
-		"	sfpc	%2\n"
-		"0:	sfpc	%1\n"
-		"	la	%0,0\n"
-		"1:\n"
-		EX_TABLE(0b,1b)
-		: "=d" (rc), "=&d" (orig_fpc)
-		: "d" (fpc), "0" (-EINVAL));
-	return rc;
+	asm volatile("\n"
+		"0:	sfpc	%[fpc]\n"
+		"1:	nopr	%%r7\n"
+		".pushsection .fixup, \"ax\"\n"
+		"2:	lghi	%[fpc],0\n"
+		"	jg	0b\n"
+		".popsection\n"
+		EX_TABLE(1b, 2b)
+		: [fpc] "+d" (fpc)
+		: : "memory");
 }
 
 #define KERNEL_FPC		1
diff --git a/arch/s390/include/asm/fpu/internal.h b/arch/s390/include/asm/fpu/internal.h
index 8634581b9011..d511c4cf5afb 100644
--- a/arch/s390/include/asm/fpu/internal.h
+++ b/arch/s390/include/asm/fpu/internal.h
@@ -10,8 +10,14 @@
 #define _ASM_S390_FPU_INTERNAL_H
 
 #include <linux/string.h>
+#include <asm/facility.h>
 #include <asm/fpu/types.h>
 
+static inline bool cpu_has_vx(void)
+{
+	return likely(test_facility(129));
+}
+
 static inline void save_vx_regs(__vector128 *vxrs)
 {
 	asm volatile(
@@ -41,7 +47,7 @@ static inline void fpregs_store(_s390_fp_regs *fpregs, struct fpu *fpu)
 {
 	fpregs->pad = 0;
 	fpregs->fpc = fpu->fpc;
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_vx_to_fp((freg_t *)&fpregs->fprs, fpu->vxrs);
 	else
 		memcpy((freg_t *)&fpregs->fprs, fpu->fprs,
@@ -51,7 +57,7 @@ static inline void fpregs_store(_s390_fp_regs *fpregs, struct fpu *fpu)
 static inline void fpregs_load(_s390_fp_regs *fpregs, struct fpu *fpu)
 {
 	fpu->fpc = fpregs->fpc;
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_fp_to_vx(fpu->vxrs, (freg_t *)&fpregs->fprs);
 	else
 		memcpy(fpu->fprs, (freg_t *)&fpregs->fprs,
diff --git a/arch/s390/include/asm/processor.h b/arch/s390/include/asm/processor.h
index c15eadbb9983..c0b6e74d899a 100644
--- a/arch/s390/include/asm/processor.h
+++ b/arch/s390/include/asm/processor.h
@@ -184,11 +184,7 @@ struct thread_struct {
 	struct gs_cb *gs_cb;			/* Current guarded storage cb */
 	struct gs_cb *gs_bc_cb;			/* Broadcast guarded storage cb */
 	struct pgm_tdb trap_tdb;		/* Transaction abort diagnose block */
-	/*
-	 * Warning: 'fpu' is dynamically-sized. It *MUST* be at
-	 * the end.
-	 */
-	struct fpu fpu;			/* FP and VX register save area */
+	struct fpu fpu;				/* FP and VX register save area */
 };
 
 /* Flag to disable transactions. */
@@ -331,14 +327,36 @@ static inline unsigned long __extract_psw(void)
 	return (((unsigned long) reg1) << 32) | ((unsigned long) reg2);
 }
 
-static inline void local_mcck_enable(void)
+static inline unsigned long __local_mcck_save(void)
 {
-	__load_psw_mask(__extract_psw() | PSW_MASK_MCHECK);
+	unsigned long mask = __extract_psw();
+
+	__load_psw_mask(mask & ~PSW_MASK_MCHECK);
+	return mask & PSW_MASK_MCHECK;
+}
+
+#define local_mcck_save(mflags)			\
+do {						\
+	typecheck(unsigned long, mflags);	\
+	mflags = __local_mcck_save();		\
+} while (0)
+
+static inline void local_mcck_restore(unsigned long mflags)
+{
+	unsigned long mask = __extract_psw();
+
+	mask &= ~PSW_MASK_MCHECK;
+	__load_psw_mask(mask | mflags);
 }
 
 static inline void local_mcck_disable(void)
 {
-	__load_psw_mask(__extract_psw() & ~PSW_MASK_MCHECK);
+	__local_mcck_save();
+}
+
+static inline void local_mcck_enable(void)
+{
+	__load_psw_mask(__extract_psw() | PSW_MASK_MCHECK);
 }
 
 /*
diff --git a/arch/s390/include/asm/setup.h b/arch/s390/include/asm/setup.h
index df316436d2e1..03bcaa8effb2 100644
--- a/arch/s390/include/asm/setup.h
+++ b/arch/s390/include/asm/setup.h
@@ -28,7 +28,6 @@
 #define MACHINE_FLAG_TOPOLOGY	BIT(10)
 #define MACHINE_FLAG_TE		BIT(11)
 #define MACHINE_FLAG_TLB_LC	BIT(12)
-#define MACHINE_FLAG_VX		BIT(13)
 #define MACHINE_FLAG_TLB_GUEST	BIT(14)
 #define MACHINE_FLAG_NX		BIT(15)
 #define MACHINE_FLAG_GS		BIT(16)
@@ -90,7 +89,6 @@ extern unsigned long mio_wb_bit_mask;
 #define MACHINE_HAS_TOPOLOGY	(S390_lowcore.machine_flags & MACHINE_FLAG_TOPOLOGY)
 #define MACHINE_HAS_TE		(S390_lowcore.machine_flags & MACHINE_FLAG_TE)
 #define MACHINE_HAS_TLB_LC	(S390_lowcore.machine_flags & MACHINE_FLAG_TLB_LC)
-#define MACHINE_HAS_VX		(S390_lowcore.machine_flags & MACHINE_FLAG_VX)
 #define MACHINE_HAS_TLB_GUEST	(S390_lowcore.machine_flags & MACHINE_FLAG_TLB_GUEST)
 #define MACHINE_HAS_NX		(S390_lowcore.machine_flags & MACHINE_FLAG_NX)
 #define MACHINE_HAS_GS		(S390_lowcore.machine_flags & MACHINE_FLAG_GS)
diff --git a/arch/s390/include/asm/sysinfo.h b/arch/s390/include/asm/sysinfo.h
index ab1c6316055c..edca5a751df4 100644
--- a/arch/s390/include/asm/sysinfo.h
+++ b/arch/s390/include/asm/sysinfo.h
@@ -40,6 +40,10 @@ struct sysinfo_1_1_1 {
 	unsigned int ncr;
 	unsigned int npr;
 	unsigned int ntr;
+	char reserved_3[4];
+	char model_var_cap[16];
+	unsigned int model_var_cap_rating;
+	unsigned int nvr;
 };
 
 struct sysinfo_1_2_1 {
diff --git a/arch/s390/kernel/compat_signal.c b/arch/s390/kernel/compat_signal.c
index cecedd01d4ec..f8fc6c25d051 100644
--- a/arch/s390/kernel/compat_signal.c
+++ b/arch/s390/kernel/compat_signal.c
@@ -29,6 +29,7 @@
 #include <asm/lowcore.h>
 #include <asm/switch_to.h>
 #include <asm/vdso.h>
+#include <asm/fpu/api.h>
 #include "compat_linux.h"
 #include "compat_ptrace.h"
 #include "entry.h"
@@ -98,10 +99,6 @@ static int restore_sigregs32(struct pt_regs *regs,_sigregs32 __user *sregs)
 	if (!is_ri_task(current) && (user_sregs.regs.psw.mask & PSW32_MASK_RI))
 		return -EINVAL;
 
-	/* Test the floating-point-control word. */
-	if (test_fp_ctl(user_sregs.fpregs.fpc))
-		return -EINVAL;
-
 	/* Use regs->psw.mask instead of PSW_USER_BITS to preserve PER bit. */
 	regs->psw.mask = (regs->psw.mask & ~(PSW_MASK_USER | PSW_MASK_RI)) |
 		(__u64)(user_sregs.regs.psw.mask & PSW32_MASK_USER) << 32 |
@@ -137,7 +134,7 @@ static int save_sigregs_ext32(struct pt_regs *regs,
 		return -EFAULT;
 
 	/* Save vector registers to signal stack */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		for (i = 0; i < __NUM_VXRS_LOW; i++)
 			vxrs[i] = current->thread.fpu.vxrs[i].low;
 		if (__copy_to_user(&sregs_ext->vxrs_low, vxrs,
@@ -165,7 +162,7 @@ static int restore_sigregs_ext32(struct pt_regs *regs,
 		*(__u32 *)&regs->gprs[i] = gprs_high[i];
 
 	/* Restore vector registers from signal stack */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		if (__copy_from_user(vxrs, &sregs_ext->vxrs_low,
 				     sizeof(sregs_ext->vxrs_low)) ||
 		    __copy_from_user(current->thread.fpu.vxrs + __NUM_VXRS_LOW,
@@ -265,7 +262,7 @@ static int setup_frame32(struct ksignal *ksig, sigset_t *set,
 	 * the machine supports it
 	 */
 	frame_size = sizeof(*frame) - sizeof(frame->sregs_ext.__reserved);
-	if (!MACHINE_HAS_VX)
+	if (!cpu_has_vx())
 		frame_size -= sizeof(frame->sregs_ext.vxrs_low) +
 			      sizeof(frame->sregs_ext.vxrs_high);
 	frame = get_sigframe(&ksig->ka, regs, frame_size);
@@ -348,11 +345,12 @@ static int setup_rt_frame32(struct ksignal *ksig, sigset_t *set,
 	 * the machine supports it
 	 */
 	uc_flags = UC_GPRS_HIGH;
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		uc_flags |= UC_VXRS;
-	} else
+	} else {
 		frame_size -= sizeof(frame->uc.uc_mcontext_ext.vxrs_low) +
 			      sizeof(frame->uc.uc_mcontext_ext.vxrs_high);
+	}
 	frame = get_sigframe(&ksig->ka, regs, frame_size);
 	if (frame == (void __user *) -1UL)
 		return -EFAULT;
diff --git a/arch/s390/kernel/crash_dump.c b/arch/s390/kernel/crash_dump.c
index 514feadd4c58..5c46c2659305 100644
--- a/arch/s390/kernel/crash_dump.c
+++ b/arch/s390/kernel/crash_dump.c
@@ -22,6 +22,7 @@
 #include <asm/ipl.h>
 #include <asm/sclp.h>
 #include <asm/maccess.h>
+#include <asm/fpu/api.h>
 
 #define PTR_ADD(x, y) (((char *) (x)) + ((unsigned long) (y)))
 #define PTR_SUB(x, y) (((char *) (x)) - ((unsigned long) (y)))
@@ -319,7 +320,7 @@ static void *fill_cpu_elf_notes(void *ptr, int cpu, struct save_area *sa)
 	ptr = nt_init(ptr, NT_S390_TODPREG, &sa->todpreg, sizeof(sa->todpreg));
 	ptr = nt_init(ptr, NT_S390_CTRS, &sa->ctrs, sizeof(sa->ctrs));
 	ptr = nt_init(ptr, NT_S390_PREFIX, &sa->prefix, sizeof(sa->prefix));
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		ptr = nt_init(ptr, NT_S390_VXRS_HIGH,
 			      &sa->vxrs_high, sizeof(sa->vxrs_high));
 		ptr = nt_init(ptr, NT_S390_VXRS_LOW,
@@ -343,7 +344,7 @@ static size_t get_cpu_elf_notes_size(void)
 	size +=  nt_size(NT_S390_TODPREG, sizeof(sa->todpreg));
 	size +=  nt_size(NT_S390_CTRS, sizeof(sa->ctrs));
 	size +=  nt_size(NT_S390_PREFIX, sizeof(sa->prefix));
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		size += nt_size(NT_S390_VXRS_HIGH, sizeof(sa->vxrs_high));
 		size += nt_size(NT_S390_VXRS_LOW, sizeof(sa->vxrs_low));
 	}
diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c
index eb43e5922a25..2345ea332b97 100644
--- a/arch/s390/kernel/early.c
+++ b/arch/s390/kernel/early.c
@@ -229,10 +229,8 @@ static __init void detect_machine_facilities(void)
 	}
 	if (test_facility(51))
 		S390_lowcore.machine_flags |= MACHINE_FLAG_TLB_LC;
-	if (test_facility(129)) {
-		S390_lowcore.machine_flags |= MACHINE_FLAG_VX;
+	if (test_facility(129))
 		system_ctl_set_bit(0, CR0_VECTOR_BIT);
-	}
 	if (test_facility(130))
 		S390_lowcore.machine_flags |= MACHINE_FLAG_NX;
 	if (test_facility(133))
@@ -271,14 +269,6 @@ static inline void setup_access_registers(void)
 	restore_access_regs(acrs);
 }
 
-static int __init disable_vector_extension(char *str)
-{
-	S390_lowcore.machine_flags &= ~MACHINE_FLAG_VX;
-	system_ctl_clear_bit(0, CR0_VECTOR_BIT);
-	return 0;
-}
-early_param("novx", disable_vector_extension);
-
 char __bootdata(early_command_line)[COMMAND_LINE_SIZE];
 static void __init setup_boot_command_line(void)
 {
diff --git a/arch/s390/kernel/fpu.c b/arch/s390/kernel/fpu.c
index 4666b29ac8a1..9e7c15fccfea 100644
--- a/arch/s390/kernel/fpu.c
+++ b/arch/s390/kernel/fpu.c
@@ -24,7 +24,7 @@ void __kernel_fpu_begin(struct kernel_fpu *state, u32 flags)
 		/* Save floating point control */
 		asm volatile("stfpc %0" : "=Q" (state->fpc));
 
-	if (!MACHINE_HAS_VX) {
+	if (!cpu_has_vx()) {
 		if (flags & KERNEL_VXR_V0V7) {
 			/* Save floating-point registers */
 			asm volatile("std 0,%0" : "=Q" (state->fprs[0]));
@@ -106,7 +106,7 @@ void __kernel_fpu_end(struct kernel_fpu *state, u32 flags)
 		/* Restore floating-point controls */
 		asm volatile("lfpc %0" : : "Q" (state->fpc));
 
-	if (!MACHINE_HAS_VX) {
+	if (!cpu_has_vx()) {
 		if (flags & KERNEL_VXR_V0V7) {
 			/* Restore floating-point registers */
 			asm volatile("ld 0,%0" : : "Q" (state->fprs[0]));
@@ -177,11 +177,11 @@ EXPORT_SYMBOL(__kernel_fpu_end);
 
 void __load_fpu_regs(void)
 {
-	struct fpu *state = &current->thread.fpu;
 	unsigned long *regs = current->thread.fpu.regs;
+	struct fpu *state = &current->thread.fpu;
 
-	asm volatile("lfpc %0" : : "Q" (state->fpc));
-	if (likely(MACHINE_HAS_VX)) {
+	sfpc_safe(state->fpc);
+	if (likely(cpu_has_vx())) {
 		asm volatile("lgr	1,%0\n"
 			     "VLM	0,15,0,1\n"
 			     "VLM	16,31,256,1\n"
@@ -232,7 +232,7 @@ void save_fpu_regs(void)
 	regs = current->thread.fpu.regs;
 
 	asm volatile("stfpc %0" : "=Q" (state->fpc));
-	if (likely(MACHINE_HAS_VX)) {
+	if (likely(cpu_has_vx())) {
 		asm volatile("lgr	1,%0\n"
 			     "VSTM	0,15,0,1\n"
 			     "VSTM	16,31,256,1\n"
diff --git a/arch/s390/kernel/machine_kexec.c b/arch/s390/kernel/machine_kexec.c
index bb0d4d68fcbe..aa22ffc16bcd 100644
--- a/arch/s390/kernel/machine_kexec.c
+++ b/arch/s390/kernel/machine_kexec.c
@@ -91,7 +91,7 @@ static noinline void __machine_kdump(void *image)
 	}
 	/* Store status of the boot CPU */
 	mcesa = __va(S390_lowcore.mcesad & MCESA_ORIGIN_MASK);
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		save_vx_regs((__vector128 *) mcesa->vector_save_area);
 	if (MACHINE_HAS_GS) {
 		local_ctl_store(2, &cr2_old.reg);
diff --git a/arch/s390/kernel/nmi.c b/arch/s390/kernel/nmi.c
index 0daf0f1cdfc9..9ad44c26d1a2 100644
--- a/arch/s390/kernel/nmi.c
+++ b/arch/s390/kernel/nmi.c
@@ -32,6 +32,7 @@
 #include <asm/asm-offsets.h>
 #include <asm/pai.h>
 #include <asm/vx-insn.h>
+#include <asm/fpu/api.h>
 
 struct mcck_struct {
 	unsigned int kill_task : 1;
@@ -45,7 +46,7 @@ static DEFINE_PER_CPU(struct mcck_struct, cpu_mcck);
 
 static inline int nmi_needs_mcesa(void)
 {
-	return MACHINE_HAS_VX || MACHINE_HAS_GS;
+	return cpu_has_vx() || MACHINE_HAS_GS;
 }
 
 /*
@@ -159,16 +160,17 @@ NOKPROBE_SYMBOL(s390_handle_damage);
 void s390_handle_mcck(void)
 {
 	struct mcck_struct mcck;
+	unsigned long mflags;
 
 	/*
 	 * Disable machine checks and get the current state of accumulated
 	 * machine checks. Afterwards delete the old state and enable machine
 	 * checks again.
 	 */
-	local_mcck_disable();
+	local_mcck_save(mflags);
 	mcck = *this_cpu_ptr(&cpu_mcck);
 	memset(this_cpu_ptr(&cpu_mcck), 0, sizeof(mcck));
-	local_mcck_enable();
+	local_mcck_restore(mflags);
 
 	if (mcck.channel_report)
 		crw_handle_channel_report();
@@ -234,7 +236,7 @@ static int notrace s390_validate_registers(union mci mci)
 	}
 
 	mcesa = __va(S390_lowcore.mcesad & MCESA_ORIGIN_MASK);
-	if (!MACHINE_HAS_VX) {
+	if (!cpu_has_vx()) {
 		/* Validate floating point registers */
 		asm volatile(
 			"	ld	0,0(%0)\n"
diff --git a/arch/s390/kernel/perf_regs.c b/arch/s390/kernel/perf_regs.c
index 6e9e5d5e927e..3d93656bd948 100644
--- a/arch/s390/kernel/perf_regs.c
+++ b/arch/s390/kernel/perf_regs.c
@@ -20,8 +20,10 @@ u64 perf_reg_value(struct pt_regs *regs, int idx)
 			return 0;
 
 		idx -= PERF_REG_S390_FP0;
-		fp = MACHINE_HAS_VX ? *(freg_t *)(current->thread.fpu.vxrs + idx)
-				    : current->thread.fpu.fprs[idx];
+		if (cpu_has_vx())
+			fp = *(freg_t *)(current->thread.fpu.vxrs + idx);
+		else
+			fp = current->thread.fpu.fprs[idx];
 		return fp.ui;
 	}
 
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index 258000417724..4e3b366589fb 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -89,7 +89,7 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
 	 */
 	save_fpu_regs();
 
-	memcpy(dst, src, arch_task_struct_size);
+	*dst = *src;
 	dst->thread.fpu.regs = dst->thread.fpu.fprs;
 
 	/*
diff --git a/arch/s390/kernel/processor.c b/arch/s390/kernel/processor.c
index 0a999c8226d7..65c1464eea4f 100644
--- a/arch/s390/kernel/processor.c
+++ b/arch/s390/kernel/processor.c
@@ -201,11 +201,8 @@ static int __init setup_hwcaps(void)
 	if (MACHINE_HAS_TE)
 		elf_hwcap |= HWCAP_TE;
 
-	/*
-	 * Vector extension can be disabled with the "novx" parameter.
-	 * Use MACHINE_HAS_VX instead of facility bit 129.
-	 */
-	if (MACHINE_HAS_VX) {
+	/* vector */
+	if (test_facility(129)) {
 		elf_hwcap |= HWCAP_VXRS;
 		if (test_facility(134))
 			elf_hwcap |= HWCAP_VXRS_BCD;
diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c
index 046403471c5d..2e6754b62b20 100644
--- a/arch/s390/kernel/ptrace.c
+++ b/arch/s390/kernel/ptrace.c
@@ -30,6 +30,7 @@
 #include <asm/switch_to.h>
 #include <asm/runtime_instr.h>
 #include <asm/facility.h>
+#include <asm/fpu/api.h>
 
 #include "entry.h"
 
@@ -254,7 +255,7 @@ static unsigned long __peek_user(struct task_struct *child, addr_t addr)
 		 * or the child->thread.fpu.vxrs array
 		 */
 		offset = addr - offsetof(struct user, regs.fp_regs.fprs);
-		if (MACHINE_HAS_VX)
+		if (cpu_has_vx())
 			tmp = *(addr_t *)
 			       ((addr_t) child->thread.fpu.vxrs + 2*offset);
 		else
@@ -392,8 +393,7 @@ static int __poke_user(struct task_struct *child, addr_t addr, addr_t data)
 		/*
 		 * floating point control reg. is in the thread structure
 		 */
-		if ((unsigned int) data != 0 ||
-		    test_fp_ctl(data >> (BITS_PER_LONG - 32)))
+		if ((unsigned int)data != 0)
 			return -EINVAL;
 		child->thread.fpu.fpc = data >> (BITS_PER_LONG - 32);
 
@@ -403,7 +403,7 @@ static int __poke_user(struct task_struct *child, addr_t addr, addr_t data)
 		 * or the child->thread.fpu.vxrs array
 		 */
 		offset = addr - offsetof(struct user, regs.fp_regs.fprs);
-		if (MACHINE_HAS_VX)
+		if (cpu_has_vx())
 			*(addr_t *)((addr_t)
 				child->thread.fpu.vxrs + 2*offset) = data;
 		else
@@ -630,7 +630,7 @@ static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
 		 * or the child->thread.fpu.vxrs array
 		 */
 		offset = addr - offsetof(struct compat_user, regs.fp_regs.fprs);
-		if (MACHINE_HAS_VX)
+		if (cpu_has_vx())
 			tmp = *(__u32 *)
 			       ((addr_t) child->thread.fpu.vxrs + 2*offset);
 		else
@@ -748,8 +748,6 @@ static int __poke_user_compat(struct task_struct *child,
 		/*
 		 * floating point control reg. is in the thread structure
 		 */
-		if (test_fp_ctl(tmp))
-			return -EINVAL;
 		child->thread.fpu.fpc = data;
 
 	} else if (addr < offsetof(struct compat_user, regs.fp_regs) + sizeof(s390_fp_regs)) {
@@ -758,7 +756,7 @@ static int __poke_user_compat(struct task_struct *child,
 		 * or the child->thread.fpu.vxrs array
 		 */
 		offset = addr - offsetof(struct compat_user, regs.fp_regs.fprs);
-		if (MACHINE_HAS_VX)
+		if (cpu_has_vx())
 			*(__u32 *)((addr_t)
 				child->thread.fpu.vxrs + 2*offset) = tmp;
 		else
@@ -914,7 +912,7 @@ static int s390_fpregs_set(struct task_struct *target,
 	if (target == current)
 		save_fpu_regs();
 
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_vx_to_fp(fprs, target->thread.fpu.vxrs);
 	else
 		memcpy(&fprs, target->thread.fpu.fprs, sizeof(fprs));
@@ -926,7 +924,7 @@ static int s390_fpregs_set(struct task_struct *target,
 					0, offsetof(s390_fp_regs, fprs));
 		if (rc)
 			return rc;
-		if (ufpc[1] != 0 || test_fp_ctl(ufpc[0]))
+		if (ufpc[1] != 0)
 			return -EINVAL;
 		target->thread.fpu.fpc = ufpc[0];
 	}
@@ -937,7 +935,7 @@ static int s390_fpregs_set(struct task_struct *target,
 	if (rc)
 		return rc;
 
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_fp_to_vx(target->thread.fpu.vxrs, fprs);
 	else
 		memcpy(target->thread.fpu.fprs, &fprs, sizeof(fprs));
@@ -988,7 +986,7 @@ static int s390_vxrs_low_get(struct task_struct *target,
 	__u64 vxrs[__NUM_VXRS_LOW];
 	int i;
 
-	if (!MACHINE_HAS_VX)
+	if (!cpu_has_vx())
 		return -ENODEV;
 	if (target == current)
 		save_fpu_regs();
@@ -1005,7 +1003,7 @@ static int s390_vxrs_low_set(struct task_struct *target,
 	__u64 vxrs[__NUM_VXRS_LOW];
 	int i, rc;
 
-	if (!MACHINE_HAS_VX)
+	if (!cpu_has_vx())
 		return -ENODEV;
 	if (target == current)
 		save_fpu_regs();
@@ -1025,7 +1023,7 @@ static int s390_vxrs_high_get(struct task_struct *target,
 			      const struct user_regset *regset,
 			      struct membuf to)
 {
-	if (!MACHINE_HAS_VX)
+	if (!cpu_has_vx())
 		return -ENODEV;
 	if (target == current)
 		save_fpu_regs();
@@ -1040,7 +1038,7 @@ static int s390_vxrs_high_set(struct task_struct *target,
 {
 	int rc;
 
-	if (!MACHINE_HAS_VX)
+	if (!cpu_has_vx())
 		return -ENODEV;
 	if (target == current)
 		save_fpu_regs();
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
index 5701356f4f33..d1f3b56e7afc 100644
--- a/arch/s390/kernel/setup.c
+++ b/arch/s390/kernel/setup.c
@@ -408,15 +408,15 @@ static void __init setup_lowcore(void)
 
 	lc->restart_psw.mask = PSW_KERNEL_BITS & ~PSW_MASK_DAT;
 	lc->restart_psw.addr = __pa(restart_int_handler);
-	lc->external_new_psw.mask = PSW_KERNEL_BITS | PSW_MASK_MCHECK;
+	lc->external_new_psw.mask = PSW_KERNEL_BITS;
 	lc->external_new_psw.addr = (unsigned long) ext_int_handler;
-	lc->svc_new_psw.mask = PSW_KERNEL_BITS | PSW_MASK_MCHECK;
+	lc->svc_new_psw.mask = PSW_KERNEL_BITS;
 	lc->svc_new_psw.addr = (unsigned long) system_call;
-	lc->program_new_psw.mask = PSW_KERNEL_BITS | PSW_MASK_MCHECK;
+	lc->program_new_psw.mask = PSW_KERNEL_BITS;
 	lc->program_new_psw.addr = (unsigned long) pgm_check_handler;
 	lc->mcck_new_psw.mask = PSW_KERNEL_BITS;
 	lc->mcck_new_psw.addr = (unsigned long) mcck_int_handler;
-	lc->io_new_psw.mask = PSW_KERNEL_BITS | PSW_MASK_MCHECK;
+	lc->io_new_psw.mask = PSW_KERNEL_BITS;
 	lc->io_new_psw.addr = (unsigned long) io_int_handler;
 	lc->clock_comparator = clock_comparator_max;
 	lc->current_task = (unsigned long)&init_task;
@@ -820,22 +820,6 @@ static void __init setup_randomness(void)
 }
 
 /*
- * Find the correct size for the task_struct. This depends on
- * the size of the struct fpu at the end of the thread_struct
- * which is embedded in the task_struct.
- */
-static void __init setup_task_size(void)
-{
-	int task_size = sizeof(struct task_struct);
-
-	if (!MACHINE_HAS_VX) {
-		task_size -= sizeof(__vector128) * __NUM_VXRS;
-		task_size += sizeof(freg_t) * __NUM_FPRS;
-	}
-	arch_task_struct_size = task_size;
-}
-
-/*
  * Issue diagnose 318 to set the control program name and
  * version codes.
  */
@@ -927,7 +911,6 @@ void __init setup_arch(char **cmdline_p)
 
 	os_info_init();
 	setup_ipl();
-	setup_task_size();
 	setup_control_program_code();
 
 	/* Do some memory reservations *before* memory is added to memblock */
diff --git a/arch/s390/kernel/signal.c b/arch/s390/kernel/signal.c
index 48e46809ee27..43e9661cd715 100644
--- a/arch/s390/kernel/signal.c
+++ b/arch/s390/kernel/signal.c
@@ -150,10 +150,6 @@ static int restore_sigregs(struct pt_regs *regs, _sigregs __user *sregs)
 	if (!is_ri_task(current) && (user_sregs.regs.psw.mask & PSW_MASK_RI))
 		return -EINVAL;
 
-	/* Test the floating-point-control word. */
-	if (test_fp_ctl(user_sregs.fpregs.fpc))
-		return -EINVAL;
-
 	/* Use regs->psw.mask instead of PSW_USER_BITS to preserve PER bit. */
 	regs->psw.mask = (regs->psw.mask & ~(PSW_MASK_USER | PSW_MASK_RI)) |
 		(user_sregs.regs.psw.mask & (PSW_MASK_USER | PSW_MASK_RI));
@@ -183,7 +179,7 @@ static int save_sigregs_ext(struct pt_regs *regs,
 	int i;
 
 	/* Save vector registers to signal stack */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		for (i = 0; i < __NUM_VXRS_LOW; i++)
 			vxrs[i] = current->thread.fpu.vxrs[i].low;
 		if (__copy_to_user(&sregs_ext->vxrs_low, vxrs,
@@ -203,7 +199,7 @@ static int restore_sigregs_ext(struct pt_regs *regs,
 	int i;
 
 	/* Restore vector registers from signal stack */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		if (__copy_from_user(vxrs, &sregs_ext->vxrs_low,
 				     sizeof(sregs_ext->vxrs_low)) ||
 		    __copy_from_user(current->thread.fpu.vxrs + __NUM_VXRS_LOW,
@@ -301,7 +297,7 @@ static int setup_frame(int sig, struct k_sigaction *ka,
 	 * included in the signal frame on a 31-bit system.
 	 */
 	frame_size = sizeof(*frame) - sizeof(frame->sregs_ext);
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		frame_size += sizeof(frame->sregs_ext);
 	frame = get_sigframe(ka, regs, frame_size);
 	if (frame == (void __user *) -1UL)
@@ -378,7 +374,7 @@ static int setup_rt_frame(struct ksignal *ksig, sigset_t *set,
 	 * included in the signal frame on a 31-bit system.
 	 */
 	uc_flags = 0;
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		frame_size += sizeof(_sigregs_ext);
 		uc_flags |= UC_VXRS;
 	}
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index f7fcfff09acf..c39d9f0d4b1c 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -582,7 +582,7 @@ int smp_store_status(int cpu)
 	if (__pcpu_sigp_relax(pcpu->address, SIGP_STORE_STATUS_AT_ADDRESS,
 			      pa) != SIGP_CC_ORDER_CODE_ACCEPTED)
 		return -EIO;
-	if (!MACHINE_HAS_VX && !MACHINE_HAS_GS)
+	if (!cpu_has_vx() && !MACHINE_HAS_GS)
 		return 0;
 	pa = lc->mcesad & MCESA_ORIGIN_MASK;
 	if (MACHINE_HAS_GS)
@@ -638,7 +638,7 @@ void __init smp_save_dump_ipl_cpu(void)
 	copy_oldmem_kernel(regs, __LC_FPREGS_SAVE_AREA, 512);
 	save_area_add_regs(sa, regs);
 	memblock_free(regs, 512);
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		save_area_add_vxrs(sa, boot_cpu_vector_save_area);
 }
 
@@ -671,7 +671,7 @@ void __init smp_save_dump_secondary_cpus(void)
 			panic("could not allocate memory for save area\n");
 		__pcpu_sigp_relax(addr, SIGP_STORE_STATUS_AT_ADDRESS, __pa(page));
 		save_area_add_regs(sa, page);
-		if (MACHINE_HAS_VX) {
+		if (cpu_has_vx()) {
 			__pcpu_sigp_relax(addr, SIGP_STORE_ADDITIONAL_STATUS, __pa(page));
 			save_area_add_vxrs(sa, page);
 		}
diff --git a/arch/s390/kernel/sysinfo.c b/arch/s390/kernel/sysinfo.c
index b5e364358ce4..f6f8f498c9be 100644
--- a/arch/s390/kernel/sysinfo.c
+++ b/arch/s390/kernel/sysinfo.c
@@ -81,10 +81,12 @@ static bool convert_ext_name(unsigned char encoding, char *name, size_t len)
 
 static void stsi_1_1_1(struct seq_file *m, struct sysinfo_1_1_1 *info)
 {
+	bool has_var_cap;
 	int i;
 
 	if (stsi(info, 1, 1, 1))
 		return;
+	has_var_cap = !!info->model_var_cap[0];
 	EBCASC(info->manufacturer, sizeof(info->manufacturer));
 	EBCASC(info->type, sizeof(info->type));
 	EBCASC(info->model, sizeof(info->model));
@@ -93,6 +95,8 @@ static void stsi_1_1_1(struct seq_file *m, struct sysinfo_1_1_1 *info)
 	EBCASC(info->model_capacity, sizeof(info->model_capacity));
 	EBCASC(info->model_perm_cap, sizeof(info->model_perm_cap));
 	EBCASC(info->model_temp_cap, sizeof(info->model_temp_cap));
+	if (has_var_cap)
+		EBCASC(info->model_var_cap, sizeof(info->model_var_cap));
 	seq_printf(m, "Manufacturer:         %-16.16s\n", info->manufacturer);
 	seq_printf(m, "Type:                 %-4.4s\n", info->type);
 	if (info->lic)
@@ -120,12 +124,18 @@ static void stsi_1_1_1(struct seq_file *m, struct sysinfo_1_1_1 *info)
 		seq_printf(m, "Model Temp. Capacity: %-16.16s %08u\n",
 			   info->model_temp_cap,
 			   info->model_temp_cap_rating);
+	if (has_var_cap && info->model_var_cap_rating)
+		seq_printf(m, "Model Var. Capacity:  %-16.16s %08u\n",
+			   info->model_var_cap,
+			   info->model_var_cap_rating);
 	if (info->ncr)
 		seq_printf(m, "Nominal Cap. Rating:  %08u\n", info->ncr);
 	if (info->npr)
 		seq_printf(m, "Nominal Perm. Rating: %08u\n", info->npr);
 	if (info->ntr)
 		seq_printf(m, "Nominal Temp. Rating: %08u\n", info->ntr);
+	if (has_var_cap && info->nvr)
+		seq_printf(m, "Nominal Var. Rating:  %08u\n", info->nvr);
 	if (info->cai) {
 		seq_printf(m, "Capacity Adj. Ind.:   %d\n", info->cai);
 		seq_printf(m, "Capacity Ch. Reason:  %d\n", info->ccr);
diff --git a/arch/s390/kernel/traps.c b/arch/s390/kernel/traps.c
index cc3e3a01dfa5..46dac4540ca8 100644
--- a/arch/s390/kernel/traps.c
+++ b/arch/s390/kernel/traps.c
@@ -195,7 +195,7 @@ static void vector_exception(struct pt_regs *regs)
 {
 	int si_code, vic;
 
-	if (!MACHINE_HAS_VX) {
+	if (!cpu_has_vx()) {
 		do_trap(regs, SIGILL, ILL_ILLOPN, "illegal operation");
 		return;
 	}
@@ -288,6 +288,17 @@ static void __init test_monitor_call(void)
 
 void __init trap_init(void)
 {
+	unsigned long flags;
+	struct ctlreg cr0;
+
+	local_irq_save(flags);
+	cr0 = local_ctl_clear_bit(0, CR0_LOW_ADDRESS_PROTECTION_BIT);
+	psw_bits(S390_lowcore.external_new_psw).mcheck = 1;
+	psw_bits(S390_lowcore.program_new_psw).mcheck = 1;
+	psw_bits(S390_lowcore.svc_new_psw).mcheck = 1;
+	psw_bits(S390_lowcore.io_new_psw).mcheck = 1;
+	local_ctl_load(0, &cr0);
+	local_irq_restore(flags);
 	local_mcck_enable();
 	test_monitor_call();
 }
diff --git a/arch/s390/kernel/vmlinux.lds.S b/arch/s390/kernel/vmlinux.lds.S
index 2ae201ebf90b..e32ef446f451 100644
--- a/arch/s390/kernel/vmlinux.lds.S
+++ b/arch/s390/kernel/vmlinux.lds.S
@@ -52,6 +52,7 @@ SECTIONS
 		SOFTIRQENTRY_TEXT
 		FTRACE_HOTPATCH_TRAMPOLINES_TEXT
 		*(.text.*_indirect_*)
+		*(.fixup)
 		*(.gnu.warning)
 		. = ALIGN(PAGE_SIZE);
 		_etext = .;		/* End of text section */
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index efaebba5ee19..fc4007cc067a 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -639,7 +639,7 @@ static int __write_machine_check(struct kvm_vcpu *vcpu,
 	rc |= put_guest_lc(vcpu, mci.val, (u64 __user *) __LC_MCCK_CODE);
 
 	/* Register-save areas */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		convert_vx_to_fp(fprs, (__vector128 *) vcpu->run->s.regs.vrs);
 		rc |= write_guest_lc(vcpu, __LC_FPREGS_SAVE_AREA, fprs, 128);
 	} else {
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 7aa0e668488f..acc81ca6492e 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -618,7 +618,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 		r = MACHINE_HAS_ESOP;
 		break;
 	case KVM_CAP_S390_VECTOR_REGISTERS:
-		r = MACHINE_HAS_VX;
+		r = test_facility(129);
 		break;
 	case KVM_CAP_S390_RI:
 		r = test_facility(64);
@@ -767,7 +767,7 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
 		mutex_lock(&kvm->lock);
 		if (kvm->created_vcpus) {
 			r = -EBUSY;
-		} else if (MACHINE_HAS_VX) {
+		} else if (cpu_has_vx()) {
 			set_kvm_facility(kvm->arch.model.fac_mask, 129);
 			set_kvm_facility(kvm->arch.model.fac_list, 129);
 			if (test_facility(134)) {
@@ -3962,9 +3962,9 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 	if (test_kvm_facility(vcpu->kvm, 156))
 		vcpu->run->kvm_valid_regs |= KVM_SYNC_ETOKEN;
 	/* fprs can be synchronized via vrs, even if the guest has no vx. With
-	 * MACHINE_HAS_VX, (load|store)_fpu_regs() will work with vrs format.
+	 * cpu_has_vx(), (load|store)_fpu_regs() will work with vrs format.
 	 */
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		vcpu->run->kvm_valid_regs |= KVM_SYNC_VRS;
 	else
 		vcpu->run->kvm_valid_regs |= KVM_SYNC_FPRS;
@@ -4316,18 +4316,13 @@ int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu)
 
 	vcpu_load(vcpu);
 
-	if (test_fp_ctl(fpu->fpc)) {
-		ret = -EINVAL;
-		goto out;
-	}
 	vcpu->run->s.regs.fpc = fpu->fpc;
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_fp_to_vx((__vector128 *) vcpu->run->s.regs.vrs,
 				 (freg_t *) fpu->fprs);
 	else
 		memcpy(vcpu->run->s.regs.fprs, &fpu->fprs, sizeof(fpu->fprs));
 
-out:
 	vcpu_put(vcpu);
 	return ret;
 }
@@ -4336,9 +4331,7 @@ int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu)
 {
 	vcpu_load(vcpu);
 
-	/* make sure we have the latest values */
-	save_fpu_regs();
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		convert_vx_to_fp((freg_t *) fpu->fprs,
 				 (__vector128 *) vcpu->run->s.regs.vrs);
 	else
@@ -4963,14 +4956,11 @@ static void sync_regs(struct kvm_vcpu *vcpu)
 	save_fpu_regs();
 	vcpu->arch.host_fpregs.fpc = current->thread.fpu.fpc;
 	vcpu->arch.host_fpregs.regs = current->thread.fpu.regs;
-	if (MACHINE_HAS_VX)
+	if (cpu_has_vx())
 		current->thread.fpu.regs = vcpu->run->s.regs.vrs;
 	else
 		current->thread.fpu.regs = vcpu->run->s.regs.fprs;
 	current->thread.fpu.fpc = vcpu->run->s.regs.fpc;
-	if (test_fp_ctl(current->thread.fpu.fpc))
-		/* User space provided an invalid FPC, let's clear it */
-		current->thread.fpu.fpc = 0;
 
 	/* Sync fmt2 only data */
 	if (likely(!kvm_s390_pv_cpu_is_protected(vcpu))) {
@@ -5145,7 +5135,7 @@ int kvm_s390_store_status_unloaded(struct kvm_vcpu *vcpu, unsigned long gpa)
 		gpa -= __LC_FPREGS_SAVE_AREA;
 
 	/* manually convert vector registers if necessary */
-	if (MACHINE_HAS_VX) {
+	if (cpu_has_vx()) {
 		convert_vx_to_fp(fprs, (__vector128 *) vcpu->run->s.regs.vrs);
 		rc = write_guest_abs(vcpu, gpa + __LC_FPREGS_SAVE_AREA,
 				     fprs, 128);
diff --git a/arch/s390/lib/test_unwind.c b/arch/s390/lib/test_unwind.c
index 7231bf97b93a..2848e3fb2ff5 100644
--- a/arch/s390/lib/test_unwind.c
+++ b/arch/s390/lib/test_unwind.c
@@ -350,15 +350,15 @@ static noinline int unwindme_func3(struct unwindme *u)
 /* This function must appear in the backtrace. */
 static noinline int unwindme_func2(struct unwindme *u)
 {
-	unsigned long flags;
+	unsigned long flags, mflags;
 	int rc;
 
 	if (u->flags & UWM_SWITCH_STACK) {
 		local_irq_save(flags);
-		local_mcck_disable();
+		local_mcck_save(mflags);
 		rc = call_on_stack(1, S390_lowcore.nodat_stack,
 				   int, unwindme_func3, struct unwindme *, u);
-		local_mcck_enable();
+		local_mcck_restore(mflags);
 		local_irq_restore(flags);
 		return rc;
 	} else {
diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c
index 5cb92941540b..99422926efe1 100644
--- a/arch/s390/mm/pgtable.c
+++ b/arch/s390/mm/pgtable.c
@@ -125,32 +125,23 @@ static inline pte_t ptep_flush_lazy(struct mm_struct *mm,
 
 static inline pgste_t pgste_get_lock(pte_t *ptep)
 {
-	unsigned long new = 0;
+	unsigned long value = 0;
 #ifdef CONFIG_PGSTE
-	unsigned long old;
-
-	asm(
-		"	lg	%0,%2\n"
-		"0:	lgr	%1,%0\n"
-		"	nihh	%0,0xff7f\n"	/* clear PCL bit in old */
-		"	oihh	%1,0x0080\n"	/* set PCL bit in new */
-		"	csg	%0,%1,%2\n"
-		"	jl	0b\n"
-		: "=&d" (old), "=&d" (new), "=Q" (ptep[PTRS_PER_PTE])
-		: "Q" (ptep[PTRS_PER_PTE]) : "cc", "memory");
+	unsigned long *ptr = (unsigned long *)(ptep + PTRS_PER_PTE);
+
+	do {
+		value = __atomic64_or_barrier(PGSTE_PCL_BIT, ptr);
+	} while (value & PGSTE_PCL_BIT);
+	value |= PGSTE_PCL_BIT;
 #endif
-	return __pgste(new);
+	return __pgste(value);
 }
 
 static inline void pgste_set_unlock(pte_t *ptep, pgste_t pgste)
 {
 #ifdef CONFIG_PGSTE
-	asm(
-		"	nihh	%1,0xff7f\n"	/* clear PCL bit */
-		"	stg	%1,%0\n"
-		: "=Q" (ptep[PTRS_PER_PTE])
-		: "d" (pgste_val(pgste)), "Q" (ptep[PTRS_PER_PTE])
-		: "cc", "memory");
+	barrier();
+	WRITE_ONCE(*(unsigned long *)(ptep + PTRS_PER_PTE), pgste_val(pgste) & ~PGSTE_PCL_BIT);
 #endif
 }
 
diff --git a/arch/s390/tools/gen_facilities.c b/arch/s390/tools/gen_facilities.c
index cb0aff5c0187..68580cbea4e6 100644
--- a/arch/s390/tools/gen_facilities.c
+++ b/arch/s390/tools/gen_facilities.c
@@ -46,6 +46,7 @@ static struct facility_def facility_defs[] = {
 #endif
 #ifdef CONFIG_HAVE_MARCH_Z13_FEATURES
 			53, /* load-and-zero-rightmost-byte, etc. */
+			129, /* vector */
 #endif
 #ifdef CONFIG_HAVE_MARCH_Z14_FEATURES
 			58, /* miscellaneous-instruction-extension 2 */