Merge tag 's390-6.14-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull more s390 updates from Alexander Gordeev:

 - The rework that uncoupled physical and virtual address spaces
   inadvertently prevented KASAN shadow mappings from using large pages.
   Restore large page mappings for KASAN shadows

 - Add decompressor routine physmem_alloc() that may fail, unlike
   physmem_alloc_or_die(). This allows callers to implement fallback
   paths

 - Allow falling back from large pages to smaller pages (1MB or 4KB) if
   the allocation of 2GB pages in the decompressor can not be fulfilled

 - Add to the decompressor boot print support of "%%" format string,
   width and padding hadnling, length modifiers and decimal conversion
   specifiers

 - Add to the decompressor message severity levels similar to kernel
   ones. Support command-line options that control console output
   verbosity

 - Replaces boot_printk() calls with appropriate loglevel- specific
   helpers such as boot_emerg(), boot_warn(), and boot_debug().

 - Collect all boot messages into a ring buffer independent of the
   current log level. This is particularly useful for early crash
   analysis

 - If 'earlyprintk' command line parameter is not specified, store
   decompressor boot messages in a ring buffer to be printed later by
   the kernel, once the console driver is registered

 - Add 'bootdebug' command line parameter to enable printing of
   decompressor debug messages when needed. That parameters allows
   message suppressing and filtering

 - Dump boot messages on a decompressor crash, but only if 'bootdebug'
   command line parameter is enabled

 - When CONFIG_PRINTK_TIME is enabled, add timestamps to boot messages
   in the same format as regular printk()

 - Dump physical memory tracking information on boot: online ranges,
   reserved areas and vmem allocations

 - Dump virtual memory layout and randomization details

 - Improve decompression error reporting and dump the message ring
   buffer in case the boot failed and system halted

 - Add an exception handler which handles exceptions when FPU control
   register is attempted to be set to an invalid value. Remove '.fixup'
   section as result of this change

 - Use 'A', 'O', and 'R' inline assembly format flags, which allows
   recent Clang compilers to generate better FPU code

 - Rework uaccess code so it reads better and generates more efficient
   code

 - Cleanup futex inline assembly code

 - Disable KMSAN instrumention for futex inline assemblies, which
   contain dereferenced user pointers. Otherwise, shadows for the user
   pointers would be accessed

 - PFs which are not initially configured but in standby create only a
   single-function PCI domain. If they are configured later on, sibling
   PFs and their child VFs will not be added to their PCI domain
   breaking SR-IOV expectations.

   Fix that by allowing initially configured but in standby PFs create
   multi-function PCI domains

 - Add '-std=gnu11' to decompressor and purgatory CFLAGS to avoid
   compile errors caused by kernel's own definitions of 'bool', 'false',
   and 'true' conflicting with the C23 reserved keywords

 - Fix sclp subsystem failure when a sclp console is not present

 - Fix misuse of non-NULL terminated strings in vmlogrdr driver

 - Various other small improvements, cleanups and fixes

* tag 's390-6.14-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (53 commits)
  s390/vmlogrdr: Use array instead of string initializer
  s390/vmlogrdr: Use internal_name for error messages
  s390/sclp: Initialize sclp subsystem via arch_cpu_finalize_init()
  s390/tools: Use array instead of string initializer
  s390/vmem: Fix null-pointer-arithmetic warning in vmem_map_init()
  s390: Add '-std=gnu11' to decompressor and purgatory CFLAGS
  s390/bitops: Use correct constraint for arch_test_bit() inline assembly
  s390/pci: Fix SR-IOV for PFs initially in standby
  s390/futex: Avoid KMSAN instrumention for user pointers
  s390/uaccess: Rename get_put_user_noinstr_attributes to uaccess_kmsan_or_inline
  s390/futex: Cleanup futex_atomic_cmpxchg_inatomic()
  s390/futex: Generate futex atomic op functions
  s390/uaccess: Remove INLINE_COPY_FROM_USER and INLINE_COPY_TO_USER
  s390/uaccess: Use asm goto for put_user()/get_user()
  s390/uaccess: Remove usage of the oac specifier
  s390/uaccess: Replace EX_TABLE_UA_LOAD_MEM exception handling
  s390/uaccess: Cleanup noinstr __put_user()/__get_user() inline assembly constraints
  s390/uaccess: Remove __put_user_fn()/__get_user_fn() wrappers
  s390/uaccess: Move put_user() / __put_user() close to put_user() asm code
  s390/uaccess: Use asm goto for __mvc_kernel_nofault()
  ...

10 files changed, 518 insertions, 398 deletions

diff --git a/arch/s390/include/asm/asm-extable.h b/arch/s390/include/asm/asm-extable.h
index 4a6b0a8b6412..2e829c16fd8a 100644
--- a/arch/s390/include/asm/asm-extable.h
+++ b/arch/s390/include/asm/asm-extable.h
@@ -9,11 +9,11 @@
 #define EX_TYPE_NONE		0
 #define EX_TYPE_FIXUP		1
 #define EX_TYPE_BPF		2
-#define EX_TYPE_UA_STORE	3
-#define EX_TYPE_UA_LOAD_MEM	4
+#define EX_TYPE_UA_FAULT	3
 #define EX_TYPE_UA_LOAD_REG	5
 #define EX_TYPE_UA_LOAD_REGPAIR	6
 #define EX_TYPE_ZEROPAD		7
+#define EX_TYPE_FPC		8
 
 #define EX_DATA_REG_ERR_SHIFT	0
 #define EX_DATA_REG_ERR		GENMASK(3, 0)
@@ -69,11 +69,8 @@
 #define EX_TABLE_AMODE31(_fault, _target)				\
 	__EX_TABLE(.amode31.ex_table, _fault, _target, EX_TYPE_FIXUP, __stringify(%%r0), __stringify(%%r0), 0)
 
-#define EX_TABLE_UA_STORE(_fault, _target, _regerr)			\
-	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_UA_STORE, _regerr, _regerr, 0)
-
-#define EX_TABLE_UA_LOAD_MEM(_fault, _target, _regerr, _regmem, _len)	\
-	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_UA_LOAD_MEM, _regerr, _regmem, _len)
+#define EX_TABLE_UA_FAULT(_fault, _target, _regerr)			\
+	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_UA_FAULT, _regerr, _regerr, 0)
 
 #define EX_TABLE_UA_LOAD_REG(_fault, _target, _regerr, _regzero)	\
 	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_UA_LOAD_REG, _regerr, _regzero, 0)
@@ -84,4 +81,7 @@
 #define EX_TABLE_ZEROPAD(_fault, _target, _regdata, _regaddr)		\
 	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_ZEROPAD, _regdata, _regaddr, 0)
 
+#define EX_TABLE_FPC(_fault, _target)					\
+	__EX_TABLE(__ex_table, _fault, _target, EX_TYPE_FPC, __stringify(%%r0), __stringify(%%r0), 0)
+
 #endif /* __ASM_EXTABLE_H */
diff --git a/arch/s390/include/asm/asm.h b/arch/s390/include/asm/asm.h
index ec011b94af2a..e9062b01e2a2 100644
--- a/arch/s390/include/asm/asm.h
+++ b/arch/s390/include/asm/asm.h
@@ -28,7 +28,7 @@
  * [var] also contains the program mask. CC_TRANSFORM() moves the condition
  * code to the two least significant bits and sets all other bits to zero.
  */
-#if defined(__GCC_ASM_FLAG_OUTPUTS__) && !(IS_ENABLED(CONFIG_GCC_ASM_FLAG_OUTPUT_BROKEN))
+#if defined(__GCC_ASM_FLAG_OUTPUTS__) && !(IS_ENABLED(CONFIG_CC_ASM_FLAG_OUTPUT_BROKEN))
 
 #define __HAVE_ASM_FLAG_OUTPUTS__
 
diff --git a/arch/s390/include/asm/bitops.h b/arch/s390/include/asm/bitops.h
index 15aa64e3020e..d5125296ade2 100644
--- a/arch/s390/include/asm/bitops.h
+++ b/arch/s390/include/asm/bitops.h
@@ -60,7 +60,7 @@ static __always_inline bool arch_test_bit(unsigned long nr, const volatile unsig
 		asm volatile(
 			"	tm	%[addr],%[mask]\n"
 			: "=@cc" (cc)
-			: [addr] "R" (*addr), [mask] "I" (mask)
+			: [addr] "Q" (*addr), [mask] "I" (mask)
 			);
 		return cc == 3;
 	}
diff --git a/arch/s390/include/asm/boot_data.h b/arch/s390/include/asm/boot_data.h
index f7eed27b3220..f55f8227058e 100644
--- a/arch/s390/include/asm/boot_data.h
+++ b/arch/s390/include/asm/boot_data.h
@@ -1,6 +1,7 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 #ifndef _ASM_S390_BOOT_DATA_H
 
+#include <linux/string.h>
 #include <asm/setup.h>
 #include <asm/ipl.h>
 
@@ -15,4 +16,54 @@ extern unsigned long ipl_cert_list_size;
 extern unsigned long early_ipl_comp_list_addr;
 extern unsigned long early_ipl_comp_list_size;
 
+extern char boot_rb[PAGE_SIZE * 2];
+extern bool boot_earlyprintk;
+extern size_t boot_rb_off;
+extern char bootdebug_filter[128];
+extern bool bootdebug;
+
+#define boot_rb_foreach(cb)							\
+	do {									\
+		size_t off = boot_rb_off + strlen(boot_rb + boot_rb_off) + 1;	\
+		size_t len;							\
+		for (; off < sizeof(boot_rb) && (len = strlen(boot_rb + off)); off += len + 1) \
+			cb(boot_rb + off);					\
+		for (off = 0; off < boot_rb_off && (len = strlen(boot_rb + off)); off += len + 1) \
+			cb(boot_rb + off);					\
+	} while (0)
+
+/*
+ * bootdebug_filter is a comma separated list of strings,
+ * where each string can be a prefix of the message.
+ */
+static inline bool bootdebug_filter_match(const char *buf)
+{
+	char *p = bootdebug_filter, *s;
+	char *end;
+
+	if (!*p)
+		return true;
+
+	end = p + strlen(p);
+	while (p < end) {
+		p = skip_spaces(p);
+		s = memscan(p, ',', end - p);
+		if (!strncmp(p, buf, s - p))
+			return true;
+		p = s + 1;
+	}
+	return false;
+}
+
+static inline const char *skip_timestamp(const char *buf)
+{
+#ifdef CONFIG_PRINTK_TIME
+	const char *p = memchr(buf, ']', strlen(buf));
+
+	if (p && p[1] == ' ')
+		return p + 2;
+#endif
+	return buf;
+}
+
 #endif /* _ASM_S390_BOOT_DATA_H */
diff --git a/arch/s390/include/asm/fpu-insn.h b/arch/s390/include/asm/fpu-insn.h
index de510c9f6efa..f668bffd6dd3 100644
--- a/arch/s390/include/asm/fpu-insn.h
+++ b/arch/s390/include/asm/fpu-insn.h
@@ -100,19 +100,12 @@ static __always_inline void fpu_lfpc(unsigned int *fpc)
  */
 static inline void fpu_lfpc_safe(unsigned int *fpc)
 {
-	u32 tmp;
-
 	instrument_read(fpc, sizeof(*fpc));
 	asm_inline volatile(
-		"0:	lfpc	%[fpc]\n"
-		"1:	nopr	%%r7\n"
-		".pushsection .fixup, \"ax\"\n"
-		"2:	lghi	%[tmp],0\n"
-		"	sfpc	%[tmp]\n"
-		"	jg	1b\n"
-		".popsection\n"
-		EX_TABLE(1b, 2b)
-		: [tmp] "=d" (tmp)
+		"	lfpc	%[fpc]\n"
+		"0:	nopr	%%r7\n"
+		EX_TABLE_FPC(0b, 0b)
+		:
 		: [fpc] "Q" (*fpc)
 		: "memory");
 }
@@ -183,7 +176,19 @@ static __always_inline void fpu_vgfmg(u8 v1, u8 v2, u8 v3)
 		     : "memory");
 }
 
-#ifdef CONFIG_CC_IS_CLANG
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
+
+static __always_inline void fpu_vl(u8 v1, const void *vxr)
+{
+	instrument_read(vxr, sizeof(__vector128));
+	asm volatile("VL	%[v1],%O[vxr],,%R[vxr]\n"
+		     :
+		     : [vxr] "Q" (*(__vector128 *)vxr),
+		       [v1] "I" (v1)
+		     : "memory");
+}
+
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vl(u8 v1, const void *vxr)
 {
@@ -197,19 +202,7 @@ static __always_inline void fpu_vl(u8 v1, const void *vxr)
 		: "memory", "1");
 }
 
-#else /* CONFIG_CC_IS_CLANG */
-
-static __always_inline void fpu_vl(u8 v1, const void *vxr)
-{
-	instrument_read(vxr, sizeof(__vector128));
-	asm volatile("VL	%[v1],%O[vxr],,%R[vxr]\n"
-		     :
-		     : [vxr] "Q" (*(__vector128 *)vxr),
-		       [v1] "I" (v1)
-		     : "memory");
-}
-
-#endif /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vleib(u8 v, s16 val, u8 index)
 {
@@ -238,7 +231,7 @@ static __always_inline u64 fpu_vlgvf(u8 v, u16 index)
 	return val;
 }
 
-#ifdef CONFIG_CC_IS_CLANG
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
 
 static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 {
@@ -246,17 +239,15 @@ static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 
 	size = min(index + 1, sizeof(__vector128));
 	instrument_read(vxr, size);
-	asm volatile(
-		"	la	1,%[vxr]\n"
-		"	VLL	%[v1],%[index],0,1\n"
-		:
-		: [vxr] "R" (*(u8 *)vxr),
-		  [index] "d" (index),
-		  [v1] "I" (v1)
-		: "memory", "1");
+	asm volatile("VLL	%[v1],%[index],%O[vxr],%R[vxr]\n"
+		     :
+		     : [vxr] "Q" (*(u8 *)vxr),
+		       [index] "d" (index),
+		       [v1] "I" (v1)
+		     : "memory");
 }
 
-#else /* CONFIG_CC_IS_CLANG */
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 {
@@ -264,17 +255,19 @@ static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 
 	size = min(index + 1, sizeof(__vector128));
 	instrument_read(vxr, size);
-	asm volatile("VLL	%[v1],%[index],%O[vxr],%R[vxr]\n"
-		     :
-		     : [vxr] "Q" (*(u8 *)vxr),
-		       [index] "d" (index),
-		       [v1] "I" (v1)
-		     : "memory");
+	asm volatile(
+		"	la	1,%[vxr]\n"
+		"	VLL	%[v1],%[index],0,1\n"
+		:
+		: [vxr] "R" (*(u8 *)vxr),
+		  [index] "d" (index),
+		  [v1] "I" (v1)
+		: "memory", "1");
 }
 
-#endif /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
-#ifdef CONFIG_CC_IS_CLANG
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
 
 #define fpu_vlm(_v1, _v3, _vxrs)					\
 ({									\
@@ -284,17 +277,15 @@ static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 	} *_v = (void *)(_vxrs);					\
 									\
 	instrument_read(_v, size);					\
-	asm volatile(							\
-		"	la	1,%[vxrs]\n"				\
-		"	VLM	%[v1],%[v3],0,1\n"			\
-		:							\
-		: [vxrs] "R" (*_v),					\
-		  [v1] "I" (_v1), [v3] "I" (_v3)			\
-		: "memory", "1");					\
+	asm volatile("VLM	%[v1],%[v3],%O[vxrs],%R[vxrs]\n"	\
+		     :							\
+		     : [vxrs] "Q" (*_v),				\
+		       [v1] "I" (_v1), [v3] "I" (_v3)			\
+		     : "memory");					\
 	(_v3) - (_v1) + 1;						\
 })
 
-#else /* CONFIG_CC_IS_CLANG */
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 #define fpu_vlm(_v1, _v3, _vxrs)					\
 ({									\
@@ -304,15 +295,17 @@ static __always_inline void fpu_vll(u8 v1, u32 index, const void *vxr)
 	} *_v = (void *)(_vxrs);					\
 									\
 	instrument_read(_v, size);					\
-	asm volatile("VLM	%[v1],%[v3],%O[vxrs],%R[vxrs]\n"	\
-		     :							\
-		     : [vxrs] "Q" (*_v),				\
-		       [v1] "I" (_v1), [v3] "I" (_v3)			\
-		     : "memory");					\
+	asm volatile(							\
+		"	la	1,%[vxrs]\n"				\
+		"	VLM	%[v1],%[v3],0,1\n"			\
+		:							\
+		: [vxrs] "R" (*_v),					\
+		  [v1] "I" (_v1), [v3] "I" (_v3)			\
+		: "memory", "1");					\
 	(_v3) - (_v1) + 1;						\
 })
 
-#endif /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vlr(u8 v1, u8 v2)
 {
@@ -362,7 +355,18 @@ static __always_inline void fpu_vsrlb(u8 v1, u8 v2, u8 v3)
 		     : "memory");
 }
 
-#ifdef CONFIG_CC_IS_CLANG
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
+
+static __always_inline void fpu_vst(u8 v1, const void *vxr)
+{
+	instrument_write(vxr, sizeof(__vector128));
+	asm volatile("VST	%[v1],%O[vxr],,%R[vxr]\n"
+		     : [vxr] "=Q" (*(__vector128 *)vxr)
+		     : [v1] "I" (v1)
+		     : "memory");
+}
+
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vst(u8 v1, const void *vxr)
 {
@@ -375,20 +379,23 @@ static __always_inline void fpu_vst(u8 v1, const void *vxr)
 		: "memory", "1");
 }
 
-#else /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
-static __always_inline void fpu_vst(u8 v1, const void *vxr)
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
+
+static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
 {
-	instrument_write(vxr, sizeof(__vector128));
-	asm volatile("VST	%[v1],%O[vxr],,%R[vxr]\n"
-		     : [vxr] "=Q" (*(__vector128 *)vxr)
-		     : [v1] "I" (v1)
+	unsigned int size;
+
+	size = min(index + 1, sizeof(__vector128));
+	instrument_write(vxr, size);
+	asm volatile("VSTL	%[v1],%[index],%O[vxr],%R[vxr]\n"
+		     : [vxr] "=Q" (*(u8 *)vxr)
+		     : [index] "d" (index), [v1] "I" (v1)
 		     : "memory");
 }
 
-#endif /* CONFIG_CC_IS_CLANG */
-
-#ifdef CONFIG_CC_IS_CLANG
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
 {
@@ -404,23 +411,9 @@ static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
 		: "memory", "1");
 }
 
-#else /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
-static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
-{
-	unsigned int size;
-
-	size = min(index + 1, sizeof(__vector128));
-	instrument_write(vxr, size);
-	asm volatile("VSTL	%[v1],%[index],%O[vxr],%R[vxr]\n"
-		     : [vxr] "=Q" (*(u8 *)vxr)
-		     : [index] "d" (index), [v1] "I" (v1)
-		     : "memory");
-}
-
-#endif /* CONFIG_CC_IS_CLANG */
-
-#ifdef CONFIG_CC_IS_CLANG
+#ifdef CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS
 
 #define fpu_vstm(_v1, _v3, _vxrs)					\
 ({									\
@@ -430,16 +423,14 @@ static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
 	} *_v = (void *)(_vxrs);					\
 									\
 	instrument_write(_v, size);					\
-	asm volatile(							\
-		"	la	1,%[vxrs]\n"				\
-		"	VSTM	%[v1],%[v3],0,1\n"			\
-		: [vxrs] "=R" (*_v)					\
-		: [v1] "I" (_v1), [v3] "I" (_v3)			\
-		: "memory", "1");					\
+	asm volatile("VSTM	%[v1],%[v3],%O[vxrs],%R[vxrs]\n"	\
+		     : [vxrs] "=Q" (*_v)				\
+		     : [v1] "I" (_v1), [v3] "I" (_v3)			\
+		     : "memory");					\
 	(_v3) - (_v1) + 1;						\
 })
 
-#else /* CONFIG_CC_IS_CLANG */
+#else /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 #define fpu_vstm(_v1, _v3, _vxrs)					\
 ({									\
@@ -449,14 +440,16 @@ static __always_inline void fpu_vstl(u8 v1, u32 index, const void *vxr)
 	} *_v = (void *)(_vxrs);					\
 									\
 	instrument_write(_v, size);					\
-	asm volatile("VSTM	%[v1],%[v3],%O[vxrs],%R[vxrs]\n"	\
-		     : [vxrs] "=Q" (*_v)				\
-		     : [v1] "I" (_v1), [v3] "I" (_v3)			\
-		     : "memory");					\
+	asm volatile(							\
+		"	la	1,%[vxrs]\n"				\
+		"	VSTM	%[v1],%[v3],0,1\n"			\
+		: [vxrs] "=R" (*_v)					\
+		: [v1] "I" (_v1), [v3] "I" (_v3)			\
+		: "memory", "1");					\
 	(_v3) - (_v1) + 1;						\
 })
 
-#endif /* CONFIG_CC_IS_CLANG */
+#endif /* CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
 static __always_inline void fpu_vupllf(u8 v1, u8 v2)
 {
diff --git a/arch/s390/include/asm/futex.h b/arch/s390/include/asm/futex.h
index 752a2310f0d6..f5781794356b 100644
--- a/arch/s390/include/asm/futex.h
+++ b/arch/s390/include/asm/futex.h
@@ -2,80 +2,95 @@
 #ifndef _ASM_S390_FUTEX_H
 #define _ASM_S390_FUTEX_H
 
+#include <linux/instrumented.h>
 #include <linux/uaccess.h>
 #include <linux/futex.h>
 #include <asm/asm-extable.h>
 #include <asm/mmu_context.h>
 #include <asm/errno.h>
 
-#define __futex_atomic_op(insn, ret, oldval, newval, uaddr, oparg)	\
-	asm volatile(							\
-		"   sacf  256\n"					\
-		"0: l     %1,0(%6)\n"					\
-		"1:"insn						\
-		"2: cs    %1,%2,0(%6)\n"				\
-		"3: jl    1b\n"						\
-		"   lhi   %0,0\n"					\
-		"4: sacf  768\n"					\
-		EX_TABLE(0b,4b) EX_TABLE(1b,4b)				\
-		EX_TABLE(2b,4b) EX_TABLE(3b,4b)				\
-		: "=d" (ret), "=&d" (oldval), "=&d" (newval),		\
-		  "=m" (*uaddr)						\
-		: "0" (-EFAULT), "d" (oparg), "a" (uaddr),		\
-		  "m" (*uaddr) : "cc");
+#define FUTEX_OP_FUNC(name, insn)						\
+static uaccess_kmsan_or_inline int						\
+__futex_atomic_##name(int oparg, int *old, u32 __user *uaddr)			\
+{										\
+	int rc, new;								\
+										\
+	instrument_copy_from_user_before(old, uaddr, sizeof(*old));		\
+	asm_inline volatile(							\
+		"	sacf	256\n"						\
+		"0:	l	%[old],%[uaddr]\n"				\
+		"1:"insn							\
+		"2:	cs	%[old],%[new],%[uaddr]\n"			\
+		"3:	jl	1b\n"						\
+		"	lhi	%[rc],0\n"					\
+		"4:	sacf	768\n"						\
+		EX_TABLE_UA_FAULT(0b, 4b, %[rc])				\
+		EX_TABLE_UA_FAULT(1b, 4b, %[rc])				\
+		EX_TABLE_UA_FAULT(2b, 4b, %[rc])				\
+		EX_TABLE_UA_FAULT(3b, 4b, %[rc])				\
+		: [rc] "=d" (rc), [old] "=&d" (*old),				\
+		  [new] "=&d" (new), [uaddr] "+Q" (*uaddr)			\
+		: [oparg] "d" (oparg)						\
+		: "cc");							\
+	if (!rc)								\
+		instrument_copy_from_user_after(old, uaddr, sizeof(*old), 0);	\
+	return rc;								\
+}
+
+FUTEX_OP_FUNC(set, "lr %[new],%[oparg]\n")
+FUTEX_OP_FUNC(add, "lr %[new],%[old]\n ar %[new],%[oparg]\n")
+FUTEX_OP_FUNC(or,  "lr %[new],%[old]\n or %[new],%[oparg]\n")
+FUTEX_OP_FUNC(and, "lr %[new],%[old]\n nr %[new],%[oparg]\n")
+FUTEX_OP_FUNC(xor, "lr %[new],%[old]\n xr %[new],%[oparg]\n")
 
-static inline int arch_futex_atomic_op_inuser(int op, int oparg, int *oval,
-		u32 __user *uaddr)
+static inline
+int arch_futex_atomic_op_inuser(int op, int oparg, int *oval, u32 __user *uaddr)
 {
-	int oldval = 0, newval, ret;
+	int old, rc;
 
 	switch (op) {
 	case FUTEX_OP_SET:
-		__futex_atomic_op("lr %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
+		rc = __futex_atomic_set(oparg, &old, uaddr);
 		break;
 	case FUTEX_OP_ADD:
-		__futex_atomic_op("lr %2,%1\nar %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
+		rc = __futex_atomic_add(oparg, &old, uaddr);
 		break;
 	case FUTEX_OP_OR:
-		__futex_atomic_op("lr %2,%1\nor %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
+		rc = __futex_atomic_or(oparg, &old, uaddr);
 		break;
 	case FUTEX_OP_ANDN:
-		__futex_atomic_op("lr %2,%1\nnr %2,%5\n",
-				  ret, oldval, newval, uaddr, ~oparg);
+		rc = __futex_atomic_and(~oparg, &old, uaddr);
 		break;
 	case FUTEX_OP_XOR:
-		__futex_atomic_op("lr %2,%1\nxr %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
+		rc = __futex_atomic_xor(oparg, &old, uaddr);
 		break;
 	default:
-		ret = -ENOSYS;
+		rc = -ENOSYS;
 	}
-
-	if (!ret)
-		*oval = oldval;
-
-	return ret;
+	if (!rc)
+		*oval = old;
+	return rc;
 }
 
-static inline int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
-						u32 oldval, u32 newval)
+static uaccess_kmsan_or_inline
+int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr, u32 oldval, u32 newval)
 {
-	int ret;
+	int rc;
 
-	asm volatile(
-		"   sacf 256\n"
-		"0: cs   %1,%4,0(%5)\n"
-		"1: la   %0,0\n"
-		"2: sacf 768\n"
-		EX_TABLE(0b,2b) EX_TABLE(1b,2b)
-		: "=d" (ret), "+d" (oldval), "=m" (*uaddr)
-		: "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr)
+	instrument_copy_from_user_before(uval, uaddr, sizeof(*uval));
+	asm_inline volatile(
+		"	sacf	256\n"
+		"0:	cs	%[old],%[new],%[uaddr]\n"
+		"1:	lhi	%[rc],0\n"
+		"2:	sacf	768\n"
+		EX_TABLE_UA_FAULT(0b, 2b, %[rc])
+		EX_TABLE_UA_FAULT(1b, 2b, %[rc])
+		: [rc] "=d" (rc), [old] "+d" (oldval), [uaddr] "+Q" (*uaddr)
+		: [new] "d" (newval)
 		: "cc", "memory");
 	*uval = oldval;
-	return ret;
+	instrument_copy_from_user_after(uval, uaddr, sizeof(*uval), 0);
+	return rc;
 }
 
 #endif /* _ASM_S390_FUTEX_H */
diff --git a/arch/s390/include/asm/page.h b/arch/s390/include/asm/page.h
index 4f43cdd9835b..1ff145f7b52b 100644
--- a/arch/s390/include/asm/page.h
+++ b/arch/s390/include/asm/page.h
@@ -184,7 +184,11 @@ extern struct vm_layout vm_layout;
 
 #define __kaslr_offset		vm_layout.kaslr_offset
 #define __kaslr_offset_phys	vm_layout.kaslr_offset_phys
+#ifdef CONFIG_RANDOMIZE_IDENTITY_BASE
 #define __identity_base		vm_layout.identity_base
+#else
+#define __identity_base		0UL
+#endif
 #define ident_map_size		vm_layout.identity_size
 
 static inline unsigned long kaslr_offset(void)
diff --git a/arch/s390/include/asm/physmem_info.h b/arch/s390/include/asm/physmem_info.h
index 51b68a43e195..7ef3bbec98b0 100644
--- a/arch/s390/include/asm/physmem_info.h
+++ b/arch/s390/include/asm/physmem_info.h
@@ -26,7 +26,7 @@ enum reserved_range_type {
 	RR_AMODE31,
 	RR_IPLREPORT,
 	RR_CERT_COMP_LIST,
-	RR_MEM_DETECT_EXTENDED,
+	RR_MEM_DETECT_EXT,
 	RR_VMEM,
 	RR_MAX
 };
@@ -128,7 +128,7 @@ static inline const char *get_rr_type_name(enum reserved_range_type t)
 	RR_TYPE_NAME(AMODE31);
 	RR_TYPE_NAME(IPLREPORT);
 	RR_TYPE_NAME(CERT_COMP_LIST);
-	RR_TYPE_NAME(MEM_DETECT_EXTENDED);
+	RR_TYPE_NAME(MEM_DETECT_EXT);
 	RR_TYPE_NAME(VMEM);
 	default:
 		return "UNKNOWN";
diff --git a/arch/s390/include/asm/sclp.h b/arch/s390/include/asm/sclp.h
index 4da3b2956285..18f37dff03c9 100644
--- a/arch/s390/include/asm/sclp.h
+++ b/arch/s390/include/asm/sclp.h
@@ -172,6 +172,7 @@ void sclp_early_printk(const char *s);
 void __sclp_early_printk(const char *s, unsigned int len);
 void sclp_emergency_printk(const char *s);
 
+int sclp_init(void);
 int sclp_early_get_memsize(unsigned long *mem);
 int sclp_early_get_hsa_size(unsigned long *hsa_size);
 int _sclp_get_core_info(struct sclp_core_info *info);
diff --git a/arch/s390/include/asm/uaccess.h b/arch/s390/include/asm/uaccess.h
index a81f897a81ce..f5920163ee97 100644
--- a/arch/s390/include/asm/uaccess.h
+++ b/arch/s390/include/asm/uaccess.h
@@ -22,16 +22,117 @@
 
 void debug_user_asce(int exit);
 
-unsigned long __must_check
-raw_copy_from_user(void *to, const void __user *from, unsigned long n);
+union oac {
+	unsigned int val;
+	struct {
+		struct {
+			unsigned short key : 4;
+			unsigned short	   : 4;
+			unsigned short as  : 2;
+			unsigned short	   : 4;
+			unsigned short k   : 1;
+			unsigned short a   : 1;
+		} oac1;
+		struct {
+			unsigned short key : 4;
+			unsigned short	   : 4;
+			unsigned short as  : 2;
+			unsigned short	   : 4;
+			unsigned short k   : 1;
+			unsigned short a   : 1;
+		} oac2;
+	};
+};
 
-unsigned long __must_check
-raw_copy_to_user(void __user *to, const void *from, unsigned long n);
+static __always_inline __must_check unsigned long
+raw_copy_from_user_key(void *to, const void __user *from, unsigned long size, unsigned long key)
+{
+	unsigned long rem;
+	union oac spec = {
+		.oac2.key = key,
+		.oac2.as = PSW_BITS_AS_SECONDARY,
+		.oac2.k = 1,
+		.oac2.a = 1,
+	};
 
-#ifndef CONFIG_KASAN
-#define INLINE_COPY_FROM_USER
-#define INLINE_COPY_TO_USER
-#endif
+	asm_inline volatile(
+		"	lr	%%r0,%[spec]\n"
+		"0:	mvcos	0(%[to]),0(%[from]),%[size]\n"
+		"1:	jz	5f\n"
+		"	algr	%[size],%[val]\n"
+		"	slgr	%[from],%[val]\n"
+		"	slgr	%[to],%[val]\n"
+		"	j	0b\n"
+		"2:	la	%[rem],4095(%[from])\n" /* rem = from + 4095 */
+		"	nr	%[rem],%[val]\n"	/* rem = (from + 4095) & -4096 */
+		"	slgr	%[rem],%[from]\n"
+		"	clgr	%[size],%[rem]\n"	/* copy crosses next page boundary? */
+		"	jnh	6f\n"
+		"3:	mvcos	0(%[to]),0(%[from]),%[rem]\n"
+		"4:	slgr	%[size],%[rem]\n"
+		"	j	6f\n"
+		"5:	lghi	%[size],0\n"
+		"6:\n"
+		EX_TABLE(0b, 2b)
+		EX_TABLE(1b, 2b)
+		EX_TABLE(3b, 6b)
+		EX_TABLE(4b, 6b)
+		: [size] "+&a" (size), [from] "+&a" (from), [to] "+&a" (to), [rem] "=&a" (rem)
+		: [val] "a" (-4096UL), [spec] "d" (spec.val)
+		: "cc", "memory", "0");
+	return size;
+}
+
+static __always_inline __must_check unsigned long
+raw_copy_from_user(void *to, const void __user *from, unsigned long n)
+{
+	return raw_copy_from_user_key(to, from, n, 0);
+}
+
+static __always_inline __must_check unsigned long
+raw_copy_to_user_key(void __user *to, const void *from, unsigned long size, unsigned long key)
+{
+	unsigned long rem;
+	union oac spec = {
+		.oac1.key = key,
+		.oac1.as = PSW_BITS_AS_SECONDARY,
+		.oac1.k = 1,
+		.oac1.a = 1,
+	};
+
+	asm_inline volatile(
+		"	lr	%%r0,%[spec]\n"
+		"0:	mvcos	0(%[to]),0(%[from]),%[size]\n"
+		"1:	jz	5f\n"
+		"	algr	%[size],%[val]\n"
+		"	slgr	%[to],%[val]\n"
+		"	slgr	%[from],%[val]\n"
+		"	j	0b\n"
+		"2:	la	%[rem],4095(%[to])\n"	/* rem = to + 4095 */
+		"	nr	%[rem],%[val]\n"	/* rem = (to + 4095) & -4096 */
+		"	slgr	%[rem],%[to]\n"
+		"	clgr	%[size],%[rem]\n"	/* copy crosses next page boundary? */
+		"	jnh	6f\n"
+		"3:	mvcos	0(%[to]),0(%[from]),%[rem]\n"
+		"4:	slgr	%[size],%[rem]\n"
+		"	j	6f\n"
+		"5:	lghi	%[size],0\n"
+		"6:\n"
+		EX_TABLE(0b, 2b)
+		EX_TABLE(1b, 2b)
+		EX_TABLE(3b, 6b)
+		EX_TABLE(4b, 6b)
+		: [size] "+&a" (size), [to] "+&a" (to), [from] "+&a" (from), [rem] "=&a" (rem)
+		: [val] "a" (-4096UL), [spec] "d" (spec.val)
+		: "cc", "memory", "0");
+	return size;
+}
+
+static __always_inline __must_check unsigned long
+raw_copy_to_user(void __user *to, const void *from, unsigned long n)
+{
+	return raw_copy_to_user_key(to, from, n, 0);
+}
 
 unsigned long __must_check
 _copy_from_user_key(void *to, const void __user *from, unsigned long n, unsigned long key);
@@ -55,63 +156,71 @@ copy_to_user_key(void __user *to, const void *from, unsigned long n, unsigned lo
 	return n;
 }
 
-union oac {
-	unsigned int val;
-	struct {
-		struct {
-			unsigned short key : 4;
-			unsigned short	   : 4;
-			unsigned short as  : 2;
-			unsigned short	   : 4;
-			unsigned short k   : 1;
-			unsigned short a   : 1;
-		} oac1;
-		struct {
-			unsigned short key : 4;
-			unsigned short	   : 4;
-			unsigned short as  : 2;
-			unsigned short	   : 4;
-			unsigned short k   : 1;
-			unsigned short a   : 1;
-		} oac2;
-	};
-};
-
 int __noreturn __put_user_bad(void);
 
 #ifdef CONFIG_KMSAN
-#define get_put_user_noinstr_attributes \
-	noinline __maybe_unused __no_sanitize_memory
+#define uaccess_kmsan_or_inline noinline __maybe_unused __no_sanitize_memory
 #else
-#define get_put_user_noinstr_attributes __always_inline
+#define uaccess_kmsan_or_inline __always_inline
 #endif
 
-#define DEFINE_PUT_USER(type)						\
-static get_put_user_noinstr_attributes int				\
+#ifdef CONFIG_CC_HAS_ASM_GOTO_OUTPUT
+
+#define DEFINE_PUT_USER_NOINSTR(type)					\
+static uaccess_kmsan_or_inline int					\
+__put_user_##type##_noinstr(unsigned type __user *to,			\
+			    unsigned type *from,			\
+			    unsigned long size)				\
+{									\
+	asm goto(							\
+		"	llilh	%%r0,%[spec]\n"				\
+		"0:	mvcos	%[to],%[from],%[size]\n"		\
+		"1:	nopr	%%r7\n"					\
+		EX_TABLE(0b, %l[Efault])				\
+		EX_TABLE(1b, %l[Efault])				\
+		: [to] "+Q" (*to)					\
+		: [size] "d" (size), [from] "Q" (*from),		\
+		  [spec] "I" (0x81)					\
+		: "cc", "0"						\
+		: Efault						\
+		);							\
+	return 0;							\
+Efault:									\
+	return -EFAULT;							\
+}
+
+#else /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT */
+
+#define DEFINE_PUT_USER_NOINSTR(type)					\
+static uaccess_kmsan_or_inline int					\
 __put_user_##type##_noinstr(unsigned type __user *to,			\
 			    unsigned type *from,			\
 			    unsigned long size)				\
 {									\
-	union oac __oac_spec = {					\
-		.oac1.as = PSW_BITS_AS_SECONDARY,			\
-		.oac1.a = 1,						\
-	};								\
 	int rc;								\
 									\
 	asm volatile(							\
-		"	lr	0,%[spec]\n"				\
-		"0:	mvcos	%[_to],%[_from],%[_size]\n"		\
-		"1:	xr	%[rc],%[rc]\n"				\
+		"	llilh	%%r0,%[spec]\n"				\
+		"0:	mvcos	%[to],%[from],%[size]\n"		\
+		"1:	lhi	%[rc],0\n"				\
 		"2:\n"							\
-		EX_TABLE_UA_STORE(0b, 2b, %[rc])			\
-		EX_TABLE_UA_STORE(1b, 2b, %[rc])			\
-		: [rc] "=&d" (rc), [_to] "+Q" (*(to))			\
-		: [_size] "d" (size), [_from] "Q" (*(from)),		\
-		  [spec] "d" (__oac_spec.val)				\
+		EX_TABLE_UA_FAULT(0b, 2b, %[rc])			\
+		EX_TABLE_UA_FAULT(1b, 2b, %[rc])			\
+		: [rc] "=d" (rc), [to] "+Q" (*to)			\
+		: [size] "d" (size), [from] "Q" (*from),		\
+		  [spec] "I" (0x81)					\
 		: "cc", "0");						\
 	return rc;							\
-}									\
-									\
+}
+
+#endif /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT */
+
+DEFINE_PUT_USER_NOINSTR(char);
+DEFINE_PUT_USER_NOINSTR(short);
+DEFINE_PUT_USER_NOINSTR(int);
+DEFINE_PUT_USER_NOINSTR(long);
+
+#define DEFINE_PUT_USER(type)						\
 static __always_inline int						\
 __put_user_##type(unsigned type __user *to, unsigned type *from,	\
 		  unsigned long size)					\
@@ -128,69 +237,111 @@ DEFINE_PUT_USER(short);
 DEFINE_PUT_USER(int);
 DEFINE_PUT_USER(long);
 
-static __always_inline int __put_user_fn(void *x, void __user *ptr, unsigned long size)
-{
-	int rc;
+#define __put_user(x, ptr)						\
+({									\
+	__typeof__(*(ptr)) __x = (x);					\
+	int __prc;							\
+									\
+	__chk_user_ptr(ptr);						\
+	switch (sizeof(*(ptr))) {					\
+	case 1:								\
+		__prc = __put_user_char((unsigned char __user *)(ptr),	\
+					(unsigned char *)&__x,		\
+					sizeof(*(ptr)));		\
+		break;							\
+	case 2:								\
+		__prc = __put_user_short((unsigned short __user *)(ptr),\
+					 (unsigned short *)&__x,	\
+					 sizeof(*(ptr)));		\
+		break;							\
+	case 4:								\
+		__prc = __put_user_int((unsigned int __user *)(ptr),	\
+				       (unsigned int *)&__x,		\
+				       sizeof(*(ptr)));			\
+		break;							\
+	case 8:								\
+		__prc = __put_user_long((unsigned long __user *)(ptr),	\
+					(unsigned long *)&__x,		\
+					sizeof(*(ptr)));		\
+		break;							\
+	default:							\
+		__prc = __put_user_bad();				\
+		break;							\
+	}								\
+	__builtin_expect(__prc, 0);					\
+})
 
-	switch (size) {
-	case 1:
-		rc = __put_user_char((unsigned char __user *)ptr,
-				     (unsigned char *)x,
-				     size);
-		break;
-	case 2:
-		rc = __put_user_short((unsigned short __user *)ptr,
-				      (unsigned short *)x,
-				      size);
-		break;
-	case 4:
-		rc = __put_user_int((unsigned int __user *)ptr,
-				    (unsigned int *)x,
-				    size);
-		break;
-	case 8:
-		rc = __put_user_long((unsigned long __user *)ptr,
-				     (unsigned long *)x,
-				     size);
-		break;
-	default:
-		__put_user_bad();
-		break;
-	}
-	return rc;
-}
+#define put_user(x, ptr)						\
+({									\
+	might_fault();							\
+	__put_user(x, ptr);						\
+})
 
 int __noreturn __get_user_bad(void);
 
-#define DEFINE_GET_USER(type)						\
-static get_put_user_noinstr_attributes int				\
+#ifdef CONFIG_CC_HAS_ASM_GOTO_OUTPUT
+
+#define DEFINE_GET_USER_NOINSTR(type)					\
+static uaccess_kmsan_or_inline int					\
 __get_user_##type##_noinstr(unsigned type *to,				\
-			    unsigned type __user *from,			\
+			    const unsigned type __user *from,		\
+			    unsigned long size)				\
+{									\
+	asm goto(							\
+		"	lhi	%%r0,%[spec]\n"				\
+		"0:	mvcos	%[to],%[from],%[size]\n"		\
+		"1:	nopr	%%r7\n"					\
+		EX_TABLE(0b, %l[Efault])				\
+		EX_TABLE(1b, %l[Efault])				\
+		: [to] "=Q" (*to)					\
+		: [size] "d" (size), [from] "Q" (*from),		\
+		  [spec] "I" (0x81)					\
+		: "cc", "0"						\
+		: Efault						\
+		);							\
+	return 0;							\
+Efault:									\
+	*to = 0;							\
+	return -EFAULT;							\
+}
+
+#else /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT */
+
+#define DEFINE_GET_USER_NOINSTR(type)					\
+static uaccess_kmsan_or_inline int					\
+__get_user_##type##_noinstr(unsigned type *to,				\
+			    const unsigned type __user *from,		\
 			    unsigned long size)				\
 {									\
-	union oac __oac_spec = {					\
-		.oac2.as = PSW_BITS_AS_SECONDARY,			\
-		.oac2.a = 1,						\
-	};								\
 	int rc;								\
 									\
 	asm volatile(							\
-		"	lr	0,%[spec]\n"				\
-		"0:	mvcos	0(%[_to]),%[_from],%[_size]\n"		\
-		"1:	xr	%[rc],%[rc]\n"				\
+		"	lhi	%%r0,%[spec]\n"				\
+		"0:	mvcos	%[to],%[from],%[size]\n"		\
+		"1:	lhi	%[rc],0\n"				\
 		"2:\n"							\
-		EX_TABLE_UA_LOAD_MEM(0b, 2b, %[rc], %[_to], %[_ksize])	\
-		EX_TABLE_UA_LOAD_MEM(1b, 2b, %[rc], %[_to], %[_ksize])	\
-		: [rc] "=&d" (rc), "=Q" (*(to))				\
-		: [_size] "d" (size), [_from] "Q" (*(from)),		\
-		  [spec] "d" (__oac_spec.val), [_to] "a" (to),		\
-		  [_ksize] "K" (size)					\
+		EX_TABLE_UA_FAULT(0b, 2b, %[rc])			\
+		EX_TABLE_UA_FAULT(1b, 2b, %[rc])			\
+		: [rc] "=d" (rc), [to] "=Q" (*to)			\
+		: [size] "d" (size), [from] "Q" (*from),		\
+		  [spec] "I" (0x81)					\
 		: "cc", "0");						\
+	if (likely(!rc))						\
+		return 0;						\
+	*to = 0;							\
 	return rc;							\
-}									\
-									\
+}
+
+#endif /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT */
+
+DEFINE_GET_USER_NOINSTR(char);
+DEFINE_GET_USER_NOINSTR(short);
+DEFINE_GET_USER_NOINSTR(int);
+DEFINE_GET_USER_NOINSTR(long);
+
+#define DEFINE_GET_USER(type)						\
 static __always_inline int						\
-__get_user_##type(unsigned type *to, unsigned type __user *from,	\
+__get_user_##type(unsigned type *to, const unsigned type __user *from,	\
 		  unsigned long size)					\
 {									\
 	int rc;								\
@@ -205,107 +356,50 @@ DEFINE_GET_USER(short);
 DEFINE_GET_USER(int);
 DEFINE_GET_USER(long);
 
-static __always_inline int __get_user_fn(void *x, const void __user *ptr, unsigned long size)
-{
-	int rc;
-
-	switch (size) {
-	case 1:
-		rc = __get_user_char((unsigned char *)x,
-				     (unsigned char __user *)ptr,
-				     size);
-		break;
-	case 2:
-		rc = __get_user_short((unsigned short *)x,
-				      (unsigned short __user *)ptr,
-				      size);
-		break;
-	case 4:
-		rc = __get_user_int((unsigned int *)x,
-				    (unsigned int __user *)ptr,
-				    size);
-		break;
-	case 8:
-		rc = __get_user_long((unsigned long *)x,
-				     (unsigned long __user *)ptr,
-				     size);
-		break;
-	default:
-		__get_user_bad();
-		break;
-	}
-	return rc;
-}
-
-/*
- * These are the main single-value transfer routines.  They automatically
- * use the right size if we just have the right pointer type.
- */
-#define __put_user(x, ptr)						\
-({									\
-	__typeof__(*(ptr)) __x = (x);					\
-	int __pu_err = -EFAULT;						\
-									\
-	__chk_user_ptr(ptr);						\
-	switch (sizeof(*(ptr))) {					\
-	case 1:								\
-	case 2:								\
-	case 4:								\
-	case 8:								\
-		__pu_err = __put_user_fn(&__x, ptr, sizeof(*(ptr)));	\
-		break;							\
-	default:							\
-		__put_user_bad();					\
-		break;							\
-	}								\
-	__builtin_expect(__pu_err, 0);					\
-})
-
-#define put_user(x, ptr)						\
-({									\
-	might_fault();							\
-	__put_user(x, ptr);						\
-})
-
 #define __get_user(x, ptr)						\
 ({									\
-	int __gu_err = -EFAULT;						\
+	const __user void *____guptr = (ptr);				\
+	int __grc;							\
 									\
 	__chk_user_ptr(ptr);						\
 	switch (sizeof(*(ptr))) {					\
 	case 1: {							\
+		const unsigned char __user *__guptr = ____guptr;	\
 		unsigned char __x;					\
 									\
-		__gu_err = __get_user_fn(&__x, ptr, sizeof(*(ptr)));	\
+		__grc = __get_user_char(&__x, __guptr, sizeof(*(ptr)));	\
 		(x) = *(__force __typeof__(*(ptr)) *)&__x;		\
 		break;							\
 	};								\
 	case 2: {							\
+		const unsigned short __user *__guptr = ____guptr;	\
 		unsigned short __x;					\
 									\
-		__gu_err = __get_user_fn(&__x, ptr, sizeof(*(ptr)));	\
+		__grc = __get_user_short(&__x, __guptr, sizeof(*(ptr)));\
 		(x) = *(__force __typeof__(*(ptr)) *)&__x;		\
 		break;							\
 	};								\
 	case 4: {							\
+		const unsigned int __user *__guptr = ____guptr;		\
 		unsigned int __x;					\
 									\
-		__gu_err = __get_user_fn(&__x, ptr, sizeof(*(ptr)));	\
+		__grc = __get_user_int(&__x, __guptr, sizeof(*(ptr)));	\
 		(x) = *(__force __typeof__(*(ptr)) *)&__x;		\
 		break;							\
 	};								\
 	case 8: {							\
+		const unsigned long __user *__guptr = ____guptr;	\
 		unsigned long __x;					\
 									\
-		__gu_err = __get_user_fn(&__x, ptr, sizeof(*(ptr)));	\
+		__grc = __get_user_long(&__x, __guptr, sizeof(*(ptr)));	\
 		(x) = *(__force __typeof__(*(ptr)) *)&__x;		\
 		break;							\
 	};								\
 	default:							\
-		__get_user_bad();					\
+		__grc = __get_user_bad();				\
 		break;							\
 	}								\
-	__builtin_expect(__gu_err, 0);					\
+	__builtin_expect(__grc, 0);					\
 })
 
 #define get_user(x, ptr)						\
@@ -341,109 +435,71 @@ static inline void *s390_kernel_write(void *dst, const void *src, size_t size)
 	return __s390_kernel_write(dst, src, size);
 }
 
-int __noreturn __put_kernel_bad(void);
+void __noreturn __mvc_kernel_nofault_bad(void);
 
-#define __put_kernel_asm(val, to, insn)					\
-({									\
-	int __rc;							\
-									\
-	asm volatile(							\
-		"0:   " insn "  %[_val],%[_to]\n"			\
-		"1:	xr	%[rc],%[rc]\n"				\
-		"2:\n"							\
-		EX_TABLE_UA_STORE(0b, 2b, %[rc])			\
-		EX_TABLE_UA_STORE(1b, 2b, %[rc])			\
-		: [rc] "=d" (__rc), [_to] "+Q" (*(to))			\
-		: [_val] "d" (val)					\
-		: "cc");						\
-	__rc;								\
-})
+#if defined(CONFIG_CC_HAS_ASM_GOTO_OUTPUT) && defined(CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS)
 
-#define __put_kernel_nofault(dst, src, type, err_label)			\
+#define __mvc_kernel_nofault(dst, src, type, err_label)			\
 do {									\
-	unsigned long __x = (unsigned long)(*((type *)(src)));		\
-	int __pk_err;							\
-									\
 	switch (sizeof(type)) {						\
 	case 1:								\
-		__pk_err = __put_kernel_asm(__x, (type *)(dst), "stc"); \
-		break;							\
 	case 2:								\
-		__pk_err = __put_kernel_asm(__x, (type *)(dst), "sth"); \
-		break;							\
 	case 4:								\
-		__pk_err = __put_kernel_asm(__x, (type *)(dst), "st");	\
-		break;							\
 	case 8:								\
-		__pk_err = __put_kernel_asm(__x, (type *)(dst), "stg"); \
+		asm goto(						\
+			"0:	mvc	%O[_dst](%[_len],%R[_dst]),%[_src]\n" \
+			"1:	nopr	%%r7\n"				\
+			EX_TABLE(0b, %l[err_label])			\
+			EX_TABLE(1b, %l[err_label])			\
+			: [_dst] "=Q" (*(type *)dst)			\
+			: [_src] "Q" (*(type *)(src)),			\
+			  [_len] "I" (sizeof(type))			\
+			:						\
+			: err_label);					\
 		break;							\
 	default:							\
-		__pk_err = __put_kernel_bad();				\
+		__mvc_kernel_nofault_bad();				\
 		break;							\
 	}								\
-	if (unlikely(__pk_err))						\
-		goto err_label;						\
 } while (0)
 
-int __noreturn __get_kernel_bad(void);
-
-#define __get_kernel_asm(val, from, insn)				\
-({									\
-	int __rc;							\
-									\
-	asm volatile(							\
-		"0:   " insn "  %[_val],%[_from]\n"			\
-		"1:	xr	%[rc],%[rc]\n"				\
-		"2:\n"							\
-		EX_TABLE_UA_LOAD_REG(0b, 2b, %[rc], %[_val])		\
-		EX_TABLE_UA_LOAD_REG(1b, 2b, %[rc], %[_val])		\
-		: [rc] "=d" (__rc), [_val] "=d" (val)			\
-		: [_from] "Q" (*(from))					\
-		: "cc");						\
-	__rc;								\
-})
+#else /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT) && CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
 
-#define __get_kernel_nofault(dst, src, type, err_label)			\
+#define __mvc_kernel_nofault(dst, src, type, err_label)			\
 do {									\
-	int __gk_err;							\
+	type *(__dst) = (type *)(dst);					\
+	int __rc;							\
 									\
 	switch (sizeof(type)) {						\
-	case 1: {							\
-		unsigned char __x;					\
-									\
-		__gk_err = __get_kernel_asm(__x, (type *)(src), "ic");	\
-		*((type *)(dst)) = (type)__x;				\
-		break;							\
-	};								\
-	case 2: {							\
-		unsigned short __x;					\
-									\
-		__gk_err = __get_kernel_asm(__x, (type *)(src), "lh");	\
-		*((type *)(dst)) = (type)__x;				\
-		break;							\
-	};								\
-	case 4: {							\
-		unsigned int __x;					\
-									\
-		__gk_err = __get_kernel_asm(__x, (type *)(src), "l");	\
-		*((type *)(dst)) = (type)__x;				\
-		break;							\
-	};								\
-	case 8: {							\
-		unsigned long __x;					\
-									\
-		__gk_err = __get_kernel_asm(__x, (type *)(src), "lg");	\
-		*((type *)(dst)) = (type)__x;				\
+	case 1:								\
+	case 2:								\
+	case 4:								\
+	case 8:								\
+		asm_inline volatile(					\
+			"0:	mvc	0(%[_len],%[_dst]),%[_src]\n"	\
+			"1:	lhi	%[_rc],0\n"			\
+			"2:\n"						\
+			EX_TABLE_UA_FAULT(0b, 2b, %[_rc])		\
+			EX_TABLE_UA_FAULT(1b, 2b, %[_rc])		\
+			: [_rc] "=d" (__rc),				\
+			  "=m" (*__dst)					\
+			: [_src] "Q" (*(type *)(src)),			\
+			[_dst] "a" (__dst),				\
+			[_len] "I" (sizeof(type)));			\
+		if (__rc)						\
+			goto err_label;					\
 		break;							\
-	};								\
 	default:							\
-		__gk_err = __get_kernel_bad();				\
+		__mvc_kernel_nofault_bad();				\
 		break;							\
 	}								\
-	if (unlikely(__gk_err))						\
-		goto err_label;						\
 } while (0)
 
+#endif /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT && CONFIG_CC_HAS_ASM_AOR_FORMAT_FLAGS */
+
+#define __get_kernel_nofault __mvc_kernel_nofault
+#define __put_kernel_nofault __mvc_kernel_nofault
+
 void __cmpxchg_user_key_called_with_bad_pointer(void);
 
 #define CMPXCHG_USER_KEY_MAX_LOOPS 128