s390: add a system call for guarded storage

This adds a new system call to enable the use of guarded storage for
user space processes. The system call takes two arguments, a command
and pointer to a guarded storage control block:

    s390_guarded_storage(int command, struct gs_cb *gs_cb);

The second argument is relevant only for the GS_SET_BC_CB command.

The commands in detail:

0 - GS_ENABLE
    Enable the guarded storage facility for the current task. The
    initial content of the guarded storage control block will be
    all zeros. After the enablement the user space code can use
    load-guarded-storage-controls instruction (LGSC) to load an
    arbitrary control block. While a task is enabled the kernel
    will save and restore the current content of the guarded
    storage registers on context switch.
1 - GS_DISABLE
    Disables the use of the guarded storage facility for the current
    task. The kernel will cease to save and restore the content of
    the guarded storage registers, the task specific content of
    these registers is lost.
2 - GS_SET_BC_CB
    Set a broadcast guarded storage control block. This is called
    per thread and stores a specific guarded storage control block
    in the task struct of the current task. This control block will
    be used for the broadcast event GS_BROADCAST.
3 - GS_CLEAR_BC_CB
    Clears the broadcast guarded storage control block. The guarded-
    storage control block is removed from the task struct that was
    established by GS_SET_BC_CB.
4 - GS_BROADCAST
    Sends a broadcast to all thread siblings of the current task.
    Every sibling that has established a broadcast guarded storage
    control block will load this control block and will be enabled
    for guarded storage. The broadcast guarded storage control block
    is used up, a second broadcast without a refresh of the stored
    control block with GS_SET_BC_CB will not have any effect.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/include/asm/elf.h

@@ -105,6 +105,7 @@
 #define HWCAP_S390_VXRS		2048
 #define HWCAP_S390_VXRS_BCD	4096
 #define HWCAP_S390_VXRS_EXT	8192
+#define HWCAP_S390_GS		16384
 
 /* Internal bits, not exposed via elf */
 #define HWCAP_INT_SIE		1UL

arch/s390/include/asm/lowcore.h

@@ -157,8 +157,8 @@ struct lowcore {
 	__u64	stfle_fac_list[32];		/* 0x0f00 */
 	__u8	pad_0x1000[0x11b0-0x1000];	/* 0x1000 */
 
-	/* Pointer to vector register save area */
-	__u64	vector_save_area_addr;		/* 0x11b0 */
+	/* Pointer to the machine check extended save area */
+	__u64	mcesad;				/* 0x11b0 */
 
 	/* 64 bit extparam used for pfault/diag 250: defined by architecture */
 	__u64	ext_params2;			/* 0x11B8 */
@@ -182,10 +182,7 @@ struct lowcore {
 
 	/* Transaction abort diagnostic block */
 	__u8	pgm_tdb[256];			/* 0x1800 */
-	__u8	pad_0x1900[0x1c00-0x1900];	/* 0x1900 */
-
-	/* Software defined save area for vector registers */
-	__u8	vector_save_area[1024];		/* 0x1c00 */
+	__u8	pad_0x1900[0x2000-0x1900];	/* 0x1900 */
 } __packed;
 
 #define S390_lowcore (*((struct lowcore *) 0))

arch/s390/include/asm/nmi.h

@@ -58,7 +58,9 @@ union mci {
 		u64 ie :  1; /* 32 indirect storage error */
 		u64 ar :  1; /* 33 access register validity */
 		u64 da :  1; /* 34 delayed access exception */
-		u64    :  7; /* 35-41 */
+		u64    :  1; /* 35 */
+		u64 gs :  1; /* 36 guarded storage registers */
+		u64    :  5; /* 37-41 */
 		u64 pr :  1; /* 42 tod programmable register validity */
 		u64 fc :  1; /* 43 fp control register validity */
 		u64 ap :  1; /* 44 ancillary report */
@@ -69,6 +71,14 @@ union mci {
 	};
 };
 
+#define MCESA_ORIGIN_MASK	(~0x3ffUL)
+#define MCESA_LC_MASK		(0xfUL)
+
+struct mcesa {
+	u8 vector_save_area[1024];
+	u8 guarded_storage_save_area[32];
+};
+
 struct pt_regs;
 
 extern void s390_handle_mcck(void);

arch/s390/include/asm/processor.h

@@ -135,6 +135,8 @@ struct thread_struct {
 	struct list_head list;
 	/* cpu runtime instrumentation */
 	struct runtime_instr_cb *ri_cb;
+	struct gs_cb *gs_cb;		/* Current guarded storage cb */
+	struct gs_cb *gs_bc_cb;		/* Broadcast guarded storage cb */
 	unsigned char trap_tdb[256];	/* Transaction abort diagnose block */
 	/*
 	 * Warning: 'fpu' is dynamically-sized. It *MUST* be at
@@ -215,6 +217,9 @@ void show_cacheinfo(struct seq_file *m);
 /* Free all resources held by a thread. */
 extern void release_thread(struct task_struct *);
 
+/* Free guarded storage control block for current */
+void exit_thread_gs(void);
+
 /*
  * Return saved PC of a blocked thread.
  */

arch/s390/include/asm/setup.h

@@ -31,6 +31,7 @@
 #define MACHINE_FLAG_VX		_BITUL(13)
 #define MACHINE_FLAG_CAD	_BITUL(14)
 #define MACHINE_FLAG_NX		_BITUL(15)
+#define MACHINE_FLAG_GS		_BITUL(16)
 
 #define LPP_MAGIC		_BITUL(31)
 #define LPP_PFAULT_PID_MASK	_AC(0xffffffff, UL)
@@ -70,6 +71,7 @@ extern void detect_memory_memblock(void);
 #define MACHINE_HAS_VX		(S390_lowcore.machine_flags & MACHINE_FLAG_VX)
 #define MACHINE_HAS_CAD		(S390_lowcore.machine_flags & MACHINE_FLAG_CAD)
 #define MACHINE_HAS_NX		(S390_lowcore.machine_flags & MACHINE_FLAG_NX)
+#define MACHINE_HAS_GS		(S390_lowcore.machine_flags & MACHINE_FLAG_GS)
 
 /*
  * Console mode. Override with conmode=

arch/s390/include/asm/switch_to.h

@@ -10,6 +10,7 @@
 #include <linux/thread_info.h>
 #include <asm/fpu/api.h>
 #include <asm/ptrace.h>
+#include <asm/guarded_storage.h>
 
 extern struct task_struct *__switch_to(void *, void *);
 extern void update_cr_regs(struct task_struct *task);
@@ -33,12 +34,14 @@ static inline void restore_access_regs(unsigned int *acrs)
 		save_fpu_regs();					\
 		save_access_regs(&prev->thread.acrs[0]);		\
 		save_ri_cb(prev->thread.ri_cb);				\
+		save_gs_cb(prev->thread.gs_cb);				\
 	}								\
 	if (next->mm) {							\
 		update_cr_regs(next);					\
 		set_cpu_flag(CIF_FPU);					\
 		restore_access_regs(&next->thread.acrs[0]);		\
 		restore_ri_cb(next->thread.ri_cb, prev->thread.ri_cb);	\
+		restore_gs_cb(next->thread.gs_cb);			\
 	}								\
 	prev = __switch_to(prev,next);					\
 } while (0)

arch/s390/include/asm/thread_info.h

@@ -54,11 +54,12 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src);
 #define TIF_NOTIFY_RESUME	0	/* callback before returning to user */
 #define TIF_SIGPENDING		1	/* signal pending */
 #define TIF_NEED_RESCHED	2	/* rescheduling necessary */
-#define TIF_SYSCALL_TRACE	3	/* syscall trace active */
-#define TIF_SYSCALL_AUDIT	4	/* syscall auditing active */
-#define TIF_SECCOMP		5	/* secure computing */
-#define TIF_SYSCALL_TRACEPOINT	6	/* syscall tracepoint instrumentation */
-#define TIF_UPROBE		7	/* breakpointed or single-stepping */
+#define TIF_UPROBE		3	/* breakpointed or single-stepping */
+#define TIF_GUARDED_STORAGE	4	/* load guarded storage control block */
+#define TIF_SYSCALL_TRACE	8	/* syscall trace active */
+#define TIF_SYSCALL_AUDIT	9	/* syscall auditing active */
+#define TIF_SECCOMP		10	/* secure computing */
+#define TIF_SYSCALL_TRACEPOINT	11	/* syscall tracepoint instrumentation */
 #define TIF_31BIT		16	/* 32bit process */
 #define TIF_MEMDIE		17	/* is terminating due to OOM killer */
 #define TIF_RESTORE_SIGMASK	18	/* restore signal mask in do_signal() */
@@ -76,5 +77,6 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src);
 #define _TIF_UPROBE		_BITUL(TIF_UPROBE)
 #define _TIF_31BIT		_BITUL(TIF_31BIT)
 #define _TIF_SINGLE_STEP	_BITUL(TIF_SINGLE_STEP)
+#define _TIF_GUARDED_STORAGE	_BITUL(TIF_GUARDED_STORAGE)
 
 #endif /* _ASM_THREAD_INFO_H */

arch/s390/include/uapi/asm/Kbuild

@@ -12,6 +12,7 @@ header-y += dasd.h
 header-y += debug.h
 header-y += errno.h
 header-y += fcntl.h
+header-y += guarded_storage.h
 header-y += hypfs.h
 header-y += ioctl.h
 header-y += ioctls.h

arch/s390/include/uapi/asm/guarded_storage.h

+#ifndef _GUARDED_STORAGE_H
+#define _GUARDED_STORAGE_H
+
+#include <linux/types.h>
+
+struct gs_cb {
+	__u64 reserved;
+	__u64 gsd;
+	__u64 gssm;
+	__u64 gs_epl_a;
+};
+
+struct gs_epl {
+	__u8 pad1;
+	union {
+		__u8 gs_eam;
+		struct {
+			__u8	: 6;
+			__u8 e	: 1;
+			__u8 b	: 1;
+		};
+	};
+	union {
+		__u8 gs_eci;
+		struct {
+			__u8 tx	: 1;
+			__u8 cx	: 1;
+			__u8	: 5;
+			__u8 in	: 1;
+		};
+	};
+	union {
+		__u8 gs_eai;
+		struct {
+			__u8	: 1;
+			__u8 t	: 1;
+			__u8 as	: 2;
+			__u8 ar	: 4;
+		};
+	};
+	__u32 pad2;
+	__u64 gs_eha;
+	__u64 gs_eia;
+	__u64 gs_eoa;
+	__u64 gs_eir;
+	__u64 gs_era;
+};
+
+#define GS_ENABLE	0
+#define	GS_DISABLE	1
+#define GS_SET_BC_CB	2
+#define GS_CLEAR_BC_CB	3
+#define GS_BROADCAST	4
+
+static inline void load_gs_cb(struct gs_cb *gs_cb)
+{
+	asm volatile(".insn rxy,0xe3000000004d,0,%0" : : "Q" (*gs_cb));
+}
+
+static inline void store_gs_cb(struct gs_cb *gs_cb)
+{
+	asm volatile(".insn rxy,0xe30000000049,0,%0" : : "Q" (*gs_cb));
+}
+
+static inline void save_gs_cb(struct gs_cb *gs_cb)
+{
+	if (gs_cb)
+		store_gs_cb(gs_cb);
+}
+
+static inline void restore_gs_cb(struct gs_cb *gs_cb)
+{
+	if (gs_cb)
+		load_gs_cb(gs_cb);
+}
+
+#endif /* _GUARDED_STORAGE_H */

arch/s390/include/uapi/asm/unistd.h

@@ -313,7 +313,7 @@
 #define __NR_copy_file_range	375
 #define __NR_preadv2		376
 #define __NR_pwritev2		377
-/* Number 378 is reserved for guarded storage */
+#define __NR_s390_guarded_storage	378
 #define __NR_statx		379
 #define NR_syscalls 380
 

arch/s390/kernel/Makefile

@@ -57,7 +57,7 @@ obj-y	:= traps.o time.o process.o base.o early.o setup.o idle.o vtime.o
 obj-y	+= processor.o sys_s390.o ptrace.o signal.o cpcmd.o ebcdic.o nmi.o
 obj-y	+= debug.o irq.o ipl.o dis.o diag.o vdso.o als.o
 obj-y	+= sysinfo.o jump_label.o lgr.o os_info.o machine_kexec.o pgm_check.o
-obj-y	+= runtime_instr.o cache.o fpu.o dumpstack.o
+obj-y	+= runtime_instr.o cache.o fpu.o dumpstack.o guarded_storage.o
 obj-y	+= entry.o reipl.o relocate_kernel.o
 
 extra-y				+= head.o head64.o vmlinux.lds

arch/s390/kernel/asm-offsets.c

@@ -175,7 +175,7 @@ int main(void)
 	/* software defined ABI-relevant lowcore locations 0xe00 - 0xe20 */
 	OFFSET(__LC_DUMP_REIPL, lowcore, ipib);
 	/* hardware defined lowcore locations 0x1000 - 0x18ff */
-	OFFSET(__LC_VX_SAVE_AREA_ADDR, lowcore, vector_save_area_addr);
+	OFFSET(__LC_MCESAD, lowcore, mcesad);
 	OFFSET(__LC_EXT_PARAMS2, lowcore, ext_params2);
 	OFFSET(__LC_FPREGS_SAVE_AREA, lowcore, floating_pt_save_area);
 	OFFSET(__LC_GPREGS_SAVE_AREA, lowcore, gpregs_save_area);

arch/s390/kernel/compat_wrapper.c

@@ -178,4 +178,5 @@ COMPAT_SYSCALL_WRAP3(getpeername, int, fd, struct sockaddr __user *, usockaddr,
 COMPAT_SYSCALL_WRAP6(sendto, int, fd, void __user *, buff, size_t, len, unsigned int, flags, struct sockaddr __user *, addr, int, addr_len);
 COMPAT_SYSCALL_WRAP3(mlock2, unsigned long, start, size_t, len, int, flags);
 COMPAT_SYSCALL_WRAP6(copy_file_range, int, fd_in, loff_t __user *, off_in, int, fd_out, loff_t __user *, off_out, size_t, len, unsigned int, flags);
+COMPAT_SYSCALL_WRAP2(s390_guarded_storage, int, command, struct gs_cb *, gs_cb);
 COMPAT_SYSCALL_WRAP5(statx, int, dfd, const char __user *, path, unsigned, flags, unsigned, mask, struct statx __user *, buffer);

arch/s390/kernel/early.c

@@ -358,6 +358,8 @@ static __init void detect_machine_facilities(void)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_NX;
 		__ctl_set_bit(0, 20);
 	}
+	if (test_facility(133))
+		S390_lowcore.machine_flags |= MACHINE_FLAG_GS;
 }
 
 static inline void save_vector_registers(void)

arch/s390/kernel/entry.S

@@ -47,7 +47,7 @@ STACK_SIZE  = 1 << STACK_SHIFT
 STACK_INIT = STACK_SIZE - STACK_FRAME_OVERHEAD - __PT_SIZE
 
 _TIF_WORK	= (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
-		   _TIF_UPROBE)
+		   _TIF_UPROBE | _TIF_GUARDED_STORAGE)
 _TIF_TRACE	= (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | _TIF_SECCOMP | \
 		   _TIF_SYSCALL_TRACEPOINT)
 _CIF_WORK	= (_CIF_MCCK_PENDING | _CIF_ASCE_PRIMARY | \
@@ -332,6 +332,8 @@ ENTRY(system_call)
 	TSTMSK	__TI_flags(%r12),_TIF_UPROBE
 	jo	.Lsysc_uprobe_notify
 #endif
+	TSTMSK	__TI_flags(%r12),_TIF_GUARDED_STORAGE
+	jo	.Lsysc_guarded_storage
 	TSTMSK	__PT_FLAGS(%r11),_PIF_PER_TRAP
 	jo	.Lsysc_singlestep
 	TSTMSK	__TI_flags(%r12),_TIF_SIGPENDING
@@ -408,6 +410,14 @@ ENTRY(system_call)
 	jg	uprobe_notify_resume
 #endif
 
+#
+# _TIF_GUARDED_STORAGE is set, call guarded_storage_load
+#
+.Lsysc_guarded_storage:
+	lgr	%r2,%r11		# pass pointer to pt_regs
+	larl	%r14,.Lsysc_return
+	jg	gs_load_bc_cb
+
 #
 # _PIF_PER_TRAP is set, call do_per_trap
 #
@@ -663,6 +673,8 @@ ENTRY(io_int_handler)
 	jo	.Lio_sigpending
 	TSTMSK	__TI_flags(%r12),_TIF_NOTIFY_RESUME
 	jo	.Lio_notify_resume
+	TSTMSK	__TI_flags(%r12),_TIF_GUARDED_STORAGE
+	jo	.Lio_guarded_storage
 	TSTMSK	__LC_CPU_FLAGS,_CIF_FPU
 	jo	.Lio_vxrs
 	TSTMSK	__LC_CPU_FLAGS,(_CIF_ASCE_PRIMARY|_CIF_ASCE_SECONDARY)
@@ -696,6 +708,18 @@ ENTRY(io_int_handler)
 	larl	%r14,.Lio_return
 	jg	load_fpu_regs
 
+#
+# _TIF_GUARDED_STORAGE is set, call guarded_storage_load
+#
+.Lio_guarded_storage:
+	# TRACE_IRQS_ON already done at .Lio_return
+	ssm	__LC_SVC_NEW_PSW	# reenable interrupts
+	lgr	%r2,%r11		# pass pointer to pt_regs
+	brasl	%r14,gs_load_bc_cb
+	ssm	__LC_PGM_NEW_PSW	# disable I/O and ext. interrupts
+	TRACE_IRQS_OFF
+	j	.Lio_return
+
 #
 # _TIF_NEED_RESCHED is set, call schedule
 #

arch/s390/kernel/entry.h

@@ -74,12 +74,14 @@ long sys_sigreturn(void);
 
 long sys_s390_personality(unsigned int personality);
 long sys_s390_runtime_instr(int command, int signum);
+long sys_s390_guarded_storage(int command, struct gs_cb __user *);
 long sys_s390_pci_mmio_write(unsigned long, const void __user *, size_t);
 long sys_s390_pci_mmio_read(unsigned long, void __user *, size_t);
 
 DECLARE_PER_CPU(u64, mt_cycles[8]);
 
 void verify_facilities(void);
+void gs_load_bc_cb(struct pt_regs *regs);
 void set_fs_fixup(void);
 
 #endif /* _ENTRY_H */

arch/s390/kernel/guarded_storage.c

+/*
+ * Copyright IBM Corp. 2016
+ * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/syscalls.h>
+#include <linux/signal.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <asm/guarded_storage.h>
+#include "entry.h"
+
+void exit_thread_gs(void)
+{
+	kfree(current->thread.gs_cb);
+	kfree(current->thread.gs_bc_cb);
+	current->thread.gs_cb = current->thread.gs_bc_cb = NULL;
+}
+
+static int gs_enable(void)
+{
+	struct gs_cb *gs_cb;
+
+	if (!current->thread.gs_cb) {
+		gs_cb = kzalloc(sizeof(*gs_cb), GFP_KERNEL);
+		if (!gs_cb)
+			return -ENOMEM;
+		gs_cb->gsd = 25;
+		preempt_disable();
+		__ctl_set_bit(2, 4);
+		load_gs_cb(gs_cb);
+		current->thread.gs_cb = gs_cb;
+		preempt_enable();
+	}
+	return 0;
+}
+
+static int gs_disable(void)
+{
+	if (current->thread.gs_cb) {
+		preempt_disable();
+		kfree(current->thread.gs_cb);
+		current->thread.gs_cb = NULL;
+		__ctl_clear_bit(2, 4);
+		preempt_enable();
+	}
+	return 0;
+}
+
+static int gs_set_bc_cb(struct gs_cb __user *u_gs_cb)
+{
+	struct gs_cb *gs_cb;
+
+	gs_cb = current->thread.gs_bc_cb;
+	if (!gs_cb) {
+		gs_cb = kzalloc(sizeof(*gs_cb), GFP_KERNEL);
+		if (!gs_cb)
+			return -ENOMEM;
+		current->thread.gs_bc_cb = gs_cb;
+	}
+	if (copy_from_user(gs_cb, u_gs_cb, sizeof(*gs_cb)))
+		return -EFAULT;
+	return 0;
+}
+
+static int gs_clear_bc_cb(void)
+{
+	struct gs_cb *gs_cb;
+
+	gs_cb = current->thread.gs_bc_cb;
+	current->thread.gs_bc_cb = NULL;
+	kfree(gs_cb);
+	return 0;
+}
+
+void gs_load_bc_cb(struct pt_regs *regs)
+{
+	struct gs_cb *gs_cb;
+
+	preempt_disable();
+	clear_thread_flag(TIF_GUARDED_STORAGE);
+	gs_cb = current->thread.gs_bc_cb;
+	if (gs_cb) {
+		kfree(current->thread.gs_cb);
+		current->thread.gs_bc_cb = NULL;
+		__ctl_set_bit(2, 4);
+		load_gs_cb(gs_cb);
+		current->thread.gs_cb = gs_cb;
+	}
+	preempt_enable();
+}
+
+static int gs_broadcast(void)
+{
+	struct task_struct *sibling;
+
+	read_lock(&tasklist_lock);
+	for_each_thread(current, sibling) {
+		if (!sibling->thread.gs_bc_cb)
+			continue;
+		if (test_and_set_tsk_thread_flag(sibling, TIF_GUARDED_STORAGE))
+			kick_process(sibling);
+	}
+	read_unlock(&tasklist_lock);
+	return 0;
+}
+
+SYSCALL_DEFINE2(s390_guarded_storage, int, command,
+		struct gs_cb __user *, gs_cb)
+{
+	if (!MACHINE_HAS_GS)
+		return -EOPNOTSUPP;
+	switch (command) {
+	case GS_ENABLE:
+		return gs_enable();
+	case GS_DISABLE:
+		return gs_disable();
+	case GS_SET_BC_CB:
+		return gs_set_bc_cb(gs_cb);
+	case GS_CLEAR_BC_CB:
+		return gs_clear_bc_cb();
+	case GS_BROADCAST:
+		return gs_broadcast();
+	default:
+		return -EINVAL;
+	}
+}

arch/s390/kernel/machine_kexec.c

@@ -27,6 +27,7 @@
 #include <asm/cacheflush.h>
 #include <asm/os_info.h>
 #include <asm/switch_to.h>
+#include <asm/nmi.h>
 
 typedef void (*relocate_kernel_t)(kimage_entry_t *, unsigned long);
 
@@ -102,6 +103,8 @@ static void __do_machine_kdump(void *image)
  */
 static noinline void __machine_kdump(void *image)
 {
+	struct mcesa *mcesa;
+	unsigned long cr2_old, cr2_new;
 	int this_cpu, cpu;
 
 	lgr_info_log();
@@ -114,8 +117,16 @@ static noinline void __machine_kdump(void *image)
 			continue;
 	}
 	/* Store status of the boot CPU */
+	mcesa = (struct mcesa *)(S390_lowcore.mcesad & MCESA_ORIGIN_MASK);
 	if (MACHINE_HAS_VX)
-		save_vx_regs((void *) &S390_lowcore.vector_save_area);
+		save_vx_regs((__vector128 *) mcesa->vector_save_area);
+	if (MACHINE_HAS_GS) {
+		__ctl_store(cr2_old, 2, 2);
+		cr2_new = cr2_old | (1UL << 4);
+		__ctl_load(cr2_new, 2, 2);
+		save_gs_cb((struct gs_cb *) mcesa->guarded_storage_save_area);
+		__ctl_load(cr2_old, 2, 2);
+	}
 	/*
 	 * To create a good backchain for this CPU in the dump store_status
 	 * is passed the address of a function. The address is saved into

arch/s390/kernel/nmi.c

@@ -106,6 +106,7 @@ static int notrace s390_validate_registers(union mci mci, int umode)
 	int kill_task;
 	u64 zero;
 	void *fpt_save_area;
+	struct mcesa *mcesa;
 
 	kill_task = 0;
 	zero = 0;
@@ -165,6 +166,7 @@ static int notrace s390_validate_registers(union mci mci, int umode)
 			     : : "Q" (S390_lowcore.fpt_creg_save_area));
 	}
 
+	mcesa = (struct mcesa *)(S390_lowcore.mcesad & MCESA_ORIGIN_MASK);
 	if (!MACHINE_HAS_VX) {
 		/* Validate floating point registers */
 		asm volatile(
@@ -209,8 +211,8 @@ static int notrace s390_validate_registers(union mci mci, int umode)
 			"	la	1,%0\n"
 			"	.word	0xe70f,0x1000,0x0036\n"	/* vlm 0,15,0(1) */
 			"	.word	0xe70f,0x1100,0x0c36\n"	/* vlm 16,31,256(1) */
-			: : "Q" (*(struct vx_array *)
-				 &S390_lowcore.vector_save_area) : "1");
+			: : "Q" (*(struct vx_array *) mcesa->vector_save_area)
+			: "1");
 		__ctl_load(S390_lowcore.cregs_save_area[0], 0, 0);
 	}
 	/* Validate access registers */
@@ -224,6 +226,19 @@ static int notrace s390_validate_registers(union mci mci, int umode)
 		 */
 		kill_task = 1;
 	}
+	/* Validate guarded storage registers */
+	if (MACHINE_HAS_GS && (S390_lowcore.cregs_save_area[2] & (1UL << 4))) {
+		if (!mci.gs)
+			/*
+			 * Guarded storage register can't be restored and
+			 * the current processes uses guarded storage.
+			 * It has to be terminated.
+			 */
+			kill_task = 1;
+		else
+			load_gs_cb((struct gs_cb *)
+				   mcesa->guarded_storage_save_area);
+	}
 	/*
 	 * We don't even try to validate the TOD register, since we simply
 	 * can't write something sensible into that register.

arch/s390/kernel/process.c

@@ -73,8 +73,10 @@ extern void kernel_thread_starter(void);
  */
 void exit_thread(struct task_struct *tsk)
 {
-	if (tsk == current)
+	if (tsk == current) {
 		exit_thread_runtime_instr();
+		exit_thread_gs();
+	}
 }
 
 void flush_thread(void)
@@ -159,6 +161,9 @@ int copy_thread_tls(unsigned long clone_flags, unsigned long new_stackp,
 	/* Don't copy runtime instrumentation info */
 	p->thread.ri_cb = NULL;
 	frame->childregs.psw.mask &= ~PSW_MASK_RI;
+	/* Don't copy guarded storage control block */
+	p->thread.gs_cb = NULL;
+	p->thread.gs_bc_cb = NULL;
 
 	/* Set a new TLS ?  */
 	if (clone_flags & CLONE_SETTLS) {

arch/s390/kernel/processor.c

@@ -95,7 +95,7 @@ static void show_cpu_summary(struct seq_file *m, void *v)
 {
 	static const char *hwcap_str[] = {
 		"esan3", "zarch", "stfle", "msa", "ldisp", "eimm", "dfp",
-		"edat", "etf3eh", "highgprs", "te", "vx", "vxd", "vxe"
+		"edat", "etf3eh", "highgprs", "te", "vx", "vxd", "vxe", "gs"
 	};
 	static const char * const int_hwcap_str[] = {
 		"sie"

arch/s390/kernel/ptrace.c

@@ -44,30 +44,42 @@ void update_cr_regs(struct task_struct *task)
 	struct pt_regs *regs = task_pt_regs(task);
 	struct thread_struct *thread = &task->thread;
 	struct per_regs old, new;
-
+	unsigned long cr0_old, cr0_new;
+	unsigned long cr2_old, cr2_new;
+	int cr0_changed, cr2_changed;
+
+	__ctl_store(cr0_old, 0, 0);
+	__ctl_store(cr2_old, 2, 2);
+	cr0_new = cr0_old;
+	cr2_new = cr2_old;
 	/* Take care of the enable/disable of transactional execution. */
 	if (MACHINE_HAS_TE) {
-		unsigned long cr, cr_new;
-
-		__ctl_store(cr, 0, 0);
 		/* Set or clear transaction execution TXC bit 8. */
-		cr_new = cr | (1UL << 55);
+		cr0_new |= (1UL << 55);
 		if (task->thread.per_flags & PER_FLAG_NO_TE)
-			cr_new &= ~(1UL << 55);
-		if (cr_new != cr)
-			__ctl_load(cr_new, 0, 0);
+			cr0_new &= ~(1UL << 55);
 		/* Set or clear transaction execution TDC bits 62 and 63. */
-		__ctl_store(cr, 2, 2);
-		cr_new = cr & ~3UL;
+		cr2_new &= ~3UL;
 		if (task->thread.per_flags & PER_FLAG_TE_ABORT_RAND) {
 			if (task->thread.per_flags & PER_FLAG_TE_ABORT_RAND_TEND)
-				cr_new |= 1UL;
+				cr2_new |= 1UL;
 			else
-				cr_new |= 2UL;
+				cr2_new |= 2UL;
 		}
-		if (cr_new != cr)
-			__ctl_load(cr_new, 2, 2);
 	}
+	/* Take care of enable/disable of guarded storage. */
+	if (MACHINE_HAS_GS) {
+		cr2_new &= ~(1UL << 4);
+		if (task->thread.gs_cb)
+			cr2_new |= (1UL << 4);
+	}
+	/* Load control register 0/2 iff changed */
+	cr0_changed = cr0_new != cr0_old;
+	cr2_changed = cr2_new != cr2_old;
+	if (cr0_changed)
+		__ctl_load(cr0_new, 0, 0);
+	if (cr2_changed)
+		__ctl_load(cr2_new, 2, 2);
 	/* Copy user specified PER registers */
 	new.control = thread->per_user.control;
 	new.start = thread->per_user.start;
@@ -1137,6 +1149,36 @@ static int s390_system_call_set(struct task_struct *target,
 				  data, 0, sizeof(unsigned int));
 }
 
+static int s390_gs_cb_get(struct task_struct *target,
+			  const struct user_regset *regset,
+			  unsigned int pos, unsigned int count,
+			  void *kbuf, void __user *ubuf)
+{
+	struct gs_cb *data = target->thread.gs_cb;
+
+	if (!MACHINE_HAS_GS)
+		return -ENODEV;
+	if (!data)
+		return -ENODATA;
+	return user_regset_copyout(&pos, &count, &kbuf, &ubuf,
+				   data, 0, sizeof(struct gs_cb));
+}
+
+static int s390_gs_cb_set(struct task_struct *target,
+			  const struct user_regset *regset,
+			  unsigned int pos, unsigned int count,
+			  const void *kbuf, const void __user *ubuf)
+{
+	struct gs_cb *data = target->thread.gs_cb;
+
+	if (!MACHINE_HAS_GS)
+		return -ENODEV;
+	if (!data)
+		return -ENODATA;
+	return user_regset_copyin(&pos, &count, &kbuf, &ubuf,
+				  data, 0, sizeof(struct gs_cb));
+}
+
 static const struct user_regset s390_regsets[] = {
 	{
 		.core_note_type = NT_PRSTATUS,
@@ -1194,6 +1236,14 @@ static const struct user_regset s390_regsets[] = {
 		.get = s390_vxrs_high_get,
 		.set = s390_vxrs_high_set,
 	},
+	{
+		.core_note_type = NT_S390_GS_CB,
+		.n = sizeof(struct gs_cb) / sizeof(__u64),
+		.size = sizeof(__u64),
+		.align = sizeof(__u64),
+		.get = s390_gs_cb_get,
+		.set = s390_gs_cb_set,
+	},
 };
 
 static const struct user_regset_view user_s390_view = {
@@ -1422,6 +1472,14 @@ static const struct user_regset s390_compat_regsets[] = {
 		.get = s390_compat_regs_high_get,
 		.set = s390_compat_regs_high_set,
 	},
+	{
+		.core_note_type = NT_S390_GS_CB,
+		.n = sizeof(struct gs_cb) / sizeof(__u64),
+		.size = sizeof(__u64),
+		.align = sizeof(__u64),
+		.get = s390_gs_cb_get,
+		.set = s390_gs_cb_set,
+	},
 };
 
 static const struct user_regset_view user_s390_compat_view = {

arch/s390/kernel/setup.c

@@ -339,9 +339,15 @@ static void __init setup_lowcore(void)
 	lc->stfl_fac_list = S390_lowcore.stfl_fac_list;
 	memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
 	       MAX_FACILITY_BIT/8);
-	if (MACHINE_HAS_VX)
-		lc->vector_save_area_addr =
-			(unsigned long) &lc->vector_save_area;
+	if (MACHINE_HAS_VX || MACHINE_HAS_GS) {
+		unsigned long bits, size;
+
+		bits = MACHINE_HAS_GS ? 11 : 10;
+		size = 1UL << bits;
+		lc->mcesad = (__u64) memblock_virt_alloc(size, size);
+		if (MACHINE_HAS_GS)
+			lc->mcesad |= bits;
+	}
 	lc->vdso_per_cpu_data = (unsigned long) &lc->paste[0];
 	lc->sync_enter_timer = S390_lowcore.sync_enter_timer;
 	lc->async_enter_timer = S390_lowcore.async_enter_timer;
@@ -779,6 +785,12 @@ static int __init setup_hwcaps(void)
 			elf_hwcap |= HWCAP_S390_VXRS_BCD;
 	}
 
+	/*
+	 * Guarded storage support HWCAP_S390_GS is bit 12.
+	 */
+	if (MACHINE_HAS_GS)
+		elf_hwcap |= HWCAP_S390_GS;
+
 	get_cpu_id(&cpu_id);
 	add_device_randomness(&cpu_id, sizeof(cpu_id));
 	switch (cpu_id.machine) {

arch/s390/kernel/smp.c

@@ -51,6 +51,7 @@
 #include <asm/os_info.h>
 #include <asm/sigp.h>
 #include <asm/idle.h>
+#include <asm/nmi.h>
 #include "entry.h"
 
 enum {
@@ -78,6 +79,8 @@ struct pcpu {
 static u8 boot_core_type;
 static struct pcpu pcpu_devices[NR_CPUS];
 
+static struct kmem_cache *pcpu_mcesa_cache;
+
 unsigned int smp_cpu_mt_shift;
 EXPORT_SYMBOL(smp_cpu_mt_shift);
 
@@ -188,8 +191,10 @@ static void pcpu_ec_call(struct pcpu *pcpu, int ec_bit)
 static int pcpu_alloc_lowcore(struct pcpu *pcpu, int cpu)
 {
 	unsigned long async_stack, panic_stack;
+	unsigned long mcesa_origin, mcesa_bits;
 	struct lowcore *lc;
 
+	mcesa_origin = mcesa_bits = 0;
 	if (pcpu != &pcpu_devices[0]) {
 		pcpu->lowcore =	(struct lowcore *)
 			__get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER);
@@ -197,20 +202,27 @@ static int pcpu_alloc_lowcore(struct pcpu *pcpu, int cpu)
 		panic_stack = __get_free_page(GFP_KERNEL);
 		if (!pcpu->lowcore || !panic_stack || !async_stack)
 			goto out;
+		if (MACHINE_HAS_VX || MACHINE_HAS_GS) {
+			mcesa_origin = (unsigned long)
+				kmem_cache_alloc(pcpu_mcesa_cache, GFP_KERNEL);
+			if (!mcesa_origin)
+				goto out;
+			mcesa_bits = MACHINE_HAS_GS ? 11 : 0;
+		}
 	} else {
 		async_stack = pcpu->lowcore->async_stack - ASYNC_FRAME_OFFSET;
 		panic_stack = pcpu->lowcore->panic_stack - PANIC_FRAME_OFFSET;
+		mcesa_origin = pcpu->lowcore->mcesad & MCESA_ORIGIN_MASK;
+		mcesa_bits = pcpu->lowcore->mcesad & MCESA_LC_MASK;
 	}
 	lc = pcpu->lowcore;
 	memcpy(lc, &S390_lowcore, 512);
 	memset((char *) lc + 512, 0, sizeof(*lc) - 512);
 	lc->async_stack = async_stack + ASYNC_FRAME_OFFSET;
 	lc->panic_stack = panic_stack + PANIC_FRAME_OFFSET;
+	lc->mcesad = mcesa_origin | mcesa_bits;
 	lc->cpu_nr = cpu;
 	lc->spinlock_lockval = arch_spin_lockval(cpu);
-	if (MACHINE_HAS_VX)
-		lc->vector_save_area_addr =
-			(unsigned long) &lc->vector_save_area;
 	if (vdso_alloc_per_cpu(lc))
 		goto out;
 	lowcore_ptr[cpu] = lc;
@@ -218,6 +230,9 @@ static int pcpu_alloc_lowcore(struct pcpu *pcpu, int cpu)
 	return 0;
 out:
 	if (pcpu != &pcpu_devices[0]) {
+		if (mcesa_origin)
+			kmem_cache_free(pcpu_mcesa_cache,
+					(void *) mcesa_origin);
 		free_page(panic_stack);
 		free_pages(async_stack, ASYNC_ORDER);
 		free_pages((unsigned long) pcpu->lowcore, LC_ORDER);
@@ -229,11 +244,17 @@ static int pcpu_alloc_lowcore(struct pcpu *pcpu, int cpu)
 
 static void pcpu_free_lowcore(struct pcpu *pcpu)
 {
+	unsigned long mcesa_origin;
+
 	pcpu_sigp_retry(pcpu, SIGP_SET_PREFIX, 0);
 	lowcore_ptr[pcpu - pcpu_devices] = NULL;
 	vdso_free_per_cpu(pcpu->lowcore);
 	if (pcpu == &pcpu_devices[0])
 		return;
+	if (MACHINE_HAS_VX || MACHINE_HAS_GS) {
+		mcesa_origin = pcpu->lowcore->mcesad & MCESA_ORIGIN_MASK;
+		kmem_cache_free(pcpu_mcesa_cache, (void *) mcesa_origin);
+	}
 	free_page(pcpu->lowcore->panic_stack-PANIC_FRAME_OFFSET);
 	free_pages(pcpu->lowcore->async_stack-ASYNC_FRAME_OFFSET, ASYNC_ORDER);
 	free_pages((unsigned long) pcpu->lowcore, LC_ORDER);
@@ -550,9 +571,11 @@ 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)
+	if (!MACHINE_HAS_VX && !MACHINE_HAS_GS)
 		return 0;
-	pa = __pa(pcpu->lowcore->vector_save_area_addr);
+	pa = __pa(pcpu->lowcore->mcesad & MCESA_ORIGIN_MASK);
+	if (MACHINE_HAS_GS)
+		pa |= pcpu->lowcore->mcesad & MCESA_LC_MASK;
 	if (__pcpu_sigp_relax(pcpu->address, SIGP_STORE_ADDITIONAL_STATUS,
 			      pa) != SIGP_CC_ORDER_CODE_ACCEPTED)
 		return -EIO;
@@ -897,12 +920,22 @@ void __init smp_fill_possible_mask(void)
 
 void __init smp_prepare_cpus(unsigned int max_cpus)
 {
+	unsigned long size;
+
 	/* request the 0x1201 emergency signal external interrupt */
 	if (register_external_irq(EXT_IRQ_EMERGENCY_SIG, do_ext_call_interrupt))
 		panic("Couldn't request external interrupt 0x1201");
 	/* request the 0x1202 external call external interrupt */
 	if (register_external_irq(EXT_IRQ_EXTERNAL_CALL, do_ext_call_interrupt))
 		panic("Couldn't request external interrupt 0x1202");
+	/* create slab cache for the machine-check-extended-save-areas */
+	if (MACHINE_HAS_VX || MACHINE_HAS_GS) {
+		size = 1UL << (MACHINE_HAS_GS ? 11 : 10);
+		pcpu_mcesa_cache = kmem_cache_create("nmi_save_areas",
+						     size, size, 0, NULL);
+		if (!pcpu_mcesa_cache)
+			panic("Couldn't create nmi save area cache");
+	}
 }
 
 void __init smp_prepare_boot_cpu(void)

arch/s390/kernel/syscalls.S

@@ -386,5 +386,5 @@ SYSCALL(sys_mlock2,compat_sys_mlock2)
 SYSCALL(sys_copy_file_range,compat_sys_copy_file_range) /* 375 */
 SYSCALL(sys_preadv2,compat_sys_preadv2)
 SYSCALL(sys_pwritev2,compat_sys_pwritev2)
-NI_SYSCALL
+SYSCALL(sys_s390_guarded_storage,compat_sys_s390_guarded_storage) /* 378 */
 SYSCALL(sys_statx,compat_sys_statx)

arch/s390/kvm/interrupt.c

@@ -420,8 +420,8 @@ static int __write_machine_check(struct kvm_vcpu *vcpu,
 	save_access_regs(vcpu->run->s.regs.acrs);
 
 	/* Extended save area */
-	rc = read_guest_lc(vcpu, __LC_VX_SAVE_AREA_ADDR, &ext_sa_addr,
-			    sizeof(unsigned long));
+	rc = read_guest_lc(vcpu, __LC_MCESAD, &ext_sa_addr,
+			   sizeof(unsigned long));
 	/* Only bits 0-53 are used for address formation */
 	ext_sa_addr &= ~0x3ffUL;
 	if (!rc && mci.vr && ext_sa_addr && test_kvm_facility(vcpu->kvm, 129)) {

include/uapi/linux/elf.h

@@ -409,6 +409,7 @@ typedef struct elf64_shdr {
 #define NT_S390_TDB	0x308		/* s390 transaction diagnostic block */
 #define NT_S390_VXRS_LOW	0x309	/* s390 vector registers 0-15 upper half */
 #define NT_S390_VXRS_HIGH	0x30a	/* s390 vector registers 16-31 */
+#define NT_S390_GS_CB	0x30b		/* s390 guarded storage registers */
 #define NT_ARM_VFP	0x400		/* ARM VFP/NEON registers */
 #define NT_ARM_TLS	0x401		/* ARM TLS register */
 #define NT_ARM_HW_BREAK	0x402		/* ARM hardware breakpoint registers */