Merge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6: (65 commits)
  [S390] prevent unneccesary loops_per_jiffy recalculation
  [S390] cpuinfo: use get_online_cpus() instead of preempt_disable()
  [S390] smp: remove cpu hotplug messages
  [S390] mutex: enable spinning mutex on s390
  [S390] mutex: Introduce arch_mutex_cpu_relax()
  [S390] cio: fix ccwgroup unregistration race condition
  [S390] perf: add DWARF register lookup for s390
  [S390] cleanup ftrace backend functions
  [S390] ptrace cleanup
  [S390] smp/idle: call init_idle() before starting a new cpu
  [S390] smp: delay idle task creation
  [S390] dasd: Correct retry counter for terminated I/O.
  [S390] dasd: Add support for raw ECKD access.
  [S390] dasd: Prevent deadlock during suspend/resume.
  [S390] dasd: Improve handling of stolen DASD reservation
  [S390] dasd: do path verification for paths added at runtime
  [S390] dasd: add High Performance FICON multitrack support
  [S390] cio: reduce memory consumption of itcw structures
  [S390] nmi: enable machine checks early
  [S390] qeth: buffer count imbalance
  ...

arch/s390/Kconfig

@@ -1,13 +1,8 @@
-config SCHED_MC
-	def_bool y
-	depends on SMP
-
 config MMU
 	def_bool y
 
 config ZONE_DMA
-	def_bool y
-	depends on 64BIT
+	def_bool y if 64BIT
 
 config LOCKDEP_SUPPORT
 	def_bool y
@@ -25,12 +20,10 @@ config RWSEM_XCHGADD_ALGORITHM
 	def_bool y
 
 config ARCH_HAS_ILOG2_U32
-	bool
-	default n
+	def_bool n
 
 config ARCH_HAS_ILOG2_U64
-	bool
-	default n
+	def_bool n
 
 config GENERIC_HWEIGHT
 	def_bool y
@@ -42,9 +35,7 @@ config GENERIC_CLOCKEVENTS
 	def_bool y
 
 config GENERIC_BUG
-	bool
-	depends on BUG
-	default y
+	def_bool y if BUG
 
 config GENERIC_BUG_RELATIVE_POINTERS
 	def_bool y
@@ -59,13 +50,10 @@ config ARCH_DMA_ADDR_T_64BIT
 	def_bool 64BIT
 
 config GENERIC_LOCKBREAK
-	bool
-	default y
-	depends on SMP && PREEMPT
+	def_bool y if SMP && PREEMPT
 
 config PGSTE
-	bool
-	default y if KVM
+	def_bool y if KVM
 
 config VIRT_CPU_ACCOUNTING
 	def_bool y
@@ -85,7 +73,6 @@ config S390
 	select HAVE_DYNAMIC_FTRACE
 	select HAVE_FUNCTION_GRAPH_TRACER
 	select HAVE_REGS_AND_STACK_ACCESS_API
-	select HAVE_DEFAULT_NO_SPIN_MUTEXES
 	select HAVE_OPROFILE
 	select HAVE_KPROBES
 	select HAVE_KRETPROBES
@@ -130,8 +117,7 @@ config S390
 	select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE
 
 config SCHED_OMIT_FRAME_POINTER
-	bool
-	default y
+	def_bool y
 
 source "init/Kconfig"
 
@@ -144,20 +130,21 @@ comment "Processor type and features"
 source "kernel/time/Kconfig"
 
 config 64BIT
-	bool "64 bit kernel"
+	def_bool y
+	prompt "64 bit kernel"
 	help
 	  Select this option if you have an IBM z/Architecture machine
 	  and want to use the 64 bit addressing mode.
 
 config 32BIT
-	bool
-	default y if !64BIT
+	def_bool y if !64BIT
 
 config KTIME_SCALAR
 	def_bool 32BIT
 
 config SMP
-	bool "Symmetric multi-processing support"
+	def_bool y
+	prompt "Symmetric multi-processing support"
 	---help---
 	  This enables support for systems with more than one CPU. If you have
 	  a system with only one CPU, like most personal computers, say N. If
@@ -189,10 +176,10 @@ config NR_CPUS
 	  approximately sixteen kilobytes to the kernel image.
 
 config HOTPLUG_CPU
-	bool "Support for hot-pluggable CPUs"
+	def_bool y
+	prompt "Support for hot-pluggable CPUs"
 	depends on SMP
 	select HOTPLUG
-	default n
 	help
 	  Say Y here to be able to turn CPUs off and on. CPUs
 	  can be controlled through /sys/devices/system/cpu/cpu#.
@@ -208,14 +195,16 @@ config SCHED_MC
 	  increased overhead in some places.
 
 config SCHED_BOOK
-	bool "Book scheduler support"
+	def_bool y
+	prompt "Book scheduler support"
 	depends on SMP && SCHED_MC
 	help
 	  Book scheduler support improves the CPU scheduler's decision making
 	  when dealing with machines that have several books.
 
 config MATHEMU
-	bool "IEEE FPU emulation"
+	def_bool y
+	prompt "IEEE FPU emulation"
 	depends on MARCH_G5
 	help
 	  This option is required for IEEE compliant floating point arithmetic
@@ -223,7 +212,8 @@ config MATHEMU
 	  need this.
 
 config COMPAT
-	bool "Kernel support for 31 bit emulation"
+	def_bool y
+	prompt "Kernel support for 31 bit emulation"
 	depends on 64BIT
 	select COMPAT_BINFMT_ELF
 	help
@@ -233,16 +223,14 @@ config COMPAT
 	  executing 31 bit applications.  It is safe to say "Y".
 
 config SYSVIPC_COMPAT
-	bool
-	depends on COMPAT && SYSVIPC
-	default y
+	def_bool y if COMPAT && SYSVIPC
 
 config AUDIT_ARCH
-	bool
-	default y
+	def_bool y
 
 config S390_EXEC_PROTECT
-	bool "Data execute protection"
+	def_bool y
+	prompt "Data execute protection"
 	help
 	  This option allows to enable a buffer overflow protection for user
 	  space programs and it also selects the addressing mode option above.
@@ -302,7 +290,8 @@ config MARCH_Z196
 endchoice
 
 config PACK_STACK
-	bool "Pack kernel stack"
+	def_bool y
+	prompt "Pack kernel stack"
 	help
 	  This option enables the compiler option -mkernel-backchain if it
 	  is available. If the option is available the compiler supports
@@ -315,7 +304,8 @@ config PACK_STACK
 	  Say Y if you are unsure.
 
 config SMALL_STACK
-	bool "Use 8kb for kernel stack instead of 16kb"
+	def_bool n
+	prompt "Use 8kb for kernel stack instead of 16kb"
 	depends on PACK_STACK && 64BIT && !LOCKDEP
 	help
 	  If you say Y here and the compiler supports the -mkernel-backchain
@@ -327,7 +317,8 @@ config SMALL_STACK
 	  Say N if you are unsure.
 
 config CHECK_STACK
-	bool "Detect kernel stack overflow"
+	def_bool y
+	prompt "Detect kernel stack overflow"
 	help
 	  This option enables the compiler option -mstack-guard and
 	  -mstack-size if they are available. If the compiler supports them
@@ -351,7 +342,8 @@ config STACK_GUARD
 	  512 for 64 bit.
 
 config WARN_STACK
-	bool "Emit compiler warnings for function with broken stack usage"
+	def_bool n
+	prompt "Emit compiler warnings for function with broken stack usage"
 	help
 	  This option enables the compiler options -mwarn-framesize and
 	  -mwarn-dynamicstack. If the compiler supports these options it
@@ -386,24 +378,24 @@ config ARCH_SPARSEMEM_DEFAULT
 	def_bool y
 
 config ARCH_SELECT_MEMORY_MODEL
-       def_bool y
+	def_bool y
 
 config ARCH_ENABLE_MEMORY_HOTPLUG
-	def_bool y
-	depends on SPARSEMEM
+	def_bool y if SPARSEMEM
 
 config ARCH_ENABLE_MEMORY_HOTREMOVE
 	def_bool y
 
 config ARCH_HIBERNATION_POSSIBLE
-       def_bool y if 64BIT
+	def_bool y if 64BIT
 
 source "mm/Kconfig"
 
 comment "I/O subsystem configuration"
 
 config QDIO
-	tristate "QDIO support"
+	def_tristate y
+	prompt "QDIO support"
 	---help---
 	  This driver provides the Queued Direct I/O base support for
 	  IBM System z.
@@ -414,7 +406,8 @@ config QDIO
 	  If unsure, say Y.
 
 config CHSC_SCH
-	tristate "Support for CHSC subchannels"
+	def_tristate y
+	prompt "Support for CHSC subchannels"
 	help
 	  This driver allows usage of CHSC subchannels. A CHSC subchannel
 	  is usually present on LPAR only.
@@ -432,7 +425,8 @@ config CHSC_SCH
 comment "Misc"
 
 config IPL
-	bool "Builtin IPL record support"
+	def_bool y
+	prompt "Builtin IPL record support"
 	help
 	  If you want to use the produced kernel to IPL directly from a
 	  device, you have to merge a bootsector specific to the device
@@ -464,7 +458,8 @@ config FORCE_MAX_ZONEORDER
 	default "9"
 
 config PFAULT
-	bool "Pseudo page fault support"
+	def_bool y
+	prompt "Pseudo page fault support"
 	help
 	  Select this option, if you want to use PFAULT pseudo page fault
 	  handling under VM. If running native or in LPAR, this option
@@ -476,7 +471,8 @@ config PFAULT
 	  this option.
 
 config SHARED_KERNEL
-	bool "VM shared kernel support"
+	def_bool y
+	prompt "VM shared kernel support"
 	help
 	  Select this option, if you want to share the text segment of the
 	  Linux kernel between different VM guests. This reduces memory
@@ -487,7 +483,8 @@ config SHARED_KERNEL
 	  doing and want to exploit this feature.
 
 config CMM
-	tristate "Cooperative memory management"
+	def_tristate n
+	prompt "Cooperative memory management"
 	help
 	  Select this option, if you want to enable the kernel interface
 	  to reduce the memory size of the system. This is accomplished
@@ -499,14 +496,16 @@ config CMM
 	  option.
 
 config CMM_IUCV
-	bool "IUCV special message interface to cooperative memory management"
+	def_bool y
+	prompt "IUCV special message interface to cooperative memory management"
 	depends on CMM && (SMSGIUCV=y || CMM=SMSGIUCV)
 	help
 	  Select this option to enable the special message interface to
 	  the cooperative memory management.
 
 config APPLDATA_BASE
-	bool "Linux - VM Monitor Stream, base infrastructure"
+	def_bool n
+	prompt "Linux - VM Monitor Stream, base infrastructure"
 	depends on PROC_FS
 	help
 	  This provides a kernel interface for creating and updating z/VM APPLDATA
@@ -521,7 +520,8 @@ config APPLDATA_BASE
 	  The /proc entries can also be read from, showing the current settings.
 
 config APPLDATA_MEM
-	tristate "Monitor memory management statistics"
+	def_tristate m
+	prompt "Monitor memory management statistics"
 	depends on APPLDATA_BASE && VM_EVENT_COUNTERS
 	help
 	  This provides memory management related data to the Linux - VM Monitor
@@ -537,7 +537,8 @@ config APPLDATA_MEM
 	  appldata_mem.o.
 
 config APPLDATA_OS
-	tristate "Monitor OS statistics"
+	def_tristate m
+	prompt "Monitor OS statistics"
 	depends on APPLDATA_BASE
 	help
 	  This provides OS related data to the Linux - VM Monitor Stream, like
@@ -551,7 +552,8 @@ config APPLDATA_OS
 	  appldata_os.o.
 
 config APPLDATA_NET_SUM
-	tristate "Monitor overall network statistics"
+	def_tristate m
+	prompt "Monitor overall network statistics"
 	depends on APPLDATA_BASE && NET
 	help
 	  This provides network related data to the Linux - VM Monitor Stream,
@@ -568,30 +570,32 @@ config APPLDATA_NET_SUM
 source kernel/Kconfig.hz
 
 config S390_HYPFS_FS
-	bool "s390 hypervisor file system support"
+	def_bool y
+	prompt "s390 hypervisor file system support"
 	select SYS_HYPERVISOR
-	default y
 	help
 	  This is a virtual file system intended to provide accounting
 	  information in an s390 hypervisor environment.
 
 config KEXEC
-	bool "kexec system call"
+	def_bool n
+	prompt "kexec system call"
 	help
 	  kexec is a system call that implements the ability to shutdown your
 	  current kernel, and to start another kernel.  It is like a reboot
 	  but is independent of hardware/microcode support.
 
 config ZFCPDUMP
-	bool "zfcpdump support"
+	def_bool n
+	prompt "zfcpdump support"
 	select SMP
-	default n
 	help
 	  Select this option if you want to build an zfcpdump enabled kernel.
 	  Refer to <file:Documentation/s390/zfcpdump.txt> for more details on this.
 
 config S390_GUEST
-bool "s390 guest support for KVM (EXPERIMENTAL)"
+	def_bool y
+	prompt "s390 guest support for KVM (EXPERIMENTAL)"
 	depends on 64BIT && EXPERIMENTAL
 	select VIRTIO
 	select VIRTIO_RING
@@ -603,9 +607,9 @@ bool "s390 guest support for KVM (EXPERIMENTAL)"
 	  the default console.
 
 config SECCOMP
-	bool "Enable seccomp to safely compute untrusted bytecode"
+	def_bool y
+	prompt "Enable seccomp to safely compute untrusted bytecode"
 	depends on PROC_FS
-	default y
 	help
 	  This kernel feature is useful for number crunching applications
 	  that may need to compute untrusted bytecode during their

arch/s390/Kconfig.debug

@@ -1,8 +1,7 @@
 menu "Kernel hacking"
 
 config TRACE_IRQFLAGS_SUPPORT
-	bool
-	default y
+	def_bool y
 
 source "lib/Kconfig.debug"
 
@@ -19,7 +18,8 @@ config STRICT_DEVMEM
 	  If you are unsure, say Y.
 
 config DEBUG_STRICT_USER_COPY_CHECKS
-	bool "Strict user copy size checks"
+	def_bool n
+	prompt "Strict user copy size checks"
 	---help---
 	  Enabling this option turns a certain set of sanity checks for user
 	  copy operations into compile time warnings.

arch/s390/defconfig

@@ -2,16 +2,12 @@ CONFIG_EXPERIMENTAL=y
 CONFIG_SYSVIPC=y
 CONFIG_POSIX_MQUEUE=y
 CONFIG_AUDIT=y
+CONFIG_RCU_TRACE=y
 CONFIG_IKCONFIG=y
 CONFIG_IKCONFIG_PROC=y
-CONFIG_CGROUPS=y
-CONFIG_CGROUP_NS=y
-CONFIG_SYSFS_DEPRECATED_V2=y
-CONFIG_UTS_NS=y
-CONFIG_IPC_NS=y
 CONFIG_BLK_DEV_INITRD=y
 # CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
-# CONFIG_COMPAT_BRK is not set
+CONFIG_PERF_EVENTS=y
 CONFIG_SLAB=y
 CONFIG_KPROBES=y
 CONFIG_MODULES=y
@@ -20,24 +16,12 @@ CONFIG_MODVERSIONS=y
 CONFIG_DEFAULT_DEADLINE=y
 CONFIG_NO_HZ=y
 CONFIG_HIGH_RES_TIMERS=y
-CONFIG_64BIT=y
-CONFIG_SMP=y
-CONFIG_NR_CPUS=32
-CONFIG_COMPAT=y
-CONFIG_S390_EXEC_PROTECT=y
-CONFIG_PACK_STACK=y
-CONFIG_CHECK_STACK=y
 CONFIG_PREEMPT=y
 CONFIG_MEMORY_HOTPLUG=y
 CONFIG_MEMORY_HOTREMOVE=y
-CONFIG_QDIO=y
-CONFIG_CHSC_SCH=m
-CONFIG_IPL=y
 CONFIG_BINFMT_MISC=m
-CONFIG_PFAULT=y
 CONFIG_HZ_100=y
 CONFIG_KEXEC=y
-CONFIG_S390_GUEST=y
 CONFIG_PM=y
 CONFIG_HIBERNATION=y
 CONFIG_PACKET=y
@@ -46,16 +30,15 @@ CONFIG_NET_KEY=y
 CONFIG_AFIUCV=m
 CONFIG_INET=y
 CONFIG_IP_MULTICAST=y
+# CONFIG_INET_LRO is not set
 CONFIG_IPV6=y
-CONFIG_NETFILTER=y
-CONFIG_NETFILTER_NETLINK_QUEUE=m
-CONFIG_NETFILTER_NETLINK_LOG=m
-CONFIG_NF_CONNTRACK=m
-# CONFIG_NF_CT_PROTO_SCTP is not set
+CONFIG_NET_SCTPPROBE=m
+CONFIG_L2TP=m
+CONFIG_L2TP_DEBUGFS=m
+CONFIG_VLAN_8021Q=y
 CONFIG_NET_SCHED=y
 CONFIG_NET_SCH_CBQ=m
 CONFIG_NET_SCH_PRIO=m
-CONFIG_NET_SCH_MULTIQ=y
 CONFIG_NET_SCH_RED=m
 CONFIG_NET_SCH_SFQ=m
 CONFIG_NET_SCH_TEQL=m
@@ -69,28 +52,14 @@ CONFIG_NET_CLS_U32=m
 CONFIG_CLS_U32_MARK=y
 CONFIG_NET_CLS_RSVP=m
 CONFIG_NET_CLS_RSVP6=m
-CONFIG_NET_CLS_FLOW=m
 CONFIG_NET_CLS_ACT=y
 CONFIG_NET_ACT_POLICE=y
-CONFIG_NET_ACT_NAT=m
-CONFIG_CAN=m
-CONFIG_CAN_RAW=m
-CONFIG_CAN_BCM=m
-CONFIG_CAN_VCAN=m
 CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
 # CONFIG_FIRMWARE_IN_KERNEL is not set
 CONFIG_BLK_DEV_LOOP=m
 CONFIG_BLK_DEV_NBD=m
 CONFIG_BLK_DEV_RAM=y
-CONFIG_BLK_DEV_XIP=y
-CONFIG_BLK_DEV_XPRAM=m
-CONFIG_DASD=y
-CONFIG_DASD_PROFILE=y
-CONFIG_DASD_ECKD=y
-CONFIG_DASD_FBA=y
-CONFIG_DASD_DIAG=y
-CONFIG_DASD_EER=y
-CONFIG_VIRTIO_BLK=m
+CONFIG_VIRTIO_BLK=y
 CONFIG_SCSI=y
 CONFIG_BLK_DEV_SD=y
 CONFIG_CHR_DEV_ST=y
@@ -102,101 +71,92 @@ CONFIG_SCSI_CONSTANTS=y
 CONFIG_SCSI_LOGGING=y
 CONFIG_SCSI_SCAN_ASYNC=y
 CONFIG_ZFCP=y
-CONFIG_SCSI_DH=m
-CONFIG_SCSI_DH_RDAC=m
-CONFIG_SCSI_DH_HP_SW=m
-CONFIG_SCSI_DH_EMC=m
-CONFIG_SCSI_DH_ALUA=m
-CONFIG_SCSI_OSD_INITIATOR=m
-CONFIG_SCSI_OSD_ULD=m
-CONFIG_MD=y
-CONFIG_BLK_DEV_MD=y
-CONFIG_MD_LINEAR=m
-CONFIG_MD_RAID0=m
-CONFIG_MD_RAID1=m
-CONFIG_MD_MULTIPATH=m
-CONFIG_BLK_DEV_DM=y
-CONFIG_DM_CRYPT=y
-CONFIG_DM_SNAPSHOT=y
-CONFIG_DM_MIRROR=y
-CONFIG_DM_ZERO=y
-CONFIG_DM_MULTIPATH=m
+CONFIG_ZFCP_DIF=y
 CONFIG_NETDEVICES=y
 CONFIG_DUMMY=m
 CONFIG_BONDING=m
 CONFIG_EQUALIZER=m
 CONFIG_TUN=m
-CONFIG_VETH=m
 CONFIG_NET_ETHERNET=y
-CONFIG_LCS=m
-CONFIG_CTCM=m
-CONFIG_QETH=y
-CONFIG_QETH_L2=y
-CONFIG_QETH_L3=y
-CONFIG_VIRTIO_NET=m
-CONFIG_HW_RANDOM_VIRTIO=m
+CONFIG_VIRTIO_NET=y
 CONFIG_RAW_DRIVER=m
-CONFIG_TN3270=y
-CONFIG_TN3270_TTY=y
-CONFIG_TN3270_FS=m
-CONFIG_TN3270_CONSOLE=y
-CONFIG_TN3215=y
-CONFIG_TN3215_CONSOLE=y
-CONFIG_SCLP_TTY=y
-CONFIG_SCLP_CONSOLE=y
-CONFIG_SCLP_VT220_TTY=y
-CONFIG_SCLP_VT220_CONSOLE=y
-CONFIG_SCLP_CPI=m
-CONFIG_SCLP_ASYNC=m
-CONFIG_S390_TAPE=m
-CONFIG_S390_TAPE_BLOCK=y
-CONFIG_S390_TAPE_34XX=m
-CONFIG_ACCESSIBILITY=y
 CONFIG_EXT2_FS=y
 CONFIG_EXT3_FS=y
 # CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set
+CONFIG_EXT4_FS=y
+CONFIG_EXT4_FS_POSIX_ACL=y
+CONFIG_EXT4_FS_SECURITY=y
 CONFIG_PROC_KCORE=y
 CONFIG_TMPFS=y
 CONFIG_TMPFS_POSIX_ACL=y
-CONFIG_NFS_FS=y
-CONFIG_NFS_V3=y
-CONFIG_NFSD=y
-CONFIG_NFSD_V3=y
+# CONFIG_NETWORK_FILESYSTEMS is not set
 CONFIG_PARTITION_ADVANCED=y
 CONFIG_IBM_PARTITION=y
 CONFIG_DLM=m
 CONFIG_MAGIC_SYSRQ=y
 CONFIG_DEBUG_KERNEL=y
-# CONFIG_SCHED_DEBUG is not set
-CONFIG_DEBUG_SPINLOCK=y
-CONFIG_DEBUG_MUTEXES=y
+CONFIG_TIMER_STATS=y
+CONFIG_PROVE_LOCKING=y
+CONFIG_PROVE_RCU=y
+CONFIG_LOCK_STAT=y
+CONFIG_DEBUG_LOCKDEP=y
 CONFIG_DEBUG_SPINLOCK_SLEEP=y
+CONFIG_DEBUG_LIST=y
+CONFIG_DEBUG_NOTIFIERS=y
 # CONFIG_RCU_CPU_STALL_DETECTOR is not set
-CONFIG_DEBUG_FORCE_WEAK_PER_CPU=y
+CONFIG_KPROBES_SANITY_TEST=y
+CONFIG_CPU_NOTIFIER_ERROR_INJECT=m
+CONFIG_LATENCYTOP=y
 CONFIG_SYSCTL_SYSCALL_CHECK=y
-CONFIG_SAMPLES=y
-CONFIG_CRYPTO_FIPS=y
+CONFIG_DEBUG_PAGEALLOC=y
+# CONFIG_FTRACE is not set
+# CONFIG_STRICT_DEVMEM is not set
+CONFIG_CRYPTO_NULL=m
+CONFIG_CRYPTO_CRYPTD=m
 CONFIG_CRYPTO_AUTHENC=m
+CONFIG_CRYPTO_TEST=m
 CONFIG_CRYPTO_CCM=m
 CONFIG_CRYPTO_GCM=m
+CONFIG_CRYPTO_CBC=y
 CONFIG_CRYPTO_CTS=m
 CONFIG_CRYPTO_ECB=m
+CONFIG_CRYPTO_LRW=m
 CONFIG_CRYPTO_PCBC=m
+CONFIG_CRYPTO_XTS=m
+CONFIG_CRYPTO_XCBC=m
 CONFIG_CRYPTO_VMAC=m
+CONFIG_CRYPTO_MD4=m
+CONFIG_CRYPTO_MICHAEL_MIC=m
 CONFIG_CRYPTO_RMD128=m
 CONFIG_CRYPTO_RMD160=m
 CONFIG_CRYPTO_RMD256=m
 CONFIG_CRYPTO_RMD320=m
+CONFIG_CRYPTO_SHA256=m
+CONFIG_CRYPTO_SHA512=m
+CONFIG_CRYPTO_TGR192=m
+CONFIG_CRYPTO_WP512=m
+CONFIG_CRYPTO_ANUBIS=m
+CONFIG_CRYPTO_ARC4=m
+CONFIG_CRYPTO_BLOWFISH=m
 CONFIG_CRYPTO_CAMELLIA=m
+CONFIG_CRYPTO_CAST5=m
+CONFIG_CRYPTO_CAST6=m
+CONFIG_CRYPTO_DES=m
 CONFIG_CRYPTO_FCRYPT=m
+CONFIG_CRYPTO_KHAZAD=m
 CONFIG_CRYPTO_SALSA20=m
 CONFIG_CRYPTO_SEED=m
+CONFIG_CRYPTO_SERPENT=m
+CONFIG_CRYPTO_TEA=m
+CONFIG_CRYPTO_TWOFISH=m
+CONFIG_CRYPTO_DEFLATE=m
 CONFIG_CRYPTO_ZLIB=m
 CONFIG_CRYPTO_LZO=m
 CONFIG_ZCRYPT=m
+CONFIG_CRYPTO_SHA1_S390=m
+CONFIG_CRYPTO_SHA256_S390=m
 CONFIG_CRYPTO_SHA512_S390=m
-CONFIG_CRC_T10DIF=y
-CONFIG_CRC32=m
+CONFIG_CRYPTO_DES_S390=m
+CONFIG_CRYPTO_AES_S390=m
 CONFIG_CRC7=m
-CONFIG_KVM=m
-CONFIG_VIRTIO_BALLOON=m
+CONFIG_VIRTIO_BALLOON=y

arch/s390/hypfs/Makefile

@@ -4,4 +4,4 @@
 
 obj-$(CONFIG_S390_HYPFS_FS) += s390_hypfs.o
 
-s390_hypfs-objs := inode.o hypfs_diag.o hypfs_vm.o
+s390_hypfs-objs := inode.o hypfs_diag.o hypfs_vm.o hypfs_dbfs.o

arch/s390/hypfs/hypfs.h

@@ -12,6 +12,8 @@
 #include <linux/fs.h>
 #include <linux/types.h>
 #include <linux/debugfs.h>
+#include <linux/workqueue.h>
+#include <linux/kref.h>
 
 #define REG_FILE_MODE    0440
 #define UPDATE_FILE_MODE 0220
@@ -38,6 +40,33 @@ extern int hypfs_vm_init(void);
 extern void hypfs_vm_exit(void);
 extern int hypfs_vm_create_files(struct super_block *sb, struct dentry *root);
 
-/* Directory for debugfs files */
-extern struct dentry *hypfs_dbfs_dir;
+/* debugfs interface */
+struct hypfs_dbfs_file;
+
+struct hypfs_dbfs_data {
+	void			*buf;
+	void			*buf_free_ptr;
+	size_t			size;
+	struct hypfs_dbfs_file	*dbfs_file;;
+	struct kref		kref;
+};
+
+struct hypfs_dbfs_file {
+	const char	*name;
+	int		(*data_create)(void **data, void **data_free_ptr,
+				       size_t *size);
+	void		(*data_free)(const void *buf_free_ptr);
+
+	/* Private data for hypfs_dbfs.c */
+	struct hypfs_dbfs_data	*data;
+	struct delayed_work	data_free_work;
+	struct mutex		lock;
+	struct dentry		*dentry;
+};
+
+extern int hypfs_dbfs_init(void);
+extern void hypfs_dbfs_exit(void);
+extern int hypfs_dbfs_create_file(struct hypfs_dbfs_file *df);
+extern void hypfs_dbfs_remove_file(struct hypfs_dbfs_file *df);
+
 #endif /* _HYPFS_H_ */

arch/s390/hypfs/hypfs_dbfs.c

@@ -0,0 +1,116 @@
+/*
+ * Hypervisor filesystem for Linux on s390 - debugfs interface
+ *
+ * Copyright (C) IBM Corp. 2010
+ * Author(s): Michael Holzheu <holzheu@linux.vnet.ibm.com>
+ */
+
+#include <linux/slab.h>
+#include "hypfs.h"
+
+static struct dentry *dbfs_dir;
+
+static struct hypfs_dbfs_data *hypfs_dbfs_data_alloc(struct hypfs_dbfs_file *f)
+{
+	struct hypfs_dbfs_data *data;
+
+	data = kmalloc(sizeof(*data), GFP_KERNEL);
+	if (!data)
+		return NULL;
+	kref_init(&data->kref);
+	data->dbfs_file = f;
+	return data;
+}
+
+static void hypfs_dbfs_data_free(struct kref *kref)
+{
+	struct hypfs_dbfs_data *data;
+
+	data = container_of(kref, struct hypfs_dbfs_data, kref);
+	data->dbfs_file->data_free(data->buf_free_ptr);
+	kfree(data);
+}
+
+static void data_free_delayed(struct work_struct *work)
+{
+	struct hypfs_dbfs_data *data;
+	struct hypfs_dbfs_file *df;
+
+	df = container_of(work, struct hypfs_dbfs_file, data_free_work.work);
+	mutex_lock(&df->lock);
+	data = df->data;
+	df->data = NULL;
+	mutex_unlock(&df->lock);
+	kref_put(&data->kref, hypfs_dbfs_data_free);
+}
+
+static ssize_t dbfs_read(struct file *file, char __user *buf,
+			 size_t size, loff_t *ppos)
+{
+	struct hypfs_dbfs_data *data;
+	struct hypfs_dbfs_file *df;
+	ssize_t rc;
+
+	if (*ppos != 0)
+		return 0;
+
+	df = file->f_path.dentry->d_inode->i_private;
+	mutex_lock(&df->lock);
+	if (!df->data) {
+		data = hypfs_dbfs_data_alloc(df);
+		if (!data) {
+			mutex_unlock(&df->lock);
+			return -ENOMEM;
+		}
+		rc = df->data_create(&data->buf, &data->buf_free_ptr,
+				     &data->size);
+		if (rc) {
+			mutex_unlock(&df->lock);
+			kfree(data);
+			return rc;
+		}
+		df->data = data;
+		schedule_delayed_work(&df->data_free_work, HZ);
+	}
+	data = df->data;
+	kref_get(&data->kref);
+	mutex_unlock(&df->lock);
+
+	rc = simple_read_from_buffer(buf, size, ppos, data->buf, data->size);
+	kref_put(&data->kref, hypfs_dbfs_data_free);
+	return rc;
+}
+
+static const struct file_operations dbfs_ops = {
+	.read		= dbfs_read,
+	.llseek		= no_llseek,
+};
+
+int hypfs_dbfs_create_file(struct hypfs_dbfs_file *df)
+{
+	df->dentry = debugfs_create_file(df->name, 0400, dbfs_dir, df,
+					 &dbfs_ops);
+	if (IS_ERR(df->dentry))
+		return PTR_ERR(df->dentry);
+	mutex_init(&df->lock);
+	INIT_DELAYED_WORK(&df->data_free_work, data_free_delayed);
+	return 0;
+}
+
+void hypfs_dbfs_remove_file(struct hypfs_dbfs_file *df)
+{
+	debugfs_remove(df->dentry);
+}
+
+int hypfs_dbfs_init(void)
+{
+	dbfs_dir = debugfs_create_dir("s390_hypfs", NULL);
+	if (IS_ERR(dbfs_dir))
+		return PTR_ERR(dbfs_dir);
+	return 0;
+}
+
+void hypfs_dbfs_exit(void)
+{
+	debugfs_remove(dbfs_dir);
+}

arch/s390/hypfs/hypfs_diag.c

@@ -555,81 +555,38 @@ struct dbfs_d204 {
 	char			buf[];	/* d204 buffer */
 } __attribute__ ((packed));
 
-struct dbfs_d204_private {
-	struct dbfs_d204	*d204;	/* Aligned d204 data with header */
-	void			*base;	/* Base pointer (needed for vfree) */
-};
-
-static int dbfs_d204_open(struct inode *inode, struct file *file)
+static int dbfs_d204_create(void **data, void **data_free_ptr, size_t *size)
 {
-	struct dbfs_d204_private *data;
 	struct dbfs_d204 *d204;
 	int rc, buf_size;
+	void *base;
 
-	data = kzalloc(sizeof(*data), GFP_KERNEL);
-	if (!data)
-		return -ENOMEM;
 	buf_size = PAGE_SIZE * (diag204_buf_pages + 1) + sizeof(d204->hdr);
-	data->base = vmalloc(buf_size);
-	if (!data->base) {
-		rc = -ENOMEM;
-		goto fail_kfree_data;
+	base = vmalloc(buf_size);
+	if (!base)
+		return -ENOMEM;
+	memset(base, 0, buf_size);
+	d204 = page_align_ptr(base + sizeof(d204->hdr)) - sizeof(d204->hdr);
+	rc = diag204_do_store(d204->buf, diag204_buf_pages);
+	if (rc) {
+		vfree(base);
+		return rc;
 	}
-	memset(data->base, 0, buf_size);
-	d204 = page_align_ptr(data->base + sizeof(d204->hdr))
-		- sizeof(d204->hdr);
-	rc = diag204_do_store(&d204->buf, diag204_buf_pages);
-	if (rc)
-		goto fail_vfree_base;
 	d204->hdr.version = DBFS_D204_HDR_VERSION;
 	d204->hdr.len = PAGE_SIZE * diag204_buf_pages;
 	d204->hdr.sc = diag204_store_sc;
-	data->d204 = d204;
-	file->private_data = data;
-	return nonseekable_open(inode, file);
-
-fail_vfree_base:
-	vfree(data->base);
-fail_kfree_data:
-	kfree(data);
-	return rc;
-}
-
-static int dbfs_d204_release(struct inode *inode, struct file *file)
-{
-	struct dbfs_d204_private *data = file->private_data;
-
-	vfree(data->base);
-	kfree(data);
+	*data = d204;
+	*data_free_ptr = base;
+	*size = d204->hdr.len + sizeof(struct dbfs_d204_hdr);
 	return 0;
 }
 
-static ssize_t dbfs_d204_read(struct file *file, char __user *buf,
-			      size_t size, loff_t *ppos)
-{
-	struct dbfs_d204_private *data = file->private_data;
-
-	return simple_read_from_buffer(buf, size, ppos, data->d204,
-				       data->d204->hdr.len +
-				       sizeof(data->d204->hdr));
-}
-
-static const struct file_operations dbfs_d204_ops = {
-	.open		= dbfs_d204_open,
-	.read		= dbfs_d204_read,
-	.release	= dbfs_d204_release,
-	.llseek		= no_llseek,
+static struct hypfs_dbfs_file dbfs_file_d204 = {
+	.name		= "diag_204",
+	.data_create	= dbfs_d204_create,
+	.data_free	= vfree,
 };
 
-static int hypfs_dbfs_init(void)
-{
-	dbfs_d204_file = debugfs_create_file("diag_204", 0400, hypfs_dbfs_dir,
-					     NULL, &dbfs_d204_ops);
-	if (IS_ERR(dbfs_d204_file))
-		return PTR_ERR(dbfs_d204_file);
-	return 0;
-}
-
 __init int hypfs_diag_init(void)
 {
 	int rc;
@@ -639,7 +596,7 @@ __init int hypfs_diag_init(void)
 		return -ENODATA;
 	}
 	if (diag204_info_type == INFO_EXT) {
-		rc = hypfs_dbfs_init();
+		rc = hypfs_dbfs_create_file(&dbfs_file_d204);
 		if (rc)
 			return rc;
 	}
@@ -660,6 +617,7 @@ void hypfs_diag_exit(void)
 	debugfs_remove(dbfs_d204_file);
 	diag224_delete_name_table();
 	diag204_free_buffer();
+	hypfs_dbfs_remove_file(&dbfs_file_d204);
 }
 
 /*

arch/s390/hypfs/hypfs_vm.c

@@ -20,8 +20,6 @@ static char local_guest[] = "        ";
 static char all_guests[] = "*       ";
 static char *guest_query;
 
-static struct dentry *dbfs_d2fc_file;
-
 struct diag2fc_data {
 	__u32 version;
 	__u32 flags;
@@ -104,7 +102,7 @@ static void *diag2fc_store(char *query, unsigned int *count, int offset)
 	return data;
 }
 
-static void diag2fc_free(void *data)
+static void diag2fc_free(const void *data)
 {
 	vfree(data);
 }
@@ -239,43 +237,29 @@ struct dbfs_d2fc {
 	char			buf[];	/* d2fc buffer */
 } __attribute__ ((packed));
 
-static int dbfs_d2fc_open(struct inode *inode, struct file *file)
+static int dbfs_diag2fc_create(void **data, void **data_free_ptr, size_t *size)
 {
-	struct dbfs_d2fc *data;
+	struct dbfs_d2fc *d2fc;
 	unsigned int count;
 
-	data = diag2fc_store(guest_query, &count, sizeof(data->hdr));
-	if (IS_ERR(data))
-		return PTR_ERR(data);
-	get_clock_ext(data->hdr.tod_ext);
-	data->hdr.len = count * sizeof(struct diag2fc_data);
-	data->hdr.version = DBFS_D2FC_HDR_VERSION;
-	data->hdr.count = count;
-	memset(&data->hdr.reserved, 0, sizeof(data->hdr.reserved));
-	file->private_data = data;
-	return nonseekable_open(inode, file);
-}
-
-static int dbfs_d2fc_release(struct inode *inode, struct file *file)
-{
-	diag2fc_free(file->private_data);
+	d2fc = diag2fc_store(guest_query, &count, sizeof(d2fc->hdr));
+	if (IS_ERR(d2fc))
+		return PTR_ERR(d2fc);
+	get_clock_ext(d2fc->hdr.tod_ext);
+	d2fc->hdr.len = count * sizeof(struct diag2fc_data);
+	d2fc->hdr.version = DBFS_D2FC_HDR_VERSION;
+	d2fc->hdr.count = count;
+	memset(&d2fc->hdr.reserved, 0, sizeof(d2fc->hdr.reserved));
+	*data = d2fc;
+	*data_free_ptr = d2fc;
+	*size = d2fc->hdr.len + sizeof(struct dbfs_d2fc_hdr);
 	return 0;
 }
 
-static ssize_t dbfs_d2fc_read(struct file *file, char __user *buf,
-				    size_t size, loff_t *ppos)
-{
-	struct dbfs_d2fc *data = file->private_data;
-
-	return simple_read_from_buffer(buf, size, ppos, data, data->hdr.len +
-				       sizeof(struct dbfs_d2fc_hdr));
-}
-
-static const struct file_operations dbfs_d2fc_ops = {
-	.open		= dbfs_d2fc_open,
-	.read		= dbfs_d2fc_read,
-	.release	= dbfs_d2fc_release,
-	.llseek		= no_llseek,
+static struct hypfs_dbfs_file dbfs_file_2fc = {
+	.name		= "diag_2fc",
+	.data_create	= dbfs_diag2fc_create,
+	.data_free	= diag2fc_free,
 };
 
 int hypfs_vm_init(void)
@@ -288,18 +272,12 @@ int hypfs_vm_init(void)
 		guest_query = local_guest;
 	else
 		return -EACCES;
-
-	dbfs_d2fc_file = debugfs_create_file("diag_2fc", 0400, hypfs_dbfs_dir,
-					     NULL, &dbfs_d2fc_ops);
-	if (IS_ERR(dbfs_d2fc_file))
-		return PTR_ERR(dbfs_d2fc_file);
-
-	return 0;
+	return hypfs_dbfs_create_file(&dbfs_file_2fc);
 }
 
 void hypfs_vm_exit(void)
 {
 	if (!MACHINE_IS_VM)
 		return;
-	debugfs_remove(dbfs_d2fc_file);
+	hypfs_dbfs_remove_file(&dbfs_file_2fc);
 }

arch/s390/hypfs/inode.c

@@ -46,8 +46,6 @@ static const struct super_operations hypfs_s_ops;
 /* start of list of all dentries, which have to be deleted on update */
 static struct dentry *hypfs_last_dentry;
 
-struct dentry *hypfs_dbfs_dir;
-
 static void hypfs_update_update(struct super_block *sb)
 {
 	struct hypfs_sb_info *sb_info = sb->s_fs_info;
@@ -471,13 +469,12 @@ static int __init hypfs_init(void)
 {
 	int rc;
 
-	hypfs_dbfs_dir = debugfs_create_dir("s390_hypfs", NULL);
-	if (IS_ERR(hypfs_dbfs_dir))
-		return PTR_ERR(hypfs_dbfs_dir);
-
+	rc = hypfs_dbfs_init();
+	if (rc)
+		return rc;
 	if (hypfs_diag_init()) {
 		rc = -ENODATA;
-		goto fail_debugfs_remove;
+		goto fail_dbfs_exit;
 	}
 	if (hypfs_vm_init()) {
 		rc = -ENODATA;
@@ -499,9 +496,8 @@ fail_hypfs_vm_exit:
 	hypfs_vm_exit();
 fail_hypfs_diag_exit:
 	hypfs_diag_exit();
-fail_debugfs_remove:
-	debugfs_remove(hypfs_dbfs_dir);
-
+fail_dbfs_exit:
+	hypfs_dbfs_exit();
 	pr_err("Initialization of hypfs failed with rc=%i\n", rc);
 	return rc;
 }
@@ -510,7 +506,7 @@ static void __exit hypfs_exit(void)
 {
 	hypfs_diag_exit();
 	hypfs_vm_exit();
-	debugfs_remove(hypfs_dbfs_dir);
+	hypfs_dbfs_exit();
 	unregister_filesystem(&hypfs_type);
 	kobject_put(s390_kobj);
 }

arch/s390/include/asm/ccwdev.h

@@ -204,6 +204,8 @@ int ccw_device_tm_start_timeout(struct ccw_device *, struct tcw *,
 			    unsigned long, u8, int);
 int ccw_device_tm_intrg(struct ccw_device *cdev);
 
+int ccw_device_get_mdc(struct ccw_device *cdev, u8 mask);
+
 extern int ccw_device_set_online(struct ccw_device *cdev);
 extern int ccw_device_set_offline(struct ccw_device *cdev);
 

arch/s390/include/asm/cputime.h

@@ -202,7 +202,7 @@ static inline void s390_idle_check(struct pt_regs *regs, __u64 int_clock,
 
 static inline int s390_nohz_delay(int cpu)
 {
-	return per_cpu(s390_idle, cpu).nohz_delay != 0;
+	return __get_cpu_var(s390_idle).nohz_delay != 0;
 }
 
 #define arch_needs_cpu(cpu) s390_nohz_delay(cpu)

arch/s390/include/asm/dasd.h

@@ -73,6 +73,7 @@ typedef struct dasd_information2_t {
  * 0x02: use diag discipline (diag)
  * 0x04: set the device initially online (internal use only)
  * 0x08: enable ERP related logging
+ * 0x20: give access to raw eckd data
  */
 #define DASD_FEATURE_DEFAULT	     0x00
 #define DASD_FEATURE_READONLY	     0x01
@@ -80,6 +81,8 @@ typedef struct dasd_information2_t {
 #define DASD_FEATURE_INITIAL_ONLINE  0x04
 #define DASD_FEATURE_ERPLOG	     0x08
 #define DASD_FEATURE_FAILFAST	     0x10
+#define DASD_FEATURE_FAILONSLCK      0x20
+#define DASD_FEATURE_USERAW	     0x40
 
 #define DASD_PARTN_BITS 2
 

arch/s390/include/asm/ftrace.h

@@ -4,20 +4,17 @@
 #ifndef __ASSEMBLY__
 
 extern void _mcount(void);
-extern unsigned long ftrace_dyn_func;
 
 struct dyn_arch_ftrace { };
 
 #define MCOUNT_ADDR ((long)_mcount)
 
 #ifdef CONFIG_64BIT
-#define MCOUNT_OFFSET_RET 18
-#define MCOUNT_INSN_SIZE  24
-#define MCOUNT_OFFSET	  14
-#else
-#define MCOUNT_OFFSET_RET 26
-#define MCOUNT_INSN_SIZE  30
+#define MCOUNT_INSN_SIZE  12
 #define MCOUNT_OFFSET	   8
+#else
+#define MCOUNT_INSN_SIZE  20
+#define MCOUNT_OFFSET	   4
 #endif
 
 static inline unsigned long ftrace_call_adjust(unsigned long addr)

arch/s390/include/asm/hardirq.h

@@ -21,20 +21,4 @@
 
 #define HARDIRQ_BITS	8
 
-void clock_comparator_work(void);
-
-static inline unsigned long long local_tick_disable(void)
-{
-	unsigned long long old;
-
-	old = S390_lowcore.clock_comparator;
-	S390_lowcore.clock_comparator = -1ULL;
-	return old;
-}
-
-static inline void local_tick_enable(unsigned long long comp)
-{
-	S390_lowcore.clock_comparator = comp;
-}
-
 #endif /* __ASM_HARDIRQ_H */

arch/s390/include/asm/irq.h

@@ -1,23 +1,33 @@
 #ifndef _ASM_IRQ_H
 #define _ASM_IRQ_H
 
-#ifdef __KERNEL__
 #include <linux/hardirq.h>
 
-/*
- * the definition of irqs has changed in 2.5.46:
- * NR_IRQS is no longer the number of i/o
- * interrupts (65536), but rather the number
- * of interrupt classes (2).
- * Only external and i/o interrupts make much sense here (CH).
- */
-
 enum interruption_class {
 	EXTERNAL_INTERRUPT,
 	IO_INTERRUPT,
-
+	EXTINT_CLK,
+	EXTINT_IPI,
+	EXTINT_TMR,
+	EXTINT_TLA,
+	EXTINT_PFL,
+	EXTINT_DSD,
+	EXTINT_VRT,
+	EXTINT_SCP,
+	EXTINT_IUC,
+	IOINT_QAI,
+	IOINT_QDI,
+	IOINT_DAS,
+	IOINT_C15,
+	IOINT_C70,
+	IOINT_TAP,
+	IOINT_VMR,
+	IOINT_LCS,
+	IOINT_CLW,
+	IOINT_CTC,
+	IOINT_APB,
+	NMI_NMI,
 	NR_IRQS,
 };
 
-#endif /* __KERNEL__ */
-#endif
+#endif /* _ASM_IRQ_H */

arch/s390/include/asm/kprobes.h

@@ -31,7 +31,6 @@
 #include <linux/ptrace.h>
 #include <linux/percpu.h>
 
-#define  __ARCH_WANT_KPROBES_INSN_SLOT
 struct pt_regs;
 struct kprobe;
 
@@ -58,23 +57,12 @@ typedef u16 kprobe_opcode_t;
 /* Architecture specific copy of original instruction */
 struct arch_specific_insn {
 	/* copy of original instruction */
-	kprobe_opcode_t *insn;
-	int fixup;
-	int ilen;
-	int reg;
+	kprobe_opcode_t insn[MAX_INSN_SIZE];
 };
 
-struct ins_replace_args {
-	kprobe_opcode_t *ptr;
-	kprobe_opcode_t old;
-	kprobe_opcode_t new;
-};
 struct prev_kprobe {
 	struct kprobe *kp;
 	unsigned long status;
-	unsigned long saved_psw;
-	unsigned long kprobe_saved_imask;
-	unsigned long kprobe_saved_ctl[3];
 };
 
 /* per-cpu kprobe control block */
@@ -82,17 +70,13 @@ struct kprobe_ctlblk {
 	unsigned long kprobe_status;
 	unsigned long kprobe_saved_imask;
 	unsigned long kprobe_saved_ctl[3];
-	struct pt_regs jprobe_saved_regs;
-	unsigned long jprobe_saved_r14;
-	unsigned long jprobe_saved_r15;
 	struct prev_kprobe prev_kprobe;
+	struct pt_regs jprobe_saved_regs;
 	kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE];
 };
 
 void arch_remove_kprobe(struct kprobe *p);
 void kretprobe_trampoline(void);
-int  is_prohibited_opcode(kprobe_opcode_t *instruction);
-void get_instruction_type(struct arch_specific_insn *ainsn);
 
 int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
 int kprobe_exceptions_notify(struct notifier_block *self,

arch/s390/include/asm/processor.h

@@ -32,7 +32,6 @@ static inline void get_cpu_id(struct cpuid *ptr)
 }
 
 extern void s390_adjust_jiffies(void);
-extern void print_cpu_info(void);
 extern int get_cpu_capability(unsigned int *);
 
 /*
@@ -81,7 +80,8 @@ struct thread_struct {
 	mm_segment_t mm_segment;
         unsigned long prot_addr;        /* address of protection-excep.     */
         unsigned int trap_no;
-        per_struct per_info;
+	struct per_regs per_user;	/* User specified PER registers */
+	struct per_event per_event;	/* Cause of the last PER trap */
         /* pfault_wait is used to block the process on a pfault event */
 	unsigned long pfault_wait;
 };

arch/s390/include/asm/ptrace.h

@@ -331,10 +331,60 @@ struct pt_regs
 	unsigned short ilc;
 	unsigned short svcnr;
 };
+
+/*
+ * Program event recording (PER) register set.
+ */
+struct per_regs {
+	unsigned long control;		/* PER control bits */
+	unsigned long start;		/* PER starting address */
+	unsigned long end;		/* PER ending address */
+};
+
+/*
+ * PER event contains information about the cause of the last PER exception.
+ */
+struct per_event {
+	unsigned short cause;		/* PER code, ATMID and AI */
+	unsigned long address;		/* PER address */
+	unsigned char paid;		/* PER access identification */
+};
+
+/*
+ * Simplified per_info structure used to decode the ptrace user space ABI.
+ */
+struct per_struct_kernel {
+	unsigned long cr9;		/* PER control bits */
+	unsigned long cr10;		/* PER starting address */
+	unsigned long cr11;		/* PER ending address */
+	unsigned long bits;		/* Obsolete software bits */
+	unsigned long starting_addr;	/* User specified start address */
+	unsigned long ending_addr;	/* User specified end address */
+	unsigned short perc_atmid;	/* PER trap ATMID */
+	unsigned long address;		/* PER trap instruction address */
+	unsigned char access_id;	/* PER trap access identification */
+};
+
+#define PER_EVENT_MASK			0xE9000000UL
+
+#define PER_EVENT_BRANCH		0x80000000UL
+#define PER_EVENT_IFETCH		0x40000000UL
+#define PER_EVENT_STORE			0x20000000UL
+#define PER_EVENT_STORE_REAL		0x08000000UL
+#define PER_EVENT_NULLIFICATION		0x01000000UL
+
+#define PER_CONTROL_MASK		0x00a00000UL
+
+#define PER_CONTROL_BRANCH_ADDRESS	0x00800000UL
+#define PER_CONTROL_ALTERATION		0x00200000UL
+
 #endif
 
 /*
- * Now for the program event recording (trace) definitions.
+ * Now for the user space program event recording (trace) definitions.
+ * The following structures are used only for the ptrace interface, don't
+ * touch or even look at it if you don't want to modify the user-space
+ * ptrace interface. In particular stay away from it for in-kernel PER.
  */
 typedef struct
 {

arch/s390/include/asm/qdio.h

@@ -361,6 +361,7 @@ struct qdio_initialize {
 	qdio_handler_t *input_handler;
 	qdio_handler_t *output_handler;
 	void (*queue_start_poll) (struct ccw_device *, int, unsigned long);
+	int scan_threshold;
 	unsigned long int_parm;
 	void **input_sbal_addr_array;
 	void **output_sbal_addr_array;

arch/s390/include/asm/s390_ext.h

@@ -1,32 +1,17 @@
-#ifndef _S390_EXTINT_H
-#define _S390_EXTINT_H
-
 /*
- *  include/asm-s390/s390_ext.h
- *
- *  S390 version
- *    Copyright IBM Corp. 1999,2007
- *    Author(s): Holger Smolinski (Holger.Smolinski@de.ibm.com),
- *               Martin Schwidefsky (schwidefsky@de.ibm.com)
+ *    Copyright IBM Corp. 1999,2010
+ *    Author(s): Holger Smolinski <Holger.Smolinski@de.ibm.com>,
+ *		 Martin Schwidefsky <schwidefsky@de.ibm.com>,
  */
 
+#ifndef _S390_EXTINT_H
+#define _S390_EXTINT_H
+
 #include <linux/types.h>
 
 typedef void (*ext_int_handler_t)(unsigned int, unsigned int, unsigned long);
 
-typedef struct ext_int_info_t {
-	struct ext_int_info_t *next;
-	ext_int_handler_t handler;
-	__u16 code;
-} ext_int_info_t;
-
-extern ext_int_info_t *ext_int_hash[];
-
 int register_external_interrupt(__u16 code, ext_int_handler_t handler);
-int register_early_external_interrupt(__u16 code, ext_int_handler_t handler,
-				      ext_int_info_t *info);
 int unregister_external_interrupt(__u16 code, ext_int_handler_t handler);
-int unregister_early_external_interrupt(__u16 code, ext_int_handler_t handler,
-					ext_int_info_t *info);
 
-#endif
+#endif /* _S390_EXTINT_H */

arch/s390/include/asm/smp.h

@@ -20,7 +20,6 @@ extern void machine_power_off_smp(void);
 
 extern int __cpu_disable (void);
 extern void __cpu_die (unsigned int cpu);
-extern void cpu_die (void) __attribute__ ((noreturn));
 extern int __cpu_up (unsigned int cpu);
 
 extern struct mutex smp_cpu_state_mutex;
@@ -71,8 +70,10 @@ static inline void smp_switch_to_ipl_cpu(void (*func)(void *), void *data)
 
 #ifdef CONFIG_HOTPLUG_CPU
 extern int smp_rescan_cpus(void);
+extern void __noreturn cpu_die(void);
 #else
 static inline int smp_rescan_cpus(void) { return 0; }
+static inline void cpu_die(void) { }
 #endif
 
 #endif /* __ASM_SMP_H */

arch/s390/include/asm/system.h

@@ -20,6 +20,7 @@
 struct task_struct;
 
 extern struct task_struct *__switch_to(void *, void *);
+extern void update_per_regs(struct task_struct *task);
 
 static inline void save_fp_regs(s390_fp_regs *fpregs)
 {
@@ -93,6 +94,7 @@ static inline void restore_access_regs(unsigned int *acrs)
 	if (next->mm) {							\
 		restore_fp_regs(&next->thread.fp_regs);			\
 		restore_access_regs(&next->thread.acrs[0]);		\
+		update_per_regs(next);					\
 	}								\
 	prev = __switch_to(prev,next);					\
 } while (0)
@@ -101,11 +103,9 @@ extern void account_vtime(struct task_struct *, struct task_struct *);
 extern void account_tick_vtime(struct task_struct *);
 
 #ifdef CONFIG_PFAULT
-extern void pfault_irq_init(void);
 extern int pfault_init(void);
 extern void pfault_fini(void);
 #else /* CONFIG_PFAULT */
-#define pfault_irq_init()	do { } while (0)
 #define pfault_init()		({-1;})
 #define pfault_fini()		do { } while (0)
 #endif /* CONFIG_PFAULT */

arch/s390/include/asm/thread_info.h

@@ -74,7 +74,7 @@ struct thread_info {
 /* how to get the thread information struct from C */
 static inline struct thread_info *current_thread_info(void)
 {
-	return (struct thread_info *)(S390_lowcore.kernel_stack - THREAD_SIZE);
+	return (struct thread_info *) S390_lowcore.thread_info;
 }
 
 #define THREAD_SIZE_ORDER THREAD_ORDER
@@ -88,7 +88,7 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_SIGPENDING		2	/* signal pending */
 #define TIF_NEED_RESCHED	3	/* rescheduling necessary */
 #define TIF_RESTART_SVC		4	/* restart svc with new svc number */
-#define TIF_SINGLE_STEP		6	/* deliver sigtrap on return to user */
+#define TIF_PER_TRAP		6	/* deliver sigtrap on return to user */
 #define TIF_MCCK_PENDING	7	/* machine check handling is pending */
 #define TIF_SYSCALL_TRACE	8	/* syscall trace active */
 #define TIF_SYSCALL_AUDIT	9	/* syscall auditing active */
@@ -99,14 +99,15 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_31BIT		17	/* 32bit process */
 #define TIF_MEMDIE		18	/* is terminating due to OOM killer */
 #define TIF_RESTORE_SIGMASK	19	/* restore signal mask in do_signal() */
-#define TIF_FREEZE		20	/* thread is freezing for suspend */
+#define TIF_SINGLE_STEP		20	/* This task is single stepped */
+#define TIF_FREEZE		21	/* thread is freezing for suspend */
 
 #define _TIF_NOTIFY_RESUME	(1<<TIF_NOTIFY_RESUME)
 #define _TIF_RESTORE_SIGMASK	(1<<TIF_RESTORE_SIGMASK)
 #define _TIF_SIGPENDING		(1<<TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED	(1<<TIF_NEED_RESCHED)
 #define _TIF_RESTART_SVC	(1<<TIF_RESTART_SVC)
-#define _TIF_SINGLE_STEP	(1<<TIF_SINGLE_STEP)
+#define _TIF_PER_TRAP		(1<<TIF_PER_TRAP)
 #define _TIF_MCCK_PENDING	(1<<TIF_MCCK_PENDING)
 #define _TIF_SYSCALL_TRACE	(1<<TIF_SYSCALL_TRACE)
 #define _TIF_SYSCALL_AUDIT	(1<<TIF_SYSCALL_AUDIT)
@@ -114,6 +115,7 @@ static inline struct thread_info *current_thread_info(void)
 #define _TIF_SYSCALL_TRACEPOINT	(1<<TIF_SYSCALL_TRACEPOINT)
 #define _TIF_POLLING_NRFLAG	(1<<TIF_POLLING_NRFLAG)
 #define _TIF_31BIT		(1<<TIF_31BIT)
+#define _TIF_SINGLE_STEP	(1<<TIF_FREEZE)
 #define _TIF_FREEZE		(1<<TIF_FREEZE)
 
 #endif /* __KERNEL__ */

arch/s390/include/asm/timex.h

@@ -11,6 +11,8 @@
 #ifndef _ASM_S390_TIMEX_H
 #define _ASM_S390_TIMEX_H
 
+#include <asm/lowcore.h>
+
 /* The value of the TOD clock for 1.1.1970. */
 #define TOD_UNIX_EPOCH 0x7d91048bca000000ULL
 
@@ -49,6 +51,24 @@ static inline void store_clock_comparator(__u64 *time)
 	asm volatile("stckc %0" : "=Q" (*time));
 }
 
+void clock_comparator_work(void);
+
+static inline unsigned long long local_tick_disable(void)
+{
+	unsigned long long old;
+
+	old = S390_lowcore.clock_comparator;
+	S390_lowcore.clock_comparator = -1ULL;
+	set_clock_comparator(S390_lowcore.clock_comparator);
+	return old;
+}
+
+static inline void local_tick_enable(unsigned long long comp)
+{
+	S390_lowcore.clock_comparator = comp;
+	set_clock_comparator(S390_lowcore.clock_comparator);
+}
+
 #define CLOCK_TICK_RATE	1193180 /* Underlying HZ */
 
 typedef unsigned long long cycles_t;

arch/s390/kernel/asm-offsets.c

@@ -23,14 +23,16 @@ int main(void)
 {
 	DEFINE(__THREAD_info, offsetof(struct task_struct, stack));
 	DEFINE(__THREAD_ksp, offsetof(struct task_struct, thread.ksp));
-	DEFINE(__THREAD_per, offsetof(struct task_struct, thread.per_info));
 	DEFINE(__THREAD_mm_segment, offsetof(struct task_struct, thread.mm_segment));
 	BLANK();
 	DEFINE(__TASK_pid, offsetof(struct task_struct, pid));
 	BLANK();
-	DEFINE(__PER_atmid, offsetof(per_struct, lowcore.words.perc_atmid));
-	DEFINE(__PER_address, offsetof(per_struct, lowcore.words.address));
-	DEFINE(__PER_access_id, offsetof(per_struct, lowcore.words.access_id));
+	DEFINE(__THREAD_per_cause,
+	       offsetof(struct task_struct, thread.per_event.cause));
+	DEFINE(__THREAD_per_address,
+	       offsetof(struct task_struct, thread.per_event.address));
+	DEFINE(__THREAD_per_paid,
+	       offsetof(struct task_struct, thread.per_event.paid));
 	BLANK();
 	DEFINE(__TI_task, offsetof(struct thread_info, task));
 	DEFINE(__TI_domain, offsetof(struct thread_info, exec_domain));
@@ -85,9 +87,9 @@ int main(void)
 	DEFINE(__LC_PGM_ILC, offsetof(struct _lowcore, pgm_ilc));
 	DEFINE(__LC_PGM_INT_CODE, offsetof(struct _lowcore, pgm_code));
 	DEFINE(__LC_TRANS_EXC_CODE, offsetof(struct _lowcore, trans_exc_code));
-	DEFINE(__LC_PER_ATMID, offsetof(struct _lowcore, per_perc_atmid));
+	DEFINE(__LC_PER_CAUSE, offsetof(struct _lowcore, per_perc_atmid));
 	DEFINE(__LC_PER_ADDRESS, offsetof(struct _lowcore, per_address));
-	DEFINE(__LC_PER_ACCESS_ID, offsetof(struct _lowcore, per_access_id));
+	DEFINE(__LC_PER_PAID, offsetof(struct _lowcore, per_access_id));
 	DEFINE(__LC_AR_MODE_ID, offsetof(struct _lowcore, ar_access_id));
 	DEFINE(__LC_SUBCHANNEL_ID, offsetof(struct _lowcore, subchannel_id));
 	DEFINE(__LC_SUBCHANNEL_NR, offsetof(struct _lowcore, subchannel_nr));

arch/s390/kernel/compat_ptrace.h

@@ -4,40 +4,19 @@
 #include <asm/ptrace.h>    /* needed for NUM_CR_WORDS */
 #include "compat_linux.h"  /* needed for psw_compat_t */
 
-typedef struct {
-	__u32 cr[NUM_CR_WORDS];
-} per_cr_words32;
-
-typedef struct {
-	__u16          perc_atmid;          /* 0x096 */
-	__u32          address;             /* 0x098 */
-	__u8           access_id;           /* 0x0a1 */
-} per_lowcore_words32;
-
-typedef struct {
-	union {
-		per_cr_words32   words;
-	} control_regs;
-	/*
-	 * Use these flags instead of setting em_instruction_fetch
-	 * directly they are used so that single stepping can be
-	 * switched on & off while not affecting other tracing
-	 */
-	unsigned  single_step       : 1;
-	unsigned  instruction_fetch : 1;
-	unsigned                    : 30;
-	/*
-	 * These addresses are copied into cr10 & cr11 if single
-	 * stepping is switched off
-	 */
-	__u32     starting_addr;
-	__u32     ending_addr;
-	union {
-		per_lowcore_words32 words;
-	} lowcore; 
-} per_struct32;
+struct compat_per_struct_kernel {
+	__u32 cr9;		/* PER control bits */
+	__u32 cr10;		/* PER starting address */
+	__u32 cr11;		/* PER ending address */
+	__u32 bits;		/* Obsolete software bits */
+	__u32 starting_addr;	/* User specified start address */
+	__u32 ending_addr;	/* User specified end address */
+	__u16 perc_atmid;	/* PER trap ATMID */
+	__u32 address;		/* PER trap instruction address */
+	__u8  access_id;	/* PER trap access identification */
+};
 
-struct user_regs_struct32
+struct compat_user_regs_struct
 {
 	psw_compat_t psw;
 	u32 gprs[NUM_GPRS];
@@ -50,14 +29,14 @@ struct user_regs_struct32
 	 * itself as there is no "official" ptrace interface for hardware
 	 * watchpoints. This is the way intel does it.
 	 */
-	per_struct32 per_info;
+	struct compat_per_struct_kernel per_info;
 	u32  ieee_instruction_pointer;	/* obsolete, always 0 */
 };
 
-struct user32 {
+struct compat_user {
 	/* We start with the registers, to mimic the way that "memory"
 	   is returned from the ptrace(3,...) function.  */
-	struct user_regs_struct32 regs; /* Where the registers are actually stored */
+	struct compat_user_regs_struct regs;
 	/* The rest of this junk is to help gdb figure out what goes where */
 	u32 u_tsize;		/* Text segment size (pages). */
 	u32 u_dsize;	        /* Data segment size (pages). */
@@ -79,6 +58,6 @@ typedef struct
 	__u32   len;
 	__u32   kernel_addr;
 	__u32   process_addr;
-} ptrace_area_emu31;
+} compat_ptrace_area;
 
 #endif /* _PTRACE32_H */

arch/s390/kernel/entry.S

@@ -9,7 +9,6 @@
  *		 Heiko Carstens <heiko.carstens@de.ibm.com>
  */
 
-#include <linux/sys.h>
 #include <linux/linkage.h>
 #include <linux/init.h>
 #include <asm/cache.h>
@@ -49,7 +48,7 @@ SP_SVCNR     =	STACK_FRAME_OVERHEAD + __PT_SVCNR
 SP_SIZE      =	STACK_FRAME_OVERHEAD + __PT_SIZE
 
 _TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
-		 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_SINGLE_STEP )
+		 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_PER_TRAP )
 _TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
 		 _TIF_MCCK_PENDING)
 _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
@@ -110,31 +109,36 @@ STACK_SIZE  = 1 << STACK_SHIFT
 1:	stm	%r10,%r11,\lc_sum
 	.endm
 
-	.macro	SAVE_ALL_BASE savearea
+	.macro	SAVE_ALL_SVC psworg,savearea
 	stm	%r12,%r15,\savearea
 	l	%r13,__LC_SVC_NEW_PSW+4	# load &system_call to %r13
+	l	%r15,__LC_KERNEL_STACK	# problem state -> load ksp
+	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
 	.endm
 
-	.macro	SAVE_ALL_SVC psworg,savearea
-	la	%r12,\psworg
-	l	%r15,__LC_KERNEL_STACK	# problem state -> load ksp
+	.macro	SAVE_ALL_BASE savearea
+	stm	%r12,%r15,\savearea
+	l	%r13,__LC_SVC_NEW_PSW+4	# load &system_call to %r13
 	.endm
 
-	.macro	SAVE_ALL_SYNC psworg,savearea
-	la	%r12,\psworg
+	.macro	SAVE_ALL_PGM psworg,savearea
 	tm	\psworg+1,0x01		# test problem state bit
-	bz	BASED(2f)		# skip stack setup save
-	l	%r15,__LC_KERNEL_STACK	# problem state -> load ksp
 #ifdef CONFIG_CHECK_STACK
-	b	BASED(3f)
-2:	tml	%r15,STACK_SIZE - CONFIG_STACK_GUARD
-	bz	BASED(stack_overflow)
-3:
+	bnz	BASED(1f)
+	tml	%r15,STACK_SIZE - CONFIG_STACK_GUARD
+	bnz	BASED(2f)
+	la	%r12,\psworg
+	b	BASED(stack_overflow)
+#else
+	bz	BASED(2f)
 #endif
-2:
+1:	l	%r15,__LC_KERNEL_STACK	# problem state -> load ksp
+2:	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
 	.endm
 
 	.macro	SAVE_ALL_ASYNC psworg,savearea
+	stm	%r12,%r15,\savearea
+	l	%r13,__LC_SVC_NEW_PSW+4	# load &system_call to %r13
 	la	%r12,\psworg
 	tm	\psworg+1,0x01		# test problem state bit
 	bnz	BASED(1f)		# from user -> load async stack
@@ -149,27 +153,23 @@ STACK_SIZE  = 1 << STACK_SHIFT
 0:	l	%r14,__LC_ASYNC_STACK	# are we already on the async stack ?
 	slr	%r14,%r15
 	sra	%r14,STACK_SHIFT
-	be	BASED(2f)
-1:	l	%r15,__LC_ASYNC_STACK
 #ifdef CONFIG_CHECK_STACK
-	b	BASED(3f)
-2:	tml	%r15,STACK_SIZE - CONFIG_STACK_GUARD
-	bz	BASED(stack_overflow)
-3:
+	bnz	BASED(1f)
+	tml	%r15,STACK_SIZE - CONFIG_STACK_GUARD
+	bnz	BASED(2f)
+	b	BASED(stack_overflow)
+#else
+	bz	BASED(2f)
 #endif
-2:
+1:	l	%r15,__LC_ASYNC_STACK
+2:	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
 	.endm
 
-	.macro	CREATE_STACK_FRAME psworg,savearea
-	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
-	mvc	SP_PSW(8,%r15),0(%r12)	# move user PSW to stack
+	.macro	CREATE_STACK_FRAME savearea
+	xc	__SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15)
 	st	%r2,SP_ORIG_R2(%r15)	# store original content of gpr 2
-	icm	%r12,12,__LC_SVC_ILC
-	stm	%r0,%r11,SP_R0(%r15)	# store gprs %r0-%r11 to kernel stack
-	st	%r12,SP_ILC(%r15)
 	mvc	SP_R12(16,%r15),\savearea # move %r12-%r15 to stack
-	la	%r12,0
-	st	%r12,__SF_BACKCHAIN(%r15)	# clear back chain
+	stm	%r0,%r11,SP_R0(%r15)	# store gprs %r0-%r11 to kernel stack
 	.endm
 
 	.macro	RESTORE_ALL psworg,sync
@@ -188,6 +188,8 @@ STACK_SIZE  = 1 << STACK_SHIFT
 	ssm	__SF_EMPTY(%r15)
 	.endm
 
+	.section .kprobes.text, "ax"
+
 /*
  * Scheduler resume function, called by switch_to
  *  gpr2 = (task_struct *) prev
@@ -198,31 +200,21 @@ STACK_SIZE  = 1 << STACK_SHIFT
 	.globl	__switch_to
 __switch_to:
 	basr	%r1,0
-__switch_to_base:
-	tm	__THREAD_per(%r3),0xe8		# new process is using per ?
-	bz	__switch_to_noper-__switch_to_base(%r1)	# if not we're fine
-	stctl	%c9,%c11,__SF_EMPTY(%r15)	# We are using per stuff
-	clc	__THREAD_per(12,%r3),__SF_EMPTY(%r15)
-	be	__switch_to_noper-__switch_to_base(%r1)	# we got away w/o bashing TLB's
-	lctl	%c9,%c11,__THREAD_per(%r3)	# Nope we didn't
-__switch_to_noper:
-	l	%r4,__THREAD_info(%r2)		# get thread_info of prev
+0:	l	%r4,__THREAD_info(%r2)		# get thread_info of prev
+	l	%r5,__THREAD_info(%r3)		# get thread_info of next
 	tm	__TI_flags+3(%r4),_TIF_MCCK_PENDING # machine check pending?
-	bz	__switch_to_no_mcck-__switch_to_base(%r1)
-	ni	__TI_flags+3(%r4),255-_TIF_MCCK_PENDING # clear flag in prev
-	l	%r4,__THREAD_info(%r3)		# get thread_info of next
-	oi	__TI_flags+3(%r4),_TIF_MCCK_PENDING # set it in next
-__switch_to_no_mcck:
-	stm	%r6,%r15,__SF_GPRS(%r15)# store __switch_to registers of prev task
-	st	%r15,__THREAD_ksp(%r2)	# store kernel stack to prev->tss.ksp
-	l	%r15,__THREAD_ksp(%r3)	# load kernel stack from next->tss.ksp
-	lm	%r6,%r15,__SF_GPRS(%r15)# load __switch_to registers of next task
-	st	%r3,__LC_CURRENT	# __LC_CURRENT = current task struct
-	lctl	%c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4
-	l	%r3,__THREAD_info(%r3)	# load thread_info from task struct
-	st	%r3,__LC_THREAD_INFO
-	ahi	%r3,STACK_SIZE
-	st	%r3,__LC_KERNEL_STACK	# __LC_KERNEL_STACK = new kernel stack
+	bz	1f-0b(%r1)
+	ni	__TI_flags+3(%r4),255-_TIF_MCCK_PENDING	# clear flag in prev
+	oi	__TI_flags+3(%r5),_TIF_MCCK_PENDING	# set it in next
+1:	stm	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
+	st	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
+	l	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
+	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
+	lm	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
+	st	%r3,__LC_CURRENT		# store task struct of next
+	st	%r5,__LC_THREAD_INFO		# store thread info of next
+	ahi	%r5,STACK_SIZE			# end of kernel stack of next
+	st	%r5,__LC_KERNEL_STACK		# store end of kernel stack
 	br	%r14
 
 __critical_start:
@@ -235,10 +227,11 @@ __critical_start:
 system_call:
 	stpt	__LC_SYNC_ENTER_TIMER
 sysc_saveall:
-	SAVE_ALL_BASE __LC_SAVE_AREA
 	SAVE_ALL_SVC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
-	CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
-	lh	%r7,0x8a	  # get svc number from lowcore
+	CREATE_STACK_FRAME __LC_SAVE_AREA
+	mvc	SP_PSW(8,%r15),__LC_SVC_OLD_PSW
+	mvc	SP_ILC(4,%r15),__LC_SVC_ILC
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 sysc_vtime:
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
 sysc_stime:
@@ -246,20 +239,20 @@ sysc_stime:
 sysc_update:
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
 sysc_do_svc:
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
-	ltr	%r7,%r7			# test for svc 0
+	xr	%r7,%r7
+	icm	%r7,3,SP_SVCNR(%r15)	# load svc number and test for svc 0
 	bnz	BASED(sysc_nr_ok)	# svc number > 0
 	# svc 0: system call number in %r1
 	cl	%r1,BASED(.Lnr_syscalls)
 	bnl	BASED(sysc_nr_ok)
+	sth	%r1,SP_SVCNR(%r15)
 	lr	%r7,%r1 	  # copy svc number to %r7
 sysc_nr_ok:
-	sth	%r7,SP_SVCNR(%r15)
 	sll	%r7,2		  # svc number *4
-	l	%r8,BASED(.Lsysc_table)
-	tm	__TI_flags+2(%r9),_TIF_SYSCALL
+	l	%r10,BASED(.Lsysc_table)
+	tm	__TI_flags+2(%r12),_TIF_SYSCALL
 	mvc	SP_ARGS(4,%r15),SP_R7(%r15)
-	l	%r8,0(%r7,%r8)	  # get system call addr.
+	l	%r8,0(%r7,%r10)	  # get system call addr.
 	bnz	BASED(sysc_tracesys)
 	basr	%r14,%r8	  # call sys_xxxx
 	st	%r2,SP_R2(%r15)   # store return value (change R2 on stack)
@@ -267,7 +260,7 @@ sysc_nr_ok:
 sysc_return:
 	LOCKDEP_SYS_EXIT
 sysc_tif:
-	tm	__TI_flags+3(%r9),_TIF_WORK_SVC
+	tm	__TI_flags+3(%r12),_TIF_WORK_SVC
 	bnz	BASED(sysc_work)  # there is work to do (signals etc.)
 sysc_restore:
 	RESTORE_ALL __LC_RETURN_PSW,1
@@ -284,17 +277,17 @@ sysc_work:
 # One of the work bits is on. Find out which one.
 #
 sysc_work_tif:
-	tm	__TI_flags+3(%r9),_TIF_MCCK_PENDING
+	tm	__TI_flags+3(%r12),_TIF_MCCK_PENDING
 	bo	BASED(sysc_mcck_pending)
-	tm	__TI_flags+3(%r9),_TIF_NEED_RESCHED
+	tm	__TI_flags+3(%r12),_TIF_NEED_RESCHED
 	bo	BASED(sysc_reschedule)
-	tm	__TI_flags+3(%r9),_TIF_SIGPENDING
+	tm	__TI_flags+3(%r12),_TIF_SIGPENDING
 	bo	BASED(sysc_sigpending)
-	tm	__TI_flags+3(%r9),_TIF_NOTIFY_RESUME
+	tm	__TI_flags+3(%r12),_TIF_NOTIFY_RESUME
 	bo	BASED(sysc_notify_resume)
-	tm	__TI_flags+3(%r9),_TIF_RESTART_SVC
+	tm	__TI_flags+3(%r12),_TIF_RESTART_SVC
 	bo	BASED(sysc_restart)
-	tm	__TI_flags+3(%r9),_TIF_SINGLE_STEP
+	tm	__TI_flags+3(%r12),_TIF_PER_TRAP
 	bo	BASED(sysc_singlestep)
 	b	BASED(sysc_return)	# beware of critical section cleanup
 
@@ -318,13 +311,13 @@ sysc_mcck_pending:
 # _TIF_SIGPENDING is set, call do_signal
 #
 sysc_sigpending:
-	ni	__TI_flags+3(%r9),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP
+	ni	__TI_flags+3(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
 	la	%r2,SP_PTREGS(%r15)	# load pt_regs
 	l	%r1,BASED(.Ldo_signal)
 	basr	%r14,%r1		# call do_signal
-	tm	__TI_flags+3(%r9),_TIF_RESTART_SVC
+	tm	__TI_flags+3(%r12),_TIF_RESTART_SVC
 	bo	BASED(sysc_restart)
-	tm	__TI_flags+3(%r9),_TIF_SINGLE_STEP
+	tm	__TI_flags+3(%r12),_TIF_PER_TRAP
 	bo	BASED(sysc_singlestep)
 	b	BASED(sysc_return)
 
@@ -342,23 +335,23 @@ sysc_notify_resume:
 # _TIF_RESTART_SVC is set, set up registers and restart svc
 #
 sysc_restart:
-	ni	__TI_flags+3(%r9),255-_TIF_RESTART_SVC # clear TIF_RESTART_SVC
+	ni	__TI_flags+3(%r12),255-_TIF_RESTART_SVC # clear TIF_RESTART_SVC
 	l	%r7,SP_R2(%r15) 	# load new svc number
 	mvc	SP_R2(4,%r15),SP_ORIG_R2(%r15) # restore first argument
 	lm	%r2,%r6,SP_R2(%r15)	# load svc arguments
+	sth	%r7,SP_SVCNR(%r15)
 	b	BASED(sysc_nr_ok)	# restart svc
 
 #
-# _TIF_SINGLE_STEP is set, call do_single_step
+# _TIF_PER_TRAP is set, call do_per_trap
 #
 sysc_singlestep:
-	ni	__TI_flags+3(%r9),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP
-	mvi	SP_SVCNR(%r15),0xff	# set trap indication to pgm check
-	mvi	SP_SVCNR+1(%r15),0xff
+	ni	__TI_flags+3(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
+	xc	SP_SVCNR(2,%r15),SP_SVCNR(%r15)		# clear svc number
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
 	l	%r1,BASED(.Lhandle_per)	# load adr. of per handler
 	la	%r14,BASED(sysc_return)	# load adr. of system return
-	br	%r1			# branch to do_single_step
+	br	%r1			# branch to do_per_trap
 
 #
 # call tracehook_report_syscall_entry/tracehook_report_syscall_exit before
@@ -368,15 +361,15 @@ sysc_tracesys:
 	l	%r1,BASED(.Ltrace_entry)
 	la	%r2,SP_PTREGS(%r15)	# load pt_regs
 	la	%r3,0
-	srl	%r7,2
-	st	%r7,SP_R2(%r15)
+	xr	%r0,%r0
+	icm	%r0,3,SP_SVCNR(%r15)
+	st	%r0,SP_R2(%r15)
 	basr	%r14,%r1
 	cl	%r2,BASED(.Lnr_syscalls)
 	bnl	BASED(sysc_tracenogo)
-	l	%r8,BASED(.Lsysc_table)
 	lr	%r7,%r2
 	sll	%r7,2			# svc number *4
-	l	%r8,0(%r7,%r8)
+	l	%r8,0(%r7,%r10)
 sysc_tracego:
 	lm	%r3,%r6,SP_R3(%r15)
 	mvc	SP_ARGS(4,%r15),SP_R7(%r15)
@@ -384,7 +377,7 @@ sysc_tracego:
 	basr	%r14,%r8		# call sys_xxx
 	st	%r2,SP_R2(%r15)		# store return value
 sysc_tracenogo:
-	tm	__TI_flags+2(%r9),_TIF_SYSCALL
+	tm	__TI_flags+2(%r12),_TIF_SYSCALL
 	bz	BASED(sysc_return)
 	l	%r1,BASED(.Ltrace_exit)
 	la	%r2,SP_PTREGS(%r15)	# load pt_regs
@@ -397,7 +390,7 @@ sysc_tracenogo:
 	.globl	ret_from_fork
 ret_from_fork:
 	l	%r13,__LC_SVC_NEW_PSW+4
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	SP_PSW+1(%r15),0x01	# forking a kernel thread ?
 	bo	BASED(0f)
 	st	%r15,SP_R15(%r15)	# store stack pointer for new kthread
@@ -432,8 +425,8 @@ kernel_execve:
 0:	stnsm	__SF_EMPTY(%r15),0xfc	# disable interrupts
 	l	%r15,__LC_KERNEL_STACK	# load ksp
 	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
-	l	%r9,__LC_THREAD_INFO
 	mvc	SP_PTREGS(__PT_SIZE,%r15),0(%r12)	# copy pt_regs
+	l	%r12,__LC_THREAD_INFO
 	xc	__SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15)
 	stosm	__SF_EMPTY(%r15),0x03	# reenable interrupts
 	l	%r1,BASED(.Lexecve_tail)
@@ -463,26 +456,27 @@ pgm_check_handler:
 	SAVE_ALL_BASE __LC_SAVE_AREA
 	tm	__LC_PGM_INT_CODE+1,0x80 # check whether we got a per exception
 	bnz	BASED(pgm_per)		# got per exception -> special case
-	SAVE_ALL_SYNC __LC_PGM_OLD_PSW,__LC_SAVE_AREA
-	CREATE_STACK_FRAME __LC_PGM_OLD_PSW,__LC_SAVE_AREA
+	SAVE_ALL_PGM __LC_PGM_OLD_PSW,__LC_SAVE_AREA
+	CREATE_STACK_FRAME __LC_SAVE_AREA
+	xc	SP_ILC(4,%r15),SP_ILC(%r15)
+	mvc	SP_PSW(8,%r15),__LC_PGM_OLD_PSW
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	SP_PSW+1(%r15),0x01	# interrupting from user ?
 	bz	BASED(pgm_no_vtime)
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
 	UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
 pgm_no_vtime:
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
 	l	%r3,__LC_PGM_ILC	# load program interruption code
 	l	%r4,__LC_TRANS_EXC_CODE
 	REENABLE_IRQS
 	la	%r8,0x7f
 	nr	%r8,%r3
-pgm_do_call:
-	l	%r7,BASED(.Ljump_table)
 	sll	%r8,2
-	l	%r7,0(%r8,%r7)		# load address of handler routine
+	l	%r1,BASED(.Ljump_table)
+	l	%r1,0(%r8,%r1)		# load address of handler routine
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
-	basr	%r14,%r7		# branch to interrupt-handler
+	basr	%r14,%r1		# branch to interrupt-handler
 pgm_exit:
 	b	BASED(sysc_return)
 
@@ -503,33 +497,34 @@ pgm_per:
 # Normal per exception
 #
 pgm_per_std:
-	SAVE_ALL_SYNC __LC_PGM_OLD_PSW,__LC_SAVE_AREA
-	CREATE_STACK_FRAME __LC_PGM_OLD_PSW,__LC_SAVE_AREA
+	SAVE_ALL_PGM __LC_PGM_OLD_PSW,__LC_SAVE_AREA
+	CREATE_STACK_FRAME __LC_SAVE_AREA
+	mvc	SP_PSW(8,%r15),__LC_PGM_OLD_PSW
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	SP_PSW+1(%r15),0x01	# interrupting from user ?
 	bz	BASED(pgm_no_vtime2)
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
 	UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
 pgm_no_vtime2:
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
-	l	%r1,__TI_task(%r9)
+	l	%r1,__TI_task(%r12)
 	tm	SP_PSW+1(%r15),0x01	# kernel per event ?
 	bz	BASED(kernel_per)
-	mvc	__THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID
-	mvc	__THREAD_per+__PER_address(4,%r1),__LC_PER_ADDRESS
-	mvc	__THREAD_per+__PER_access_id(1,%r1),__LC_PER_ACCESS_ID
-	oi	__TI_flags+3(%r9),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP
+	mvc	__THREAD_per_cause(2,%r1),__LC_PER_CAUSE
+	mvc	__THREAD_per_address(4,%r1),__LC_PER_ADDRESS
+	mvc	__THREAD_per_paid(1,%r1),__LC_PER_PAID
+	oi	__TI_flags+3(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
 	l	%r3,__LC_PGM_ILC	# load program interruption code
 	l	%r4,__LC_TRANS_EXC_CODE
 	REENABLE_IRQS
 	la	%r8,0x7f
 	nr	%r8,%r3 		# clear per-event-bit and ilc
 	be	BASED(pgm_exit2)	# only per or per+check ?
-	l	%r7,BASED(.Ljump_table)
 	sll	%r8,2
-	l	%r7,0(%r8,%r7)		# load address of handler routine
+	l	%r1,BASED(.Ljump_table)
+	l	%r1,0(%r8,%r1)		# load address of handler routine
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
-	basr	%r14,%r7		# branch to interrupt-handler
+	basr	%r14,%r1		# branch to interrupt-handler
 pgm_exit2:
 	b	BASED(sysc_return)
 
@@ -537,18 +532,19 @@ pgm_exit2:
 # it was a single stepped SVC that is causing all the trouble
 #
 pgm_svcper:
-	SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
-	CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
+	SAVE_ALL_PGM __LC_SVC_OLD_PSW,__LC_SAVE_AREA
+	CREATE_STACK_FRAME __LC_SAVE_AREA
+	mvc	SP_PSW(8,%r15),__LC_SVC_OLD_PSW
+	mvc	SP_ILC(4,%r15),__LC_SVC_ILC
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
 	UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
-	lh	%r7,0x8a		# get svc number from lowcore
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
-	l	%r8,__TI_task(%r9)
-	mvc	__THREAD_per+__PER_atmid(2,%r8),__LC_PER_ATMID
-	mvc	__THREAD_per+__PER_address(4,%r8),__LC_PER_ADDRESS
-	mvc	__THREAD_per+__PER_access_id(1,%r8),__LC_PER_ACCESS_ID
-	oi	__TI_flags+3(%r9),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP
+	l	%r8,__TI_task(%r12)
+	mvc	__THREAD_per_cause(2,%r8),__LC_PER_CAUSE
+	mvc	__THREAD_per_address(4,%r8),__LC_PER_ADDRESS
+	mvc	__THREAD_per_paid(1,%r8),__LC_PER_PAID
+	oi	__TI_flags+3(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
 	stosm	__SF_EMPTY(%r15),0x03	# reenable interrupts
 	lm	%r2,%r6,SP_R2(%r15)	# load svc arguments
 	b	BASED(sysc_do_svc)
@@ -558,8 +554,7 @@ pgm_svcper:
 #
 kernel_per:
 	REENABLE_IRQS
-	mvi	SP_SVCNR(%r15),0xff	# set trap indication to pgm check
-	mvi	SP_SVCNR+1(%r15),0xff
+	xc	SP_SVCNR(2,%r15),SP_SVCNR(%r15)
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
 	l	%r1,BASED(.Lhandle_per)	# load adr. of per handler
 	basr	%r14,%r1		# branch to do_single_step
@@ -573,9 +568,10 @@ kernel_per:
 io_int_handler:
 	stck	__LC_INT_CLOCK
 	stpt	__LC_ASYNC_ENTER_TIMER
-	SAVE_ALL_BASE __LC_SAVE_AREA+16
 	SAVE_ALL_ASYNC __LC_IO_OLD_PSW,__LC_SAVE_AREA+16
-	CREATE_STACK_FRAME __LC_IO_OLD_PSW,__LC_SAVE_AREA+16
+	CREATE_STACK_FRAME __LC_SAVE_AREA+16
+	mvc	SP_PSW(8,%r15),0(%r12)	# move user PSW to stack
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	SP_PSW+1(%r15),0x01	# interrupting from user ?
 	bz	BASED(io_no_vtime)
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER
@@ -583,7 +579,6 @@ io_int_handler:
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER
 io_no_vtime:
 	TRACE_IRQS_OFF
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
 	l	%r1,BASED(.Ldo_IRQ)	# load address of do_IRQ
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
 	basr	%r14,%r1		# branch to standard irq handler
@@ -591,7 +586,7 @@ io_return:
 	LOCKDEP_SYS_EXIT
 	TRACE_IRQS_ON
 io_tif:
-	tm	__TI_flags+3(%r9),_TIF_WORK_INT
+	tm	__TI_flags+3(%r12),_TIF_WORK_INT
 	bnz	BASED(io_work)		# there is work to do (signals etc.)
 io_restore:
 	RESTORE_ALL __LC_RETURN_PSW,0
@@ -609,9 +604,9 @@ io_work:
 	bo	BASED(io_work_user)	# yes -> do resched & signal
 #ifdef CONFIG_PREEMPT
 	# check for preemptive scheduling
-	icm	%r0,15,__TI_precount(%r9)
+	icm	%r0,15,__TI_precount(%r12)
 	bnz	BASED(io_restore)	# preemption disabled
-	tm	__TI_flags+3(%r9),_TIF_NEED_RESCHED
+	tm	__TI_flags+3(%r12),_TIF_NEED_RESCHED
 	bno	BASED(io_restore)
 	# switch to kernel stack
 	l	%r1,SP_R15(%r15)
@@ -645,13 +640,13 @@ io_work_user:
 #		and _TIF_MCCK_PENDING
 #
 io_work_tif:
-	tm	__TI_flags+3(%r9),_TIF_MCCK_PENDING
+	tm	__TI_flags+3(%r12),_TIF_MCCK_PENDING
 	bo	BASED(io_mcck_pending)
-	tm	__TI_flags+3(%r9),_TIF_NEED_RESCHED
+	tm	__TI_flags+3(%r12),_TIF_NEED_RESCHED
 	bo	BASED(io_reschedule)
-	tm	__TI_flags+3(%r9),_TIF_SIGPENDING
+	tm	__TI_flags+3(%r12),_TIF_SIGPENDING
 	bo	BASED(io_sigpending)
-	tm	__TI_flags+3(%r9),_TIF_NOTIFY_RESUME
+	tm	__TI_flags+3(%r12),_TIF_NOTIFY_RESUME
 	bo	BASED(io_notify_resume)
 	b	BASED(io_return)	# beware of critical section cleanup
 
@@ -711,16 +706,16 @@ io_notify_resume:
 ext_int_handler:
 	stck	__LC_INT_CLOCK
 	stpt	__LC_ASYNC_ENTER_TIMER
-	SAVE_ALL_BASE __LC_SAVE_AREA+16
 	SAVE_ALL_ASYNC __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16
-	CREATE_STACK_FRAME __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16
+	CREATE_STACK_FRAME __LC_SAVE_AREA+16
+	mvc	SP_PSW(8,%r15),0(%r12)	# move user PSW to stack
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	SP_PSW+1(%r15),0x01	# interrupting from user ?
 	bz	BASED(ext_no_vtime)
 	UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER
 	UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER
 ext_no_vtime:
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
 	TRACE_IRQS_OFF
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
 	l	%r3,__LC_CPU_ADDRESS	# get cpu address + interruption code
@@ -775,7 +770,10 @@ mcck_int_main:
 	sra	%r14,PAGE_SHIFT
 	be	BASED(0f)
 	l	%r15,__LC_PANIC_STACK	# load panic stack
-0:	CREATE_STACK_FRAME __LC_MCK_OLD_PSW,__LC_SAVE_AREA+32
+0:	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
+	CREATE_STACK_FRAME __LC_SAVE_AREA+32
+	mvc	SP_PSW(8,%r15),0(%r12)
+	l	%r12,__LC_THREAD_INFO	# load pointer to thread_info struct
 	tm	__LC_MCCK_CODE+2,0x08	# mwp of old psw valid?
 	bno	BASED(mcck_no_vtime)	# no -> skip cleanup critical
 	tm	SP_PSW+1(%r15),0x01	# interrupting from user ?
@@ -784,7 +782,6 @@ mcck_int_main:
 	UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_MCCK_ENTER_TIMER
 mcck_no_vtime:
-	l	%r9,__LC_THREAD_INFO	# load pointer to thread_info struct
 	la	%r2,SP_PTREGS(%r15)	# load pt_regs
 	l	%r1,BASED(.Ls390_mcck)
 	basr	%r14,%r1		# call machine check handler
@@ -796,7 +793,7 @@ mcck_no_vtime:
 	xc	__SF_BACKCHAIN(4,%r1),__SF_BACKCHAIN(%r1) # clear back chain
 	lr	%r15,%r1
 	stosm	__SF_EMPTY(%r15),0x04	# turn dat on
-	tm	__TI_flags+3(%r9),_TIF_MCCK_PENDING
+	tm	__TI_flags+3(%r12),_TIF_MCCK_PENDING
 	bno	BASED(mcck_return)
 	TRACE_IRQS_OFF
 	l	%r1,BASED(.Ls390_handle_mcck)
@@ -861,6 +858,8 @@ restart_crash:
 restart_go:
 #endif
 
+	.section .kprobes.text, "ax"
+
 #ifdef CONFIG_CHECK_STACK
 /*
  * The synchronous or the asynchronous stack overflowed. We are dead.
@@ -943,12 +942,13 @@ cleanup_system_call:
 	bh	BASED(0f)
 	mvc	__LC_SAVE_AREA(16),0(%r12)
 0:	st	%r13,4(%r12)
-	st	%r12,__LC_SAVE_AREA+48	# argh
-	SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
-	CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
-	l	%r12,__LC_SAVE_AREA+48	# argh
+	l	%r15,__LC_KERNEL_STACK	# problem state -> load ksp
+	s	%r15,BASED(.Lc_spsize)	# make room for registers & psw
 	st	%r15,12(%r12)
-	lh	%r7,0x8a
+	CREATE_STACK_FRAME __LC_SAVE_AREA
+	mvc	SP_PSW(8,%r15),__LC_SVC_OLD_PSW
+	mvc	SP_ILC(4,%r15),__LC_SVC_ILC
+	mvc	0(4,%r12),__LC_THREAD_INFO
 cleanup_vtime:
 	clc	__LC_RETURN_PSW+4(4),BASED(cleanup_system_call_insn+12)
 	bhe	BASED(cleanup_stime)
@@ -1046,7 +1046,7 @@ cleanup_io_restore_insn:
 .Ldo_signal:	.long	do_signal
 .Ldo_notify_resume:
 		.long	do_notify_resume
-.Lhandle_per:	.long	do_single_step
+.Lhandle_per:	.long	do_per_trap
 .Ldo_execve:	.long	do_execve
 .Lexecve_tail:	.long	execve_tail
 .Ljump_table:	.long	pgm_check_table

arch/s390/kernel/entry.h

@@ -12,7 +12,7 @@ pgm_check_handler_t do_dat_exception;
 
 extern int sysctl_userprocess_debug;
 
-void do_single_step(struct pt_regs *regs);
+void do_per_trap(struct pt_regs *regs);
 void syscall_trace(struct pt_regs *regs, int entryexit);
 void kernel_stack_overflow(struct pt_regs * regs);
 void do_signal(struct pt_regs *regs);

arch/s390/kernel/entry64.S

@@ -51,7 +51,7 @@ STACK_SHIFT = PAGE_SHIFT + THREAD_ORDER
 STACK_SIZE  = 1 << STACK_SHIFT
 
 _TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
-		 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_SINGLE_STEP )
+		 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_PER_TRAP )
 _TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
 		 _TIF_MCCK_PENDING)
 _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
@@ -197,6 +197,8 @@ _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
 	ssm	__SF_EMPTY(%r15)
 	.endm
 
+	.section .kprobes.text, "ax"
+
 /*
  * Scheduler resume function, called by switch_to
  *  gpr2 = (task_struct *) prev
@@ -206,30 +208,21 @@ _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
  */
 	.globl	__switch_to
 __switch_to:
-	tm	__THREAD_per+4(%r3),0xe8 # is the new process using per ?
-	jz	__switch_to_noper		# if not we're fine
-	stctg	%c9,%c11,__SF_EMPTY(%r15)# We are using per stuff
-	clc	__THREAD_per(24,%r3),__SF_EMPTY(%r15)
-	je	__switch_to_noper	     # we got away without bashing TLB's
-	lctlg	%c9,%c11,__THREAD_per(%r3)	# Nope we didn't
-__switch_to_noper:
-	lg	%r4,__THREAD_info(%r2)		    # get thread_info of prev
+	lg	%r4,__THREAD_info(%r2)		# get thread_info of prev
+	lg	%r5,__THREAD_info(%r3)		# get thread_info of next
 	tm	__TI_flags+7(%r4),_TIF_MCCK_PENDING # machine check pending?
-	jz	__switch_to_no_mcck
-	ni	__TI_flags+7(%r4),255-_TIF_MCCK_PENDING # clear flag in prev
-	lg	%r4,__THREAD_info(%r3)		    # get thread_info of next
-	oi	__TI_flags+7(%r4),_TIF_MCCK_PENDING # set it in next
-__switch_to_no_mcck:
-	stmg	%r6,%r15,__SF_GPRS(%r15)# store __switch_to registers of prev task
-	stg	%r15,__THREAD_ksp(%r2)	# store kernel stack to prev->tss.ksp
-	lg	%r15,__THREAD_ksp(%r3)	# load kernel stack from next->tss.ksp
-	lmg	%r6,%r15,__SF_GPRS(%r15)# load __switch_to registers of next task
-	stg	%r3,__LC_CURRENT	# __LC_CURRENT = current task struct
-	lctl	%c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4
-	lg	%r3,__THREAD_info(%r3)	# load thread_info from task struct
-	stg	%r3,__LC_THREAD_INFO
-	aghi	%r3,STACK_SIZE
-	stg	%r3,__LC_KERNEL_STACK	# __LC_KERNEL_STACK = new kernel stack
+	jz	0f
+	ni	__TI_flags+7(%r4),255-_TIF_MCCK_PENDING	# clear flag in prev
+	oi	__TI_flags+7(%r5),_TIF_MCCK_PENDING	# set it in next
+0:	stmg	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
+	stg	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
+	lg	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
+	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
+	lmg	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
+	stg	%r3,__LC_CURRENT		# store task struct of next
+	stg	%r5,__LC_THREAD_INFO		# store thread info of next
+	aghi	%r5,STACK_SIZE			# end of kernel stack of next
+	stg	%r5,__LC_KERNEL_STACK		# store end of kernel stack
 	br	%r14
 
 __critical_start:
@@ -309,7 +302,7 @@ sysc_work_tif:
 	jo	sysc_notify_resume
 	tm	__TI_flags+7(%r12),_TIF_RESTART_SVC
 	jo	sysc_restart
-	tm	__TI_flags+7(%r12),_TIF_SINGLE_STEP
+	tm	__TI_flags+7(%r12),_TIF_PER_TRAP
 	jo	sysc_singlestep
 	j	sysc_return		# beware of critical section cleanup
 
@@ -331,12 +324,12 @@ sysc_mcck_pending:
 # _TIF_SIGPENDING is set, call do_signal
 #
 sysc_sigpending:
-	ni	__TI_flags+7(%r12),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP
+	ni	__TI_flags+7(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
 	la	%r2,SP_PTREGS(%r15)	# load pt_regs
 	brasl	%r14,do_signal		# call do_signal
 	tm	__TI_flags+7(%r12),_TIF_RESTART_SVC
 	jo	sysc_restart
-	tm	__TI_flags+7(%r12),_TIF_SINGLE_STEP
+	tm	__TI_flags+7(%r12),_TIF_PER_TRAP
 	jo	sysc_singlestep
 	j	sysc_return
 
@@ -361,14 +354,14 @@ sysc_restart:
 	j	sysc_nr_ok		# restart svc
 
 #
-# _TIF_SINGLE_STEP is set, call do_single_step
+# _TIF_PER_TRAP is set, call do_per_trap
 #
 sysc_singlestep:
-	ni	__TI_flags+7(%r12),255-_TIF_SINGLE_STEP	# clear TIF_SINGLE_STEP
+	ni	__TI_flags+7(%r12),255-_TIF_PER_TRAP	# clear TIF_PER_TRAP
 	xc	SP_SVCNR(2,%r15),SP_SVCNR(%r15)		# clear svc number
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
 	larl	%r14,sysc_return	# load adr. of system return
-	jg	do_single_step		# branch to do_sigtrap
+	jg	do_per_trap
 
 #
 # call tracehook_report_syscall_entry/tracehook_report_syscall_exit before
@@ -524,10 +517,10 @@ pgm_no_vtime2:
 	lg	%r1,__TI_task(%r12)
 	tm	SP_PSW+1(%r15),0x01	# kernel per event ?
 	jz	kernel_per
-	mvc	__THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID
-	mvc	__THREAD_per+__PER_address(8,%r1),__LC_PER_ADDRESS
-	mvc	__THREAD_per+__PER_access_id(1,%r1),__LC_PER_ACCESS_ID
-	oi	__TI_flags+7(%r12),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP
+	mvc	__THREAD_per_cause(2,%r1),__LC_PER_CAUSE
+	mvc	__THREAD_per_address(8,%r1),__LC_PER_ADDRESS
+	mvc	__THREAD_per_paid(1,%r1),__LC_PER_PAID
+	oi	__TI_flags+7(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
 	lgf	%r3,__LC_PGM_ILC	# load program interruption code
 	lg	%r4,__LC_TRANS_EXC_CODE
 	REENABLE_IRQS
@@ -556,10 +549,10 @@ pgm_svcper:
 	mvc	__LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
 	LAST_BREAK
 	lg	%r8,__TI_task(%r12)
-	mvc	__THREAD_per+__PER_atmid(2,%r8),__LC_PER_ATMID
-	mvc	__THREAD_per+__PER_address(8,%r8),__LC_PER_ADDRESS
-	mvc	__THREAD_per+__PER_access_id(1,%r8),__LC_PER_ACCESS_ID
-	oi	__TI_flags+7(%r12),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP
+	mvc	__THREAD_per_cause(2,%r8),__LC_PER_CAUSE
+	mvc	__THREAD_per_address(8,%r8),__LC_PER_ADDRESS
+	mvc	__THREAD_per_paid(1,%r8),__LC_PER_PAID
+	oi	__TI_flags+7(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
 	stosm	__SF_EMPTY(%r15),0x03	# reenable interrupts
 	lmg	%r2,%r6,SP_R2(%r15)	# load svc arguments
 	j	sysc_do_svc
@@ -571,7 +564,7 @@ kernel_per:
 	REENABLE_IRQS
 	xc	SP_SVCNR(2,%r15),SP_SVCNR(%r15)	# clear svc number
 	la	%r2,SP_PTREGS(%r15)	# address of register-save area
-	brasl	%r14,do_single_step
+	brasl	%r14,do_per_trap
 	j	pgm_exit
 
 /*
@@ -868,6 +861,8 @@ restart_crash:
 restart_go:
 #endif
 
+	.section .kprobes.text, "ax"
+
 #ifdef CONFIG_CHECK_STACK
 /*
  * The synchronous or the asynchronous stack overflowed. We are dead.

arch/s390/kernel/ftrace.c

@@ -4,7 +4,7 @@
  * Copyright IBM Corp. 2009
  *
  *   Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>,
- *
+ *		Martin Schwidefsky <schwidefsky@de.ibm.com>
  */
 
 #include <linux/hardirq.h>
@@ -12,176 +12,144 @@
 #include <linux/ftrace.h>
 #include <linux/kernel.h>
 #include <linux/types.h>
+#include <linux/kprobes.h>
 #include <trace/syscall.h>
 #include <asm/asm-offsets.h>
 
+#ifdef CONFIG_64BIT
+#define MCOUNT_OFFSET_RET 12
+#else
+#define MCOUNT_OFFSET_RET 22
+#endif
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 
 void ftrace_disable_code(void);
-void ftrace_disable_return(void);
-void ftrace_call_code(void);
-void ftrace_nop_code(void);
-
-#define FTRACE_INSN_SIZE 4
+void ftrace_enable_insn(void);
 
 #ifdef CONFIG_64BIT
-
+/*
+ * The 64-bit mcount code looks like this:
+ *	stg	%r14,8(%r15)		# offset 0
+ * >	larl	%r1,<&counter>		# offset 6
+ * >	brasl	%r14,_mcount		# offset 12
+ *	lg	%r14,8(%r15)		# offset 18
+ * Total length is 24 bytes. The middle two instructions of the mcount
+ * block get overwritten by ftrace_make_nop / ftrace_make_call.
+ * The 64-bit enabled ftrace code block looks like this:
+ *	stg	%r14,8(%r15)		# offset 0
+ * >	lg	%r1,__LC_FTRACE_FUNC	# offset 6
+ * >	lgr	%r0,%r0			# offset 12
+ * >	basr	%r14,%r1		# offset 16
+ *	lg	%r14,8(%15)		# offset 18
+ * The return points of the mcount/ftrace function have the same offset 18.
+ * The 64-bit disable ftrace code block looks like this:
+ *	stg	%r14,8(%r15)		# offset 0
+ * >	jg	.+18			# offset 6
+ * >	lgr	%r0,%r0			# offset 12
+ * >	basr	%r14,%r1		# offset 16
+ *	lg	%r14,8(%15)		# offset 18
+ * The jg instruction branches to offset 24 to skip as many instructions
+ * as possible.
+ */
 asm(
 	"	.align	4\n"
 	"ftrace_disable_code:\n"
-	"	j	0f\n"
-	"	.word	0x0024\n"
-	"	lg	%r1,"__stringify(__LC_FTRACE_FUNC)"\n"
-	"	basr	%r14,%r1\n"
-	"ftrace_disable_return:\n"
-	"	lg	%r14,8(15)\n"
+	"	jg	0f\n"
 	"	lgr	%r0,%r0\n"
-	"0:\n");
-
-asm(
+	"	basr	%r14,%r1\n"
+	"0:\n"
 	"	.align	4\n"
-	"ftrace_nop_code:\n"
-	"	j	.+"__stringify(MCOUNT_INSN_SIZE)"\n");
+	"ftrace_enable_insn:\n"
+	"	lg	%r1,"__stringify(__LC_FTRACE_FUNC)"\n");
 
-asm(
-	"	.align	4\n"
-	"ftrace_call_code:\n"
-	"	stg	%r14,8(%r15)\n");
+#define FTRACE_INSN_SIZE	6
 
 #else /* CONFIG_64BIT */
-
+/*
+ * The 31-bit mcount code looks like this:
+ *	st	%r14,4(%r15)		# offset 0
+ * >	bras	%r1,0f			# offset 4
+ * >	.long	_mcount			# offset 8
+ * >	.long	<&counter>		# offset 12
+ * > 0:	l	%r14,0(%r1)		# offset 16
+ * >	l	%r1,4(%r1)		# offset 20
+ *	basr	%r14,%r14		# offset 24
+ *	l	%r14,4(%r15)		# offset 26
+ * Total length is 30 bytes. The twenty bytes starting from offset 4
+ * to offset 24 get overwritten by ftrace_make_nop / ftrace_make_call.
+ * The 31-bit enabled ftrace code block looks like this:
+ *	st	%r14,4(%r15)		# offset 0
+ * >	l	%r14,__LC_FTRACE_FUNC	# offset 4
+ * >	j	0f			# offset 8
+ * >	.fill	12,1,0x07		# offset 12
+ *   0:	basr	%r14,%r14		# offset 24
+ *	l	%r14,4(%r14)		# offset 26
+ * The return points of the mcount/ftrace function have the same offset 26.
+ * The 31-bit disabled ftrace code block looks like this:
+ *	st	%r14,4(%r15)		# offset 0
+ * >	j	.+26			# offset 4
+ * >	j	0f			# offset 8
+ * >	.fill	12,1,0x07		# offset 12
+ *   0:	basr	%r14,%r14		# offset 24
+ *	l	%r14,4(%r14)		# offset 26
+ * The j instruction branches to offset 30 to skip as many instructions
+ * as possible.
+ */
 asm(
 	"	.align	4\n"
 	"ftrace_disable_code:\n"
+	"	j	1f\n"
 	"	j	0f\n"
-	"	l	%r1,"__stringify(__LC_FTRACE_FUNC)"\n"
-	"	basr	%r14,%r1\n"
-	"ftrace_disable_return:\n"
-	"	l	%r14,4(%r15)\n"
-	"	j	0f\n"
-	"	bcr	0,%r7\n"
-	"	bcr	0,%r7\n"
-	"	bcr	0,%r7\n"
-	"	bcr	0,%r7\n"
-	"	bcr	0,%r7\n"
-	"	bcr	0,%r7\n"
-	"0:\n");
-
-asm(
+	"	.fill	12,1,0x07\n"
+	"0:	basr	%r14,%r14\n"
+	"1:\n"
 	"	.align	4\n"
-	"ftrace_nop_code:\n"
-	"	j	.+"__stringify(MCOUNT_INSN_SIZE)"\n");
+	"ftrace_enable_insn:\n"
+	"	l	%r14,"__stringify(__LC_FTRACE_FUNC)"\n");
 
-asm(
-	"	.align	4\n"
-	"ftrace_call_code:\n"
-	"	st	%r14,4(%r15)\n");
+#define FTRACE_INSN_SIZE	4
 
 #endif /* CONFIG_64BIT */
 
-static int ftrace_modify_code(unsigned long ip,
-			      void *old_code, int old_size,
-			      void *new_code, int new_size)
-{
-	unsigned char replaced[MCOUNT_INSN_SIZE];
-
-	/*
-	 * Note: Due to modules code can disappear and change.
-	 *  We need to protect against faulting as well as code
-	 *  changing. We do this by using the probe_kernel_*
-	 *  functions.
-	 *  This however is just a simple sanity check.
-	 */
-	if (probe_kernel_read(replaced, (void *)ip, old_size))
-		return -EFAULT;
-	if (memcmp(replaced, old_code, old_size) != 0)
-		return -EINVAL;
-	if (probe_kernel_write((void *)ip, new_code, new_size))
-		return -EPERM;
-	return 0;
-}
-
-static int ftrace_make_initial_nop(struct module *mod, struct dyn_ftrace *rec,
-				   unsigned long addr)
-{
-	return ftrace_modify_code(rec->ip,
-				  ftrace_call_code, FTRACE_INSN_SIZE,
-				  ftrace_disable_code, MCOUNT_INSN_SIZE);
-}
 
 int ftrace_make_nop(struct module *mod, struct dyn_ftrace *rec,
 		    unsigned long addr)
 {
-	if (addr == MCOUNT_ADDR)
-		return ftrace_make_initial_nop(mod, rec, addr);
-	return ftrace_modify_code(rec->ip,
-				  ftrace_call_code, FTRACE_INSN_SIZE,
-				  ftrace_nop_code, FTRACE_INSN_SIZE);
+	if (probe_kernel_write((void *) rec->ip, ftrace_disable_code,
+			       MCOUNT_INSN_SIZE))
+		return -EPERM;
+	return 0;
 }
 
 int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 {
-	return ftrace_modify_code(rec->ip,
-				  ftrace_nop_code, FTRACE_INSN_SIZE,
-				  ftrace_call_code, FTRACE_INSN_SIZE);
+	if (probe_kernel_write((void *) rec->ip, ftrace_enable_insn,
+			       FTRACE_INSN_SIZE))
+		return -EPERM;
+	return 0;
 }
 
 int ftrace_update_ftrace_func(ftrace_func_t func)
 {
-	ftrace_dyn_func = (unsigned long)func;
 	return 0;
 }
 
 int __init ftrace_dyn_arch_init(void *data)
 {
-	*(unsigned long *)data = 0;
+	*(unsigned long *) data = 0;
 	return 0;
 }
 
 #endif /* CONFIG_DYNAMIC_FTRACE */
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
-#ifdef CONFIG_DYNAMIC_FTRACE
-/*
- * Patch the kernel code at ftrace_graph_caller location:
- * The instruction there is branch relative on condition. The condition mask
- * is either all ones (always branch aka disable ftrace_graph_caller) or all
- * zeroes (nop aka enable ftrace_graph_caller).
- * Instruction format for brc is a7m4xxxx where m is the condition mask.
- */
-int ftrace_enable_ftrace_graph_caller(void)
-{
-	unsigned short opcode = 0xa704;
-
-	return probe_kernel_write(ftrace_graph_caller, &opcode, sizeof(opcode));
-}
-
-int ftrace_disable_ftrace_graph_caller(void)
-{
-	unsigned short opcode = 0xa7f4;
-
-	return probe_kernel_write(ftrace_graph_caller, &opcode, sizeof(opcode));
-}
-
-static inline unsigned long ftrace_mcount_call_adjust(unsigned long addr)
-{
-	return addr - (ftrace_disable_return - ftrace_disable_code);
-}
-
-#else /* CONFIG_DYNAMIC_FTRACE */
-
-static inline unsigned long ftrace_mcount_call_adjust(unsigned long addr)
-{
-	return addr - MCOUNT_OFFSET_RET;
-}
-
-#endif /* CONFIG_DYNAMIC_FTRACE */
-
 /*
  * Hook the return address and push it in the stack of return addresses
  * in current thread info.
  */
-unsigned long prepare_ftrace_return(unsigned long ip, unsigned long parent)
+unsigned long __kprobes prepare_ftrace_return(unsigned long parent,
+					      unsigned long ip)
 {
 	struct ftrace_graph_ent trace;
 
@@ -189,14 +157,42 @@ unsigned long prepare_ftrace_return(unsigned long ip, unsigned long parent)
 		goto out;
 	if (ftrace_push_return_trace(parent, ip, &trace.depth, 0) == -EBUSY)
 		goto out;
-	trace.func = ftrace_mcount_call_adjust(ip) & PSW_ADDR_INSN;
+	trace.func = (ip & PSW_ADDR_INSN) - MCOUNT_OFFSET_RET;
 	/* Only trace if the calling function expects to. */
 	if (!ftrace_graph_entry(&trace)) {
 		current->curr_ret_stack--;
 		goto out;
 	}
-	parent = (unsigned long)return_to_handler;
+	parent = (unsigned long) return_to_handler;
 out:
 	return parent;
 }
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+/*
+ * Patch the kernel code at ftrace_graph_caller location. The instruction
+ * there is branch relative and save to prepare_ftrace_return. To disable
+ * the call to prepare_ftrace_return we patch the bras offset to point
+ * directly after the instructions. To enable the call we calculate
+ * the original offset to prepare_ftrace_return and put it back.
+ */
+int ftrace_enable_ftrace_graph_caller(void)
+{
+	unsigned short offset;
+
+	offset = ((void *) prepare_ftrace_return -
+		  (void *) ftrace_graph_caller) / 2;
+	return probe_kernel_write(ftrace_graph_caller + 2,
+				  &offset, sizeof(offset));
+}
+
+int ftrace_disable_ftrace_graph_caller(void)
+{
+	static unsigned short offset = 0x0002;
+
+	return probe_kernel_write(ftrace_graph_caller + 2,
+				  &offset, sizeof(offset));
+}
+
+#endif /* CONFIG_DYNAMIC_FTRACE */
 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */

arch/s390/kernel/irq.c

@@ -1,7 +1,5 @@
 /*
- *  arch/s390/kernel/irq.c
- *
- *    Copyright IBM Corp. 2004,2007
+ *    Copyright IBM Corp. 2004,2010
  *    Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com),
  *		 Thomas Spatzier (tspat@de.ibm.com)
  *
@@ -17,12 +15,42 @@
 #include <linux/proc_fs.h>
 #include <linux/profile.h>
 
+struct irq_class {
+	char *name;
+	char *desc;
+};
+
+static const struct irq_class intrclass_names[] = {
+	{.name = "EXT" },
+	{.name = "I/O" },
+	{.name = "CLK", .desc = "[EXT] Clock Comparator" },
+	{.name = "IPI", .desc = "[EXT] Signal Processor" },
+	{.name = "TMR", .desc = "[EXT] CPU Timer" },
+	{.name = "TAL", .desc = "[EXT] Timing Alert" },
+	{.name = "PFL", .desc = "[EXT] Pseudo Page Fault" },
+	{.name = "DSD", .desc = "[EXT] DASD Diag" },
+	{.name = "VRT", .desc = "[EXT] Virtio" },
+	{.name = "SCP", .desc = "[EXT] Service Call" },
+	{.name = "IUC", .desc = "[EXT] IUCV" },
+	{.name = "QAI", .desc = "[I/O] QDIO Adapter Interrupt" },
+	{.name = "QDI", .desc = "[I/O] QDIO Interrupt" },
+	{.name = "DAS", .desc = "[I/O] DASD" },
+	{.name = "C15", .desc = "[I/O] 3215" },
+	{.name = "C70", .desc = "[I/O] 3270" },
+	{.name = "TAP", .desc = "[I/O] Tape" },
+	{.name = "VMR", .desc = "[I/O] Unit Record Devices" },
+	{.name = "LCS", .desc = "[I/O] LCS" },
+	{.name = "CLW", .desc = "[I/O] CLAW" },
+	{.name = "CTC", .desc = "[I/O] CTC" },
+	{.name = "APB", .desc = "[I/O] AP Bus" },
+	{.name = "NMI", .desc = "[NMI] Machine Check" },
+};
+
 /*
  * show_interrupts is needed by /proc/interrupts.
  */
 int show_interrupts(struct seq_file *p, void *v)
 {
-	static const char *intrclass_names[] = { "EXT", "I/O", };
 	int i = *(loff_t *) v, j;
 
 	get_online_cpus();
@@ -34,15 +62,16 @@ int show_interrupts(struct seq_file *p, void *v)
 	}
 
 	if (i < NR_IRQS) {
-		seq_printf(p, "%s: ", intrclass_names[i]);
+		seq_printf(p, "%s: ", intrclass_names[i].name);
 #ifndef CONFIG_SMP
 		seq_printf(p, "%10u ", kstat_irqs(i));
 #else
 		for_each_online_cpu(j)
 			seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]);
 #endif
+		if (intrclass_names[i].desc)
+			seq_printf(p, "  %s", intrclass_names[i].desc);
                 seq_putc(p, '\n');
-
         }
 	put_online_cpus();
         return 0;

arch/s390/kernel/kprobes.c

@@ -32,34 +32,14 @@
 #include <linux/slab.h>
 #include <linux/hardirq.h>
 
-DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
+DEFINE_PER_CPU(struct kprobe *, current_kprobe);
 DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
 
-struct kretprobe_blackpoint kretprobe_blacklist[] = {{NULL, NULL}};
+struct kretprobe_blackpoint kretprobe_blacklist[] = { };
 
-int __kprobes arch_prepare_kprobe(struct kprobe *p)
-{
-	/* Make sure the probe isn't going on a difficult instruction */
-	if (is_prohibited_opcode((kprobe_opcode_t *) p->addr))
-		return -EINVAL;
-
-	if ((unsigned long)p->addr & 0x01)
-		return -EINVAL;
-
-	/* Use the get_insn_slot() facility for correctness */
-	if (!(p->ainsn.insn = get_insn_slot()))
-		return -ENOMEM;
-
-	memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
-
-	get_instruction_type(&p->ainsn);
-	p->opcode = *p->addr;
-	return 0;
-}
-
-int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
+static int __kprobes is_prohibited_opcode(kprobe_opcode_t *insn)
 {
-	switch (*(__u8 *) instruction) {
+	switch (insn[0] >> 8) {
 	case 0x0c:	/* bassm */
 	case 0x0b:	/* bsm	 */
 	case 0x83:	/* diag  */
@@ -68,7 +48,7 @@ int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
 	case 0xad:	/* stosm */
 		return -EINVAL;
 	}
-	switch (*(__u16 *) instruction) {
+	switch (insn[0]) {
 	case 0x0101:	/* pr	 */
 	case 0xb25a:	/* bsa	 */
 	case 0xb240:	/* bakr  */
@@ -81,93 +61,92 @@ int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
 	return 0;
 }
 
-void __kprobes get_instruction_type(struct arch_specific_insn *ainsn)
+static int __kprobes get_fixup_type(kprobe_opcode_t *insn)
 {
 	/* default fixup method */
-	ainsn->fixup = FIXUP_PSW_NORMAL;
-
-	/* save r1 operand */
-	ainsn->reg = (*ainsn->insn & 0xf0) >> 4;
+	int fixup = FIXUP_PSW_NORMAL;
 
-	/* save the instruction length (pop 5-5) in bytes */
-	switch (*(__u8 *) (ainsn->insn) >> 6) {
-	case 0:
-		ainsn->ilen = 2;
-		break;
-	case 1:
-	case 2:
-		ainsn->ilen = 4;
-		break;
-	case 3:
-		ainsn->ilen = 6;
-		break;
-	}
-
-	switch (*(__u8 *) ainsn->insn) {
+	switch (insn[0] >> 8) {
 	case 0x05:	/* balr	*/
 	case 0x0d:	/* basr */
-		ainsn->fixup = FIXUP_RETURN_REGISTER;
+		fixup = FIXUP_RETURN_REGISTER;
 		/* if r2 = 0, no branch will be taken */
-		if ((*ainsn->insn & 0x0f) == 0)
-			ainsn->fixup |= FIXUP_BRANCH_NOT_TAKEN;
+		if ((insn[0] & 0x0f) == 0)
+			fixup |= FIXUP_BRANCH_NOT_TAKEN;
 		break;
 	case 0x06:	/* bctr	*/
 	case 0x07:	/* bcr	*/
-		ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
+		fixup = FIXUP_BRANCH_NOT_TAKEN;
 		break;
 	case 0x45:	/* bal	*/
 	case 0x4d:	/* bas	*/
-		ainsn->fixup = FIXUP_RETURN_REGISTER;
+		fixup = FIXUP_RETURN_REGISTER;
 		break;
 	case 0x47:	/* bc	*/
 	case 0x46:	/* bct	*/
 	case 0x86:	/* bxh	*/
 	case 0x87:	/* bxle	*/
-		ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
+		fixup = FIXUP_BRANCH_NOT_TAKEN;
 		break;
 	case 0x82:	/* lpsw	*/
-		ainsn->fixup = FIXUP_NOT_REQUIRED;
+		fixup = FIXUP_NOT_REQUIRED;
 		break;
 	case 0xb2:	/* lpswe */
-		if (*(((__u8 *) ainsn->insn) + 1) == 0xb2) {
-			ainsn->fixup = FIXUP_NOT_REQUIRED;
-		}
+		if ((insn[0] & 0xff) == 0xb2)
+			fixup = FIXUP_NOT_REQUIRED;
 		break;
 	case 0xa7:	/* bras	*/
-		if ((*ainsn->insn & 0x0f) == 0x05) {
-			ainsn->fixup |= FIXUP_RETURN_REGISTER;
-		}
+		if ((insn[0] & 0x0f) == 0x05)
+			fixup |= FIXUP_RETURN_REGISTER;
 		break;
 	case 0xc0:
-		if ((*ainsn->insn & 0x0f) == 0x00  /* larl  */
-			|| (*ainsn->insn & 0x0f) == 0x05) /* brasl */
-		ainsn->fixup |= FIXUP_RETURN_REGISTER;
+		if ((insn[0] & 0x0f) == 0x00 ||	/* larl  */
+		    (insn[0] & 0x0f) == 0x05)	/* brasl */
+		fixup |= FIXUP_RETURN_REGISTER;
 		break;
 	case 0xeb:
-		if (*(((__u8 *) ainsn->insn) + 5 ) == 0x44 ||	/* bxhg  */
-			*(((__u8 *) ainsn->insn) + 5) == 0x45) {/* bxleg */
-			ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
-		}
+		if ((insn[2] & 0xff) == 0x44 ||	/* bxhg  */
+		    (insn[2] & 0xff) == 0x45)	/* bxleg */
+			fixup = FIXUP_BRANCH_NOT_TAKEN;
 		break;
 	case 0xe3:	/* bctg	*/
-		if (*(((__u8 *) ainsn->insn) + 5) == 0x46) {
-			ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
-		}
+		if ((insn[2] & 0xff) == 0x46)
+			fixup = FIXUP_BRANCH_NOT_TAKEN;
 		break;
 	}
+	return fixup;
+}
+
+int __kprobes arch_prepare_kprobe(struct kprobe *p)
+{
+	if ((unsigned long) p->addr & 0x01)
+		return -EINVAL;
+
+	/* Make sure the probe isn't going on a difficult instruction */
+	if (is_prohibited_opcode(p->addr))
+		return -EINVAL;
+
+	p->opcode = *p->addr;
+	memcpy(p->ainsn.insn, p->addr, ((p->opcode >> 14) + 3) & -2);
+
+	return 0;
 }
 
+struct ins_replace_args {
+	kprobe_opcode_t *ptr;
+	kprobe_opcode_t opcode;
+};
+
 static int __kprobes swap_instruction(void *aref)
 {
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
 	unsigned long status = kcb->kprobe_status;
 	struct ins_replace_args *args = aref;
-	int rc;
 
 	kcb->kprobe_status = KPROBE_SWAP_INST;
-	rc = probe_kernel_write(args->ptr, &args->new, sizeof(args->new));
+	probe_kernel_write(args->ptr, &args->opcode, sizeof(args->opcode));
 	kcb->kprobe_status = status;
-	return rc;
+	return 0;
 }
 
 void __kprobes arch_arm_kprobe(struct kprobe *p)
@@ -175,8 +154,7 @@ void __kprobes arch_arm_kprobe(struct kprobe *p)
 	struct ins_replace_args args;
 
 	args.ptr = p->addr;
-	args.old = p->opcode;
-	args.new = BREAKPOINT_INSTRUCTION;
+	args.opcode = BREAKPOINT_INSTRUCTION;
 	stop_machine(swap_instruction, &args, NULL);
 }
 
@@ -185,64 +163,69 @@ void __kprobes arch_disarm_kprobe(struct kprobe *p)
 	struct ins_replace_args args;
 
 	args.ptr = p->addr;
-	args.old = BREAKPOINT_INSTRUCTION;
-	args.new = p->opcode;
+	args.opcode = p->opcode;
 	stop_machine(swap_instruction, &args, NULL);
 }
 
 void __kprobes arch_remove_kprobe(struct kprobe *p)
 {
-	if (p->ainsn.insn) {
-		free_insn_slot(p->ainsn.insn, 0);
-		p->ainsn.insn = NULL;
-	}
 }
 
-static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
+static void __kprobes enable_singlestep(struct kprobe_ctlblk *kcb,
+					struct pt_regs *regs,
+					unsigned long ip)
 {
-	per_cr_bits kprobe_per_regs[1];
+	struct per_regs per_kprobe;
 
-	memset(kprobe_per_regs, 0, sizeof(per_cr_bits));
-	regs->psw.addr = (unsigned long)p->ainsn.insn | PSW_ADDR_AMODE;
+	/* Set up the PER control registers %cr9-%cr11 */
+	per_kprobe.control = PER_EVENT_IFETCH;
+	per_kprobe.start = ip;
+	per_kprobe.end = ip;
 
-	/* Set up the per control reg info, will pass to lctl */
-	kprobe_per_regs[0].em_instruction_fetch = 1;
-	kprobe_per_regs[0].starting_addr = (unsigned long)p->ainsn.insn;
-	kprobe_per_regs[0].ending_addr = (unsigned long)p->ainsn.insn + 1;
+	/* Save control regs and psw mask */
+	__ctl_store(kcb->kprobe_saved_ctl, 9, 11);
+	kcb->kprobe_saved_imask = regs->psw.mask &
+		(PSW_MASK_PER | PSW_MASK_IO | PSW_MASK_EXT);
 
-	/* Set the PER control regs, turns on single step for this address */
-	__ctl_load(kprobe_per_regs, 9, 11);
+	/* Set PER control regs, turns on single step for the given address */
+	__ctl_load(per_kprobe, 9, 11);
 	regs->psw.mask |= PSW_MASK_PER;
 	regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT);
+	regs->psw.addr = ip | PSW_ADDR_AMODE;
 }
 
-static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
+static void __kprobes disable_singlestep(struct kprobe_ctlblk *kcb,
+					 struct pt_regs *regs,
+					 unsigned long ip)
 {
-	kcb->prev_kprobe.kp = kprobe_running();
-	kcb->prev_kprobe.status = kcb->kprobe_status;
-	kcb->prev_kprobe.kprobe_saved_imask = kcb->kprobe_saved_imask;
-	memcpy(kcb->prev_kprobe.kprobe_saved_ctl, kcb->kprobe_saved_ctl,
-					sizeof(kcb->kprobe_saved_ctl));
+	/* Restore control regs and psw mask, set new psw address */
+	__ctl_load(kcb->kprobe_saved_ctl, 9, 11);
+	regs->psw.mask &= ~PSW_MASK_PER;
+	regs->psw.mask |= kcb->kprobe_saved_imask;
+	regs->psw.addr = ip | PSW_ADDR_AMODE;
 }
 
-static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
+/*
+ * Activate a kprobe by storing its pointer to current_kprobe. The
+ * previous kprobe is stored in kcb->prev_kprobe. A stack of up to
+ * two kprobes can be active, see KPROBE_REENTER.
+ */
+static void __kprobes push_kprobe(struct kprobe_ctlblk *kcb, struct kprobe *p)
 {
-	__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
-	kcb->kprobe_status = kcb->prev_kprobe.status;
-	kcb->kprobe_saved_imask = kcb->prev_kprobe.kprobe_saved_imask;
-	memcpy(kcb->kprobe_saved_ctl, kcb->prev_kprobe.kprobe_saved_ctl,
-					sizeof(kcb->kprobe_saved_ctl));
+	kcb->prev_kprobe.kp = __get_cpu_var(current_kprobe);
+	kcb->prev_kprobe.status = kcb->kprobe_status;
+	__get_cpu_var(current_kprobe) = p;
 }
 
-static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
-						struct kprobe_ctlblk *kcb)
+/*
+ * Deactivate a kprobe by backing up to the previous state. If the
+ * current state is KPROBE_REENTER prev_kprobe.kp will be non-NULL,
+ * for any other state prev_kprobe.kp will be NULL.
+ */
+static void __kprobes pop_kprobe(struct kprobe_ctlblk *kcb)
 {
-	__get_cpu_var(current_kprobe) = p;
-	/* Save the interrupt and per flags */
-	kcb->kprobe_saved_imask = regs->psw.mask &
-		(PSW_MASK_PER | PSW_MASK_IO | PSW_MASK_EXT);
-	/* Save the control regs that govern PER */
-	__ctl_store(kcb->kprobe_saved_ctl, 9, 11);
+	__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
+	kcb->kprobe_status = kcb->prev_kprobe.status;
 }
 
 void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
@@ -251,79 +234,104 @@ void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
 	ri->ret_addr = (kprobe_opcode_t *) regs->gprs[14];
 
 	/* Replace the return addr with trampoline addr */
-	regs->gprs[14] = (unsigned long)&kretprobe_trampoline;
+	regs->gprs[14] = (unsigned long) &kretprobe_trampoline;
+}
+
+static void __kprobes kprobe_reenter_check(struct kprobe_ctlblk *kcb,
+					   struct kprobe *p)
+{
+	switch (kcb->kprobe_status) {
+	case KPROBE_HIT_SSDONE:
+	case KPROBE_HIT_ACTIVE:
+		kprobes_inc_nmissed_count(p);
+		break;
+	case KPROBE_HIT_SS:
+	case KPROBE_REENTER:
+	default:
+		/*
+		 * A kprobe on the code path to single step an instruction
+		 * is a BUG. The code path resides in the .kprobes.text
+		 * section and is executed with interrupts disabled.
+		 */
+		printk(KERN_EMERG "Invalid kprobe detected at %p.\n", p->addr);
+		dump_kprobe(p);
+		BUG();
+	}
 }
 
 static int __kprobes kprobe_handler(struct pt_regs *regs)
 {
-	struct kprobe *p;
-	int ret = 0;
-	unsigned long *addr = (unsigned long *)
-		((regs->psw.addr & PSW_ADDR_INSN) - 2);
 	struct kprobe_ctlblk *kcb;
+	struct kprobe *p;
 
 	/*
-	 * We don't want to be preempted for the entire
-	 * duration of kprobe processing
+	 * We want to disable preemption for the entire duration of kprobe
+	 * processing. That includes the calls to the pre/post handlers
+	 * and single stepping the kprobe instruction.
 	 */
 	preempt_disable();
 	kcb = get_kprobe_ctlblk();
+	p = get_kprobe((void *)((regs->psw.addr & PSW_ADDR_INSN) - 2));
 
-	/* Check we're not actually recursing */
-	if (kprobe_running()) {
-		p = get_kprobe(addr);
-		if (p) {
-			if (kcb->kprobe_status == KPROBE_HIT_SS &&
-			    *p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
-				regs->psw.mask &= ~PSW_MASK_PER;
-				regs->psw.mask |= kcb->kprobe_saved_imask;
-				goto no_kprobe;
-			}
-			/* We have reentered the kprobe_handler(), since
-			 * another probe was hit while within the handler.
-			 * We here save the original kprobes variables and
-			 * just single step on the instruction of the new probe
-			 * without calling any user handlers.
+	if (p) {
+		if (kprobe_running()) {
+			/*
+			 * We have hit a kprobe while another is still
+			 * active. This can happen in the pre and post
+			 * handler. Single step the instruction of the
+			 * new probe but do not call any handler function
+			 * of this secondary kprobe.
+			 * push_kprobe and pop_kprobe saves and restores
+			 * the currently active kprobe.
 			 */
-			save_previous_kprobe(kcb);
-			set_current_kprobe(p, regs, kcb);
-			kprobes_inc_nmissed_count(p);
-			prepare_singlestep(p, regs);
+			kprobe_reenter_check(kcb, p);
+			push_kprobe(kcb, p);
 			kcb->kprobe_status = KPROBE_REENTER;
-			return 1;
 		} else {
-			p = __get_cpu_var(current_kprobe);
-			if (p->break_handler && p->break_handler(p, regs)) {
-				goto ss_probe;
-			}
+			/*
+			 * If we have no pre-handler or it returned 0, we
+			 * continue with single stepping. If we have a
+			 * pre-handler and it returned non-zero, it prepped
+			 * for calling the break_handler below on re-entry
+			 * for jprobe processing, so get out doing nothing
+			 * more here.
+			 */
+			push_kprobe(kcb, p);
+			kcb->kprobe_status = KPROBE_HIT_ACTIVE;
+			if (p->pre_handler && p->pre_handler(p, regs))
+				return 1;
+			kcb->kprobe_status = KPROBE_HIT_SS;
 		}
-		goto no_kprobe;
-	}
-
-	p = get_kprobe(addr);
-	if (!p)
-		/*
-		 * No kprobe at this address. The fault has not been
-		 * caused by a kprobe breakpoint. The race of breakpoint
-		 * vs. kprobe remove does not exist because on s390 we
-		 * use stop_machine to arm/disarm the breakpoints.
-		 */
-		goto no_kprobe;
-
-	kcb->kprobe_status = KPROBE_HIT_ACTIVE;
-	set_current_kprobe(p, regs, kcb);
-	if (p->pre_handler && p->pre_handler(p, regs))
-		/* handler has already set things up, so skip ss setup */
+		enable_singlestep(kcb, regs, (unsigned long) p->ainsn.insn);
 		return 1;
-
-ss_probe:
-	prepare_singlestep(p, regs);
-	kcb->kprobe_status = KPROBE_HIT_SS;
-	return 1;
-
-no_kprobe:
+	} else if (kprobe_running()) {
+		p = __get_cpu_var(current_kprobe);
+		if (p->break_handler && p->break_handler(p, regs)) {
+			/*
+			 * Continuation after the jprobe completed and
+			 * caused the jprobe_return trap. The jprobe
+			 * break_handler "returns" to the original
+			 * function that still has the kprobe breakpoint
+			 * installed. We continue with single stepping.
+			 */
+			kcb->kprobe_status = KPROBE_HIT_SS;
+			enable_singlestep(kcb, regs,
+					  (unsigned long) p->ainsn.insn);
+			return 1;
+		} /* else:
+		   * No kprobe at this address and the current kprobe
+		   * has no break handler (no jprobe!). The kernel just
+		   * exploded, let the standard trap handler pick up the
+		   * pieces.
+		   */
+	} /* else:
+	   * No kprobe at this address and no active kprobe. The trap has
+	   * not been caused by a kprobe breakpoint. The race of breakpoint
+	   * vs. kprobe remove does not exist because on s390 as we use
+	   * stop_machine to arm/disarm the breakpoints.
+	   */
 	preempt_enable_no_resched();
-	return ret;
+	return 0;
 }
 
 /*
@@ -344,12 +352,12 @@ static void __used kretprobe_trampoline_holder(void)
 static int __kprobes trampoline_probe_handler(struct kprobe *p,
 					      struct pt_regs *regs)
 {
-	struct kretprobe_instance *ri = NULL;
+	struct kretprobe_instance *ri;
 	struct hlist_head *head, empty_rp;
 	struct hlist_node *node, *tmp;
-	unsigned long flags, orig_ret_address = 0;
-	unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline;
-	kprobe_opcode_t *correct_ret_addr = NULL;
+	unsigned long flags, orig_ret_address;
+	unsigned long trampoline_address;
+	kprobe_opcode_t *correct_ret_addr;
 
 	INIT_HLIST_HEAD(&empty_rp);
 	kretprobe_hash_lock(current, &head, &flags);
@@ -367,12 +375,16 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
 	 *	 real return address, and all the rest will point to
 	 *	 kretprobe_trampoline
 	 */
+	ri = NULL;
+	orig_ret_address = 0;
+	correct_ret_addr = NULL;
+	trampoline_address = (unsigned long) &kretprobe_trampoline;
 	hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
 		if (ri->task != current)
 			/* another task is sharing our hash bucket */
 			continue;
 
-		orig_ret_address = (unsigned long)ri->ret_addr;
+		orig_ret_address = (unsigned long) ri->ret_addr;
 
 		if (orig_ret_address != trampoline_address)
 			/*
@@ -391,7 +403,7 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
 			/* another task is sharing our hash bucket */
 			continue;
 
-		orig_ret_address = (unsigned long)ri->ret_addr;
+		orig_ret_address = (unsigned long) ri->ret_addr;
 
 		if (ri->rp && ri->rp->handler) {
 			ri->ret_addr = correct_ret_addr;
@@ -400,19 +412,18 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
 
 		recycle_rp_inst(ri, &empty_rp);
 
-		if (orig_ret_address != trampoline_address) {
+		if (orig_ret_address != trampoline_address)
 			/*
 			 * This is the real return address. Any other
 			 * instances associated with this task are for
 			 * other calls deeper on the call stack
 			 */
 			break;
-		}
 	}
 
 	regs->psw.addr = orig_ret_address | PSW_ADDR_AMODE;
 
-	reset_current_kprobe();
+	pop_kprobe(get_kprobe_ctlblk());
 	kretprobe_hash_unlock(current, &flags);
 	preempt_enable_no_resched();
 
@@ -439,55 +450,42 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
 static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs)
 {
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+	unsigned long ip = regs->psw.addr & PSW_ADDR_INSN;
+	int fixup = get_fixup_type(p->ainsn.insn);
 
-	regs->psw.addr &= PSW_ADDR_INSN;
-
-	if (p->ainsn.fixup & FIXUP_PSW_NORMAL)
-		regs->psw.addr = (unsigned long)p->addr +
-				((unsigned long)regs->psw.addr -
-				 (unsigned long)p->ainsn.insn);
+	if (fixup & FIXUP_PSW_NORMAL)
+		ip += (unsigned long) p->addr - (unsigned long) p->ainsn.insn;
 
-	if (p->ainsn.fixup & FIXUP_BRANCH_NOT_TAKEN)
-		if ((unsigned long)regs->psw.addr -
-		    (unsigned long)p->ainsn.insn == p->ainsn.ilen)
-			regs->psw.addr = (unsigned long)p->addr + p->ainsn.ilen;
+	if (fixup & FIXUP_BRANCH_NOT_TAKEN) {
+		int ilen = ((p->ainsn.insn[0] >> 14) + 3) & -2;
+		if (ip - (unsigned long) p->ainsn.insn == ilen)
+			ip = (unsigned long) p->addr + ilen;
+	}
 
-	if (p->ainsn.fixup & FIXUP_RETURN_REGISTER)
-		regs->gprs[p->ainsn.reg] = ((unsigned long)p->addr +
-						(regs->gprs[p->ainsn.reg] -
-						(unsigned long)p->ainsn.insn))
-						| PSW_ADDR_AMODE;
+	if (fixup & FIXUP_RETURN_REGISTER) {
+		int reg = (p->ainsn.insn[0] & 0xf0) >> 4;
+		regs->gprs[reg] += (unsigned long) p->addr -
+				   (unsigned long) p->ainsn.insn;
+	}
 
-	regs->psw.addr |= PSW_ADDR_AMODE;
-	/* turn off PER mode */
-	regs->psw.mask &= ~PSW_MASK_PER;
-	/* Restore the original per control regs */
-	__ctl_load(kcb->kprobe_saved_ctl, 9, 11);
-	regs->psw.mask |= kcb->kprobe_saved_imask;
+	disable_singlestep(kcb, regs, ip);
 }
 
 static int __kprobes post_kprobe_handler(struct pt_regs *regs)
 {
-	struct kprobe *cur = kprobe_running();
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+	struct kprobe *p = kprobe_running();
 
-	if (!cur)
+	if (!p)
 		return 0;
 
-	if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
+	if (kcb->kprobe_status != KPROBE_REENTER && p->post_handler) {
 		kcb->kprobe_status = KPROBE_HIT_SSDONE;
-		cur->post_handler(cur, regs, 0);
+		p->post_handler(p, regs, 0);
 	}
 
-	resume_execution(cur, regs);
-
-	/*Restore back the original saved kprobes variables and continue. */
-	if (kcb->kprobe_status == KPROBE_REENTER) {
-		restore_previous_kprobe(kcb);
-		goto out;
-	}
-	reset_current_kprobe();
-out:
+	resume_execution(p, regs);
+	pop_kprobe(kcb);
 	preempt_enable_no_resched();
 
 	/*
@@ -495,17 +493,16 @@ out:
 	 * will have PER set, in which case, continue the remaining processing
 	 * of do_single_step, as if this is not a probe hit.
 	 */
-	if (regs->psw.mask & PSW_MASK_PER) {
+	if (regs->psw.mask & PSW_MASK_PER)
 		return 0;
-	}
 
 	return 1;
 }
 
 static int __kprobes kprobe_trap_handler(struct pt_regs *regs, int trapnr)
 {
-	struct kprobe *cur = kprobe_running();
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+	struct kprobe *p = kprobe_running();
 	const struct exception_table_entry *entry;
 
 	switch(kcb->kprobe_status) {
@@ -521,14 +518,8 @@ static int __kprobes kprobe_trap_handler(struct pt_regs *regs, int trapnr)
 		 * and allow the page fault handler to continue as a
 		 * normal page fault.
 		 */
-		regs->psw.addr = (unsigned long)cur->addr | PSW_ADDR_AMODE;
-		regs->psw.mask &= ~PSW_MASK_PER;
-		regs->psw.mask |= kcb->kprobe_saved_imask;
-		if (kcb->kprobe_status == KPROBE_REENTER)
-			restore_previous_kprobe(kcb);
-		else {
-			reset_current_kprobe();
-		}
+		disable_singlestep(kcb, regs, (unsigned long) p->addr);
+		pop_kprobe(kcb);
 		preempt_enable_no_resched();
 		break;
 	case KPROBE_HIT_ACTIVE:
@@ -538,7 +529,7 @@ static int __kprobes kprobe_trap_handler(struct pt_regs *regs, int trapnr)
 		 * we can also use npre/npostfault count for accouting
 		 * these specific fault cases.
 		 */
-		kprobes_inc_nmissed_count(cur);
+		kprobes_inc_nmissed_count(p);
 
 		/*
 		 * We come here because instructions in the pre/post
@@ -547,7 +538,7 @@ static int __kprobes kprobe_trap_handler(struct pt_regs *regs, int trapnr)
 		 * copy_from_user(), get_user() etc. Let the
 		 * user-specified handler try to fix it first.
 		 */
-		if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
+		if (p->fault_handler && p->fault_handler(p, regs, trapnr))
 			return 1;
 
 		/*
@@ -589,7 +580,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
 int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
 				       unsigned long val, void *data)
 {
-	struct die_args *args = (struct die_args *)data;
+	struct die_args *args = (struct die_args *) data;
 	struct pt_regs *regs = args->regs;
 	int ret = NOTIFY_DONE;
 
@@ -598,16 +589,16 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
 
 	switch (val) {
 	case DIE_BPT:
-		if (kprobe_handler(args->regs))
+		if (kprobe_handler(regs))
 			ret = NOTIFY_STOP;
 		break;
 	case DIE_SSTEP:
-		if (post_kprobe_handler(args->regs))
+		if (post_kprobe_handler(regs))
 			ret = NOTIFY_STOP;
 		break;
 	case DIE_TRAP:
 		if (!preemptible() && kprobe_running() &&
-		    kprobe_trap_handler(args->regs, args->trapnr))
+		    kprobe_trap_handler(regs, args->trapnr))
 			ret = NOTIFY_STOP;
 		break;
 	default:
@@ -623,23 +614,19 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
 int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
 {
 	struct jprobe *jp = container_of(p, struct jprobe, kp);
-	unsigned long addr;
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+	unsigned long stack;
 
 	memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs));
 
 	/* setup return addr to the jprobe handler routine */
-	regs->psw.addr = (unsigned long)(jp->entry) | PSW_ADDR_AMODE;
+	regs->psw.addr = (unsigned long) jp->entry | PSW_ADDR_AMODE;
 	regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT);
 
-	/* r14 is the function return address */
-	kcb->jprobe_saved_r14 = (unsigned long)regs->gprs[14];
 	/* r15 is the stack pointer */
-	kcb->jprobe_saved_r15 = (unsigned long)regs->gprs[15];
-	addr = (unsigned long)kcb->jprobe_saved_r15;
+	stack = (unsigned long) regs->gprs[15];
 
-	memcpy(kcb->jprobes_stack, (kprobe_opcode_t *) addr,
-	       MIN_STACK_SIZE(addr));
+	memcpy(kcb->jprobes_stack, (void *) stack, MIN_STACK_SIZE(stack));
 	return 1;
 }
 
@@ -656,30 +643,29 @@ void __kprobes jprobe_return_end(void)
 int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
 {
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
-	unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_r15);
+	unsigned long stack;
+
+	stack = (unsigned long) kcb->jprobe_saved_regs.gprs[15];
 
 	/* Put the regs back */
 	memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs));
 	/* put the stack back */
-	memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack,
-	       MIN_STACK_SIZE(stack_addr));
+	memcpy((void *) stack, kcb->jprobes_stack, MIN_STACK_SIZE(stack));
 	preempt_enable_no_resched();
 	return 1;
 }
 
-static struct kprobe trampoline_p = {
-	.addr = (kprobe_opcode_t *) & kretprobe_trampoline,
+static struct kprobe trampoline = {
+	.addr = (kprobe_opcode_t *) &kretprobe_trampoline,
 	.pre_handler = trampoline_probe_handler
 };
 
 int __init arch_init_kprobes(void)
 {
-	return register_kprobe(&trampoline_p);
+	return register_kprobe(&trampoline);
 }
 
 int __kprobes arch_trampoline_kprobe(struct kprobe *p)
 {
-	if (p->addr == (kprobe_opcode_t *) & kretprobe_trampoline)
-		return 1;
-	return 0;
+	return p->addr == (kprobe_opcode_t *) &kretprobe_trampoline;
 }

arch/s390/kernel/mcount.S

@@ -7,6 +7,8 @@
 
 #include <asm/asm-offsets.h>
 
+	.section .kprobes.text, "ax"
+
 	.globl ftrace_stub
 ftrace_stub:
 	br	%r14
@@ -16,22 +18,12 @@ _mcount:
 #ifdef CONFIG_DYNAMIC_FTRACE
 	br	%r14
 
-	.data
-	.globl	ftrace_dyn_func
-ftrace_dyn_func:
-	.long	ftrace_stub
-	.previous
-
 	.globl ftrace_caller
 ftrace_caller:
 #endif
 	stm	%r2,%r5,16(%r15)
 	bras	%r1,2f
-#ifdef CONFIG_DYNAMIC_FTRACE
-0:	.long	ftrace_dyn_func
-#else
 0:	.long	ftrace_trace_function
-#endif
 1:	.long	function_trace_stop
 2:	l	%r2,1b-0b(%r1)
 	icm	%r2,0xf,0(%r2)
@@ -47,21 +39,15 @@ ftrace_caller:
 	l	%r14,0(%r14)
 	basr	%r14,%r14
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
-#ifdef CONFIG_DYNAMIC_FTRACE
+	l	%r2,100(%r15)
+	l	%r3,152(%r15)
 	.globl	ftrace_graph_caller
 ftrace_graph_caller:
-	# This unconditional branch gets runtime patched. Change only if
-	# you know what you are doing. See ftrace_enable_graph_caller().
-	j	1f
-#endif
-	bras	%r1,0f
-	.long	prepare_ftrace_return
-0:	l	%r2,152(%r15)
-	l	%r4,0(%r1)
-	l	%r3,100(%r15)
-	basr	%r14,%r4
-	st	%r2,100(%r15)
-1:
+# The bras instruction gets runtime patched to call prepare_ftrace_return.
+# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
+#	bras	%r14,prepare_ftrace_return
+	bras	%r14,0f
+0:	st	%r2,100(%r15)
 #endif
 	ahi	%r15,96
 	l	%r14,56(%r15)

arch/s390/kernel/mcount64.S

@@ -7,6 +7,8 @@
 
 #include <asm/asm-offsets.h>
 
+	.section .kprobes.text, "ax"
+
 	.globl ftrace_stub
 ftrace_stub:
 	br	%r14
@@ -16,12 +18,6 @@ _mcount:
 #ifdef CONFIG_DYNAMIC_FTRACE
 	br	%r14
 
-	.data
-	.globl	ftrace_dyn_func
-ftrace_dyn_func:
-	.quad	ftrace_stub
-	.previous
-
 	.globl ftrace_caller
 ftrace_caller:
 #endif
@@ -35,26 +31,19 @@ ftrace_caller:
 	stg	%r1,__SF_BACKCHAIN(%r15)
 	lgr	%r2,%r14
 	lg	%r3,168(%r15)
-#ifdef CONFIG_DYNAMIC_FTRACE
-	larl	%r14,ftrace_dyn_func
-#else
 	larl	%r14,ftrace_trace_function
-#endif
 	lg	%r14,0(%r14)
 	basr	%r14,%r14
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
-#ifdef CONFIG_DYNAMIC_FTRACE
+	lg	%r2,168(%r15)
+	lg	%r3,272(%r15)
 	.globl	ftrace_graph_caller
 ftrace_graph_caller:
-	# This unconditional branch gets runtime patched. Change only if
-	# you know what you are doing. See ftrace_enable_graph_caller().
-	j	0f
-#endif
-	lg	%r2,272(%r15)
-	lg	%r3,168(%r15)
-	brasl	%r14,prepare_ftrace_return
-	stg	%r2,168(%r15)
-0:
+# The bras instruction gets runtime patched to call prepare_ftrace_return.
+# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
+#	bras	%r14,prepare_ftrace_return
+	bras	%r14,0f
+0:	stg	%r2,168(%r15)
 #endif
 	aghi	%r15,160
 	lmg	%r2,%r5,32(%r15)

arch/s390/kernel/nmi.c

@@ -8,6 +8,7 @@
  *		 Heiko Carstens <heiko.carstens@de.ibm.com>,
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/init.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
@@ -255,7 +256,7 @@ void notrace s390_do_machine_check(struct pt_regs *regs)
 	nmi_enter();
 	s390_idle_check(regs, S390_lowcore.mcck_clock,
 			S390_lowcore.mcck_enter_timer);
-
+	kstat_cpu(smp_processor_id()).irqs[NMI_NMI]++;
 	mci = (struct mci *) &S390_lowcore.mcck_interruption_code;
 	mcck = &__get_cpu_var(cpu_mcck);
 	umode = user_mode(regs);

arch/s390/kernel/process.c

@@ -32,6 +32,7 @@
 #include <linux/kernel_stat.h>
 #include <linux/syscalls.h>
 #include <linux/compat.h>
+#include <linux/kprobes.h>
 #include <asm/compat.h>
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -41,6 +42,7 @@
 #include <asm/irq.h>
 #include <asm/timer.h>
 #include <asm/nmi.h>
+#include <asm/smp.h>
 #include "entry.h"
 
 asmlinkage void ret_from_fork(void) asm ("ret_from_fork");
@@ -75,13 +77,8 @@ unsigned long thread_saved_pc(struct task_struct *tsk)
  */
 static void default_idle(void)
 {
-	/* CPU is going idle. */
-#ifdef CONFIG_HOTPLUG_CPU
-	if (cpu_is_offline(smp_processor_id())) {
-		preempt_enable_no_resched();
+	if (cpu_is_offline(smp_processor_id()))
 		cpu_die();
-	}
-#endif
 	local_irq_disable();
 	if (need_resched()) {
 		local_irq_enable();
@@ -116,15 +113,17 @@ void cpu_idle(void)
 	}
 }
 
-extern void kernel_thread_starter(void);
+extern void __kprobes kernel_thread_starter(void);
 
 asm(
-	".align 4\n"
+	".section .kprobes.text, \"ax\"\n"
+	".global kernel_thread_starter\n"
 	"kernel_thread_starter:\n"
 	"    la    2,0(10)\n"
 	"    basr  14,9\n"
 	"    la    2,0\n"
-	"    br    11\n");
+	"    br    11\n"
+	".previous\n");
 
 int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
 {
@@ -214,8 +213,10 @@ int copy_thread(unsigned long clone_flags, unsigned long new_stackp,
 	/* start new process with ar4 pointing to the correct address space */
 	p->thread.mm_segment = get_fs();
 	/* Don't copy debug registers */
-	memset(&p->thread.per_info, 0, sizeof(p->thread.per_info));
+	memset(&p->thread.per_user, 0, sizeof(p->thread.per_user));
+	memset(&p->thread.per_event, 0, sizeof(p->thread.per_event));
 	clear_tsk_thread_flag(p, TIF_SINGLE_STEP);
+	clear_tsk_thread_flag(p, TIF_PER_TRAP);
 	/* Initialize per thread user and system timer values */
 	ti = task_thread_info(p);
 	ti->user_timer = 0;

arch/s390/kernel/processor.c

@@ -13,7 +13,7 @@
 #include <linux/smp.h>
 #include <linux/seq_file.h>
 #include <linux/delay.h>
-
+#include <linux/cpu.h>
 #include <asm/elf.h>
 #include <asm/lowcore.h>
 #include <asm/param.h>
@@ -34,17 +34,6 @@ void __cpuinit cpu_init(void)
 	enter_lazy_tlb(&init_mm, current);
 }
 
-/*
- * print_cpu_info - print basic information about a cpu
- */
-void __cpuinit print_cpu_info(void)
-{
-	struct cpuid *id = &per_cpu(cpu_id, smp_processor_id());
-
-	pr_info("Processor %d started, address %d, identification %06X\n",
-		S390_lowcore.cpu_nr, stap(), id->ident);
-}
-
 /*
  * show_cpuinfo - Get information on one CPU for use by procfs.
  */
@@ -57,9 +46,8 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 	unsigned long n = (unsigned long) v - 1;
 	int i;
 
-	s390_adjust_jiffies();
-	preempt_disable();
 	if (!n) {
+		s390_adjust_jiffies();
 		seq_printf(m, "vendor_id       : IBM/S390\n"
 			   "# processors    : %i\n"
 			   "bogomips per cpu: %lu.%02lu\n",
@@ -71,7 +59,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 				seq_printf(m, "%s ", hwcap_str[i]);
 		seq_puts(m, "\n");
 	}
-
+	get_online_cpus();
 	if (cpu_online(n)) {
 		struct cpuid *id = &per_cpu(cpu_id, n);
 		seq_printf(m, "processor %li: "
@@ -80,7 +68,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 			   "machine = %04X\n",
 			   n, id->version, id->ident, id->machine);
 	}
-	preempt_enable();
+	put_online_cpus();
 	return 0;
 }
 

arch/s390/kernel/ptrace.c

@@ -1,25 +1,9 @@
 /*
- *  arch/s390/kernel/ptrace.c
+ *  Ptrace user space interface.
  *
- *  S390 version
- *    Copyright (C) 1999,2000 IBM Deutschland Entwicklung GmbH, IBM Corporation
- *    Author(s): Denis Joseph Barrow (djbarrow@de.ibm.com,barrow_dj@yahoo.com),
+ *    Copyright IBM Corp. 1999,2010
+ *    Author(s): Denis Joseph Barrow
  *               Martin Schwidefsky (schwidefsky@de.ibm.com)
- *
- *  Based on PowerPC version 
- *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
- *
- *  Derived from "arch/m68k/kernel/ptrace.c"
- *  Copyright (C) 1994 by Hamish Macdonald
- *  Taken from linux/kernel/ptrace.c and modified for M680x0.
- *  linux/kernel/ptrace.c is by Ross Biro 1/23/92, edited by Linus Torvalds
- *
- * Modified by Cort Dougan (cort@cs.nmt.edu) 
- *
- *
- * This file is subject to the terms and conditions of the GNU General
- * Public License.  See the file README.legal in the main directory of
- * this archive for more details.
  */
 
 #include <linux/kernel.h>
@@ -61,76 +45,58 @@ enum s390_regset {
 	REGSET_GENERAL_EXTENDED,
 };
 
-static void
-FixPerRegisters(struct task_struct *task)
+void update_per_regs(struct task_struct *task)
 {
-	struct pt_regs *regs;
-	per_struct *per_info;
-	per_cr_words cr_words;
-
-	regs = task_pt_regs(task);
-	per_info = (per_struct *) &task->thread.per_info;
-	per_info->control_regs.bits.em_instruction_fetch =
-		per_info->single_step | per_info->instruction_fetch;
-	
-	if (per_info->single_step) {
-		per_info->control_regs.bits.starting_addr = 0;
-#ifdef CONFIG_COMPAT
-		if (is_compat_task())
-			per_info->control_regs.bits.ending_addr = 0x7fffffffUL;
-		else
-#endif
-			per_info->control_regs.bits.ending_addr = PSW_ADDR_INSN;
-	} else {
-		per_info->control_regs.bits.starting_addr =
-			per_info->starting_addr;
-		per_info->control_regs.bits.ending_addr =
-			per_info->ending_addr;
-	}
-	/*
-	 * if any of the control reg tracing bits are on 
-	 * we switch on per in the psw
-	 */
-	if (per_info->control_regs.words.cr[0] & PER_EM_MASK)
-		regs->psw.mask |= PSW_MASK_PER;
-	else
+	static const struct per_regs per_single_step = {
+		.control = PER_EVENT_IFETCH,
+		.start = 0,
+		.end = PSW_ADDR_INSN,
+	};
+	struct pt_regs *regs = task_pt_regs(task);
+	struct thread_struct *thread = &task->thread;
+	const struct per_regs *new;
+	struct per_regs old;
+
+	/* TIF_SINGLE_STEP overrides the user specified PER registers. */
+	new = test_tsk_thread_flag(task, TIF_SINGLE_STEP) ?
+		&per_single_step : &thread->per_user;
+
+	/* Take care of the PER enablement bit in the PSW. */
+	if (!(new->control & PER_EVENT_MASK)) {
 		regs->psw.mask &= ~PSW_MASK_PER;
-
-	if (per_info->control_regs.bits.em_storage_alteration)
-		per_info->control_regs.bits.storage_alt_space_ctl = 1;
-	else
-		per_info->control_regs.bits.storage_alt_space_ctl = 0;
-
-	if (task == current) {
-		__ctl_store(cr_words, 9, 11);
-		if (memcmp(&cr_words, &per_info->control_regs.words,
-			   sizeof(cr_words)) != 0)
-			__ctl_load(per_info->control_regs.words, 9, 11);
+		return;
 	}
+	regs->psw.mask |= PSW_MASK_PER;
+	__ctl_store(old, 9, 11);
+	if (memcmp(new, &old, sizeof(struct per_regs)) != 0)
+		__ctl_load(*new, 9, 11);
 }
 
 void user_enable_single_step(struct task_struct *task)
 {
-	task->thread.per_info.single_step = 1;
-	FixPerRegisters(task);
+	set_tsk_thread_flag(task, TIF_SINGLE_STEP);
+	if (task == current)
+		update_per_regs(task);
 }
 
 void user_disable_single_step(struct task_struct *task)
 {
-	task->thread.per_info.single_step = 0;
-	FixPerRegisters(task);
+	clear_tsk_thread_flag(task, TIF_SINGLE_STEP);
+	if (task == current)
+		update_per_regs(task);
 }
 
 /*
  * Called by kernel/ptrace.c when detaching..
  *
- * Make sure single step bits etc are not set.
+ * Clear all debugging related fields.
  */
-void
-ptrace_disable(struct task_struct *child)
+void ptrace_disable(struct task_struct *task)
 {
-	/* make sure the single step bit is not set. */
-	user_disable_single_step(child);
+	memset(&task->thread.per_user, 0, sizeof(task->thread.per_user));
+	memset(&task->thread.per_event, 0, sizeof(task->thread.per_event));
+	clear_tsk_thread_flag(task, TIF_SINGLE_STEP);
+	clear_tsk_thread_flag(task, TIF_PER_TRAP);
 }
 
 #ifndef CONFIG_64BIT
@@ -139,6 +105,47 @@ ptrace_disable(struct task_struct *child)
 # define __ADDR_MASK 7
 #endif
 
+static inline unsigned long __peek_user_per(struct task_struct *child,
+					    addr_t addr)
+{
+	struct per_struct_kernel *dummy = NULL;
+
+	if (addr == (addr_t) &dummy->cr9)
+		/* Control bits of the active per set. */
+		return test_thread_flag(TIF_SINGLE_STEP) ?
+			PER_EVENT_IFETCH : child->thread.per_user.control;
+	else if (addr == (addr_t) &dummy->cr10)
+		/* Start address of the active per set. */
+		return test_thread_flag(TIF_SINGLE_STEP) ?
+			0 : child->thread.per_user.start;
+	else if (addr == (addr_t) &dummy->cr11)
+		/* End address of the active per set. */
+		return test_thread_flag(TIF_SINGLE_STEP) ?
+			PSW_ADDR_INSN : child->thread.per_user.end;
+	else if (addr == (addr_t) &dummy->bits)
+		/* Single-step bit. */
+		return test_thread_flag(TIF_SINGLE_STEP) ?
+			(1UL << (BITS_PER_LONG - 1)) : 0;
+	else if (addr == (addr_t) &dummy->starting_addr)
+		/* Start address of the user specified per set. */
+		return child->thread.per_user.start;
+	else if (addr == (addr_t) &dummy->ending_addr)
+		/* End address of the user specified per set. */
+		return child->thread.per_user.end;
+	else if (addr == (addr_t) &dummy->perc_atmid)
+		/* PER code, ATMID and AI of the last PER trap */
+		return (unsigned long)
+			child->thread.per_event.cause << (BITS_PER_LONG - 16);
+	else if (addr == (addr_t) &dummy->address)
+		/* Address of the last PER trap */
+		return child->thread.per_event.address;
+	else if (addr == (addr_t) &dummy->access_id)
+		/* Access id of the last PER trap */
+		return (unsigned long)
+			child->thread.per_event.paid << (BITS_PER_LONG - 8);
+	return 0;
+}
+
 /*
  * Read the word at offset addr from the user area of a process. The
  * trouble here is that the information is littered over different
@@ -204,10 +211,10 @@ static unsigned long __peek_user(struct task_struct *child, addr_t addr)
 
 	} else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
 		/*
-		 * per_info is found in the thread structure
+		 * Handle access to the per_info structure.
 		 */
-		offset = addr - (addr_t) &dummy->regs.per_info;
-		tmp = *(addr_t *)((addr_t) &child->thread.per_info + offset);
+		addr -= (addr_t) &dummy->regs.per_info;
+		tmp = __peek_user_per(child, addr);
 
 	} else
 		tmp = 0;
@@ -237,6 +244,35 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data)
 	return put_user(tmp, (addr_t __user *) data);
 }
 
+static inline void __poke_user_per(struct task_struct *child,
+				   addr_t addr, addr_t data)
+{
+	struct per_struct_kernel *dummy = NULL;
+
+	/*
+	 * There are only three fields in the per_info struct that the
+	 * debugger user can write to.
+	 * 1) cr9: the debugger wants to set a new PER event mask
+	 * 2) starting_addr: the debugger wants to set a new starting
+	 *    address to use with the PER event mask.
+	 * 3) ending_addr: the debugger wants to set a new ending
+	 *    address to use with the PER event mask.
+	 * The user specified PER event mask and the start and end
+	 * addresses are used only if single stepping is not in effect.
+	 * Writes to any other field in per_info are ignored.
+	 */
+	if (addr == (addr_t) &dummy->cr9)
+		/* PER event mask of the user specified per set. */
+		child->thread.per_user.control =
+			data & (PER_EVENT_MASK | PER_CONTROL_MASK);
+	else if (addr == (addr_t) &dummy->starting_addr)
+		/* Starting address of the user specified per set. */
+		child->thread.per_user.start = data;
+	else if (addr == (addr_t) &dummy->ending_addr)
+		/* Ending address of the user specified per set. */
+		child->thread.per_user.end = data;
+}
+
 /*
  * Write a word to the user area of a process at location addr. This
  * operation does have an additional problem compared to peek_user.
@@ -311,19 +347,17 @@ static int __poke_user(struct task_struct *child, addr_t addr, addr_t data)
 
 	} else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
 		/*
-		 * per_info is found in the thread structure 
+		 * Handle access to the per_info structure.
 		 */
-		offset = addr - (addr_t) &dummy->regs.per_info;
-		*(addr_t *)((addr_t) &child->thread.per_info + offset) = data;
+		addr -= (addr_t) &dummy->regs.per_info;
+		__poke_user_per(child, addr, data);
 
 	}
 
-	FixPerRegisters(child);
 	return 0;
 }
 
-static int
-poke_user(struct task_struct *child, addr_t addr, addr_t data)
+static int poke_user(struct task_struct *child, addr_t addr, addr_t data)
 {
 	addr_t mask;
 
@@ -409,13 +443,54 @@ long arch_ptrace(struct task_struct *child, long request,
  * a 64 bit program is a no-no.
  */
 
+/*
+ * Same as peek_user_per but for a 31 bit program.
+ */
+static inline __u32 __peek_user_per_compat(struct task_struct *child,
+					   addr_t addr)
+{
+	struct compat_per_struct_kernel *dummy32 = NULL;
+
+	if (addr == (addr_t) &dummy32->cr9)
+		/* Control bits of the active per set. */
+		return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
+			PER_EVENT_IFETCH : child->thread.per_user.control;
+	else if (addr == (addr_t) &dummy32->cr10)
+		/* Start address of the active per set. */
+		return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
+			0 : child->thread.per_user.start;
+	else if (addr == (addr_t) &dummy32->cr11)
+		/* End address of the active per set. */
+		return test_thread_flag(TIF_SINGLE_STEP) ?
+			PSW32_ADDR_INSN : child->thread.per_user.end;
+	else if (addr == (addr_t) &dummy32->bits)
+		/* Single-step bit. */
+		return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
+			0x80000000 : 0;
+	else if (addr == (addr_t) &dummy32->starting_addr)
+		/* Start address of the user specified per set. */
+		return (__u32) child->thread.per_user.start;
+	else if (addr == (addr_t) &dummy32->ending_addr)
+		/* End address of the user specified per set. */
+		return (__u32) child->thread.per_user.end;
+	else if (addr == (addr_t) &dummy32->perc_atmid)
+		/* PER code, ATMID and AI of the last PER trap */
+		return (__u32) child->thread.per_event.cause << 16;
+	else if (addr == (addr_t) &dummy32->address)
+		/* Address of the last PER trap */
+		return (__u32) child->thread.per_event.address;
+	else if (addr == (addr_t) &dummy32->access_id)
+		/* Access id of the last PER trap */
+		return (__u32) child->thread.per_event.paid << 24;
+	return 0;
+}
+
 /*
  * Same as peek_user but for a 31 bit program.
  */
 static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
 {
-	struct user32 *dummy32 = NULL;
-	per_struct32 *dummy_per32 = NULL;
+	struct compat_user *dummy32 = NULL;
 	addr_t offset;
 	__u32 tmp;
 
@@ -465,19 +540,10 @@ static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
 
 	} else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
 		/*
-		 * per_info is found in the thread structure
+		 * Handle access to the per_info structure.
 		 */
-		offset = addr - (addr_t) &dummy32->regs.per_info;
-		/* This is magic. See per_struct and per_struct32. */
-		if ((offset >= (addr_t) &dummy_per32->control_regs &&
-		     offset < (addr_t) (&dummy_per32->control_regs + 1)) ||
-		    (offset >= (addr_t) &dummy_per32->starting_addr &&
-		     offset <= (addr_t) &dummy_per32->ending_addr) ||
-		    offset == (addr_t) &dummy_per32->lowcore.words.address)
-			offset = offset*2 + 4;
-		else
-			offset = offset*2;
-		tmp = *(__u32 *)((addr_t) &child->thread.per_info + offset);
+		addr -= (addr_t) &dummy32->regs.per_info;
+		tmp = __peek_user_per_compat(child, addr);
 
 	} else
 		tmp = 0;
@@ -497,14 +563,33 @@ static int peek_user_compat(struct task_struct *child,
 	return put_user(tmp, (__u32 __user *) data);
 }
 
+/*
+ * Same as poke_user_per but for a 31 bit program.
+ */
+static inline void __poke_user_per_compat(struct task_struct *child,
+					  addr_t addr, __u32 data)
+{
+	struct compat_per_struct_kernel *dummy32 = NULL;
+
+	if (addr == (addr_t) &dummy32->cr9)
+		/* PER event mask of the user specified per set. */
+		child->thread.per_user.control =
+			data & (PER_EVENT_MASK | PER_CONTROL_MASK);
+	else if (addr == (addr_t) &dummy32->starting_addr)
+		/* Starting address of the user specified per set. */
+		child->thread.per_user.start = data;
+	else if (addr == (addr_t) &dummy32->ending_addr)
+		/* Ending address of the user specified per set. */
+		child->thread.per_user.end = data;
+}
+
 /*
  * Same as poke_user but for a 31 bit program.
  */
 static int __poke_user_compat(struct task_struct *child,
 			      addr_t addr, addr_t data)
 {
-	struct user32 *dummy32 = NULL;
-	per_struct32 *dummy_per32 = NULL;
+	struct compat_user *dummy32 = NULL;
 	__u32 tmp = (__u32) data;
 	addr_t offset;
 
@@ -561,37 +646,20 @@ static int __poke_user_compat(struct task_struct *child,
 
 	} else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
 		/*
-		 * per_info is found in the thread structure.
-		 */
-		offset = addr - (addr_t) &dummy32->regs.per_info;
-		/*
-		 * This is magic. See per_struct and per_struct32.
-		 * By incident the offsets in per_struct are exactly
-		 * twice the offsets in per_struct32 for all fields.
-		 * The 8 byte fields need special handling though,
-		 * because the second half (bytes 4-7) is needed and
-		 * not the first half.
+		 * Handle access to the per_info structure.
 		 */
-		if ((offset >= (addr_t) &dummy_per32->control_regs &&
-		     offset < (addr_t) (&dummy_per32->control_regs + 1)) ||
-		    (offset >= (addr_t) &dummy_per32->starting_addr &&
-		     offset <= (addr_t) &dummy_per32->ending_addr) ||
-		    offset == (addr_t) &dummy_per32->lowcore.words.address)
-			offset = offset*2 + 4;
-		else
-			offset = offset*2;
-		*(__u32 *)((addr_t) &child->thread.per_info + offset) = tmp;
-
+		addr -= (addr_t) &dummy32->regs.per_info;
+		__poke_user_per_compat(child, addr, data);
 	}
 
-	FixPerRegisters(child);
 	return 0;
 }
 
 static int poke_user_compat(struct task_struct *child,
 			    addr_t addr, addr_t data)
 {
-	if (!is_compat_task() || (addr & 3) || addr > sizeof(struct user32) - 3)
+	if (!is_compat_task() || (addr & 3) ||
+	    addr > sizeof(struct compat_user) - 3)
 		return -EIO;
 
 	return __poke_user_compat(child, addr, data);
@@ -602,7 +670,7 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
 {
 	unsigned long addr = caddr;
 	unsigned long data = cdata;
-	ptrace_area_emu31 parea; 
+	compat_ptrace_area parea;
 	int copied, ret;
 
 	switch (request) {

arch/s390/kernel/s390_ext.c

@@ -1,33 +1,36 @@
 /*
- *  arch/s390/kernel/s390_ext.c
- *
- *  S390 version
- *    Copyright (C) 1999,2000 IBM Deutschland Entwicklung GmbH, IBM Corporation
- *    Author(s): Holger Smolinski (Holger.Smolinski@de.ibm.com),
- *               Martin Schwidefsky (schwidefsky@de.ibm.com)
+ *    Copyright IBM Corp. 1999,2010
+ *    Author(s): Holger Smolinski <Holger.Smolinski@de.ibm.com>,
+ *		 Martin Schwidefsky <schwidefsky@de.ibm.com>,
  */
 
+#include <linux/kernel_stat.h>
+#include <linux/interrupt.h>
 #include <linux/module.h>
 #include <linux/kernel.h>
-#include <linux/slab.h>
 #include <linux/ftrace.h>
 #include <linux/errno.h>
-#include <linux/kernel_stat.h>
-#include <linux/interrupt.h>
-#include <asm/cputime.h>
-#include <asm/lowcore.h>
+#include <linux/slab.h>
 #include <asm/s390_ext.h>
 #include <asm/irq_regs.h>
+#include <asm/cputime.h>
+#include <asm/lowcore.h>
 #include <asm/irq.h>
 #include "entry.h"
 
+struct ext_int_info {
+	struct ext_int_info *next;
+	ext_int_handler_t handler;
+	__u16 code;
+};
+
 /*
  * ext_int_hash[index] is the start of the list for all external interrupts
  * that hash to this index. With the current set of external interrupts 
  * (0x1202 external call, 0x1004 cpu timer, 0x2401 hwc console, 0x4000
  * iucv and 0x2603 pfault) this is always the first element. 
  */
-ext_int_info_t *ext_int_hash[256] = { NULL, };
+static struct ext_int_info *ext_int_hash[256];
 
 static inline int ext_hash(__u16 code)
 {
@@ -36,90 +39,53 @@ static inline int ext_hash(__u16 code)
 
 int register_external_interrupt(__u16 code, ext_int_handler_t handler)
 {
-        ext_int_info_t *p;
-        int index;
-
-	p = kmalloc(sizeof(ext_int_info_t), GFP_ATOMIC);
-        if (p == NULL)
-                return -ENOMEM;
-        p->code = code;
-        p->handler = handler;
-	index = ext_hash(code);
-        p->next = ext_int_hash[index];
-        ext_int_hash[index] = p;
-        return 0;
-}
-
-int register_early_external_interrupt(__u16 code, ext_int_handler_t handler,
-				      ext_int_info_t *p)
-{
-        int index;
+	struct ext_int_info *p;
+	int index;
 
-        if (p == NULL)
-                return -EINVAL;
-        p->code = code;
-        p->handler = handler;
+	p = kmalloc(sizeof(*p), GFP_ATOMIC);
+	if (!p)
+		return -ENOMEM;
+	p->code = code;
+	p->handler = handler;
 	index = ext_hash(code);
-        p->next = ext_int_hash[index];
-        ext_int_hash[index] = p;
-        return 0;
+	p->next = ext_int_hash[index];
+	ext_int_hash[index] = p;
+	return 0;
 }
+EXPORT_SYMBOL(register_external_interrupt);
 
 int unregister_external_interrupt(__u16 code, ext_int_handler_t handler)
 {
-        ext_int_info_t *p, *q;
-        int index;
-
-	index = ext_hash(code);
-        q = NULL;
-        p = ext_int_hash[index];
-        while (p != NULL) {
-                if (p->code == code && p->handler == handler)
-                        break;
-                q = p;
-                p = p->next;
-        }
-        if (p == NULL)
-                return -ENOENT;
-        if (q != NULL)
-                q->next = p->next;
-        else
-                ext_int_hash[index] = p->next;
-	kfree(p);
-        return 0;
-}
-
-int unregister_early_external_interrupt(__u16 code, ext_int_handler_t handler,
-					ext_int_info_t *p)
-{
-	ext_int_info_t *q;
+	struct ext_int_info *p, *q;
 	int index;
 
-	if (p == NULL || p->code != code || p->handler != handler)
-		return -EINVAL;
 	index = ext_hash(code);
-	q = ext_int_hash[index];
-	if (p != q) {
-		while (q != NULL) {
-			if (q->next == p)
-				break;
-			q = q->next;
-		}
-		if (q == NULL)
-			return -ENOENT;
+	q = NULL;
+	p = ext_int_hash[index];
+	while (p) {
+		if (p->code == code && p->handler == handler)
+			break;
+		q = p;
+		p = p->next;
+	}
+	if (!p)
+		return -ENOENT;
+	if (q)
 		q->next = p->next;
-	} else
+	else
 		ext_int_hash[index] = p->next;
+	kfree(p);
 	return 0;
 }
+EXPORT_SYMBOL(unregister_external_interrupt);
 
 void __irq_entry do_extint(struct pt_regs *regs, unsigned int ext_int_code,
 			   unsigned int param32, unsigned long param64)
 {
 	struct pt_regs *old_regs;
 	unsigned short code;
-        ext_int_info_t *p;
-        int index;
+	struct ext_int_info *p;
+	int index;
 
 	code = (unsigned short) ext_int_code;
 	old_regs = set_irq_regs(regs);
@@ -132,7 +98,7 @@ void __irq_entry do_extint(struct pt_regs *regs, unsigned int ext_int_code,
 	kstat_cpu(smp_processor_id()).irqs[EXTERNAL_INTERRUPT]++;
 	if (code != 0x1004)
 		__get_cpu_var(s390_idle).nohz_delay = 1;
-        index = ext_hash(code);
+	index = ext_hash(code);
 	for (p = ext_int_hash[index]; p; p = p->next) {
 		if (likely(p->code == code))
 			p->handler(ext_int_code, param32, param64);
@@ -140,6 +106,3 @@ void __irq_entry do_extint(struct pt_regs *regs, unsigned int ext_int_code,
 	irq_exit();
 	set_irq_regs(old_regs);
 }
-
-EXPORT_SYMBOL(register_external_interrupt);
-EXPORT_SYMBOL(unregister_external_interrupt);

arch/s390/kernel/signal.c

@@ -505,7 +505,7 @@ void do_signal(struct pt_regs *regs)
 			 * Let tracing know that we've done the handler setup.
 			 */
 			tracehook_signal_handler(signr, &info, &ka, regs,
-					current->thread.per_info.single_step);
+					test_thread_flag(TIF_SINGLE_STEP));
 		}
 		return;
 	}

arch/s390/kernel/smp.c

@@ -23,6 +23,7 @@
 #define KMSG_COMPONENT "cpu"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/workqueue.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/mm.h>
@@ -161,6 +162,7 @@ static void do_ext_call_interrupt(unsigned int ext_int_code,
 {
 	unsigned long bits;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_IPI]++;
 	/*
 	 * handle bit signal external calls
 	 *
@@ -469,25 +471,25 @@ int __cpuinit start_secondary(void *cpuvoid)
 	ipi_call_unlock();
 	/* Switch on interrupts */
 	local_irq_enable();
-	/* Print info about this processor */
-	print_cpu_info();
 	/* cpu_idle will call schedule for us */
 	cpu_idle();
 	return 0;
 }
 
-static void __init smp_create_idle(unsigned int cpu)
+struct create_idle {
+	struct work_struct work;
+	struct task_struct *idle;
+	struct completion done;
+	int cpu;
+};
+
+static void __cpuinit smp_fork_idle(struct work_struct *work)
 {
-	struct task_struct *p;
+	struct create_idle *c_idle;
 
-	/*
-	 *  don't care about the psw and regs settings since we'll never
-	 *  reschedule the forked task.
-	 */
-	p = fork_idle(cpu);
-	if (IS_ERR(p))
-		panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
-	current_set[cpu] = p;
+	c_idle = container_of(work, struct create_idle, work);
+	c_idle->idle = fork_idle(c_idle->cpu);
+	complete(&c_idle->done);
 }
 
 static int __cpuinit smp_alloc_lowcore(int cpu)
@@ -551,6 +553,7 @@ static void smp_free_lowcore(int cpu)
 int __cpuinit __cpu_up(unsigned int cpu)
 {
 	struct _lowcore *cpu_lowcore;
+	struct create_idle c_idle;
 	struct task_struct *idle;
 	struct stack_frame *sf;
 	u32 lowcore;
@@ -558,6 +561,19 @@ int __cpuinit __cpu_up(unsigned int cpu)
 
 	if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED)
 		return -EIO;
+	idle = current_set[cpu];
+	if (!idle) {
+		c_idle.done = COMPLETION_INITIALIZER_ONSTACK(c_idle.done);
+		INIT_WORK_ONSTACK(&c_idle.work, smp_fork_idle);
+		c_idle.cpu = cpu;
+		schedule_work(&c_idle.work);
+		wait_for_completion(&c_idle.done);
+		if (IS_ERR(c_idle.idle))
+			return PTR_ERR(c_idle.idle);
+		idle = c_idle.idle;
+		current_set[cpu] = c_idle.idle;
+	}
+	init_idle(idle, cpu);
 	if (smp_alloc_lowcore(cpu))
 		return -ENOMEM;
 	do {
@@ -572,7 +588,6 @@ int __cpuinit __cpu_up(unsigned int cpu)
 	while (sigp_p(lowcore, cpu, sigp_set_prefix) == sigp_busy)
 		udelay(10);
 
-	idle = current_set[cpu];
 	cpu_lowcore = lowcore_ptr[cpu];
 	cpu_lowcore->kernel_stack = (unsigned long)
 		task_stack_page(idle) + THREAD_SIZE;
@@ -664,7 +679,6 @@ void __cpu_die(unsigned int cpu)
 		udelay(10);
 	smp_free_lowcore(cpu);
 	atomic_dec(&init_mm.context.attach_count);
-	pr_info("Processor %d stopped\n", cpu);
 }
 
 void cpu_die(void)
@@ -684,14 +698,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 #endif
 	unsigned long async_stack, panic_stack;
 	struct _lowcore *lowcore;
-	unsigned int cpu;
 
 	smp_detect_cpus();
 
 	/* request the 0x1201 emergency signal external interrupt */
 	if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0)
 		panic("Couldn't request external interrupt 0x1201");
-	print_cpu_info();
 
 	/* Reallocate current lowcore, but keep its contents. */
 	lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER);
@@ -719,9 +731,6 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore))
 		BUG();
 #endif
-	for_each_possible_cpu(cpu)
-		if (cpu != smp_processor_id())
-			smp_create_idle(cpu);
 }
 
 void __init smp_prepare_boot_cpu(void)

arch/s390/kernel/time.c

@@ -15,6 +15,7 @@
 #define KMSG_COMPONENT "time"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/errno.h>
 #include <linux/module.h>
 #include <linux/sched.h>
@@ -37,6 +38,7 @@
 #include <linux/clocksource.h>
 #include <linux/clockchips.h>
 #include <linux/gfp.h>
+#include <linux/kprobes.h>
 #include <asm/uaccess.h>
 #include <asm/delay.h>
 #include <asm/s390_ext.h>
@@ -60,7 +62,7 @@ static DEFINE_PER_CPU(struct clock_event_device, comparators);
 /*
  * Scheduler clock - returns current time in nanosec units.
  */
-unsigned long long notrace sched_clock(void)
+unsigned long long notrace __kprobes sched_clock(void)
 {
 	return (get_clock_monotonic() * 125) >> 9;
 }
@@ -159,6 +161,7 @@ static void clock_comparator_interrupt(unsigned int ext_int_code,
 				       unsigned int param32,
 				       unsigned long param64)
 {
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_CLK]++;
 	if (S390_lowcore.clock_comparator == -1ULL)
 		set_clock_comparator(S390_lowcore.clock_comparator);
 }
@@ -169,6 +172,7 @@ static void stp_timing_alert(struct stp_irq_parm *);
 static void timing_alert_interrupt(unsigned int ext_int_code,
 				   unsigned int param32, unsigned long param64)
 {
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_TLA]++;
 	if (param32 & 0x00c40000)
 		etr_timing_alert((struct etr_irq_parm *) &param32);
 	if (param32 & 0x00038000)

arch/s390/kernel/traps.c

@@ -365,12 +365,10 @@ static inline void __user *get_psw_address(struct pt_regs *regs,
 		((regs->psw.addr - (pgm_int_code >> 16)) & PSW_ADDR_INSN);
 }
 
-void __kprobes do_single_step(struct pt_regs *regs)
+void __kprobes do_per_trap(struct pt_regs *regs)
 {
-	if (notify_die(DIE_SSTEP, "sstep", regs, 0, 0,
-					SIGTRAP) == NOTIFY_STOP){
+	if (notify_die(DIE_SSTEP, "sstep", regs, 0, 0, SIGTRAP) == NOTIFY_STOP)
 		return;
-	}
 	if (tracehook_consider_fatal_signal(current, SIGTRAP))
 		force_sig(SIGTRAP, current);
 }
@@ -451,8 +449,8 @@ static inline void do_fp_trap(struct pt_regs *regs, void __user *location,
 		"floating point exception", regs, &si);
 }
 
-static void illegal_op(struct pt_regs *regs, long pgm_int_code,
-		       unsigned long trans_exc_code)
+static void __kprobes illegal_op(struct pt_regs *regs, long pgm_int_code,
+				 unsigned long trans_exc_code)
 {
 	siginfo_t info;
         __u8 opcode[6];
@@ -688,7 +686,7 @@ static void space_switch_exception(struct pt_regs *regs, long pgm_int_code,
 	do_trap(pgm_int_code, SIGILL, "space switch event", regs, &info);
 }
 
-asmlinkage void kernel_stack_overflow(struct pt_regs * regs)
+asmlinkage void __kprobes kernel_stack_overflow(struct pt_regs * regs)
 {
 	bust_spinlocks(1);
 	printk("Kernel stack overflow.\n");
@@ -733,5 +731,6 @@ void __init trap_init(void)
         pgm_check_table[0x15] = &operand_exception;
         pgm_check_table[0x1C] = &space_switch_exception;
         pgm_check_table[0x1D] = &hfp_sqrt_exception;
-	pfault_irq_init();
+	/* Enable machine checks early. */
+	local_mcck_enable();
 }

arch/s390/kernel/vtime.c

@@ -20,6 +20,7 @@
 #include <linux/rcupdate.h>
 #include <linux/posix-timers.h>
 #include <linux/cpu.h>
+#include <linux/kprobes.h>
 
 #include <asm/s390_ext.h>
 #include <asm/timer.h>
@@ -122,7 +123,7 @@ void account_system_vtime(struct task_struct *tsk)
 }
 EXPORT_SYMBOL_GPL(account_system_vtime);
 
-void vtime_start_cpu(__u64 int_clock, __u64 enter_timer)
+void __kprobes vtime_start_cpu(__u64 int_clock, __u64 enter_timer)
 {
 	struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
 	struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
@@ -162,7 +163,7 @@ void vtime_start_cpu(__u64 int_clock, __u64 enter_timer)
 	idle->sequence++;
 }
 
-void vtime_stop_cpu(void)
+void __kprobes vtime_stop_cpu(void)
 {
 	struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
 	struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
@@ -323,6 +324,7 @@ static void do_cpu_timer_interrupt(unsigned int ext_int_code,
 	struct list_head cb_list;	/* the callback queue */
 	__u64 elapsed, next;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_TMR]++;
 	INIT_LIST_HEAD(&cb_list);
 	vq = &__get_cpu_var(virt_cpu_timer);
 

arch/s390/kvm/Kconfig

@@ -4,8 +4,8 @@
 source "virt/kvm/Kconfig"
 
 menuconfig VIRTUALIZATION
-	bool "Virtualization"
-	default y
+	def_bool y
+	prompt "Virtualization"
 	---help---
 	  Say Y here to get to see options for using your Linux host to run other
 	  operating systems inside virtual machines (guests).
@@ -16,7 +16,8 @@ menuconfig VIRTUALIZATION
 if VIRTUALIZATION
 
 config KVM
-	tristate "Kernel-based Virtual Machine (KVM) support"
+	def_tristate y
+	prompt "Kernel-based Virtual Machine (KVM) support"
 	depends on HAVE_KVM && EXPERIMENTAL
 	select PREEMPT_NOTIFIERS
 	select ANON_INODES

arch/s390/lib/delay.c

@@ -47,7 +47,6 @@ static void __udelay_disabled(unsigned long long usecs)
 	lockdep_on();
 	__ctl_load(cr0_saved, 0, 0);
 	local_tick_enable(clock_saved);
-	set_clock_comparator(S390_lowcore.clock_comparator);
 }
 
 static void __udelay_enabled(unsigned long long usecs)
@@ -70,7 +69,6 @@ static void __udelay_enabled(unsigned long long usecs)
 		if (clock_saved)
 			local_tick_enable(clock_saved);
 	} while (get_clock() < end);
-	set_clock_comparator(S390_lowcore.clock_comparator);
 }
 
 /*

arch/s390/mm/fault.c

@@ -10,6 +10,7 @@
  *    Copyright (C) 1995  Linus Torvalds
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/perf_event.h>
 #include <linux/signal.h>
 #include <linux/sched.h>
@@ -234,13 +235,13 @@ static noinline int signal_return(struct pt_regs *regs, long int_code,
 	rc = __get_user(instruction, (u16 __user *) regs->psw.addr);
 
 	if (!rc && instruction == 0x0a77) {
-		clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
+		clear_tsk_thread_flag(current, TIF_PER_TRAP);
 		if (is_compat_task())
 			sys32_sigreturn();
 		else
 			sys_sigreturn();
 	} else if (!rc && instruction == 0x0aad) {
-		clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
+		clear_tsk_thread_flag(current, TIF_PER_TRAP);
 		if (is_compat_task())
 			sys32_rt_sigreturn();
 		else
@@ -378,7 +379,7 @@ static inline int do_exception(struct pt_regs *regs, int access,
 	 * The instruction that caused the program check will
 	 * be repeated. Don't signal single step via SIGTRAP.
 	 */
-	clear_tsk_thread_flag(tsk, TIF_SINGLE_STEP);
+	clear_tsk_thread_flag(tsk, TIF_PER_TRAP);
 	fault = 0;
 out_up:
 	up_read(&mm->mmap_sem);
@@ -480,8 +481,7 @@ int __handle_fault(unsigned long uaddr, unsigned long pgm_int_code, int write)
 /*
  * 'pfault' pseudo page faults routines.
  */
-static ext_int_info_t ext_int_pfault;
-static int pfault_disable = 0;
+static int pfault_disable;
 
 static int __init nopfault(char *str)
 {
@@ -543,6 +543,7 @@ static void pfault_interrupt(unsigned int ext_int_code,
 	struct task_struct *tsk;
 	__u16 subcode;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_PFL]++;
 	/*
 	 * Get the external interruption subcode & pfault
 	 * initial/completion signal bit. VM stores this 
@@ -592,24 +593,28 @@ static void pfault_interrupt(unsigned int ext_int_code,
 	}
 }
 
-void __init pfault_irq_init(void)
+static int __init pfault_irq_init(void)
 {
-	if (!MACHINE_IS_VM)
-		return;
+	int rc;
 
+	if (!MACHINE_IS_VM)
+		return 0;
 	/*
 	 * Try to get pfault pseudo page faults going.
 	 */
-	if (register_early_external_interrupt(0x2603, pfault_interrupt,
-					      &ext_int_pfault) != 0)
-		panic("Couldn't request external interrupt 0x2603");
-
+	rc = register_external_interrupt(0x2603, pfault_interrupt);
+	if (rc) {
+		pfault_disable = 1;
+		return rc;
+	}
 	if (pfault_init() == 0)
-		return;
+		return 0;
 
 	/* Tough luck, no pfault. */
 	pfault_disable = 1;
-	unregister_early_external_interrupt(0x2603, pfault_interrupt,
-					    &ext_int_pfault);
+	unregister_external_interrupt(0x2603, pfault_interrupt);
+	return 0;
 }
+early_initcall(pfault_irq_init);
+
 #endif

drivers/s390/block/Kconfig

@@ -2,7 +2,8 @@ comment "S/390 block device drivers"
 	depends on S390 && BLOCK
 
 config BLK_DEV_XPRAM
-	tristate "XPRAM disk support"
+	def_tristate m
+	prompt "XPRAM disk support"
 	depends on S390 && BLOCK
 	help
 	  Select this option if you want to use your expanded storage on S/390
@@ -12,13 +13,15 @@ config BLK_DEV_XPRAM
 	  xpram.  If unsure, say "N".
 
 config DCSSBLK
-	tristate "DCSSBLK support"
+	def_tristate m
+	prompt "DCSSBLK support"
 	depends on S390 && BLOCK
 	help
 	  Support for dcss block device
 
 config DASD
-	tristate "Support for DASD devices"
+	def_tristate y
+	prompt "Support for DASD devices"
 	depends on CCW && BLOCK
 	select IOSCHED_DEADLINE
 	help
@@ -27,28 +30,32 @@ config DASD
 	  natively on a single image or an LPAR.
 
 config DASD_PROFILE
-	bool "Profiling support for dasd devices"
+	def_bool y
+	prompt "Profiling support for dasd devices"
 	depends on DASD
 	help
 	  Enable this option if you want to see profiling information
           in /proc/dasd/statistics.
 
 config DASD_ECKD
-	tristate "Support for ECKD Disks"
+	def_tristate y
+	prompt "Support for ECKD Disks"
 	depends on DASD
 	help
 	  ECKD devices are the most commonly used devices. You should enable
 	  this option unless you are very sure to have no ECKD device.
 
 config DASD_FBA
-	tristate "Support for FBA  Disks"
+	def_tristate y
+	prompt "Support for FBA  Disks"
 	depends on DASD
 	help
 	  Select this option to be able to access FBA devices. It is safe to
 	  say "Y".
 
 config DASD_DIAG
-	tristate "Support for DIAG access to Disks"
+	def_tristate y
+	prompt "Support for DIAG access to Disks"
 	depends on DASD
 	help
 	  Select this option if you want to use Diagnose250 command to access
@@ -56,7 +63,8 @@ config DASD_DIAG
 	  say "N".
 
 config DASD_EER
-	bool "Extended error reporting (EER)"
+	def_bool y
+	prompt "Extended error reporting (EER)"
 	depends on DASD
 	help
 	  This driver provides a character device interface to the

drivers/s390/block/dasd.c

@@ -11,6 +11,7 @@
 #define KMSG_COMPONENT "dasd"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/kmod.h>
 #include <linux/init.h>
 #include <linux/interrupt.h>
@@ -368,6 +369,11 @@ dasd_state_ready_to_online(struct dasd_device * device)
 	device->state = DASD_STATE_ONLINE;
 	if (device->block) {
 		dasd_schedule_block_bh(device->block);
+		if ((device->features & DASD_FEATURE_USERAW)) {
+			disk = device->block->gdp;
+			kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
+			return 0;
+		}
 		disk = device->block->bdev->bd_disk;
 		disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0);
 		while ((part = disk_part_iter_next(&piter)))
@@ -393,7 +399,7 @@ static int dasd_state_online_to_ready(struct dasd_device *device)
 			return rc;
 	}
 	device->state = DASD_STATE_READY;
-	if (device->block) {
+	if (device->block && !(device->features & DASD_FEATURE_USERAW)) {
 		disk = device->block->bdev->bd_disk;
 		disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0);
 		while ((part = disk_part_iter_next(&piter)))
@@ -744,10 +750,6 @@ struct dasd_ccw_req *dasd_smalloc_request(int magic, int cplength,
 	char *data;
 	int size;
 
-	/* Sanity checks */
-	BUG_ON(datasize > PAGE_SIZE ||
-	     (cplength*sizeof(struct ccw1)) > PAGE_SIZE);
-
 	size = (sizeof(struct dasd_ccw_req) + 7L) & -8L;
 	if (cplength > 0)
 		size += cplength * sizeof(struct ccw1);
@@ -853,7 +855,6 @@ int dasd_term_IO(struct dasd_ccw_req *cqr)
 		rc = ccw_device_clear(device->cdev, (long) cqr);
 		switch (rc) {
 		case 0:	/* termination successful */
-			cqr->retries--;
 			cqr->status = DASD_CQR_CLEAR_PENDING;
 			cqr->stopclk = get_clock();
 			cqr->starttime = 0;
@@ -905,6 +906,16 @@ int dasd_start_IO(struct dasd_ccw_req *cqr)
 		return rc;
 	}
 	device = (struct dasd_device *) cqr->startdev;
+	if (((cqr->block &&
+	      test_bit(DASD_FLAG_LOCK_STOLEN, &cqr->block->base->flags)) ||
+	     test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags)) &&
+	    !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
+		DBF_DEV_EVENT(DBF_DEBUG, device, "start_IO: return request %p "
+			      "because of stolen lock", cqr);
+		cqr->status = DASD_CQR_ERROR;
+		cqr->intrc = -EPERM;
+		return -EPERM;
+	}
 	if (cqr->retries < 0) {
 		/* internal error 14 - start_IO run out of retries */
 		sprintf(errorstring, "14 %p", cqr);
@@ -916,6 +927,11 @@ int dasd_start_IO(struct dasd_ccw_req *cqr)
 	cqr->startclk = get_clock();
 	cqr->starttime = jiffies;
 	cqr->retries--;
+	if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) {
+		cqr->lpm &= device->path_data.opm;
+		if (!cqr->lpm)
+			cqr->lpm = device->path_data.opm;
+	}
 	if (cqr->cpmode == 1) {
 		rc = ccw_device_tm_start(device->cdev, cqr->cpaddr,
 					 (long) cqr, cqr->lpm);
@@ -928,35 +944,53 @@ int dasd_start_IO(struct dasd_ccw_req *cqr)
 		cqr->status = DASD_CQR_IN_IO;
 		break;
 	case -EBUSY:
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
 			      "start_IO: device busy, retry later");
 		break;
 	case -ETIMEDOUT:
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
 			      "start_IO: request timeout, retry later");
 		break;
 	case -EACCES:
-		/* -EACCES indicates that the request used only a
-		 * subset of the available pathes and all these
-		 * pathes are gone.
-		 * Do a retry with all available pathes.
+		/* -EACCES indicates that the request used only a subset of the
+		 * available paths and all these paths are gone. If the lpm of
+		 * this request was only a subset of the opm (e.g. the ppm) then
+		 * we just do a retry with all available paths.
+		 * If we already use the full opm, something is amiss, and we
+		 * need a full path verification.
 		 */
-		cqr->lpm = LPM_ANYPATH;
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
-			      "start_IO: selected pathes gone,"
-			      " retry on all pathes");
+		if (test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) {
+			DBF_DEV_EVENT(DBF_WARNING, device,
+				      "start_IO: selected paths gone (%x)",
+				      cqr->lpm);
+		} else if (cqr->lpm != device->path_data.opm) {
+			cqr->lpm = device->path_data.opm;
+			DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+				      "start_IO: selected paths gone,"
+				      " retry on all paths");
+		} else {
+			DBF_DEV_EVENT(DBF_WARNING, device, "%s",
+				      "start_IO: all paths in opm gone,"
+				      " do path verification");
+			dasd_generic_last_path_gone(device);
+			device->path_data.opm = 0;
+			device->path_data.ppm = 0;
+			device->path_data.npm = 0;
+			device->path_data.tbvpm =
+				ccw_device_get_path_mask(device->cdev);
+		}
 		break;
 	case -ENODEV:
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
 			      "start_IO: -ENODEV device gone, retry");
 		break;
 	case -EIO:
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
 			      "start_IO: -EIO device gone, retry");
 		break;
 	case -EINVAL:
 		/* most likely caused in power management context */
-		DBF_DEV_EVENT(DBF_DEBUG, device, "%s",
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
 			      "start_IO: -EINVAL device currently "
 			      "not accessible");
 		break;
@@ -1076,6 +1110,7 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 	unsigned long long now;
 	int expires;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_DAS]++;
 	if (IS_ERR(irb)) {
 		switch (PTR_ERR(irb)) {
 		case -EIO:
@@ -1094,16 +1129,11 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 	}
 
 	now = get_clock();
-
-	/* check for unsolicited interrupts */
 	cqr = (struct dasd_ccw_req *) intparm;
-	if (!cqr || ((scsw_cc(&irb->scsw) == 1) &&
-		     (scsw_fctl(&irb->scsw) & SCSW_FCTL_START_FUNC) &&
-		     ((scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND) ||
-		      (scsw_stctl(&irb->scsw) == (SCSW_STCTL_STATUS_PEND |
-						  SCSW_STCTL_ALERT_STATUS))))) {
-		if (cqr && cqr->status == DASD_CQR_IN_IO)
-			cqr->status = DASD_CQR_QUEUED;
+	/* check for conditions that should be handled immediately */
+	if (!cqr ||
+	    !(scsw_dstat(&irb->scsw) == (DEV_STAT_CHN_END | DEV_STAT_DEV_END) &&
+	      scsw_cstat(&irb->scsw) == 0)) {
 		if (cqr)
 			memcpy(&cqr->irb, irb, sizeof(*irb));
 		device = dasd_device_from_cdev_locked(cdev);
@@ -1114,17 +1144,14 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 			dasd_put_device(device);
 			return;
 		}
-		device->discipline->dump_sense_dbf(device, irb,
-						   "unsolicited");
-		if ((device->features & DASD_FEATURE_ERPLOG))
-			device->discipline->dump_sense(device, cqr,
-						       irb);
-		dasd_device_clear_timer(device);
-		device->discipline->handle_unsolicited_interrupt(device,
-								 irb);
+		device->discipline->dump_sense_dbf(device, irb, "int");
+		if (device->features & DASD_FEATURE_ERPLOG)
+			device->discipline->dump_sense(device, cqr, irb);
+		device->discipline->check_for_device_change(device, cqr, irb);
 		dasd_put_device(device);
-		return;
 	}
+	if (!cqr)
+		return;
 
 	device = (struct dasd_device *) cqr->startdev;
 	if (!device ||
@@ -1164,25 +1191,19 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 					  struct dasd_ccw_req, devlist);
 		}
 	} else {  /* error */
-		memcpy(&cqr->irb, irb, sizeof(struct irb));
-		/* log sense for every failed I/O to s390 debugfeature */
-		dasd_log_sense_dbf(cqr, irb);
-		if (device->features & DASD_FEATURE_ERPLOG) {
-			dasd_log_sense(cqr, irb);
-		}
-
 		/*
 		 * If we don't want complex ERP for this request, then just
 		 * reset this and retry it in the fastpath
 		 */
 		if (!test_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags) &&
 		    cqr->retries > 0) {
-			if (cqr->lpm == LPM_ANYPATH)
+			if (cqr->lpm == device->path_data.opm)
 				DBF_DEV_EVENT(DBF_DEBUG, device,
 					      "default ERP in fastpath "
 					      "(%i retries left)",
 					      cqr->retries);
-			cqr->lpm    = LPM_ANYPATH;
+			if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags))
+				cqr->lpm = device->path_data.opm;
 			cqr->status = DASD_CQR_QUEUED;
 			next = cqr;
 		} else
@@ -1210,13 +1231,13 @@ enum uc_todo dasd_generic_uc_handler(struct ccw_device *cdev, struct irb *irb)
 		goto out;
 	if (test_bit(DASD_FLAG_OFFLINE, &device->flags) ||
 	   device->state != device->target ||
-	   !device->discipline->handle_unsolicited_interrupt){
+	   !device->discipline->check_for_device_change){
 		dasd_put_device(device);
 		goto out;
 	}
-
-	dasd_device_clear_timer(device);
-	device->discipline->handle_unsolicited_interrupt(device, irb);
+	if (device->discipline->dump_sense_dbf)
+		device->discipline->dump_sense_dbf(device, irb, "uc");
+	device->discipline->check_for_device_change(device, NULL, irb);
 	dasd_put_device(device);
 out:
 	return UC_TODO_RETRY;
@@ -1366,8 +1387,14 @@ static void __dasd_device_start_head(struct dasd_device *device)
 	cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist);
 	if (cqr->status != DASD_CQR_QUEUED)
 		return;
-	/* when device is stopped, return request to previous layer */
-	if (device->stopped) {
+	/* when device is stopped, return request to previous layer
+	 * exception: only the disconnect or unresumed bits are set and the
+	 * cqr is a path verification request
+	 */
+	if (device->stopped &&
+	    !(!(device->stopped & ~(DASD_STOPPED_DC_WAIT | DASD_UNRESUMED_PM))
+	      && test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags))) {
+		cqr->intrc = -EAGAIN;
 		cqr->status = DASD_CQR_CLEARED;
 		dasd_schedule_device_bh(device);
 		return;
@@ -1383,6 +1410,23 @@ static void __dasd_device_start_head(struct dasd_device *device)
 		dasd_device_set_timer(device, 50);
 }
 
+static void __dasd_device_check_path_events(struct dasd_device *device)
+{
+	int rc;
+
+	if (device->path_data.tbvpm) {
+		if (device->stopped & ~(DASD_STOPPED_DC_WAIT |
+					DASD_UNRESUMED_PM))
+			return;
+		rc = device->discipline->verify_path(
+			device, device->path_data.tbvpm);
+		if (rc)
+			dasd_device_set_timer(device, 50);
+		else
+			device->path_data.tbvpm = 0;
+	}
+};
+
 /*
  * Go through all request on the dasd_device request queue,
  * terminate them on the cdev if necessary, and return them to the
@@ -1457,6 +1501,7 @@ static void dasd_device_tasklet(struct dasd_device *device)
 	__dasd_device_check_expire(device);
 	/* find final requests on ccw queue */
 	__dasd_device_process_ccw_queue(device, &final_queue);
+	__dasd_device_check_path_events(device);
 	spin_unlock_irq(get_ccwdev_lock(device->cdev));
 	/* Now call the callback function of requests with final status */
 	__dasd_device_process_final_queue(device, &final_queue);
@@ -1613,7 +1658,12 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible)
 			continue;
 		if (cqr->status != DASD_CQR_FILLED) /* could be failed */
 			continue;
-
+		if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) &&
+		    !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
+			cqr->status = DASD_CQR_FAILED;
+			cqr->intrc = -EPERM;
+			continue;
+		}
 		/* Non-temporary stop condition will trigger fail fast */
 		if (device->stopped & ~DASD_STOPPED_PENDING &&
 		    test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&
@@ -1621,7 +1671,6 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible)
 			cqr->status = DASD_CQR_FAILED;
 			continue;
 		}
-
 		/* Don't try to start requests if device is stopped */
 		if (interruptible) {
 			rc = wait_event_interruptible(
@@ -1706,13 +1755,18 @@ int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr)
 	int rc;
 
 	device = cqr->startdev;
+	if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) &&
+	    !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
+		cqr->status = DASD_CQR_FAILED;
+		cqr->intrc = -EPERM;
+		return -EIO;
+	}
 	spin_lock_irq(get_ccwdev_lock(device->cdev));
 	rc = _dasd_term_running_cqr(device);
 	if (rc) {
 		spin_unlock_irq(get_ccwdev_lock(device->cdev));
 		return rc;
 	}
-
 	cqr->callback = dasd_wakeup_cb;
 	cqr->callback_data = DASD_SLEEPON_START_TAG;
 	cqr->status = DASD_CQR_QUEUED;
@@ -2016,6 +2070,13 @@ static void __dasd_block_start_head(struct dasd_block *block)
 	list_for_each_entry(cqr, &block->ccw_queue, blocklist) {
 		if (cqr->status != DASD_CQR_FILLED)
 			continue;
+		if (test_bit(DASD_FLAG_LOCK_STOLEN, &block->base->flags) &&
+		    !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
+			cqr->status = DASD_CQR_FAILED;
+			cqr->intrc = -EPERM;
+			dasd_schedule_block_bh(block);
+			continue;
+		}
 		/* Non-temporary stop condition will trigger fail fast */
 		if (block->base->stopped & ~DASD_STOPPED_PENDING &&
 		    test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&
@@ -2201,8 +2262,20 @@ static void dasd_setup_queue(struct dasd_block *block)
 {
 	int max;
 
-	blk_queue_logical_block_size(block->request_queue, block->bp_block);
-	max = block->base->discipline->max_blocks << block->s2b_shift;
+	if (block->base->features & DASD_FEATURE_USERAW) {
+		/*
+		 * the max_blocks value for raw_track access is 256
+		 * it is higher than the native ECKD value because we
+		 * only need one ccw per track
+		 * so the max_hw_sectors are
+		 * 2048 x 512B = 1024kB = 16 tracks
+		 */
+		max = 2048;
+	} else {
+		max = block->base->discipline->max_blocks << block->s2b_shift;
+	}
+	blk_queue_logical_block_size(block->request_queue,
+				     block->bp_block);
 	blk_queue_max_hw_sectors(block->request_queue, max);
 	blk_queue_max_segments(block->request_queue, -1L);
 	/* with page sized segments we can translate each segement into
@@ -2588,10 +2661,53 @@ int dasd_generic_set_offline(struct ccw_device *cdev)
 	return 0;
 }
 
+int dasd_generic_last_path_gone(struct dasd_device *device)
+{
+	struct dasd_ccw_req *cqr;
+
+	dev_warn(&device->cdev->dev, "No operational channel path is left "
+		 "for the device\n");
+	DBF_DEV_EVENT(DBF_WARNING, device, "%s", "last path gone");
+	/* First of all call extended error reporting. */
+	dasd_eer_write(device, NULL, DASD_EER_NOPATH);
+
+	if (device->state < DASD_STATE_BASIC)
+		return 0;
+	/* Device is active. We want to keep it. */
+	list_for_each_entry(cqr, &device->ccw_queue, devlist)
+		if ((cqr->status == DASD_CQR_IN_IO) ||
+		    (cqr->status == DASD_CQR_CLEAR_PENDING)) {
+			cqr->status = DASD_CQR_QUEUED;
+			cqr->retries++;
+		}
+	dasd_device_set_stop_bits(device, DASD_STOPPED_DC_WAIT);
+	dasd_device_clear_timer(device);
+	dasd_schedule_device_bh(device);
+	return 1;
+}
+EXPORT_SYMBOL_GPL(dasd_generic_last_path_gone);
+
+int dasd_generic_path_operational(struct dasd_device *device)
+{
+	dev_info(&device->cdev->dev, "A channel path to the device has become "
+		 "operational\n");
+	DBF_DEV_EVENT(DBF_WARNING, device, "%s", "path operational");
+	dasd_device_remove_stop_bits(device, DASD_STOPPED_DC_WAIT);
+	if (device->stopped & DASD_UNRESUMED_PM) {
+		dasd_device_remove_stop_bits(device, DASD_UNRESUMED_PM);
+		dasd_restore_device(device);
+		return 1;
+	}
+	dasd_schedule_device_bh(device);
+	if (device->block)
+		dasd_schedule_block_bh(device->block);
+	return 1;
+}
+EXPORT_SYMBOL_GPL(dasd_generic_path_operational);
+
 int dasd_generic_notify(struct ccw_device *cdev, int event)
 {
 	struct dasd_device *device;
-	struct dasd_ccw_req *cqr;
 	int ret;
 
 	device = dasd_device_from_cdev_locked(cdev);
@@ -2602,41 +2718,64 @@ int dasd_generic_notify(struct ccw_device *cdev, int event)
 	case CIO_GONE:
 	case CIO_BOXED:
 	case CIO_NO_PATH:
-		/* First of all call extended error reporting. */
-		dasd_eer_write(device, NULL, DASD_EER_NOPATH);
-
-		if (device->state < DASD_STATE_BASIC)
-			break;
-		/* Device is active. We want to keep it. */
-		list_for_each_entry(cqr, &device->ccw_queue, devlist)
-			if (cqr->status == DASD_CQR_IN_IO) {
-				cqr->status = DASD_CQR_QUEUED;
-				cqr->retries++;
-			}
-		dasd_device_set_stop_bits(device, DASD_STOPPED_DC_WAIT);
-		dasd_device_clear_timer(device);
-		dasd_schedule_device_bh(device);
-		ret = 1;
+		device->path_data.opm = 0;
+		device->path_data.ppm = 0;
+		device->path_data.npm = 0;
+		ret = dasd_generic_last_path_gone(device);
 		break;
 	case CIO_OPER:
-		/* FIXME: add a sanity check. */
-		dasd_device_remove_stop_bits(device, DASD_STOPPED_DC_WAIT);
-		if (device->stopped & DASD_UNRESUMED_PM) {
-			dasd_device_remove_stop_bits(device, DASD_UNRESUMED_PM);
-			dasd_restore_device(device);
-			ret = 1;
-			break;
-		}
-		dasd_schedule_device_bh(device);
-		if (device->block)
-			dasd_schedule_block_bh(device->block);
 		ret = 1;
+		if (device->path_data.opm)
+			ret = dasd_generic_path_operational(device);
 		break;
 	}
 	dasd_put_device(device);
 	return ret;
 }
 
+void dasd_generic_path_event(struct ccw_device *cdev, int *path_event)
+{
+	int chp;
+	__u8 oldopm, eventlpm;
+	struct dasd_device *device;
+
+	device = dasd_device_from_cdev_locked(cdev);
+	if (IS_ERR(device))
+		return;
+	for (chp = 0; chp < 8; chp++) {
+		eventlpm = 0x80 >> chp;
+		if (path_event[chp] & PE_PATH_GONE) {
+			oldopm = device->path_data.opm;
+			device->path_data.opm &= ~eventlpm;
+			device->path_data.ppm &= ~eventlpm;
+			device->path_data.npm &= ~eventlpm;
+			if (oldopm && !device->path_data.opm)
+				dasd_generic_last_path_gone(device);
+		}
+		if (path_event[chp] & PE_PATH_AVAILABLE) {
+			device->path_data.opm &= ~eventlpm;
+			device->path_data.ppm &= ~eventlpm;
+			device->path_data.npm &= ~eventlpm;
+			device->path_data.tbvpm |= eventlpm;
+			dasd_schedule_device_bh(device);
+		}
+	}
+	dasd_put_device(device);
+}
+EXPORT_SYMBOL_GPL(dasd_generic_path_event);
+
+int dasd_generic_verify_path(struct dasd_device *device, __u8 lpm)
+{
+	if (!device->path_data.opm && lpm) {
+		device->path_data.opm = lpm;
+		dasd_generic_path_operational(device);
+	} else
+		device->path_data.opm |= lpm;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(dasd_generic_verify_path);
+
+
 int dasd_generic_pm_freeze(struct ccw_device *cdev)
 {
 	struct dasd_ccw_req *cqr, *n;
@@ -2646,6 +2785,10 @@ int dasd_generic_pm_freeze(struct ccw_device *cdev)
 
 	if (IS_ERR(device))
 		return PTR_ERR(device);
+
+	if (device->discipline->freeze)
+		rc = device->discipline->freeze(device);
+
 	/* disallow new I/O  */
 	dasd_device_set_stop_bits(device, DASD_STOPPED_PM);
 	/* clear active requests */
@@ -2682,9 +2825,6 @@ int dasd_generic_pm_freeze(struct ccw_device *cdev)
 	list_splice_tail(&freeze_queue, &device->ccw_queue);
 	spin_unlock_irq(get_ccwdev_lock(cdev));
 
-	if (device->discipline->freeze)
-		rc = device->discipline->freeze(device);
-
 	dasd_put_device(device);
 	return rc;
 }

drivers/s390/block/dasd_3990_erp.c

@@ -152,9 +152,9 @@ dasd_3990_erp_alternate_path(struct dasd_ccw_req * erp)
 	spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
 	opm = ccw_device_get_path_mask(device->cdev);
 	spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
-	//FIXME: start with get_opm ?
 	if (erp->lpm == 0)
-		erp->lpm = LPM_ANYPATH & ~(erp->irb.esw.esw0.sublog.lpum);
+		erp->lpm = device->path_data.opm &
+			~(erp->irb.esw.esw0.sublog.lpum);
 	else
 		erp->lpm &= ~(erp->irb.esw.esw0.sublog.lpum);
 
@@ -270,10 +270,11 @@ static struct dasd_ccw_req *dasd_3990_erp_action_1(struct dasd_ccw_req *erp)
 {
 	erp->function = dasd_3990_erp_action_1;
 	dasd_3990_erp_alternate_path(erp);
-	if (erp->status == DASD_CQR_FAILED) {
+	if (erp->status == DASD_CQR_FAILED &&
+	    !test_bit(DASD_CQR_VERIFY_PATH, &erp->flags)) {
 		erp->status = DASD_CQR_FILLED;
 		erp->retries = 10;
-		erp->lpm = LPM_ANYPATH;
+		erp->lpm = erp->startdev->path_data.opm;
 		erp->function = dasd_3990_erp_action_1_sec;
 	}
 	return erp;
@@ -1907,15 +1908,14 @@ dasd_3990_erp_compound_retry(struct dasd_ccw_req * erp, char *sense)
 static void
 dasd_3990_erp_compound_path(struct dasd_ccw_req * erp, char *sense)
 {
-
 	if (sense[25] & DASD_SENSE_BIT_3) {
 		dasd_3990_erp_alternate_path(erp);
 
-		if (erp->status == DASD_CQR_FAILED) {
+		if (erp->status == DASD_CQR_FAILED &&
+		    !test_bit(DASD_CQR_VERIFY_PATH, &erp->flags)) {
 			/* reset the lpm and the status to be able to
 			 * try further actions. */
-
-			erp->lpm = 0;
+			erp->lpm = erp->startdev->path_data.opm;
 			erp->status = DASD_CQR_NEED_ERP;
 		}
 	}

drivers/s390/block/dasd_devmap.c

@@ -208,6 +208,8 @@ dasd_feature_list(char *str, char **endp)
 			features |= DASD_FEATURE_READONLY;
 		else if (len == 4 && !strncmp(str, "diag", 4))
 			features |= DASD_FEATURE_USEDIAG;
+		else if (len == 3 && !strncmp(str, "raw", 3))
+			features |= DASD_FEATURE_USERAW;
 		else if (len == 6 && !strncmp(str, "erplog", 6))
 			features |= DASD_FEATURE_ERPLOG;
 		else if (len == 8 && !strncmp(str, "failfast", 8))
@@ -639,6 +641,7 @@ dasd_put_device_wake(struct dasd_device *device)
 {
 	wake_up(&dasd_delete_wq);
 }
+EXPORT_SYMBOL_GPL(dasd_put_device_wake);
 
 /*
  * Return dasd_device structure associated with cdev.
@@ -856,7 +859,7 @@ dasd_use_diag_store(struct device *dev, struct device_attribute *attr,
 	spin_lock(&dasd_devmap_lock);
 	/* Changing diag discipline flag is only allowed in offline state. */
 	rc = count;
-	if (!devmap->device) {
+	if (!devmap->device && !(devmap->features & DASD_FEATURE_USERAW)) {
 		if (val)
 			devmap->features |= DASD_FEATURE_USEDIAG;
 		else
@@ -869,6 +872,56 @@ dasd_use_diag_store(struct device *dev, struct device_attribute *attr,
 
 static DEVICE_ATTR(use_diag, 0644, dasd_use_diag_show, dasd_use_diag_store);
 
+/*
+ * use_raw controls whether the driver should give access to raw eckd data or
+ * operate in standard mode
+ */
+static ssize_t
+dasd_use_raw_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct dasd_devmap *devmap;
+	int use_raw;
+
+	devmap = dasd_find_busid(dev_name(dev));
+	if (!IS_ERR(devmap))
+		use_raw = (devmap->features & DASD_FEATURE_USERAW) != 0;
+	else
+		use_raw = (DASD_FEATURE_DEFAULT & DASD_FEATURE_USERAW) != 0;
+	return sprintf(buf, use_raw ? "1\n" : "0\n");
+}
+
+static ssize_t
+dasd_use_raw_store(struct device *dev, struct device_attribute *attr,
+		    const char *buf, size_t count)
+{
+	struct dasd_devmap *devmap;
+	ssize_t rc;
+	unsigned long val;
+
+	devmap = dasd_devmap_from_cdev(to_ccwdev(dev));
+	if (IS_ERR(devmap))
+		return PTR_ERR(devmap);
+
+	if ((strict_strtoul(buf, 10, &val) != 0) || val > 1)
+		return -EINVAL;
+
+	spin_lock(&dasd_devmap_lock);
+	/* Changing diag discipline flag is only allowed in offline state. */
+	rc = count;
+	if (!devmap->device && !(devmap->features & DASD_FEATURE_USEDIAG)) {
+		if (val)
+			devmap->features |= DASD_FEATURE_USERAW;
+		else
+			devmap->features &= ~DASD_FEATURE_USERAW;
+	} else
+		rc = -EPERM;
+	spin_unlock(&dasd_devmap_lock);
+	return rc;
+}
+
+static DEVICE_ATTR(raw_track_access, 0644, dasd_use_raw_show,
+		   dasd_use_raw_store);
+
 static ssize_t
 dasd_discipline_show(struct device *dev, struct device_attribute *attr,
 		     char *buf)
@@ -1126,6 +1179,103 @@ dasd_expires_store(struct device *dev, struct device_attribute *attr,
 
 static DEVICE_ATTR(expires, 0644, dasd_expires_show, dasd_expires_store);
 
+static ssize_t dasd_reservation_policy_show(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	struct dasd_devmap *devmap;
+	int rc = 0;
+
+	devmap = dasd_find_busid(dev_name(dev));
+	if (IS_ERR(devmap)) {
+		rc = snprintf(buf, PAGE_SIZE, "ignore\n");
+	} else {
+		spin_lock(&dasd_devmap_lock);
+		if (devmap->features & DASD_FEATURE_FAILONSLCK)
+			rc = snprintf(buf, PAGE_SIZE, "fail\n");
+		else
+			rc = snprintf(buf, PAGE_SIZE, "ignore\n");
+		spin_unlock(&dasd_devmap_lock);
+	}
+	return rc;
+}
+
+static ssize_t dasd_reservation_policy_store(struct device *dev,
+					     struct device_attribute *attr,
+					     const char *buf, size_t count)
+{
+	struct dasd_devmap *devmap;
+	int rc;
+
+	devmap = dasd_devmap_from_cdev(to_ccwdev(dev));
+	if (IS_ERR(devmap))
+		return PTR_ERR(devmap);
+	rc = 0;
+	spin_lock(&dasd_devmap_lock);
+	if (sysfs_streq("ignore", buf))
+		devmap->features &= ~DASD_FEATURE_FAILONSLCK;
+	else if (sysfs_streq("fail", buf))
+		devmap->features |= DASD_FEATURE_FAILONSLCK;
+	else
+		rc = -EINVAL;
+	if (devmap->device)
+		devmap->device->features = devmap->features;
+	spin_unlock(&dasd_devmap_lock);
+	if (rc)
+		return rc;
+	else
+		return count;
+}
+
+static DEVICE_ATTR(reservation_policy, 0644,
+		   dasd_reservation_policy_show, dasd_reservation_policy_store);
+
+static ssize_t dasd_reservation_state_show(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	struct dasd_device *device;
+	int rc = 0;
+
+	device = dasd_device_from_cdev(to_ccwdev(dev));
+	if (IS_ERR(device))
+		return snprintf(buf, PAGE_SIZE, "none\n");
+
+	if (test_bit(DASD_FLAG_IS_RESERVED, &device->flags))
+		rc = snprintf(buf, PAGE_SIZE, "reserved\n");
+	else if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags))
+		rc = snprintf(buf, PAGE_SIZE, "lost\n");
+	else
+		rc = snprintf(buf, PAGE_SIZE, "none\n");
+	dasd_put_device(device);
+	return rc;
+}
+
+static ssize_t dasd_reservation_state_store(struct device *dev,
+					    struct device_attribute *attr,
+					    const char *buf, size_t count)
+{
+	struct dasd_device *device;
+	int rc = 0;
+
+	device = dasd_device_from_cdev(to_ccwdev(dev));
+	if (IS_ERR(device))
+		return -ENODEV;
+	if (sysfs_streq("reset", buf))
+		clear_bit(DASD_FLAG_LOCK_STOLEN, &device->flags);
+	else
+		rc = -EINVAL;
+	dasd_put_device(device);
+
+	if (rc)
+		return rc;
+	else
+		return count;
+}
+
+static DEVICE_ATTR(last_known_reservation_state, 0644,
+		   dasd_reservation_state_show, dasd_reservation_state_store);
+
 static struct attribute * dasd_attrs[] = {
 	&dev_attr_readonly.attr,
 	&dev_attr_discipline.attr,
@@ -1134,10 +1284,13 @@ static struct attribute * dasd_attrs[] = {
 	&dev_attr_vendor.attr,
 	&dev_attr_uid.attr,
 	&dev_attr_use_diag.attr,
+	&dev_attr_raw_track_access.attr,
 	&dev_attr_eer_enabled.attr,
 	&dev_attr_erplog.attr,
 	&dev_attr_failfast.attr,
 	&dev_attr_expires.attr,
+	&dev_attr_reservation_policy.attr,
+	&dev_attr_last_known_reservation_state.attr,
 	NULL,
 };
 

drivers/s390/block/dasd_diag.c

@@ -10,6 +10,7 @@
 
 #define KMSG_COMPONENT "dasd"
 
+#include <linux/kernel_stat.h>
 #include <linux/stddef.h>
 #include <linux/kernel.h>
 #include <linux/slab.h>
@@ -238,6 +239,7 @@ static void dasd_ext_handler(unsigned int ext_int_code,
 	addr_t ip;
 	int rc;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_DSD]++;
 	switch (ext_int_code >> 24) {
 	case DASD_DIAG_CODE_31BIT:
 		ip = (addr_t) param32;
@@ -617,6 +619,7 @@ static struct dasd_discipline dasd_diag_discipline = {
 	.ebcname = "DIAG",
 	.max_blocks = DIAG_MAX_BLOCKS,
 	.check_device = dasd_diag_check_device,
+	.verify_path = dasd_generic_verify_path,
 	.fill_geometry = dasd_diag_fill_geometry,
 	.start_IO = dasd_start_diag,
 	.term_IO = dasd_diag_term_IO,

drivers/s390/block/dasd_eckd.c

@@ -54,6 +54,15 @@
 #define ECKD_F7(i) (i->factor7)
 #define ECKD_F8(i) (i->factor8)
 
+/*
+ * raw track access always map to 64k in memory
+ * so it maps to 16 blocks of 4k per track
+ */
+#define DASD_RAW_BLOCK_PER_TRACK 16
+#define DASD_RAW_BLOCKSIZE 4096
+/* 64k are 128 x 512 byte sectors  */
+#define DASD_RAW_SECTORS_PER_TRACK 128
+
 MODULE_LICENSE("GPL");
 
 static struct dasd_discipline dasd_eckd_discipline;
@@ -90,6 +99,18 @@ static struct {
 } *dasd_reserve_req;
 static DEFINE_MUTEX(dasd_reserve_mutex);
 
+/* definitions for the path verification worker */
+struct path_verification_work_data {
+	struct work_struct worker;
+	struct dasd_device *device;
+	struct dasd_ccw_req cqr;
+	struct ccw1 ccw;
+	__u8 rcd_buffer[DASD_ECKD_RCD_DATA_SIZE];
+	int isglobal;
+	__u8 tbvpm;
+};
+static struct path_verification_work_data *path_verification_worker;
+static DEFINE_MUTEX(dasd_path_verification_mutex);
 
 /* initial attempt at a probe function. this can be simplified once
  * the other detection code is gone */
@@ -373,6 +394,23 @@ static void fill_LRE_data(struct LRE_eckd_data *data, unsigned int trk,
 		data->length = reclen;
 		data->operation.operation = 0x03;
 		break;
+	case DASD_ECKD_CCW_WRITE_FULL_TRACK:
+		data->operation.orientation = 0x0;
+		data->operation.operation = 0x3F;
+		data->extended_operation = 0x11;
+		data->length = 0;
+		data->extended_parameter_length = 0x02;
+		if (data->count > 8) {
+			data->extended_parameter[0] = 0xFF;
+			data->extended_parameter[1] = 0xFF;
+			data->extended_parameter[1] <<= (16 - count);
+		} else {
+			data->extended_parameter[0] = 0xFF;
+			data->extended_parameter[0] <<= (8 - count);
+			data->extended_parameter[1] = 0x00;
+		}
+		data->sector = 0xFF;
+		break;
 	case DASD_ECKD_CCW_WRITE_TRACK_DATA:
 		data->auxiliary.length_valid = 0x1;
 		data->length = reclen;	/* not tlf, as one might think */
@@ -396,6 +434,12 @@ static void fill_LRE_data(struct LRE_eckd_data *data, unsigned int trk,
 	case DASD_ECKD_CCW_READ_COUNT:
 		data->operation.operation = 0x06;
 		break;
+	case DASD_ECKD_CCW_READ_TRACK:
+		data->operation.orientation = 0x1;
+		data->operation.operation = 0x0C;
+		data->extended_parameter_length = 0;
+		data->sector = 0xFF;
+		break;
 	case DASD_ECKD_CCW_READ_TRACK_DATA:
 		data->auxiliary.length_valid = 0x1;
 		data->length = tlf;
@@ -439,10 +483,16 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata,
 
 	ccw->cmd_code = DASD_ECKD_CCW_PFX;
 	ccw->flags = 0;
-	ccw->count = sizeof(*pfxdata);
-	ccw->cda = (__u32) __pa(pfxdata);
+	if (cmd == DASD_ECKD_CCW_WRITE_FULL_TRACK) {
+		ccw->count = sizeof(*pfxdata) + 2;
+		ccw->cda = (__u32) __pa(pfxdata);
+		memset(pfxdata, 0, sizeof(*pfxdata) + 2);
+	} else {
+		ccw->count = sizeof(*pfxdata);
+		ccw->cda = (__u32) __pa(pfxdata);
+		memset(pfxdata, 0, sizeof(*pfxdata));
+	}
 
-	memset(pfxdata, 0, sizeof(*pfxdata));
 	/* prefix data */
 	if (format > 1) {
 		DBF_DEV_EVENT(DBF_ERR, basedev,
@@ -476,6 +526,7 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata,
 		dedata->mask.perm = 0x1;
 		dedata->attributes.operation = basepriv->attrib.operation;
 		break;
+	case DASD_ECKD_CCW_READ_TRACK:
 	case DASD_ECKD_CCW_READ_TRACK_DATA:
 		dedata->mask.perm = 0x1;
 		dedata->attributes.operation = basepriv->attrib.operation;
@@ -502,6 +553,11 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata,
 		dedata->attributes.operation = DASD_BYPASS_CACHE;
 		rc = check_XRC_on_prefix(pfxdata, basedev);
 		break;
+	case DASD_ECKD_CCW_WRITE_FULL_TRACK:
+		dedata->mask.perm = 0x03;
+		dedata->attributes.operation = basepriv->attrib.operation;
+		dedata->blk_size = 0;
+		break;
 	case DASD_ECKD_CCW_WRITE_TRACK_DATA:
 		dedata->mask.perm = 0x02;
 		dedata->attributes.operation = basepriv->attrib.operation;
@@ -755,26 +811,27 @@ static int dasd_eckd_get_uid(struct dasd_device *device, struct dasd_uid *uid)
 	return -EINVAL;
 }
 
-static struct dasd_ccw_req *dasd_eckd_build_rcd_lpm(struct dasd_device *device,
-						    void *rcd_buffer,
-						    struct ciw *ciw, __u8 lpm)
+static void dasd_eckd_fill_rcd_cqr(struct dasd_device *device,
+				   struct dasd_ccw_req *cqr,
+				   __u8 *rcd_buffer,
+				   __u8 lpm)
 {
-	struct dasd_ccw_req *cqr;
 	struct ccw1 *ccw;
-
-	cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, 1 /* RCD */, ciw->count,
-				   device);
-
-	if (IS_ERR(cqr)) {
-		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
-			      "Could not allocate RCD request");
-		return cqr;
-	}
+	/*
+	 * buffer has to start with EBCDIC "V1.0" to show
+	 * support for virtual device SNEQ
+	 */
+	rcd_buffer[0] = 0xE5;
+	rcd_buffer[1] = 0xF1;
+	rcd_buffer[2] = 0x4B;
+	rcd_buffer[3] = 0xF0;
 
 	ccw = cqr->cpaddr;
-	ccw->cmd_code = ciw->cmd;
+	ccw->cmd_code = DASD_ECKD_CCW_RCD;
+	ccw->flags = 0;
 	ccw->cda = (__u32)(addr_t)rcd_buffer;
-	ccw->count = ciw->count;
+	ccw->count = DASD_ECKD_RCD_DATA_SIZE;
+	cqr->magic = DASD_ECKD_MAGIC;
 
 	cqr->startdev = device;
 	cqr->memdev = device;
@@ -784,7 +841,30 @@ static struct dasd_ccw_req *dasd_eckd_build_rcd_lpm(struct dasd_device *device,
 	cqr->retries = 256;
 	cqr->buildclk = get_clock();
 	cqr->status = DASD_CQR_FILLED;
-	return cqr;
+	set_bit(DASD_CQR_VERIFY_PATH, &cqr->flags);
+}
+
+static int dasd_eckd_read_conf_immediately(struct dasd_device *device,
+					   struct dasd_ccw_req *cqr,
+					   __u8 *rcd_buffer,
+					   __u8 lpm)
+{
+	struct ciw *ciw;
+	int rc;
+	/*
+	 * sanity check: scan for RCD command in extended SenseID data
+	 * some devices do not support RCD
+	 */
+	ciw = ccw_device_get_ciw(device->cdev, CIW_TYPE_RCD);
+	if (!ciw || ciw->cmd != DASD_ECKD_CCW_RCD)
+		return -EOPNOTSUPP;
+
+	dasd_eckd_fill_rcd_cqr(device, cqr, rcd_buffer, lpm);
+	clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+	set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
+	cqr->retries = 5;
+	rc = dasd_sleep_on_immediatly(cqr);
+	return rc;
 }
 
 static int dasd_eckd_read_conf_lpm(struct dasd_device *device,
@@ -797,32 +877,29 @@ static int dasd_eckd_read_conf_lpm(struct dasd_device *device,
 	struct dasd_ccw_req *cqr;
 
 	/*
-	 * scan for RCD command in extended SenseID data
+	 * sanity check: scan for RCD command in extended SenseID data
+	 * some devices do not support RCD
 	 */
 	ciw = ccw_device_get_ciw(device->cdev, CIW_TYPE_RCD);
-	if (!ciw || ciw->cmd == 0) {
+	if (!ciw || ciw->cmd != DASD_ECKD_CCW_RCD) {
 		ret = -EOPNOTSUPP;
 		goto out_error;
 	}
-	rcd_buf = kzalloc(ciw->count, GFP_KERNEL | GFP_DMA);
+	rcd_buf = kzalloc(DASD_ECKD_RCD_DATA_SIZE, GFP_KERNEL | GFP_DMA);
 	if (!rcd_buf) {
 		ret = -ENOMEM;
 		goto out_error;
 	}
-
-	/*
-	 * buffer has to start with EBCDIC "V1.0" to show
-	 * support for virtual device SNEQ
-	 */
-	rcd_buf[0] = 0xE5;
-	rcd_buf[1] = 0xF1;
-	rcd_buf[2] = 0x4B;
-	rcd_buf[3] = 0xF0;
-	cqr = dasd_eckd_build_rcd_lpm(device, rcd_buf, ciw, lpm);
+	cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, 1 /* RCD */,
+				   0, /* use rcd_buf as data ara */
+				   device);
 	if (IS_ERR(cqr)) {
-		ret =  PTR_ERR(cqr);
+		DBF_DEV_EVENT(DBF_WARNING, device, "%s",
+			      "Could not allocate RCD request");
+		ret = -ENOMEM;
 		goto out_error;
 	}
+	dasd_eckd_fill_rcd_cqr(device, cqr, rcd_buf, lpm);
 	ret = dasd_sleep_on(cqr);
 	/*
 	 * on success we update the user input parms
@@ -831,7 +908,7 @@ static int dasd_eckd_read_conf_lpm(struct dasd_device *device,
 	if (ret)
 		goto out_error;
 
-	*rcd_buffer_size = ciw->count;
+	*rcd_buffer_size = DASD_ECKD_RCD_DATA_SIZE;
 	*rcd_buffer = rcd_buf;
 	return 0;
 out_error:
@@ -901,18 +978,18 @@ static int dasd_eckd_read_conf(struct dasd_device *device)
 	void *conf_data;
 	int conf_len, conf_data_saved;
 	int rc;
-	__u8 lpm;
+	__u8 lpm, opm;
 	struct dasd_eckd_private *private;
-	struct dasd_eckd_path *path_data;
+	struct dasd_path *path_data;
 
 	private = (struct dasd_eckd_private *) device->private;
-	path_data = (struct dasd_eckd_path *) &private->path_data;
-	path_data->opm = ccw_device_get_path_mask(device->cdev);
+	path_data = &device->path_data;
+	opm = ccw_device_get_path_mask(device->cdev);
 	lpm = 0x80;
 	conf_data_saved = 0;
 	/* get configuration data per operational path */
 	for (lpm = 0x80; lpm; lpm>>= 1) {
-		if (lpm & path_data->opm){
+		if (lpm & opm) {
 			rc = dasd_eckd_read_conf_lpm(device, &conf_data,
 						     &conf_len, lpm);
 			if (rc && rc != -EOPNOTSUPP) {	/* -EOPNOTSUPP is ok */
@@ -925,6 +1002,8 @@ static int dasd_eckd_read_conf(struct dasd_device *device)
 				DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s",
 						"No configuration data "
 						"retrieved");
+				/* no further analysis possible */
+				path_data->opm |= lpm;
 				continue;	/* no error */
 			}
 			/* save first valid configuration data */
@@ -948,6 +1027,7 @@ static int dasd_eckd_read_conf(struct dasd_device *device)
 				path_data->ppm |= lpm;
 				break;
 			}
+			path_data->opm |= lpm;
 			if (conf_data != private->conf_data)
 				kfree(conf_data);
 		}
@@ -955,6 +1035,140 @@ static int dasd_eckd_read_conf(struct dasd_device *device)
 	return 0;
 }
 
+static int verify_fcx_max_data(struct dasd_device *device, __u8 lpm)
+{
+	struct dasd_eckd_private *private;
+	int mdc;
+	u32 fcx_max_data;
+
+	private = (struct dasd_eckd_private *) device->private;
+	if (private->fcx_max_data) {
+		mdc = ccw_device_get_mdc(device->cdev, lpm);
+		if ((mdc < 0)) {
+			dev_warn(&device->cdev->dev,
+				 "Detecting the maximum data size for zHPF "
+				 "requests failed (rc=%d) for a new path %x\n",
+				 mdc, lpm);
+			return mdc;
+		}
+		fcx_max_data = mdc * FCX_MAX_DATA_FACTOR;
+		if (fcx_max_data < private->fcx_max_data) {
+			dev_warn(&device->cdev->dev,
+				 "The maximum data size for zHPF requests %u "
+				 "on a new path %x is below the active maximum "
+				 "%u\n", fcx_max_data, lpm,
+				 private->fcx_max_data);
+			return -EACCES;
+		}
+	}
+	return 0;
+}
+
+static void do_path_verification_work(struct work_struct *work)
+{
+	struct path_verification_work_data *data;
+	struct dasd_device *device;
+	__u8 lpm, opm, npm, ppm, epm;
+	unsigned long flags;
+	int rc;
+
+	data = container_of(work, struct path_verification_work_data, worker);
+	device = data->device;
+
+	opm = 0;
+	npm = 0;
+	ppm = 0;
+	epm = 0;
+	for (lpm = 0x80; lpm; lpm >>= 1) {
+		if (lpm & data->tbvpm) {
+			memset(data->rcd_buffer, 0, sizeof(data->rcd_buffer));
+			memset(&data->cqr, 0, sizeof(data->cqr));
+			data->cqr.cpaddr = &data->ccw;
+			rc = dasd_eckd_read_conf_immediately(device, &data->cqr,
+							     data->rcd_buffer,
+							     lpm);
+			if (!rc) {
+				switch (dasd_eckd_path_access(data->rcd_buffer,
+						     DASD_ECKD_RCD_DATA_SIZE)) {
+				case 0x02:
+					npm |= lpm;
+					break;
+				case 0x03:
+					ppm |= lpm;
+					break;
+				}
+				opm |= lpm;
+			} else if (rc == -EOPNOTSUPP) {
+				DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s",
+				       "path verification: No configuration "
+				       "data retrieved");
+				opm |= lpm;
+			} else if (rc == -EAGAIN) {
+				DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s",
+					"path verification: device is stopped,"
+					" try again later");
+				epm |= lpm;
+			} else {
+				dev_warn(&device->cdev->dev,
+					 "Reading device feature codes failed "
+					 "(rc=%d) for new path %x\n", rc, lpm);
+				continue;
+			}
+			if (verify_fcx_max_data(device, lpm)) {
+				opm &= ~lpm;
+				npm &= ~lpm;
+				ppm &= ~lpm;
+			}
+		}
+	}
+	/*
+	 * There is a small chance that a path is lost again between
+	 * above path verification and the following modification of
+	 * the device opm mask. We could avoid that race here by using
+	 * yet another path mask, but we rather deal with this unlikely
+	 * situation in dasd_start_IO.
+	 */
+	spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+	if (!device->path_data.opm && opm) {
+		device->path_data.opm = opm;
+		dasd_generic_path_operational(device);
+	} else
+		device->path_data.opm |= opm;
+	device->path_data.npm |= npm;
+	device->path_data.ppm |= ppm;
+	device->path_data.tbvpm |= epm;
+	spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+
+	dasd_put_device(device);
+	if (data->isglobal)
+		mutex_unlock(&dasd_path_verification_mutex);
+	else
+		kfree(data);
+}
+
+static int dasd_eckd_verify_path(struct dasd_device *device, __u8 lpm)
+{
+	struct path_verification_work_data *data;
+
+	data = kmalloc(sizeof(*data), GFP_ATOMIC | GFP_DMA);
+	if (!data) {
+		if (mutex_trylock(&dasd_path_verification_mutex)) {
+			data = path_verification_worker;
+			data->isglobal = 1;
+		} else
+			return -ENOMEM;
+	} else {
+		memset(data, 0, sizeof(*data));
+		data->isglobal = 0;
+	}
+	INIT_WORK(&data->worker, do_path_verification_work);
+	dasd_get_device(device);
+	data->device = device;
+	data->tbvpm = lpm;
+	schedule_work(&data->worker);
+	return 0;
+}
+
 static int dasd_eckd_read_features(struct dasd_device *device)
 {
 	struct dasd_psf_prssd_data *prssdp;
@@ -1105,6 +1319,37 @@ static void dasd_eckd_validate_server(struct dasd_device *device)
 			"returned rc=%d", private->uid.ssid, rc);
 }
 
+static u32 get_fcx_max_data(struct dasd_device *device)
+{
+#if defined(CONFIG_64BIT)
+	int tpm, mdc;
+	int fcx_in_css, fcx_in_gneq, fcx_in_features;
+	struct dasd_eckd_private *private;
+
+	if (dasd_nofcx)
+		return 0;
+	/* is transport mode supported? */
+	private = (struct dasd_eckd_private *) device->private;
+	fcx_in_css = css_general_characteristics.fcx;
+	fcx_in_gneq = private->gneq->reserved2[7] & 0x04;
+	fcx_in_features = private->features.feature[40] & 0x80;
+	tpm = fcx_in_css && fcx_in_gneq && fcx_in_features;
+
+	if (!tpm)
+		return 0;
+
+	mdc = ccw_device_get_mdc(device->cdev, 0);
+	if (mdc < 0) {
+		dev_warn(&device->cdev->dev, "Detecting the maximum supported"
+			 " data size for zHPF requests failed\n");
+		return 0;
+	} else
+		return mdc * FCX_MAX_DATA_FACTOR;
+#else
+	return 0;
+#endif
+}
+
 /*
  * Check device characteristics.
  * If the device is accessible using ECKD discipline, the device is enabled.
@@ -1223,6 +1468,8 @@ dasd_eckd_check_characteristics(struct dasd_device *device)
 	else
 		private->real_cyl = private->rdc_data.no_cyl;
 
+	private->fcx_max_data = get_fcx_max_data(device);
+
 	readonly = dasd_device_is_ro(device);
 	if (readonly)
 		set_bit(DASD_FLAG_DEVICE_RO, &device->flags);
@@ -1404,6 +1651,13 @@ static int dasd_eckd_end_analysis(struct dasd_block *block)
 		dasd_sfree_request(init_cqr, device);
 	}
 
+	if (device->features & DASD_FEATURE_USERAW) {
+		block->bp_block = DASD_RAW_BLOCKSIZE;
+		blk_per_trk = DASD_RAW_BLOCK_PER_TRACK;
+		block->s2b_shift = 3;
+		goto raw;
+	}
+
 	if (status == INIT_CQR_UNFORMATTED) {
 		dev_warn(&device->cdev->dev, "The DASD is not formatted\n");
 		return -EMEDIUMTYPE;
@@ -1441,6 +1695,7 @@ static int dasd_eckd_end_analysis(struct dasd_block *block)
 			dev_warn(&device->cdev->dev,
 				 "Track 0 has no records following the VTOC\n");
 	}
+
 	if (count_area != NULL && count_area->kl == 0) {
 		/* we found notthing violating our disk layout */
 		if (dasd_check_blocksize(count_area->dl) == 0)
@@ -1456,6 +1711,8 @@ static int dasd_eckd_end_analysis(struct dasd_block *block)
 		block->s2b_shift++;
 
 	blk_per_trk = recs_per_track(&private->rdc_data, 0, block->bp_block);
+
+raw:
 	block->blocks = (private->real_cyl *
 			  private->rdc_data.trk_per_cyl *
 			  blk_per_trk);
@@ -1716,6 +1973,7 @@ static void dasd_eckd_handle_terminated_request(struct dasd_ccw_req *cqr)
 	if (cqr->block && (cqr->startdev != cqr->block->base)) {
 		dasd_eckd_reset_ccw_to_base_io(cqr);
 		cqr->startdev = cqr->block->base;
+		cqr->lpm = cqr->block->base->path_data.opm;
 	}
 };
 
@@ -1744,9 +2002,9 @@ dasd_eckd_erp_postaction(struct dasd_ccw_req * cqr)
 	return dasd_default_erp_postaction;
 }
 
-
-static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device,
-						   struct irb *irb)
+static void dasd_eckd_check_for_device_change(struct dasd_device *device,
+					      struct dasd_ccw_req *cqr,
+					      struct irb *irb)
 {
 	char mask;
 	char *sense = NULL;
@@ -1770,40 +2028,41 @@ static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device,
 			/* schedule worker to reload device */
 			dasd_reload_device(device);
 		}
-
 		dasd_generic_handle_state_change(device);
 		return;
 	}
 
-	/* summary unit check */
 	sense = dasd_get_sense(irb);
-	if (sense && (sense[7] == 0x0D) &&
+	if (!sense)
+		return;
+
+	/* summary unit check */
+	if ((sense[7] == 0x0D) &&
 	    (scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK)) {
 		dasd_alias_handle_summary_unit_check(device, irb);
 		return;
 	}
 
 	/* service information message SIM */
-	if (sense && !(sense[27] & DASD_SENSE_BIT_0) &&
+	if (!cqr && !(sense[27] & DASD_SENSE_BIT_0) &&
 	    ((sense[6] & DASD_SIM_SENSE) == DASD_SIM_SENSE)) {
 		dasd_3990_erp_handle_sim(device, sense);
-		dasd_schedule_device_bh(device);
 		return;
 	}
 
-	if ((scsw_cc(&irb->scsw) == 1) && !sense &&
-	    (scsw_fctl(&irb->scsw) == SCSW_FCTL_START_FUNC) &&
-	    (scsw_actl(&irb->scsw) == SCSW_ACTL_START_PEND) &&
-	    (scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND)) {
-		/* fake irb do nothing, they are handled elsewhere */
-		dasd_schedule_device_bh(device);
-		return;
+	/* loss of device reservation is handled via base devices only
+	 * as alias devices may be used with several bases
+	 */
+	if (device->block && (sense[7] == 0x3F) &&
+	    (scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK) &&
+	    test_bit(DASD_FLAG_IS_RESERVED, &device->flags)) {
+		if (device->features & DASD_FEATURE_FAILONSLCK)
+			set_bit(DASD_FLAG_LOCK_STOLEN, &device->flags);
+		clear_bit(DASD_FLAG_IS_RESERVED, &device->flags);
+		dev_err(&device->cdev->dev,
+			"The device reservation was lost\n");
 	}
-
-	dasd_schedule_device_bh(device);
-	return;
-};
-
+}
 
 static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single(
 					       struct dasd_device *startdev,
@@ -1984,7 +2243,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single(
 	cqr->memdev = startdev;
 	cqr->block = block;
 	cqr->expires = startdev->default_expires * HZ;	/* default 5 minutes */
-	cqr->lpm = private->path_data.ppm;
+	cqr->lpm = startdev->path_data.ppm;
 	cqr->retries = 256;
 	cqr->buildclk = get_clock();
 	cqr->status = DASD_CQR_FILLED;
@@ -2161,7 +2420,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track(
 	cqr->memdev = startdev;
 	cqr->block = block;
 	cqr->expires = startdev->default_expires * HZ;	/* default 5 minutes */
-	cqr->lpm = private->path_data.ppm;
+	cqr->lpm = startdev->path_data.ppm;
 	cqr->retries = 256;
 	cqr->buildclk = get_clock();
 	cqr->status = DASD_CQR_FILLED;
@@ -2326,6 +2585,12 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track(
 	struct tidaw *last_tidaw = NULL;
 	int itcw_op;
 	size_t itcw_size;
+	u8 tidaw_flags;
+	unsigned int seg_len, part_len, len_to_track_end;
+	unsigned char new_track;
+	sector_t recid, trkid;
+	unsigned int offs;
+	unsigned int count, count_to_trk_end;
 
 	basedev = block->base;
 	private = (struct dasd_eckd_private *) basedev->private;
@@ -2341,12 +2606,16 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track(
 	/* trackbased I/O needs address all memory via TIDAWs,
 	 * not just for 64 bit addresses. This allows us to map
 	 * each segment directly to one tidaw.
+	 * In the case of write requests, additional tidaws may
+	 * be needed when a segment crosses a track boundary.
 	 */
 	trkcount = last_trk - first_trk + 1;
 	ctidaw = 0;
 	rq_for_each_segment(bv, req, iter) {
 		++ctidaw;
 	}
+	if (rq_data_dir(req) == WRITE)
+		ctidaw += (last_trk - first_trk);
 
 	/* Allocate the ccw request. */
 	itcw_size = itcw_calc_size(0, ctidaw, 0);
@@ -2354,15 +2623,6 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track(
 	if (IS_ERR(cqr))
 		return cqr;
 
-	cqr->cpmode = 1;
-	cqr->startdev = startdev;
-	cqr->memdev = startdev;
-	cqr->block = block;
-	cqr->expires = 100*HZ;
-	cqr->buildclk = get_clock();
-	cqr->status = DASD_CQR_FILLED;
-	cqr->retries = 10;
-
 	/* transfer length factor: how many bytes to read from the last track */
 	if (first_trk == last_trk)
 		tlf = last_offs - first_offs + 1;
@@ -2371,8 +2631,11 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track(
 	tlf *= blksize;
 
 	itcw = itcw_init(cqr->data, itcw_size, itcw_op, 0, ctidaw, 0);
+	if (IS_ERR(itcw)) {
+		dasd_sfree_request(cqr, startdev);
+		return ERR_PTR(-EINVAL);
+	}
 	cqr->cpaddr = itcw_get_tcw(itcw);
-
 	if (prepare_itcw(itcw, first_trk, last_trk,
 			 cmd, basedev, startdev,
 			 first_offs + 1,
@@ -2385,31 +2648,69 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track(
 		dasd_sfree_request(cqr, startdev);
 		return ERR_PTR(-EAGAIN);
 	}
-
 	/*
 	 * A tidaw can address 4k of memory, but must not cross page boundaries
 	 * We can let the block layer handle this by setting
 	 * blk_queue_segment_boundary to page boundaries and
 	 * blk_max_segment_size to page size when setting up the request queue.
+	 * For write requests, a TIDAW must not cross track boundaries, because
+	 * we have to set the CBC flag on the last tidaw for each track.
 	 */
-	rq_for_each_segment(bv, req, iter) {
-		dst = page_address(bv->bv_page) + bv->bv_offset;
-		last_tidaw = itcw_add_tidaw(itcw, 0x00, dst, bv->bv_len);
-		if (IS_ERR(last_tidaw))
-			return (struct dasd_ccw_req *)last_tidaw;
+	if (rq_data_dir(req) == WRITE) {
+		new_track = 1;
+		recid = first_rec;
+		rq_for_each_segment(bv, req, iter) {
+			dst = page_address(bv->bv_page) + bv->bv_offset;
+			seg_len = bv->bv_len;
+			while (seg_len) {
+				if (new_track) {
+					trkid = recid;
+					offs = sector_div(trkid, blk_per_trk);
+					count_to_trk_end = blk_per_trk - offs;
+					count = min((last_rec - recid + 1),
+						    (sector_t)count_to_trk_end);
+					len_to_track_end = count * blksize;
+					recid += count;
+					new_track = 0;
+				}
+				part_len = min(seg_len, len_to_track_end);
+				seg_len -= part_len;
+				len_to_track_end -= part_len;
+				/* We need to end the tidaw at track end */
+				if (!len_to_track_end) {
+					new_track = 1;
+					tidaw_flags = TIDAW_FLAGS_INSERT_CBC;
+				} else
+					tidaw_flags = 0;
+				last_tidaw = itcw_add_tidaw(itcw, tidaw_flags,
+							    dst, part_len);
+				if (IS_ERR(last_tidaw))
+					return ERR_PTR(-EINVAL);
+				dst += part_len;
+			}
+		}
+	} else {
+		rq_for_each_segment(bv, req, iter) {
+			dst = page_address(bv->bv_page) + bv->bv_offset;
+			last_tidaw = itcw_add_tidaw(itcw, 0x00,
+						    dst, bv->bv_len);
+			if (IS_ERR(last_tidaw))
+				return ERR_PTR(-EINVAL);
+		}
 	}
-
-	last_tidaw->flags |= 0x80;
+	last_tidaw->flags |= TIDAW_FLAGS_LAST;
+	last_tidaw->flags &= ~TIDAW_FLAGS_INSERT_CBC;
 	itcw_finalize(itcw);
 
 	if (blk_noretry_request(req) ||
 	    block->base->features & DASD_FEATURE_FAILFAST)
 		set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
+	cqr->cpmode = 1;
 	cqr->startdev = startdev;
 	cqr->memdev = startdev;
 	cqr->block = block;
 	cqr->expires = startdev->default_expires * HZ;	/* default 5 minutes */
-	cqr->lpm = private->path_data.ppm;
+	cqr->lpm = startdev->path_data.ppm;
 	cqr->retries = 256;
 	cqr->buildclk = get_clock();
 	cqr->status = DASD_CQR_FILLED;
@@ -2420,11 +2721,9 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 					       struct dasd_block *block,
 					       struct request *req)
 {
-	int tpm, cmdrtd, cmdwtd;
+	int cmdrtd, cmdwtd;
 	int use_prefix;
-#if defined(CONFIG_64BIT)
-	int fcx_in_css, fcx_in_gneq, fcx_in_features;
-#endif
+	int fcx_multitrack;
 	struct dasd_eckd_private *private;
 	struct dasd_device *basedev;
 	sector_t first_rec, last_rec;
@@ -2432,6 +2731,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 	unsigned int first_offs, last_offs;
 	unsigned int blk_per_trk, blksize;
 	int cdlspecial;
+	unsigned int data_size;
 	struct dasd_ccw_req *cqr;
 
 	basedev = block->base;
@@ -2450,15 +2750,11 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 	last_offs = sector_div(last_trk, blk_per_trk);
 	cdlspecial = (private->uses_cdl && first_rec < 2*blk_per_trk);
 
-	/* is transport mode supported? */
-#if defined(CONFIG_64BIT)
-	fcx_in_css = css_general_characteristics.fcx;
-	fcx_in_gneq = private->gneq->reserved2[7] & 0x04;
-	fcx_in_features = private->features.feature[40] & 0x80;
-	tpm = fcx_in_css && fcx_in_gneq && fcx_in_features;
-#else
-	tpm = 0;
-#endif
+	fcx_multitrack = private->features.feature[40] & 0x20;
+	data_size = blk_rq_bytes(req);
+	/* tpm write request add CBC data on each track boundary */
+	if (rq_data_dir(req) == WRITE)
+		data_size += (last_trk - first_trk) * 4;
 
 	/* is read track data and write track data in command mode supported? */
 	cmdrtd = private->features.feature[9] & 0x20;
@@ -2468,13 +2764,15 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 	cqr = NULL;
 	if (cdlspecial || dasd_page_cache) {
 		/* do nothing, just fall through to the cmd mode single case */
-	} else if (!dasd_nofcx && tpm && (first_trk == last_trk)) {
+	} else if ((data_size <= private->fcx_max_data)
+		   && (fcx_multitrack || (first_trk == last_trk))) {
 		cqr = dasd_eckd_build_cp_tpm_track(startdev, block, req,
 						    first_rec, last_rec,
 						    first_trk, last_trk,
 						    first_offs, last_offs,
 						    blk_per_trk, blksize);
-		if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN)
+		if (IS_ERR(cqr) && (PTR_ERR(cqr) != -EAGAIN) &&
+		    (PTR_ERR(cqr) != -ENOMEM))
 			cqr = NULL;
 	} else if (use_prefix &&
 		   (((rq_data_dir(req) == READ) && cmdrtd) ||
@@ -2484,7 +2782,8 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 						   first_trk, last_trk,
 						   first_offs, last_offs,
 						   blk_per_trk, blksize);
-		if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN)
+		if (IS_ERR(cqr) && (PTR_ERR(cqr) != -EAGAIN) &&
+		    (PTR_ERR(cqr) != -ENOMEM))
 			cqr = NULL;
 	}
 	if (!cqr)
@@ -2496,6 +2795,135 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
 	return cqr;
 }
 
+static struct dasd_ccw_req *dasd_raw_build_cp(struct dasd_device *startdev,
+					       struct dasd_block *block,
+					       struct request *req)
+{
+	struct dasd_eckd_private *private;
+	unsigned long *idaws;
+	struct dasd_device *basedev;
+	struct dasd_ccw_req *cqr;
+	struct ccw1 *ccw;
+	struct req_iterator iter;
+	struct bio_vec *bv;
+	char *dst;
+	unsigned char cmd;
+	unsigned int trkcount;
+	unsigned int seg_len, len_to_track_end;
+	unsigned int first_offs;
+	unsigned int cidaw, cplength, datasize;
+	sector_t first_trk, last_trk;
+	unsigned int pfx_datasize;
+
+	/*
+	 * raw track access needs to be mutiple of 64k and on 64k boundary
+	 */
+	if ((blk_rq_pos(req) % DASD_RAW_SECTORS_PER_TRACK) != 0) {
+		cqr = ERR_PTR(-EINVAL);
+		goto out;
+	}
+	if (((blk_rq_pos(req) + blk_rq_sectors(req)) %
+	     DASD_RAW_SECTORS_PER_TRACK) != 0) {
+		cqr = ERR_PTR(-EINVAL);
+		goto out;
+	}
+
+	first_trk = blk_rq_pos(req) / DASD_RAW_SECTORS_PER_TRACK;
+	last_trk = (blk_rq_pos(req) + blk_rq_sectors(req) - 1) /
+		DASD_RAW_SECTORS_PER_TRACK;
+	trkcount = last_trk - first_trk + 1;
+	first_offs = 0;
+	basedev = block->base;
+	private = (struct dasd_eckd_private *) basedev->private;
+
+	if (rq_data_dir(req) == READ)
+		cmd = DASD_ECKD_CCW_READ_TRACK;
+	else if (rq_data_dir(req) == WRITE)
+		cmd = DASD_ECKD_CCW_WRITE_FULL_TRACK;
+	else {
+		cqr = ERR_PTR(-EINVAL);
+		goto out;
+	}
+
+	/*
+	 * Raw track based I/O needs IDAWs for each page,
+	 * and not just for 64 bit addresses.
+	 */
+	cidaw = trkcount * DASD_RAW_BLOCK_PER_TRACK;
+
+	/* 1x prefix + one read/write ccw per track */
+	cplength = 1 + trkcount;
+
+	/*
+	 * struct PFX_eckd_data has up to 2 byte as extended parameter
+	 * this is needed for write full track and has to be mentioned
+	 * seperately
+	 * add 8 instead of 2 to keep 8 byte boundary
+	 */
+	pfx_datasize = sizeof(struct PFX_eckd_data) + 8;
+
+	datasize = pfx_datasize + cidaw * sizeof(unsigned long long);
+
+	/* Allocate the ccw request. */
+	cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength,
+				   datasize, startdev);
+	if (IS_ERR(cqr))
+		goto out;
+	ccw = cqr->cpaddr;
+
+	if (prefix_LRE(ccw++, cqr->data, first_trk, last_trk, cmd,
+		       basedev, startdev, 1 /* format */, first_offs + 1,
+		       trkcount, 0, 0) == -EAGAIN) {
+		/* Clock not in sync and XRC is enabled.
+		 * Try again later.
+		 */
+		dasd_sfree_request(cqr, startdev);
+		cqr = ERR_PTR(-EAGAIN);
+		goto out;
+	}
+
+	idaws = (unsigned long *)(cqr->data + pfx_datasize);
+
+	len_to_track_end = 0;
+
+	rq_for_each_segment(bv, req, iter) {
+		dst = page_address(bv->bv_page) + bv->bv_offset;
+		seg_len = bv->bv_len;
+		if (!len_to_track_end) {
+			ccw[-1].flags |= CCW_FLAG_CC;
+			ccw->cmd_code = cmd;
+			/* maximum 3390 track size */
+			ccw->count = 57326;
+			/* 64k map to one track */
+			len_to_track_end = 65536;
+			ccw->cda = (__u32)(addr_t)idaws;
+			ccw->flags |= CCW_FLAG_IDA;
+			ccw->flags |= CCW_FLAG_SLI;
+			ccw++;
+		}
+		len_to_track_end -= seg_len;
+		idaws = idal_create_words(idaws, dst, seg_len);
+	}
+
+	if (blk_noretry_request(req) ||
+	    block->base->features & DASD_FEATURE_FAILFAST)
+		set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
+	cqr->startdev = startdev;
+	cqr->memdev = startdev;
+	cqr->block = block;
+	cqr->expires = startdev->default_expires * HZ;
+	cqr->lpm = startdev->path_data.ppm;
+	cqr->retries = 256;
+	cqr->buildclk = get_clock();
+	cqr->status = DASD_CQR_FILLED;
+
+	if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN)
+		cqr = NULL;
+out:
+	return cqr;
+}
+
+
 static int
 dasd_eckd_free_cp(struct dasd_ccw_req *cqr, struct request *req)
 {
@@ -2600,7 +3028,10 @@ static struct dasd_ccw_req *dasd_eckd_build_alias_cp(struct dasd_device *base,
 
 	spin_lock_irqsave(get_ccwdev_lock(startdev->cdev), flags);
 	private->count++;
-	cqr = dasd_eckd_build_cp(startdev, block, req);
+	if ((base->features & DASD_FEATURE_USERAW))
+		cqr = dasd_raw_build_cp(startdev, block, req);
+	else
+		cqr = dasd_eckd_build_cp(startdev, block, req);
 	if (IS_ERR(cqr))
 		private->count--;
 	spin_unlock_irqrestore(get_ccwdev_lock(startdev->cdev), flags);
@@ -2688,6 +3119,8 @@ dasd_eckd_release(struct dasd_device *device)
 	cqr->status = DASD_CQR_FILLED;
 
 	rc = dasd_sleep_on_immediatly(cqr);
+	if (!rc)
+		clear_bit(DASD_FLAG_IS_RESERVED, &device->flags);
 
 	if (useglobal)
 		mutex_unlock(&dasd_reserve_mutex);
@@ -2741,6 +3174,8 @@ dasd_eckd_reserve(struct dasd_device *device)
 	cqr->status = DASD_CQR_FILLED;
 
 	rc = dasd_sleep_on_immediatly(cqr);
+	if (!rc)
+		set_bit(DASD_FLAG_IS_RESERVED, &device->flags);
 
 	if (useglobal)
 		mutex_unlock(&dasd_reserve_mutex);
@@ -2793,6 +3228,8 @@ dasd_eckd_steal_lock(struct dasd_device *device)
 	cqr->status = DASD_CQR_FILLED;
 
 	rc = dasd_sleep_on_immediatly(cqr);
+	if (!rc)
+		set_bit(DASD_FLAG_IS_RESERVED, &device->flags);
 
 	if (useglobal)
 		mutex_unlock(&dasd_reserve_mutex);
@@ -2845,6 +3282,7 @@ static int dasd_eckd_snid(struct dasd_device *device,
 	cqr->memdev = device;
 	clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
 	set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
+	set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
 	cqr->retries = 5;
 	cqr->expires = 10 * HZ;
 	cqr->buildclk = get_clock();
@@ -3279,10 +3717,8 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device,
 {
 	char *page;
 	int len, sl, sct, residual;
-
 	struct tsb *tsb;
-	u8 *sense;
-
+	u8 *sense, *rcq;
 
 	page = (char *) get_zeroed_page(GFP_ATOMIC);
 	if (page == NULL) {
@@ -3348,12 +3784,15 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device,
 		case 2: /* ts_ddpc */
 			len += sprintf(page + len, KERN_ERR PRINTK_HEADER
 			       " tsb->tsa.ddpc.rc %d\n", tsb->tsa.ddpc.rc);
-			len += sprintf(page + len, KERN_ERR PRINTK_HEADER
-			       " tsb->tsa.ddpc.rcq:  ");
-			for (sl = 0; sl < 16; sl++) {
+			for (sl = 0; sl < 2; sl++) {
+				len += sprintf(page + len,
+					       KERN_ERR PRINTK_HEADER
+					       " tsb->tsa.ddpc.rcq %2d-%2d: ",
+					       (8 * sl), ((8 * sl) + 7));
+				rcq = tsb->tsa.ddpc.rcq;
 				for (sct = 0; sct < 8; sct++) {
 					len += sprintf(page + len, " %02x",
-						       tsb->tsa.ddpc.rcq[sl]);
+						       rcq[8 * sl + sct]);
 				}
 				len += sprintf(page + len, "\n");
 			}
@@ -3550,6 +3989,7 @@ static struct ccw_driver dasd_eckd_driver = {
 	.set_offline = dasd_generic_set_offline,
 	.set_online  = dasd_eckd_set_online,
 	.notify      = dasd_generic_notify,
+	.path_event  = dasd_generic_path_event,
 	.freeze      = dasd_generic_pm_freeze,
 	.thaw	     = dasd_generic_restore_device,
 	.restore     = dasd_generic_restore_device,
@@ -3573,10 +4013,11 @@ static struct dasd_discipline dasd_eckd_discipline = {
 	.owner = THIS_MODULE,
 	.name = "ECKD",
 	.ebcname = "ECKD",
-	.max_blocks = 240,
+	.max_blocks = 190,
 	.check_device = dasd_eckd_check_characteristics,
 	.uncheck_device = dasd_eckd_uncheck_device,
 	.do_analysis = dasd_eckd_do_analysis,
+	.verify_path = dasd_eckd_verify_path,
 	.ready_to_online = dasd_eckd_ready_to_online,
 	.online_to_ready = dasd_eckd_online_to_ready,
 	.fill_geometry = dasd_eckd_fill_geometry,
@@ -3586,7 +4027,7 @@ static struct dasd_discipline dasd_eckd_discipline = {
 	.format_device = dasd_eckd_format_device,
 	.erp_action = dasd_eckd_erp_action,
 	.erp_postaction = dasd_eckd_erp_postaction,
-	.handle_unsolicited_interrupt = dasd_eckd_handle_unsolicited_interrupt,
+	.check_for_device_change = dasd_eckd_check_for_device_change,
 	.build_cp = dasd_eckd_build_alias_cp,
 	.free_cp = dasd_eckd_free_alias_cp,
 	.dump_sense = dasd_eckd_dump_sense,
@@ -3609,11 +4050,19 @@ dasd_eckd_init(void)
 				   GFP_KERNEL | GFP_DMA);
 	if (!dasd_reserve_req)
 		return -ENOMEM;
+	path_verification_worker = kmalloc(sizeof(*path_verification_worker),
+				   GFP_KERNEL | GFP_DMA);
+	if (!path_verification_worker) {
+		kfree(dasd_reserve_req);
+		return -ENOMEM;
+	}
 	ret = ccw_driver_register(&dasd_eckd_driver);
 	if (!ret)
 		wait_for_device_probe();
-	else
+	else {
+		kfree(path_verification_worker);
 		kfree(dasd_reserve_req);
+	}
 	return ret;
 }
 
@@ -3621,6 +4070,7 @@ static void __exit
 dasd_eckd_cleanup(void)
 {
 	ccw_driver_unregister(&dasd_eckd_driver);
+	kfree(path_verification_worker);
 	kfree(dasd_reserve_req);
 }
 

drivers/s390/block/dasd_eckd.h

@@ -37,14 +37,17 @@
 #define DASD_ECKD_CCW_WRITE_KD_MT	 0x8d
 #define DASD_ECKD_CCW_READ_KD_MT	 0x8e
 #define DASD_ECKD_CCW_RELEASE		 0x94
+#define DASD_ECKD_CCW_WRITE_FULL_TRACK	 0x95
 #define DASD_ECKD_CCW_READ_CKD_MT	 0x9e
 #define DASD_ECKD_CCW_WRITE_CKD_MT	 0x9d
 #define DASD_ECKD_CCW_WRITE_TRACK_DATA	 0xA5
 #define DASD_ECKD_CCW_READ_TRACK_DATA	 0xA6
 #define DASD_ECKD_CCW_RESERVE		 0xB4
+#define DASD_ECKD_CCW_READ_TRACK	 0xDE
 #define DASD_ECKD_CCW_PFX		 0xE7
 #define DASD_ECKD_CCW_PFX_READ		 0xEA
 #define DASD_ECKD_CCW_RSCK		 0xF9
+#define DASD_ECKD_CCW_RCD		 0xFA
 
 /*
  * Perform Subsystem Function / Sub-Orders
@@ -57,6 +60,11 @@
  */
 #define LV_COMPAT_CYL 0xFFFE
 
+
+#define FCX_MAX_DATA_FACTOR 65536
+#define DASD_ECKD_RCD_DATA_SIZE 256
+
+
 /*****************************************************************************
  * SECTION: Type Definitions
  ****************************************************************************/
@@ -331,12 +339,6 @@ struct dasd_gneq {
 	__u8 reserved2[22];
 } __attribute__ ((packed));
 
-struct dasd_eckd_path {
-	__u8 opm;
-	__u8 ppm;
-	__u8 npm;
-};
-
 struct dasd_rssd_features {
 	char feature[256];
 } __attribute__((packed));
@@ -442,7 +444,6 @@ struct dasd_eckd_private {
 	struct vd_sneq *vdsneq;
 	struct dasd_gneq *gneq;
 
-	struct dasd_eckd_path path_data;
 	struct eckd_count count_area[5];
 	int init_cqr_status;
 	int uses_cdl;
@@ -455,6 +456,8 @@ struct dasd_eckd_private {
 	struct alias_pav_group *pavgroup;
 	struct alias_lcu *lcu;
 	int count;
+
+	u32 fcx_max_data;
 };
 
 

drivers/s390/block/dasd_eer.c

@@ -473,6 +473,7 @@ int dasd_eer_enable(struct dasd_device *device)
 	cqr->retries = 255;
 	cqr->expires = 10 * HZ;
 	clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+	set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
 
 	ccw = cqr->cpaddr;
 	ccw->cmd_code = DASD_ECKD_CCW_SNSS;

drivers/s390/block/dasd_erp.c

@@ -96,7 +96,8 @@ dasd_default_erp_action(struct dasd_ccw_req *cqr)
 		DBF_DEV_EVENT(DBF_DEBUG, device,
                              "default ERP called (%i retries left)",
                              cqr->retries);
-		cqr->lpm    = LPM_ANYPATH;
+		if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags))
+			cqr->lpm = device->path_data.opm;
 		cqr->status = DASD_CQR_FILLED;
         } else {
 		pr_err("%s: default ERP has run out of retries and failed\n",

drivers/s390/block/dasd_fba.c

@@ -73,6 +73,7 @@ static struct ccw_driver dasd_fba_driver = {
 	.set_offline = dasd_generic_set_offline,
 	.set_online  = dasd_fba_set_online,
 	.notify      = dasd_generic_notify,
+	.path_event  = dasd_generic_path_event,
 	.freeze      = dasd_generic_pm_freeze,
 	.thaw	     = dasd_generic_restore_device,
 	.restore     = dasd_generic_restore_device,
@@ -164,6 +165,7 @@ dasd_fba_check_characteristics(struct dasd_device *device)
 	}
 
 	device->default_expires = DASD_EXPIRES;
+	device->path_data.opm = LPM_ANYPATH;
 
 	readonly = dasd_device_is_ro(device);
 	if (readonly)
@@ -231,24 +233,16 @@ dasd_fba_erp_postaction(struct dasd_ccw_req * cqr)
 	return NULL;
 }
 
-static void dasd_fba_handle_unsolicited_interrupt(struct dasd_device *device,
-						   struct irb *irb)
+static void dasd_fba_check_for_device_change(struct dasd_device *device,
+					     struct dasd_ccw_req *cqr,
+					     struct irb *irb)
 {
 	char mask;
 
 	/* first of all check for state change pending interrupt */
 	mask = DEV_STAT_ATTENTION | DEV_STAT_DEV_END | DEV_STAT_UNIT_EXCEP;
-	if ((irb->scsw.cmd.dstat & mask) == mask) {
+	if ((irb->scsw.cmd.dstat & mask) == mask)
 		dasd_generic_handle_state_change(device);
-		return;
-	}
-
-	/* check for unsolicited interrupts */
-	DBF_DEV_EVENT(DBF_WARNING, device, "%s",
-		    "unsolicited interrupt received");
-	device->discipline->dump_sense_dbf(device, irb, "unsolicited");
-	dasd_schedule_device_bh(device);
-	return;
 };
 
 static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev,
@@ -596,13 +590,14 @@ static struct dasd_discipline dasd_fba_discipline = {
 	.max_blocks = 96,
 	.check_device = dasd_fba_check_characteristics,
 	.do_analysis = dasd_fba_do_analysis,
+	.verify_path = dasd_generic_verify_path,
 	.fill_geometry = dasd_fba_fill_geometry,
 	.start_IO = dasd_start_IO,
 	.term_IO = dasd_term_IO,
 	.handle_terminated_request = dasd_fba_handle_terminated_request,
 	.erp_action = dasd_fba_erp_action,
 	.erp_postaction = dasd_fba_erp_postaction,
-	.handle_unsolicited_interrupt = dasd_fba_handle_unsolicited_interrupt,
+	.check_for_device_change = dasd_fba_check_for_device_change,
 	.build_cp = dasd_fba_build_cp,
 	.free_cp = dasd_fba_free_cp,
 	.dump_sense = dasd_fba_dump_sense,

drivers/s390/block/dasd_int.h

@@ -231,6 +231,11 @@ struct dasd_ccw_req {
 /* per dasd_ccw_req flags */
 #define DASD_CQR_FLAGS_USE_ERP   0	/* use ERP for this request */
 #define DASD_CQR_FLAGS_FAILFAST  1	/* FAILFAST */
+#define DASD_CQR_VERIFY_PATH	 2	/* path verification request */
+#define DASD_CQR_ALLOW_SLOCK	 3	/* Try this request even when lock was
+					 * stolen. Should not be combined with
+					 * DASD_CQR_FLAGS_USE_ERP
+					 */
 
 /* Signature for error recovery functions. */
 typedef struct dasd_ccw_req *(*dasd_erp_fn_t) (struct dasd_ccw_req *);
@@ -286,6 +291,14 @@ struct dasd_discipline {
 	 */
 	int (*do_analysis) (struct dasd_block *);
 
+	/*
+	 * This function is called, when new paths become available.
+	 * Disciplins may use this callback to do necessary setup work,
+	 * e.g. verify that new path is compatible with the current
+	 * configuration.
+	 */
+	int (*verify_path)(struct dasd_device *, __u8);
+
 	/*
 	 * Last things to do when a device is set online, and first things
 	 * when it is set offline.
@@ -325,9 +338,9 @@ struct dasd_discipline {
 	void (*dump_sense) (struct dasd_device *, struct dasd_ccw_req *,
 			    struct irb *);
 	void (*dump_sense_dbf) (struct dasd_device *, struct irb *, char *);
-
-	void (*handle_unsolicited_interrupt) (struct dasd_device *,
-					      struct irb *);
+	void (*check_for_device_change) (struct dasd_device *,
+					 struct dasd_ccw_req *,
+					 struct irb *);
 
         /* i/o control functions. */
 	int (*fill_geometry) (struct dasd_block *, struct hd_geometry *);
@@ -362,6 +375,13 @@ extern struct dasd_discipline *dasd_diag_discipline_pointer;
 #define DASD_EER_STATECHANGE 3
 #define DASD_EER_PPRCSUSPEND 4
 
+struct dasd_path {
+	__u8 opm;
+	__u8 tbvpm;
+	__u8 ppm;
+	__u8 npm;
+};
+
 struct dasd_device {
 	/* Block device stuff. */
 	struct dasd_block *block;
@@ -377,6 +397,7 @@ struct dasd_device {
 	struct dasd_discipline *discipline;
 	struct dasd_discipline *base_discipline;
 	char *private;
+	struct dasd_path path_data;
 
 	/* Device state and target state. */
 	int state, target;
@@ -456,6 +477,9 @@ struct dasd_block {
 					 * confuse this with the user specified
 					 * read-only feature.
 					 */
+#define DASD_FLAG_IS_RESERVED	7	/* The device is reserved */
+#define DASD_FLAG_LOCK_STOLEN	8	/* The device lock was stolen */
+
 
 void dasd_put_device_wake(struct dasd_device *);
 
@@ -620,10 +644,15 @@ void dasd_generic_remove (struct ccw_device *cdev);
 int dasd_generic_set_online(struct ccw_device *, struct dasd_discipline *);
 int dasd_generic_set_offline (struct ccw_device *cdev);
 int dasd_generic_notify(struct ccw_device *, int);
+int dasd_generic_last_path_gone(struct dasd_device *);
+int dasd_generic_path_operational(struct dasd_device *);
+
 void dasd_generic_handle_state_change(struct dasd_device *);
 int dasd_generic_pm_freeze(struct ccw_device *);
 int dasd_generic_restore_device(struct ccw_device *);
 enum uc_todo dasd_generic_uc_handler(struct ccw_device *, struct irb *);
+void dasd_generic_path_event(struct ccw_device *, int *);
+int dasd_generic_verify_path(struct dasd_device *, __u8);
 
 int dasd_generic_read_dev_chars(struct dasd_device *, int, void *, int);
 char *dasd_get_sense(struct irb *);

drivers/s390/char/Kconfig

@@ -2,76 +2,85 @@ comment "S/390 character device drivers"
 	depends on S390
 
 config TN3270
-	tristate "Support for locally attached 3270 terminals"
+	def_tristate y
+	prompt "Support for locally attached 3270 terminals"
 	depends on CCW
 	help
 	  Include support for IBM 3270 terminals.
 
 config TN3270_TTY
-	tristate "Support for tty input/output on 3270 terminals"
+	def_tristate y
+	prompt "Support for tty input/output on 3270 terminals"
 	depends on TN3270
 	help
 	  Include support for using an IBM 3270 terminal as a Linux tty.
 
 config TN3270_FS
-	tristate "Support for fullscreen applications on 3270 terminals"
+	def_tristate m
+	prompt "Support for fullscreen applications on 3270 terminals"
 	depends on TN3270
 	help
 	  Include support for fullscreen applications on an IBM 3270 terminal.
 
 config TN3270_CONSOLE
-	bool "Support for console on 3270 terminal"
+	def_bool y
+	prompt "Support for console on 3270 terminal"
 	depends on TN3270=y && TN3270_TTY=y
 	help
 	  Include support for using an IBM 3270 terminal as a Linux system
 	  console.  Available only if 3270 support is compiled in statically.
 
 config TN3215
-	bool "Support for 3215 line mode terminal"
+	def_bool y
+	prompt "Support for 3215 line mode terminal"
 	depends on CCW
 	help
 	  Include support for IBM 3215 line-mode terminals.
 
 config TN3215_CONSOLE
-	bool "Support for console on 3215 line mode terminal"
+	def_bool y
+	prompt "Support for console on 3215 line mode terminal"
 	depends on TN3215
 	help
 	  Include support for using an IBM 3215 line-mode terminal as a
 	  Linux system console.
 
 config CCW_CONSOLE
-	bool
-	depends on TN3215_CONSOLE || TN3270_CONSOLE
-	default y
+	def_bool y if TN3215_CONSOLE || TN3270_CONSOLE
 
 config SCLP_TTY
-	bool "Support for SCLP line mode terminal"
+	def_bool y
+	prompt "Support for SCLP line mode terminal"
 	depends on S390
 	help
 	  Include support for IBM SCLP line-mode terminals.
 
 config SCLP_CONSOLE
-	bool "Support for console on SCLP line mode terminal"
+	def_bool y
+	prompt "Support for console on SCLP line mode terminal"
 	depends on SCLP_TTY
 	help
 	  Include support for using an IBM HWC line-mode terminal as the Linux
 	  system console.
 
 config SCLP_VT220_TTY
-	bool "Support for SCLP VT220-compatible terminal"
+	def_bool y
+	prompt "Support for SCLP VT220-compatible terminal"
 	depends on S390
 	help
 	  Include support for an IBM SCLP VT220-compatible terminal.
 
 config SCLP_VT220_CONSOLE
-	bool "Support for console on SCLP VT220-compatible terminal"
+	def_bool y
+	prompt "Support for console on SCLP VT220-compatible terminal"
 	depends on SCLP_VT220_TTY
 	help
 	  Include support for using an IBM SCLP VT220-compatible terminal as a
 	  Linux system console.
 
 config SCLP_CPI
-	tristate "Control-Program Identification"
+	def_tristate m
+	prompt "Control-Program Identification"
 	depends on S390
 	help
 	  This option enables the hardware console interface for system
@@ -83,7 +92,8 @@ config SCLP_CPI
 	  need this feature and intend to run your kernel in LPAR.
 
 config SCLP_ASYNC
-	tristate "Support for Call Home via Asynchronous SCLP Records"
+	def_tristate m
+	prompt "Support for Call Home via Asynchronous SCLP Records"
 	depends on S390
 	help
 	  This option enables the call home function, which is able to inform
@@ -93,7 +103,8 @@ config SCLP_ASYNC
 	  need this feature and intend to run your kernel in LPAR.
 
 config S390_TAPE
-	tristate "S/390 tape device support"
+	def_tristate m
+	prompt "S/390 tape device support"
 	depends on CCW
 	help
 	  Select this option if you want to access channel-attached tape
@@ -109,7 +120,8 @@ comment "S/390 tape interface support"
 	depends on S390_TAPE
 
 config S390_TAPE_BLOCK
-	bool "Support for tape block devices"
+	def_bool y
+	prompt "Support for tape block devices"
 	depends on S390_TAPE && BLOCK
 	help
 	  Select this option if you want to access your channel-attached tape
@@ -123,7 +135,8 @@ comment "S/390 tape hardware support"
 	depends on S390_TAPE
 
 config S390_TAPE_34XX
-	tristate "Support for 3480/3490 tape hardware"
+	def_tristate m
+	prompt "Support for 3480/3490 tape hardware"
 	depends on S390_TAPE
 	help
 	  Select this option if you want to access IBM 3480/3490 magnetic
@@ -131,7 +144,8 @@ config S390_TAPE_34XX
 	  It is safe to say "Y" here.
 
 config S390_TAPE_3590
-	tristate "Support for 3590 tape hardware"
+	def_tristate m
+	prompt "Support for 3590 tape hardware"
 	depends on S390_TAPE
 	help
 	  Select this option if you want to access IBM 3590 magnetic
@@ -139,7 +153,8 @@ config S390_TAPE_3590
 	  It is safe to say "Y" here.
 
 config VMLOGRDR
-	tristate "Support for the z/VM recording system services (VM only)"
+	def_tristate m
+	prompt "Support for the z/VM recording system services (VM only)"
 	depends on IUCV
 	help
 	  Select this option if you want to be able to receive records collected
@@ -148,29 +163,31 @@ config VMLOGRDR
 	  This driver depends on the IUCV support driver.
 
 config VMCP
-	bool "Support for the z/VM CP interface"
+	def_bool y
+	prompt "Support for the z/VM CP interface"
 	depends on S390
 	help
 	  Select this option if you want to be able to interact with the control
 	  program on z/VM
 
 config MONREADER
-	tristate "API for reading z/VM monitor service records"
+	def_tristate m
+	prompt "API for reading z/VM monitor service records"
 	depends on IUCV
 	help
 	  Character device driver for reading z/VM monitor service records
 
 config MONWRITER
-	tristate "API for writing z/VM monitor service records"
+	def_tristate m
+	prompt "API for writing z/VM monitor service records"
 	depends on S390
-	default "m"
 	help
 	  Character device driver for writing z/VM monitor service records
 
 config S390_VMUR
-	tristate "z/VM unit record device driver"
+	def_tristate m
+	prompt "z/VM unit record device driver"
 	depends on S390
-	default "m"
 	help
 	  Character device driver for z/VM reader, puncher and printer.
 

drivers/s390/char/con3215.c

@@ -9,6 +9,7 @@
  *	      Dan Morrison, IBM Corporation <dmorriso@cse.buffalo.edu>
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/types.h>
 #include <linux/kdev_t.h>
@@ -361,6 +362,7 @@ static void raw3215_irq(struct ccw_device *cdev, unsigned long intparm,
 	int cstat, dstat;
 	int count;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_C15]++;
 	raw = dev_get_drvdata(&cdev->dev);
 	req = (struct raw3215_req *) intparm;
 	cstat = irb->scsw.cmd.cstat;

drivers/s390/char/raw3270.c

@@ -7,6 +7,7 @@
  *     Copyright IBM Corp. 2003, 2009
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/err.h>
 #include <linux/init.h>
@@ -329,6 +330,7 @@ raw3270_irq (struct ccw_device *cdev, unsigned long intparm, struct irb *irb)
 	struct raw3270_request *rq;
 	int rc;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_C70]++;
 	rp = dev_get_drvdata(&cdev->dev);
 	if (!rp)
 		return;

drivers/s390/char/sclp.c

@@ -7,6 +7,7 @@
  *	      Martin Schwidefsky <schwidefsky@de.ibm.com>
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/err.h>
 #include <linux/spinlock.h>
@@ -18,16 +19,14 @@
 #include <linux/suspend.h>
 #include <linux/completion.h>
 #include <linux/platform_device.h>
-#include <asm/types.h>
 #include <asm/s390_ext.h>
+#include <asm/types.h>
+#include <asm/irq.h>
 
 #include "sclp.h"
 
 #define SCLP_HEADER		"sclp: "
 
-/* Structure for register_early_external_interrupt. */
-static ext_int_info_t ext_int_info_hwc;
-
 /* Lock to protect internal data consistency. */
 static DEFINE_SPINLOCK(sclp_lock);
 
@@ -402,6 +401,7 @@ static void sclp_interrupt_handler(unsigned int ext_int_code,
 	u32 finished_sccb;
 	u32 evbuf_pending;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_SCP]++;
 	spin_lock(&sclp_lock);
 	finished_sccb = param32 & 0xfffffff8;
 	evbuf_pending = param32 & 0x3;
@@ -824,6 +824,7 @@ static void sclp_check_handler(unsigned int ext_int_code,
 {
 	u32 finished_sccb;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_SCP]++;
 	finished_sccb = param32 & 0xfffffff8;
 	/* Is this the interrupt we are waiting for? */
 	if (finished_sccb == 0)
@@ -866,8 +867,7 @@ sclp_check_interface(void)
 
 	spin_lock_irqsave(&sclp_lock, flags);
 	/* Prepare init mask command */
-	rc = register_early_external_interrupt(0x2401, sclp_check_handler,
-					       &ext_int_info_hwc);
+	rc = register_external_interrupt(0x2401, sclp_check_handler);
 	if (rc) {
 		spin_unlock_irqrestore(&sclp_lock, flags);
 		return rc;
@@ -900,8 +900,7 @@ sclp_check_interface(void)
 		} else
 			rc = -EBUSY;
 	}
-	unregister_early_external_interrupt(0x2401, sclp_check_handler,
-					    &ext_int_info_hwc);
+	unregister_external_interrupt(0x2401, sclp_check_handler);
 	spin_unlock_irqrestore(&sclp_lock, flags);
 	return rc;
 }
@@ -1064,8 +1063,7 @@ sclp_init(void)
 	if (rc)
 		goto fail_init_state_uninitialized;
 	/* Register interrupt handler */
-	rc = register_early_external_interrupt(0x2401, sclp_interrupt_handler,
-					       &ext_int_info_hwc);
+	rc = register_external_interrupt(0x2401, sclp_interrupt_handler);
 	if (rc)
 		goto fail_unregister_reboot_notifier;
 	sclp_init_state = sclp_init_state_initialized;

drivers/s390/char/sclp_config.c

@@ -33,6 +33,7 @@ static void sclp_cpu_capability_notify(struct work_struct *work)
 	int cpu;
 	struct sys_device *sysdev;
 
+	s390_adjust_jiffies();
 	pr_warning("cpu capability changed.\n");
 	get_online_cpus();
 	for_each_online_cpu(cpu) {

drivers/s390/char/tape_core.c

@@ -14,6 +14,7 @@
 #define KMSG_COMPONENT "tape"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/init.h>	     // for kernel parameters
 #include <linux/kmod.h>	     // for requesting modules
@@ -1114,6 +1115,7 @@ __tape_do_irq (struct ccw_device *cdev, unsigned long intparm, struct irb *irb)
 	struct tape_request *request;
 	int rc;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_TAP]++;
 	device = dev_get_drvdata(&cdev->dev);
 	if (device == NULL) {
 		return;

drivers/s390/char/vmur.c

@@ -11,6 +11,7 @@
 #define KMSG_COMPONENT "vmur"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/cdev.h>
 #include <linux/slab.h>
 
@@ -302,6 +303,7 @@ static void ur_int_handler(struct ccw_device *cdev, unsigned long intparm,
 {
 	struct urdev *urd;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_VMR]++;
 	TRACE("ur_int_handler: intparm=0x%lx cstat=%02x dstat=%02x res=%u\n",
 	      intparm, irb->scsw.cmd.cstat, irb->scsw.cmd.dstat,
 	      irb->scsw.cmd.count);

drivers/s390/cio/ccwgroup.c

@@ -66,6 +66,27 @@ __ccwgroup_remove_symlinks(struct ccwgroup_device *gdev)
 	
 }
 
+/*
+ * Remove references from ccw devices to ccw group device and from
+ * ccw group device to ccw devices.
+ */
+static void __ccwgroup_remove_cdev_refs(struct ccwgroup_device *gdev)
+{
+	struct ccw_device *cdev;
+	int i;
+
+	for (i = 0; i < gdev->count; i++) {
+		cdev = gdev->cdev[i];
+		if (!cdev)
+			continue;
+		spin_lock_irq(cdev->ccwlock);
+		dev_set_drvdata(&cdev->dev, NULL);
+		spin_unlock_irq(cdev->ccwlock);
+		gdev->cdev[i] = NULL;
+		put_device(&cdev->dev);
+	}
+}
+
 /*
  * Provide an 'ungroup' attribute so the user can remove group devices no
  * longer needed or accidentially created. Saves memory :)
@@ -78,6 +99,7 @@ static void ccwgroup_ungroup_callback(struct device *dev)
 	if (device_is_registered(&gdev->dev)) {
 		__ccwgroup_remove_symlinks(gdev);
 		device_unregister(dev);
+		__ccwgroup_remove_cdev_refs(gdev);
 	}
 	mutex_unlock(&gdev->reg_mutex);
 }
@@ -116,21 +138,7 @@ static DEVICE_ATTR(ungroup, 0200, NULL, ccwgroup_ungroup_store);
 static void
 ccwgroup_release (struct device *dev)
 {
-	struct ccwgroup_device *gdev;
-	int i;
-
-	gdev = to_ccwgroupdev(dev);
-
-	for (i = 0; i < gdev->count; i++) {
-		if (gdev->cdev[i]) {
-			spin_lock_irq(gdev->cdev[i]->ccwlock);
-			if (dev_get_drvdata(&gdev->cdev[i]->dev) == gdev)
-				dev_set_drvdata(&gdev->cdev[i]->dev, NULL);
-			spin_unlock_irq(gdev->cdev[i]->ccwlock);
-			put_device(&gdev->cdev[i]->dev);
-		}
-	}
-	kfree(gdev);
+	kfree(to_ccwgroupdev(dev));
 }
 
 static int
@@ -639,6 +647,7 @@ void ccwgroup_driver_unregister(struct ccwgroup_driver *cdriver)
 		mutex_lock(&gdev->reg_mutex);
 		__ccwgroup_remove_symlinks(gdev);
 		device_unregister(dev);
+		__ccwgroup_remove_cdev_refs(gdev);
 		mutex_unlock(&gdev->reg_mutex);
 		put_device(dev);
 	}
@@ -660,25 +669,6 @@ int ccwgroup_probe_ccwdev(struct ccw_device *cdev)
 	return 0;
 }
 
-static struct ccwgroup_device *
-__ccwgroup_get_gdev_by_cdev(struct ccw_device *cdev)
-{
-	struct ccwgroup_device *gdev;
-
-	gdev = dev_get_drvdata(&cdev->dev);
-	if (gdev) {
-		if (get_device(&gdev->dev)) {
-			mutex_lock(&gdev->reg_mutex);
-			if (device_is_registered(&gdev->dev))
-				return gdev;
-			mutex_unlock(&gdev->reg_mutex);
-			put_device(&gdev->dev);
-		}
-		return NULL;
-	}
-	return NULL;
-}
-
 /**
  * ccwgroup_remove_ccwdev() - remove function for slave devices
  * @cdev: ccw device to be removed
@@ -694,13 +684,25 @@ void ccwgroup_remove_ccwdev(struct ccw_device *cdev)
 	/* Ignore offlining errors, device is gone anyway. */
 	ccw_device_set_offline(cdev);
 	/* If one of its devices is gone, the whole group is done for. */
-	gdev = __ccwgroup_get_gdev_by_cdev(cdev);
-	if (gdev) {
+	spin_lock_irq(cdev->ccwlock);
+	gdev = dev_get_drvdata(&cdev->dev);
+	if (!gdev) {
+		spin_unlock_irq(cdev->ccwlock);
+		return;
+	}
+	/* Get ccwgroup device reference for local processing. */
+	get_device(&gdev->dev);
+	spin_unlock_irq(cdev->ccwlock);
+	/* Unregister group device. */
+	mutex_lock(&gdev->reg_mutex);
+	if (device_is_registered(&gdev->dev)) {
 		__ccwgroup_remove_symlinks(gdev);
 		device_unregister(&gdev->dev);
-		mutex_unlock(&gdev->reg_mutex);
-		put_device(&gdev->dev);
+		__ccwgroup_remove_cdev_refs(gdev);
 	}
+	mutex_unlock(&gdev->reg_mutex);
+	/* Release ccwgroup device reference for local processing. */
+	put_device(&gdev->dev);
 }
 
 MODULE_LICENSE("GPL");

drivers/s390/cio/chsc.c

@@ -695,6 +695,25 @@ out:
 	return ret;
 }
 
+int chsc_determine_fmt1_channel_path_desc(struct chp_id chpid,
+					  struct channel_path_desc_fmt1 *desc)
+{
+	struct chsc_response_struct *chsc_resp;
+	struct chsc_scpd *scpd_area;
+	int ret;
+
+	spin_lock_irq(&chsc_page_lock);
+	scpd_area = chsc_page;
+	ret = chsc_determine_channel_path_desc(chpid, 0, 0, 1, 0, scpd_area);
+	if (ret)
+		goto out;
+	chsc_resp = (void *)&scpd_area->response;
+	memcpy(desc, &chsc_resp->data, sizeof(*desc));
+out:
+	spin_unlock_irq(&chsc_page_lock);
+	return ret;
+}
+
 static void
 chsc_initialize_cmg_chars(struct channel_path *chp, u8 cmcv,
 			  struct cmg_chars *chars)

drivers/s390/cio/chsc.h

@@ -35,6 +35,22 @@ struct channel_path_desc {
 	u8 chpp;
 } __attribute__ ((packed));
 
+struct channel_path_desc_fmt1 {
+	u8 flags;
+	u8 lsn;
+	u8 desc;
+	u8 chpid;
+	u32:24;
+	u8 chpp;
+	u32 unused[3];
+	u16 mdc;
+	u16:13;
+	u8 r:1;
+	u8 s:1;
+	u8 f:1;
+	u32 zeros[2];
+} __attribute__ ((packed));
+
 struct channel_path;
 
 struct css_chsc_char {
@@ -92,6 +108,8 @@ int chsc_determine_channel_path_desc(struct chp_id chpid, int fmt, int rfmt,
 				     int c, int m, void *page);
 int chsc_determine_base_channel_path_desc(struct chp_id chpid,
 					  struct channel_path_desc *desc);
+int chsc_determine_fmt1_channel_path_desc(struct chp_id chpid,
+					  struct channel_path_desc_fmt1 *desc);
 void chsc_chp_online(struct chp_id chpid);
 void chsc_chp_offline(struct chp_id chpid);
 int chsc_get_channel_measurement_chars(struct channel_path *chp);

drivers/s390/cio/css.c

@@ -618,6 +618,7 @@ EXPORT_SYMBOL_GPL(css_schedule_reprobe);
 static void css_process_crw(struct crw *crw0, struct crw *crw1, int overflow)
 {
 	struct subchannel_id mchk_schid;
+	struct subchannel *sch;
 
 	if (overflow) {
 		css_schedule_eval_all();
@@ -637,6 +638,13 @@ static void css_process_crw(struct crw *crw0, struct crw *crw1, int overflow)
 	if (crw1)
 		mchk_schid.ssid = (crw1->rsid >> 4) & 3;
 
+	if (crw0->erc == CRW_ERC_PMOD) {
+		sch = get_subchannel_by_schid(mchk_schid);
+		if (sch) {
+			css_update_ssd_info(sch);
+			put_device(&sch->dev);
+		}
+	}
 	/*
 	 * Since we are always presented with IPI in the CRW, we have to
 	 * use stsch() to find out if the subchannel in question has come

drivers/s390/cio/device_ops.c

@@ -686,6 +686,46 @@ int ccw_device_tm_start_timeout(struct ccw_device *cdev, struct tcw *tcw,
 }
 EXPORT_SYMBOL(ccw_device_tm_start_timeout);
 
+/**
+ * ccw_device_get_mdc - accumulate max data count
+ * @cdev: ccw device for which the max data count is accumulated
+ * @mask: mask of paths to use
+ *
+ * Return the number of 64K-bytes blocks all paths at least support
+ * for a transport command. Return values <= 0 indicate failures.
+ */
+int ccw_device_get_mdc(struct ccw_device *cdev, u8 mask)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct channel_path_desc_fmt1 desc;
+	struct chp_id chpid;
+	int mdc = 0, ret, i;
+
+	/* Adjust requested path mask to excluded varied off paths. */
+	if (mask)
+		mask &= sch->lpm;
+	else
+		mask = sch->lpm;
+
+	chp_id_init(&chpid);
+	for (i = 0; i < 8; i++) {
+		if (!(mask & (0x80 >> i)))
+			continue;
+		chpid.id = sch->schib.pmcw.chpid[i];
+		ret = chsc_determine_fmt1_channel_path_desc(chpid, &desc);
+		if (ret)
+			return ret;
+		if (!desc.f)
+			return 0;
+		if (!desc.r)
+			mdc = 1;
+		mdc = mdc ? min(mdc, (int)desc.mdc) : desc.mdc;
+	}
+
+	return mdc;
+}
+EXPORT_SYMBOL(ccw_device_get_mdc);
+
 /**
  * ccw_device_tm_intrg - perform interrogate function
  * @cdev: ccw device on which to perform the interrogate function

drivers/s390/cio/itcw.c

@@ -93,6 +93,7 @@ EXPORT_SYMBOL(itcw_get_tcw);
 size_t itcw_calc_size(int intrg, int max_tidaws, int intrg_max_tidaws)
 {
 	size_t len;
+	int cross_count;
 
 	/* Main data. */
 	len = sizeof(struct itcw);
@@ -105,12 +106,27 @@ size_t itcw_calc_size(int intrg, int max_tidaws, int intrg_max_tidaws)
 		       /* TSB */ sizeof(struct tsb) +
 		       /* TIDAL */ intrg_max_tidaws * sizeof(struct tidaw);
 	}
+
 	/* Maximum required alignment padding. */
 	len += /* Initial TCW */ 63 + /* Interrogate TCCB */ 7;
-	/* Maximum padding for structures that may not cross 4k boundary. */
-	if ((max_tidaws > 0) || (intrg_max_tidaws > 0))
-		len += max(max_tidaws, intrg_max_tidaws) *
-		       sizeof(struct tidaw) - 1;
+
+	/* TIDAW lists may not cross a 4k boundary. To cross a
+	 * boundary we need to add a TTIC TIDAW. We need to reserve
+	 * one additional TIDAW for a TTIC that we may need to add due
+	 * to the placement of the data chunk in memory, and a further
+	 * TIDAW for each page boundary that the TIDAW list may cross
+	 * due to it's own size.
+	 */
+	if (max_tidaws) {
+		cross_count = 1 + ((max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+		len += cross_count * sizeof(struct tidaw);
+	}
+	if (intrg_max_tidaws) {
+		cross_count = 1 + ((intrg_max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+		len += cross_count * sizeof(struct tidaw);
+	}
 	return len;
 }
 EXPORT_SYMBOL(itcw_calc_size);
@@ -165,6 +181,7 @@ struct itcw *itcw_init(void *buffer, size_t size, int op, int intrg,
 	void *chunk;
 	addr_t start;
 	addr_t end;
+	int cross_count;
 
 	/* Check for 2G limit. */
 	start = (addr_t) buffer;
@@ -177,8 +194,17 @@ struct itcw *itcw_init(void *buffer, size_t size, int op, int intrg,
 	if (IS_ERR(chunk))
 		return chunk;
 	itcw = chunk;
-	itcw->max_tidaws = max_tidaws;
-	itcw->intrg_max_tidaws = intrg_max_tidaws;
+	/* allow for TTIC tidaws that may be needed to cross a page boundary */
+	cross_count = 0;
+	if (max_tidaws)
+		cross_count = 1 + ((max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+	itcw->max_tidaws = max_tidaws + cross_count;
+	cross_count = 0;
+	if (intrg_max_tidaws)
+		cross_count = 1 + ((intrg_max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+	itcw->intrg_max_tidaws = intrg_max_tidaws + cross_count;
 	/* Main TCW. */
 	chunk = fit_chunk(&start, end, sizeof(struct tcw), 64, 0);
 	if (IS_ERR(chunk))
@@ -198,7 +224,7 @@ struct itcw *itcw_init(void *buffer, size_t size, int op, int intrg,
 	/* Data TIDAL. */
 	if (max_tidaws > 0) {
 		chunk = fit_chunk(&start, end, sizeof(struct tidaw) *
-				  max_tidaws, 16, 1);
+				  itcw->max_tidaws, 16, 0);
 		if (IS_ERR(chunk))
 			return chunk;
 		tcw_set_data(itcw->tcw, chunk, 1);
@@ -206,7 +232,7 @@ struct itcw *itcw_init(void *buffer, size_t size, int op, int intrg,
 	/* Interrogate data TIDAL. */
 	if (intrg && (intrg_max_tidaws > 0)) {
 		chunk = fit_chunk(&start, end, sizeof(struct tidaw) *
-				  intrg_max_tidaws, 16, 1);
+				  itcw->intrg_max_tidaws, 16, 0);
 		if (IS_ERR(chunk))
 			return chunk;
 		tcw_set_data(itcw->intrg_tcw, chunk, 1);
@@ -283,13 +309,29 @@ EXPORT_SYMBOL(itcw_add_dcw);
  * the new tidaw on success or -%ENOSPC if the new tidaw would exceed the
  * available space.
  *
- * Note: the tidaw-list is assumed to be contiguous with no ttics. The
- * last-tidaw flag for the last tidaw in the list will be set by itcw_finalize.
+ * Note: TTIC tidaws are automatically added when needed, so explicitly calling
+ * this interface with the TTIC flag is not supported. The last-tidaw flag
+ * for the last tidaw in the list will be set by itcw_finalize.
  */
 struct tidaw *itcw_add_tidaw(struct itcw *itcw, u8 flags, void *addr, u32 count)
 {
+	struct tidaw *following;
+
 	if (itcw->num_tidaws >= itcw->max_tidaws)
 		return ERR_PTR(-ENOSPC);
+	/*
+	 * Is the tidaw, which follows the one we are about to fill, on the next
+	 * page? Then we have to insert a TTIC tidaw first, that points to the
+	 * tidaw on the new page.
+	 */
+	following = ((struct tidaw *) tcw_get_data(itcw->tcw))
+		+ itcw->num_tidaws + 1;
+	if (itcw->num_tidaws && !((unsigned long) following & ~PAGE_MASK)) {
+		tcw_add_tidaw(itcw->tcw, itcw->num_tidaws++,
+			      TIDAW_FLAGS_TTIC, following, 0);
+		if (itcw->num_tidaws >= itcw->max_tidaws)
+			return ERR_PTR(-ENOSPC);
+	}
 	return tcw_add_tidaw(itcw->tcw, itcw->num_tidaws++, flags, addr, count);
 }
 EXPORT_SYMBOL(itcw_add_tidaw);

drivers/s390/cio/qdio.h

@@ -91,6 +91,12 @@ enum qdio_irq_states {
 #define AC1_SC_QEBSM_AVAILABLE		0x02	/* available for subchannel */
 #define AC1_SC_QEBSM_ENABLED		0x01	/* enabled for subchannel */
 
+/* SIGA flags */
+#define QDIO_SIGA_WRITE		0x00
+#define QDIO_SIGA_READ		0x01
+#define QDIO_SIGA_SYNC		0x02
+#define QDIO_SIGA_QEBSM_FLAG	0x80
+
 #ifdef CONFIG_64BIT
 static inline int do_sqbs(u64 token, unsigned char state, int queue,
 			  int *start, int *count)
@@ -142,10 +148,9 @@ struct siga_flag {
 	u8 input:1;
 	u8 output:1;
 	u8 sync:1;
-	u8 no_sync_ti:1;
-	u8 no_sync_out_ti:1;
-	u8 no_sync_out_pci:1;
-	u8:2;
+	u8 sync_after_ai:1;
+	u8 sync_out_after_pci:1;
+	u8:3;
 } __attribute__ ((packed));
 
 struct chsc_ssqd_area {
@@ -202,6 +207,7 @@ struct qdio_dev_perf_stat {
 	unsigned int inbound_queue_full;
 	unsigned int outbound_call;
 	unsigned int outbound_handler;
+	unsigned int outbound_queue_full;
 	unsigned int fast_requeue;
 	unsigned int target_full;
 	unsigned int eqbs;
@@ -245,10 +251,10 @@ struct qdio_input_q {
 struct qdio_output_q {
 	/* PCIs are enabled for the queue */
 	int pci_out_enabled;
-	/* IQDIO: output multiple buffers (enhanced SIGA) */
-	int use_enh_siga;
 	/* timer to check for more outbound work */
 	struct timer_list timer;
+	/* used SBALs before tasklet schedule */
+	int scan_threshold;
 };
 
 /*
@@ -383,12 +389,13 @@ static inline int multicast_outbound(struct qdio_q *q)
 	(q->irq_ptr->qib.ac & QIB_AC_OUTBOUND_PCI_SUPPORTED)
 #define is_qebsm(q)			(q->irq_ptr->sch_token != 0)
 
-#define need_siga_sync_thinint(q)	(!q->irq_ptr->siga_flag.no_sync_ti)
-#define need_siga_sync_out_thinint(q)	(!q->irq_ptr->siga_flag.no_sync_out_ti)
 #define need_siga_in(q)			(q->irq_ptr->siga_flag.input)
 #define need_siga_out(q)		(q->irq_ptr->siga_flag.output)
-#define need_siga_sync(q)		(q->irq_ptr->siga_flag.sync)
-#define siga_syncs_out_pci(q)		(q->irq_ptr->siga_flag.no_sync_out_pci)
+#define need_siga_sync(q)		(unlikely(q->irq_ptr->siga_flag.sync))
+#define need_siga_sync_after_ai(q)	\
+	(unlikely(q->irq_ptr->siga_flag.sync_after_ai))
+#define need_siga_sync_out_after_pci(q)	\
+	(unlikely(q->irq_ptr->siga_flag.sync_out_after_pci))
 
 #define for_each_input_queue(irq_ptr, q, i)	\
 	for (i = 0, q = irq_ptr->input_qs[0];	\
@@ -423,9 +430,9 @@ struct indicator_t {
 
 extern struct indicator_t *q_indicators;
 
-static inline int shared_ind(struct qdio_irq *irq_ptr)
+static inline int shared_ind(u32 *dsci)
 {
-	return irq_ptr->dsci == &q_indicators[TIQDIO_SHARED_IND].ind;
+	return dsci == &q_indicators[TIQDIO_SHARED_IND].ind;
 }
 
 /* prototypes for thin interrupt */

drivers/s390/cio/qdio_debug.c

@@ -151,6 +151,7 @@ static char *qperf_names[] = {
 	"Inbound queue full",
 	"Outbound calls",
 	"Outbound handler",
+	"Outbound queue full",
 	"Outbound fast_requeue",
 	"Outbound target_full",
 	"QEBSM eqbs",

drivers/s390/cio/qdio_main.c

@@ -14,6 +14,7 @@
 #include <linux/timer.h>
 #include <linux/delay.h>
 #include <linux/gfp.h>
+#include <linux/kernel_stat.h>
 #include <asm/atomic.h>
 #include <asm/debug.h>
 #include <asm/qdio.h>
@@ -29,11 +30,12 @@ MODULE_AUTHOR("Utz Bacher <utz.bacher@de.ibm.com>,"\
 MODULE_DESCRIPTION("QDIO base support");
 MODULE_LICENSE("GPL");
 
-static inline int do_siga_sync(struct subchannel_id schid,
-			       unsigned int out_mask, unsigned int in_mask)
+static inline int do_siga_sync(unsigned long schid,
+			       unsigned int out_mask, unsigned int in_mask,
+			       unsigned int fc)
 {
-	register unsigned long __fc asm ("0") = 2;
-	register struct subchannel_id __schid asm ("1") = schid;
+	register unsigned long __fc asm ("0") = fc;
+	register unsigned long __schid asm ("1") = schid;
 	register unsigned long out asm ("2") = out_mask;
 	register unsigned long in asm ("3") = in_mask;
 	int cc;
@@ -47,10 +49,11 @@ static inline int do_siga_sync(struct subchannel_id schid,
 	return cc;
 }
 
-static inline int do_siga_input(struct subchannel_id schid, unsigned int mask)
+static inline int do_siga_input(unsigned long schid, unsigned int mask,
+				unsigned int fc)
 {
-	register unsigned long __fc asm ("0") = 1;
-	register struct subchannel_id __schid asm ("1") = schid;
+	register unsigned long __fc asm ("0") = fc;
+	register unsigned long __schid asm ("1") = schid;
 	register unsigned long __mask asm ("2") = mask;
 	int cc;
 
@@ -279,16 +282,20 @@ void qdio_init_buf_states(struct qdio_irq *irq_ptr)
 static inline int qdio_siga_sync(struct qdio_q *q, unsigned int output,
 			  unsigned int input)
 {
+	unsigned long schid = *((u32 *) &q->irq_ptr->schid);
+	unsigned int fc = QDIO_SIGA_SYNC;
 	int cc;
 
-	if (!need_siga_sync(q))
-		return 0;
-
 	DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "siga-s:%1d", q->nr);
 	qperf_inc(q, siga_sync);
 
-	cc = do_siga_sync(q->irq_ptr->schid, output, input);
-	if (cc)
+	if (is_qebsm(q)) {
+		schid = q->irq_ptr->sch_token;
+		fc |= QDIO_SIGA_QEBSM_FLAG;
+	}
+
+	cc = do_siga_sync(schid, output, input, fc);
+	if (unlikely(cc))
 		DBF_ERROR("%4x SIGA-S:%2d", SCH_NO(q), cc);
 	return cc;
 }
@@ -301,38 +308,22 @@ static inline int qdio_siga_sync_q(struct qdio_q *q)
 		return qdio_siga_sync(q, q->mask, 0);
 }
 
-static inline int qdio_siga_sync_out(struct qdio_q *q)
-{
-	return qdio_siga_sync(q, ~0U, 0);
-}
-
-static inline int qdio_siga_sync_all(struct qdio_q *q)
-{
-	return qdio_siga_sync(q, ~0U, ~0U);
-}
-
 static int qdio_siga_output(struct qdio_q *q, unsigned int *busy_bit)
 {
-	unsigned long schid;
-	unsigned int fc = 0;
+	unsigned long schid = *((u32 *) &q->irq_ptr->schid);
+	unsigned int fc = QDIO_SIGA_WRITE;
 	u64 start_time = 0;
 	int cc;
 
-	if (q->u.out.use_enh_siga)
-		fc = 3;
-
 	if (is_qebsm(q)) {
 		schid = q->irq_ptr->sch_token;
-		fc |= 0x80;
+		fc |= QDIO_SIGA_QEBSM_FLAG;
 	}
-	else
-		schid = *((u32 *)&q->irq_ptr->schid);
-
 again:
 	cc = do_siga_output(schid, q->mask, busy_bit, fc);
 
 	/* hipersocket busy condition */
-	if (*busy_bit) {
+	if (unlikely(*busy_bit)) {
 		WARN_ON(queue_type(q) != QDIO_IQDIO_QFMT || cc != 2);
 
 		if (!start_time) {
@@ -347,32 +338,41 @@ again:
 
 static inline int qdio_siga_input(struct qdio_q *q)
 {
+	unsigned long schid = *((u32 *) &q->irq_ptr->schid);
+	unsigned int fc = QDIO_SIGA_READ;
 	int cc;
 
 	DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "siga-r:%1d", q->nr);
 	qperf_inc(q, siga_read);
 
-	cc = do_siga_input(q->irq_ptr->schid, q->mask);
-	if (cc)
+	if (is_qebsm(q)) {
+		schid = q->irq_ptr->sch_token;
+		fc |= QDIO_SIGA_QEBSM_FLAG;
+	}
+
+	cc = do_siga_input(schid, q->mask, fc);
+	if (unlikely(cc))
 		DBF_ERROR("%4x SIGA-R:%2d", SCH_NO(q), cc);
 	return cc;
 }
 
-static inline void qdio_sync_after_thinint(struct qdio_q *q)
+#define qdio_siga_sync_out(q) qdio_siga_sync(q, ~0U, 0)
+#define qdio_siga_sync_all(q) qdio_siga_sync(q, ~0U, ~0U)
+
+static inline void qdio_sync_queues(struct qdio_q *q)
 {
-	if (pci_out_supported(q)) {
-		if (need_siga_sync_thinint(q))
-			qdio_siga_sync_all(q);
-		else if (need_siga_sync_out_thinint(q))
-			qdio_siga_sync_out(q);
-	} else
+	/* PCI capable outbound queues will also be scanned so sync them too */
+	if (pci_out_supported(q))
+		qdio_siga_sync_all(q);
+	else
 		qdio_siga_sync_q(q);
 }
 
 int debug_get_buf_state(struct qdio_q *q, unsigned int bufnr,
 			unsigned char *state)
 {
-	qdio_siga_sync_q(q);
+	if (need_siga_sync(q))
+		qdio_siga_sync_q(q);
 	return get_buf_states(q, bufnr, state, 1, 0);
 }
 
@@ -549,7 +549,8 @@ static inline int qdio_inbound_q_done(struct qdio_q *q)
 	if (!atomic_read(&q->nr_buf_used))
 		return 1;
 
-	qdio_siga_sync_q(q);
+	if (need_siga_sync(q))
+		qdio_siga_sync_q(q);
 	get_buf_state(q, q->first_to_check, &state, 0);
 
 	if (state == SLSB_P_INPUT_PRIMED || state == SLSB_P_INPUT_ERROR)
@@ -644,9 +645,12 @@ static int get_outbound_buffer_frontier(struct qdio_q *q)
 	int count, stop;
 	unsigned char state;
 
-	if (((queue_type(q) != QDIO_IQDIO_QFMT) && !pci_out_supported(q)) ||
-	    (queue_type(q) == QDIO_IQDIO_QFMT && multicast_outbound(q)))
-		qdio_siga_sync_q(q);
+	if (need_siga_sync(q))
+		if (((queue_type(q) != QDIO_IQDIO_QFMT) &&
+		    !pci_out_supported(q)) ||
+		    (queue_type(q) == QDIO_IQDIO_QFMT &&
+		    multicast_outbound(q)))
+			qdio_siga_sync_q(q);
 
 	/*
 	 * Don't check 128 buffers, as otherwise qdio_inbound_q_moved
@@ -818,7 +822,8 @@ static inline void qdio_check_outbound_after_thinint(struct qdio_q *q)
 static void __tiqdio_inbound_processing(struct qdio_q *q)
 {
 	qperf_inc(q, tasklet_inbound);
-	qdio_sync_after_thinint(q);
+	if (need_siga_sync(q) && need_siga_sync_after_ai(q))
+		qdio_sync_queues(q);
 
 	/*
 	 * The interrupt could be caused by a PCI request. Check the
@@ -898,16 +903,14 @@ static void qdio_int_handler_pci(struct qdio_irq *irq_ptr)
 			tasklet_schedule(&q->tasklet);
 	}
 
-	if (!(irq_ptr->qib.ac & QIB_AC_OUTBOUND_PCI_SUPPORTED))
+	if (!pci_out_supported(q))
 		return;
 
 	for_each_output_queue(irq_ptr, q, i) {
 		if (qdio_outbound_q_done(q))
 			continue;
-
-		if (!siga_syncs_out_pci(q))
+		if (need_siga_sync(q) && need_siga_sync_out_after_pci(q))
 			qdio_siga_sync_q(q);
-
 		tasklet_schedule(&q->tasklet);
 	}
 }
@@ -970,6 +973,7 @@ void qdio_int_handler(struct ccw_device *cdev, unsigned long intparm,
 		return;
 	}
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_QDI]++;
 	if (irq_ptr->perf_stat_enabled)
 		irq_ptr->perf_stat.qdio_int++;
 
@@ -1273,7 +1277,6 @@ int qdio_establish(struct qdio_initialize *init_data)
 	}
 
 	qdio_setup_ssqd_info(irq_ptr);
-	DBF_EVENT("qDmmwc:%2x", irq_ptr->ssqd_desc.mmwc);
 	DBF_EVENT("qib ac:%4x", irq_ptr->qib.ac);
 
 	/* qebsm is now setup if available, initialize buffer states */
@@ -1445,52 +1448,38 @@ static int handle_outbound(struct qdio_q *q, unsigned int callflags,
 	used = atomic_add_return(count, &q->nr_buf_used);
 	BUG_ON(used > QDIO_MAX_BUFFERS_PER_Q);
 
+	if (used == QDIO_MAX_BUFFERS_PER_Q)
+		qperf_inc(q, outbound_queue_full);
+
 	if (callflags & QDIO_FLAG_PCI_OUT) {
 		q->u.out.pci_out_enabled = 1;
 		qperf_inc(q, pci_request_int);
-	}
-	else
+	} else
 		q->u.out.pci_out_enabled = 0;
 
 	if (queue_type(q) == QDIO_IQDIO_QFMT) {
-		if (multicast_outbound(q))
+		/* One SIGA-W per buffer required for unicast HiperSockets. */
+		WARN_ON_ONCE(count > 1 && !multicast_outbound(q));
+
+		rc = qdio_kick_outbound_q(q);
+	} else if (need_siga_sync(q)) {
+		rc = qdio_siga_sync_q(q);
+	} else {
+		/* try to fast requeue buffers */
+		get_buf_state(q, prev_buf(bufnr), &state, 0);
+		if (state != SLSB_CU_OUTPUT_PRIMED)
 			rc = qdio_kick_outbound_q(q);
 		else
-			if ((q->irq_ptr->ssqd_desc.mmwc > 1) &&
-			    (count > 1) &&
-			    (count <= q->irq_ptr->ssqd_desc.mmwc)) {
-				/* exploit enhanced SIGA */
-				q->u.out.use_enh_siga = 1;
-				rc = qdio_kick_outbound_q(q);
-			} else {
-				/*
-				* One siga-w per buffer required for unicast
-				* HiperSockets.
-				*/
-				q->u.out.use_enh_siga = 0;
-				while (count--) {
-					rc = qdio_kick_outbound_q(q);
-					if (rc)
-						goto out;
-				}
-			}
-		goto out;
-	}
-
-	if (need_siga_sync(q)) {
-		qdio_siga_sync_q(q);
-		goto out;
+			qperf_inc(q, fast_requeue);
 	}
 
-	/* try to fast requeue buffers */
-	get_buf_state(q, prev_buf(bufnr), &state, 0);
-	if (state != SLSB_CU_OUTPUT_PRIMED)
-		rc = qdio_kick_outbound_q(q);
+	/* in case of SIGA errors we must process the error immediately */
+	if (used >= q->u.out.scan_threshold || rc)
+		tasklet_schedule(&q->tasklet);
 	else
-		qperf_inc(q, fast_requeue);
-
-out:
-	tasklet_schedule(&q->tasklet);
+		/* free the SBALs in case of no further traffic */
+		if (!timer_pending(&q->u.out.timer))
+			mod_timer(&q->u.out.timer, jiffies + HZ);
 	return rc;
 }
 
@@ -1550,7 +1539,7 @@ int qdio_start_irq(struct ccw_device *cdev, int nr)
 
 	WARN_ON(queue_irqs_enabled(q));
 
-	if (!shared_ind(q->irq_ptr))
+	if (!shared_ind(q->irq_ptr->dsci))
 		xchg(q->irq_ptr->dsci, 0);
 
 	qdio_stop_polling(q);
@@ -1560,7 +1549,7 @@ int qdio_start_irq(struct ccw_device *cdev, int nr)
 	 * We need to check again to not lose initiative after
 	 * resetting the ACK state.
 	 */
-	if (!shared_ind(q->irq_ptr) && *q->irq_ptr->dsci)
+	if (!shared_ind(q->irq_ptr->dsci) && *q->irq_ptr->dsci)
 		goto rescan;
 	if (!qdio_inbound_q_done(q))
 		goto rescan;
@@ -1600,12 +1589,14 @@ int qdio_get_next_buffers(struct ccw_device *cdev, int nr, int *bufnr,
 	q = irq_ptr->input_qs[nr];
 	WARN_ON(queue_irqs_enabled(q));
 
-	qdio_sync_after_thinint(q);
-
 	/*
-	 * The interrupt could be caused by a PCI request. Check the
-	 * PCI capable outbound queues.
+	 * Cannot rely on automatic sync after interrupt since queues may
+	 * also be examined without interrupt.
 	 */
+	if (need_siga_sync(q))
+		qdio_sync_queues(q);
+
+	/* check the PCI capable outbound queues. */
 	qdio_check_outbound_after_thinint(q);
 
 	if (!qdio_inbound_q_moved(q))

drivers/s390/cio/qdio_setup.c

@@ -178,6 +178,7 @@ static void setup_queues(struct qdio_irq *irq_ptr,
 		setup_queues_misc(q, irq_ptr, qdio_init->output_handler, i);
 
 		q->is_input_q = 0;
+		q->u.out.scan_threshold = qdio_init->scan_threshold;
 		setup_storage_lists(q, irq_ptr, output_sbal_array, i);
 		output_sbal_array += QDIO_MAX_BUFFERS_PER_Q;
 
@@ -196,14 +197,10 @@ static void process_ac_flags(struct qdio_irq *irq_ptr, unsigned char qdioac)
 		irq_ptr->siga_flag.output = 1;
 	if (qdioac & AC1_SIGA_SYNC_NEEDED)
 		irq_ptr->siga_flag.sync = 1;
-	if (qdioac & AC1_AUTOMATIC_SYNC_ON_THININT)
-		irq_ptr->siga_flag.no_sync_ti = 1;
-	if (qdioac & AC1_AUTOMATIC_SYNC_ON_OUT_PCI)
-		irq_ptr->siga_flag.no_sync_out_pci = 1;
-
-	if (irq_ptr->siga_flag.no_sync_out_pci &&
-	    irq_ptr->siga_flag.no_sync_ti)
-		irq_ptr->siga_flag.no_sync_out_ti = 1;
+	if (!(qdioac & AC1_AUTOMATIC_SYNC_ON_THININT))
+		irq_ptr->siga_flag.sync_after_ai = 1;
+	if (!(qdioac & AC1_AUTOMATIC_SYNC_ON_OUT_PCI))
+		irq_ptr->siga_flag.sync_out_after_pci = 1;
 }
 
 static void check_and_setup_qebsm(struct qdio_irq *irq_ptr,
@@ -451,7 +448,7 @@ void qdio_print_subchannel_info(struct qdio_irq *irq_ptr,
 	char s[80];
 
 	snprintf(s, 80, "qdio: %s %s on SC %x using "
-		 "AI:%d QEBSM:%d PCI:%d TDD:%d SIGA:%s%s%s%s%s%s\n",
+		 "AI:%d QEBSM:%d PCI:%d TDD:%d SIGA:%s%s%s%s%s\n",
 		 dev_name(&cdev->dev),
 		 (irq_ptr->qib.qfmt == QDIO_QETH_QFMT) ? "OSA" :
 			((irq_ptr->qib.qfmt == QDIO_ZFCP_QFMT) ? "ZFCP" : "HS"),
@@ -463,9 +460,8 @@ void qdio_print_subchannel_info(struct qdio_irq *irq_ptr,
 		 (irq_ptr->siga_flag.input) ? "R" : " ",
 		 (irq_ptr->siga_flag.output) ? "W" : " ",
 		 (irq_ptr->siga_flag.sync) ? "S" : " ",
-		 (!irq_ptr->siga_flag.no_sync_ti) ? "A" : " ",
-		 (!irq_ptr->siga_flag.no_sync_out_ti) ? "O" : " ",
-		 (!irq_ptr->siga_flag.no_sync_out_pci) ? "P" : " ");
+		 (irq_ptr->siga_flag.sync_after_ai) ? "A" : " ",
+		 (irq_ptr->siga_flag.sync_out_after_pci) ? "P" : " ");
 	printk(KERN_INFO "%s", s);
 }
 

drivers/s390/cio/qdio_thinint.c

@@ -8,6 +8,7 @@
  */
 #include <linux/io.h>
 #include <linux/slab.h>
+#include <linux/kernel_stat.h>
 #include <asm/atomic.h>
 #include <asm/debug.h>
 #include <asm/qdio.h>
@@ -35,22 +36,8 @@ static u8 *tiqdio_alsi;
 
 struct indicator_t *q_indicators;
 
-static int css_qdio_omit_svs;
-
 static u64 last_ai_time;
 
-static inline unsigned long do_clear_global_summary(void)
-{
-	register unsigned long __fn asm("1") = 3;
-	register unsigned long __tmp asm("2");
-	register unsigned long __time asm("3");
-
-	asm volatile(
-		"	.insn	rre,0xb2650000,2,0"
-		: "+d" (__fn), "=d" (__tmp), "=d" (__time));
-	return __time;
-}
-
 /* returns addr for the device state change indicator */
 static u32 *get_indicator(void)
 {
@@ -83,10 +70,6 @@ void tiqdio_add_input_queues(struct qdio_irq *irq_ptr)
 	struct qdio_q *q;
 	int i;
 
-	/* No TDD facility? If we must use SIGA-s we can also omit SVS. */
-	if (!css_qdio_omit_svs && irq_ptr->siga_flag.sync)
-		css_qdio_omit_svs = 1;
-
 	mutex_lock(&tiq_list_lock);
 	for_each_input_queue(irq_ptr, q, i)
 		list_add_rcu(&q->entry, &tiq_list);
@@ -112,9 +95,9 @@ void tiqdio_remove_input_queues(struct qdio_irq *irq_ptr)
 	}
 }
 
-static inline int shared_ind_used(void)
+static inline u32 shared_ind_set(void)
 {
-	return atomic_read(&q_indicators[TIQDIO_SHARED_IND].count);
+	return q_indicators[TIQDIO_SHARED_IND].ind;
 }
 
 /**
@@ -124,20 +107,11 @@ static inline int shared_ind_used(void)
  */
 static void tiqdio_thinint_handler(void *alsi, void *data)
 {
+	u32 si_used = shared_ind_set();
 	struct qdio_q *q;
 
 	last_ai_time = S390_lowcore.int_clock;
-
-	/*
-	 * SVS only when needed: issue SVS to benefit from iqdio interrupt
-	 * avoidance (SVS clears adapter interrupt suppression overwrite).
-	 */
-	if (!css_qdio_omit_svs)
-		do_clear_global_summary();
-
-	/* reset local summary indicator */
-	if (shared_ind_used())
-		xchg(tiqdio_alsi, 0);
+	kstat_cpu(smp_processor_id()).irqs[IOINT_QAI]++;
 
 	/* protect tiq_list entries, only changed in activate or shutdown */
 	rcu_read_lock();
@@ -146,7 +120,10 @@ static void tiqdio_thinint_handler(void *alsi, void *data)
 	list_for_each_entry_rcu(q, &tiq_list, entry) {
 
 		/* only process queues from changed sets */
-		if (!*q->irq_ptr->dsci)
+		if (unlikely(shared_ind(q->irq_ptr->dsci))) {
+			if (!si_used)
+				continue;
+		} else if (!*q->irq_ptr->dsci)
 			continue;
 
 		if (q->u.in.queue_start_poll) {
@@ -162,7 +139,7 @@ static void tiqdio_thinint_handler(void *alsi, void *data)
 						 q->irq_ptr->int_parm);
 		} else {
 			/* only clear it if the indicator is non-shared */
-			if (!shared_ind(q->irq_ptr))
+			if (!shared_ind(q->irq_ptr->dsci))
 				xchg(q->irq_ptr->dsci, 0);
 			/*
 			 * Call inbound processing but not directly
@@ -178,13 +155,8 @@ static void tiqdio_thinint_handler(void *alsi, void *data)
 	 * If the shared indicator was used clear it now after all queues
 	 * were processed.
 	 */
-	if (shared_ind_used()) {
+	if (si_used && shared_ind_set())
 		xchg(&q_indicators[TIQDIO_SHARED_IND].ind, 0);
-
-		/* prevent racing */
-		if (*tiqdio_alsi)
-			xchg(&q_indicators[TIQDIO_SHARED_IND].ind, 1 << 7);
-	}
 }
 
 static int set_subchannel_ind(struct qdio_irq *irq_ptr, int reset)
@@ -269,12 +241,6 @@ int qdio_establish_thinint(struct qdio_irq *irq_ptr)
 {
 	if (!is_thinint_irq(irq_ptr))
 		return 0;
-
-	/* Check for aif time delay disablement. If installed,
-	 * omit SVS even under LPAR
-	 */
-	if (css_general_characteristics.aif_tdd)
-		css_qdio_omit_svs = 1;
 	return set_subchannel_ind(irq_ptr, 0);
 }
 

drivers/s390/crypto/ap_bus.c

@@ -27,6 +27,7 @@
 #define KMSG_COMPONENT "ap"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/delay.h>
@@ -154,7 +155,7 @@ static inline int ap_instructions_available(void)
  */
 static int ap_interrupts_available(void)
 {
-	return test_facility(1) && test_facility(2);
+	return test_facility(2) && test_facility(65);
 }
 
 /**
@@ -221,6 +222,69 @@ ap_queue_interruption_control(ap_qid_t qid, void *ind)
 }
 #endif
 
+static inline struct ap_queue_status __ap_4096_commands_available(ap_qid_t qid,
+								  int *support)
+{
+	register unsigned long reg0 asm ("0") = 0UL | qid | (1UL << 23);
+	register struct ap_queue_status reg1 asm ("1");
+	register unsigned long reg2 asm ("2") = 0UL;
+
+	asm volatile(
+		".long 0xb2af0000\n"
+		"0: la    %1,0\n"
+		"1:\n"
+		EX_TABLE(0b, 1b)
+		: "+d" (reg0), "=d" (reg1), "=d" (reg2)
+		:
+		: "cc");
+
+	if (reg2 & 0x6000000000000000ULL)
+		*support = 1;
+	else
+		*support = 0;
+
+	return reg1;
+}
+
+/**
+ * ap_4096_commands_availablen(): Check for availability of 4096 bit RSA
+ * support.
+ * @qid: The AP queue number
+ *
+ * Returns 1 if 4096 bit RSA keys are support fo the AP, returns 0 if not.
+ */
+int ap_4096_commands_available(ap_qid_t qid)
+{
+	struct ap_queue_status status;
+	int i, support = 0;
+	status = __ap_4096_commands_available(qid, &support);
+
+	for (i = 0; i < AP_MAX_RESET; i++) {
+		switch (status.response_code) {
+		case AP_RESPONSE_NORMAL:
+			return support;
+		case AP_RESPONSE_RESET_IN_PROGRESS:
+		case AP_RESPONSE_BUSY:
+			break;
+		case AP_RESPONSE_Q_NOT_AVAIL:
+		case AP_RESPONSE_DECONFIGURED:
+		case AP_RESPONSE_CHECKSTOPPED:
+		case AP_RESPONSE_INVALID_ADDRESS:
+			return 0;
+		case AP_RESPONSE_OTHERWISE_CHANGED:
+			break;
+		default:
+			break;
+		}
+		if (i < AP_MAX_RESET - 1) {
+			udelay(5);
+			status = __ap_4096_commands_available(qid, &support);
+		}
+	}
+	return support;
+}
+EXPORT_SYMBOL(ap_4096_commands_available);
+
 /**
  * ap_queue_enable_interruption(): Enable interruption on an AP.
  * @qid: The AP queue number
@@ -1042,6 +1106,7 @@ out:
 
 static void ap_interrupt_handler(void *unused1, void *unused2)
 {
+	kstat_cpu(smp_processor_id()).irqs[IOINT_APB]++;
 	tasklet_schedule(&ap_tasklet);
 }
 

drivers/s390/crypto/ap_bus.h

@@ -196,4 +196,6 @@ void ap_flush_queue(struct ap_device *ap_dev);
 int ap_module_init(void);
 void ap_module_exit(void);
 
+int ap_4096_commands_available(ap_qid_t qid);
+
 #endif /* _AP_BUS_H_ */

drivers/s390/crypto/zcrypt_api.c

@@ -396,8 +396,15 @@ static long zcrypt_rsa_crt(struct ica_rsa_modexpo_crt *crt)
 			if (copied == 0) {
 				unsigned int len;
 				spin_unlock_bh(&zcrypt_device_lock);
-				/* len is max 256 / 2 - 120 = 8 */
-				len = crt->inputdatalength / 2 - 120;
+				/* len is max 256 / 2 - 120 = 8
+				 * For bigger device just assume len of leading
+				 * 0s is 8 as stated in the requirements for
+				 * ica_rsa_modexpo_crt struct in zcrypt.h.
+				 */
+				if (crt->inputdatalength <= 256)
+					len = crt->inputdatalength / 2 - 120;
+				else
+					len = 8;
 				if (len > sizeof(z1))
 					return -EFAULT;
 				z1 = z2 = z3 = 0;
@@ -405,6 +412,7 @@ static long zcrypt_rsa_crt(struct ica_rsa_modexpo_crt *crt)
 				    copy_from_user(&z2, crt->bp_key, len) ||
 				    copy_from_user(&z3, crt->u_mult_inv, len))
 					return -EFAULT;
+				z1 = z2 = z3 = 0;
 				copied = 1;
 				/*
 				 * We have to restart device lookup -

drivers/s390/crypto/zcrypt_api.h

@@ -109,6 +109,7 @@ struct zcrypt_device {
 	int request_count;		/* # current requests. */
 
 	struct ap_message reply;	/* Per-device reply structure. */
+	int max_exp_bit_length;
 };
 
 struct zcrypt_device *zcrypt_device_alloc(size_t);

drivers/s390/crypto/zcrypt_cex2a.c

@@ -41,7 +41,7 @@
 #define CEX2A_MIN_MOD_SIZE	  1	/*    8 bits	*/
 #define CEX2A_MAX_MOD_SIZE	256	/* 2048 bits	*/
 #define CEX3A_MIN_MOD_SIZE	CEX2A_MIN_MOD_SIZE
-#define CEX3A_MAX_MOD_SIZE	CEX2A_MAX_MOD_SIZE
+#define CEX3A_MAX_MOD_SIZE	512	/* 4096 bits	*/
 
 #define CEX2A_SPEED_RATING	970
 #define CEX3A_SPEED_RATING	900 /* Fixme: Needs finetuning */
@@ -49,8 +49,10 @@
 #define CEX2A_MAX_MESSAGE_SIZE	0x390	/* sizeof(struct type50_crb2_msg)    */
 #define CEX2A_MAX_RESPONSE_SIZE 0x110	/* max outputdatalength + type80_hdr */
 
-#define CEX3A_MAX_MESSAGE_SIZE	CEX2A_MAX_MESSAGE_SIZE
-#define CEX3A_MAX_RESPONSE_SIZE	CEX2A_MAX_RESPONSE_SIZE
+#define CEX3A_MAX_RESPONSE_SIZE	0x210	/* 512 bit modulus
+					 * (max outputdatalength) +
+					 * type80_hdr*/
+#define CEX3A_MAX_MESSAGE_SIZE	sizeof(struct type50_crb3_msg)
 
 #define CEX2A_CLEANUP_TIME	(15*HZ)
 #define CEX3A_CLEANUP_TIME	CEX2A_CLEANUP_TIME
@@ -110,7 +112,7 @@ static int ICAMEX_msg_to_type50MEX_msg(struct zcrypt_device *zdev,
 		mod = meb1->modulus + sizeof(meb1->modulus) - mod_len;
 		exp = meb1->exponent + sizeof(meb1->exponent) - mod_len;
 		inp = meb1->message + sizeof(meb1->message) - mod_len;
-	} else {
+	} else if (mod_len <= 256) {
 		struct type50_meb2_msg *meb2 = ap_msg->message;
 		memset(meb2, 0, sizeof(*meb2));
 		ap_msg->length = sizeof(*meb2);
@@ -120,6 +122,17 @@ static int ICAMEX_msg_to_type50MEX_msg(struct zcrypt_device *zdev,
 		mod = meb2->modulus + sizeof(meb2->modulus) - mod_len;
 		exp = meb2->exponent + sizeof(meb2->exponent) - mod_len;
 		inp = meb2->message + sizeof(meb2->message) - mod_len;
+	} else {
+		/* mod_len > 256 = 4096 bit RSA Key */
+		struct type50_meb3_msg *meb3 = ap_msg->message;
+		memset(meb3, 0, sizeof(*meb3));
+		ap_msg->length = sizeof(*meb3);
+		meb3->header.msg_type_code = TYPE50_TYPE_CODE;
+		meb3->header.msg_len = sizeof(*meb3);
+		meb3->keyblock_type = TYPE50_MEB3_FMT;
+		mod = meb3->modulus + sizeof(meb3->modulus) - mod_len;
+		exp = meb3->exponent + sizeof(meb3->exponent) - mod_len;
+		inp = meb3->message + sizeof(meb3->message) - mod_len;
 	}
 
 	if (copy_from_user(mod, mex->n_modulus, mod_len) ||
@@ -142,7 +155,7 @@ static int ICACRT_msg_to_type50CRT_msg(struct zcrypt_device *zdev,
 				       struct ap_message *ap_msg,
 				       struct ica_rsa_modexpo_crt *crt)
 {
-	int mod_len, short_len, long_len, long_offset;
+	int mod_len, short_len, long_len, long_offset, limit;
 	unsigned char *p, *q, *dp, *dq, *u, *inp;
 
 	mod_len = crt->inputdatalength;
@@ -152,14 +165,20 @@ static int ICACRT_msg_to_type50CRT_msg(struct zcrypt_device *zdev,
 	/*
 	 * CEX2A cannot handle p, dp, or U > 128 bytes.
 	 * If we have one of these, we need to do extra checking.
+	 * For CEX3A the limit is 256 bytes.
 	 */
-	if (long_len > 128) {
+	if (zdev->max_mod_size == CEX3A_MAX_MOD_SIZE)
+		limit = 256;
+	else
+		limit = 128;
+
+	if (long_len > limit) {
 		/*
 		 * zcrypt_rsa_crt already checked for the leading
 		 * zeroes of np_prime, bp_key and u_mult_inc.
 		 */
-		long_offset = long_len - 128;
-		long_len = 128;
+		long_offset = long_len - limit;
+		long_len = limit;
 	} else
 		long_offset = 0;
 
@@ -180,7 +199,7 @@ static int ICACRT_msg_to_type50CRT_msg(struct zcrypt_device *zdev,
 		dq = crb1->dq + sizeof(crb1->dq) - short_len;
 		u = crb1->u + sizeof(crb1->u) - long_len;
 		inp = crb1->message + sizeof(crb1->message) - mod_len;
-	} else {
+	} else if (long_len <= 128) {
 		struct type50_crb2_msg *crb2 = ap_msg->message;
 		memset(crb2, 0, sizeof(*crb2));
 		ap_msg->length = sizeof(*crb2);
@@ -193,6 +212,20 @@ static int ICACRT_msg_to_type50CRT_msg(struct zcrypt_device *zdev,
 		dq = crb2->dq + sizeof(crb2->dq) - short_len;
 		u = crb2->u + sizeof(crb2->u) - long_len;
 		inp = crb2->message + sizeof(crb2->message) - mod_len;
+	} else {
+		/* long_len >= 256 */
+		struct type50_crb3_msg *crb3 = ap_msg->message;
+		memset(crb3, 0, sizeof(*crb3));
+		ap_msg->length = sizeof(*crb3);
+		crb3->header.msg_type_code = TYPE50_TYPE_CODE;
+		crb3->header.msg_len = sizeof(*crb3);
+		crb3->keyblock_type = TYPE50_CRB3_FMT;
+		p = crb3->p + sizeof(crb3->p) - long_len;
+		q = crb3->q + sizeof(crb3->q) - short_len;
+		dp = crb3->dp + sizeof(crb3->dp) - long_len;
+		dq = crb3->dq + sizeof(crb3->dq) - short_len;
+		u = crb3->u + sizeof(crb3->u) - long_len;
+		inp = crb3->message + sizeof(crb3->message) - mod_len;
 	}
 
 	if (copy_from_user(p, crt->np_prime + long_offset, long_len) ||
@@ -203,7 +236,6 @@ static int ICACRT_msg_to_type50CRT_msg(struct zcrypt_device *zdev,
 	    copy_from_user(inp, crt->inputdata, mod_len))
 		return -EFAULT;
 
-
 	return 0;
 }
 
@@ -230,7 +262,10 @@ static int convert_type80(struct zcrypt_device *zdev,
 		zdev->online = 0;
 		return -EAGAIN;	/* repeat the request on a different device. */
 	}
-	BUG_ON(t80h->len > CEX2A_MAX_RESPONSE_SIZE);
+	if (zdev->user_space_type == ZCRYPT_CEX2A)
+		BUG_ON(t80h->len > CEX2A_MAX_RESPONSE_SIZE);
+	else
+		BUG_ON(t80h->len > CEX3A_MAX_RESPONSE_SIZE);
 	data = reply->message + t80h->len - outputdatalength;
 	if (copy_to_user(outputdata, data, outputdatalength))
 		return -EFAULT;
@@ -282,7 +317,10 @@ static void zcrypt_cex2a_receive(struct ap_device *ap_dev,
 	}
 	t80h = reply->message;
 	if (t80h->type == TYPE80_RSP_CODE) {
-		length = min(CEX2A_MAX_RESPONSE_SIZE, (int) t80h->len);
+		if (ap_dev->device_type == AP_DEVICE_TYPE_CEX2A)
+			length = min(CEX2A_MAX_RESPONSE_SIZE, (int) t80h->len);
+		else
+			length = min(CEX3A_MAX_RESPONSE_SIZE, (int) t80h->len);
 		memcpy(msg->message, reply->message, length);
 	} else
 		memcpy(msg->message, reply->message, sizeof error_reply);
@@ -307,7 +345,10 @@ static long zcrypt_cex2a_modexpo(struct zcrypt_device *zdev,
 	int rc;
 
 	ap_init_message(&ap_msg);
-	ap_msg.message = kmalloc(CEX2A_MAX_MESSAGE_SIZE, GFP_KERNEL);
+	if (zdev->user_space_type == ZCRYPT_CEX2A)
+		ap_msg.message = kmalloc(CEX2A_MAX_MESSAGE_SIZE, GFP_KERNEL);
+	else
+		ap_msg.message = kmalloc(CEX3A_MAX_MESSAGE_SIZE, GFP_KERNEL);
 	if (!ap_msg.message)
 		return -ENOMEM;
 	ap_msg.psmid = (((unsigned long long) current->pid) << 32) +
@@ -345,7 +386,10 @@ static long zcrypt_cex2a_modexpo_crt(struct zcrypt_device *zdev,
 	int rc;
 
 	ap_init_message(&ap_msg);
-	ap_msg.message = kmalloc(CEX2A_MAX_MESSAGE_SIZE, GFP_KERNEL);
+	if (zdev->user_space_type == ZCRYPT_CEX2A)
+		ap_msg.message = kmalloc(CEX2A_MAX_MESSAGE_SIZE, GFP_KERNEL);
+	else
+		ap_msg.message = kmalloc(CEX3A_MAX_MESSAGE_SIZE, GFP_KERNEL);
 	if (!ap_msg.message)
 		return -ENOMEM;
 	ap_msg.psmid = (((unsigned long long) current->pid) << 32) +
@@ -397,6 +441,7 @@ static int zcrypt_cex2a_probe(struct ap_device *ap_dev)
 		zdev->max_mod_size = CEX2A_MAX_MOD_SIZE;
 		zdev->short_crt = 1;
 		zdev->speed_rating = CEX2A_SPEED_RATING;
+		zdev->max_exp_bit_length = CEX2A_MAX_MOD_SIZE;
 		break;
 	case AP_DEVICE_TYPE_CEX3A:
 		zdev = zcrypt_device_alloc(CEX3A_MAX_RESPONSE_SIZE);
@@ -404,8 +449,13 @@ static int zcrypt_cex2a_probe(struct ap_device *ap_dev)
 			return -ENOMEM;
 		zdev->user_space_type = ZCRYPT_CEX3A;
 		zdev->type_string = "CEX3A";
-		zdev->min_mod_size = CEX3A_MIN_MOD_SIZE;
-		zdev->max_mod_size = CEX3A_MAX_MOD_SIZE;
+		zdev->min_mod_size = CEX2A_MIN_MOD_SIZE;
+		zdev->max_mod_size = CEX2A_MAX_MOD_SIZE;
+		zdev->max_exp_bit_length = CEX2A_MAX_MOD_SIZE;
+		if (ap_4096_commands_available(ap_dev->qid)) {
+			zdev->max_mod_size = CEX3A_MAX_MOD_SIZE;
+			zdev->max_exp_bit_length = CEX3A_MAX_MOD_SIZE;
+		}
 		zdev->short_crt = 1;
 		zdev->speed_rating = CEX3A_SPEED_RATING;
 		break;

drivers/s390/crypto/zcrypt_cex2a.h

@@ -51,8 +51,10 @@ struct type50_hdr {
 
 #define TYPE50_MEB1_FMT		0x0001
 #define TYPE50_MEB2_FMT		0x0002
+#define TYPE50_MEB3_FMT		0x0003
 #define TYPE50_CRB1_FMT		0x0011
 #define TYPE50_CRB2_FMT		0x0012
+#define TYPE50_CRB3_FMT		0x0013
 
 /* Mod-Exp, with a small modulus */
 struct type50_meb1_msg {
@@ -74,6 +76,16 @@ struct type50_meb2_msg {
 	unsigned char	message[256];
 } __attribute__((packed));
 
+/* Mod-Exp, with a larger modulus */
+struct type50_meb3_msg {
+	struct type50_hdr header;
+	unsigned short	keyblock_type;	/* 0x0003 */
+	unsigned char	reserved[6];
+	unsigned char	exponent[512];
+	unsigned char	modulus[512];
+	unsigned char	message[512];
+} __attribute__((packed));
+
 /* CRT, with a small modulus */
 struct type50_crb1_msg {
 	struct type50_hdr header;
@@ -100,6 +112,19 @@ struct type50_crb2_msg {
 	unsigned char	message[256];
 } __attribute__((packed));
 
+/* CRT, with a larger modulus */
+struct type50_crb3_msg {
+	struct type50_hdr header;
+	unsigned short	keyblock_type;	/* 0x0013 */
+	unsigned char	reserved[6];
+	unsigned char	p[256];
+	unsigned char	q[256];
+	unsigned char	dp[256];
+	unsigned char	dq[256];
+	unsigned char	u[256];
+	unsigned char	message[512];
+} __attribute__((packed));
+
 /**
  * The type 80 response family is associated with a CEX2A card.
  *

drivers/s390/crypto/zcrypt_pcica.c

@@ -373,6 +373,7 @@ static int zcrypt_pcica_probe(struct ap_device *ap_dev)
 	zdev->min_mod_size = PCICA_MIN_MOD_SIZE;
 	zdev->max_mod_size = PCICA_MAX_MOD_SIZE;
 	zdev->speed_rating = PCICA_SPEED_RATING;
+	zdev->max_exp_bit_length = PCICA_MAX_MOD_SIZE;
 	ap_dev->reply = &zdev->reply;
 	ap_dev->private = zdev;
 	rc = zcrypt_device_register(zdev);

drivers/s390/crypto/zcrypt_pcicc.c

@@ -579,6 +579,7 @@ static int zcrypt_pcicc_probe(struct ap_device *ap_dev)
 	zdev->min_mod_size = PCICC_MIN_MOD_SIZE;
 	zdev->max_mod_size = PCICC_MAX_MOD_SIZE;
 	zdev->speed_rating = PCICC_SPEED_RATING;
+	zdev->max_exp_bit_length = PCICC_MAX_MOD_SIZE;
 	ap_dev->reply = &zdev->reply;
 	ap_dev->private = zdev;
 	rc = zcrypt_device_register(zdev);

drivers/s390/crypto/zcrypt_pcixcc.c

@@ -45,12 +45,12 @@
 #define PCIXCC_MIN_MOD_SIZE_OLD	 64	/*  512 bits	*/
 #define PCIXCC_MAX_MOD_SIZE	256	/* 2048 bits	*/
 #define CEX3C_MIN_MOD_SIZE	PCIXCC_MIN_MOD_SIZE
-#define CEX3C_MAX_MOD_SIZE	PCIXCC_MAX_MOD_SIZE
+#define CEX3C_MAX_MOD_SIZE	512	/* 4096 bits	*/
 
 #define PCIXCC_MCL2_SPEED_RATING	7870
 #define PCIXCC_MCL3_SPEED_RATING	7870
 #define CEX2C_SPEED_RATING		7000
-#define CEX3C_SPEED_RATING		6500	/* FIXME: needs finetuning */
+#define CEX3C_SPEED_RATING		6500
 
 #define PCIXCC_MAX_ICA_MESSAGE_SIZE 0x77c  /* max size type6 v2 crt message */
 #define PCIXCC_MAX_ICA_RESPONSE_SIZE 0x77c /* max size type86 v2 reply	    */
@@ -567,6 +567,15 @@ static int convert_response_ica(struct zcrypt_device *zdev,
 	case TYPE88_RSP_CODE:
 		return convert_error(zdev, reply);
 	case TYPE86_RSP_CODE:
+		if (msg->cprbx.ccp_rtcode &&
+		   (msg->cprbx.ccp_rscode == 0x14f) &&
+		   (outputdatalength > 256)) {
+			if (zdev->max_exp_bit_length <= 17) {
+				zdev->max_exp_bit_length = 17;
+				return -EAGAIN;
+			} else
+				return -EINVAL;
+		}
 		if (msg->hdr.reply_code)
 			return convert_error(zdev, reply);
 		if (msg->cprbx.cprb_ver_id == 0x02)
@@ -1052,11 +1061,13 @@ static int zcrypt_pcixcc_probe(struct ap_device *ap_dev)
 			zdev->speed_rating = PCIXCC_MCL2_SPEED_RATING;
 			zdev->min_mod_size = PCIXCC_MIN_MOD_SIZE_OLD;
 			zdev->max_mod_size = PCIXCC_MAX_MOD_SIZE;
+			zdev->max_exp_bit_length = PCIXCC_MAX_MOD_SIZE;
 		} else {
 			zdev->type_string = "PCIXCC_MCL3";
 			zdev->speed_rating = PCIXCC_MCL3_SPEED_RATING;
 			zdev->min_mod_size = PCIXCC_MIN_MOD_SIZE;
 			zdev->max_mod_size = PCIXCC_MAX_MOD_SIZE;
+			zdev->max_exp_bit_length = PCIXCC_MAX_MOD_SIZE;
 		}
 		break;
 	case AP_DEVICE_TYPE_CEX2C:
@@ -1065,6 +1076,7 @@ static int zcrypt_pcixcc_probe(struct ap_device *ap_dev)
 		zdev->speed_rating = CEX2C_SPEED_RATING;
 		zdev->min_mod_size = PCIXCC_MIN_MOD_SIZE;
 		zdev->max_mod_size = PCIXCC_MAX_MOD_SIZE;
+		zdev->max_exp_bit_length = PCIXCC_MAX_MOD_SIZE;
 		break;
 	case AP_DEVICE_TYPE_CEX3C:
 		zdev->user_space_type = ZCRYPT_CEX3C;
@@ -1072,6 +1084,7 @@ static int zcrypt_pcixcc_probe(struct ap_device *ap_dev)
 		zdev->speed_rating = CEX3C_SPEED_RATING;
 		zdev->min_mod_size = CEX3C_MIN_MOD_SIZE;
 		zdev->max_mod_size = CEX3C_MAX_MOD_SIZE;
+		zdev->max_exp_bit_length = CEX3C_MAX_MOD_SIZE;
 		break;
 	default:
 		goto out_free;

drivers/s390/kvm/kvm_virtio.c

@@ -10,6 +10,7 @@
  *    Author(s): Christian Borntraeger <borntraeger@de.ibm.com>
  */
 
+#include <linux/kernel_stat.h>
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/err.h>
@@ -25,6 +26,7 @@
 #include <asm/kvm_virtio.h>
 #include <asm/setup.h>
 #include <asm/s390_ext.h>
+#include <asm/irq.h>
 
 #define VIRTIO_SUBCODE_64 0x0D00
 
@@ -379,6 +381,7 @@ static void kvm_extint_handler(unsigned int ext_int_code,
 	u16 subcode;
 	u32 param;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_VRT]++;
 	subcode = ext_int_code >> 16;
 	if ((subcode & 0xff00) != VIRTIO_SUBCODE_64)
 		return;

drivers/s390/net/Kconfig

@@ -2,7 +2,8 @@ menu "S/390 network device drivers"
 	depends on NETDEVICES && S390
 
 config LCS
-	tristate "Lan Channel Station Interface"
+	def_tristate m
+	prompt "Lan Channel Station Interface"
 	depends on CCW && NETDEVICES && (NET_ETHERNET || TR || FDDI)
 	help
 	   Select this option if you want to use LCS networking on IBM System z.
@@ -12,7 +13,8 @@ config LCS
 	   If you do not know what it is, it's safe to choose Y.
 
 config CTCM
-	tristate "CTC and MPC SNA device support"
+	def_tristate m
+	prompt "CTC and MPC SNA device support"
 	depends on CCW && NETDEVICES
 	help
 	  Select this option if you want to use channel-to-channel
@@ -26,7 +28,8 @@ config CTCM
 	  If you do not need any channel-to-channel connection, choose N.
 
 config NETIUCV
-	tristate "IUCV network device support (VM only)"
+	def_tristate m
+	prompt "IUCV network device support (VM only)"
 	depends on IUCV && NETDEVICES
 	help
 	  Select this option if you want to use inter-user communication
@@ -37,14 +40,16 @@ config NETIUCV
 	  The module name is netiucv. If unsure, choose Y.
 
 config SMSGIUCV
-	tristate "IUCV special message support (VM only)"
+	def_tristate m
+	prompt "IUCV special message support (VM only)"
 	depends on IUCV
 	help
 	  Select this option if you want to be able to receive SMSG messages
 	  from other VM guest systems.
 
 config SMSGIUCV_EVENT
-	tristate "Deliver IUCV special messages as uevents (VM only)"
+	def_tristate m
+	prompt "Deliver IUCV special messages as uevents (VM only)"
 	depends on SMSGIUCV
 	help
 	  Select this option to deliver CP special messages (SMSGs) as
@@ -54,7 +59,8 @@ config SMSGIUCV_EVENT
 	  To compile as a module, choose M. The module name is "smsgiucv_app".
 
 config CLAW
-	tristate "CLAW device support"
+	def_tristate m
+	prompt "CLAW device support"
 	depends on CCW && NETDEVICES
 	help
 	  This driver supports channel attached CLAW devices.
@@ -64,7 +70,8 @@ config CLAW
 	  To compile into the kernel, choose Y.
 
 config QETH
-	tristate "Gigabit Ethernet device support"
+	def_tristate y
+	prompt "Gigabit Ethernet device support"
 	depends on CCW && NETDEVICES && IP_MULTICAST && QDIO
 	help
 	  This driver supports the IBM System z OSA Express adapters
@@ -78,25 +85,25 @@ config QETH
 	  The module name is qeth.
 
 config QETH_L2
-        tristate "qeth layer 2 device support"
-        depends on QETH
-        help
-          Select this option to be able to run qeth devices in layer 2 mode.
-          To compile as a module, choose M. The module name is qeth_l2.
-          If unsure, choose y.
+	def_tristate y
+	prompt "qeth layer 2 device support"
+	depends on QETH
+	help
+	  Select this option to be able to run qeth devices in layer 2 mode.
+	  To compile as a module, choose M. The module name is qeth_l2.
+	  If unsure, choose y.
 
 config QETH_L3
-        tristate "qeth layer 3 device support"
-        depends on QETH
-        help
-          Select this option to be able to run qeth devices in layer 3 mode.
-          To compile as a module choose M. The module name is qeth_l3.
-          If unsure, choose Y.
+	def_tristate y
+	prompt "qeth layer 3 device support"
+	depends on QETH
+	help
+	  Select this option to be able to run qeth devices in layer 3 mode.
+	  To compile as a module choose M. The module name is qeth_l3.
+	  If unsure, choose Y.
 
 config QETH_IPV6
-        bool
-        depends on (QETH_L3 = IPV6) || (QETH_L3 && IPV6 = 'y')
-        default y
+	def_bool y if (QETH_L3 = IPV6) || (QETH_L3 && IPV6 = 'y')
 
 config CCWGROUP
 	tristate

drivers/s390/net/claw.c

@@ -63,6 +63,7 @@
 
 #define KMSG_COMPONENT "claw"
 
+#include <linux/kernel_stat.h>
 #include <asm/ccwdev.h>
 #include <asm/ccwgroup.h>
 #include <asm/debug.h>
@@ -640,6 +641,7 @@ claw_irq_handler(struct ccw_device *cdev,
         struct claw_env  *p_env;
         struct chbk *p_ch_r=NULL;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_CLW]++;
 	CLAW_DBF_TEXT(4, trace, "clawirq");
         /* Bypass all 'unsolicited interrupts' */
 	privptr = dev_get_drvdata(&cdev->dev);

drivers/s390/net/ctcm_main.c

@@ -24,6 +24,7 @@
 #define KMSG_COMPONENT "ctcm"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/kernel.h>
@@ -1204,6 +1205,7 @@ static void ctcm_irq_handler(struct ccw_device *cdev,
 	int cstat;
 	int dstat;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_CTC]++;
 	CTCM_DBF_TEXT_(TRACE, CTC_DBF_DEBUG,
 		"Enter %s(%s)", CTCM_FUNTAIL, dev_name(&cdev->dev));
 

drivers/s390/net/lcs.c

@@ -26,6 +26,7 @@
 #define KMSG_COMPONENT		"lcs"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/if.h>
 #include <linux/netdevice.h>
@@ -1398,6 +1399,7 @@ lcs_irq(struct ccw_device *cdev, unsigned long intparm, struct irb *irb)
 	int rc, index;
 	int cstat, dstat;
 
+	kstat_cpu(smp_processor_id()).irqs[IOINT_LCS]++;
 	if (lcs_check_irb_error(cdev, irb))
 		return;
 

drivers/s390/net/qeth_core_main.c

@@ -3831,6 +3831,8 @@ static int qeth_qdio_establish(struct qeth_card *card)
 	init_data.int_parm               = (unsigned long) card;
 	init_data.input_sbal_addr_array  = (void **) in_sbal_ptrs;
 	init_data.output_sbal_addr_array = (void **) out_sbal_ptrs;
+	init_data.scan_threshold =
+		(card->info.type == QETH_CARD_TYPE_IQD) ? 8 : 32;
 
 	if (atomic_cmpxchg(&card->qdio.state, QETH_QDIO_ALLOCATED,
 		QETH_QDIO_ESTABLISHED) == QETH_QDIO_ALLOCATED) {

drivers/s390/scsi/zfcp_qdio.c

@@ -290,6 +290,8 @@ static void zfcp_qdio_setup_init_data(struct qdio_initialize *id,
 	id->int_parm = (unsigned long) qdio;
 	id->input_sbal_addr_array = (void **) (qdio->res_q);
 	id->output_sbal_addr_array = (void **) (qdio->req_q);
+	id->scan_threshold =
+		QDIO_MAX_BUFFERS_PER_Q - ZFCP_QDIO_MAX_SBALS_PER_REQ * 2;
 }
 
 /**

net/iucv/iucv.c

@@ -36,6 +36,7 @@
 #define KMSG_COMPONENT "iucv"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/kernel_stat.h>
 #include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/spinlock.h>
@@ -1804,6 +1805,7 @@ static void iucv_external_interrupt(unsigned int ext_int_code,
 	struct iucv_irq_data *p;
 	struct iucv_irq_list *work;
 
+	kstat_cpu(smp_processor_id()).irqs[EXTINT_IUC]++;
 	p = iucv_irq_data[smp_processor_id()];
 	if (p->ippathid >= iucv_max_pathid) {
 		WARN_ON(p->ippathid >= iucv_max_pathid);

tools/perf/arch/s390/Makefile

@@ -0,0 +1,4 @@
+ifndef NO_DWARF
+PERF_HAVE_DWARF_REGS := 1
+LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o
+endif

tools/perf/arch/s390/util/dwarf-regs.c

@@ -0,0 +1,22 @@
+/*
+ * Mapping of DWARF debug register numbers into register names.
+ *
+ *    Copyright IBM Corp. 2010
+ *    Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>,
+ *
+ */
+
+#include <libio.h>
+#include <dwarf-regs.h>
+
+#define NUM_GPRS 16
+
+static const char *gpr_names[NUM_GPRS] = {
+	"%r0", "%r1",  "%r2",  "%r3",  "%r4",  "%r5",  "%r6",  "%r7",
+	"%r8", "%r9", "%r10", "%r11", "%r12", "%r13", "%r14", "%r15",
+};
+
+const char *get_arch_regstr(unsigned int n)
+{
+	return (n >= NUM_GPRS) ? NULL : gpr_names[n];
+}