diff options
Diffstat (limited to 'arch/s390')
54 files changed, 3028 insertions, 302 deletions
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig index e9f312532526..65a07750f4f9 100644 --- a/arch/s390/Kconfig +++ b/arch/s390/Kconfig @@ -334,10 +334,10 @@ config SMP a system with only one CPU, like most personal computers, say N. If you have a system with more than one CPU, say Y. - If you say N here, the kernel will run on single and multiprocessor + If you say N here, the kernel will run on uni- and multiprocessor machines, but will use only one CPU of a multiprocessor machine. If you say Y here, the kernel will run on many, but not all, - singleprocessor machines. On a singleprocessor machine, the kernel + uniprocessor machines. On a uniprocessor machine, the kernel will run faster if you say N here. See also the SMP-HOWTO available at @@ -596,7 +596,7 @@ config CRASH_DUMP config ZFCPDUMP def_bool n prompt "zfcpdump support" - depends on SMP + depends on 64BIT && SMP 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. diff --git a/arch/s390/crypto/des_s390.c b/arch/s390/crypto/des_s390.c index bcca01c9989d..200f2a1b599d 100644 --- a/arch/s390/crypto/des_s390.c +++ b/arch/s390/crypto/des_s390.c @@ -237,9 +237,9 @@ static int des3_setkey(struct crypto_tfm *tfm, const u8 *key, struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm); u32 *flags = &tfm->crt_flags; - if (!(memcmp(key, &key[DES_KEY_SIZE], DES_KEY_SIZE) && - memcmp(&key[DES_KEY_SIZE], &key[DES_KEY_SIZE * 2], - DES_KEY_SIZE)) && + if (!(crypto_memneq(key, &key[DES_KEY_SIZE], DES_KEY_SIZE) && + crypto_memneq(&key[DES_KEY_SIZE], &key[DES_KEY_SIZE * 2], + DES_KEY_SIZE)) && (*flags & CRYPTO_TFM_REQ_WEAK_KEY)) { *flags |= CRYPTO_TFM_RES_WEAK_KEY; return -EINVAL; diff --git a/arch/s390/hypfs/Makefile b/arch/s390/hypfs/Makefile index 2e671d5004ca..06f8d95a16cd 100644 --- a/arch/s390/hypfs/Makefile +++ b/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 hypfs_dbfs.o +s390_hypfs-objs := inode.o hypfs_diag.o hypfs_vm.o hypfs_dbfs.o hypfs_sprp.o diff --git a/arch/s390/hypfs/hypfs.h b/arch/s390/hypfs/hypfs.h index 79f2ac55253f..b34b5ab90a31 100644 --- a/arch/s390/hypfs/hypfs.h +++ b/arch/s390/hypfs/hypfs.h @@ -13,6 +13,7 @@ #include <linux/debugfs.h> #include <linux/workqueue.h> #include <linux/kref.h> +#include <asm/hypfs.h> #define REG_FILE_MODE 0440 #define UPDATE_FILE_MODE 0220 @@ -36,6 +37,10 @@ extern int hypfs_vm_init(void); extern void hypfs_vm_exit(void); extern int hypfs_vm_create_files(struct dentry *root); +/* Set Partition-Resource Parameter */ +int hypfs_sprp_init(void); +void hypfs_sprp_exit(void); + /* debugfs interface */ struct hypfs_dbfs_file; @@ -52,6 +57,8 @@ struct hypfs_dbfs_file { int (*data_create)(void **data, void **data_free_ptr, size_t *size); void (*data_free)(const void *buf_free_ptr); + long (*unlocked_ioctl) (struct file *, unsigned int, + unsigned long); /* Private data for hypfs_dbfs.c */ struct hypfs_dbfs_data *data; diff --git a/arch/s390/hypfs/hypfs_dbfs.c b/arch/s390/hypfs/hypfs_dbfs.c index 17ab8b7b53cc..2badf2bf9cd7 100644 --- a/arch/s390/hypfs/hypfs_dbfs.c +++ b/arch/s390/hypfs/hypfs_dbfs.c @@ -81,9 +81,25 @@ static ssize_t dbfs_read(struct file *file, char __user *buf, return rc; } +static long dbfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +{ + struct hypfs_dbfs_file *df; + long rc; + + df = file->f_path.dentry->d_inode->i_private; + mutex_lock(&df->lock); + if (df->unlocked_ioctl) + rc = df->unlocked_ioctl(file, cmd, arg); + else + rc = -ENOTTY; + mutex_unlock(&df->lock); + return rc; +} + static const struct file_operations dbfs_ops = { .read = dbfs_read, .llseek = no_llseek, + .unlocked_ioctl = dbfs_ioctl, }; int hypfs_dbfs_create_file(struct hypfs_dbfs_file *df) diff --git a/arch/s390/hypfs/hypfs_sprp.c b/arch/s390/hypfs/hypfs_sprp.c new file mode 100644 index 000000000000..f043c3c7e73c --- /dev/null +++ b/arch/s390/hypfs/hypfs_sprp.c @@ -0,0 +1,141 @@ +/* + * Hypervisor filesystem for Linux on s390. + * Set Partition-Resource Parameter interface. + * + * Copyright IBM Corp. 2013 + * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com> + */ + +#include <linux/compat.h> +#include <linux/errno.h> +#include <linux/gfp.h> +#include <linux/string.h> +#include <linux/types.h> +#include <linux/uaccess.h> +#include <asm/compat.h> +#include <asm/sclp.h> +#include "hypfs.h" + +#define DIAG304_SET_WEIGHTS 0 +#define DIAG304_QUERY_PRP 1 +#define DIAG304_SET_CAPPING 2 + +#define DIAG304_CMD_MAX 2 + +static unsigned long hypfs_sprp_diag304(void *data, unsigned long cmd) +{ + register unsigned long _data asm("2") = (unsigned long) data; + register unsigned long _rc asm("3"); + register unsigned long _cmd asm("4") = cmd; + + asm volatile("diag %1,%2,0x304\n" + : "=d" (_rc) : "d" (_data), "d" (_cmd) : "memory"); + + return _rc; +} + +static void hypfs_sprp_free(const void *data) +{ + free_page((unsigned long) data); +} + +static int hypfs_sprp_create(void **data_ptr, void **free_ptr, size_t *size) +{ + unsigned long rc; + void *data; + + data = (void *) get_zeroed_page(GFP_KERNEL); + if (!data) + return -ENOMEM; + rc = hypfs_sprp_diag304(data, DIAG304_QUERY_PRP); + if (rc != 1) { + *data_ptr = *free_ptr = NULL; + *size = 0; + free_page((unsigned long) data); + return -EIO; + } + *data_ptr = *free_ptr = data; + *size = PAGE_SIZE; + return 0; +} + +static int __hypfs_sprp_ioctl(void __user *user_area) +{ + struct hypfs_diag304 diag304; + unsigned long cmd; + void __user *udata; + void *data; + int rc; + + if (copy_from_user(&diag304, user_area, sizeof(diag304))) + return -EFAULT; + if ((diag304.args[0] >> 8) != 0 || diag304.args[1] > DIAG304_CMD_MAX) + return -EINVAL; + + data = (void *) get_zeroed_page(GFP_KERNEL | GFP_DMA); + if (!data) + return -ENOMEM; + + udata = (void __user *)(unsigned long) diag304.data; + if (diag304.args[1] == DIAG304_SET_WEIGHTS || + diag304.args[1] == DIAG304_SET_CAPPING) + if (copy_from_user(data, udata, PAGE_SIZE)) { + rc = -EFAULT; + goto out; + } + + cmd = *(unsigned long *) &diag304.args[0]; + diag304.rc = hypfs_sprp_diag304(data, cmd); + + if (diag304.args[1] == DIAG304_QUERY_PRP) + if (copy_to_user(udata, data, PAGE_SIZE)) { + rc = -EFAULT; + goto out; + } + + rc = copy_to_user(user_area, &diag304, sizeof(diag304)) ? -EFAULT : 0; +out: + free_page((unsigned long) data); + return rc; +} + +static long hypfs_sprp_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + void __user *argp; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (is_compat_task()) + argp = compat_ptr(arg); + else + argp = (void __user *) arg; + switch (cmd) { + case HYPFS_DIAG304: + return __hypfs_sprp_ioctl(argp); + default: /* unknown ioctl number */ + return -ENOTTY; + } + return 0; +} + +static struct hypfs_dbfs_file hypfs_sprp_file = { + .name = "diag_304", + .data_create = hypfs_sprp_create, + .data_free = hypfs_sprp_free, + .unlocked_ioctl = hypfs_sprp_ioctl, +}; + +int hypfs_sprp_init(void) +{ + if (!sclp_has_sprp()) + return 0; + return hypfs_dbfs_create_file(&hypfs_sprp_file); +} + +void hypfs_sprp_exit(void) +{ + if (!sclp_has_sprp()) + return; + hypfs_dbfs_remove_file(&hypfs_sprp_file); +} diff --git a/arch/s390/hypfs/inode.c b/arch/s390/hypfs/inode.c index ddfe09b45134..c952b981e4f2 100644 --- a/arch/s390/hypfs/inode.c +++ b/arch/s390/hypfs/inode.c @@ -478,10 +478,14 @@ static int __init hypfs_init(void) rc = -ENODATA; goto fail_hypfs_diag_exit; } + if (hypfs_sprp_init()) { + rc = -ENODATA; + goto fail_hypfs_vm_exit; + } s390_kobj = kobject_create_and_add("s390", hypervisor_kobj); if (!s390_kobj) { rc = -ENOMEM; - goto fail_hypfs_vm_exit; + goto fail_hypfs_sprp_exit; } rc = register_filesystem(&hypfs_type); if (rc) @@ -490,6 +494,8 @@ static int __init hypfs_init(void) fail_filesystem: kobject_put(s390_kobj); +fail_hypfs_sprp_exit: + hypfs_sprp_exit(); fail_hypfs_vm_exit: hypfs_vm_exit(); fail_hypfs_diag_exit: @@ -502,11 +508,12 @@ fail_dbfs_exit: static void __exit hypfs_exit(void) { - hypfs_diag_exit(); - hypfs_vm_exit(); - hypfs_dbfs_exit(); unregister_filesystem(&hypfs_type); kobject_put(s390_kobj); + hypfs_sprp_exit(); + hypfs_vm_exit(); + hypfs_diag_exit(); + hypfs_dbfs_exit(); } module_init(hypfs_init) diff --git a/arch/s390/include/asm/Kbuild b/arch/s390/include/asm/Kbuild index 7a5288f3479a..8386a4a1f19a 100644 --- a/arch/s390/include/asm/Kbuild +++ b/arch/s390/include/asm/Kbuild @@ -3,3 +3,4 @@ generic-y += clkdev.h generic-y += trace_clock.h generic-y += preempt.h +generic-y += hash.h diff --git a/arch/s390/include/asm/barrier.h b/arch/s390/include/asm/barrier.h index 16760eeb79b0..578680f6207a 100644 --- a/arch/s390/include/asm/barrier.h +++ b/arch/s390/include/asm/barrier.h @@ -32,4 +32,19 @@ #define set_mb(var, value) do { var = value; mb(); } while (0) +#define smp_store_release(p, v) \ +do { \ + compiletime_assert_atomic_type(*p); \ + barrier(); \ + ACCESS_ONCE(*p) = (v); \ +} while (0) + +#define smp_load_acquire(p) \ +({ \ + typeof(*p) ___p1 = ACCESS_ONCE(*p); \ + compiletime_assert_atomic_type(*p); \ + barrier(); \ + ___p1; \ +}) + #endif /* __ASM_BARRIER_H */ diff --git a/arch/s390/include/asm/cmpxchg.h b/arch/s390/include/asm/cmpxchg.h index 0f636cbdf342..4236408070e5 100644 --- a/arch/s390/include/asm/cmpxchg.h +++ b/arch/s390/include/asm/cmpxchg.h @@ -185,11 +185,12 @@ static inline unsigned long long __cmpxchg64(void *ptr, { register_pair rp_old = {.pair = old}; register_pair rp_new = {.pair = new}; + unsigned long long *ullptr = ptr; asm volatile( " cds %0,%2,%1" - : "+&d" (rp_old), "=Q" (ptr) - : "d" (rp_new), "Q" (ptr) + : "+d" (rp_old), "+Q" (*ullptr) + : "d" (rp_new) : "memory", "cc"); return rp_old.pair; } diff --git a/arch/s390/include/asm/compat.h b/arch/s390/include/asm/compat.h index 4bf9da03591e..5d7e8cf83bd6 100644 --- a/arch/s390/include/asm/compat.h +++ b/arch/s390/include/asm/compat.h @@ -38,7 +38,8 @@ #define PSW32_USER_BITS (PSW32_MASK_DAT | PSW32_MASK_IO | PSW32_MASK_EXT | \ PSW32_DEFAULT_KEY | PSW32_MASK_BASE | \ - PSW32_MASK_MCHECK | PSW32_MASK_PSTATE | PSW32_ASC_HOME) + PSW32_MASK_MCHECK | PSW32_MASK_PSTATE | \ + PSW32_ASC_PRIMARY) #define COMPAT_USER_HZ 100 #define COMPAT_UTS_MACHINE "s390\0\0\0\0" diff --git a/arch/s390/include/asm/cpu_mf.h b/arch/s390/include/asm/cpu_mf.h index c879fad404c8..cb700d54bd83 100644 --- a/arch/s390/include/asm/cpu_mf.h +++ b/arch/s390/include/asm/cpu_mf.h @@ -56,6 +56,96 @@ struct cpumf_ctr_info { u32 reserved2[12]; } __packed; +/* QUERY SAMPLING INFORMATION block */ +struct hws_qsi_info_block { /* Bit(s) */ + unsigned int b0_13:14; /* 0-13: zeros */ + unsigned int as:1; /* 14: basic-sampling authorization */ + unsigned int ad:1; /* 15: diag-sampling authorization */ + unsigned int b16_21:6; /* 16-21: zeros */ + unsigned int es:1; /* 22: basic-sampling enable control */ + unsigned int ed:1; /* 23: diag-sampling enable control */ + unsigned int b24_29:6; /* 24-29: zeros */ + unsigned int cs:1; /* 30: basic-sampling activation control */ + unsigned int cd:1; /* 31: diag-sampling activation control */ + unsigned int bsdes:16; /* 4-5: size of basic sampling entry */ + unsigned int dsdes:16; /* 6-7: size of diagnostic sampling entry */ + unsigned long min_sampl_rate; /* 8-15: minimum sampling interval */ + unsigned long max_sampl_rate; /* 16-23: maximum sampling interval*/ + unsigned long tear; /* 24-31: TEAR contents */ + unsigned long dear; /* 32-39: DEAR contents */ + unsigned int rsvrd0; /* 40-43: reserved */ + unsigned int cpu_speed; /* 44-47: CPU speed */ + unsigned long long rsvrd1; /* 48-55: reserved */ + unsigned long long rsvrd2; /* 56-63: reserved */ +} __packed; + +/* SET SAMPLING CONTROLS request block */ +struct hws_lsctl_request_block { + unsigned int s:1; /* 0: maximum buffer indicator */ + unsigned int h:1; /* 1: part. level reserved for VM use*/ + unsigned long long b2_53:52;/* 2-53: zeros */ + unsigned int es:1; /* 54: basic-sampling enable control */ + unsigned int ed:1; /* 55: diag-sampling enable control */ + unsigned int b56_61:6; /* 56-61: - zeros */ + unsigned int cs:1; /* 62: basic-sampling activation control */ + unsigned int cd:1; /* 63: diag-sampling activation control */ + unsigned long interval; /* 8-15: sampling interval */ + unsigned long tear; /* 16-23: TEAR contents */ + unsigned long dear; /* 24-31: DEAR contents */ + /* 32-63: */ + unsigned long rsvrd1; /* reserved */ + unsigned long rsvrd2; /* reserved */ + unsigned long rsvrd3; /* reserved */ + unsigned long rsvrd4; /* reserved */ +} __packed; + +struct hws_basic_entry { + unsigned int def:16; /* 0-15 Data Entry Format */ + unsigned int R:4; /* 16-19 reserved */ + unsigned int U:4; /* 20-23 Number of unique instruct. */ + unsigned int z:2; /* zeros */ + unsigned int T:1; /* 26 PSW DAT mode */ + unsigned int W:1; /* 27 PSW wait state */ + unsigned int P:1; /* 28 PSW Problem state */ + unsigned int AS:2; /* 29-30 PSW address-space control */ + unsigned int I:1; /* 31 entry valid or invalid */ + unsigned int:16; + unsigned int prim_asn:16; /* primary ASN */ + unsigned long long ia; /* Instruction Address */ + unsigned long long gpp; /* Guest Program Parameter */ + unsigned long long hpp; /* Host Program Parameter */ +} __packed; + +struct hws_diag_entry { + unsigned int def:16; /* 0-15 Data Entry Format */ + unsigned int R:14; /* 16-19 and 20-30 reserved */ + unsigned int I:1; /* 31 entry valid or invalid */ + u8 data[]; /* Machine-dependent sample data */ +} __packed; + +struct hws_combined_entry { + struct hws_basic_entry basic; /* Basic-sampling data entry */ + struct hws_diag_entry diag; /* Diagnostic-sampling data entry */ +} __packed; + +struct hws_trailer_entry { + union { + struct { + unsigned int f:1; /* 0 - Block Full Indicator */ + unsigned int a:1; /* 1 - Alert request control */ + unsigned int t:1; /* 2 - Timestamp format */ + unsigned long long:61; /* 3 - 63: Reserved */ + }; + unsigned long long flags; /* 0 - 63: All indicators */ + }; + unsigned long long overflow; /* 64 - sample Overflow count */ + unsigned char timestamp[16]; /* 16 - 31 timestamp */ + unsigned long long reserved1; /* 32 -Reserved */ + unsigned long long reserved2; /* */ + unsigned long long progusage1; /* 48 - reserved for programming use */ + unsigned long long progusage2; /* */ +} __packed; + /* Query counter information */ static inline int qctri(struct cpumf_ctr_info *info) { @@ -99,4 +189,95 @@ static inline int ecctr(u64 ctr, u64 *val) return cc; } +/* Query sampling information */ +static inline int qsi(struct hws_qsi_info_block *info) +{ + int cc; + cc = 1; + + asm volatile( + "0: .insn s,0xb2860000,0(%1)\n" + "1: lhi %0,0\n" + "2:\n" + EX_TABLE(0b, 2b) EX_TABLE(1b, 2b) + : "=d" (cc), "+a" (info) + : "m" (*info) + : "cc", "memory"); + + return cc ? -EINVAL : 0; +} + +/* Load sampling controls */ +static inline int lsctl(struct hws_lsctl_request_block *req) +{ + int cc; + + cc = 1; + asm volatile( + "0: .insn s,0xb2870000,0(%1)\n" + "1: ipm %0\n" + " srl %0,28\n" + "2:\n" + EX_TABLE(0b, 2b) EX_TABLE(1b, 2b) + : "+d" (cc), "+a" (req) + : "m" (*req) + : "cc", "memory"); + + return cc ? -EINVAL : 0; +} + +/* Sampling control helper functions */ + +#include <linux/time.h> + +static inline unsigned long freq_to_sample_rate(struct hws_qsi_info_block *qsi, + unsigned long freq) +{ + return (USEC_PER_SEC / freq) * qsi->cpu_speed; +} + +static inline unsigned long sample_rate_to_freq(struct hws_qsi_info_block *qsi, + unsigned long rate) +{ + return USEC_PER_SEC * qsi->cpu_speed / rate; +} + +#define SDB_TE_ALERT_REQ_MASK 0x4000000000000000UL +#define SDB_TE_BUFFER_FULL_MASK 0x8000000000000000UL + +/* Return TOD timestamp contained in an trailer entry */ +static inline unsigned long long trailer_timestamp(struct hws_trailer_entry *te) +{ + /* TOD in STCKE format */ + if (te->t) + return *((unsigned long long *) &te->timestamp[1]); + + /* TOD in STCK format */ + return *((unsigned long long *) &te->timestamp[0]); +} + +/* Return pointer to trailer entry of an sample data block */ +static inline unsigned long *trailer_entry_ptr(unsigned long v) +{ + void *ret; + + ret = (void *) v; + ret += PAGE_SIZE; + ret -= sizeof(struct hws_trailer_entry); + + return (unsigned long *) ret; +} + +/* Return if the entry in the sample data block table (sdbt) + * is a link to the next sdbt */ +static inline int is_link_entry(unsigned long *s) +{ + return *s & 0x1ul ? 1 : 0; +} + +/* Return pointer to the linked sdbt */ +static inline unsigned long *get_next_sdbt(unsigned long *s) +{ + return (unsigned long *) (*s & ~0x1ul); +} #endif /* _ASM_S390_CPU_MF_H */ diff --git a/arch/s390/include/asm/css_chars.h b/arch/s390/include/asm/css_chars.h index 7e1c917bbba2..09d1dd46bd57 100644 --- a/arch/s390/include/asm/css_chars.h +++ b/arch/s390/include/asm/css_chars.h @@ -29,6 +29,8 @@ struct css_general_char { u32 fcx : 1; /* bit 88 */ u32 : 19; u32 alt_ssi : 1; /* bit 108 */ + u32:1; + u32 narf:1; /* bit 110 */ } __packed; extern struct css_general_char css_general_characteristics; diff --git a/arch/s390/include/asm/pci.h b/arch/s390/include/asm/pci.h index c129ab2ac731..2583466f576b 100644 --- a/arch/s390/include/asm/pci.h +++ b/arch/s390/include/asm/pci.h @@ -144,6 +144,7 @@ int clp_disable_fh(struct zpci_dev *); void zpci_event_error(void *); void zpci_event_availability(void *); void zpci_rescan(void); +bool zpci_is_enabled(void); #else /* CONFIG_PCI */ static inline void zpci_event_error(void *e) {} static inline void zpci_event_availability(void *e) {} diff --git a/arch/s390/include/asm/perf_event.h b/arch/s390/include/asm/perf_event.h index 1141fb3e7b21..159a8ec6da9a 100644 --- a/arch/s390/include/asm/perf_event.h +++ b/arch/s390/include/asm/perf_event.h @@ -1,21 +1,40 @@ /* * Performance event support - s390 specific definitions. * - * Copyright IBM Corp. 2009, 2012 + * Copyright IBM Corp. 2009, 2013 * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com> * Hendrik Brueckner <brueckner@linux.vnet.ibm.com> */ -#include <asm/cpu_mf.h> +#ifndef _ASM_S390_PERF_EVENT_H +#define _ASM_S390_PERF_EVENT_H -/* CPU-measurement counter facility */ -#define PERF_CPUM_CF_MAX_CTR 256 +#ifdef CONFIG_64BIT + +#include <linux/perf_event.h> +#include <linux/device.h> +#include <asm/cpu_mf.h> /* Per-CPU flags for PMU states */ #define PMU_F_RESERVED 0x1000 #define PMU_F_ENABLED 0x2000 +#define PMU_F_IN_USE 0x4000 +#define PMU_F_ERR_IBE 0x0100 +#define PMU_F_ERR_LSDA 0x0200 +#define PMU_F_ERR_MASK (PMU_F_ERR_IBE|PMU_F_ERR_LSDA) + +/* Perf defintions for PMU event attributes in sysfs */ +extern __init const struct attribute_group **cpumf_cf_event_group(void); +extern ssize_t cpumf_events_sysfs_show(struct device *dev, + struct device_attribute *attr, + char *page); +#define EVENT_VAR(_cat, _name) event_attr_##_cat##_##_name +#define EVENT_PTR(_cat, _name) (&EVENT_VAR(_cat, _name).attr.attr) + +#define CPUMF_EVENT_ATTR(cat, name, id) \ + PMU_EVENT_ATTR(name, EVENT_VAR(cat, name), id, cpumf_events_sysfs_show) +#define CPUMF_EVENT_PTR(cat, name) EVENT_PTR(cat, name) -#ifdef CONFIG_64BIT /* Perf callbacks */ struct pt_regs; @@ -23,4 +42,55 @@ extern unsigned long perf_instruction_pointer(struct pt_regs *regs); extern unsigned long perf_misc_flags(struct pt_regs *regs); #define perf_misc_flags(regs) perf_misc_flags(regs) +/* Perf pt_regs extension for sample-data-entry indicators */ +struct perf_sf_sde_regs { + unsigned char in_guest:1; /* guest sample */ + unsigned long reserved:63; /* reserved */ +}; + +/* Perf PMU definitions for the counter facility */ +#define PERF_CPUM_CF_MAX_CTR 256 + +/* Perf PMU definitions for the sampling facility */ +#define PERF_CPUM_SF_MAX_CTR 2 +#define PERF_EVENT_CPUM_SF 0xB0000UL /* Event: Basic-sampling */ +#define PERF_EVENT_CPUM_SF_DIAG 0xBD000UL /* Event: Combined-sampling */ +#define PERF_CPUM_SF_BASIC_MODE 0x0001 /* Basic-sampling flag */ +#define PERF_CPUM_SF_DIAG_MODE 0x0002 /* Diagnostic-sampling flag */ +#define PERF_CPUM_SF_MODE_MASK (PERF_CPUM_SF_BASIC_MODE| \ + PERF_CPUM_SF_DIAG_MODE) +#define PERF_CPUM_SF_FULL_BLOCKS 0x0004 /* Process full SDBs only */ + +#define REG_NONE 0 +#define REG_OVERFLOW 1 +#define OVERFLOW_REG(hwc) ((hwc)->extra_reg.config) +#define SFB_ALLOC_REG(hwc) ((hwc)->extra_reg.alloc) +#define RAWSAMPLE_REG(hwc) ((hwc)->config) +#define TEAR_REG(hwc) ((hwc)->last_tag) +#define SAMPL_RATE(hwc) ((hwc)->event_base) +#define SAMPL_FLAGS(hwc) ((hwc)->config_base) +#define SAMPL_DIAG_MODE(hwc) (SAMPL_FLAGS(hwc) & PERF_CPUM_SF_DIAG_MODE) +#define SDB_FULL_BLOCKS(hwc) (SAMPL_FLAGS(hwc) & PERF_CPUM_SF_FULL_BLOCKS) + +/* Structure for sampling data entries to be passed as perf raw sample data + * to user space. Note that raw sample data must be aligned and, thus, might + * be padded with zeros. + */ +struct sf_raw_sample { +#define SF_RAW_SAMPLE_BASIC PERF_CPUM_SF_BASIC_MODE +#define SF_RAW_SAMPLE_DIAG PERF_CPUM_SF_DIAG_MODE + u64 format; + u32 size; /* Size of sf_raw_sample */ + u16 bsdes; /* Basic-sampling data entry size */ + u16 dsdes; /* Diagnostic-sampling data entry size */ + struct hws_basic_entry basic; /* Basic-sampling data entry */ + struct hws_diag_entry diag; /* Diagnostic-sampling data entry */ + u8 padding[]; /* Padding to next multiple of 8 */ +} __packed; + +/* Perf hardware reserve and release functions */ +int perf_reserve_sampling(void); +void perf_release_sampling(void); + #endif /* CONFIG_64BIT */ +#endif /* _ASM_S390_PERF_EVENT_H */ diff --git a/arch/s390/include/asm/qdio.h b/arch/s390/include/asm/qdio.h index 57d0d7e794b1..d786c634e052 100644 --- a/arch/s390/include/asm/qdio.h +++ b/arch/s390/include/asm/qdio.h @@ -336,7 +336,7 @@ typedef void qdio_handler_t(struct ccw_device *, unsigned int, int, #define QDIO_FLAG_CLEANUP_USING_HALT 0x02 /** - * struct qdio_initialize - qdio initalization data + * struct qdio_initialize - qdio initialization data * @cdev: associated ccw device * @q_format: queue format * @adapter_name: name for the adapter @@ -378,6 +378,34 @@ struct qdio_initialize { struct qdio_outbuf_state *output_sbal_state_array; }; +/** + * enum qdio_brinfo_entry_type - type of address entry for qdio_brinfo_desc() + * @l3_ipv6_addr: entry contains IPv6 address + * @l3_ipv4_addr: entry contains IPv4 address + * @l2_addr_lnid: entry contains MAC address and VLAN ID + */ +enum qdio_brinfo_entry_type {l3_ipv6_addr, l3_ipv4_addr, l2_addr_lnid}; + +/** + * struct qdio_brinfo_entry_XXX - Address entry for qdio_brinfo_desc() + * @nit: Network interface token + * @addr: Address of one of the three types + * + * The struct is passed to the callback function by qdio_brinfo_desc() + */ +struct qdio_brinfo_entry_l3_ipv6 { + u64 nit; + struct { unsigned char _s6_addr[16]; } addr; +} __packed; +struct qdio_brinfo_entry_l3_ipv4 { + u64 nit; + struct { uint32_t _s_addr; } addr; +} __packed; +struct qdio_brinfo_entry_l2 { + u64 nit; + struct { u8 mac[6]; u16 lnid; } addr_lnid; +} __packed; + #define QDIO_STATE_INACTIVE 0x00000002 /* after qdio_cleanup */ #define QDIO_STATE_ESTABLISHED 0x00000004 /* after qdio_establish */ #define QDIO_STATE_ACTIVE 0x00000008 /* after qdio_activate */ @@ -399,5 +427,10 @@ extern int qdio_get_next_buffers(struct ccw_device *, int, int *, int *); extern int qdio_shutdown(struct ccw_device *, int); extern int qdio_free(struct ccw_device *); extern int qdio_get_ssqd_desc(struct ccw_device *, struct qdio_ssqd_desc *); +extern int qdio_pnso_brinfo(struct subchannel_id schid, + int cnc, u16 *response, + void (*cb)(void *priv, enum qdio_brinfo_entry_type type, + void *entry), + void *priv); #endif /* __QDIO_H__ */ diff --git a/arch/s390/include/asm/sclp.h b/arch/s390/include/asm/sclp.h index 2f390956c7c1..abaca2275c7a 100644 --- a/arch/s390/include/asm/sclp.h +++ b/arch/s390/include/asm/sclp.h @@ -52,8 +52,9 @@ int sclp_chp_configure(struct chp_id chpid); int sclp_chp_deconfigure(struct chp_id chpid); int sclp_chp_read_info(struct sclp_chp_info *info); void sclp_get_ipl_info(struct sclp_ipl_info *info); -bool sclp_has_linemode(void); -bool sclp_has_vt220(void); +bool __init sclp_has_linemode(void); +bool __init sclp_has_vt220(void); +bool sclp_has_sprp(void); int sclp_pci_configure(u32 fid); int sclp_pci_deconfigure(u32 fid); int memcpy_hsa(void *dest, unsigned long src, size_t count, int mode); diff --git a/arch/s390/include/asm/sigp.h b/arch/s390/include/asm/sigp.h index 5a87d16d3e7c..d091aa1aaf11 100644 --- a/arch/s390/include/asm/sigp.h +++ b/arch/s390/include/asm/sigp.h @@ -5,6 +5,7 @@ #define SIGP_SENSE 1 #define SIGP_EXTERNAL_CALL 2 #define SIGP_EMERGENCY_SIGNAL 3 +#define SIGP_START 4 #define SIGP_STOP 5 #define SIGP_RESTART 6 #define SIGP_STOP_AND_STORE_STATUS 9 @@ -12,6 +13,7 @@ #define SIGP_SET_PREFIX 13 #define SIGP_STORE_STATUS_AT_ADDRESS 14 #define SIGP_SET_ARCHITECTURE 18 +#define SIGP_COND_EMERGENCY_SIGNAL 19 #define SIGP_SENSE_RUNNING 21 /* SIGP condition codes */ diff --git a/arch/s390/include/uapi/asm/hypfs.h b/arch/s390/include/uapi/asm/hypfs.h new file mode 100644 index 000000000000..37998b449531 --- /dev/null +++ b/arch/s390/include/uapi/asm/hypfs.h @@ -0,0 +1,25 @@ +/* + * IOCTL interface for hypfs + * + * Copyright IBM Corp. 2013 + * + * Author: Martin Schwidefsky <schwidefsky@de.ibm.com> + */ + +#ifndef _ASM_HYPFS_CTL_H +#define _ASM_HYPFS_CTL_H + +#include <linux/types.h> + +struct hypfs_diag304 { + __u32 args[2]; + __u64 data; + __u64 rc; +} __attribute__((packed)); + +#define HYPFS_IOCTL_MAGIC 0x10 + +#define HYPFS_DIAG304 \ + _IOWR(HYPFS_IOCTL_MAGIC, 0x20, struct hypfs_diag304) + +#endif diff --git a/arch/s390/include/uapi/asm/socket.h b/arch/s390/include/uapi/asm/socket.h index c286c2e868f0..e031332096d7 100644 --- a/arch/s390/include/uapi/asm/socket.h +++ b/arch/s390/include/uapi/asm/socket.h @@ -84,4 +84,6 @@ #define SO_MAX_PACING_RATE 47 +#define SO_BPF_EXTENSIONS 48 + #endif /* _ASM_SOCKET_H */ diff --git a/arch/s390/include/uapi/asm/statfs.h b/arch/s390/include/uapi/asm/statfs.h index a61d538756f2..471eb09184d4 100644 --- a/arch/s390/include/uapi/asm/statfs.h +++ b/arch/s390/include/uapi/asm/statfs.h @@ -35,11 +35,11 @@ struct statfs { struct statfs64 { unsigned int f_type; unsigned int f_bsize; - unsigned long f_blocks; - unsigned long f_bfree; - unsigned long f_bavail; - unsigned long f_files; - unsigned long f_ffree; + unsigned long long f_blocks; + unsigned long long f_bfree; + unsigned long long f_bavail; + unsigned long long f_files; + unsigned long long f_ffree; __kernel_fsid_t f_fsid; unsigned int f_namelen; unsigned int f_frsize; diff --git a/arch/s390/include/uapi/asm/unistd.h b/arch/s390/include/uapi/asm/unistd.h index 864f693c237f..5eb5c9ddb120 100644 --- a/arch/s390/include/uapi/asm/unistd.h +++ b/arch/s390/include/uapi/asm/unistd.h @@ -280,6 +280,8 @@ #define __NR_s390_runtime_instr 342 #define __NR_kcmp 343 #define __NR_finit_module 344 +#define __NR_sched_setattr 345 +#define __NR_sched_getattr 346 #define NR_syscalls 345 /* diff --git a/arch/s390/include/uapi/asm/zcrypt.h b/arch/s390/include/uapi/asm/zcrypt.h index e83fc116f5bf..f2b18eacaca8 100644 --- a/arch/s390/include/uapi/asm/zcrypt.h +++ b/arch/s390/include/uapi/asm/zcrypt.h @@ -154,6 +154,67 @@ struct ica_xcRB { unsigned short priority_window; unsigned int status; } __attribute__((packed)); + +/** + * struct ep11_cprb - EP11 connectivity programming request block + * @cprb_len: CPRB header length [0x0020] + * @cprb_ver_id: CPRB version id. [0x04] + * @pad_000: Alignment pad bytes + * @flags: Admin cmd [0x80] or functional cmd [0x00] + * @func_id: Function id / subtype [0x5434] + * @source_id: Source id [originator id] + * @target_id: Target id [usage/ctrl domain id] + * @ret_code: Return code + * @reserved1: Reserved + * @reserved2: Reserved + * @payload_len: Payload length + */ +struct ep11_cprb { + uint16_t cprb_len; + unsigned char cprb_ver_id; + unsigned char pad_000[2]; + unsigned char flags; + unsigned char func_id[2]; + uint32_t source_id; + uint32_t target_id; + uint32_t ret_code; + uint32_t reserved1; + uint32_t reserved2; + uint32_t payload_len; +} __attribute__((packed)); + +/** + * struct ep11_target_dev - EP11 target device list + * @ap_id: AP device id + * @dom_id: Usage domain id + */ +struct ep11_target_dev { + uint16_t ap_id; + uint16_t dom_id; +}; + +/** + * struct ep11_urb - EP11 user request block + * @targets_num: Number of target adapters + * @targets: Addr to target adapter list + * @weight: Level of request priority + * @req_no: Request id/number + * @req_len: Request length + * @req: Addr to request block + * @resp_len: Response length + * @resp: Addr to response block + */ +struct ep11_urb { + uint16_t targets_num; + uint64_t targets; + uint64_t weight; + uint64_t req_no; + uint64_t req_len; + uint64_t req; + uint64_t resp_len; + uint64_t resp; +} __attribute__((packed)); + #define AUTOSELECT ((unsigned int)0xFFFFFFFF) #define ZCRYPT_IOCTL_MAGIC 'z' @@ -183,6 +244,9 @@ struct ica_xcRB { * ZSECSENDCPRB * Send an arbitrary CPRB to a crypto card. * + * ZSENDEP11CPRB + * Send an arbitrary EP11 CPRB to an EP11 coprocessor crypto card. + * * Z90STAT_STATUS_MASK * Return an 64 element array of unsigned chars for the status of * all devices. @@ -256,6 +320,7 @@ struct ica_xcRB { #define ICARSAMODEXPO _IOC(_IOC_READ|_IOC_WRITE, ZCRYPT_IOCTL_MAGIC, 0x05, 0) #define ICARSACRT _IOC(_IOC_READ|_IOC_WRITE, ZCRYPT_IOCTL_MAGIC, 0x06, 0) #define ZSECSENDCPRB _IOC(_IOC_READ|_IOC_WRITE, ZCRYPT_IOCTL_MAGIC, 0x81, 0) +#define ZSENDEP11CPRB _IOC(_IOC_READ|_IOC_WRITE, ZCRYPT_IOCTL_MAGIC, 0x04, 0) /* New status calls */ #define Z90STAT_TOTALCOUNT _IOR(ZCRYPT_IOCTL_MAGIC, 0x40, int) diff --git a/arch/s390/kernel/Makefile b/arch/s390/kernel/Makefile index 2403303cfed7..1b3ac09c11b6 100644 --- a/arch/s390/kernel/Makefile +++ b/arch/s390/kernel/Makefile @@ -60,7 +60,8 @@ obj-$(CONFIG_FTRACE_SYSCALLS) += ftrace.o obj-$(CONFIG_CRASH_DUMP) += crash_dump.o ifdef CONFIG_64BIT -obj-$(CONFIG_PERF_EVENTS) += perf_event.o perf_cpum_cf.o +obj-$(CONFIG_PERF_EVENTS) += perf_event.o perf_cpum_cf.o perf_cpum_sf.o \ + perf_cpum_cf_events.o obj-y += runtime_instr.o cache.o endif diff --git a/arch/s390/kernel/compat_linux.c b/arch/s390/kernel/compat_linux.c index e030d2bdec1b..db02052bd137 100644 --- a/arch/s390/kernel/compat_linux.c +++ b/arch/s390/kernel/compat_linux.c @@ -286,8 +286,8 @@ asmlinkage long sys32_getegid16(void) } #ifdef CONFIG_SYSVIPC -COMPAT_SYSCALL_DEFINE5(s390_ipc, uint, call, int, first, unsigned long, second, - unsigned long, third, compat_uptr_t, ptr) +COMPAT_SYSCALL_DEFINE5(s390_ipc, uint, call, int, first, compat_ulong_t, second, + compat_ulong_t, third, compat_uptr_t, ptr) { if (call >> 16) /* hack for backward compatibility */ return -EINVAL; diff --git a/arch/s390/kernel/compat_signal.c b/arch/s390/kernel/compat_signal.c index 95e7ba0fbb7e..8b84bc373e94 100644 --- a/arch/s390/kernel/compat_signal.c +++ b/arch/s390/kernel/compat_signal.c @@ -412,8 +412,9 @@ static int setup_rt_frame32(int sig, struct k_sigaction *ka, siginfo_t *info, regs->gprs[14] = (__u64 __force) ka->sa.sa_restorer | PSW32_ADDR_AMODE; } else { regs->gprs[14] = (__u64 __force) frame->retcode | PSW32_ADDR_AMODE; - err |= __put_user(S390_SYSCALL_OPCODE | __NR_rt_sigreturn, - (u16 __force __user *)(frame->retcode)); + if (__put_user(S390_SYSCALL_OPCODE | __NR_rt_sigreturn, + (u16 __force __user *)(frame->retcode))) + goto give_sigsegv; } /* Set up backchain. */ diff --git a/arch/s390/kernel/compat_wrapper.S b/arch/s390/kernel/compat_wrapper.S index 9cb1b975b353..59c8efce1b99 100644 --- a/arch/s390/kernel/compat_wrapper.S +++ b/arch/s390/kernel/compat_wrapper.S @@ -1412,3 +1412,14 @@ ENTRY(sys_finit_module_wrapper) llgtr %r3,%r3 # const char __user * lgfr %r4,%r4 # int jg sys_finit_module + +ENTRY(sys_sched_setattr_wrapper) + lgfr %r2,%r2 # pid_t + llgtr %r3,%r3 # struct sched_attr __user * + jg sys_sched_setattr + +ENTRY(sys_sched_getattr_wrapper) + lgfr %r2,%r2 # pid_t + llgtr %r3,%r3 # const char __user * + llgfr %r3,%r3 # unsigned int + jg sys_sched_getattr diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S index e5b43c97a834..384e609b4711 100644 --- a/arch/s390/kernel/entry64.S +++ b/arch/s390/kernel/entry64.S @@ -74,7 +74,7 @@ _TIF_TRACE = (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | _TIF_SECCOMP | \ .endm .macro LPP newpp -#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +#if IS_ENABLED(CONFIG_KVM) tm __LC_MACHINE_FLAGS+6,0x20 # MACHINE_FLAG_LPP jz .+8 .insn s,0xb2800000,\newpp @@ -82,7 +82,7 @@ _TIF_TRACE = (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | _TIF_SECCOMP | \ .endm .macro HANDLE_SIE_INTERCEPT scratch,reason -#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +#if IS_ENABLED(CONFIG_KVM) tmhh %r8,0x0001 # interrupting from user ? jnz .+62 lgr \scratch,%r9 @@ -946,7 +946,7 @@ cleanup_idle_insn: .quad __critical_end - __critical_start -#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +#if IS_ENABLED(CONFIG_KVM) /* * sie64a calling convention: * %r2 pointer to sie control block @@ -975,7 +975,7 @@ sie_done: lctlg %c1,%c1,__LC_USER_ASCE # load primary asce # some program checks are suppressing. C code (e.g. do_protection_exception) # will rewind the PSW by the ILC, which is 4 bytes in case of SIE. Other -# instructions beween sie64a and sie_done should not cause program +# instructions between sie64a and sie_done should not cause program # interrupts. So lets use a nop (47 00 00 00) as a landing pad. # See also HANDLE_SIE_INTERCEPT rewind_pad: diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c index 1105502bf6e9..f51214c04858 100644 --- a/arch/s390/kernel/perf_cpum_cf.c +++ b/arch/s390/kernel/perf_cpum_cf.c @@ -680,6 +680,7 @@ static int __init cpumf_pmu_init(void) goto out; } + cpumf_pmu.attr_groups = cpumf_cf_event_group(); rc = perf_pmu_register(&cpumf_pmu, "cpum_cf", PERF_TYPE_RAW); if (rc) { pr_err("Registering the cpum_cf PMU failed with rc=%i\n", rc); diff --git a/arch/s390/kernel/perf_cpum_cf_events.c b/arch/s390/kernel/perf_cpum_cf_events.c new file mode 100644 index 000000000000..4554a4bae39e --- /dev/null +++ b/arch/s390/kernel/perf_cpum_cf_events.c @@ -0,0 +1,322 @@ +/* + * Perf PMU sysfs events attributes for available CPU-measurement counters + * + */ + +#include <linux/slab.h> +#include <linux/perf_event.h> + + +/* BEGIN: CPUM_CF COUNTER DEFINITIONS =================================== */ + +CPUMF_EVENT_ATTR(cf, CPU_CYCLES, 0x0000); +CPUMF_EVENT_ATTR(cf, INSTRUCTIONS, 0x0001); +CPUMF_EVENT_ATTR(cf, L1I_DIR_WRITES, 0x0002); +CPUMF_EVENT_ATTR(cf, L1I_PENALTY_CYCLES, 0x0003); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_CPU_CYCLES, 0x0020); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_INSTRUCTIONS, 0x0021); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_L1I_DIR_WRITES, 0x0022); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_L1I_PENALTY_CYCLES, 0x0023); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_L1D_DIR_WRITES, 0x0024); +CPUMF_EVENT_ATTR(cf, PROBLEM_STATE_L1D_PENALTY_CYCLES, 0x0025); +CPUMF_EVENT_ATTR(cf, L1D_DIR_WRITES, 0x0004); +CPUMF_EVENT_ATTR(cf, L1D_PENALTY_CYCLES, 0x0005); +CPUMF_EVENT_ATTR(cf, PRNG_FUNCTIONS, 0x0040); +CPUMF_EVENT_ATTR(cf, PRNG_CYCLES, 0x0041); +CPUMF_EVENT_ATTR(cf, PRNG_BLOCKED_FUNCTIONS, 0x0042); +CPUMF_EVENT_ATTR(cf, PRNG_BLOCKED_CYCLES, 0x0043); +CPUMF_EVENT_ATTR(cf, SHA_FUNCTIONS, 0x0044); +CPUMF_EVENT_ATTR(cf, SHA_CYCLES, 0x0045); +CPUMF_EVENT_ATTR(cf, SHA_BLOCKED_FUNCTIONS, 0x0046); +CPUMF_EVENT_ATTR(cf, SHA_BLOCKED_CYCLES, 0x0047); +CPUMF_EVENT_ATTR(cf, DEA_FUNCTIONS, 0x0048); +CPUMF_EVENT_ATTR(cf, DEA_CYCLES, 0x0049); +CPUMF_EVENT_ATTR(cf, DEA_BLOCKED_FUNCTIONS, 0x004a); +CPUMF_EVENT_ATTR(cf, DEA_BLOCKED_CYCLES, 0x004b); +CPUMF_EVENT_ATTR(cf, AES_FUNCTIONS, 0x004c); +CPUMF_EVENT_ATTR(cf, AES_CYCLES, 0x004d); +CPUMF_EVENT_ATTR(cf, AES_BLOCKED_FUNCTIONS, 0x004e); +CPUMF_EVENT_ATTR(cf, AES_BLOCKED_CYCLES, 0x004f); +CPUMF_EVENT_ATTR(cf_z10, L1I_L2_SOURCED_WRITES, 0x0080); +CPUMF_EVENT_ATTR(cf_z10, L1D_L2_SOURCED_WRITES, 0x0081); +CPUMF_EVENT_ATTR(cf_z10, L1I_L3_LOCAL_WRITES, 0x0082); +CPUMF_EVENT_ATTR(cf_z10, L1D_L3_LOCAL_WRITES, 0x0083); +CPUMF_EVENT_ATTR(cf_z10, L1I_L3_REMOTE_WRITES, 0x0084); +CPUMF_EVENT_ATTR(cf_z10, L1D_L3_REMOTE_WRITES, 0x0085); +CPUMF_EVENT_ATTR(cf_z10, L1D_LMEM_SOURCED_WRITES, 0x0086); +CPUMF_EVENT_ATTR(cf_z10, L1I_LMEM_SOURCED_WRITES, 0x0087); +CPUMF_EVENT_ATTR(cf_z10, L1D_RO_EXCL_WRITES, 0x0088); +CPUMF_EVENT_ATTR(cf_z10, L1I_CACHELINE_INVALIDATES, 0x0089); +CPUMF_EVENT_ATTR(cf_z10, ITLB1_WRITES, 0x008a); +CPUMF_EVENT_ATTR(cf_z10, DTLB1_WRITES, 0x008b); +CPUMF_EVENT_ATTR(cf_z10, TLB2_PTE_WRITES, 0x008c); +CPUMF_EVENT_ATTR(cf_z10, TLB2_CRSTE_WRITES, 0x008d); +CPUMF_EVENT_ATTR(cf_z10, TLB2_CRSTE_HPAGE_WRITES, 0x008e); +CPUMF_EVENT_ATTR(cf_z10, ITLB1_MISSES, 0x0091); +CPUMF_EVENT_ATTR(cf_z10, DTLB1_MISSES, 0x0092); +CPUMF_EVENT_ATTR(cf_z10, L2C_STORES_SENT, 0x0093); +CPUMF_EVENT_ATTR(cf_z196, L1D_L2_SOURCED_WRITES, 0x0080); +CPUMF_EVENT_ATTR(cf_z196, L1I_L2_SOURCED_WRITES, 0x0081); +CPUMF_EVENT_ATTR(cf_z196, DTLB1_MISSES, 0x0082); +CPUMF_EVENT_ATTR(cf_z196, ITLB1_MISSES, 0x0083); +CPUMF_EVENT_ATTR(cf_z196, L2C_STORES_SENT, 0x0085); +CPUMF_EVENT_ATTR(cf_z196, L1D_OFFBOOK_L3_SOURCED_WRITES, 0x0086); +CPUMF_EVENT_ATTR(cf_z196, L1D_ONBOOK_L4_SOURCED_WRITES, 0x0087); +CPUMF_EVENT_ATTR(cf_z196, L1I_ONBOOK_L4_SOURCED_WRITES, 0x0088); +CPUMF_EVENT_ATTR(cf_z196, L1D_RO_EXCL_WRITES, 0x0089); +CPUMF_EVENT_ATTR(cf_z196, L1D_OFFBOOK_L4_SOURCED_WRITES, 0x008a); +CPUMF_EVENT_ATTR(cf_z196, L1I_OFFBOOK_L4_SOURCED_WRITES, 0x008b); +CPUMF_EVENT_ATTR(cf_z196, DTLB1_HPAGE_WRITES, 0x008c); +CPUMF_EVENT_ATTR(cf_z196, L1D_LMEM_SOURCED_WRITES, 0x008d); +CPUMF_EVENT_ATTR(cf_z196, L1I_LMEM_SOURCED_WRITES, 0x008e); +CPUMF_EVENT_ATTR(cf_z196, L1I_OFFBOOK_L3_SOURCED_WRITES, 0x008f); +CPUMF_EVENT_ATTR(cf_z196, DTLB1_WRITES, 0x0090); +CPUMF_EVENT_ATTR(cf_z196, ITLB1_WRITES, 0x0091); +CPUMF_EVENT_ATTR(cf_z196, TLB2_PTE_WRITES, 0x0092); +CPUMF_EVENT_ATTR(cf_z196, TLB2_CRSTE_HPAGE_WRITES, 0x0093); +CPUMF_EVENT_ATTR(cf_z196, TLB2_CRSTE_WRITES, 0x0094); +CPUMF_EVENT_ATTR(cf_z196, L1D_ONCHIP_L3_SOURCED_WRITES, 0x0096); +CPUMF_EVENT_ATTR(cf_z196, L1D_OFFCHIP_L3_SOURCED_WRITES, 0x0098); +CPUMF_EVENT_ATTR(cf_z196, L1I_ONCHIP_L3_SOURCED_WRITES, 0x0099); +CPUMF_EVENT_ATTR(cf_z196, L1I_OFFCHIP_L3_SOURCED_WRITES, 0x009b); +CPUMF_EVENT_ATTR(cf_zec12, DTLB1_MISSES, 0x0080); +CPUMF_EVENT_ATTR(cf_zec12, ITLB1_MISSES, 0x0081); +CPUMF_EVENT_ATTR(cf_zec12, L1D_L2I_SOURCED_WRITES, 0x0082); +CPUMF_EVENT_ATTR(cf_zec12, L1I_L2I_SOURCED_WRITES, 0x0083); +CPUMF_EVENT_ATTR(cf_zec12, L1D_L2D_SOURCED_WRITES, 0x0084); +CPUMF_EVENT_ATTR(cf_zec12, DTLB1_WRITES, 0x0085); +CPUMF_EVENT_ATTR(cf_zec12, L1D_LMEM_SOURCED_WRITES, 0x0087); +CPUMF_EVENT_ATTR(cf_zec12, L1I_LMEM_SOURCED_WRITES, 0x0089); +CPUMF_EVENT_ATTR(cf_zec12, L1D_RO_EXCL_WRITES, 0x008a); +CPUMF_EVENT_ATTR(cf_zec12, DTLB1_HPAGE_WRITES, 0x008b); +CPUMF_EVENT_ATTR(cf_zec12, ITLB1_WRITES, 0x008c); +CPUMF_EVENT_ATTR(cf_zec12, TLB2_PTE_WRITES, 0x008d); +CPUMF_EVENT_ATTR(cf_zec12, TLB2_CRSTE_HPAGE_WRITES, 0x008e); +CPUMF_EVENT_ATTR(cf_zec12, TLB2_CRSTE_WRITES, 0x008f); +CPUMF_EVENT_ATTR(cf_zec12, L1D_ONCHIP_L3_SOURCED_WRITES, 0x0090); +CPUMF_EVENT_ATTR(cf_zec12, L1D_OFFCHIP_L3_SOURCED_WRITES, 0x0091); +CPUMF_EVENT_ATTR(cf_zec12, L1D_OFFBOOK_L3_SOURCED_WRITES, 0x0092); +CPUMF_EVENT_ATTR(cf_zec12, L1D_ONBOOK_L4_SOURCED_WRITES, 0x0093); +CPUMF_EVENT_ATTR(cf_zec12, L1D_OFFBOOK_L4_SOURCED_WRITES, 0x0094); +CPUMF_EVENT_ATTR(cf_zec12, TX_NC_TEND, 0x0095); +CPUMF_EVENT_ATTR(cf_zec12, L1D_ONCHIP_L3_SOURCED_WRITES_IV, 0x0096); +CPUMF_EVENT_ATTR(cf_zec12, L1D_OFFCHIP_L3_SOURCED_WRITES_IV, 0x0097); +CPUMF_EVENT_ATTR(cf_zec12, L1D_OFFBOOK_L3_SOURCED_WRITES_IV, 0x0098); +CPUMF_EVENT_ATTR(cf_zec12, L1I_ONCHIP_L3_SOURCED_WRITES, 0x0099); +CPUMF_EVENT_ATTR(cf_zec12, L1I_OFFCHIP_L3_SOURCED_WRITES, 0x009a); +CPUMF_EVENT_ATTR(cf_zec12, L1I_OFFBOOK_L3_SOURCED_WRITES, 0x009b); +CPUMF_EVENT_ATTR(cf_zec12, L1I_ONBOOK_L4_SOURCED_WRITES, 0x009c); +CPUMF_EVENT_ATTR(cf_zec12, L1I_OFFBOOK_L4_SOURCED_WRITES, 0x009d); +CPUMF_EVENT_ATTR(cf_zec12, TX_C_TEND, 0x009e); +CPUMF_EVENT_ATTR(cf_zec12, L1I_ONCHIP_L3_SOURCED_WRITES_IV, 0x009f); +CPUMF_EVENT_ATTR(cf_zec12, L1I_OFFCHIP_L3_SOURCED_WRITES_IV, 0x00a0); +CPUMF_EVENT_ATTR(cf_zec12, L1I_OFFBOOK_L3_SOURCED_WRITES_IV, 0x00a1); +CPUMF_EVENT_ATTR(cf_zec12, TX_NC_TABORT, 0x00b1); +CPUMF_EVENT_ATTR(cf_zec12, TX_C_TABORT_NO_SPECIAL, 0x00b2); +CPUMF_EVENT_ATTR(cf_zec12, TX_C_TABORT_SPECIAL, 0x00b3); + +static struct attribute *cpumcf_pmu_event_attr[] = { + CPUMF_EVENT_PTR(cf, CPU_CYCLES), + CPUMF_EVENT_PTR(cf, INSTRUCTIONS), + CPUMF_EVENT_PTR(cf, L1I_DIR_WRITES), + CPUMF_EVENT_PTR(cf, L1I_PENALTY_CYCLES), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_CPU_CYCLES), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_INSTRUCTIONS), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_L1I_DIR_WRITES), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_L1I_PENALTY_CYCLES), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_L1D_DIR_WRITES), + CPUMF_EVENT_PTR(cf, PROBLEM_STATE_L1D_PENALTY_CYCLES), + CPUMF_EVENT_PTR(cf, L1D_DIR_WRITES), + CPUMF_EVENT_PTR(cf, L1D_PENALTY_CYCLES), + CPUMF_EVENT_PTR(cf, PRNG_FUNCTIONS), + CPUMF_EVENT_PTR(cf, PRNG_CYCLES), + CPUMF_EVENT_PTR(cf, PRNG_BLOCKED_FUNCTIONS), + CPUMF_EVENT_PTR(cf, PRNG_BLOCKED_CYCLES), + CPUMF_EVENT_PTR(cf, SHA_FUNCTIONS), + CPUMF_EVENT_PTR(cf, SHA_CYCLES), + CPUMF_EVENT_PTR(cf, SHA_BLOCKED_FUNCTIONS), + CPUMF_EVENT_PTR(cf, SHA_BLOCKED_CYCLES), + CPUMF_EVENT_PTR(cf, DEA_FUNCTIONS), + CPUMF_EVENT_PTR(cf, DEA_CYCLES), + CPUMF_EVENT_PTR(cf, DEA_BLOCKED_FUNCTIONS), + CPUMF_EVENT_PTR(cf, DEA_BLOCKED_CYCLES), + CPUMF_EVENT_PTR(cf, AES_FUNCTIONS), + CPUMF_EVENT_PTR(cf, AES_CYCLES), + CPUMF_EVENT_PTR(cf, AES_BLOCKED_FUNCTIONS), + CPUMF_EVENT_PTR(cf, AES_BLOCKED_CYCLES), + NULL, +}; + +static struct attribute *cpumcf_z10_pmu_event_attr[] __initdata = { + CPUMF_EVENT_PTR(cf_z10, L1I_L2_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1D_L2_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1I_L3_LOCAL_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1D_L3_LOCAL_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1I_L3_REMOTE_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1D_L3_REMOTE_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1D_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1I_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1D_RO_EXCL_WRITES), + CPUMF_EVENT_PTR(cf_z10, L1I_CACHELINE_INVALIDATES), + CPUMF_EVENT_PTR(cf_z10, ITLB1_WRITES), + CPUMF_EVENT_PTR(cf_z10, DTLB1_WRITES), + CPUMF_EVENT_PTR(cf_z10, TLB2_PTE_WRITES), + CPUMF_EVENT_PTR(cf_z10, TLB2_CRSTE_WRITES), + CPUMF_EVENT_PTR(cf_z10, TLB2_CRSTE_HPAGE_WRITES), + CPUMF_EVENT_PTR(cf_z10, ITLB1_MISSES), + CPUMF_EVENT_PTR(cf_z10, DTLB1_MISSES), + CPUMF_EVENT_PTR(cf_z10, L2C_STORES_SENT), + NULL, +}; + +static struct attribute *cpumcf_z196_pmu_event_attr[] __initdata = { + CPUMF_EVENT_PTR(cf_z196, L1D_L2_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_L2_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, DTLB1_MISSES), + CPUMF_EVENT_PTR(cf_z196, ITLB1_MISSES), + CPUMF_EVENT_PTR(cf_z196, L2C_STORES_SENT), + CPUMF_EVENT_PTR(cf_z196, L1D_OFFBOOK_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_ONBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_ONBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_RO_EXCL_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_OFFBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_OFFBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, DTLB1_HPAGE_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_OFFBOOK_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, DTLB1_WRITES), + CPUMF_EVENT_PTR(cf_z196, ITLB1_WRITES), + CPUMF_EVENT_PTR(cf_z196, TLB2_PTE_WRITES), + CPUMF_EVENT_PTR(cf_z196, TLB2_CRSTE_HPAGE_WRITES), + CPUMF_EVENT_PTR(cf_z196, TLB2_CRSTE_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_ONCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1D_OFFCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_ONCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_z196, L1I_OFFCHIP_L3_SOURCED_WRITES), + NULL, +}; + +static struct attribute *cpumcf_zec12_pmu_event_attr[] __initdata = { + CPUMF_EVENT_PTR(cf_zec12, DTLB1_MISSES), + CPUMF_EVENT_PTR(cf_zec12, ITLB1_MISSES), + CPUMF_EVENT_PTR(cf_zec12, L1D_L2I_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_L2I_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_L2D_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, DTLB1_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_LMEM_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_RO_EXCL_WRITES), + CPUMF_EVENT_PTR(cf_zec12, DTLB1_HPAGE_WRITES), + CPUMF_EVENT_PTR(cf_zec12, ITLB1_WRITES), + CPUMF_EVENT_PTR(cf_zec12, TLB2_PTE_WRITES), + CPUMF_EVENT_PTR(cf_zec12, TLB2_CRSTE_HPAGE_WRITES), + CPUMF_EVENT_PTR(cf_zec12, TLB2_CRSTE_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_ONCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_OFFCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_OFFBOOK_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_ONBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1D_OFFBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, TX_NC_TEND), + CPUMF_EVENT_PTR(cf_zec12, L1D_ONCHIP_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, L1D_OFFCHIP_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, L1D_OFFBOOK_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, L1I_ONCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_OFFCHIP_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_OFFBOOK_L3_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_ONBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, L1I_OFFBOOK_L4_SOURCED_WRITES), + CPUMF_EVENT_PTR(cf_zec12, TX_C_TEND), + CPUMF_EVENT_PTR(cf_zec12, L1I_ONCHIP_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, L1I_OFFCHIP_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, L1I_OFFBOOK_L3_SOURCED_WRITES_IV), + CPUMF_EVENT_PTR(cf_zec12, TX_NC_TABORT), + CPUMF_EVENT_PTR(cf_zec12, TX_C_TABORT_NO_SPECIAL), + CPUMF_EVENT_PTR(cf_zec12, TX_C_TABORT_SPECIAL), + NULL, +}; + +/* END: CPUM_CF COUNTER DEFINITIONS ===================================== */ + +static struct attribute_group cpumsf_pmu_events_group = { + .name = "events", + .attrs = cpumcf_pmu_event_attr, +}; + +PMU_FORMAT_ATTR(event, "config:0-63"); + +static struct attribute *cpumsf_pmu_format_attr[] = { + &format_attr_event.attr, + NULL, +}; + +static struct attribute_group cpumsf_pmu_format_group = { + .name = "format", + .attrs = cpumsf_pmu_format_attr, +}; + +static const struct attribute_group *cpumsf_pmu_attr_groups[] = { + &cpumsf_pmu_events_group, + &cpumsf_pmu_format_group, + NULL, +}; + + +static __init struct attribute **merge_attr(struct attribute **a, + struct attribute **b) +{ + struct attribute **new; + int j, i; + + for (j = 0; a[j]; j++) + ; + for (i = 0; b[i]; i++) + j++; + j++; + + new = kmalloc(sizeof(struct attribute *) * j, GFP_KERNEL); + if (!new) + return NULL; + j = 0; + for (i = 0; a[i]; i++) + new[j++] = a[i]; + for (i = 0; b[i]; i++) + new[j++] = b[i]; + new[j] = NULL; + + return new; +} + +__init const struct attribute_group **cpumf_cf_event_group(void) +{ + struct attribute **combined, **model; + struct cpuid cpu_id; + + get_cpu_id(&cpu_id); + switch (cpu_id.machine) { + case 0x2097: + case 0x2098: + model = cpumcf_z10_pmu_event_attr; + break; + case 0x2817: + case 0x2818: + model = cpumcf_z196_pmu_event_attr; + break; + case 0x2827: + case 0x2828: + model = cpumcf_zec12_pmu_event_attr; + break; + default: + model = NULL; + break; + }; + + if (!model) + goto out; + + combined = merge_attr(cpumcf_pmu_event_attr, model); + if (combined) + cpumsf_pmu_events_group.attrs = combined; +out: + return cpumsf_pmu_attr_groups; +} diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c new file mode 100644 index 000000000000..6c0d29827cb6 --- /dev/null +++ b/arch/s390/kernel/perf_cpum_sf.c @@ -0,0 +1,1641 @@ +/* + * Performance event support for the System z CPU-measurement Sampling Facility + * + * Copyright IBM Corp. 2013 + * Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License (version 2 only) + * as published by the Free Software Foundation. + */ +#define KMSG_COMPONENT "cpum_sf" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/kernel.h> +#include <linux/kernel_stat.h> +#include <linux/perf_event.h> +#include <linux/percpu.h> +#include <linux/notifier.h> +#include <linux/export.h> +#include <linux/slab.h> +#include <linux/mm.h> +#include <linux/moduleparam.h> +#include <asm/cpu_mf.h> +#include <asm/irq.h> +#include <asm/debug.h> +#include <asm/timex.h> + +/* Minimum number of sample-data-block-tables: + * At least one table is required for the sampling buffer structure. + * A single table contains up to 511 pointers to sample-data-blocks. + */ +#define CPUM_SF_MIN_SDBT 1 + +/* Number of sample-data-blocks per sample-data-block-table (SDBT): + * A table contains SDB pointers (8 bytes) and one table-link entry + * that points to the origin of the next SDBT. + */ +#define CPUM_SF_SDB_PER_TABLE ((PAGE_SIZE - 8) / 8) + +/* Maximum page offset for an SDBT table-link entry: + * If this page offset is reached, a table-link entry to the next SDBT + * must be added. + */ +#define CPUM_SF_SDBT_TL_OFFSET (CPUM_SF_SDB_PER_TABLE * 8) +static inline int require_table_link(const void *sdbt) +{ + return ((unsigned long) sdbt & ~PAGE_MASK) == CPUM_SF_SDBT_TL_OFFSET; +} + +/* Minimum and maximum sampling buffer sizes: + * + * This number represents the maximum size of the sampling buffer taking + * the number of sample-data-block-tables into account. Note that these + * numbers apply to the basic-sampling function only. + * The maximum number of SDBs is increased by CPUM_SF_SDB_DIAG_FACTOR if + * the diagnostic-sampling function is active. + * + * Sampling buffer size Buffer characteristics + * --------------------------------------------------- + * 64KB == 16 pages (4KB per page) + * 1 page for SDB-tables + * 15 pages for SDBs + * + * 32MB == 8192 pages (4KB per page) + * 16 pages for SDB-tables + * 8176 pages for SDBs + */ +static unsigned long __read_mostly CPUM_SF_MIN_SDB = 15; +static unsigned long __read_mostly CPUM_SF_MAX_SDB = 8176; +static unsigned long __read_mostly CPUM_SF_SDB_DIAG_FACTOR = 1; + +struct sf_buffer { + unsigned long *sdbt; /* Sample-data-block-table origin */ + /* buffer characteristics (required for buffer increments) */ + unsigned long num_sdb; /* Number of sample-data-blocks */ + unsigned long num_sdbt; /* Number of sample-data-block-tables */ + unsigned long *tail; /* last sample-data-block-table */ +}; + +struct cpu_hw_sf { + /* CPU-measurement sampling information block */ + struct hws_qsi_info_block qsi; + /* CPU-measurement sampling control block */ + struct hws_lsctl_request_block lsctl; + struct sf_buffer sfb; /* Sampling buffer */ + unsigned int flags; /* Status flags */ + struct perf_event *event; /* Scheduled perf event */ +}; +static DEFINE_PER_CPU(struct cpu_hw_sf, cpu_hw_sf); + +/* Debug feature */ +static debug_info_t *sfdbg; + +/* + * sf_disable() - Switch off sampling facility + */ +static int sf_disable(void) +{ + struct hws_lsctl_request_block sreq; + + memset(&sreq, 0, sizeof(sreq)); + return lsctl(&sreq); +} + +/* + * sf_buffer_available() - Check for an allocated sampling buffer + */ +static int sf_buffer_available(struct cpu_hw_sf *cpuhw) +{ + return !!cpuhw->sfb.sdbt; +} + +/* + * deallocate sampling facility buffer + */ +static void free_sampling_buffer(struct sf_buffer *sfb) +{ + unsigned long *sdbt, *curr; + + if (!sfb->sdbt) + return; + + sdbt = sfb->sdbt; + curr = sdbt; + + /* Free the SDBT after all SDBs are processed... */ + while (1) { + if (!*curr || !sdbt) + break; + + /* Process table-link entries */ + if (is_link_entry(curr)) { + curr = get_next_sdbt(curr); + if (sdbt) + free_page((unsigned long) sdbt); + + /* If the origin is reached, sampling buffer is freed */ + if (curr == sfb->sdbt) + break; + else + sdbt = curr; + } else { + /* Process SDB pointer */ + if (*curr) { + free_page(*curr); + curr++; + } + } + } + + debug_sprintf_event(sfdbg, 5, + "free_sampling_buffer: freed sdbt=%p\n", sfb->sdbt); + memset(sfb, 0, sizeof(*sfb)); +} + +static int alloc_sample_data_block(unsigned long *sdbt, gfp_t gfp_flags) +{ + unsigned long sdb, *trailer; + + /* Allocate and initialize sample-data-block */ + sdb = get_zeroed_page(gfp_flags); + if (!sdb) + return -ENOMEM; + trailer = trailer_entry_ptr(sdb); + *trailer = SDB_TE_ALERT_REQ_MASK; + + /* Link SDB into the sample-data-block-table */ + *sdbt = sdb; + + return 0; +} + +/* + * realloc_sampling_buffer() - extend sampler memory + * + * Allocates new sample-data-blocks and adds them to the specified sampling + * buffer memory. + * + * Important: This modifies the sampling buffer and must be called when the + * sampling facility is disabled. + * + * Returns zero on success, non-zero otherwise. + */ +static int realloc_sampling_buffer(struct sf_buffer *sfb, + unsigned long num_sdb, gfp_t gfp_flags) +{ + int i, rc; + unsigned long *new, *tail; + + if (!sfb->sdbt || !sfb->tail) + return -EINVAL; + + if (!is_link_entry(sfb->tail)) + return -EINVAL; + + /* Append to the existing sampling buffer, overwriting the table-link + * register. + * The tail variables always points to the "tail" (last and table-link) + * entry in an SDB-table. + */ + tail = sfb->tail; + + /* Do a sanity check whether the table-link entry points to + * the sampling buffer origin. + */ + if (sfb->sdbt != get_next_sdbt(tail)) { + debug_sprintf_event(sfdbg, 3, "realloc_sampling_buffer: " + "sampling buffer is not linked: origin=%p" + "tail=%p\n", + (void *) sfb->sdbt, (void *) tail); + return -EINVAL; + } + + /* Allocate remaining SDBs */ + rc = 0; + for (i = 0; i < num_sdb; i++) { + /* Allocate a new SDB-table if it is full. */ + if (require_table_link(tail)) { + new = (unsigned long *) get_zeroed_page(gfp_flags); + if (!new) { + rc = -ENOMEM; + break; + } + sfb->num_sdbt++; + /* Link current page to tail of chain */ + *tail = (unsigned long)(void *) new + 1; + tail = new; + } + + /* Allocate a new sample-data-block. + * If there is not enough memory, stop the realloc process + * and simply use what was allocated. If this is a temporary + * issue, a new realloc call (if required) might succeed. + */ + rc = alloc_sample_data_block(tail, gfp_flags); + if (rc) + break; + sfb->num_sdb++; + tail++; + } + + /* Link sampling buffer to its origin */ + *tail = (unsigned long) sfb->sdbt + 1; + sfb->tail = tail; + + debug_sprintf_event(sfdbg, 4, "realloc_sampling_buffer: new buffer" + " settings: sdbt=%lu sdb=%lu\n", + sfb->num_sdbt, sfb->num_sdb); + return rc; +} + +/* + * allocate_sampling_buffer() - allocate sampler memory + * + * Allocates and initializes a sampling buffer structure using the + * specified number of sample-data-blocks (SDB). For each allocation, + * a 4K page is used. The number of sample-data-block-tables (SDBT) + * are calculated from SDBs. + * Also set the ALERT_REQ mask in each SDBs trailer. + * + * Returns zero on success, non-zero otherwise. + */ +static int alloc_sampling_buffer(struct sf_buffer *sfb, unsigned long num_sdb) +{ + int rc; + + if (sfb->sdbt) + return -EINVAL; + + /* Allocate the sample-data-block-table origin */ + sfb->sdbt = (unsigned long *) get_zeroed_page(GFP_KERNEL); + if (!sfb->sdbt) + return -ENOMEM; + sfb->num_sdb = 0; + sfb->num_sdbt = 1; + + /* Link the table origin to point to itself to prepare for + * realloc_sampling_buffer() invocation. + */ + sfb->tail = sfb->sdbt; + *sfb->tail = (unsigned long)(void *) sfb->sdbt + 1; + + /* Allocate requested number of sample-data-blocks */ + rc = realloc_sampling_buffer(sfb, num_sdb, GFP_KERNEL); + if (rc) { + free_sampling_buffer(sfb); + debug_sprintf_event(sfdbg, 4, "alloc_sampling_buffer: " + "realloc_sampling_buffer failed with rc=%i\n", rc); + } else + debug_sprintf_event(sfdbg, 4, + "alloc_sampling_buffer: tear=%p dear=%p\n", + sfb->sdbt, (void *) *sfb->sdbt); + return rc; +} + +static void sfb_set_limits(unsigned long min, unsigned long max) +{ + struct hws_qsi_info_block si; + + CPUM_SF_MIN_SDB = min; + CPUM_SF_MAX_SDB = max; + + memset(&si, 0, sizeof(si)); + if (!qsi(&si)) + CPUM_SF_SDB_DIAG_FACTOR = DIV_ROUND_UP(si.dsdes, si.bsdes); +} + +static unsigned long sfb_max_limit(struct hw_perf_event *hwc) +{ + return SAMPL_DIAG_MODE(hwc) ? CPUM_SF_MAX_SDB * CPUM_SF_SDB_DIAG_FACTOR + : CPUM_SF_MAX_SDB; +} + +static unsigned long sfb_pending_allocs(struct sf_buffer *sfb, + struct hw_perf_event *hwc) +{ + if (!sfb->sdbt) + return SFB_ALLOC_REG(hwc); + if (SFB_ALLOC_REG(hwc) > sfb->num_sdb) + return SFB_ALLOC_REG(hwc) - sfb->num_sdb; + return 0; +} + +static int sfb_has_pending_allocs(struct sf_buffer *sfb, + struct hw_perf_event *hwc) +{ + return sfb_pending_allocs(sfb, hwc) > 0; +} + +static void sfb_account_allocs(unsigned long num, struct hw_perf_event *hwc) +{ + /* Limit the number of SDBs to not exceed the maximum */ + num = min_t(unsigned long, num, sfb_max_limit(hwc) - SFB_ALLOC_REG(hwc)); + if (num) + SFB_ALLOC_REG(hwc) += num; +} + +static void sfb_init_allocs(unsigned long num, struct hw_perf_event *hwc) +{ + SFB_ALLOC_REG(hwc) = 0; + sfb_account_allocs(num, hwc); +} + +static size_t event_sample_size(struct hw_perf_event *hwc) +{ + struct sf_raw_sample *sfr = (struct sf_raw_sample *) RAWSAMPLE_REG(hwc); + size_t sample_size; + + /* The sample size depends on the sampling function: The basic-sampling + * function must be always enabled, diagnostic-sampling function is + * optional. + */ + sample_size = sfr->bsdes; + if (SAMPL_DIAG_MODE(hwc)) + sample_size += sfr->dsdes; + + return sample_size; +} + +static void deallocate_buffers(struct cpu_hw_sf *cpuhw) +{ + if (cpuhw->sfb.sdbt) + free_sampling_buffer(&cpuhw->sfb); +} + +static int allocate_buffers(struct cpu_hw_sf *cpuhw, struct hw_perf_event *hwc) +{ + unsigned long n_sdb, freq, factor; + size_t sfr_size, sample_size; + struct sf_raw_sample *sfr; + + /* Allocate raw sample buffer + * + * The raw sample buffer is used to temporarily store sampling data + * entries for perf raw sample processing. The buffer size mainly + * depends on the size of diagnostic-sampling data entries which is + * machine-specific. The exact size calculation includes: + * 1. The first 4 bytes of diagnostic-sampling data entries are + * already reflected in the sf_raw_sample structure. Subtract + * these bytes. + * 2. The perf raw sample data must be 8-byte aligned (u64) and + * perf's internal data size must be considered too. So add + * an additional u32 for correct alignment and subtract before + * allocating the buffer. + * 3. Store the raw sample buffer pointer in the perf event + * hardware structure. + */ + sfr_size = ALIGN((sizeof(*sfr) - sizeof(sfr->diag) + cpuhw->qsi.dsdes) + + sizeof(u32), sizeof(u64)); + sfr_size -= sizeof(u32); + sfr = kzalloc(sfr_size, GFP_KERNEL); + if (!sfr) + return -ENOMEM; + sfr->size = sfr_size; + sfr->bsdes = cpuhw->qsi.bsdes; + sfr->dsdes = cpuhw->qsi.dsdes; + RAWSAMPLE_REG(hwc) = (unsigned long) sfr; + + /* Calculate sampling buffers using 4K pages + * + * 1. Determine the sample data size which depends on the used + * sampling functions, for example, basic-sampling or + * basic-sampling with diagnostic-sampling. + * + * 2. Use the sampling frequency as input. The sampling buffer is + * designed for almost one second. This can be adjusted through + * the "factor" variable. + * In any case, alloc_sampling_buffer() sets the Alert Request + * Control indicator to trigger a measurement-alert to harvest + * sample-data-blocks (sdb). + * + * 3. Compute the number of sample-data-blocks and ensure a minimum + * of CPUM_SF_MIN_SDB. Also ensure the upper limit does not + * exceed a "calculated" maximum. The symbolic maximum is + * designed for basic-sampling only and needs to be increased if + * diagnostic-sampling is active. + * See also the remarks for these symbolic constants. + * + * 4. Compute the number of sample-data-block-tables (SDBT) and + * ensure a minimum of CPUM_SF_MIN_SDBT (one table can manage up + * to 511 SDBs). + */ + sample_size = event_sample_size(hwc); + freq = sample_rate_to_freq(&cpuhw->qsi, SAMPL_RATE(hwc)); + factor = 1; + n_sdb = DIV_ROUND_UP(freq, factor * ((PAGE_SIZE-64) / sample_size)); + if (n_sdb < CPUM_SF_MIN_SDB) + n_sdb = CPUM_SF_MIN_SDB; + + /* If there is already a sampling buffer allocated, it is very likely + * that the sampling facility is enabled too. If the event to be + * initialized requires a greater sampling buffer, the allocation must + * be postponed. Changing the sampling buffer requires the sampling + * facility to be in the disabled state. So, account the number of + * required SDBs and let cpumsf_pmu_enable() resize the buffer just + * before the event is started. + */ + sfb_init_allocs(n_sdb, hwc); + if (sf_buffer_available(cpuhw)) + return 0; + + debug_sprintf_event(sfdbg, 3, + "allocate_buffers: rate=%lu f=%lu sdb=%lu/%lu" + " sample_size=%lu cpuhw=%p\n", + SAMPL_RATE(hwc), freq, n_sdb, sfb_max_limit(hwc), + sample_size, cpuhw); + + return alloc_sampling_buffer(&cpuhw->sfb, + sfb_pending_allocs(&cpuhw->sfb, hwc)); +} + +static unsigned long min_percent(unsigned int percent, unsigned long base, + unsigned long min) +{ + return min_t(unsigned long, min, DIV_ROUND_UP(percent * base, 100)); +} + +static unsigned long compute_sfb_extent(unsigned long ratio, unsigned long base) +{ + /* Use a percentage-based approach to extend the sampling facility + * buffer. Accept up to 5% sample data loss. + * Vary the extents between 1% to 5% of the current number of + * sample-data-blocks. + */ + if (ratio <= 5) + return 0; + if (ratio <= 25) + return min_percent(1, base, 1); + if (ratio <= 50) + return min_percent(1, base, 1); + if (ratio <= 75) + return min_percent(2, base, 2); + if (ratio <= 100) + return min_percent(3, base, 3); + if (ratio <= 250) + return min_percent(4, base, 4); + + return min_percent(5, base, 8); +} + +static void sfb_account_overflows(struct cpu_hw_sf *cpuhw, + struct hw_perf_event *hwc) +{ + unsigned long ratio, num; + + if (!OVERFLOW_REG(hwc)) + return; + + /* The sample_overflow contains the average number of sample data + * that has been lost because sample-data-blocks were full. + * + * Calculate the total number of sample data entries that has been + * discarded. Then calculate the ratio of lost samples to total samples + * per second in percent. + */ + ratio = DIV_ROUND_UP(100 * OVERFLOW_REG(hwc) * cpuhw->sfb.num_sdb, + sample_rate_to_freq(&cpuhw->qsi, SAMPL_RATE(hwc))); + + /* Compute number of sample-data-blocks */ + num = compute_sfb_extent(ratio, cpuhw->sfb.num_sdb); + if (num) + sfb_account_allocs(num, hwc); + + debug_sprintf_event(sfdbg, 5, "sfb: overflow: overflow=%llu ratio=%lu" + " num=%lu\n", OVERFLOW_REG(hwc), ratio, num); + OVERFLOW_REG(hwc) = 0; +} + +/* extend_sampling_buffer() - Extend sampling buffer + * @sfb: Sampling buffer structure (for local CPU) + * @hwc: Perf event hardware structure + * + * Use this function to extend the sampling buffer based on the overflow counter + * and postponed allocation extents stored in the specified Perf event hardware. + * + * Important: This function disables the sampling facility in order to safely + * change the sampling buffer structure. Do not call this function + * when the PMU is active. + */ +static void extend_sampling_buffer(struct sf_buffer *sfb, + struct hw_perf_event *hwc) +{ + unsigned long num, num_old; + int rc; + + num = sfb_pending_allocs(sfb, hwc); + if (!num) + return; + num_old = sfb->num_sdb; + + /* Disable the sampling facility to reset any states and also + * clear pending measurement alerts. + */ + sf_disable(); + + /* Extend the sampling buffer. + * This memory allocation typically happens in an atomic context when + * called by perf. Because this is a reallocation, it is fine if the + * new SDB-request cannot be satisfied immediately. + */ + rc = realloc_sampling_buffer(sfb, num, GFP_ATOMIC); + if (rc) + debug_sprintf_event(sfdbg, 5, "sfb: extend: realloc " + "failed with rc=%i\n", rc); + + if (sfb_has_pending_allocs(sfb, hwc)) + debug_sprintf_event(sfdbg, 5, "sfb: extend: " + "req=%lu alloc=%lu remaining=%lu\n", + num, sfb->num_sdb - num_old, + sfb_pending_allocs(sfb, hwc)); +} + + +/* Number of perf events counting hardware events */ +static atomic_t num_events; +/* Used to avoid races in calling reserve/release_cpumf_hardware */ +static DEFINE_MUTEX(pmc_reserve_mutex); + +#define PMC_INIT 0 +#define PMC_RELEASE 1 +#define PMC_FAILURE 2 +static void setup_pmc_cpu(void *flags) +{ + int err; + struct cpu_hw_sf *cpusf = &__get_cpu_var(cpu_hw_sf); + + err = 0; + switch (*((int *) flags)) { + case PMC_INIT: + memset(cpusf, 0, sizeof(*cpusf)); + err = qsi(&cpusf->qsi); + if (err) + break; + cpusf->flags |= PMU_F_RESERVED; + err = sf_disable(); + if (err) + pr_err("Switching off the sampling facility failed " + "with rc=%i\n", err); + debug_sprintf_event(sfdbg, 5, + "setup_pmc_cpu: initialized: cpuhw=%p\n", cpusf); + break; + case PMC_RELEASE: + cpusf->flags &= ~PMU_F_RESERVED; + err = sf_disable(); + if (err) { + pr_err("Switching off the sampling facility failed " + "with rc=%i\n", err); + } else + deallocate_buffers(cpusf); + debug_sprintf_event(sfdbg, 5, + "setup_pmc_cpu: released: cpuhw=%p\n", cpusf); + break; + } + if (err) + *((int *) flags) |= PMC_FAILURE; +} + +static void release_pmc_hardware(void) +{ + int flags = PMC_RELEASE; + + irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT); + on_each_cpu(setup_pmc_cpu, &flags, 1); + perf_release_sampling(); +} + +static int reserve_pmc_hardware(void) +{ + int flags = PMC_INIT; + int err; + + err = perf_reserve_sampling(); + if (err) + return err; + on_each_cpu(setup_pmc_cpu, &flags, 1); + if (flags & PMC_FAILURE) { + release_pmc_hardware(); + return -ENODEV; + } + irq_subclass_register(IRQ_SUBCLASS_MEASUREMENT_ALERT); + + return 0; +} + +static void hw_perf_event_destroy(struct perf_event *event) +{ + /* Free raw sample buffer */ + if (RAWSAMPLE_REG(&event->hw)) + kfree((void *) RAWSAMPLE_REG(&event->hw)); + + /* Release PMC if this is the last perf event */ + if (!atomic_add_unless(&num_events, -1, 1)) { + mutex_lock(&pmc_reserve_mutex); + if (atomic_dec_return(&num_events) == 0) + release_pmc_hardware(); + mutex_unlock(&pmc_reserve_mutex); + } +} + +static void hw_init_period(struct hw_perf_event *hwc, u64 period) +{ + hwc->sample_period = period; + hwc->last_period = hwc->sample_period; + local64_set(&hwc->period_left, hwc->sample_period); +} + +static void hw_reset_registers(struct hw_perf_event *hwc, + unsigned long *sdbt_origin) +{ + struct sf_raw_sample *sfr; + + /* (Re)set to first sample-data-block-table */ + TEAR_REG(hwc) = (unsigned long) sdbt_origin; + + /* (Re)set raw sampling buffer register */ + sfr = (struct sf_raw_sample *) RAWSAMPLE_REG(hwc); + memset(&sfr->basic, 0, sizeof(sfr->basic)); + memset(&sfr->diag, 0, sfr->dsdes); +} + +static unsigned long hw_limit_rate(const struct hws_qsi_info_block *si, + unsigned long rate) +{ + return clamp_t(unsigned long, rate, + si->min_sampl_rate, si->max_sampl_rate); +} + +static int __hw_perf_event_init(struct perf_event *event) +{ + struct cpu_hw_sf *cpuhw; + struct hws_qsi_info_block si; + struct perf_event_attr *attr = &event->attr; + struct hw_perf_event *hwc = &event->hw; + unsigned long rate; + int cpu, err; + + /* Reserve CPU-measurement sampling facility */ + err = 0; + if (!atomic_inc_not_zero(&num_events)) { + mutex_lock(&pmc_reserve_mutex); + if (atomic_read(&num_events) == 0 && reserve_pmc_hardware()) + err = -EBUSY; + else + atomic_inc(&num_events); + mutex_unlock(&pmc_reserve_mutex); + } + event->destroy = hw_perf_event_destroy; + + if (err) + goto out; + + /* Access per-CPU sampling information (query sampling info) */ + /* + * The event->cpu value can be -1 to count on every CPU, for example, + * when attaching to a task. If this is specified, use the query + * sampling info from the current CPU, otherwise use event->cpu to + * retrieve the per-CPU information. + * Later, cpuhw indicates whether to allocate sampling buffers for a + * particular CPU (cpuhw!=NULL) or each online CPU (cpuw==NULL). + */ + memset(&si, 0, sizeof(si)); + cpuhw = NULL; + if (event->cpu == -1) + qsi(&si); + else { + /* Event is pinned to a particular CPU, retrieve the per-CPU + * sampling structure for accessing the CPU-specific QSI. + */ + cpuhw = &per_cpu(cpu_hw_sf, event->cpu); + si = cpuhw->qsi; + } + + /* Check sampling facility authorization and, if not authorized, + * fall back to other PMUs. It is safe to check any CPU because + * the authorization is identical for all configured CPUs. + */ + if (!si.as) { + err = -ENOENT; + goto out; + } + + /* Always enable basic sampling */ + SAMPL_FLAGS(hwc) = PERF_CPUM_SF_BASIC_MODE; + + /* Check if diagnostic sampling is requested. Deny if the required + * sampling authorization is missing. + */ + if (attr->config == PERF_EVENT_CPUM_SF_DIAG) { + if (!si.ad) { + err = -EPERM; + goto out; + } + SAMPL_FLAGS(hwc) |= PERF_CPUM_SF_DIAG_MODE; + } + + /* Check and set other sampling flags */ + if (attr->config1 & PERF_CPUM_SF_FULL_BLOCKS) + SAMPL_FLAGS(hwc) |= PERF_CPUM_SF_FULL_BLOCKS; + + /* The sampling information (si) contains information about the + * min/max sampling intervals and the CPU speed. So calculate the + * correct sampling interval and avoid the whole period adjust + * feedback loop. + */ + rate = 0; + if (attr->freq) { + rate = freq_to_sample_rate(&si, attr->sample_freq); + rate = hw_limit_rate(&si, rate); + attr->freq = 0; + attr->sample_period = rate; + } else { + /* The min/max sampling rates specifies the valid range + * of sample periods. If the specified sample period is + * out of range, limit the period to the range boundary. + */ + rate = hw_limit_rate(&si, hwc->sample_period); + + /* The perf core maintains a maximum sample rate that is + * configurable through the sysctl interface. Ensure the + * sampling rate does not exceed this value. This also helps + * to avoid throttling when pushing samples with + * perf_event_overflow(). + */ + if (sample_rate_to_freq(&si, rate) > + sysctl_perf_event_sample_rate) { + err = -EINVAL; + debug_sprintf_event(sfdbg, 1, "Sampling rate exceeds maximum perf sample rate\n"); + goto out; + } + } + SAMPL_RATE(hwc) = rate; + hw_init_period(hwc, SAMPL_RATE(hwc)); + + /* Initialize sample data overflow accounting */ + hwc->extra_reg.reg = REG_OVERFLOW; + OVERFLOW_REG(hwc) = 0; + + /* Allocate the per-CPU sampling buffer using the CPU information + * from the event. If the event is not pinned to a particular + * CPU (event->cpu == -1; or cpuhw == NULL), allocate sampling + * buffers for each online CPU. + */ + if (cpuhw) + /* Event is pinned to a particular CPU */ + err = allocate_buffers(cpuhw, hwc); + else { + /* Event is not pinned, allocate sampling buffer on + * each online CPU + */ + for_each_online_cpu(cpu) { + cpuhw = &per_cpu(cpu_hw_sf, cpu); + err = allocate_buffers(cpuhw, hwc); + if (err) + break; + } + } +out: + return err; +} + +static int cpumsf_pmu_event_init(struct perf_event *event) +{ + int err; + + /* No support for taken branch sampling */ + if (has_branch_stack(event)) + return -EOPNOTSUPP; + + switch (event->attr.type) { + case PERF_TYPE_RAW: + if ((event->attr.config != PERF_EVENT_CPUM_SF) && + (event->attr.config != PERF_EVENT_CPUM_SF_DIAG)) + return -ENOENT; + break; + case PERF_TYPE_HARDWARE: + /* Support sampling of CPU cycles in addition to the + * counter facility. However, the counter facility + * is more precise and, hence, restrict this PMU to + * sampling events only. + */ + if (event->attr.config != PERF_COUNT_HW_CPU_CYCLES) + return -ENOENT; + if (!is_sampling_event(event)) + return -ENOENT; + break; + default: + return -ENOENT; + } + + /* Check online status of the CPU to which the event is pinned */ + if (event->cpu >= nr_cpumask_bits || + (event->cpu >= 0 && !cpu_online(event->cpu))) + return -ENODEV; + + /* Force reset of idle/hv excludes regardless of what the + * user requested. + */ + if (event->attr.exclude_hv) + event->attr.exclude_hv = 0; + if (event->attr.exclude_idle) + event->attr.exclude_idle = 0; + + err = __hw_perf_event_init(event); + if (unlikely(err)) + if (event->destroy) + event->destroy(event); + return err; +} + +static void cpumsf_pmu_enable(struct pmu *pmu) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + struct hw_perf_event *hwc; + int err; + + if (cpuhw->flags & PMU_F_ENABLED) + return; + + if (cpuhw->flags & PMU_F_ERR_MASK) + return; + + /* Check whether to extent the sampling buffer. + * + * Two conditions trigger an increase of the sampling buffer for a + * perf event: + * 1. Postponed buffer allocations from the event initialization. + * 2. Sampling overflows that contribute to pending allocations. + * + * Note that the extend_sampling_buffer() function disables the sampling + * facility, but it can be fully re-enabled using sampling controls that + * have been saved in cpumsf_pmu_disable(). + */ + if (cpuhw->event) { + hwc = &cpuhw->event->hw; + /* Account number of overflow-designated buffer extents */ + sfb_account_overflows(cpuhw, hwc); + if (sfb_has_pending_allocs(&cpuhw->sfb, hwc)) + extend_sampling_buffer(&cpuhw->sfb, hwc); + } + + /* (Re)enable the PMU and sampling facility */ + cpuhw->flags |= PMU_F_ENABLED; + barrier(); + + err = lsctl(&cpuhw->lsctl); + if (err) { + cpuhw->flags &= ~PMU_F_ENABLED; + pr_err("Loading sampling controls failed: op=%i err=%i\n", + 1, err); + return; + } + + debug_sprintf_event(sfdbg, 6, "pmu_enable: es=%i cs=%i ed=%i cd=%i " + "tear=%p dear=%p\n", cpuhw->lsctl.es, cpuhw->lsctl.cs, + cpuhw->lsctl.ed, cpuhw->lsctl.cd, + (void *) cpuhw->lsctl.tear, (void *) cpuhw->lsctl.dear); +} + +static void cpumsf_pmu_disable(struct pmu *pmu) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + struct hws_lsctl_request_block inactive; + struct hws_qsi_info_block si; + int err; + + if (!(cpuhw->flags & PMU_F_ENABLED)) + return; + + if (cpuhw->flags & PMU_F_ERR_MASK) + return; + + /* Switch off sampling activation control */ + inactive = cpuhw->lsctl; + inactive.cs = 0; + inactive.cd = 0; + + err = lsctl(&inactive); + if (err) { + pr_err("Loading sampling controls failed: op=%i err=%i\n", + 2, err); + return; + } + + /* Save state of TEAR and DEAR register contents */ + if (!qsi(&si)) { + /* TEAR/DEAR values are valid only if the sampling facility is + * enabled. Note that cpumsf_pmu_disable() might be called even + * for a disabled sampling facility because cpumsf_pmu_enable() + * controls the enable/disable state. + */ + if (si.es) { + cpuhw->lsctl.tear = si.tear; + cpuhw->lsctl.dear = si.dear; + } + } else + debug_sprintf_event(sfdbg, 3, "cpumsf_pmu_disable: " + "qsi() failed with err=%i\n", err); + + cpuhw->flags &= ~PMU_F_ENABLED; +} + +/* perf_exclude_event() - Filter event + * @event: The perf event + * @regs: pt_regs structure + * @sde_regs: Sample-data-entry (sde) regs structure + * + * Filter perf events according to their exclude specification. + * + * Return non-zero if the event shall be excluded. + */ +static int perf_exclude_event(struct perf_event *event, struct pt_regs *regs, + struct perf_sf_sde_regs *sde_regs) +{ + if (event->attr.exclude_user && user_mode(regs)) + return 1; + if (event->attr.exclude_kernel && !user_mode(regs)) + return 1; + if (event->attr.exclude_guest && sde_regs->in_guest) + return 1; + if (event->attr.exclude_host && !sde_regs->in_guest) + return 1; + return 0; +} + +/* perf_push_sample() - Push samples to perf + * @event: The perf event + * @sample: Hardware sample data + * + * Use the hardware sample data to create perf event sample. The sample + * is the pushed to the event subsystem and the function checks for + * possible event overflows. If an event overflow occurs, the PMU is + * stopped. + * + * Return non-zero if an event overflow occurred. + */ +static int perf_push_sample(struct perf_event *event, struct sf_raw_sample *sfr) +{ + int overflow; + struct pt_regs regs; + struct perf_sf_sde_regs *sde_regs; + struct perf_sample_data data; + struct perf_raw_record raw; + + /* Setup perf sample */ + perf_sample_data_init(&data, 0, event->hw.last_period); + raw.size = sfr->size; + raw.data = sfr; + data.raw = &raw; + + /* Setup pt_regs to look like an CPU-measurement external interrupt + * using the Program Request Alert code. The regs.int_parm_long + * field which is unused contains additional sample-data-entry related + * indicators. + */ + memset(®s, 0, sizeof(regs)); + regs.int_code = 0x1407; + regs.int_parm = CPU_MF_INT_SF_PRA; + sde_regs = (struct perf_sf_sde_regs *) ®s.int_parm_long; + + regs.psw.addr = sfr->basic.ia; + if (sfr->basic.T) + regs.psw.mask |= PSW_MASK_DAT; + if (sfr->basic.W) + regs.psw.mask |= PSW_MASK_WAIT; + if (sfr->basic.P) + regs.psw.mask |= PSW_MASK_PSTATE; + switch (sfr->basic.AS) { + case 0x0: + regs.psw.mask |= PSW_ASC_PRIMARY; + break; + case 0x1: + regs.psw.mask |= PSW_ASC_ACCREG; + break; + case 0x2: + regs.psw.mask |= PSW_ASC_SECONDARY; + break; + case 0x3: + regs.psw.mask |= PSW_ASC_HOME; + break; + } + + /* The host-program-parameter (hpp) contains the sie control + * block that is set by sie64a() in entry64.S. Check if hpp + * refers to a valid control block and set sde_regs flags + * accordingly. This would allow to use hpp values for other + * purposes too. + * For now, simply use a non-zero value as guest indicator. + */ + if (sfr->basic.hpp) + sde_regs->in_guest = 1; + + overflow = 0; + if (perf_exclude_event(event, ®s, sde_regs)) + goto out; + if (perf_event_overflow(event, &data, ®s)) { + overflow = 1; + event->pmu->stop(event, 0); + } + perf_event_update_userpage(event); +out: + return overflow; +} + +static void perf_event_count_update(struct perf_event *event, u64 count) +{ + local64_add(count, &event->count); +} + +static int sample_format_is_valid(struct hws_combined_entry *sample, + unsigned int flags) +{ + if (likely(flags & PERF_CPUM_SF_BASIC_MODE)) + /* Only basic-sampling data entries with data-entry-format + * version of 0x0001 can be processed. + */ + if (sample->basic.def != 0x0001) + return 0; + if (flags & PERF_CPUM_SF_DIAG_MODE) + /* The data-entry-format number of diagnostic-sampling data + * entries can vary. Because diagnostic data is just passed + * through, do only a sanity check on the DEF. + */ + if (sample->diag.def < 0x8001) + return 0; + return 1; +} + +static int sample_is_consistent(struct hws_combined_entry *sample, + unsigned long flags) +{ + /* This check applies only to basic-sampling data entries of potentially + * combined-sampling data entries. Invalid entries cannot be processed + * by the PMU and, thus, do not deliver an associated + * diagnostic-sampling data entry. + */ + if (unlikely(!(flags & PERF_CPUM_SF_BASIC_MODE))) + return 0; + /* + * Samples are skipped, if they are invalid or for which the + * instruction address is not predictable, i.e., the wait-state bit is + * set. + */ + if (sample->basic.I || sample->basic.W) + return 0; + return 1; +} + +static void reset_sample_slot(struct hws_combined_entry *sample, + unsigned long flags) +{ + if (likely(flags & PERF_CPUM_SF_BASIC_MODE)) + sample->basic.def = 0; + if (flags & PERF_CPUM_SF_DIAG_MODE) + sample->diag.def = 0; +} + +static void sfr_store_sample(struct sf_raw_sample *sfr, + struct hws_combined_entry *sample) +{ + if (likely(sfr->format & PERF_CPUM_SF_BASIC_MODE)) + sfr->basic = sample->basic; + if (sfr->format & PERF_CPUM_SF_DIAG_MODE) + memcpy(&sfr->diag, &sample->diag, sfr->dsdes); +} + +static void debug_sample_entry(struct hws_combined_entry *sample, + struct hws_trailer_entry *te, + unsigned long flags) +{ + debug_sprintf_event(sfdbg, 4, "hw_collect_samples: Found unknown " + "sampling data entry: te->f=%i basic.def=%04x (%p)" + " diag.def=%04x (%p)\n", te->f, + sample->basic.def, &sample->basic, + (flags & PERF_CPUM_SF_DIAG_MODE) + ? sample->diag.def : 0xFFFF, + (flags & PERF_CPUM_SF_DIAG_MODE) + ? &sample->diag : NULL); +} + +/* hw_collect_samples() - Walk through a sample-data-block and collect samples + * @event: The perf event + * @sdbt: Sample-data-block table + * @overflow: Event overflow counter + * + * Walks through a sample-data-block and collects sampling data entries that are + * then pushed to the perf event subsystem. Depending on the sampling function, + * there can be either basic-sampling or combined-sampling data entries. A + * combined-sampling data entry consists of a basic- and a diagnostic-sampling + * data entry. The sampling function is determined by the flags in the perf + * event hardware structure. The function always works with a combined-sampling + * data entry but ignores the the diagnostic portion if it is not available. + * + * Note that the implementation focuses on basic-sampling data entries and, if + * such an entry is not valid, the entire combined-sampling data entry is + * ignored. + * + * The overflow variables counts the number of samples that has been discarded + * due to a perf event overflow. + */ +static void hw_collect_samples(struct perf_event *event, unsigned long *sdbt, + unsigned long long *overflow) +{ + unsigned long flags = SAMPL_FLAGS(&event->hw); + struct hws_combined_entry *sample; + struct hws_trailer_entry *te; + struct sf_raw_sample *sfr; + size_t sample_size; + + /* Prepare and initialize raw sample data */ + sfr = (struct sf_raw_sample *) RAWSAMPLE_REG(&event->hw); + sfr->format = flags & PERF_CPUM_SF_MODE_MASK; + + sample_size = event_sample_size(&event->hw); + te = (struct hws_trailer_entry *) trailer_entry_ptr(*sdbt); + sample = (struct hws_combined_entry *) *sdbt; + while ((unsigned long *) sample < (unsigned long *) te) { + /* Check for an empty sample */ + if (!sample->basic.def) + break; + + /* Update perf event period */ + perf_event_count_update(event, SAMPL_RATE(&event->hw)); + + /* Check sampling data entry */ + if (sample_format_is_valid(sample, flags)) { + /* If an event overflow occurred, the PMU is stopped to + * throttle event delivery. Remaining sample data is + * discarded. + */ + if (!*overflow) { + if (sample_is_consistent(sample, flags)) { + /* Deliver sample data to perf */ + sfr_store_sample(sfr, sample); + *overflow = perf_push_sample(event, sfr); + } + } else + /* Count discarded samples */ + *overflow += 1; + } else { + debug_sample_entry(sample, te, flags); + /* Sample slot is not yet written or other record. + * + * This condition can occur if the buffer was reused + * from a combined basic- and diagnostic-sampling. + * If only basic-sampling is then active, entries are + * written into the larger diagnostic entries. + * This is typically the case for sample-data-blocks + * that are not full. Stop processing if the first + * invalid format was detected. + */ + if (!te->f) + break; + } + + /* Reset sample slot and advance to next sample */ + reset_sample_slot(sample, flags); + sample += sample_size; + } +} + +/* hw_perf_event_update() - Process sampling buffer + * @event: The perf event + * @flush_all: Flag to also flush partially filled sample-data-blocks + * + * Processes the sampling buffer and create perf event samples. + * The sampling buffer position are retrieved and saved in the TEAR_REG + * register of the specified perf event. + * + * Only full sample-data-blocks are processed. Specify the flash_all flag + * to also walk through partially filled sample-data-blocks. It is ignored + * if PERF_CPUM_SF_FULL_BLOCKS is set. The PERF_CPUM_SF_FULL_BLOCKS flag + * enforces the processing of full sample-data-blocks only (trailer entries + * with the block-full-indicator bit set). + */ +static void hw_perf_event_update(struct perf_event *event, int flush_all) +{ + struct hw_perf_event *hwc = &event->hw; + struct hws_trailer_entry *te; + unsigned long *sdbt; + unsigned long long event_overflow, sampl_overflow, num_sdb, te_flags; + int done; + + if (flush_all && SDB_FULL_BLOCKS(hwc)) + flush_all = 0; + + sdbt = (unsigned long *) TEAR_REG(hwc); + done = event_overflow = sampl_overflow = num_sdb = 0; + while (!done) { + /* Get the trailer entry of the sample-data-block */ + te = (struct hws_trailer_entry *) trailer_entry_ptr(*sdbt); + + /* Leave loop if no more work to do (block full indicator) */ + if (!te->f) { + done = 1; + if (!flush_all) + break; + } + + /* Check the sample overflow count */ + if (te->overflow) + /* Account sample overflows and, if a particular limit + * is reached, extend the sampling buffer. + * For details, see sfb_account_overflows(). + */ + sampl_overflow += te->overflow; + + /* Timestamps are valid for full sample-data-blocks only */ + debug_sprintf_event(sfdbg, 6, "hw_perf_event_update: sdbt=%p " + "overflow=%llu timestamp=0x%llx\n", + sdbt, te->overflow, + (te->f) ? trailer_timestamp(te) : 0ULL); + + /* Collect all samples from a single sample-data-block and + * flag if an (perf) event overflow happened. If so, the PMU + * is stopped and remaining samples will be discarded. + */ + hw_collect_samples(event, sdbt, &event_overflow); + num_sdb++; + + /* Reset trailer (using compare-double-and-swap) */ + do { + te_flags = te->flags & ~SDB_TE_BUFFER_FULL_MASK; + te_flags |= SDB_TE_ALERT_REQ_MASK; + } while (!cmpxchg_double(&te->flags, &te->overflow, + te->flags, te->overflow, + te_flags, 0ULL)); + + /* Advance to next sample-data-block */ + sdbt++; + if (is_link_entry(sdbt)) + sdbt = get_next_sdbt(sdbt); + + /* Update event hardware registers */ + TEAR_REG(hwc) = (unsigned long) sdbt; + + /* Stop processing sample-data if all samples of the current + * sample-data-block were flushed even if it was not full. + */ + if (flush_all && done) + break; + + /* If an event overflow happened, discard samples by + * processing any remaining sample-data-blocks. + */ + if (event_overflow) + flush_all = 1; + } + + /* Account sample overflows in the event hardware structure */ + if (sampl_overflow) + OVERFLOW_REG(hwc) = DIV_ROUND_UP(OVERFLOW_REG(hwc) + + sampl_overflow, 1 + num_sdb); + if (sampl_overflow || event_overflow) + debug_sprintf_event(sfdbg, 4, "hw_perf_event_update: " + "overflow stats: sample=%llu event=%llu\n", + sampl_overflow, event_overflow); +} + +static void cpumsf_pmu_read(struct perf_event *event) +{ + /* Nothing to do ... updates are interrupt-driven */ +} + +/* Activate sampling control. + * Next call of pmu_enable() starts sampling. + */ +static void cpumsf_pmu_start(struct perf_event *event, int flags) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + + if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED))) + return; + + if (flags & PERF_EF_RELOAD) + WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE)); + + perf_pmu_disable(event->pmu); + event->hw.state = 0; + cpuhw->lsctl.cs = 1; + if (SAMPL_DIAG_MODE(&event->hw)) + cpuhw->lsctl.cd = 1; + perf_pmu_enable(event->pmu); +} + +/* Deactivate sampling control. + * Next call of pmu_enable() stops sampling. + */ +static void cpumsf_pmu_stop(struct perf_event *event, int flags) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + + if (event->hw.state & PERF_HES_STOPPED) + return; + + perf_pmu_disable(event->pmu); + cpuhw->lsctl.cs = 0; + cpuhw->lsctl.cd = 0; + event->hw.state |= PERF_HES_STOPPED; + + if ((flags & PERF_EF_UPDATE) && !(event->hw.state & PERF_HES_UPTODATE)) { + hw_perf_event_update(event, 1); + event->hw.state |= PERF_HES_UPTODATE; + } + perf_pmu_enable(event->pmu); +} + +static int cpumsf_pmu_add(struct perf_event *event, int flags) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + int err; + + if (cpuhw->flags & PMU_F_IN_USE) + return -EAGAIN; + + if (!cpuhw->sfb.sdbt) + return -EINVAL; + + err = 0; + perf_pmu_disable(event->pmu); + + event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED; + + /* Set up sampling controls. Always program the sampling register + * using the SDB-table start. Reset TEAR_REG event hardware register + * that is used by hw_perf_event_update() to store the sampling buffer + * position after samples have been flushed. + */ + cpuhw->lsctl.s = 0; + cpuhw->lsctl.h = 1; + cpuhw->lsctl.tear = (unsigned long) cpuhw->sfb.sdbt; + cpuhw->lsctl.dear = *(unsigned long *) cpuhw->sfb.sdbt; + cpuhw->lsctl.interval = SAMPL_RATE(&event->hw); + hw_reset_registers(&event->hw, cpuhw->sfb.sdbt); + + /* Ensure sampling functions are in the disabled state. If disabled, + * switch on sampling enable control. */ + if (WARN_ON_ONCE(cpuhw->lsctl.es == 1 || cpuhw->lsctl.ed == 1)) { + err = -EAGAIN; + goto out; + } + cpuhw->lsctl.es = 1; + if (SAMPL_DIAG_MODE(&event->hw)) + cpuhw->lsctl.ed = 1; + + /* Set in_use flag and store event */ + event->hw.idx = 0; /* only one sampling event per CPU supported */ + cpuhw->event = event; + cpuhw->flags |= PMU_F_IN_USE; + + if (flags & PERF_EF_START) + cpumsf_pmu_start(event, PERF_EF_RELOAD); +out: + perf_event_update_userpage(event); + perf_pmu_enable(event->pmu); + return err; +} + +static void cpumsf_pmu_del(struct perf_event *event, int flags) +{ + struct cpu_hw_sf *cpuhw = &__get_cpu_var(cpu_hw_sf); + + perf_pmu_disable(event->pmu); + cpumsf_pmu_stop(event, PERF_EF_UPDATE); + + cpuhw->lsctl.es = 0; + cpuhw->lsctl.ed = 0; + cpuhw->flags &= ~PMU_F_IN_USE; + cpuhw->event = NULL; + + perf_event_update_userpage(event); + perf_pmu_enable(event->pmu); +} + +static int cpumsf_pmu_event_idx(struct perf_event *event) +{ + return event->hw.idx; +} + +CPUMF_EVENT_ATTR(SF, SF_CYCLES_BASIC, PERF_EVENT_CPUM_SF); +CPUMF_EVENT_ATTR(SF, SF_CYCLES_BASIC_DIAG, PERF_EVENT_CPUM_SF_DIAG); + +static struct attribute *cpumsf_pmu_events_attr[] = { + CPUMF_EVENT_PTR(SF, SF_CYCLES_BASIC), + CPUMF_EVENT_PTR(SF, SF_CYCLES_BASIC_DIAG), + NULL, +}; + +PMU_FORMAT_ATTR(event, "config:0-63"); + +static struct attribute *cpumsf_pmu_format_attr[] = { + &format_attr_event.attr, + NULL, +}; + +static struct attribute_group cpumsf_pmu_events_group = { + .name = "events", + .attrs = cpumsf_pmu_events_attr, +}; +static struct attribute_group cpumsf_pmu_format_group = { + .name = "format", + .attrs = cpumsf_pmu_format_attr, +}; +static const struct attribute_group *cpumsf_pmu_attr_groups[] = { + &cpumsf_pmu_events_group, + &cpumsf_pmu_format_group, + NULL, +}; + +static struct pmu cpumf_sampling = { + .pmu_enable = cpumsf_pmu_enable, + .pmu_disable = cpumsf_pmu_disable, + + .event_init = cpumsf_pmu_event_init, + .add = cpumsf_pmu_add, + .del = cpumsf_pmu_del, + + .start = cpumsf_pmu_start, + .stop = cpumsf_pmu_stop, + .read = cpumsf_pmu_read, + + .event_idx = cpumsf_pmu_event_idx, + .attr_groups = cpumsf_pmu_attr_groups, +}; + +static void cpumf_measurement_alert(struct ext_code ext_code, + unsigned int alert, unsigned long unused) +{ + struct cpu_hw_sf *cpuhw; + + if (!(alert & CPU_MF_INT_SF_MASK)) + return; + inc_irq_stat(IRQEXT_CMS); + cpuhw = &__get_cpu_var(cpu_hw_sf); + + /* Measurement alerts are shared and might happen when the PMU + * is not reserved. Ignore these alerts in this case. */ + if (!(cpuhw->flags & PMU_F_RESERVED)) + return; + + /* The processing below must take care of multiple alert events that + * might be indicated concurrently. */ + + /* Program alert request */ + if (alert & CPU_MF_INT_SF_PRA) { + if (cpuhw->flags & PMU_F_IN_USE) + hw_perf_event_update(cpuhw->event, 0); + else + WARN_ON_ONCE(!(cpuhw->flags & PMU_F_IN_USE)); + } + + /* Report measurement alerts only for non-PRA codes */ + if (alert != CPU_MF_INT_SF_PRA) + debug_sprintf_event(sfdbg, 6, "measurement alert: 0x%x\n", alert); + + /* Sampling authorization change request */ + if (alert & CPU_MF_INT_SF_SACA) + qsi(&cpuhw->qsi); + + /* Loss of sample data due to high-priority machine activities */ + if (alert & CPU_MF_INT_SF_LSDA) { + pr_err("Sample data was lost\n"); + cpuhw->flags |= PMU_F_ERR_LSDA; + sf_disable(); + } + + /* Invalid sampling buffer entry */ + if (alert & (CPU_MF_INT_SF_IAE|CPU_MF_INT_SF_ISE)) { + pr_err("A sampling buffer entry is incorrect (alert=0x%x)\n", + alert); + cpuhw->flags |= PMU_F_ERR_IBE; + sf_disable(); + } +} + +static int cpumf_pmu_notifier(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + unsigned int cpu = (long) hcpu; + int flags; + + /* Ignore the notification if no events are scheduled on the PMU. + * This might be racy... + */ + if (!atomic_read(&num_events)) + return NOTIFY_OK; + + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_ONLINE: + case CPU_ONLINE_FROZEN: + flags = PMC_INIT; + smp_call_function_single(cpu, setup_pmc_cpu, &flags, 1); + break; + case CPU_DOWN_PREPARE: + flags = PMC_RELEASE; + smp_call_function_single(cpu, setup_pmc_cpu, &flags, 1); + break; + default: + break; + } + + return NOTIFY_OK; +} + +static int param_get_sfb_size(char *buffer, const struct kernel_param *kp) +{ + if (!cpum_sf_avail()) + return -ENODEV; + return sprintf(buffer, "%lu,%lu", CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB); +} + +static int param_set_sfb_size(const char *val, const struct kernel_param *kp) +{ + int rc; + unsigned long min, max; + + if (!cpum_sf_avail()) + return -ENODEV; + if (!val || !strlen(val)) + return -EINVAL; + + /* Valid parameter values: "min,max" or "max" */ + min = CPUM_SF_MIN_SDB; + max = CPUM_SF_MAX_SDB; + if (strchr(val, ',')) + rc = (sscanf(val, "%lu,%lu", &min, &max) == 2) ? 0 : -EINVAL; + else + rc = kstrtoul(val, 10, &max); + + if (min < 2 || min >= max || max > get_num_physpages()) + rc = -EINVAL; + if (rc) + return rc; + + sfb_set_limits(min, max); + pr_info("The sampling buffer limits have changed to: " + "min=%lu max=%lu (diag=x%lu)\n", + CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB, CPUM_SF_SDB_DIAG_FACTOR); + return 0; +} + +#define param_check_sfb_size(name, p) __param_check(name, p, void) +static struct kernel_param_ops param_ops_sfb_size = { + .set = param_set_sfb_size, + .get = param_get_sfb_size, +}; + +#define RS_INIT_FAILURE_QSI 0x0001 +#define RS_INIT_FAILURE_BSDES 0x0002 +#define RS_INIT_FAILURE_ALRT 0x0003 +#define RS_INIT_FAILURE_PERF 0x0004 +static void __init pr_cpumsf_err(unsigned int reason) +{ + pr_err("Sampling facility support for perf is not available: " + "reason=%04x\n", reason); +} + +static int __init init_cpum_sampling_pmu(void) +{ + struct hws_qsi_info_block si; + int err; + + if (!cpum_sf_avail()) + return -ENODEV; + + memset(&si, 0, sizeof(si)); + if (qsi(&si)) { + pr_cpumsf_err(RS_INIT_FAILURE_QSI); + return -ENODEV; + } + + if (si.bsdes != sizeof(struct hws_basic_entry)) { + pr_cpumsf_err(RS_INIT_FAILURE_BSDES); + return -EINVAL; + } + + if (si.ad) + sfb_set_limits(CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB); + + sfdbg = debug_register(KMSG_COMPONENT, 2, 1, 80); + if (!sfdbg) + pr_err("Registering for s390dbf failed\n"); + debug_register_view(sfdbg, &debug_sprintf_view); + + err = register_external_interrupt(0x1407, cpumf_measurement_alert); + if (err) { + pr_cpumsf_err(RS_INIT_FAILURE_ALRT); + goto out; + } + + err = perf_pmu_register(&cpumf_sampling, "cpum_sf", PERF_TYPE_RAW); + if (err) { + pr_cpumsf_err(RS_INIT_FAILURE_PERF); + unregister_external_interrupt(0x1407, cpumf_measurement_alert); + goto out; + } + perf_cpu_notifier(cpumf_pmu_notifier); +out: + return err; +} +arch_initcall(init_cpum_sampling_pmu); +core_param(cpum_sfb_size, CPUM_SF_MAX_SDB, sfb_size, 0640); diff --git a/arch/s390/kernel/perf_event.c b/arch/s390/kernel/perf_event.c index 2343c218b8f9..5d2dfa31c4ef 100644 --- a/arch/s390/kernel/perf_event.c +++ b/arch/s390/kernel/perf_event.c @@ -1,7 +1,7 @@ /* * Performance event support for s390x * - * Copyright IBM Corp. 2012 + * Copyright IBM Corp. 2012, 2013 * Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com> * * This program is free software; you can redistribute it and/or modify @@ -16,15 +16,19 @@ #include <linux/kvm_host.h> #include <linux/percpu.h> #include <linux/export.h> +#include <linux/seq_file.h> +#include <linux/spinlock.h> +#include <linux/sysfs.h> #include <asm/irq.h> #include <asm/cpu_mf.h> #include <asm/lowcore.h> #include <asm/processor.h> +#include <asm/sysinfo.h> const char *perf_pmu_name(void) { if (cpum_cf_avail() || cpum_sf_avail()) - return "CPU-measurement facilities (CPUMF)"; + return "CPU-Measurement Facilities (CPU-MF)"; return "pmu"; } EXPORT_SYMBOL(perf_pmu_name); @@ -35,6 +39,8 @@ int perf_num_counters(void) if (cpum_cf_avail()) num += PERF_CPUM_CF_MAX_CTR; + if (cpum_sf_avail()) + num += PERF_CPUM_SF_MAX_CTR; return num; } @@ -54,7 +60,7 @@ static bool is_in_guest(struct pt_regs *regs) { if (user_mode(regs)) return false; -#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +#if IS_ENABLED(CONFIG_KVM) return instruction_pointer(regs) == (unsigned long) &sie_exit; #else return false; @@ -83,8 +89,31 @@ static unsigned long perf_misc_guest_flags(struct pt_regs *regs) : PERF_RECORD_MISC_GUEST_KERNEL; } +static unsigned long perf_misc_flags_sf(struct pt_regs *regs) +{ + struct perf_sf_sde_regs *sde_regs; + unsigned long flags; + + sde_regs = (struct perf_sf_sde_regs *) ®s->int_parm_long; + if (sde_regs->in_guest) + flags = user_mode(regs) ? PERF_RECORD_MISC_GUEST_USER + : PERF_RECORD_MISC_GUEST_KERNEL; + else + flags = user_mode(regs) ? PERF_RECORD_MISC_USER + : PERF_RECORD_MISC_KERNEL; + return flags; +} + unsigned long perf_misc_flags(struct pt_regs *regs) { + /* Check if the cpum_sf PMU has created the pt_regs structure. + * In this case, perf misc flags can be easily extracted. Otherwise, + * do regular checks on the pt_regs content. + */ + if (regs->int_code == 0x1407 && regs->int_parm == CPU_MF_INT_SF_PRA) + if (!regs->gprs[15]) + return perf_misc_flags_sf(regs); + if (is_in_guest(regs)) return perf_misc_guest_flags(regs); @@ -92,27 +121,107 @@ unsigned long perf_misc_flags(struct pt_regs *regs) : PERF_RECORD_MISC_KERNEL; } -void perf_event_print_debug(void) +void print_debug_cf(void) { struct cpumf_ctr_info cf_info; - unsigned long flags; - int cpu; - - if (!cpum_cf_avail()) - return; - - local_irq_save(flags); + int cpu = smp_processor_id(); - cpu = smp_processor_id(); memset(&cf_info, 0, sizeof(cf_info)); if (!qctri(&cf_info)) pr_info("CPU[%i] CPUM_CF: ver=%u.%u A=%04x E=%04x C=%04x\n", cpu, cf_info.cfvn, cf_info.csvn, cf_info.auth_ctl, cf_info.enable_ctl, cf_info.act_ctl); +} + +static void print_debug_sf(void) +{ + struct hws_qsi_info_block si; + int cpu = smp_processor_id(); + memset(&si, 0, sizeof(si)); + if (qsi(&si)) + return; + + pr_info("CPU[%i] CPUM_SF: basic=%i diag=%i min=%lu max=%lu cpu_speed=%u\n", + cpu, si.as, si.ad, si.min_sampl_rate, si.max_sampl_rate, + si.cpu_speed); + + if (si.as) + pr_info("CPU[%i] CPUM_SF: Basic-sampling: a=%i e=%i c=%i" + " bsdes=%i tear=%016lx dear=%016lx\n", cpu, + si.as, si.es, si.cs, si.bsdes, si.tear, si.dear); + if (si.ad) + pr_info("CPU[%i] CPUM_SF: Diagnostic-sampling: a=%i e=%i c=%i" + " dsdes=%i tear=%016lx dear=%016lx\n", cpu, + si.ad, si.ed, si.cd, si.dsdes, si.tear, si.dear); +} + +void perf_event_print_debug(void) +{ + unsigned long flags; + + local_irq_save(flags); + if (cpum_cf_avail()) + print_debug_cf(); + if (cpum_sf_avail()) + print_debug_sf(); local_irq_restore(flags); } +/* Service level infrastructure */ +static void sl_print_counter(struct seq_file *m) +{ + struct cpumf_ctr_info ci; + + memset(&ci, 0, sizeof(ci)); + if (qctri(&ci)) + return; + + seq_printf(m, "CPU-MF: Counter facility: version=%u.%u " + "authorization=%04x\n", ci.cfvn, ci.csvn, ci.auth_ctl); +} + +static void sl_print_sampling(struct seq_file *m) +{ + struct hws_qsi_info_block si; + + memset(&si, 0, sizeof(si)); + if (qsi(&si)) + return; + + if (!si.as && !si.ad) + return; + + seq_printf(m, "CPU-MF: Sampling facility: min_rate=%lu max_rate=%lu" + " cpu_speed=%u\n", si.min_sampl_rate, si.max_sampl_rate, + si.cpu_speed); + if (si.as) + seq_printf(m, "CPU-MF: Sampling facility: mode=basic" + " sample_size=%u\n", si.bsdes); + if (si.ad) + seq_printf(m, "CPU-MF: Sampling facility: mode=diagnostic" + " sample_size=%u\n", si.dsdes); +} + +static void service_level_perf_print(struct seq_file *m, + struct service_level *sl) +{ + if (cpum_cf_avail()) + sl_print_counter(m); + if (cpum_sf_avail()) + sl_print_sampling(m); +} + +static struct service_level service_level_perf = { + .seq_print = service_level_perf_print, +}; + +static int __init service_level_perf_register(void) +{ + return register_service_level(&service_level_perf); +} +arch_initcall(service_level_perf_register); + /* See also arch/s390/kernel/traps.c */ static unsigned long __store_trace(struct perf_callchain_entry *entry, unsigned long sp, @@ -172,3 +281,44 @@ void perf_callchain_kernel(struct perf_callchain_entry *entry, __store_trace(entry, head, S390_lowcore.thread_info, S390_lowcore.thread_info + THREAD_SIZE); } + +/* Perf defintions for PMU event attributes in sysfs */ +ssize_t cpumf_events_sysfs_show(struct device *dev, + struct device_attribute *attr, char *page) +{ + struct perf_pmu_events_attr *pmu_attr; + + pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr); + return sprintf(page, "event=0x%04llx,name=%s\n", + pmu_attr->id, attr->attr.name); +} + +/* Reserve/release functions for sharing perf hardware */ +static DEFINE_SPINLOCK(perf_hw_owner_lock); +static void *perf_sampling_owner; + +int perf_reserve_sampling(void) +{ + int err; + + err = 0; + spin_lock(&perf_hw_owner_lock); + if (perf_sampling_owner) { + pr_warn("The sampling facility is already reserved by %p\n", + perf_sampling_owner); + err = -EBUSY; + } else + perf_sampling_owner = __builtin_return_address(0); + spin_unlock(&perf_hw_owner_lock); + return err; +} +EXPORT_SYMBOL(perf_reserve_sampling); + +void perf_release_sampling(void) +{ + spin_lock(&perf_hw_owner_lock); + WARN_ON(!perf_sampling_owner); + perf_sampling_owner = NULL; + spin_unlock(&perf_hw_owner_lock); +} +EXPORT_SYMBOL(perf_release_sampling); diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index 7ed0d4e2a435..dd145321d215 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -261,20 +261,18 @@ static inline unsigned long brk_rnd(void) unsigned long arch_randomize_brk(struct mm_struct *mm) { - unsigned long ret = PAGE_ALIGN(mm->brk + brk_rnd()); + unsigned long ret; - if (ret < mm->brk) - return mm->brk; - return ret; + ret = PAGE_ALIGN(mm->brk + brk_rnd()); + return (ret > mm->brk) ? ret : mm->brk; } unsigned long randomize_et_dyn(unsigned long base) { - unsigned long ret = PAGE_ALIGN(base + brk_rnd()); + unsigned long ret; if (!(current->flags & PF_RANDOMIZE)) return base; - if (ret < base) - return base; - return ret; + ret = PAGE_ALIGN(base + brk_rnd()); + return (ret > base) ? ret : base; } diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c index e65c91c591e8..f6be6087a0e9 100644 --- a/arch/s390/kernel/ptrace.c +++ b/arch/s390/kernel/ptrace.c @@ -56,25 +56,26 @@ void update_cr_regs(struct task_struct *task) #ifdef CONFIG_64BIT /* Take care of the enable/disable of transactional execution. */ if (MACHINE_HAS_TE) { - unsigned long cr[3], cr_new[3]; + unsigned long cr, cr_new; - __ctl_store(cr, 0, 2); - cr_new[1] = cr[1]; + __ctl_store(cr, 0, 0); /* Set or clear transaction execution TXC bit 8. */ + cr_new = cr | (1UL << 55); if (task->thread.per_flags & PER_FLAG_NO_TE) - cr_new[0] = cr[0] & ~(1UL << 55); - else - cr_new[0] = cr[0] | (1UL << 55); + cr_new &= ~(1UL << 55); + if (cr_new != cr) + __ctl_load(cr, 0, 0); /* Set or clear transaction execution TDC bits 62 and 63. */ - cr_new[2] = cr[2] & ~3UL; + __ctl_store(cr, 2, 2); + cr_new = cr & ~3UL; if (task->thread.per_flags & PER_FLAG_TE_ABORT_RAND) { if (task->thread.per_flags & PER_FLAG_TE_ABORT_RAND_TEND) - cr_new[2] |= 1UL; + cr_new |= 1UL; else - cr_new[2] |= 2UL; + cr_new |= 2UL; } - if (memcmp(&cr_new, &cr, sizeof(cr))) - __ctl_load(cr_new, 0, 2); + if (cr_new != cr) + __ctl_load(cr_new, 2, 2); } #endif /* Copy user specified PER registers */ @@ -107,15 +108,11 @@ void update_cr_regs(struct task_struct *task) void user_enable_single_step(struct task_struct *task) { set_tsk_thread_flag(task, TIF_SINGLE_STEP); - if (task == current) - update_cr_regs(task); } void user_disable_single_step(struct task_struct *task) { clear_tsk_thread_flag(task, TIF_SINGLE_STEP); - if (task == current) - update_cr_regs(task); } /* diff --git a/arch/s390/kernel/s390_ksyms.c b/arch/s390/kernel/s390_ksyms.c index 3bac589844a7..9f60467938d1 100644 --- a/arch/s390/kernel/s390_ksyms.c +++ b/arch/s390/kernel/s390_ksyms.c @@ -5,7 +5,7 @@ #ifdef CONFIG_FUNCTION_TRACER EXPORT_SYMBOL(_mcount); #endif -#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +#if IS_ENABLED(CONFIG_KVM) EXPORT_SYMBOL(sie64a); EXPORT_SYMBOL(sie_exit); #endif diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c index 0f3d44ecbfc6..09e2f468f48b 100644 --- a/arch/s390/kernel/setup.c +++ b/arch/s390/kernel/setup.c @@ -373,7 +373,7 @@ static void __init setup_lowcore(void) /* * Set up PSW restart to call ipl.c:do_restart(). Copy the relevant - * restart data to the absolute zero lowcore. This is necesary if + * restart data to the absolute zero lowcore. This is necessary if * PSW restart is done on an offline CPU that has lowcore zero. */ lc->restart_stack = (unsigned long) restart_stack; diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index 958704798f4a..a7125b62a9a6 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -59,7 +59,7 @@ enum { }; struct pcpu { - struct cpu cpu; + struct cpu *cpu; struct _lowcore *lowcore; /* lowcore page(s) for the cpu */ unsigned long async_stack; /* async stack for the cpu */ unsigned long panic_stack; /* panic stack for the cpu */ @@ -159,9 +159,9 @@ static void pcpu_ec_call(struct pcpu *pcpu, int ec_bit) { int order; - set_bit(ec_bit, &pcpu->ec_mask); - order = pcpu_running(pcpu) ? - SIGP_EXTERNAL_CALL : SIGP_EMERGENCY_SIGNAL; + if (test_and_set_bit(ec_bit, &pcpu->ec_mask)) + return; + order = pcpu_running(pcpu) ? SIGP_EXTERNAL_CALL : SIGP_EMERGENCY_SIGNAL; pcpu_sigp_retry(pcpu, order, 0); } @@ -965,7 +965,7 @@ static int smp_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) { unsigned int cpu = (unsigned int)(long)hcpu; - struct cpu *c = &pcpu_devices[cpu].cpu; + struct cpu *c = pcpu_devices[cpu].cpu; struct device *s = &c->dev; int err = 0; @@ -982,10 +982,15 @@ static int smp_cpu_notify(struct notifier_block *self, unsigned long action, static int smp_add_present_cpu(int cpu) { - struct cpu *c = &pcpu_devices[cpu].cpu; - struct device *s = &c->dev; + struct device *s; + struct cpu *c; int rc; + c = kzalloc(sizeof(*c), GFP_KERNEL); + if (!c) + return -ENOMEM; + pcpu_devices[cpu].cpu = c; + s = &c->dev; c->hotpluggable = 1; rc = register_cpu(c, cpu); if (rc) diff --git a/arch/s390/kernel/syscalls.S b/arch/s390/kernel/syscalls.S index 913410bd74a3..143992152ec9 100644 --- a/arch/s390/kernel/syscalls.S +++ b/arch/s390/kernel/syscalls.S @@ -353,3 +353,5 @@ SYSCALL(sys_process_vm_writev,sys_process_vm_writev,compat_sys_process_vm_writev SYSCALL(sys_ni_syscall,sys_s390_runtime_instr,sys_s390_runtime_instr_wrapper) SYSCALL(sys_kcmp,sys_kcmp,sys_kcmp_wrapper) SYSCALL(sys_finit_module,sys_finit_module,sys_finit_module_wrapper) +SYSCALL(sys_sched_setattr,sys_sched_setattr,sys_sched_setattr_wrapper) /* 345 */ +SYSCALL(sys_sched_getattr,sys_sched_getattr,sys_sched_getattr_wrapper) diff --git a/arch/s390/kvm/diag.c b/arch/s390/kvm/diag.c index 78d967f180f4..8216c0e0b2e2 100644 --- a/arch/s390/kvm/diag.c +++ b/arch/s390/kvm/diag.c @@ -121,7 +121,7 @@ static int __diag_virtio_hypercall(struct kvm_vcpu *vcpu) * - gpr 4 contains the index on the bus (optionally) */ ret = kvm_io_bus_write_cookie(vcpu->kvm, KVM_VIRTIO_CCW_NOTIFY_BUS, - vcpu->run->s.regs.gprs[2], + vcpu->run->s.regs.gprs[2] & 0xffffffff, 8, &vcpu->run->s.regs.gprs[3], vcpu->run->s.regs.gprs[4]); @@ -137,7 +137,7 @@ static int __diag_virtio_hypercall(struct kvm_vcpu *vcpu) int kvm_s390_handle_diag(struct kvm_vcpu *vcpu) { - int code = (vcpu->arch.sie_block->ipb & 0xfff0000) >> 16; + int code = kvm_s390_get_base_disp_rs(vcpu) & 0xffff; if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE) return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP); diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 569494e01ec6..7635c00a1479 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -732,14 +732,16 @@ static int vcpu_post_run(struct kvm_vcpu *vcpu, int exit_reason) if (exit_reason >= 0) { rc = 0; + } else if (kvm_is_ucontrol(vcpu->kvm)) { + vcpu->run->exit_reason = KVM_EXIT_S390_UCONTROL; + vcpu->run->s390_ucontrol.trans_exc_code = + current->thread.gmap_addr; + vcpu->run->s390_ucontrol.pgm_code = 0x10; + rc = -EREMOTE; } else { - if (kvm_is_ucontrol(vcpu->kvm)) { - rc = SIE_INTERCEPT_UCONTROL; - } else { - VCPU_EVENT(vcpu, 3, "%s", "fault in sie instruction"); - trace_kvm_s390_sie_fault(vcpu); - rc = kvm_s390_inject_program_int(vcpu, PGM_ADDRESSING); - } + VCPU_EVENT(vcpu, 3, "%s", "fault in sie instruction"); + trace_kvm_s390_sie_fault(vcpu); + rc = kvm_s390_inject_program_int(vcpu, PGM_ADDRESSING); } memcpy(&vcpu->run->s.regs.gprs[14], &vcpu->arch.sie_block->gg14, 16); @@ -833,16 +835,6 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run) rc = -EINTR; } -#ifdef CONFIG_KVM_S390_UCONTROL - if (rc == SIE_INTERCEPT_UCONTROL) { - kvm_run->exit_reason = KVM_EXIT_S390_UCONTROL; - kvm_run->s390_ucontrol.trans_exc_code = - current->thread.gmap_addr; - kvm_run->s390_ucontrol.pgm_code = 0x10; - rc = 0; - } -#endif - if (rc == -EOPNOTSUPP) { /* intercept cannot be handled in-kernel, prepare kvm-run */ kvm_run->exit_reason = KVM_EXIT_S390_SIEIC; @@ -885,10 +877,11 @@ static int __guestcopy(struct kvm_vcpu *vcpu, u64 guestdest, void *from, * KVM_S390_STORE_STATUS_NOADDR: -> 0x1200 on 64 bit * KVM_S390_STORE_STATUS_PREFIXED: -> prefix */ -int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr) +int kvm_s390_store_status_unloaded(struct kvm_vcpu *vcpu, unsigned long addr) { unsigned char archmode = 1; int prefix; + u64 clkcomp; if (addr == KVM_S390_STORE_STATUS_NOADDR) { if (copy_to_guest_absolute(vcpu, 163ul, &archmode, 1)) @@ -903,15 +896,6 @@ int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr) } else prefix = 0; - /* - * The guest FPRS and ACRS are in the host FPRS/ACRS due to the lazy - * copying in vcpu load/put. Lets update our copies before we save - * it into the save area - */ - save_fp_ctl(&vcpu->arch.guest_fpregs.fpc); - save_fp_regs(vcpu->arch.guest_fpregs.fprs); - save_access_regs(vcpu->run->s.regs.acrs); - if (__guestcopy(vcpu, addr + offsetof(struct save_area, fp_regs), vcpu->arch.guest_fpregs.fprs, 128, prefix)) return -EFAULT; @@ -941,8 +925,9 @@ int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr) &vcpu->arch.sie_block->cputm, 8, prefix)) return -EFAULT; + clkcomp = vcpu->arch.sie_block->ckc >> 8; if (__guestcopy(vcpu, addr + offsetof(struct save_area, clk_cmp), - &vcpu->arch.sie_block->ckc, 8, prefix)) + &clkcomp, 8, prefix)) return -EFAULT; if (__guestcopy(vcpu, addr + offsetof(struct save_area, acc_regs), @@ -956,6 +941,20 @@ int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr) return 0; } +int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr) +{ + /* + * The guest FPRS and ACRS are in the host FPRS/ACRS due to the lazy + * copying in vcpu load/put. Lets update our copies before we save + * it into the save area + */ + save_fp_ctl(&vcpu->arch.guest_fpregs.fpc); + save_fp_regs(vcpu->arch.guest_fpregs.fprs); + save_access_regs(vcpu->run->s.regs.acrs); + + return kvm_s390_store_status_unloaded(vcpu, addr); +} + static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu, struct kvm_enable_cap *cap) { diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index b44912a32949..095cf51b16ec 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -19,16 +19,11 @@ #include <linux/kvm.h> #include <linux/kvm_host.h> -/* The current code can have up to 256 pages for virtio */ -#define VIRTIODESCSPACE (256ul * 4096ul) - typedef int (*intercept_handler_t)(struct kvm_vcpu *vcpu); /* declare vfacilities extern */ extern unsigned long *vfacilities; -/* negativ values are error codes, positive values for internal conditions */ -#define SIE_INTERCEPT_UCONTROL (1<<0) int kvm_handle_sie_intercept(struct kvm_vcpu *vcpu); #define VM_EVENT(d_kvm, d_loglevel, d_string, d_args...)\ @@ -133,7 +128,6 @@ int __must_check kvm_s390_inject_vm(struct kvm *kvm, int __must_check kvm_s390_inject_vcpu(struct kvm_vcpu *vcpu, struct kvm_s390_interrupt *s390int); int __must_check kvm_s390_inject_program_int(struct kvm_vcpu *vcpu, u16 code); -int __must_check kvm_s390_inject_sigp_stop(struct kvm_vcpu *vcpu, int action); struct kvm_s390_interrupt_info *kvm_s390_get_io_int(struct kvm *kvm, u64 cr6, u64 schid); @@ -150,8 +144,8 @@ int kvm_s390_handle_eb(struct kvm_vcpu *vcpu); int kvm_s390_handle_sigp(struct kvm_vcpu *vcpu); /* implemented in kvm-s390.c */ -int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, - unsigned long addr); +int kvm_s390_store_status_unloaded(struct kvm_vcpu *vcpu, unsigned long addr); +int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long addr); void s390_vcpu_block(struct kvm_vcpu *vcpu); void s390_vcpu_unblock(struct kvm_vcpu *vcpu); void exit_sie(struct kvm_vcpu *vcpu); diff --git a/arch/s390/kvm/priv.c b/arch/s390/kvm/priv.c index 2440602e6df1..75beea632a10 100644 --- a/arch/s390/kvm/priv.c +++ b/arch/s390/kvm/priv.c @@ -197,7 +197,7 @@ static int handle_tpi(struct kvm_vcpu *vcpu) if (addr & 3) return kvm_s390_inject_program_int(vcpu, PGM_SPECIFICATION); cc = 0; - inti = kvm_s390_get_io_int(vcpu->kvm, vcpu->run->s.regs.crs[6], 0); + inti = kvm_s390_get_io_int(vcpu->kvm, vcpu->arch.sie_block->gcr[6], 0); if (!inti) goto no_interrupt; cc = 1; @@ -275,7 +275,7 @@ static int handle_io_inst(struct kvm_vcpu *vcpu) return -EOPNOTSUPP; } else { /* - * Set condition code 3 to stop the guest from issueing channel + * Set condition code 3 to stop the guest from issuing channel * I/O instructions. */ kvm_s390_set_psw_cc(vcpu, 3); @@ -638,7 +638,6 @@ static int handle_pfmf(struct kvm_vcpu *vcpu) static const intercept_handler_t b9_handlers[256] = { [0x8d] = handle_epsw, - [0x9c] = handle_io_inst, [0xaf] = handle_pfmf, }; @@ -731,7 +730,6 @@ static int handle_lctlg(struct kvm_vcpu *vcpu) static const intercept_handler_t eb_handlers[256] = { [0x2f] = handle_lctlg, - [0x8a] = handle_io_inst, }; int kvm_s390_handle_eb(struct kvm_vcpu *vcpu) diff --git a/arch/s390/kvm/sigp.c b/arch/s390/kvm/sigp.c index bec398c57acf..87c2b3a3bd3e 100644 --- a/arch/s390/kvm/sigp.c +++ b/arch/s390/kvm/sigp.c @@ -1,7 +1,7 @@ /* * handling interprocessor communication * - * Copyright IBM Corp. 2008, 2009 + * Copyright IBM Corp. 2008, 2013 * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License (version 2 only) @@ -89,6 +89,37 @@ unlock: return rc; } +static int __sigp_conditional_emergency(struct kvm_vcpu *vcpu, u16 cpu_addr, + u16 asn, u64 *reg) +{ + struct kvm_vcpu *dst_vcpu = NULL; + const u64 psw_int_mask = PSW_MASK_IO | PSW_MASK_EXT; + u16 p_asn, s_asn; + psw_t *psw; + u32 flags; + + if (cpu_addr < KVM_MAX_VCPUS) + dst_vcpu = kvm_get_vcpu(vcpu->kvm, cpu_addr); + if (!dst_vcpu) + return SIGP_CC_NOT_OPERATIONAL; + flags = atomic_read(&dst_vcpu->arch.sie_block->cpuflags); + psw = &dst_vcpu->arch.sie_block->gpsw; + p_asn = dst_vcpu->arch.sie_block->gcr[4] & 0xffff; /* Primary ASN */ + s_asn = dst_vcpu->arch.sie_block->gcr[3] & 0xffff; /* Secondary ASN */ + + /* Deliver the emergency signal? */ + if (!(flags & CPUSTAT_STOPPED) + || (psw->mask & psw_int_mask) != psw_int_mask + || ((flags & CPUSTAT_WAIT) && psw->addr != 0) + || (!(flags & CPUSTAT_WAIT) && (asn == p_asn || asn == s_asn))) { + return __sigp_emergency(vcpu, cpu_addr); + } else { + *reg &= 0xffffffff00000000UL; + *reg |= SIGP_STATUS_INCORRECT_STATE; + return SIGP_CC_STATUS_STORED; + } +} + static int __sigp_external_call(struct kvm_vcpu *vcpu, u16 cpu_addr) { struct kvm_s390_float_interrupt *fi = &vcpu->kvm->arch.float_int; @@ -130,6 +161,7 @@ unlock: static int __inject_sigp_stop(struct kvm_s390_local_interrupt *li, int action) { struct kvm_s390_interrupt_info *inti; + int rc = SIGP_CC_ORDER_CODE_ACCEPTED; inti = kzalloc(sizeof(*inti), GFP_ATOMIC); if (!inti) @@ -139,6 +171,8 @@ static int __inject_sigp_stop(struct kvm_s390_local_interrupt *li, int action) spin_lock_bh(&li->lock); if ((atomic_read(li->cpuflags) & CPUSTAT_STOPPED)) { kfree(inti); + if ((action & ACTION_STORE_ON_STOP) != 0) + rc = -ESHUTDOWN; goto out; } list_add_tail(&inti->list, &li->list); @@ -150,7 +184,7 @@ static int __inject_sigp_stop(struct kvm_s390_local_interrupt *li, int action) out: spin_unlock_bh(&li->lock); - return SIGP_CC_ORDER_CODE_ACCEPTED; + return rc; } static int __sigp_stop(struct kvm_vcpu *vcpu, u16 cpu_addr, int action) @@ -174,13 +208,17 @@ static int __sigp_stop(struct kvm_vcpu *vcpu, u16 cpu_addr, int action) unlock: spin_unlock(&fi->lock); VCPU_EVENT(vcpu, 4, "sent sigp stop to cpu %x", cpu_addr); - return rc; -} -int kvm_s390_inject_sigp_stop(struct kvm_vcpu *vcpu, int action) -{ - struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; - return __inject_sigp_stop(li, action); + if ((action & ACTION_STORE_ON_STOP) != 0 && rc == -ESHUTDOWN) { + /* If the CPU has already been stopped, we still have + * to save the status when doing stop-and-store. This + * has to be done after unlocking all spinlocks. */ + struct kvm_vcpu *dst_vcpu = kvm_get_vcpu(vcpu->kvm, cpu_addr); + rc = kvm_s390_store_status_unloaded(dst_vcpu, + KVM_S390_STORE_STATUS_NOADDR); + } + + return rc; } static int __sigp_set_arch(struct kvm_vcpu *vcpu, u32 parameter) @@ -262,6 +300,37 @@ out_fi: return rc; } +static int __sigp_store_status_at_addr(struct kvm_vcpu *vcpu, u16 cpu_id, + u32 addr, u64 *reg) +{ + struct kvm_vcpu *dst_vcpu = NULL; + int flags; + int rc; + + if (cpu_id < KVM_MAX_VCPUS) + dst_vcpu = kvm_get_vcpu(vcpu->kvm, cpu_id); + if (!dst_vcpu) + return SIGP_CC_NOT_OPERATIONAL; + + spin_lock_bh(&dst_vcpu->arch.local_int.lock); + flags = atomic_read(dst_vcpu->arch.local_int.cpuflags); + spin_unlock_bh(&dst_vcpu->arch.local_int.lock); + if (!(flags & CPUSTAT_STOPPED)) { + *reg &= 0xffffffff00000000UL; + *reg |= SIGP_STATUS_INCORRECT_STATE; + return SIGP_CC_STATUS_STORED; + } + + addr &= 0x7ffffe00; + rc = kvm_s390_store_status_unloaded(dst_vcpu, addr); + if (rc == -EFAULT) { + *reg &= 0xffffffff00000000UL; + *reg |= SIGP_STATUS_INVALID_PARAMETER; + rc = SIGP_CC_STATUS_STORED; + } + return rc; +} + static int __sigp_sense_running(struct kvm_vcpu *vcpu, u16 cpu_addr, u64 *reg) { @@ -294,7 +363,8 @@ static int __sigp_sense_running(struct kvm_vcpu *vcpu, u16 cpu_addr, return rc; } -static int __sigp_restart(struct kvm_vcpu *vcpu, u16 cpu_addr) +/* Test whether the destination CPU is available and not busy */ +static int sigp_check_callable(struct kvm_vcpu *vcpu, u16 cpu_addr) { struct kvm_s390_float_interrupt *fi = &vcpu->kvm->arch.float_int; struct kvm_s390_local_interrupt *li; @@ -313,9 +383,6 @@ static int __sigp_restart(struct kvm_vcpu *vcpu, u16 cpu_addr) spin_lock_bh(&li->lock); if (li->action_bits & ACTION_STOP_ON_STOP) rc = SIGP_CC_BUSY; - else - VCPU_EVENT(vcpu, 4, "sigp restart %x to handle userspace", - cpu_addr); spin_unlock_bh(&li->lock); out: spin_unlock(&fi->lock); @@ -366,6 +433,10 @@ int kvm_s390_handle_sigp(struct kvm_vcpu *vcpu) rc = __sigp_stop(vcpu, cpu_addr, ACTION_STORE_ON_STOP | ACTION_STOP_ON_STOP); break; + case SIGP_STORE_STATUS_AT_ADDRESS: + rc = __sigp_store_status_at_addr(vcpu, cpu_addr, parameter, + &vcpu->run->s.regs.gprs[r1]); + break; case SIGP_SET_ARCHITECTURE: vcpu->stat.instruction_sigp_arch++; rc = __sigp_set_arch(vcpu, parameter); @@ -375,17 +446,31 @@ int kvm_s390_handle_sigp(struct kvm_vcpu *vcpu) rc = __sigp_set_prefix(vcpu, cpu_addr, parameter, &vcpu->run->s.regs.gprs[r1]); break; + case SIGP_COND_EMERGENCY_SIGNAL: + rc = __sigp_conditional_emergency(vcpu, cpu_addr, parameter, + &vcpu->run->s.regs.gprs[r1]); + break; case SIGP_SENSE_RUNNING: vcpu->stat.instruction_sigp_sense_running++; rc = __sigp_sense_running(vcpu, cpu_addr, &vcpu->run->s.regs.gprs[r1]); break; + case SIGP_START: + rc = sigp_check_callable(vcpu, cpu_addr); + if (rc == SIGP_CC_ORDER_CODE_ACCEPTED) + rc = -EOPNOTSUPP; /* Handle START in user space */ + break; case SIGP_RESTART: vcpu->stat.instruction_sigp_restart++; - rc = __sigp_restart(vcpu, cpu_addr); - if (rc == SIGP_CC_BUSY) - break; - /* user space must know about restart */ + rc = sigp_check_callable(vcpu, cpu_addr); + if (rc == SIGP_CC_ORDER_CODE_ACCEPTED) { + VCPU_EVENT(vcpu, 4, + "sigp restart %x to handle userspace", + cpu_addr); + /* user space must know about restart */ + rc = -EOPNOTSUPP; + } + break; default: return -EOPNOTSUPP; } @@ -393,7 +478,6 @@ int kvm_s390_handle_sigp(struct kvm_vcpu *vcpu) if (rc < 0) return rc; - vcpu->arch.sie_block->gpsw.mask &= ~(3ul << 44); - vcpu->arch.sie_block->gpsw.mask |= (rc & 3ul) << 44; + kvm_s390_set_psw_cc(vcpu, rc); return 0; } diff --git a/arch/s390/kvm/trace.h b/arch/s390/kvm/trace.h index 0c991c6748ab..3db76b2daed7 100644 --- a/arch/s390/kvm/trace.h +++ b/arch/s390/kvm/trace.h @@ -175,6 +175,7 @@ TRACE_EVENT(kvm_s390_intercept_validity, {SIGP_STOP_AND_STORE_STATUS, "stop and store status"}, \ {SIGP_SET_ARCHITECTURE, "set architecture"}, \ {SIGP_SET_PREFIX, "set prefix"}, \ + {SIGP_STORE_STATUS_AT_ADDRESS, "store status at addr"}, \ {SIGP_SENSE_RUNNING, "sense running"}, \ {SIGP_RESTART, "restart"} diff --git a/arch/s390/lib/uaccess.h b/arch/s390/lib/uaccess.h index 315dbe09983e..b1a22173d027 100644 --- a/arch/s390/lib/uaccess.h +++ b/arch/s390/lib/uaccess.h @@ -6,15 +6,6 @@ #ifndef __ARCH_S390_LIB_UACCESS_H #define __ARCH_S390_LIB_UACCESS_H -extern size_t copy_from_user_std(size_t, const void __user *, void *); -extern size_t copy_to_user_std(size_t, void __user *, const void *); -extern size_t strnlen_user_std(size_t, const char __user *); -extern size_t strncpy_from_user_std(size_t, const char __user *, char *); -extern int futex_atomic_cmpxchg_std(u32 *, u32 __user *, u32, u32); -extern int futex_atomic_op_std(int, u32 __user *, int, int *); - -extern size_t copy_from_user_pt(size_t, const void __user *, void *); -extern size_t copy_to_user_pt(size_t, void __user *, const void *); extern int futex_atomic_op_pt(int, u32 __user *, int, int *); extern int futex_atomic_cmpxchg_pt(u32 *, u32 __user *, u32, u32); diff --git a/arch/s390/lib/uaccess_pt.c b/arch/s390/lib/uaccess_pt.c index dbdab3e7a1a6..61ebcc9ccb34 100644 --- a/arch/s390/lib/uaccess_pt.c +++ b/arch/s390/lib/uaccess_pt.c @@ -74,8 +74,8 @@ static size_t copy_in_kernel(size_t count, void __user *to, /* * Returns kernel address for user virtual address. If the returned address is - * >= -4095 (IS_ERR_VALUE(x) returns true), a fault has occured and the address - * contains the (negative) exception code. + * >= -4095 (IS_ERR_VALUE(x) returns true), a fault has occurred and the + * address contains the (negative) exception code. */ #ifdef CONFIG_64BIT @@ -153,6 +153,8 @@ static __always_inline size_t __user_copy_pt(unsigned long uaddr, void *kptr, unsigned long offset, done, size, kaddr; void *from, *to; + if (!mm) + return n; done = 0; retry: spin_lock(&mm->page_table_lock); @@ -209,7 +211,7 @@ fault: return 0; } -size_t copy_from_user_pt(size_t n, const void __user *from, void *to) +static size_t copy_from_user_pt(size_t n, const void __user *from, void *to) { size_t rc; @@ -221,7 +223,7 @@ size_t copy_from_user_pt(size_t n, const void __user *from, void *to) return rc; } -size_t copy_to_user_pt(size_t n, void __user *to, const void *from) +static size_t copy_to_user_pt(size_t n, void __user *to, const void *from) { if (segment_eq(get_fs(), KERNEL_DS)) return copy_in_kernel(n, to, (void __user *) from); @@ -262,6 +264,8 @@ static size_t strnlen_user_pt(size_t count, const char __user *src) return 0; if (segment_eq(get_fs(), KERNEL_DS)) return strnlen_kernel(count, src); + if (!mm) + return 0; done = 0; retry: spin_lock(&mm->page_table_lock); @@ -323,6 +327,8 @@ static size_t copy_in_user_pt(size_t n, void __user *to, if (segment_eq(get_fs(), KERNEL_DS)) return copy_in_kernel(n, to, from); + if (!mm) + return n; done = 0; retry: spin_lock(&mm->page_table_lock); @@ -411,6 +417,8 @@ int futex_atomic_op_pt(int op, u32 __user *uaddr, int oparg, int *old) if (segment_eq(get_fs(), KERNEL_DS)) return __futex_atomic_op_pt(op, uaddr, oparg, old); + if (unlikely(!current->mm)) + return -EFAULT; spin_lock(¤t->mm->page_table_lock); uaddr = (u32 __force __user *) __dat_user_addr((__force unsigned long) uaddr, 1); @@ -448,6 +456,8 @@ int futex_atomic_cmpxchg_pt(u32 *uval, u32 __user *uaddr, if (segment_eq(get_fs(), KERNEL_DS)) return __futex_atomic_cmpxchg_pt(uval, uaddr, oldval, newval); + if (unlikely(!current->mm)) + return -EFAULT; spin_lock(¤t->mm->page_table_lock); uaddr = (u32 __force __user *) __dat_user_addr((__force unsigned long) uaddr, 1); diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c index e794c88f699a..3584ed9b20a1 100644 --- a/arch/s390/mm/pgtable.c +++ b/arch/s390/mm/pgtable.c @@ -293,7 +293,7 @@ static int gmap_alloc_table(struct gmap *gmap, * @addr: address in the guest address space * @len: length of the memory area to unmap * - * Returns 0 if the unmap succeded, -EINVAL if not. + * Returns 0 if the unmap succeeded, -EINVAL if not. */ int gmap_unmap_segment(struct gmap *gmap, unsigned long to, unsigned long len) { @@ -344,7 +344,7 @@ EXPORT_SYMBOL_GPL(gmap_unmap_segment); * @from: source address in the parent address space * @to: target address in the guest address space * - * Returns 0 if the mmap succeded, -EINVAL or -ENOMEM if not. + * Returns 0 if the mmap succeeded, -EINVAL or -ENOMEM if not. */ int gmap_map_segment(struct gmap *gmap, unsigned long from, unsigned long to, unsigned long len) diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c index 16871da37371..708d60e40066 100644 --- a/arch/s390/net/bpf_jit_comp.c +++ b/arch/s390/net/bpf_jit_comp.c @@ -368,14 +368,16 @@ static int bpf_jit_insn(struct bpf_jit *jit, struct sock_filter *filter, EMIT4_PCREL(0xa7840000, (jit->ret0_ip - jit->prg)); /* lhi %r4,0 */ EMIT4(0xa7480000); - /* dr %r4,%r12 */ - EMIT2(0x1d4c); + /* dlr %r4,%r12 */ + EMIT4(0xb997004c); break; - case BPF_S_ALU_DIV_K: /* A = reciprocal_divide(A, K) */ - /* m %r4,<d(K)>(%r13) */ - EMIT4_DISP(0x5c40d000, EMIT_CONST(K)); - /* lr %r5,%r4 */ - EMIT2(0x1854); + case BPF_S_ALU_DIV_K: /* A /= K */ + if (K == 1) + break; + /* lhi %r4,0 */ + EMIT4(0xa7480000); + /* dl %r4,<d(K)>(%r13) */ + EMIT6_DISP(0xe340d000, 0x0097, EMIT_CONST(K)); break; case BPF_S_ALU_MOD_X: /* A %= X */ jit->seen |= SEEN_XREG | SEEN_RET0; @@ -385,16 +387,21 @@ static int bpf_jit_insn(struct bpf_jit *jit, struct sock_filter *filter, EMIT4_PCREL(0xa7840000, (jit->ret0_ip - jit->prg)); /* lhi %r4,0 */ EMIT4(0xa7480000); - /* dr %r4,%r12 */ - EMIT2(0x1d4c); + /* dlr %r4,%r12 */ + EMIT4(0xb997004c); /* lr %r5,%r4 */ EMIT2(0x1854); break; case BPF_S_ALU_MOD_K: /* A %= K */ + if (K == 1) { + /* lhi %r5,0 */ + EMIT4(0xa7580000); + break; + } /* lhi %r4,0 */ EMIT4(0xa7480000); - /* d %r4,<d(K)>(%r13) */ - EMIT4_DISP(0x5d40d000, EMIT_CONST(K)); + /* dl %r4,<d(K)>(%r13) */ + EMIT6_DISP(0xe340d000, 0x0097, EMIT_CONST(K)); /* lr %r5,%r4 */ EMIT2(0x1854); break; diff --git a/arch/s390/oprofile/hwsampler.c b/arch/s390/oprofile/hwsampler.c index 231cecafc2f1..a32c96761eab 100644 --- a/arch/s390/oprofile/hwsampler.c +++ b/arch/s390/oprofile/hwsampler.c @@ -26,9 +26,6 @@ #define MAX_NUM_SDB 511 #define MIN_NUM_SDB 1 -#define ALERT_REQ_MASK 0x4000000000000000ul -#define BUFFER_FULL_MASK 0x8000000000000000ul - DECLARE_PER_CPU(struct hws_cpu_buffer, sampler_cpu_buffer); struct hws_execute_parms { @@ -44,6 +41,7 @@ static DEFINE_MUTEX(hws_sem_oom); static unsigned char hws_flush_all; static unsigned int hws_oom; +static unsigned int hws_alert; static struct workqueue_struct *hws_wq; static unsigned int hws_state; @@ -65,43 +63,6 @@ static unsigned long interval; static unsigned long min_sampler_rate; static unsigned long max_sampler_rate; -static int ssctl(void *buffer) -{ - int cc; - - /* set in order to detect a program check */ - cc = 1; - - asm volatile( - "0: .insn s,0xB2870000,0(%1)\n" - "1: ipm %0\n" - " srl %0,28\n" - "2:\n" - EX_TABLE(0b, 2b) EX_TABLE(1b, 2b) - : "+d" (cc), "+a" (buffer) - : "m" (*((struct hws_ssctl_request_block *)buffer)) - : "cc", "memory"); - - return cc ? -EINVAL : 0 ; -} - -static int qsi(void *buffer) -{ - int cc; - cc = 1; - - asm volatile( - "0: .insn s,0xB2860000,0(%1)\n" - "1: lhi %0,0\n" - "2:\n" - EX_TABLE(0b, 2b) EX_TABLE(1b, 2b) - : "=d" (cc), "+a" (buffer) - : "m" (*((struct hws_qsi_info_block *)buffer)) - : "cc", "memory"); - - return cc ? -EINVAL : 0; -} - static void execute_qsi(void *parms) { struct hws_execute_parms *ep = parms; @@ -113,7 +74,7 @@ static void execute_ssctl(void *parms) { struct hws_execute_parms *ep = parms; - ep->rc = ssctl(ep->buffer); + ep->rc = lsctl(ep->buffer); } static int smp_ctl_ssctl_stop(int cpu) @@ -214,17 +175,6 @@ static int smp_ctl_qsi(int cpu) return ep.rc; } -static inline unsigned long *trailer_entry_ptr(unsigned long v) -{ - void *ret; - - ret = (void *)v; - ret += PAGE_SIZE; - ret -= sizeof(struct hws_trailer_entry); - - return (unsigned long *) ret; -} - static void hws_ext_handler(struct ext_code ext_code, unsigned int param32, unsigned long param64) { @@ -233,6 +183,9 @@ static void hws_ext_handler(struct ext_code ext_code, if (!(param32 & CPU_MF_INT_SF_MASK)) return; + if (!hws_alert) + return; + inc_irq_stat(IRQEXT_CMS); atomic_xchg(&cb->ext_params, atomic_read(&cb->ext_params) | param32); @@ -256,16 +209,6 @@ static void init_all_cpu_buffers(void) } } -static int is_link_entry(unsigned long *s) -{ - return *s & 0x1ul ? 1 : 0; -} - -static unsigned long *get_next_sdbt(unsigned long *s) -{ - return (unsigned long *) (*s & ~0x1ul); -} - static int prepare_cpu_buffers(void) { int cpu; @@ -353,7 +296,7 @@ static int allocate_sdbt(int cpu) } *sdbt = sdb; trailer = trailer_entry_ptr(*sdbt); - *trailer = ALERT_REQ_MASK; + *trailer = SDB_TE_ALERT_REQ_MASK; sdbt++; mutex_unlock(&hws_sem_oom); } @@ -829,7 +772,7 @@ static void worker_on_interrupt(unsigned int cpu) trailer = trailer_entry_ptr(*sdbt); /* leave loop if no more work to do */ - if (!(*trailer & BUFFER_FULL_MASK)) { + if (!(*trailer & SDB_TE_BUFFER_FULL_MASK)) { done = 1; if (!hws_flush_all) continue; @@ -856,7 +799,7 @@ static void worker_on_interrupt(unsigned int cpu) static void add_samples_to_oprofile(unsigned int cpu, unsigned long *sdbt, unsigned long *dear) { - struct hws_data_entry *sample_data_ptr; + struct hws_basic_entry *sample_data_ptr; unsigned long *trailer; trailer = trailer_entry_ptr(*sdbt); @@ -866,7 +809,7 @@ static void add_samples_to_oprofile(unsigned int cpu, unsigned long *sdbt, trailer = dear; } - sample_data_ptr = (struct hws_data_entry *)(*sdbt); + sample_data_ptr = (struct hws_basic_entry *)(*sdbt); while ((unsigned long *)sample_data_ptr < trailer) { struct pt_regs *regs = NULL; @@ -1002,6 +945,7 @@ int hwsampler_deallocate(void) goto deallocate_exit; irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT); + hws_alert = 0; deallocate_sdbt(); hws_state = HWS_DEALLOCATED; @@ -1116,6 +1060,7 @@ int hwsampler_shutdown(void) if (hws_state == HWS_STOPPED) { irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT); + hws_alert = 0; deallocate_sdbt(); } if (hws_wq) { @@ -1190,6 +1135,7 @@ start_all_exit: hws_oom = 1; hws_flush_all = 0; /* now let them in, 1407 CPUMF external interrupts */ + hws_alert = 1; irq_subclass_register(IRQ_SUBCLASS_MEASUREMENT_ALERT); return 0; diff --git a/arch/s390/oprofile/hwsampler.h b/arch/s390/oprofile/hwsampler.h index 0022e1ebfbde..a483d06f2fa7 100644 --- a/arch/s390/oprofile/hwsampler.h +++ b/arch/s390/oprofile/hwsampler.h @@ -9,27 +9,7 @@ #define HWSAMPLER_H_ #include <linux/workqueue.h> - -struct hws_qsi_info_block /* QUERY SAMPLING information block */ -{ /* Bit(s) */ - unsigned int b0_13:14; /* 0-13: zeros */ - unsigned int as:1; /* 14: sampling authorisation control*/ - unsigned int b15_21:7; /* 15-21: zeros */ - unsigned int es:1; /* 22: sampling enable control */ - unsigned int b23_29:7; /* 23-29: zeros */ - unsigned int cs:1; /* 30: sampling activation control */ - unsigned int:1; /* 31: reserved */ - unsigned int bsdes:16; /* 4-5: size of sampling entry */ - unsigned int:16; /* 6-7: reserved */ - unsigned long min_sampl_rate; /* 8-15: minimum sampling interval */ - unsigned long max_sampl_rate; /* 16-23: maximum sampling interval*/ - unsigned long tear; /* 24-31: TEAR contents */ - unsigned long dear; /* 32-39: DEAR contents */ - unsigned int rsvrd0; /* 40-43: reserved */ - unsigned int cpu_speed; /* 44-47: CPU speed */ - unsigned long long rsvrd1; /* 48-55: reserved */ - unsigned long long rsvrd2; /* 56-63: reserved */ -}; +#include <asm/cpu_mf.h> struct hws_ssctl_request_block /* SET SAMPLING CONTROLS req block */ { /* bytes 0 - 7 Bit(s) */ @@ -68,36 +48,6 @@ struct hws_cpu_buffer { unsigned int stop_mode:1; }; -struct hws_data_entry { - unsigned int def:16; /* 0-15 Data Entry Format */ - unsigned int R:4; /* 16-19 reserved */ - unsigned int U:4; /* 20-23 Number of unique instruct. */ - unsigned int z:2; /* zeros */ - unsigned int T:1; /* 26 PSW DAT mode */ - unsigned int W:1; /* 27 PSW wait state */ - unsigned int P:1; /* 28 PSW Problem state */ - unsigned int AS:2; /* 29-30 PSW address-space control */ - unsigned int I:1; /* 31 entry valid or invalid */ - unsigned int:16; - unsigned int prim_asn:16; /* primary ASN */ - unsigned long long ia; /* Instruction Address */ - unsigned long long gpp; /* Guest Program Parameter */ - unsigned long long hpp; /* Host Program Parameter */ -}; - -struct hws_trailer_entry { - unsigned int f:1; /* 0 - Block Full Indicator */ - unsigned int a:1; /* 1 - Alert request control */ - unsigned long:62; /* 2 - 63: Reserved */ - unsigned long overflow; /* 64 - sample Overflow count */ - unsigned long timestamp; /* 16 - time-stamp */ - unsigned long timestamp1; /* */ - unsigned long reserved1; /* 32 -Reserved */ - unsigned long reserved2; /* */ - unsigned long progusage1; /* 48 - reserved for programming use */ - unsigned long progusage2; /* */ -}; - int hwsampler_setup(void); int hwsampler_shutdown(void); int hwsampler_allocate(unsigned long sdbt, unsigned long sdb); diff --git a/arch/s390/oprofile/init.c b/arch/s390/oprofile/init.c index 04e1b6a85362..9ffe645d5989 100644 --- a/arch/s390/oprofile/init.c +++ b/arch/s390/oprofile/init.c @@ -10,6 +10,7 @@ */ #include <linux/oprofile.h> +#include <linux/perf_event.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/fs.h> @@ -67,6 +68,21 @@ module_param_call(cpu_type, set_cpu_type, NULL, NULL, 0); MODULE_PARM_DESC(cpu_type, "Force legacy basic mode sampling" "(report cpu_type \"timer\""); +static int __oprofile_hwsampler_start(void) +{ + int retval; + + retval = hwsampler_allocate(oprofile_sdbt_blocks, oprofile_sdb_blocks); + if (retval) + return retval; + + retval = hwsampler_start_all(oprofile_hw_interval); + if (retval) + hwsampler_deallocate(); + + return retval; +} + static int oprofile_hwsampler_start(void) { int retval; @@ -76,13 +92,13 @@ static int oprofile_hwsampler_start(void) if (!hwsampler_running) return timer_ops.start(); - retval = hwsampler_allocate(oprofile_sdbt_blocks, oprofile_sdb_blocks); + retval = perf_reserve_sampling(); if (retval) return retval; - retval = hwsampler_start_all(oprofile_hw_interval); + retval = __oprofile_hwsampler_start(); if (retval) - hwsampler_deallocate(); + perf_release_sampling(); return retval; } @@ -96,6 +112,7 @@ static void oprofile_hwsampler_stop(void) hwsampler_stop_all(); hwsampler_deallocate(); + perf_release_sampling(); return; } diff --git a/arch/s390/pci/pci.c b/arch/s390/pci/pci.c index bf7c73d71eef..66670ff262a0 100644 --- a/arch/s390/pci/pci.c +++ b/arch/s390/pci/pci.c @@ -407,8 +407,8 @@ int arch_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type) struct msi_msg msg; int rc; - if (type != PCI_CAP_ID_MSIX && type != PCI_CAP_ID_MSI) - return -EINVAL; + if (type == PCI_CAP_ID_MSI && nvec > 1) + return 1; msi_vecs = min(nvec, ZPCI_MSI_VEC_MAX); msi_vecs = min_t(unsigned int, msi_vecs, CONFIG_PCI_NR_MSI); @@ -919,17 +919,23 @@ static void zpci_mem_exit(void) kmem_cache_destroy(zdev_fmb_cache); } -static unsigned int s390_pci_probe; +static unsigned int s390_pci_probe = 1; +static unsigned int s390_pci_initialized; char * __init pcibios_setup(char *str) { - if (!strcmp(str, "on")) { - s390_pci_probe = 1; + if (!strcmp(str, "off")) { + s390_pci_probe = 0; return NULL; } return str; } +bool zpci_is_enabled(void) +{ + return s390_pci_initialized; +} + static int __init pci_base_init(void) { int rc; @@ -961,6 +967,7 @@ static int __init pci_base_init(void) if (rc) goto out_find; + s390_pci_initialized = 1; return 0; out_find: @@ -978,5 +985,6 @@ subsys_initcall_sync(pci_base_init); void zpci_rescan(void) { - clp_rescan_pci_devices_simple(); + if (zpci_is_enabled()) + clp_rescan_pci_devices_simple(); } diff --git a/arch/s390/pci/pci_dma.c b/arch/s390/pci/pci_dma.c index 9b83d080902d..60c11a629d96 100644 --- a/arch/s390/pci/pci_dma.c +++ b/arch/s390/pci/pci_dma.c @@ -285,7 +285,7 @@ static dma_addr_t s390_dma_map_pages(struct device *dev, struct page *page, flags |= ZPCI_TABLE_PROTECTED; if (!dma_update_trans(zdev, pa, dma_addr, size, flags)) { - atomic64_add(nr_pages, (atomic64_t *) &zdev->fmb->mapped_pages); + atomic64_add(nr_pages, &zdev->fmb->mapped_pages); return dma_addr + (offset & ~PAGE_MASK); } @@ -313,7 +313,7 @@ static void s390_dma_unmap_pages(struct device *dev, dma_addr_t dma_addr, zpci_err_hex(&dma_addr, sizeof(dma_addr)); } - atomic64_add(npages, (atomic64_t *) &zdev->fmb->unmapped_pages); + atomic64_add(npages, &zdev->fmb->unmapped_pages); iommu_page_index = (dma_addr - zdev->start_dma) >> PAGE_SHIFT; dma_free_iommu(zdev, iommu_page_index, npages); } @@ -332,7 +332,6 @@ static void *s390_dma_alloc(struct device *dev, size_t size, if (!page) return NULL; - atomic64_add(size / PAGE_SIZE, (atomic64_t *) &zdev->fmb->allocated_pages); pa = page_to_phys(page); memset((void *) pa, 0, size); @@ -343,6 +342,7 @@ static void *s390_dma_alloc(struct device *dev, size_t size, return NULL; } + atomic64_add(size / PAGE_SIZE, &zdev->fmb->allocated_pages); if (dma_handle) *dma_handle = map; return (void *) pa; @@ -352,8 +352,11 @@ static void s390_dma_free(struct device *dev, size_t size, void *pa, dma_addr_t dma_handle, struct dma_attrs *attrs) { - s390_dma_unmap_pages(dev, dma_handle, PAGE_ALIGN(size), - DMA_BIDIRECTIONAL, NULL); + struct zpci_dev *zdev = get_zdev(to_pci_dev(dev)); + + size = PAGE_ALIGN(size); + atomic64_sub(size / PAGE_SIZE, &zdev->fmb->allocated_pages); + s390_dma_unmap_pages(dev, dma_handle, size, DMA_BIDIRECTIONAL, NULL); free_pages((unsigned long) pa, get_order(size)); } diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c index 069607209a30..01e251b1da0c 100644 --- a/arch/s390/pci/pci_event.c +++ b/arch/s390/pci/pci_event.c @@ -43,9 +43,8 @@ struct zpci_ccdf_avail { u16 pec; /* PCI event code */ } __packed; -void zpci_event_error(void *data) +static void __zpci_event_error(struct zpci_ccdf_err *ccdf) { - struct zpci_ccdf_err *ccdf = data; struct zpci_dev *zdev = get_zdev_by_fid(ccdf->fid); zpci_err("error CCDF:\n"); @@ -58,9 +57,14 @@ void zpci_event_error(void *data) pci_name(zdev->pdev), ccdf->pec, ccdf->fid); } -void zpci_event_availability(void *data) +void zpci_event_error(void *data) +{ + if (zpci_is_enabled()) + __zpci_event_error(data); +} + +static void __zpci_event_availability(struct zpci_ccdf_avail *ccdf) { - struct zpci_ccdf_avail *ccdf = data; struct zpci_dev *zdev = get_zdev_by_fid(ccdf->fid); struct pci_dev *pdev = zdev ? zdev->pdev : NULL; int ret; @@ -99,8 +103,12 @@ void zpci_event_availability(void *data) break; case 0x0304: /* Configured -> Standby */ - if (pdev) + if (pdev) { + /* Give the driver a hint that the function is + * already unusable. */ + pdev->error_state = pci_channel_io_perm_failure; pci_stop_and_remove_bus_device(pdev); + } zdev->fh = ccdf->fh; zpci_disable_device(zdev); @@ -110,6 +118,8 @@ void zpci_event_availability(void *data) clp_rescan_pci_devices(); break; case 0x0308: /* Standby -> Reserved */ + if (!zdev) + break; pci_stop_root_bus(zdev->bus); pci_remove_root_bus(zdev->bus); break; @@ -117,3 +127,9 @@ void zpci_event_availability(void *data) break; } } + +void zpci_event_availability(void *data) +{ + if (zpci_is_enabled()) + __zpci_event_availability(data); +} |