diff options
| -rw-r--r-- | Documentation/admin-guide/bootconfig.rst | 11 | ||||
| -rw-r--r-- | arch/x86/kernel/ftrace.c | 14 | ||||
| -rw-r--r-- | arch/x86/kernel/ftrace_64.S | 29 | ||||
| -rw-r--r-- | include/linux/kprobes.h | 1 | ||||
| -rw-r--r-- | include/linux/ring_buffer.h | 1 | ||||
| -rw-r--r-- | include/linux/trace.h | 1 | ||||
| -rw-r--r-- | include/linux/tracepoint.h | 11 | ||||
| -rw-r--r-- | include/trace/trace_events.h | 19 | ||||
| -rw-r--r-- | init/main.c | 14 | ||||
| -rw-r--r-- | kernel/kprobes.c | 24 | ||||
| -rw-r--r-- | kernel/trace/ftrace.c | 34 | ||||
| -rw-r--r-- | kernel/trace/ring_buffer.c | 694 | ||||
| -rw-r--r-- | kernel/trace/trace.c | 80 | ||||
| -rw-r--r-- | kernel/trace/trace.h | 9 | ||||
| -rw-r--r-- | kernel/trace/trace_events.c | 4 | ||||
| -rw-r--r-- | kernel/trace/trace_hwlat.c | 6 | ||||
| -rw-r--r-- | kernel/trace/trace_output.c | 14 | ||||
| -rw-r--r-- | kernel/trace/trace_uprobe.c | 1 | ||||
| -rw-r--r-- | lib/bootconfig.c | 33 | ||||
| -rw-r--r-- | tools/bootconfig/samples/bad-override.bconf | 3 | ||||
| -rw-r--r-- | tools/bootconfig/samples/bad-override2.bconf | 3 | ||||
| -rw-r--r-- | tools/bootconfig/samples/good-override.bconf | 6 | ||||
| -rwxr-xr-x | tools/bootconfig/test-bootconfig.sh | 13 |
23 files changed, 775 insertions, 250 deletions
diff --git a/Documentation/admin-guide/bootconfig.rst b/Documentation/admin-guide/bootconfig.rst index d6b3b77a4129..a22024f9175e 100644 --- a/Documentation/admin-guide/bootconfig.rst +++ b/Documentation/admin-guide/bootconfig.rst @@ -71,6 +71,16 @@ For example,:: foo = bar, baz foo = qux # !ERROR! we can not re-define same key +If you want to update the value, you must use the override operator +``:=`` explicitly. For example:: + + foo = bar, baz + foo := qux + +then, the ``qux`` is assigned to ``foo`` key. This is useful for +overriding the default value by adding (partial) custom bootconfigs +without parsing the default bootconfig. + If you want to append the value to existing key as an array member, you can use ``+=`` operator. For example:: @@ -84,6 +94,7 @@ For example, following config is NOT allowed.:: foo = value1 foo.bar = value2 # !ERROR! subkey "bar" and value "value1" can NOT co-exist + foo.bar := value2 # !ERROR! even with the override operator, this is NOT allowed. Comments diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 51504566b3a6..7edbd5ee5ed4 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -286,6 +286,7 @@ extern void ftrace_regs_caller_ret(void); extern void ftrace_caller_end(void); extern void ftrace_caller_op_ptr(void); extern void ftrace_regs_caller_op_ptr(void); +extern void ftrace_regs_caller_jmp(void); /* movq function_trace_op(%rip), %rdx */ /* 0x48 0x8b 0x15 <offset-to-ftrace_trace_op (4 bytes)> */ @@ -316,6 +317,7 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size) unsigned long end_offset; unsigned long op_offset; unsigned long call_offset; + unsigned long jmp_offset; unsigned long offset; unsigned long npages; unsigned long size; @@ -333,11 +335,13 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size) end_offset = (unsigned long)ftrace_regs_caller_end; op_offset = (unsigned long)ftrace_regs_caller_op_ptr; call_offset = (unsigned long)ftrace_regs_call; + jmp_offset = (unsigned long)ftrace_regs_caller_jmp; } else { start_offset = (unsigned long)ftrace_caller; end_offset = (unsigned long)ftrace_caller_end; op_offset = (unsigned long)ftrace_caller_op_ptr; call_offset = (unsigned long)ftrace_call; + jmp_offset = 0; } size = end_offset - start_offset; @@ -367,10 +371,14 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size) if (WARN_ON(ret < 0)) goto fail; + /* No need to test direct calls on created trampolines */ if (ops->flags & FTRACE_OPS_FL_SAVE_REGS) { - ip = trampoline + (ftrace_regs_caller_ret - ftrace_regs_caller); - ret = copy_from_kernel_nofault(ip, (void *)retq, RET_SIZE); - if (WARN_ON(ret < 0)) + /* NOP the jnz 1f; but make sure it's a 2 byte jnz */ + ip = trampoline + (jmp_offset - start_offset); + if (WARN_ON(*(char *)ip != 0x75)) + goto fail; + ret = copy_from_kernel_nofault(ip, ideal_nops[2], 2); + if (ret < 0) goto fail; } diff --git a/arch/x86/kernel/ftrace_64.S b/arch/x86/kernel/ftrace_64.S index 083a3da7bb73..ac3d5f22fe64 100644 --- a/arch/x86/kernel/ftrace_64.S +++ b/arch/x86/kernel/ftrace_64.S @@ -241,22 +241,10 @@ SYM_INNER_LABEL(ftrace_regs_call, SYM_L_GLOBAL) */ movq ORIG_RAX(%rsp), %rax testq %rax, %rax - jz 1f +SYM_INNER_LABEL(ftrace_regs_caller_jmp, SYM_L_GLOBAL) + jnz 1f - /* Swap the flags with orig_rax */ - movq MCOUNT_REG_SIZE(%rsp), %rdi - movq %rdi, MCOUNT_REG_SIZE-8(%rsp) - movq %rax, MCOUNT_REG_SIZE(%rsp) - - restore_mcount_regs 8 - /* Restore flags */ - popfq - -SYM_INNER_LABEL(ftrace_regs_caller_ret, SYM_L_GLOBAL); - UNWIND_HINT_RET_OFFSET - jmp ftrace_epilogue - -1: restore_mcount_regs + restore_mcount_regs /* Restore flags */ popfq @@ -269,6 +257,17 @@ SYM_INNER_LABEL(ftrace_regs_caller_ret, SYM_L_GLOBAL); SYM_INNER_LABEL(ftrace_regs_caller_end, SYM_L_GLOBAL) jmp ftrace_epilogue + /* Swap the flags with orig_rax */ +1: movq MCOUNT_REG_SIZE(%rsp), %rdi + movq %rdi, MCOUNT_REG_SIZE-8(%rsp) + movq %rax, MCOUNT_REG_SIZE(%rsp) + + restore_mcount_regs 8 + /* Restore flags */ + popfq + UNWIND_HINT_RET_OFFSET + jmp ftrace_epilogue + SYM_FUNC_END(ftrace_regs_caller) diff --git a/include/linux/kprobes.h b/include/linux/kprobes.h index 45b8cdc9fad7..9be1bff4f586 100644 --- a/include/linux/kprobes.h +++ b/include/linux/kprobes.h @@ -227,7 +227,6 @@ extern int arch_prepare_kprobe(struct kprobe *p); extern void arch_arm_kprobe(struct kprobe *p); extern void arch_disarm_kprobe(struct kprobe *p); extern int arch_init_kprobes(void); -extern void show_registers(struct pt_regs *regs); extern void kprobes_inc_nmissed_count(struct kprobe *p); extern bool arch_within_kprobe_blacklist(unsigned long addr); extern int arch_populate_kprobe_blacklist(void); diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index c76b2f3b3ac4..136ea0997e6d 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -143,6 +143,7 @@ bool ring_buffer_iter_dropped(struct ring_buffer_iter *iter); unsigned long ring_buffer_size(struct trace_buffer *buffer, int cpu); void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu); +void ring_buffer_reset_online_cpus(struct trace_buffer *buffer); void ring_buffer_reset(struct trace_buffer *buffer); #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP diff --git a/include/linux/trace.h b/include/linux/trace.h index 7fd86d3c691f..36d255d66f88 100644 --- a/include/linux/trace.h +++ b/include/linux/trace.h @@ -29,6 +29,7 @@ struct trace_array; void trace_printk_init_buffers(void); int trace_array_printk(struct trace_array *tr, unsigned long ip, const char *fmt, ...); +int trace_array_init_printk(struct trace_array *tr); void trace_array_put(struct trace_array *tr); struct trace_array *trace_array_get_by_name(const char *name); int trace_array_destroy(struct trace_array *tr); diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h index a1fecf311621..598fec9f9dbf 100644 --- a/include/linux/tracepoint.h +++ b/include/linux/tracepoint.h @@ -116,8 +116,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p) #define __TRACEPOINT_ENTRY(name) \ static tracepoint_ptr_t __tracepoint_ptr_##name __used \ - __attribute__((section("__tracepoints_ptrs"))) = \ - &__tracepoint_##name + __section(__tracepoints_ptrs) = &__tracepoint_##name #endif #endif /* _LINUX_TRACEPOINT_H */ @@ -280,9 +279,9 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p) */ #define DEFINE_TRACE_FN(name, reg, unreg) \ static const char __tpstrtab_##name[] \ - __attribute__((section("__tracepoints_strings"))) = #name; \ - struct tracepoint __tracepoint_##name \ - __attribute__((section("__tracepoints"), used)) = \ + __section(__tracepoints_strings) = #name; \ + struct tracepoint __tracepoint_##name __used \ + __section(__tracepoints) = \ { __tpstrtab_##name, STATIC_KEY_INIT_FALSE, reg, unreg, NULL };\ __TRACEPOINT_ENTRY(name); @@ -361,7 +360,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p) static const char *___tp_str __tracepoint_string = str; \ ___tp_str; \ }) -#define __tracepoint_string __attribute__((section("__tracepoint_str"))) +#define __tracepoint_string __used __section(__tracepoint_str) #else /* * tracepoint_string() is used to save the string address for userspace diff --git a/include/trace/trace_events.h b/include/trace/trace_events.h index 502c7be50b8d..1bc3e7bba9a4 100644 --- a/include/trace/trace_events.h +++ b/include/trace/trace_events.h @@ -210,8 +210,7 @@ TRACE_MAKE_SYSTEM_STR(); #define DEFINE_EVENT(template, name, proto, args) #undef DEFINE_EVENT_PRINT -#define DEFINE_EVENT_PRINT(template, name, proto, args, print) \ - DEFINE_EVENT(template, name, PARAMS(proto), PARAMS(args)) +#define DEFINE_EVENT_PRINT(template, name, proto, args, print) #undef TRACE_EVENT_FLAGS #define TRACE_EVENT_FLAGS(event, flag) @@ -443,12 +442,8 @@ static struct trace_event_fields trace_event_fields_##call[] = { \ tstruct \ {} }; -#undef DEFINE_EVENT -#define DEFINE_EVENT(template, name, proto, args) - #undef DEFINE_EVENT_PRINT -#define DEFINE_EVENT_PRINT(template, name, proto, args, print) \ - DEFINE_EVENT(template, name, PARAMS(proto), PARAMS(args)) +#define DEFINE_EVENT_PRINT(template, name, proto, args, print) #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) @@ -523,13 +518,6 @@ static inline notrace int trace_event_get_offsets_##call( \ return __data_size; \ } -#undef DEFINE_EVENT -#define DEFINE_EVENT(template, name, proto, args) - -#undef DEFINE_EVENT_PRINT -#define DEFINE_EVENT_PRINT(template, name, proto, args, print) \ - DEFINE_EVENT(template, name, PARAMS(proto), PARAMS(args)) - #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) /* @@ -721,9 +709,6 @@ static inline void ftrace_test_probe_##call(void) \ check_trace_callback_type_##call(trace_event_raw_event_##template); \ } -#undef DEFINE_EVENT_PRINT -#define DEFINE_EVENT_PRINT(template, name, proto, args, print) - #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) #undef __entry diff --git a/init/main.c b/init/main.c index de2f9fa6bb4a..ae78fb68d231 100644 --- a/init/main.c +++ b/init/main.c @@ -388,8 +388,6 @@ static int __init bootconfig_params(char *param, char *val, { if (strcmp(param, "bootconfig") == 0) { bootconfig_found = true; - } else if (strcmp(param, "--") == 0) { - initargs_found = true; } return 0; } @@ -400,19 +398,23 @@ static void __init setup_boot_config(const char *cmdline) const char *msg; int pos; u32 size, csum; - char *data, *copy; + char *data, *copy, *err; int ret; /* Cut out the bootconfig data even if we have no bootconfig option */ data = get_boot_config_from_initrd(&size, &csum); strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); - parse_args("bootconfig", tmp_cmdline, NULL, 0, 0, 0, NULL, - bootconfig_params); + err = parse_args("bootconfig", tmp_cmdline, NULL, 0, 0, 0, NULL, + bootconfig_params); - if (!bootconfig_found) + if (IS_ERR(err) || !bootconfig_found) return; + /* parse_args() stops at '--' and returns an address */ + if (err) + initargs_found = true; + if (!data) { pr_err("'bootconfig' found on command line, but no bootconfig found\n"); return; diff --git a/kernel/kprobes.c b/kernel/kprobes.c index e87679a48ba2..287b263c9cb9 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1111,9 +1111,20 @@ static int disarm_kprobe_ftrace(struct kprobe *p) ipmodify ? &kprobe_ipmodify_enabled : &kprobe_ftrace_enabled); } #else /* !CONFIG_KPROBES_ON_FTRACE */ -#define prepare_kprobe(p) arch_prepare_kprobe(p) -#define arm_kprobe_ftrace(p) (-ENODEV) -#define disarm_kprobe_ftrace(p) (-ENODEV) +static inline int prepare_kprobe(struct kprobe *p) +{ + return arch_prepare_kprobe(p); +} + +static inline int arm_kprobe_ftrace(struct kprobe *p) +{ + return -ENODEV; +} + +static inline int disarm_kprobe_ftrace(struct kprobe *p) +{ + return -ENODEV; +} #endif /* Arm a kprobe with text_mutex */ @@ -2145,6 +2156,13 @@ static void kill_kprobe(struct kprobe *p) * the original probed function (which will be freed soon) any more. */ arch_remove_kprobe(p); + + /* + * The module is going away. We should disarm the kprobe which + * is using ftrace. + */ + if (kprobe_ftrace(p)) + disarm_kprobe_ftrace(p); } /* Disable one kprobe */ diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 72064541bef2..275441254bb5 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -139,9 +139,6 @@ static inline void ftrace_ops_init(struct ftrace_ops *ops) #endif } -#define FTRACE_PID_IGNORE -1 -#define FTRACE_PID_TRACE -2 - static void ftrace_pid_func(unsigned long ip, unsigned long parent_ip, struct ftrace_ops *op, struct pt_regs *regs) { @@ -2388,6 +2385,14 @@ struct ftrace_ops direct_ops = { .flags = FTRACE_OPS_FL_IPMODIFY | FTRACE_OPS_FL_RECURSION_SAFE | FTRACE_OPS_FL_DIRECT | FTRACE_OPS_FL_SAVE_REGS | FTRACE_OPS_FL_PERMANENT, + /* + * By declaring the main trampoline as this trampoline + * it will never have one allocated for it. Allocated + * trampolines should not call direct functions. + * The direct_ops should only be called by the builtin + * ftrace_regs_caller trampoline. + */ + .trampoline = FTRACE_REGS_ADDR, }; #endif /* CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS */ @@ -6255,8 +6260,19 @@ static int referenced_filters(struct dyn_ftrace *rec) int cnt = 0; for (ops = ftrace_ops_list; ops != &ftrace_list_end; ops = ops->next) { - if (ops_references_rec(ops, rec)) - cnt++; + if (ops_references_rec(ops, rec)) { + if (WARN_ON_ONCE(ops->flags & FTRACE_OPS_FL_DIRECT)) + continue; + if (WARN_ON_ONCE(ops->flags & FTRACE_OPS_FL_IPMODIFY)) + continue; + cnt++; + if (ops->flags & FTRACE_OPS_FL_SAVE_REGS) + rec->flags |= FTRACE_FL_REGS; + if (cnt == 1 && ops->trampoline) + rec->flags |= FTRACE_FL_TRAMP; + else + rec->flags &= ~FTRACE_FL_TRAMP; + } } return cnt; @@ -6435,8 +6451,8 @@ void ftrace_module_enable(struct module *mod) if (ftrace_start_up) cnt += referenced_filters(rec); - /* This clears FTRACE_FL_DISABLED */ - rec->flags = cnt; + rec->flags &= ~FTRACE_FL_DISABLED; + rec->flags += cnt; if (ftrace_start_up && cnt) { int failed = __ftrace_replace_code(rec, 1); @@ -7066,12 +7082,12 @@ void ftrace_pid_follow_fork(struct trace_array *tr, bool enable) if (enable) { register_trace_sched_process_fork(ftrace_pid_follow_sched_process_fork, tr); - register_trace_sched_process_exit(ftrace_pid_follow_sched_process_exit, + register_trace_sched_process_free(ftrace_pid_follow_sched_process_exit, tr); } else { unregister_trace_sched_process_fork(ftrace_pid_follow_sched_process_fork, tr); - unregister_trace_sched_process_exit(ftrace_pid_follow_sched_process_exit, + unregister_trace_sched_process_free(ftrace_pid_follow_sched_process_exit, tr); } } diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index f15471ce969e..93ef0ab6ea20 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -270,6 +270,9 @@ EXPORT_SYMBOL_GPL(ring_buffer_event_data); #define for_each_buffer_cpu(buffer, cpu) \ for_each_cpu(cpu, buffer->cpumask) +#define for_each_online_buffer_cpu(buffer, cpu) \ + for_each_cpu_and(cpu, buffer->cpumask, cpu_online_mask) + #define TS_SHIFT 27 #define TS_MASK ((1ULL << TS_SHIFT) - 1) #define TS_DELTA_TEST (~TS_MASK) @@ -413,12 +416,27 @@ struct rb_irq_work { struct rb_event_info { u64 ts; u64 delta; + u64 before; + u64 after; unsigned long length; struct buffer_page *tail_page; int add_timestamp; }; /* + * Used for the add_timestamp + * NONE + * EXTEND - wants a time extend + * ABSOLUTE - the buffer requests all events to have absolute time stamps + * FORCE - force a full time stamp. + */ +enum { + RB_ADD_STAMP_NONE = 0, + RB_ADD_STAMP_EXTEND = BIT(1), + RB_ADD_STAMP_ABSOLUTE = BIT(2), + RB_ADD_STAMP_FORCE = BIT(3) +}; +/* * Used for which event context the event is in. * NMI = 0 * IRQ = 1 @@ -435,6 +453,28 @@ enum { RB_CTX_MAX }; +#if BITS_PER_LONG == 32 +#define RB_TIME_32 +#endif + +/* To test on 64 bit machines */ +//#define RB_TIME_32 + +#ifdef RB_TIME_32 + +struct rb_time_struct { + local_t cnt; + local_t top; + local_t bottom; +}; +#else +#include <asm/local64.h> +struct rb_time_struct { + local64_t time; +}; +#endif +typedef struct rb_time_struct rb_time_t; + /* * head_page == tail_page && head == tail then buffer is empty. */ @@ -470,7 +510,8 @@ struct ring_buffer_per_cpu { size_t shortest_full; unsigned long read; unsigned long read_bytes; - u64 write_stamp; + rb_time_t write_stamp; + rb_time_t before_stamp; u64 read_stamp; /* ring buffer pages to update, > 0 to add, < 0 to remove */ long nr_pages_to_update; @@ -513,6 +554,189 @@ struct ring_buffer_iter { int missed_events; }; +#ifdef RB_TIME_32 + +/* + * On 32 bit machines, local64_t is very expensive. As the ring + * buffer doesn't need all the features of a true 64 bit atomic, + * on 32 bit, it uses these functions (64 still uses local64_t). + * + * For the ring buffer, 64 bit required operations for the time is + * the following: + * + * - Only need 59 bits (uses 60 to make it even). + * - Reads may fail if it interrupted a modification of the time stamp. + * It will succeed if it did not interrupt another write even if + * the read itself is interrupted by a write. + * It returns whether it was successful or not. + * + * - Writes always succeed and will overwrite other writes and writes + * that were done by events interrupting the current write. + * + * - A write followed by a read of the same time stamp will always succeed, + * but may not contain the same value. + * + * - A cmpxchg will fail if it interrupted another write or cmpxchg. + * Other than that, it acts like a normal cmpxchg. + * + * The 60 bit time stamp is broken up by 30 bits in a top and bottom half + * (bottom being the least significant 30 bits of the 60 bit time stamp). + * + * The two most significant bits of each half holds a 2 bit counter (0-3). + * Each update will increment this counter by one. + * When reading the top and bottom, if the two counter bits match then the + * top and bottom together make a valid 60 bit number. + */ +#define RB_TIME_SHIFT 30 +#define RB_TIME_VAL_MASK ((1 << RB_TIME_SHIFT) - 1) + +static inline int rb_time_cnt(unsigned long val) +{ + return (val >> RB_TIME_SHIFT) & 3; +} + +static inline u64 rb_time_val(unsigned long top, unsigned long bottom) +{ + u64 val; + + val = top & RB_TIME_VAL_MASK; + val <<= RB_TIME_SHIFT; + val |= bottom & RB_TIME_VAL_MASK; + + return val; +} + +static inline bool __rb_time_read(rb_time_t *t, u64 *ret, unsigned long *cnt) +{ + unsigned long top, bottom; + unsigned long c; + + /* + * If the read is interrupted by a write, then the cnt will + * be different. Loop until both top and bottom have been read + * without interruption. + */ + do { + c = local_read(&t->cnt); + top = local_read(&t->top); + bottom = local_read(&t->bottom); + } while (c != local_read(&t->cnt)); + + *cnt = rb_time_cnt(top); + + /* If top and bottom counts don't match, this interrupted a write */ + if (*cnt != rb_time_cnt(bottom)) + return false; + + *ret = rb_time_val(top, bottom); + return true; +} + +static bool rb_time_read(rb_time_t *t, u64 *ret) +{ + unsigned long cnt; + + return __rb_time_read(t, ret, &cnt); +} + +static inline unsigned long rb_time_val_cnt(unsigned long val, unsigned long cnt) +{ + return (val & RB_TIME_VAL_MASK) | ((cnt & 3) << RB_TIME_SHIFT); +} + +static inline void rb_time_split(u64 val, unsigned long *top, unsigned long *bottom) +{ + *top = (unsigned long)((val >> RB_TIME_SHIFT) & RB_TIME_VAL_MASK); + *bottom = (unsigned long)(val & RB_TIME_VAL_MASK); +} + +static inline void rb_time_val_set(local_t *t, unsigned long val, unsigned long cnt) +{ + val = rb_time_val_cnt(val, cnt); + local_set(t, val); +} + +static void rb_time_set(rb_time_t *t, u64 val) +{ + unsigned long cnt, top, bottom; + + rb_time_split(val, &top, &bottom); + + /* Writes always succeed with a valid number even if it gets interrupted. */ + do { + cnt = local_inc_return(&t->cnt); + rb_time_val_set(&t->top, top, cnt); + rb_time_val_set(&t->bottom, bottom, cnt); + } while (cnt != local_read(&t->cnt)); +} + +static inline bool +rb_time_read_cmpxchg(local_t *l, unsigned long expect, unsigned long set) +{ + unsigned long ret; + + ret = local_cmpxchg(l, expect, set); + return ret == expect; +} + +static int rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set) +{ + unsigned long cnt, top, bottom; + unsigned long cnt2, top2, bottom2; + u64 val; + + /* The cmpxchg always fails if it interrupted an update */ + if (!__rb_time_read(t, &val, &cnt2)) + return false; + + if (val != expect) + return false; + + cnt = local_read(&t->cnt); + if ((cnt & 3) != cnt2) + return false; + + cnt2 = cnt + 1; + + rb_time_split(val, &top, &bottom); + top = rb_time_val_cnt(top, cnt); + bottom = rb_time_val_cnt(bottom, cnt); + + rb_time_split(set, &top2, &bottom2); + top2 = rb_time_val_cnt(top2, cnt2); + bottom2 = rb_time_val_cnt(bottom2, cnt2); + + if (!rb_time_read_cmpxchg(&t->cnt, cnt, cnt2)) + return false; + if (!rb_time_read_cmpxchg(&t->top, top, top2)) + return false; + if (!rb_time_read_cmpxchg(&t->bottom, bottom, bottom2)) + return false; + return true; +} + +#else /* 64 bits */ + +/* local64_t always succeeds */ + +static inline bool rb_time_read(rb_time_t *t, u64 *ret) +{ + *ret = local64_read(&t->time); + return true; +} +static void rb_time_set(rb_time_t *t, u64 val) +{ + local64_set(&t->time, val); +} + +static bool rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set) +{ + u64 val; + val = local64_cmpxchg(&t->time, expect, set); + return val == expect; +} +#endif + /** * ring_buffer_nr_pages - get the number of buffer pages in the ring buffer * @buffer: The ring_buffer to get the number of pages from @@ -746,8 +970,16 @@ __poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu, static inline u64 rb_time_stamp(struct trace_buffer *buffer) { + u64 ts; + + /* Skip retpolines :-( */ + if (IS_ENABLED(CONFIG_RETPOLINE) && likely(buffer->clock == trace_clock_local)) + ts = trace_clock_local(); + else + ts = buffer->clock(); + /* shift to debug/test normalization and TIME_EXTENTS */ - return buffer->clock() << DEBUG_SHIFT; + return ts << DEBUG_SHIFT; } u64 ring_buffer_time_stamp(struct trace_buffer *buffer, int cpu) @@ -2372,8 +2604,8 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, return NULL; } -/* Slow path, do not inline */ -static noinline struct ring_buffer_event * +/* Slow path */ +static struct ring_buffer_event * rb_add_time_stamp(struct ring_buffer_event *event, u64 delta, bool abs) { if (abs) @@ -2397,6 +2629,66 @@ rb_add_time_stamp(struct ring_buffer_event *event, u64 delta, bool abs) static inline bool rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer, struct ring_buffer_event *event); +#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK +static inline bool sched_clock_stable(void) +{ + return true; +} +#endif + +static void +rb_check_timestamp(struct ring_buffer_per_cpu *cpu_buffer, + struct rb_event_info *info) +{ + u64 write_stamp; + + WARN_ONCE(1, "Delta way too big! %llu ts=%llu before=%llu after=%llu write stamp=%llu\n%s", + (unsigned long long)info->delta, + (unsigned long long)info->ts, + (unsigned long long)info->before, + (unsigned long long)info->after, + (unsigned long long)(rb_time_read(&cpu_buffer->write_stamp, &write_stamp) ? write_stamp : 0), + sched_clock_stable() ? "" : + "If you just came from a suspend/resume,\n" + "please switch to the trace global clock:\n" + " echo global > /sys/kernel/debug/tracing/trace_clock\n" + "or add trace_clock=global to the kernel command line\n"); +} + +static void rb_add_timestamp(struct ring_buffer_per_cpu *cpu_buffer, + struct ring_buffer_event **event, + struct rb_event_info *info, + u64 *delta, + unsigned int *length) +{ + bool abs = info->add_timestamp & + (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_ABSOLUTE); + + if (unlikely(info->delta > (1ULL << 59))) { + /* did the clock go backwards */ + if (info->before == info->after && info->before > info->ts) { + /* not interrupted */ + static int once; + + /* + * This is possible with a recalibrating of the TSC. + * Do not produce a call stack, but just report it. + */ + if (!once) { + once++; + pr_warn("Ring buffer clock went backwards: %llu -> %llu\n", + info->before, info->ts); + } + } else + rb_check_timestamp(cpu_buffer, info); + if (!abs) + info->delta = 0; + } + *event = rb_add_time_stamp(*event, info->delta, abs); + *length -= RB_LEN_TIME_EXTEND; + *delta = 0; +} + /** * rb_update_event - update event type and data * @cpu_buffer: The per cpu buffer of the @event @@ -2416,21 +2708,12 @@ rb_update_event(struct ring_buffer_per_cpu *cpu_buffer, unsigned length = info->length; u64 delta = info->delta; - /* Only a commit updates the timestamp */ - if (unlikely(!rb_event_is_commit(cpu_buffer, event))) - delta = 0; - /* * If we need to add a timestamp, then we * add it to the start of the reserved space. */ - if (unlikely(info->add_timestamp)) { - bool abs = ring_buffer_time_stamp_abs(cpu_buffer->buffer); - - event = rb_add_time_stamp(event, abs ? info->delta : delta, abs); - length -= RB_LEN_TIME_EXTEND; - delta = 0; - } + if (unlikely(info->add_timestamp)) + rb_add_timestamp(cpu_buffer, &event, info, &delta, &length); event->time_delta = delta; length -= RB_EVNT_HDR_SIZE; @@ -2473,12 +2756,38 @@ static unsigned rb_calculate_event_length(unsigned length) return length; } -#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK -static inline bool sched_clock_stable(void) +static __always_inline bool +rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer, + struct ring_buffer_event *event) { - return true; + unsigned long addr = (unsigned long)event; + unsigned long index; + + index = rb_event_index(event); + addr &= PAGE_MASK; + + return cpu_buffer->commit_page->page == (void *)addr && + rb_commit_index(cpu_buffer) == index; +} + +static u64 rb_time_delta(struct ring_buffer_event *event) +{ + switch (event->type_len) { + case RINGBUF_TYPE_PADDING: + return 0; + + case RINGBUF_TYPE_TIME_EXTEND: + return ring_buffer_event_time_stamp(event); + + case RINGBUF_TYPE_TIME_STAMP: + return 0; + + case RINGBUF_TYPE_DATA: + return event->time_delta; + default: + return 0; + } } -#endif static inline int rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, @@ -2488,6 +2797,8 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, struct buffer_page *bpage; unsigned long index; unsigned long addr; + u64 write_stamp; + u64 delta; new_index = rb_event_index(event); old_index = new_index + rb_event_ts_length(event); @@ -2496,10 +2807,32 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, bpage = READ_ONCE(cpu_buffer->tail_page); + delta = rb_time_delta(event); + + if (!rb_time_read(&cpu_buffer->write_stamp, &write_stamp)) + return 0; + + /* Make sure the write stamp is read before testing the location */ + barrier(); + if (bpage->page == (void *)addr && rb_page_write(bpage) == old_index) { unsigned long write_mask = local_read(&bpage->write) & ~RB_WRITE_MASK; unsigned long event_length = rb_event_length(event); + + /* Something came in, can't discard */ + if (!rb_time_cmpxchg(&cpu_buffer->write_stamp, + write_stamp, write_stamp - delta)) + return 0; + + /* + * If an event were to come in now, it would see that the + * write_stamp and the before_stamp are different, and assume + * that this event just added itself before updating + * the write stamp. The interrupting event will fix the + * write stamp for us, and use the before stamp as its delta. + */ + /* * This is on the tail page. It is possible that * a write could come in and move the tail page @@ -2551,10 +2884,6 @@ rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer) local_set(&cpu_buffer->commit_page->page->commit, rb_page_write(cpu_buffer->commit_page)); rb_inc_page(cpu_buffer, &cpu_buffer->commit_page); - /* Only update the write stamp if the page has an event */ - if (rb_page_write(cpu_buffer->commit_page)) - cpu_buffer->write_stamp = - cpu_buffer->commit_page->page->time_stamp; /* add barrier to keep gcc from optimizing too much */ barrier(); } @@ -2626,54 +2955,10 @@ static inline void rb_event_discard(struct ring_buffer_event *event) event->time_delta = 1; } -static __always_inline bool -rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer, - struct ring_buffer_event *event) -{ - unsigned long addr = (unsigned long)event; - unsigned long index; - - index = rb_event_index(event); - addr &= PAGE_MASK; - - return cpu_buffer->commit_page->page == (void *)addr && - rb_commit_index(cpu_buffer) == index; -} - -static __always_inline void -rb_update_write_stamp(struct ring_buffer_per_cpu *cpu_buffer, - struct ring_buffer_event *event) -{ - u64 delta; - - /* - * The event first in the commit queue updates the - * time stamp. - */ - if (rb_event_is_commit(cpu_buffer, event)) { - /* - * A commit event that is first on a page - * updates the write timestamp with the page stamp - */ - if (!rb_event_index(event)) - cpu_buffer->write_stamp = - cpu_buffer->commit_page->page->time_stamp; - else if (event->type_len == RINGBUF |