From ac9721f3f54b27a16c7e1afb2481e7ee95a70318 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 27 May 2010 12:54:41 +0200 Subject: perf_events: Fix races and clean up perf_event and perf_mmap_data interaction In order to move toward separate buffer objects, rework the whole perf_mmap_data construct to be a more self-sufficient entity, one with its own lifetime rules. This greatly sanitizes the whole output redirection code, which was riddled with bugs and races. Signed-off-by: Peter Zijlstra Cc: LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 224 +++++++++++++++++++++++++++++----------------------- 1 file changed, 126 insertions(+), 98 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index bd7ce8ca5bb9..848d49a043e9 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1841,6 +1841,7 @@ static void free_event_rcu(struct rcu_head *head) } static void perf_pending_sync(struct perf_event *event); +static void perf_mmap_data_put(struct perf_mmap_data *data); static void free_event(struct perf_event *event) { @@ -1856,9 +1857,9 @@ static void free_event(struct perf_event *event) atomic_dec(&nr_task_events); } - if (event->output) { - fput(event->output->filp); - event->output = NULL; + if (event->data) { + perf_mmap_data_put(event->data); + event->data = NULL; } if (event->destroy) @@ -2175,7 +2176,27 @@ unlock: return ret; } -static int perf_event_set_output(struct perf_event *event, int output_fd); +static const struct file_operations perf_fops; + +static struct perf_event *perf_fget_light(int fd, int *fput_needed) +{ + struct file *file; + + file = fget_light(fd, fput_needed); + if (!file) + return ERR_PTR(-EBADF); + + if (file->f_op != &perf_fops) { + fput_light(file, *fput_needed); + *fput_needed = 0; + return ERR_PTR(-EBADF); + } + + return file->private_data; +} + +static int perf_event_set_output(struct perf_event *event, + struct perf_event *output_event); static int perf_event_set_filter(struct perf_event *event, void __user *arg); static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg) @@ -2202,7 +2223,23 @@ static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg) return perf_event_period(event, (u64 __user *)arg); case PERF_EVENT_IOC_SET_OUTPUT: - return perf_event_set_output(event, arg); + { + struct perf_event *output_event = NULL; + int fput_needed = 0; + int ret; + + if (arg != -1) { + output_event = perf_fget_light(arg, &fput_needed); + if (IS_ERR(output_event)) + return PTR_ERR(output_event); + } + + ret = perf_event_set_output(event, output_event); + if (output_event) + fput_light(output_event->filp, fput_needed); + + return ret; + } case PERF_EVENT_IOC_SET_FILTER: return perf_event_set_filter(event, (void __user *)arg); @@ -2335,8 +2372,6 @@ perf_mmap_data_alloc(struct perf_event *event, int nr_pages) unsigned long size; int i; - WARN_ON(atomic_read(&event->mmap_count)); - size = sizeof(struct perf_mmap_data); size += nr_pages * sizeof(void *); @@ -2452,8 +2487,6 @@ perf_mmap_data_alloc(struct perf_event *event, int nr_pages) unsigned long size; void *all_buf; - WARN_ON(atomic_read(&event->mmap_count)); - size = sizeof(struct perf_mmap_data); size += sizeof(void *); @@ -2536,7 +2569,7 @@ perf_mmap_data_init(struct perf_event *event, struct perf_mmap_data *data) if (!data->watermark) data->watermark = max_size / 2; - + atomic_set(&data->refcount, 1); rcu_assign_pointer(event->data, data); } @@ -2548,13 +2581,26 @@ static void perf_mmap_data_free_rcu(struct rcu_head *rcu_head) perf_mmap_data_free(data); } -static void perf_mmap_data_release(struct perf_event *event) +static struct perf_mmap_data *perf_mmap_data_get(struct perf_event *event) { - struct perf_mmap_data *data = event->data; + struct perf_mmap_data *data; + + rcu_read_lock(); + data = rcu_dereference(event->data); + if (data) { + if (!atomic_inc_not_zero(&data->refcount)) + data = NULL; + } + rcu_read_unlock(); + + return data; +} - WARN_ON(atomic_read(&event->mmap_count)); +static void perf_mmap_data_put(struct perf_mmap_data *data) +{ + if (!atomic_dec_and_test(&data->refcount)) + return; - rcu_assign_pointer(event->data, NULL); call_rcu(&data->rcu_head, perf_mmap_data_free_rcu); } @@ -2569,15 +2615,18 @@ static void perf_mmap_close(struct vm_area_struct *vma) { struct perf_event *event = vma->vm_file->private_data; - WARN_ON_ONCE(event->ctx->parent_ctx); if (atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex)) { unsigned long size = perf_data_size(event->data); - struct user_struct *user = current_user(); + struct user_struct *user = event->mmap_user; + struct perf_mmap_data *data = event->data; atomic_long_sub((size >> PAGE_SHIFT) + 1, &user->locked_vm); - vma->vm_mm->locked_vm -= event->data->nr_locked; - perf_mmap_data_release(event); + vma->vm_mm->locked_vm -= event->mmap_locked; + rcu_assign_pointer(event->data, NULL); mutex_unlock(&event->mmap_mutex); + + perf_mmap_data_put(data); + free_uid(user); } } @@ -2629,13 +2678,10 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) WARN_ON_ONCE(event->ctx->parent_ctx); mutex_lock(&event->mmap_mutex); - if (event->output) { - ret = -EINVAL; - goto unlock; - } - - if (atomic_inc_not_zero(&event->mmap_count)) { - if (nr_pages != event->data->nr_pages) + if (event->data) { + if (event->data->nr_pages == nr_pages) + atomic_inc(&event->data->refcount); + else ret = -EINVAL; goto unlock; } @@ -2667,21 +2713,23 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) WARN_ON(event->data); data = perf_mmap_data_alloc(event, nr_pages); - ret = -ENOMEM; - if (!data) + if (!data) { + ret = -ENOMEM; goto unlock; + } - ret = 0; perf_mmap_data_init(event, data); - - atomic_set(&event->mmap_count, 1); - atomic_long_add(user_extra, &user->locked_vm); - vma->vm_mm->locked_vm += extra; - event->data->nr_locked = extra; if (vma->vm_flags & VM_WRITE) event->data->writable = 1; + atomic_long_add(user_extra, &user->locked_vm); + event->mmap_locked = extra; + event->mmap_user = get_current_user(); + vma->vm_mm->locked_vm += event->mmap_locked; + unlock: + if (!ret) + atomic_inc(&event->mmap_count); mutex_unlock(&event->mmap_mutex); vma->vm_flags |= VM_RESERVED; @@ -2993,7 +3041,6 @@ int perf_output_begin(struct perf_output_handle *handle, struct perf_event *event, unsigned int size, int nmi, int sample) { - struct perf_event *output_event; struct perf_mmap_data *data; unsigned long tail, offset, head; int have_lost; @@ -3010,10 +3057,6 @@ int perf_output_begin(struct perf_output_handle *handle, if (event->parent) event = event->parent; - output_event = rcu_dereference(event->output); - if (output_event) - event = output_event; - data = rcu_dereference(event->data); if (!data) goto out; @@ -4912,39 +4955,17 @@ err_size: goto out; } -static int perf_event_set_output(struct perf_event *event, int output_fd) +static int +perf_event_set_output(struct perf_event *event, struct perf_event *output_event) { - struct perf_event *output_event = NULL; - struct file *output_file = NULL; - struct perf_event *old_output; - int fput_needed = 0; + struct perf_mmap_data *data = NULL, *old_data = NULL; int ret = -EINVAL; - /* - * Don't allow output of inherited per-task events. This would - * create performance issues due to cross cpu access. - */ - if (event->cpu == -1 && event->attr.inherit) - return -EINVAL; - - if (!output_fd) + if (!output_event) goto set; - output_file = fget_light(output_fd, &fput_needed); - if (!output_file) - return -EBADF; - - if (output_file->f_op != &perf_fops) - goto out; - - output_event = output_file->private_data; - - /* Don't chain output fds */ - if (output_event->output) - goto out; - - /* Don't set an output fd when we already have an output channel */ - if (event->data) + /* don't allow circular references */ + if (event == output_event) goto out; /* @@ -4959,26 +4980,28 @@ static int perf_event_set_output(struct perf_event *event, int output_fd) if (output_event->cpu == -1 && output_event->ctx != event->ctx) goto out; - atomic_long_inc(&output_file->f_count); - set: mutex_lock(&event->mmap_mutex); - old_output = event->output; - rcu_assign_pointer(event->output, output_event); - mutex_unlock(&event->mmap_mutex); + /* Can't redirect output if we've got an active mmap() */ + if (atomic_read(&event->mmap_count)) + goto unlock; - if (old_output) { - /* - * we need to make sure no existing perf_output_*() - * is still referencing this event. - */ - synchronize_rcu(); - fput(old_output->filp); + if (output_event) { + /* get the buffer we want to redirect to */ + data = perf_mmap_data_get(output_event); + if (!data) + goto unlock; } + old_data = event->data; + rcu_assign_pointer(event->data, data); ret = 0; +unlock: + mutex_unlock(&event->mmap_mutex); + + if (old_data) + perf_mmap_data_put(old_data); out: - fput_light(output_file, fput_needed); return ret; } @@ -4994,7 +5017,7 @@ SYSCALL_DEFINE5(perf_event_open, struct perf_event_attr __user *, attr_uptr, pid_t, pid, int, cpu, int, group_fd, unsigned long, flags) { - struct perf_event *event, *group_leader; + struct perf_event *event, *group_leader = NULL, *output_event = NULL; struct perf_event_attr attr; struct perf_event_context *ctx; struct file *event_file = NULL; @@ -5034,19 +5057,25 @@ SYSCALL_DEFINE5(perf_event_open, goto err_fd; } + if (group_fd != -1) { + group_leader = perf_fget_light(group_fd, &fput_needed); + if (IS_ERR(group_leader)) { + err = PTR_ERR(group_leader); + goto err_put_context; + } + group_file = group_leader->filp; + if (flags & PERF_FLAG_FD_OUTPUT) + output_event = group_leader; + if (flags & PERF_FLAG_FD_NO_GROUP) + group_leader = NULL; + } + /* * Look up the group leader (we will attach this event to it): */ - group_leader = NULL; - if (group_fd != -1 && !(flags & PERF_FLAG_FD_NO_GROUP)) { + if (group_leader) { err = -EINVAL; - group_file = fget_light(group_fd, &fput_needed); - if (!group_file) - goto err_put_context; - if (group_file->f_op != &perf_fops) - goto err_put_context; - group_leader = group_file->private_data; /* * Do not allow a recursive hierarchy (this new sibling * becoming part of another group-sibling): @@ -5068,9 +5097,16 @@ SYSCALL_DEFINE5(perf_event_open, event = perf_event_alloc(&attr, cpu, ctx, group_leader, NULL, NULL, GFP_KERNEL); - err = PTR_ERR(event); - if (IS_ERR(event)) + if (IS_ERR(event)) { + err = PTR_ERR(event); goto err_put_context; + } + + if (output_event) { + err = perf_event_set_output(event, output_event); + if (err) + goto err_free_put_context; + } event_file = anon_inode_getfile("[perf_event]", &perf_fops, event, O_RDWR); if (IS_ERR(event_file)) { @@ -5078,12 +5114,6 @@ SYSCALL_DEFINE5(perf_event_open, goto err_free_put_context; } - if (flags & PERF_FLAG_FD_OUTPUT) { - err = perf_event_set_output(event, group_fd); - if (err) - goto err_fput_free_put_context; - } - event->filp = event_file; WARN_ON_ONCE(ctx->parent_ctx); mutex_lock(&ctx->mutex); @@ -5101,8 +5131,6 @@ SYSCALL_DEFINE5(perf_event_open, fd_install(event_fd, event_file); return event_fd; -err_fput_free_put_context: - fput(event_file); err_free_put_context: free_event(event); err_put_context: -- cgit v1.2.3 From 8a49542c0554af7d0073aac0ee73ee65b807ef34 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 27 May 2010 15:47:49 +0200 Subject: perf_events: Fix races in group composition Group siblings don't pin each-other or the parent, so when we destroy events we must make sure to clean up all cross referencing pointers. In particular, for destruction of a group leader we must be able to find all its siblings and remove their reference to it. This means that detaching an event from its context must not detach it from the group, otherwise we can end up failing to clear all pointers. Solve this by clearly separating the attachment to a context and attachment to a group, and keep the group composed until we destroy the events. Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 4 ++ kernel/perf_event.c | 91 ++++++++++++++++++++++++++++++++++------------ 2 files changed, 71 insertions(+), 24 deletions(-) (limited to 'kernel') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 490698590d6e..5d0266d94985 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -631,6 +631,9 @@ struct swevent_hlist { struct rcu_head rcu_head; }; +#define PERF_ATTACH_CONTEXT 0x01 +#define PERF_ATTACH_GROUP 0x02 + /** * struct perf_event - performance event kernel representation: */ @@ -646,6 +649,7 @@ struct perf_event { const struct pmu *pmu; enum perf_event_active_state state; + unsigned int attach_state; atomic64_t count; /* diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 848d49a043e9..10a1aee2309e 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -283,14 +283,15 @@ ctx_group_list(struct perf_event *event, struct perf_event_context *ctx) static void list_add_event(struct perf_event *event, struct perf_event_context *ctx) { - struct perf_event *group_leader = event->group_leader; + WARN_ON_ONCE(event->attach_state & PERF_ATTACH_CONTEXT); + event->attach_state |= PERF_ATTACH_CONTEXT; /* - * Depending on whether it is a standalone or sibling event, - * add it straight to the context's event list, or to the group - * leader's sibling list: + * If we're a stand alone event or group leader, we go to the context + * list, group events are kept attached to the group so that + * perf_group_detach can, at all times, locate all siblings. */ - if (group_leader == event) { + if (event->group_leader == event) { struct list_head *list; if (is_software_event(event)) @@ -298,13 +299,6 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) list = ctx_group_list(event, ctx); list_add_tail(&event->group_entry, list); - } else { - if (group_leader->group_flags & PERF_GROUP_SOFTWARE && - !is_software_event(event)) - group_leader->group_flags &= ~PERF_GROUP_SOFTWARE; - - list_add_tail(&event->group_entry, &group_leader->sibling_list); - group_leader->nr_siblings++; } list_add_rcu(&event->event_entry, &ctx->event_list); @@ -313,6 +307,24 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) ctx->nr_stat++; } +static void perf_group_attach(struct perf_event *event) +{ + struct perf_event *group_leader = event->group_leader; + + WARN_ON_ONCE(event->attach_state & PERF_ATTACH_GROUP); + event->attach_state |= PERF_ATTACH_GROUP; + + if (group_leader == event) + return; + + if (group_leader->group_flags & PERF_GROUP_SOFTWARE && + !is_software_event(event)) + group_leader->group_flags &= ~PERF_GROUP_SOFTWARE; + + list_add_tail(&event->group_entry, &group_leader->sibling_list); + group_leader->nr_siblings++; +} + /* * Remove a event from the lists for its context. * Must be called with ctx->mutex and ctx->lock held. @@ -320,17 +332,22 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) static void list_del_event(struct perf_event *event, struct perf_event_context *ctx) { - if (list_empty(&event->group_entry)) + /* + * We can have double detach due to exit/hot-unplug + close. + */ + if (!(event->attach_state & PERF_ATTACH_CONTEXT)) return; + + event->attach_state &= ~PERF_ATTACH_CONTEXT; + ctx->nr_events--; if (event->attr.inherit_stat) ctx->nr_stat--; - list_del_init(&event->group_entry); list_del_rcu(&event->event_entry); - if (event->group_leader != event) - event->group_leader->nr_siblings--; + if (event->group_leader == event) + list_del_init(&event->group_entry); update_group_times(event); @@ -345,21 +362,39 @@ list_del_event(struct perf_event *event, struct perf_event_context *ctx) event->state = PERF_EVENT_STATE_OFF; } -static void -perf_destroy_group(struct perf_event *event, struct perf_event_context *ctx) +static void perf_group_detach(struct perf_event *event) { struct perf_event *sibling, *tmp; + struct list_head *list = NULL; + + /* + * We can have double detach due to exit/hot-unplug + close. + */ + if (!(event->attach_state & PERF_ATTACH_GROUP)) + return; + + event->attach_state &= ~PERF_ATTACH_GROUP; + + /* + * If this is a sibling, remove it from its group. + */ + if (event->group_leader != event) { + list_del_init(&event->group_entry); + event->group_leader->nr_siblings--; + return; + } + + if (!list_empty(&event->group_entry)) + list = &event->group_entry; /* * If this was a group event with sibling events then * upgrade the siblings to singleton events by adding them - * to the context list directly: + * to whatever list we are on. */ list_for_each_entry_safe(sibling, tmp, &event->sibling_list, group_entry) { - struct list_head *list; - - list = ctx_group_list(event, ctx); - list_move_tail(&sibling->group_entry, list); + if (list) + list_move_tail(&sibling->group_entry, list); sibling->group_leader = sibling; /* Inherit group flags from the previous leader */ @@ -727,6 +762,7 @@ static void add_event_to_ctx(struct perf_event *event, struct perf_event_context *ctx) { list_add_event(event, ctx); + perf_group_attach(event); event->tstamp_enabled = ctx->time; event->tstamp_running = ctx->time; event->tstamp_stopped = ctx->time; @@ -1894,8 +1930,8 @@ int perf_event_release_kernel(struct perf_event *event) */ mutex_lock_nested(&ctx->mutex, SINGLE_DEPTH_NESTING); raw_spin_lock_irq(&ctx->lock); + perf_group_detach(event); list_del_event(event, ctx); - perf_destroy_group(event, ctx); raw_spin_unlock_irq(&ctx->lock); mutex_unlock(&ctx->mutex); @@ -5127,6 +5163,12 @@ SYSCALL_DEFINE5(perf_event_open, list_add_tail(&event->owner_entry, ¤t->perf_event_list); mutex_unlock(¤t->perf_event_mutex); + /* + * Drop the reference on the group_event after placing the + * new event on the sibling_list. This ensures destruction + * of the group leader will find the pointer to itself in + * perf_group_detach(). + */ fput_light(group_file, fput_needed); fd_install(event_fd, event_file); return event_fd; @@ -5448,6 +5490,7 @@ static void perf_free_event(struct perf_event *event, fput(parent->filp); + perf_group_detach(event); list_del_event(event, ctx); free_event(event); } -- cgit v1.2.3 From 3771f0771154675d4a0ca780be2411f3cc357208 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 21 May 2010 12:31:09 +0200 Subject: perf_events, trace: Fix probe unregister race tracepoint_probe_unregister() does not synchronize against the probe callbacks, so do that explicitly. This properly serializes the callbacks and the free of the data used therein. Also, use this_cpu_ptr() where possible. Acked-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <1274438476.1674.1702.camel@laptop> Signed-off-by: Ingo Molnar --- include/trace/ftrace.h | 2 +- kernel/trace/trace_event_perf.c | 10 ++++++++-- kernel/trace/trace_kprobe.c | 4 ++-- kernel/trace/trace_syscalls.c | 4 ++-- 4 files changed, 13 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h index 3d685d1f2a03..5a64905d7278 100644 --- a/include/trace/ftrace.h +++ b/include/trace/ftrace.h @@ -725,7 +725,7 @@ perf_trace_##call(void *__data, proto) \ \ { assign; } \ \ - head = per_cpu_ptr(event_call->perf_events, smp_processor_id());\ + head = this_cpu_ptr(event_call->perf_events); \ perf_trace_buf_submit(entry, __entry_size, rctx, __addr, \ __count, &__regs, head); \ } diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index cb6f365016e4..49c7abf2ba5c 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -116,7 +116,7 @@ int perf_trace_enable(struct perf_event *p_event) if (WARN_ON_ONCE(!list)) return -EINVAL; - list = per_cpu_ptr(list, smp_processor_id()); + list = this_cpu_ptr(list); hlist_add_head_rcu(&p_event->hlist_entry, list); return 0; @@ -142,6 +142,12 @@ void perf_trace_destroy(struct perf_event *p_event) tp_event->class->perf_probe, tp_event); + /* + * Ensure our callback won't be called anymore. See + * tracepoint_probe_unregister() and __DO_TRACE(). + */ + synchronize_sched(); + free_percpu(tp_event->perf_events); tp_event->perf_events = NULL; @@ -169,7 +175,7 @@ __kprobes void *perf_trace_buf_prepare(int size, unsigned short type, if (*rctxp < 0) return NULL; - raw_data = per_cpu_ptr(perf_trace_buf[*rctxp], smp_processor_id()); + raw_data = this_cpu_ptr(perf_trace_buf[*rctxp]); /* zero the dead bytes from align to not leak stack to user */ memset(&raw_data[size - sizeof(u64)], 0, sizeof(u64)); diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index faf7cefd15da..f52b5f50299d 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -1359,7 +1359,7 @@ static __kprobes void kprobe_perf_func(struct kprobe *kp, for (i = 0; i < tp->nr_args; i++) call_fetch(&tp->args[i].fetch, regs, data + tp->args[i].offset); - head = per_cpu_ptr(call->perf_events, smp_processor_id()); + head = this_cpu_ptr(call->perf_events); perf_trace_buf_submit(entry, size, rctx, entry->ip, 1, regs, head); } @@ -1392,7 +1392,7 @@ static __kprobes void kretprobe_perf_func(struct kretprobe_instance *ri, for (i = 0; i < tp->nr_args; i++) call_fetch(&tp->args[i].fetch, regs, data + tp->args[i].offset); - head = per_cpu_ptr(call->perf_events, smp_processor_id()); + head = this_cpu_ptr(call->perf_events); perf_trace_buf_submit(entry, size, rctx, entry->ret_ip, 1, regs, head); } diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c index d2c859cec9ea..34e35804304b 100644 --- a/kernel/trace/trace_syscalls.c +++ b/kernel/trace/trace_syscalls.c @@ -519,7 +519,7 @@ static void perf_syscall_enter(void *ignore, struct pt_regs *regs, long id) syscall_get_arguments(current, regs, 0, sys_data->nb_args, (unsigned long *)&rec->args); - head = per_cpu_ptr(sys_data->enter_event->perf_events, smp_processor_id()); + head = this_cpu_ptr(sys_data->enter_event->perf_events); perf_trace_buf_submit(rec, size, rctx, 0, 1, regs, head); } @@ -595,7 +595,7 @@ static void perf_syscall_exit(void *ignore, struct pt_regs *regs, long ret) rec->nr = syscall_nr; rec->ret = syscall_get_return_value(current, regs); - head = per_cpu_ptr(sys_data->exit_event->perf_events, smp_processor_id()); + head = this_cpu_ptr(sys_data->exit_event->perf_events); perf_trace_buf_submit(rec, size, rctx, 0, 1, regs, head); } -- cgit v1.2.3 From 2e97942fe57864588774f173cf4cd7bb68968b76 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 21 May 2010 16:22:33 +0200 Subject: perf_events, trace: Fix perf_trace_destroy(), mutex went missing Steve spotted I forgot to do the destroy under event_mutex. Reported-by: Steven Rostedt Signed-off-by: Peter Zijlstra LKML-Reference: <1274451913.1674.1707.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/trace/trace_event_perf.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 49c7abf2ba5c..e6f65887842c 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -132,8 +132,9 @@ void perf_trace_destroy(struct perf_event *p_event) struct ftrace_event_call *tp_event = p_event->tp_event; int i; + mutex_lock(&event_mutex); if (--tp_event->perf_refcount > 0) - return; + goto out; if (tp_event->class->reg) tp_event->class->reg(tp_event, TRACE_REG_PERF_UNREGISTER); @@ -157,6 +158,8 @@ void perf_trace_destroy(struct perf_event *p_event) perf_trace_buf[i] = NULL; } } +out: + mutex_unlock(&event_mutex); } __kprobes void *perf_trace_buf_prepare(int size, unsigned short type, -- cgit v1.2.3 From 90151c35b19633e0cab5a6c80f1ba4a51e7c913b Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Tue, 25 May 2010 16:23:10 +0200 Subject: perf_events: Fix event scheduling issues introduced by transactional API The transactional API patch between the generic and model-specific code introduced several important bugs with event scheduling, at least on X86. If you had pinned events, e.g., watchdog, and were over-committing the PMU, you would get bogus counts. The bug was showing up on Intel CPU because events would move around more often that on AMD. But the problem also existed on AMD, though harder to expose. The issues were: - group_sched_in() was missing a cancel_txn() in the error path - cpuc->n_added was not properly maintained, leading to missing actions in hw_perf_enable(), i.e., n_running being 0. You cannot update n_added until you know the transaction has succeeded. In case of failed transaction n_added was not adjusted back. - in case of failed transactions, event_sched_out() was called and eventually invoked x86_disable_event() to touch the HW reg. But with transactions, on X86, event_sched_in() does not touch HW registers, it simply collects events into a list. Thus, you could end up calling x86_disable_event() on a counter which did not correspond to the current event when idx != -1. The patch modifies the generic and X86 code to avoid all those problems. First, we keep track of the number of events added last. In case the transaction fails, we substract them from n_added. This approach is necessary (as opposed to delaying updates to n_added) because not all event updates use the transaction API, e.g., single events. Second, we encapsulate the event_sched_in() and event_sched_out() in group_sched_in() inside the transaction. That makes the operations symmetrical and you can also detect that you are inside a transaction and skip the HW reg access by checking cpuc->group_flag. With this patch, you can now overcommit the PMU even with pinned system-wide events present and still get valid counts. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <1274796225.5882.1389.camel@twins> Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event.c | 22 ++++++++++++++++++++++ kernel/perf_event.c | 11 +++++++---- 2 files changed, 29 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index c77586061bcb..5db5b7d65a18 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -106,6 +106,7 @@ struct cpu_hw_events { int n_events; int n_added; + int n_txn; int assign[X86_PMC_IDX_MAX]; /* event to counter assignment */ u64 tags[X86_PMC_IDX_MAX]; struct perf_event *event_list[X86_PMC_IDX_MAX]; /* in enabled order */ @@ -983,6 +984,7 @@ static int x86_pmu_enable(struct perf_event *event) out: cpuc->n_events = n; cpuc->n_added += n - n0; + cpuc->n_txn += n - n0; return 0; } @@ -1089,6 +1091,14 @@ static void x86_pmu_disable(struct perf_event *event) struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); int i; + /* + * If we're called during a txn, we don't need to do anything. + * The events never got scheduled and ->cancel_txn will truncate + * the event_list. + */ + if (cpuc->group_flag & PERF_EVENT_TXN_STARTED) + return; + x86_pmu_stop(event); for (i = 0; i < cpuc->n_events; i++) { @@ -1379,6 +1389,7 @@ static void x86_pmu_start_txn(const struct pmu *pmu) struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); cpuc->group_flag |= PERF_EVENT_TXN_STARTED; + cpuc->n_txn = 0; } /* @@ -1391,6 +1402,11 @@ static void x86_pmu_cancel_txn(const struct pmu *pmu) struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); cpuc->group_flag &= ~PERF_EVENT_TXN_STARTED; + /* + * Truncate the collected events. + */ + cpuc->n_added -= cpuc->n_txn; + cpuc->n_events -= cpuc->n_txn; } /* @@ -1419,6 +1435,12 @@ static int x86_pmu_commit_txn(const struct pmu *pmu) */ memcpy(cpuc->assign, assign, n*sizeof(int)); + /* + * Clear out the txn count so that ->cancel_txn() which gets + * run after ->commit_txn() doesn't undo things. + */ + cpuc->n_txn = 0; + return 0; } diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 10a1aee2309e..42a0e9191af5 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -687,8 +687,11 @@ group_sched_in(struct perf_event *group_event, if (txn) pmu->start_txn(pmu); - if (event_sched_in(group_event, cpuctx, ctx)) + if (event_sched_in(group_event, cpuctx, ctx)) { + if (txn) + pmu->cancel_txn(pmu); return -EAGAIN; + } /* * Schedule in siblings as one group (if any): @@ -710,9 +713,6 @@ group_sched_in(struct perf_event *group_event, } group_error: - if (txn) - pmu->cancel_txn(pmu); - /* * Groups can be scheduled in as one unit only, so undo any * partial group before returning: @@ -724,6 +724,9 @@ group_error: } event_sched_out(group_event, cpuctx, ctx); + if (txn) + pmu->cancel_txn(pmu); + return -EAGAIN; } -- cgit v1.2.3 From 74048f895fa8cbf8119b4999f1f44881a825f954 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 27 May 2010 21:34:58 +0200 Subject: perf_events: Fix unincremented buffer base on partial copy If a sample size crosses to the next page boundary, the copy will be made in more than one step. However we forget to advance the source offset for the next copy, leading to unexpected double copies that completely mess up the traces. This fixes various kinds of bad traces that have irrelevant data inside, as an example: geany-4979 [001] 5758.077775: sched_switch: prev_comm=! prev_pid=121 prev_prio=0 prev_state=S|D|Z|X|x ==> next_comm= next_pid=7497072 next_prio=0 Signed-off-by: Frederic Weisbecker Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras Signed-off-by: Peter Zijlstra LKML-Reference: <1274988898-5639-1-git-send-regression-fweisbec@gmail.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 42a0e9191af5..858f56fa2432 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -3064,6 +3064,7 @@ __always_inline void perf_output_copy(struct perf_output_handle *handle, len -= size; handle->addr += size; + buf += size; handle->size -= size; if (!handle->size) { struct perf_mmap_data *data = handle->data; -- cgit v1.2.3 From 546cf44a1b507c1cbb5cf42bbe6169780567f36f Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Sat, 29 May 2010 11:45:07 -0700 Subject: blktrace: Fix new kernel-doc warnings Fix blktrace.c kernel-doc warnings: Warning(kernel/trace/blktrace.c:858): No description found for parameter 'ignore' Warning(kernel/trace/blktrace.c:890): No description found for parameter 'ignore' Signed-off-by: Randy Dunlap Cc: Jens Axboe Cc: Steven Rostedt Cc: Frederic Weisbecker LKML-Reference: <20100529114507.c466fc1e.randy.dunlap@oracle.com> Signed-off-by: Ingo Molnar --- kernel/trace/blktrace.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 36ea2b65dcdc..638711c17504 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -842,6 +842,7 @@ static void blk_add_trace_split(void *ignore, /** * blk_add_trace_remap - Add a trace for a remap operation + * @ignore: trace callback data parameter (not used) * @q: queue the io is for * @bio: the source bio * @dev: target device @@ -873,6 +874,7 @@ static void blk_add_trace_remap(void *ignore, /** * blk_add_trace_rq_remap - Add a trace for a request-remap operation + * @ignore: trace callback data parameter (not used) * @q: queue the io is for * @rq: the source request * @dev: target device -- cgit v1.2.3 From c6df8d5ab87a246942d138321e1721edbb69f6e1 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 3 Jun 2010 11:21:20 +0200 Subject: perf: Fix crash in swevents Frederic reported that because swevents handling doesn't disable IRQs anymore, we can get a recursion of perf_adjust_period(), once from overflow handling and once from the tick. If both call ->disable, we get a double hlist_del_rcu() and trigger a LIST_POISON2 dereference. Since we don't actually need to stop/start a swevent to re-programm the hardware (lack of hardware to program), simply nop out these callbacks for the swevent pmu. Reported-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <1275557609.27810.35218.camel@twins> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 858f56fa2432..31d6afe92594 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4055,13 +4055,6 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow, } } -static void perf_swevent_unthrottle(struct perf_event *event) -{ - /* - * Nothing to do, we already reset hwc->interrupts. - */ -} - static void perf_swevent_add(struct perf_event *event, u64 nr, int nmi, struct perf_sample_data *data, struct pt_regs *regs) @@ -4276,11 +4269,22 @@ static void perf_swevent_disable(struct perf_event *event) hlist_del_rcu(&event->hlist_entry); } +static void perf_swevent_void(struct perf_event *event) +{ +} + +static int perf_swevent_int(struct perf_event *event) +{ + return 0; +} + static const struct pmu perf_ops_generic = { .enable = perf_swevent_enable, .disable = perf_swevent_disable, + .start = perf_swevent_int, + .stop = perf_swevent_void, .read = perf_swevent_read, - .unthrottle = perf_swevent_unthrottle, + .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */ }; /* @@ -4561,8 +4565,10 @@ static int swevent_hlist_get(struct perf_event *event) static const struct pmu perf_ops_tracepoint = { .enable = perf_trace_enable, .disable = perf_trace_disable, + .start = perf_swevent_int, + .stop = perf_swevent_void, .read = perf_swevent_read, - .unthrottle = perf_swevent_unthrottle, + .unthrottle = perf_swevent_void, }; static int perf_tp_filter_match(struct perf_event *event, -- cgit v1.2.3