From 486efe9f8e30bac1e236f867df164f4966f3e207 Mon Sep 17 00:00:00 2001 From: Andrew Murray Date: Thu, 10 Jan 2019 13:53:24 +0000 Subject: perf/core: Add function to test for event exclusion flags Add a function that tests if any of the perf event exclusion flags are set on a given event. Signed-off-by: Andrew Murray Signed-off-by: Peter Zijlstra (Intel) Cc: Arnaldo Carvalho de Melo Cc: Benjamin Herrenschmidt Cc: Borislav Petkov Cc: Ivan Kokshaysky Cc: Linus Torvalds Cc: Mark Rutland Cc: Matt Turner Cc: Michael Ellerman Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Richard Henderson Cc: Russell King Cc: Sascha Hauer Cc: Shawn Guo Cc: Thomas Gleixner Cc: Will Deacon Cc: linux-arm-kernel@lists.infradead.org Cc: linuxppc-dev@lists.ozlabs.org Cc: robin.murphy@arm.com Cc: suzuki.poulose@arm.com Link: https://lkml.kernel.org/r/1547128414-50693-3-git-send-email-andrew.murray@arm.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 1d5c551a5add..54a78d22f0a6 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -1004,6 +1004,15 @@ perf_event__output_id_sample(struct perf_event *event, extern void perf_log_lost_samples(struct perf_event *event, u64 lost); +static inline bool event_has_any_exclude_flag(struct perf_event *event) +{ + struct perf_event_attr *attr = &event->attr; + + return attr->exclude_idle || attr->exclude_user || + attr->exclude_kernel || attr->exclude_hv || + attr->exclude_guest || attr->exclude_host; +} + static inline bool is_sampling_event(struct perf_event *event) { return event->attr.sample_period != 0; -- cgit v1.3-14-g43fede From cc6795aeffea0a80d0baf9ad31ba926a6c42cef5 Mon Sep 17 00:00:00 2001 From: Andrew Murray Date: Thu, 10 Jan 2019 13:53:25 +0000 Subject: perf/core: Add PERF_PMU_CAP_NO_EXCLUDE for exclusion incapable PMUs Many PMU drivers do not have the capability to exclude counting events that occur in specific contexts such as idle, kernel, guest, etc. These drivers indicate this by returning an error in their event_init upon testing the events attribute flags. This approach is error prone and often inconsistent. Let's instead allow PMU drivers to advertise their inability to exclude based on context via a new capability: PERF_PMU_CAP_NO_EXCLUDE. This allows the perf core to reject requests for exclusion events where there is no support in the PMU. Signed-off-by: Andrew Murray Signed-off-by: Peter Zijlstra (Intel) Cc: Arnaldo Carvalho de Melo Cc: Benjamin Herrenschmidt Cc: Borislav Petkov Cc: Ivan Kokshaysky Cc: Linus Torvalds Cc: Mark Rutland Cc: Matt Turner Cc: Michael Ellerman Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Richard Henderson Cc: Russell King Cc: Sascha Hauer Cc: Shawn Guo Cc: Thomas Gleixner Cc: Will Deacon Cc: linux-arm-kernel@lists.infradead.org Cc: linuxppc-dev@lists.ozlabs.org Cc: robin.murphy@arm.com Cc: suzuki.poulose@arm.com Link: https://lkml.kernel.org/r/1547128414-50693-4-git-send-email-andrew.murray@arm.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 1 + kernel/events/core.c | 9 +++++++++ 2 files changed, 10 insertions(+) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 54a78d22f0a6..cec02dc63b51 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -244,6 +244,7 @@ struct perf_event; #define PERF_PMU_CAP_EXCLUSIVE 0x10 #define PERF_PMU_CAP_ITRACE 0x20 #define PERF_PMU_CAP_HETEROGENEOUS_CPUS 0x40 +#define PERF_PMU_CAP_NO_EXCLUDE 0x80 /** * struct pmu - generic performance monitoring unit diff --git a/kernel/events/core.c b/kernel/events/core.c index 3cd13a30f732..fbe59b793b36 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -9772,6 +9772,15 @@ static int perf_try_init_event(struct pmu *pmu, struct perf_event *event) if (ctx) perf_event_ctx_unlock(event->group_leader, ctx); + if (!ret) { + if (pmu->capabilities & PERF_PMU_CAP_NO_EXCLUDE && + event_has_any_exclude_flag(event)) { + if (event->destroy) + event->destroy(event); + ret = -EINVAL; + } + } + if (ret) module_put(pmu->module); -- cgit v1.3-14-g43fede From ad07c8ceb6631a83b62d405a61448bba92adac68 Mon Sep 17 00:00:00 2001 From: Andrew Murray Date: Thu, 10 Jan 2019 13:53:34 +0000 Subject: perf/core: Remove unused perf_flags Now that perf_flags is not used we remove it. Signed-off-by: Andrew Murray Signed-off-by: Peter Zijlstra (Intel) Cc: Arnaldo Carvalho de Melo Cc: Benjamin Herrenschmidt Cc: Borislav Petkov Cc: Ivan Kokshaysky Cc: Linus Torvalds Cc: Mark Rutland Cc: Matt Turner Cc: Michael Ellerman Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Richard Henderson Cc: Russell King Cc: Sascha Hauer Cc: Shawn Guo Cc: Thomas Gleixner Cc: Will Deacon Cc: linux-arm-kernel@lists.infradead.org Cc: linuxppc-dev@lists.ozlabs.org Cc: robin.murphy@arm.com Cc: suzuki.poulose@arm.com Link: https://lkml.kernel.org/r/1547128414-50693-13-git-send-email-andrew.murray@arm.com Signed-off-by: Ingo Molnar --- include/uapi/linux/perf_event.h | 2 -- tools/include/uapi/linux/perf_event.h | 2 -- 2 files changed, 4 deletions(-) (limited to 'include') diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h index 9de8780ac8d9..ea19b5d491bf 100644 --- a/include/uapi/linux/perf_event.h +++ b/include/uapi/linux/perf_event.h @@ -445,8 +445,6 @@ struct perf_event_query_bpf { __u32 ids[0]; }; -#define perf_flags(attr) (*(&(attr)->read_format + 1)) - /* * Ioctls that can be done on a perf event fd: */ diff --git a/tools/include/uapi/linux/perf_event.h b/tools/include/uapi/linux/perf_event.h index 9de8780ac8d9..ea19b5d491bf 100644 --- a/tools/include/uapi/linux/perf_event.h +++ b/tools/include/uapi/linux/perf_event.h @@ -445,8 +445,6 @@ struct perf_event_query_bpf { __u32 ids[0]; }; -#define perf_flags(attr) (*(&(attr)->read_format + 1)) - /* * Ioctls that can be done on a perf event fd: */ -- cgit v1.3-14-g43fede From cf5c6c211b7e9eb4f4219f83671432c9ef257187 Mon Sep 17 00:00:00 2001 From: YueHaibing Date: Thu, 17 Jan 2019 15:25:04 +0800 Subject: perf: Remove duplicated workqueue.h include from perf_event.h It is already included a little bit higher up in that file. Signed-off-by: YueHaibing Cc: Alexander Shishkin Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20190117072504.14428-1-yuehaibing@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- include/linux/perf_event.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index cec02dc63b51..f8ec36197718 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -53,7 +53,6 @@ struct perf_guest_info_callbacks { #include #include #include -#include #include #include -- cgit v1.3-14-g43fede From 5620196951192f7cd2da0a04e7c0113f40bfc14e Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Jan 2019 13:20:20 -0300 Subject: perf: Make perf_event_output() propagate the output() return For the original mode of operation it isn't needed, since we report back errors via PERF_RECORD_LOST records in the ring buffer, but for use in bpf_perf_event_output() it is convenient to return the errors, basically -ENOSPC. Currently bpf_perf_event_output() returns an error indication, the last thing it does, which is to push it to the ring buffer is that can fail and if so, this failure won't be reported back to its users, fix it. Reported-by: Jamal Hadi Salim Tested-by: Jamal Hadi Salim Acked-by: Peter Zijlstra (Intel) Cc: Adrian Hunter Cc: Alexei Starovoitov Cc: Daniel Borkmann Cc: Jiri Olsa Cc: Namhyung Kim Link: https://lkml.kernel.org/r/20190118150938.GN5823@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- include/linux/perf_event.h | 6 +++--- kernel/events/core.c | 11 +++++++---- kernel/trace/bpf_trace.c | 3 +-- tools/perf/examples/bpf/augmented_raw_syscalls.c | 4 ++-- tools/perf/examples/bpf/augmented_syscalls.c | 14 +++++++------- tools/perf/examples/bpf/etcsnoop.c | 10 +++++----- 6 files changed, 25 insertions(+), 23 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index f8ec36197718..4eb88065a9b5 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -978,9 +978,9 @@ extern void perf_event_output_forward(struct perf_event *event, extern void perf_event_output_backward(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs); -extern void perf_event_output(struct perf_event *event, - struct perf_sample_data *data, - struct pt_regs *regs); +extern int perf_event_output(struct perf_event *event, + struct perf_sample_data *data, + struct pt_regs *regs); static inline bool is_default_overflow_handler(struct perf_event *event) diff --git a/kernel/events/core.c b/kernel/events/core.c index fbe59b793b36..bc525cd1615c 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -6489,7 +6489,7 @@ void perf_prepare_sample(struct perf_event_header *header, data->phys_addr = perf_virt_to_phys(data->addr); } -static __always_inline void +static __always_inline int __perf_event_output(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs, @@ -6499,13 +6499,15 @@ __perf_event_output(struct perf_event *event, { struct perf_output_handle handle; struct perf_event_header header; + int err; /* protect the callchain buffers */ rcu_read_lock(); perf_prepare_sample(&header, data, event, regs); - if (output_begin(&handle, event, header.size)) + err = output_begin(&handle, event, header.size); + if (err) goto exit; perf_output_sample(&handle, &header, data, event); @@ -6514,6 +6516,7 @@ __perf_event_output(struct perf_event *event, exit: rcu_read_unlock(); + return err; } void @@ -6532,12 +6535,12 @@ perf_event_output_backward(struct perf_event *event, __perf_event_output(event, data, regs, perf_output_begin_backward); } -void +int perf_event_output(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs) { - __perf_event_output(event, data, regs, perf_output_begin); + return __perf_event_output(event, data, regs, perf_output_begin); } /* diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 8b068adb9da1..088c2032ceaf 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -431,8 +431,7 @@ __bpf_perf_event_output(struct pt_regs *regs, struct bpf_map *map, if (unlikely(event->oncpu != cpu)) return -EOPNOTSUPP; - perf_event_output(event, sd, regs); - return 0; + return perf_event_output(event, sd, regs); } BPF_CALL_5(bpf_perf_event_output, struct pt_regs *, regs, struct bpf_map *, map, diff --git a/tools/perf/examples/bpf/augmented_raw_syscalls.c b/tools/perf/examples/bpf/augmented_raw_syscalls.c index 53c233370fae..9e9d4c66e53c 100644 --- a/tools/perf/examples/bpf/augmented_raw_syscalls.c +++ b/tools/perf/examples/bpf/augmented_raw_syscalls.c @@ -141,8 +141,8 @@ int sys_enter(struct syscall_enter_args *args) len = sizeof(augmented_args.args); } - perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, &augmented_args, len); - return 0; + /* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */ + return perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, &augmented_args, len); } SEC("raw_syscalls:sys_exit") diff --git a/tools/perf/examples/bpf/augmented_syscalls.c b/tools/perf/examples/bpf/augmented_syscalls.c index 2ae44813ef2d..b7dba114e36c 100644 --- a/tools/perf/examples/bpf/augmented_syscalls.c +++ b/tools/perf/examples/bpf/augmented_syscalls.c @@ -55,9 +55,9 @@ int syscall_enter(syscall)(struct syscall_enter_##syscall##_args *args) \ len -= sizeof(augmented_args.filename.value) - augmented_args.filename.size; \ len &= sizeof(augmented_args.filename.value) - 1; \ } \ - perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ - &augmented_args, len); \ - return 0; \ + /* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */ \ + return perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ + &augmented_args, len); \ } \ int syscall_exit(syscall)(struct syscall_exit_args *args) \ { \ @@ -125,10 +125,10 @@ int syscall_enter(syscall)(struct syscall_enter_##syscall##_args *args) \ /* addrlen = augmented_args.args.addrlen; */ \ /* */ \ probe_read(&augmented_args.addr, addrlen, args->addr_ptr); \ - perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ - &augmented_args, \ - sizeof(augmented_args) - sizeof(augmented_args.addr) + addrlen); \ - return 0; \ + /* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */ \ + return perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ + &augmented_args, \ + sizeof(augmented_args) - sizeof(augmented_args.addr) + addrlen);\ } \ int syscall_exit(syscall)(struct syscall_exit_args *args) \ { \ diff --git a/tools/perf/examples/bpf/etcsnoop.c b/tools/perf/examples/bpf/etcsnoop.c index b59e8812ee8c..550e69c2e8d1 100644 --- a/tools/perf/examples/bpf/etcsnoop.c +++ b/tools/perf/examples/bpf/etcsnoop.c @@ -49,11 +49,11 @@ int syscall_enter(syscall)(struct syscall_enter_##syscall##_args *args) \ args->filename_ptr); \ if (__builtin_memcmp(augmented_args.filename.value, etc, 4) != 0) \ return 0; \ - perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ - &augmented_args, \ - (sizeof(augmented_args) - sizeof(augmented_args.filename.value) + \ - augmented_args.filename.size)); \ - return 0; \ + /* If perf_event_output fails, return non-zero so that it gets recorded unaugmented */ \ + return perf_event_output(args, &__augmented_syscalls__, BPF_F_CURRENT_CPU, \ + &augmented_args, \ + (sizeof(augmented_args) - sizeof(augmented_args.filename.value) + \ + augmented_args.filename.size)); \ } struct syscall_enter_openat_args { -- cgit v1.3-14-g43fede From 76193a94522f1d4edf2447a536f3f796ce56343b Mon Sep 17 00:00:00 2001 From: Song Liu Date: Thu, 17 Jan 2019 08:15:13 -0800 Subject: perf, bpf: Introduce PERF_RECORD_KSYMBOL For better performance analysis of dynamically JITed and loaded kernel functions, such as BPF programs, this patch introduces PERF_RECORD_KSYMBOL, a new perf_event_type that exposes kernel symbol register/unregister information to user space. The following data structure is used for PERF_RECORD_KSYMBOL. /* * struct { * struct perf_event_header header; * u64 addr; * u32 len; * u16 ksym_type; * u16 flags; * char name[]; * struct sample_id sample_id; * }; */ Signed-off-by: Song Liu Reviewed-by: Arnaldo Carvalho de Melo Tested-by: Arnaldo Carvalho de Melo Acked-by: Peter Zijlstra Cc: Alexei Starovoitov Cc: Daniel Borkmann Cc: Peter Zijlstra Cc: kernel-team@fb.com Cc: netdev@vger.kernel.org Link: http://lkml.kernel.org/r/20190117161521.1341602-2-songliubraving@fb.com Signed-off-by: Arnaldo Carvalho de Melo --- include/linux/perf_event.h | 8 ++++ include/uapi/linux/perf_event.h | 26 ++++++++++- kernel/events/core.c | 98 ++++++++++++++++++++++++++++++++++++++++- 3 files changed, 130 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 4eb88065a9b5..136fe0495374 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -1122,6 +1122,10 @@ static inline void perf_event_task_sched_out(struct task_struct *prev, } extern void perf_event_mmap(struct vm_area_struct *vma); + +extern void perf_event_ksymbol(u16 ksym_type, u64 addr, u32 len, + bool unregister, const char *sym); + extern struct perf_guest_info_callbacks *perf_guest_cbs; extern int perf_register_guest_info_callbacks(struct perf_guest_info_callbacks *callbacks); extern int perf_unregister_guest_info_callbacks(struct perf_guest_info_callbacks *callbacks); @@ -1342,6 +1346,10 @@ static inline int perf_unregister_guest_info_callbacks (struct perf_guest_info_callbacks *callbacks) { return 0; } static inline void perf_event_mmap(struct vm_area_struct *vma) { } + +typedef int (perf_ksymbol_get_name_f)(char *name, int name_len, void *data); +static inline void perf_event_ksymbol(u16 ksym_type, u64 addr, u32 len, + bool unregister, const char *sym) { } static inline void perf_event_exec(void) { } static inline void perf_event_comm(struct task_struct *tsk, bool exec) { } static inline void perf_event_namespaces(struct task_struct *tsk) { } diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h index ea19b5d491bf..1dee5c8f166b 100644 --- a/include/uapi/linux/perf_event.h +++ b/include/uapi/linux/perf_event.h @@ -372,7 +372,8 @@ struct perf_event_attr { context_switch : 1, /* context switch data */ write_backward : 1, /* Write ring buffer from end to beginning */ namespaces : 1, /* include namespaces data */ - __reserved_1 : 35; + ksymbol : 1, /* include ksymbol events */ + __reserved_1 : 34; union { __u32 wakeup_events; /* wakeup every n events */ @@ -963,9 +964,32 @@ enum perf_event_type { */ PERF_RECORD_NAMESPACES = 16, + /* + * Record ksymbol register/unregister events: + * + * struct { + * struct perf_event_header header; + * u64 addr; + * u32 len; + * u16 ksym_type; + * u16 flags; + * char name[]; + * struct sample_id sample_id; + * }; + */ + PERF_RECORD_KSYMBOL = 17, + PERF_RECORD_MAX, /* non-ABI */ }; +enum perf_record_ksymbol_type { + PERF_RECORD_KSYMBOL_TYPE_UNKNOWN = 0, + PERF_RECORD_KSYMBOL_TYPE_BPF = 1, + PERF_RECORD_KSYMBOL_TYPE_MAX /* non-ABI */ +}; + +#define PERF_RECORD_KSYMBOL_FLAGS_UNREGISTER (1 << 0) + #define PERF_MAX_STACK_DEPTH 127 #define PERF_MAX_CONTEXTS_PER_STACK 8 diff --git a/kernel/events/core.c b/kernel/events/core.c index bc525cd1615c..e04ab5f325cf 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -385,6 +385,7 @@ static atomic_t nr_namespaces_events __read_mostly; static atomic_t nr_task_events __read_mostly; static atomic_t nr_freq_events __read_mostly; static atomic_t nr_switch_events __read_mostly; +static atomic_t nr_ksymbol_events __read_mostly; static LIST_HEAD(pmus); static DEFINE_MUTEX(pmus_lock); @@ -4235,7 +4236,7 @@ static bool is_sb_event(struct perf_event *event) if (attr->mmap || attr->mmap_data || attr->mmap2 || attr->comm || attr->comm_exec || - attr->task || + attr->task || attr->ksymbol || attr->context_switch) return true; return false; @@ -4305,6 +4306,8 @@ static void unaccount_event(struct perf_event *event) dec = true; if (has_branch_stack(event)) dec = true; + if (event->attr.ksymbol) + atomic_dec(&nr_ksymbol_events); if (dec) { if (!atomic_add_unless(&perf_sched_count, -1, 1)) @@ -7653,6 +7656,97 @@ static void perf_log_throttle(struct perf_event *event, int enable) perf_output_end(&handle); } +/* + * ksymbol register/unregister tracking + */ + +struct perf_ksymbol_event { + const char *name; + int name_len; + struct { + struct perf_event_header header; + u64 addr; + u32 len; + u16 ksym_type; + u16 flags; + } event_id; +}; + +static int perf_event_ksymbol_match(struct perf_event *event) +{ + return event->attr.ksymbol; +} + +static void perf_event_ksymbol_output(struct perf_event *event, void *data) +{ + struct perf_ksymbol_event *ksymbol_event = data; + struct perf_output_handle handle; + struct perf_sample_data sample; + int ret; + + if (!perf_event_ksymbol_match(event)) + return; + + perf_event_header__init_id(&ksymbol_event->event_id.header, + &sample, event); + ret = perf_output_begin(&handle, event, + ksymbol_event->event_id.header.size); + if (ret) + return; + + perf_output_put(&handle, ksymbol_event->event_id); + __output_copy(&handle, ksymbol_event->name, ksymbol_event->name_len); + perf_event__output_id_sample(event, &handle, &sample); + + perf_output_end(&handle); +} + +void perf_event_ksymbol(u16 ksym_type, u64 addr, u32 len, bool unregister, + const char *sym) +{ + struct perf_ksymbol_event ksymbol_event; + char name[KSYM_NAME_LEN]; + u16 flags = 0; + int name_len; + + if (!atomic_read(&nr_ksymbol_events)) + return; + + if (ksym_type >= PERF_RECORD_KSYMBOL_TYPE_MAX || + ksym_type == PERF_RECORD_KSYMBOL_TYPE_UNKNOWN) + goto err; + + strlcpy(name, sym, KSYM_NAME_LEN); + name_len = strlen(name) + 1; + while (!IS_ALIGNED(name_len, sizeof(u64))) + name[name_len++] = '\0'; + BUILD_BUG_ON(KSYM_NAME_LEN % sizeof(u64)); + + if (unregister) + flags |= PERF_RECORD_KSYMBOL_FLAGS_UNREGISTER; + + ksymbol_event = (struct perf_ksymbol_event){ + .name = name, + .name_len = name_len, + .event_id = { + .header = { + .type = PERF_RECORD_KSYMBOL, + .size = sizeof(ksymbol_event.event_id) + + name_len, + }, + .addr = addr, + .len = len, + .ksym_type = ksym_type, + .flags = flags, + }, + }; + + perf_iterate_sb(perf_event_ksymbol_output, &ksymbol_event, NULL); + return; +err: + WARN_ONCE(1, "%s: Invalid KSYMBOL type 0x%x\n", __func__, ksym_type); +} + void perf_event_itrace_started(struct perf_event *event) { event->attach_state |= PERF_ATTACH_ITRACE; @@ -9912,6 +10006,8 @@ static void account_event(struct perf_event *event) inc = true; if (is_cgroup_event(event)) inc = true; + if (event->attr.ksymbol) + atomic_inc(&nr_ksymbol_events); if (inc) { /* -- cgit v1.3-14-g43fede From 6ee52e2a3fe4ea35520720736e6791df1fb67106 Mon Sep 17 00:00:00 2001 From: Song Liu Date: Thu, 17 Jan 2019 08:15:15 -0800 Subject: perf, bpf: Introduce PERF_RECORD_BPF_EVENT For better performance analysis of BPF programs, this patch introduces PERF_RECORD_BPF_EVENT, a new perf_event_type that exposes BPF program load/unload information to user space. Each BPF program may contain up to BPF_MAX_SUBPROGS (256) sub programs. The following example shows kernel symbols for a BPF program with 7 sub programs: ffffffffa0257cf9 t bpf_prog_b07ccb89267cf242_F ffffffffa02592e1 t bpf_prog_2dcecc18072623fc_F ffffffffa025b0e9 t bpf_prog_bb7a405ebaec5d5c_F ffffffffa025dd2c t bpf_prog_a7540d4a39ec1fc7_F ffffffffa025fcca t bpf_prog_05762d4ade0e3737_F ffffffffa026108f t bpf_prog_db4bd11e35df90d4_F ffffffffa0263f00 t bpf_prog_89d64e4abf0f0126_F ffffffffa0257cf9 t bpf_prog_ae31629322c4b018__dummy_tracepoi When a bpf program is loaded, PERF_RECORD_KSYMBOL is generated for each of these sub programs. Therefore, PERF_RECORD_BPF_EVENT is not needed for simple profiling. For annotation, user space need to listen to PERF_RECORD_BPF_EVENT and gather more information about these (sub) programs via sys_bpf. Signed-off-by: Song Liu Reviewed-by: Arnaldo Carvalho de Melo Acked-by: Alexei Starovoitov Acked-by: Peter Zijlstra (Intel) Tested-by: Arnaldo Carvalho de Melo Cc: Daniel Borkmann Cc: Peter Zijlstra Cc: kernel-team@fb.com Cc: netdev@vger.kernel.org Link: http://lkml.kernel.org/r/20190117161521.1341602-4-songliubraving@fb.com Signed-off-by: Arnaldo Carvalho de Melo --- include/linux/filter.h | 7 +++ include/linux/perf_event.h | 6 +++ include/uapi/linux/perf_event.h | 29 +++++++++- kernel/bpf/core.c | 2 +- kernel/bpf/syscall.c | 2 + kernel/events/core.c | 115 ++++++++++++++++++++++++++++++++++++++++ 6 files changed, 159 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/filter.h b/include/linux/filter.h index ad106d845b22..d531d4250bff 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -951,6 +951,7 @@ bpf_address_lookup(unsigned long addr, unsigned long *size, void bpf_prog_kallsyms_add(struct bpf_prog *fp); void bpf_prog_kallsyms_del(struct bpf_prog *fp); +void bpf_get_prog_name(const struct bpf_prog *prog, char *sym); #else /* CONFIG_BPF_JIT */ @@ -1006,6 +1007,12 @@ static inline void bpf_prog_kallsyms_add(struct bpf_prog *fp) static inline void bpf_prog_kallsyms_del(struct bpf_prog *fp) { } + +static inline void bpf_get_prog_name(const struct bpf_prog *prog, char *sym) +{ + sym[0] = '\0'; +} + #endif /* CONFIG_BPF_JIT */ void bpf_prog_kallsyms_del_subprogs(struct bpf_prog *fp); diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 136fe0495374..a79e59fc3b7d 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -1125,6 +1125,9 @@ extern void perf_event_mmap(struct vm_area_struct *vma); extern void perf_event_ksymbol(u16 ksym_type, u64 addr, u32 len, bool unregister, const char *sym); +extern void perf_event_bpf_event(struct bpf_prog *prog, + enum perf_bpf_event_type type, + u16 flags); extern struct perf_guest_info_callbacks *perf_guest_cbs; extern int perf_register_guest_info_callbacks(struct perf_guest_info_callbacks *callbacks); @@ -1350,6 +1353,9 @@ static inline void perf_event_mmap(struct vm_area_struct *vma) { } typedef int (perf_ksymbol_get_name_f)(char *name, int name_len, void *data); static inline void perf_event_ksymbol(u16 ksym_type, u64 addr, u32 len, bool unregister, const char *sym) { } +static inline void perf_event_bpf_event(struct bpf_prog *prog, + enum perf_bpf_event_type type, + u16 flags) { } static inline void perf_event_exec(void) { } static inline void perf_event_comm(struct task_struct *tsk, bool exec) { } static inline void perf_event_namespaces(struct task_struct *tsk) { } diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h index 1dee5c8f166b..7198ddd0c6b1 100644 --- a/include/uapi/linux/perf_event.h +++ b/include/uapi/linux/perf_event.h @@ -373,7 +373,8 @@ struct perf_event_attr { write_backward : 1, /* Write ring buffer from end to beginning */ namespaces : 1, /* include namespaces data */ ksymbol : 1, /* include ksymbol events */ - __reserved_1 : 34; + bpf_event : 1, /* include bpf events */ + __reserved_1 : 33; union { __u32 wakeup_events; /* wakeup every n events */ @@ -979,6 +980,25 @@ enum perf_event_type { */ PERF_RECORD_KSYMBOL = 17, + /* + * Record bpf events: + * enum perf_bpf_event_type { + * PERF_BPF_EVENT_UNKNOWN = 0, + * PERF_BPF_EVENT_PROG_LOAD = 1, + * PERF_BPF_EVENT_PROG_UNLOAD = 2, + * }; + * + * struct { + * struct perf_event_header header; + * u16 type; + * u16 flags; + * u32 id; + * u8 tag[BPF_TAG_SIZE]; + * struct sample_id sample_id; + * }; + */ + PERF_RECORD_BPF_EVENT = 18, + PERF_RECORD_MAX, /* non-ABI */ }; @@ -990,6 +1010,13 @@ enum perf_record_ksymbol_type { #define PERF_RECORD_KSYMBOL_FLAGS_UNREGISTER (1 << 0) +enum perf_bpf_event_type { + PERF_BPF_EVENT_UNKNOWN = 0, + PERF_BPF_EVENT_PROG_LOAD = 1, + PERF_BPF_EVENT_PROG_UNLOAD = 2, + PERF_BPF_EVENT_MAX, /* non-ABI */ +}; + #define PERF_MAX_STACK_DEPTH 127 #define PERF_MAX_CONTEXTS_PER_STACK 8 diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index f908b9356025..19c49313c709 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -495,7 +495,7 @@ bpf_get_prog_addr_region(const struct bpf_prog *prog, *symbol_end = addr + hdr->pages * PAGE_SIZE; } -static void bpf_get_prog_name(const struct bpf_prog *prog, char *sym) +void bpf_get_prog_name(const struct bpf_prog *prog, char *sym) { const char *end = sym + KSYM_NAME_LEN; const struct btf_type *type; diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index b155cd17c1bd..30ebd085790b 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1211,6 +1211,7 @@ static void __bpf_prog_put_rcu(struct rcu_head *rcu) static void __bpf_prog_put(struct bpf_prog *prog, bool do_idr_lock) { if (atomic_dec_and_test(&prog->aux->refcnt)) { + perf_event_bpf_event(prog, PERF_BPF_EVENT_PROG_UNLOAD, 0); /* bpf_prog_free_id() must be called first */ bpf_prog_free_id(prog, do_idr_lock); bpf_prog_kallsyms_del_all(prog); @@ -1554,6 +1555,7 @@ static int bpf_prog_load(union bpf_attr *attr, union bpf_attr __user *uattr) } bpf_prog_kallsyms_add(prog); + perf_event_bpf_event(prog, PERF_BPF_EVENT_PROG_LOAD, 0); return err; free_used_maps: diff --git a/kernel/events/core.c b/kernel/events/core.c index e04ab5f325cf..236bb8ddb7bc 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -386,6 +386,7 @@ static atomic_t nr_task_events __read_mostly; static atomic_t nr_freq_events __read_mostly; static atomic_t nr_switch_events __read_mostly; static atomic_t nr_ksymbol_events __read_mostly; +static atomic_t nr_bpf_events __read_mostly; static LIST_HEAD(pmus); static DEFINE_MUTEX(pmus_lock); @@ -4308,6 +4309,8 @@ static void unaccount_event(struct perf_event *event) dec = true; if (event->attr.ksymbol) atomic_dec(&nr_ksymbol_events); + if (event->attr.bpf_event) + atomic_dec(&nr_bpf_events); if (dec) { if (!atomic_add_unless(&perf_sched_count, -1, 1)) @@ -7747,6 +7750,116 @@ err: WARN_ONCE(1, "%s: Invalid KSYMBOL type 0x%x\n", __func__, ksym_type); } +/* + * bpf program load/unload tracking + */ + +struct perf_bpf_event { + struct bpf_prog *prog; + struct { + struct perf_event_header header; + u16 type; + u16 flags; + u32 id; + u8 tag[BPF_TAG_SIZE]; + } event_id; +}; + +static int perf_event_bpf_match(struct perf_event *event) +{ + return event->attr.bpf_event; +} + +static void perf_event_bpf_output(struct perf_event *event, void *data) +{ + struct perf_bpf_event *bpf_event = data; + struct perf_output_handle handle; + struct perf_sample_data sample; + int ret; + + if (!perf_event_bpf_match(event)) + return; + + perf_event_header__init_id(&bpf_event->event_id.header, + &sample, event); + ret = perf_output_begin(&handle, event, + bpf_event->event_id.header.size); + if (ret) + return; + + perf_output_put(&handle, bpf_event->event_id); + perf_event__output_id_sample(event, &handle, &sample); + + perf_output_end(&handle); +} + +static void perf_event_bpf_emit_ksymbols(struct bpf_prog *prog, + enum perf_bpf_event_type type) +{ + bool unregister = type == PERF_BPF_EVENT_PROG_UNLOAD; + char sym[KSYM_NAME_LEN]; + int i; + + if (prog->aux->func_cnt == 0) { + bpf_get_prog_name(prog, sym); + perf_event_ksymbol(PERF_RECORD_KSYMBOL_TYPE_BPF, + (u64)(unsigned long)prog->bpf_func, + prog->jited_len, unregister, sym); + } else { + for (i = 0; i < prog->aux->func_cnt; i++) { + struct bpf_prog *subprog = prog->aux->func[i]; + + bpf_get_prog_name(subprog, sym); + perf_event_ksymbol( + PERF_RECORD_KSYMBOL_TYPE_BPF, + (u64)(unsigned long)subprog->bpf_func, + subprog->jited_len, unregister, sym); + } + } +} + +void perf_event_bpf_event(struct bpf_prog *prog, + enum perf_bpf_event_type type, + u16 flags) +{ + struct perf_bpf_event bpf_event; + + if (type <= PERF_BPF_EVENT_UNKNOWN || + type >= PERF_BPF_EVENT_MAX) + return; + + switch (type) { + case PERF_BPF_EVENT_PROG_LOAD: + case PERF_BPF_EVENT_PROG_UNLOAD: + if (atomic_read(&nr_ksymbol_events)) + perf_event_bpf_emit_ksymbols(prog, type); + break; + default: + break; + } + + if (!atomic_read(&nr_bpf_events)) + return; + + bpf_event = (struct perf_bpf_event){ + .prog = prog, + .event_id = { + .header = { + .type = PERF_RECORD_BPF_EVENT, + .size = sizeof(bpf_event.event_id), + }, + .type = type, + .flags = flags, + .id = prog->aux->id, + }, + }; + + BUILD_BUG_ON(BPF_TAG_SIZE % sizeof(u64)); + + memcpy(bpf_event.event_id.tag, prog->tag, BPF_TAG_SIZE); + perf_iterate_sb(perf_event_bpf_output, &bpf_event, NULL); +} + void perf_event_itrace_started(struct perf_event *event) { event->attach_state |= PERF_ATTACH_ITRACE; @@ -10008,6 +10121,8 @@ static void account_event(struct perf_event *event) inc = true; if (event->attr.ksymbol) atomic_inc(&nr_ksymbol_events); + if (event->attr.bpf_event) + atomic_inc(&nr_bpf_events); if (inc) { /* -- cgit v1.3-14-g43fede From 8c94abbbe1ba24961278055434504b7dc3595415 Mon Sep 17 00:00:00 2001 From: Elena Reshetova Date: Mon, 28 Jan 2019 14:27:26 +0200 Subject: perf: Convert perf_event_context.refcount to refcount_t atomic_t variables are currently used to implement reference counters with the following properties: - counter is initialized to 1 using atomic_set() - a resource is freed upon counter reaching zero - once counter reaches zero, its further increments aren't allowed - counter schema uses basic atomic operations (set, inc, inc_not_zero, dec_and_test, etc.) Such atomic variables should be converted to a newly provided refcount_t type and API that prevents accidental counter overflows and underflows. This is important since overflows and underflows can lead to use-after-free situation and be exploitable. The variable perf_event_context.refcount is used as pure reference counter. Convert it to refcount_t and fix up the operations. ** Important note for maintainers: Some functions from refcount_t API defined in lib/refcount.c have different memory ordering guarantees than their atomic counterparts. Please check Documentation/core-api/refcount-vs-atomic.rst for more information. Normally the differences should not matter since refcount_t provides enough guarantees to satisfy the refcounting use cases, but in some rare cases it might matter. Please double check that you don't have some undocumented memory guarantees for this variable usage. For the perf_event_context.refcount it might make a difference in following places: - get_ctx(), perf_event_ctx_lock_nested(), perf_lock_task_context() and __perf_event_ctx_lock_double(): increment in refcount_inc_not_zero() only guarantees control dependency on success vs. fully ordered atomic counterpart - put_ctx(): decrement in refcount_dec_and_test() provides RELEASE ordering and ACQUIRE ordering + control dependency on success vs. fully ordered atomic counterpart Suggested-by: Kees Cook Signed-off-by: Elena Reshetova Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: David Windsor Reviewed-by: Hans Liljestrand Cc: Alexander Shishkin Cc: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: acme@kernel.org Cc: namhyung@kernel.org Link: https://lkml.kernel.org/r/1548678448-24458-2-git-send-email-elena.reshetova@intel.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 3 ++- kernel/events/core.c | 12 ++++++------ 2 files changed, 8 insertions(+), 7 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index a79e59fc3b7d..6cb5d483ab34 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -54,6 +54,7 @@ struct perf_guest_info_callbacks { #include #include #include +#include #include struct perf_callchain_entry { @@ -737,7 +738,7 @@ struct perf_event_context { int nr_stat; int nr_freq; int rotate_disable; - atomic_t refcount; + refcount_t refcount; struct task_struct *task; /* diff --git a/kernel/events/core.c b/kernel/events/core.c index 5b89de7918d0..677164d54547 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -1172,7 +1172,7 @@ static void perf_event_ctx_deactivate(struct perf_event_context *ctx) static void get_ctx(struct perf_event_context *ctx) { - WARN_ON(!atomic_inc_not_zero(&ctx->refcount)); + refcount_inc(&ctx->refcount); } static void free_ctx(struct rcu_head *head) @@ -1186,7 +1186,7 @@ static void free_ctx(struct rcu_head *head) static void put_ctx(struct perf_event_context *ctx) { - if (atomic_dec_and_test(&ctx->refcount)) { + if (refcount_dec_and_test(&ctx->refcount)) { if (ctx->parent_ctx) put_ctx(ctx->parent_ctx); if (ctx->task && ctx->task != TASK_TOMBSTONE) @@ -1268,7 +1268,7 @@ perf_event_ctx_lock_nested(struct perf_event *event, int nesting) again: rcu_read_lock(); ctx = READ_ONCE(event->ctx); - if (!atomic_inc_not_zero(&ctx->refcount)) { + if (!refcount_inc_not_zero(&ctx->refcount)) { rcu_read_unlock(); goto again; } @@ -1401,7 +1401,7 @@ retry: } if (ctx->task == TASK_TOMBSTONE || - !atomic_inc_not_zero(&ctx->refcount)) { + !refcount_inc_not_zero(&ctx->refcount)) { raw_spin_unlock(&ctx->lock); ctx = NULL; } else { @@ -4057,7 +4057,7 @@ static void __perf_event_init_context(struct perf_event_context *ctx) INIT_LIST_HEAD(&ctx->event_list); INIT_LIST_HEAD(&ctx->pinned_active); INIT_LIST_HEAD(&ctx->flexible_active); - atomic_set(&ctx->refcount, 1); + refcount_set(&ctx->refcount, 1); } static struct perf_event_context * @@ -10613,7 +10613,7 @@ __perf_event_ctx_lock_double(struct perf_event *group_leader, again: rcu_read_lock(); gctx = READ_ONCE(group_leader->ctx); - if (!atomic_inc_not_zero(&gctx->refcount)) { + if (!refcount_inc_not_zero(&gctx->refcount)) { rcu_read_unlock(); goto again; } -- cgit v1.3-14-g43fede From 840018668ce2d96783356204ff282d6c9b0e5f66 Mon Sep 17 00:00:00 2001 From: Mathieu Poirier Date: Thu, 31 Jan 2019 11:47:08 -0700 Subject: perf/aux: Make perf_event accessible to setup_aux() When pmu::setup_aux() is called the coresight PMU needs to know which sink to use for the session by looking up the information in the event's attr::config2 field. As such simply replace the cpu information by the complete perf_event structure and change all affected customers. Signed-off-by: Mathieu Poirier Reviewed-by: Suzuki Poulouse Acked-by: Peter Zijlstra Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Greg Kroah-Hartman Cc: H. Peter Anvin Cc: Heiko Carstens Cc: Jiri Olsa Cc: Mark Rutland Cc: Martin Schwidefsky Cc: Namhyung Kim Cc: Thomas Gleixner Cc: Will Deacon Cc: linux-arm-kernel@lists.infradead.org Cc: linux-s390@vger.kernel.org Link: http://lkml.kernel.org/r/20190131184714.20388-2-mathieu.poirier@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- arch/s390/kernel/perf_cpum_sf.c | 6 +++--- arch/x86/events/intel/bts.c | 4 +++- arch/x86/events/intel/pt.c | 5 +++-- drivers/hwtracing/coresight/coresight-etm-perf.c | 6 +++--- drivers/perf/arm_spe_pmu.c | 6 +++--- include/linux/perf_event.h | 2 +- kernel/events/ring_buffer.c | 2 +- 7 files changed, 17 insertions(+), 14 deletions(-) (limited to 'include') diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c index bfabeb1889cc..1266194afb02 100644 --- a/arch/s390/kernel/perf_cpum_sf.c +++ b/arch/s390/kernel/perf_cpum_sf.c @@ -1600,7 +1600,7 @@ static void aux_sdb_init(unsigned long sdb) /* * aux_buffer_setup() - Setup AUX buffer for diagnostic mode sampling - * @cpu: On which to allocate, -1 means current + * @event: Event the buffer is setup for, event->cpu == -1 means current * @pages: Array of pointers to buffer pages passed from perf core * @nr_pages: Total pages * @snapshot: Flag for snapshot mode @@ -1612,8 +1612,8 @@ static void aux_sdb_init(unsigned long sdb) * * Return the private AUX buffer structure if success or NULL if fails. */ -static void *aux_buffer_setup(int cpu, void **pages, int nr_pages, - bool snapshot) +static void *aux_buffer_setup(struct perf_event *event, void **pages, + int nr_pages, bool snapshot) { struct sf_buffer *sfb; struct aux_buffer *aux; diff --git a/arch/x86/events/intel/bts.c b/arch/x86/events/intel/bts.c index a01ef1b0f883..7cdd7b13bbda 100644 --- a/arch/x86/events/intel/bts.c +++ b/arch/x86/events/intel/bts.c @@ -77,10 +77,12 @@ static size_t buf_size(struct page *page) } static void * -bts_buffer_setup_aux(int cpu, void **pages, int nr_pages, bool overwrite) +bts_buffer_setup_aux(struct perf_event *event, void **pages, + int nr_pages, bool overwrite) { struct bts_buffer *buf; struct page *page; + int cpu = event->cpu; int node = (cpu == -1) ? cpu : cpu_to_node(cpu); unsigned long offset; size_t size = nr_pages << PAGE_SHIFT; diff --git a/arch/x86/events/intel/pt.c b/arch/x86/events/intel/pt.c index 9494ca68fd9d..c0e86ff21f81 100644 --- a/arch/x86/events/intel/pt.c +++ b/arch/x86/events/intel/pt.c @@ -1114,10 +1114,11 @@ static int pt_buffer_init_topa(struct pt_buffer *buf, unsigned long nr_pages, * Return: Our private PT buffer structure. */ static void * -pt_buffer_setup_aux(int cpu, void **pages, int nr_pages, bool snapshot) +pt_buffer_setup_aux(struct perf_event *event, void **pages, + int nr_pages, bool snapshot) { struct pt_buffer *buf; - int node, ret; + int node, ret, cpu = event->cpu; if (!nr_pages) return NULL; diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index abe8249b893b..f21eb28b6782 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -177,15 +177,15 @@ static void etm_free_aux(void *data) schedule_work(&event_data->work); } -static void *etm_setup_aux(int event_cpu, void **pages, +static void *etm_setup_aux(struct perf_event *event, void **pages, int nr_pages, bool overwrite) { - int cpu; + int cpu = event->cpu; cpumask_t *mask; struct coresight_device *sink; struct etm_event_data *event_data = NULL; - event_data = alloc_event_data(event_cpu); + event_data = alloc_event_data(cpu); if (!event_data) return NULL; INIT_WORK(&event_data->work, free_event_data); diff --git a/drivers/perf/arm_spe_pmu.c b/drivers/perf/arm_spe_pmu.c index 8e46a9dad2fa..7cb766dafe85 100644 --- a/drivers/perf/arm_spe_pmu.c +++ b/drivers/perf/arm_spe_pmu.c @@ -824,10 +824,10 @@ static void arm_spe_pmu_read(struct perf_event *event) { } -static void *arm_spe_pmu_setup_aux(int cpu, void **pages, int nr_pages, - bool snapshot) +static void *arm_spe_pmu_setup_aux(struct perf_event *event, void **pages, + int nr_pages, bool snapshot) { - int i; + int i, cpu = event->cpu; struct page **pglist; struct arm_spe_pmu_buf *buf; diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 6cb5d483ab34..d9c3610e0e25 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -410,7 +410,7 @@ struct pmu { /* * Set up pmu-private data structures for an AUX area */ - void *(*setup_aux) (int cpu, void **pages, + void *(*setup_aux) (struct perf_event *event, void **pages, int nr_pages, bool overwrite); /* optional */ diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c index 805f0423ee0b..70ae2422cbaf 100644 --- a/kernel/events/ring_buffer.c +++ b/kernel/events/ring_buffer.c @@ -657,7 +657,7 @@ int rb_alloc_aux(struct ring_buffer *rb, struct perf_event *event, goto out; } - rb->aux_priv = event->pmu->setup_aux(event->cpu, rb->aux_pages, nr_pages, + rb->aux_priv = event->pmu->setup_aux(event, rb->aux_pages, nr_pages, overwrite); if (!rb->aux_priv) goto out; -- cgit v1.3-14-g43fede From bb8e370bdc141ddff526e5e5ee74210c91fee0b8 Mon Sep 17 00:00:00 2001 From: Mathieu Poirier Date: Thu, 31 Jan 2019 11:47:09 -0700 Subject: coresight: perf: Add "sinks" group to PMU directory Add a "sinks" directory entry so that users can see all the sinks available in the system in a single place. Individual sink are added as they are registered with the coresight bus. Committer tests: Test built on a ubuntu 18.04 container with a cross build environment to arm64, the new field is there, need to find a machine with this feature to do further testing in the future. root@d15263e5734a:/git/perf# grep CORESIGHT /tmp/build/v5.0-rc2+/.config CONFIG_CORESIGHT=y CONFIG_CORESIGHT_LINKS_AND_SINKS=y CONFIG_CORESIGHT_LINK_AND_SINK_TMC=y CONFIG_CORESIGHT_CATU=y CONFIG_CORESIGHT_SINK_TPIU=y CONFIG_CORESIGHT_SINK_ETBV10=y CONFIG_CORESIGHT_SOURCE_ETM4X=y CONFIG_CORESIGHT_DYNAMIC_REPLICATOR=y CONFIG_CORESIGHT_STM=y CONFIG_CORESIGHT_CPU_DEBUG=m root@d15263e5734a:/git/perf# root@d15263e5734a:/git/perf# file /tmp/build/v5.0-rc2+/drivers/hwtracing/coresight/*.o .../coresight/coresight-catu.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-cpu-debug.mod.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-cpu-debug.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-dynamic-replicator.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-etb10.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-etm-perf.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-etm4x-sysfs.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-etm4x.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-funnel.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-replicator.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-stm.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-tmc-etf.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-tmc-etr.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-tmc.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight-tpiu.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/coresight.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped .../coresight/of_coresight.o: ELF 64-bit MSB relocatable, ARM aarch64, version 1 (SYSV), not stripped root@d15263e5734a:/git/perf# root@d15263e5734a:/git/perf# pahole -C coresight_device /tmp/build/v5.0-rc2+/drivers/hwtracing/coresight/coresight.o struct coresight_device { struct coresight_connection * conns; /* 0 8 */ int nr_inport; /* 8 4 */ int nr_outport; /* 12 4 */ enum coresight_dev_type type; /* 16 4 */ union coresight_dev_subtype subtype; /* 20 8 */ /* XXX 4 bytes hole, try to pack */ const struct coresight_ops * ops; /* 32 8 */ struct device dev; /* 40 1408 */ /* XXX last struct has 7 bytes of padding */ /* --- cacheline 22 boundary (1408 bytes) was 40 bytes ago --- */ atomic_t * refcnt; /* 1448 8 */ bool orphan; /* 1456 1 */ bool enable; /* 1457 1 */ bool activated; /* 1458 1 */ /* XXX 5 bytes hole, try to pack */ struct dev_ext_attribute * ea; /* 1464 8 */ /* size: 1472, cachelines: 23, members: 12 */ /* sum members: 1463, holes: 2, sum holes: 9 */ /* paddings: 1, sum paddings: 7 */ }; root@d15263e5734a:/git/perf# Signed-off-by: Mathieu Poirier Reviewed-by: Suzuki K Poulose Acked-by: Peter Zijlstra Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Greg Kroah-Hartman Cc: H. Peter Anvin Cc: Heiko Carstens Cc: Jiri Olsa Cc: Mark Rutland Cc: Martin Schwidefsky Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Will Deacon Cc: linux-arm-kernel@lists.infradead.org Cc: linux-s390@vger.kernel.org Link: http://lkml.kernel.org/r/20190131184714.20388-3-mathieu.poirier@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- drivers/hwtracing/coresight/coresight-etm-perf.c | 82 ++++++++++++++++++++++++ drivers/hwtracing/coresight/coresight-etm-perf.h | 6 +- drivers/hwtracing/coresight/coresight.c | 18 ++++++ include/linux/coresight.h | 7 +- 4 files changed, 110 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index f21eb28b6782..cdbdb28dc175 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -14,6 +14,7 @@ #include #include #include +#include #include #include @@ -43,8 +44,18 @@ static const struct attribute_group etm_pmu_format_group = { .attrs = etm_config_formats_attr, }; +static struct attribute *etm_config_sinks_attr[] = { + NULL, +}; + +static const struct attribute_group etm_pmu_sinks_group = { + .name = "sinks", + .attrs = etm_config_sinks_attr, +}; + static const struct attribute_group *etm_pmu_attr_groups[] = { &etm_pmu_format_group, + &etm_pmu_sinks_group, NULL, }; @@ -479,6 +490,77 @@ int etm_perf_symlink(struct coresight_device *csdev, bool link) return 0; } +static ssize_t etm_perf_sink_name_show(struct device *dev, + struct device_attribute *dattr, + char *buf) +{ + struct dev_ext_attribute *ea; + + ea = container_of(dattr, struct dev_ext_attribute, attr); + return scnprintf(buf, PAGE_SIZE, "0x%lx\n", (unsigned long)(ea->var)); +} + +int etm_perf_add_symlink_sink(struct coresight_device *csdev) +{ + int ret; + unsigned long hash; + const char *name; + struct device *pmu_dev = etm_pmu.dev; + struct device *pdev = csdev->dev.parent; + struct dev_ext_attribute *ea; + + if (csdev->type != CORESIGHT_DEV_TYPE_SINK && + csdev->type != CORESIGHT_DEV_TYPE_LINKSINK) + return -EINVAL; + + if (csdev->ea != NULL) + return -EINVAL; + + if (!etm_perf_up) + return -EPROBE_DEFER; + + ea = devm_kzalloc(pdev, sizeof(*ea), GFP_KERNEL); + if (!ea) + return -ENOMEM; + + name = dev_name(pdev); + /* See function coresight_get_sink_by_id() to know where this is used */ + hash = hashlen_hash(hashlen_string(NULL, name)); + + ea->attr.attr.name = devm_kstrdup(pdev, name, GFP_KERNEL); + if (!ea->attr.attr.name) + return -ENOMEM; + + ea->attr.attr.mode = 0444; + ea->attr.show = etm_perf_sink_name_show; + ea->var = (unsigned long *)hash; + + ret = sysfs_add_file_to_group(&pmu_dev->kobj, + &ea->attr.attr, "sinks"); + + if (!ret) + csdev->ea = ea; + + return ret; +} + +void etm_perf_del_symlink_sink(struct coresight_device *csdev) +{ + struct device *pmu_dev = etm_pmu.dev; + struct dev_ext_attribute *ea = csdev->ea; + + if (csdev->type != CORESIGHT_DEV_TYPE_SINK && + csdev->type != CORESIGHT_DEV_TYPE_LINKSINK) + return; + + if (!ea) + return; + + sysfs_remove_file_from_group(&pmu_dev->kobj, + &ea->attr.attr, "sinks"); + csdev->ea = NULL; +} + static int __init etm_perf_init(void) { int ret; diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h b/drivers/hwtracing/coresight/coresight-etm-perf.h index da7d9336a15c..015213abe00a 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -59,6 +59,8 @@ struct etm_event_data { #ifdef CONFIG_CORESIGHT int etm_perf_symlink(struct coresight_device *csdev, bool link); +int etm_perf_add_symlink_sink(struct coresight_device *csdev); +void etm_perf_del_symlink_sink(struct coresight_device *csdev); static inline void *etm_perf_sink_config(struct perf_output_handle *handle) { struct etm_event_data *data = perf_get_aux(handle); @@ -70,7 +72,9 @@ static inline void *etm_perf_sink_config(struct perf_output_handle *handle) #else static inline int etm_perf_symlink(struct coresight_device *csdev, bool link) { return -EINVAL; } - +int etm_perf_add_symlink_sink(struct coresight_device *csdev) +{ return -EINVAL; } +void etm_perf_del_symlink_sink(struct coresight_device *csdev) {} static inline void *etm_perf_sink_config(struct perf_output_handle *handle) { return NULL; diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 2b0df1a0a8df..d7fa90be6f42 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -18,6 +18,7 @@ #include #include +#include "coresight-etm-perf.h" #include "coresight-priv.h" static DEFINE_MUTEX(coresight_mutex); @@ -1167,6 +1168,22 @@ struct coresight_device *coresight_register(struct coresight_desc *desc) goto err_out; } + if (csdev->type == CORESIGHT_DEV_TYPE_SINK || + csdev->type == CORESIGHT_DEV_TYPE_LINKSINK) { + ret = etm_perf_add_symlink_sink(csdev); + + if (ret) { + device_unregister(&csdev->dev); + /* + * As with the above, all resources are free'd + * explicitly via coresight_device_release() triggered + * from put_device(), which is in turn called from + * function device_unregister(). + */ + goto err_out; + } + } + mutex_lock(&coresight_mutex); coresight_fixup_device_conns(csdev); @@ -1185,6 +1202,7 @@ EXPORT_SYMBOL_GPL(coresight_register); void coresight_unregister(struct coresight_device *csdev) { + etm_perf_del_symlink_sink(csdev); /* Remove references of that device in the topology */ coresight_remove_conns(csdev); device_unregister(&csdev->dev); diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 46c67a764877..7b87965f7a65 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -154,8 +154,9 @@ struct coresight_connection { * @orphan: true if the component has connections that haven't been linked. * @enable: 'true' if component is currently part of an active path. * @activated: 'true' only if a _sink_ has been activated. A sink can be - activated but not yet enabled. Enabling for a _sink_ - happens when a source has been selected for that it. + * activated but not yet enabled. Enabling for a _sink_ + * appens when a source has been selected for that it. + * @ea: Device attribute for sink representation under PMU directory. */ struct coresight_device { struct coresight_connection *conns; @@ -168,7 +169,9 @@ struct coresight_device { atomic_t *refcnt; bool orphan; bool enable; /* true only if configured as part of a path */ + /* sink specific fields */ bool activated; /* true only if a sink is part of a path */ + struct dev_ext_attribute *ea; }; #define to_coresight_device(d) container_of(d, struct coresight_device, dev) -- cgit v1.3-14-g43fede From c60f83b813e5b25ccd5de7e8c8925c31b3aebcc1 Mon Sep 17 00:00:00 2001 From: Alexander Shishkin Date: Fri, 15 Feb 2019 13:56:55 +0200 Subject: perf, pt, coresight: Fix address filters for vmas with non-zero offset Currently, the address range calculation for file-based filters works as long as the vma that maps the matching part of the object file starts from offset zero into the file (vm_pgoff==0). Otherwise, the resulting filter range would be off by vm_pgoff pages. Another related problem is that in case of a partially matching vma, that is, a vma that matches part of a filter region, the filter range size wouldn't be adjusted. Fix the arithmetics around address filter range calculations, taking into account vma offset, so that the entire calculation is done before the filter configuration is passed to the PMU drivers instead of having those drivers do the final bit of arithmetics. Based on the patch by Adrian Hunter . Reported-by: Adrian Hunter Signed-off-by: Alexander Shishkin Tested-by: Mathieu Poirier Acked-by: Peter Zijlstra Cc: Jiri Olsa Fixes: 375637bc5249 ("perf/core: Introduce address range filtering") Link: http://lkml.kernel.org/r/20190215115655.63469-3-alexander.shishkin@linux.intel.com Signed-off-by: Arnaldo Carvalho de Melo --- arch/x86/events/intel/pt.c | 9 +-- drivers/hwtracing/coresight/coresight-etm-perf.c | 7 +- include/linux/perf_event.h | 7 +- kernel/events/core.c | 81 ++++++++++++++---------- 4 files changed, 62 insertions(+), 42 deletions(-) (limited to 'include') diff --git a/arch/x86/events/intel/pt.c b/arch/x86/events/intel/pt.c index c0e86ff21f81..fb3a2f13fc70 100644 --- a/arch/x86/events/intel/pt.c +++ b/arch/x86/events/intel/pt.c @@ -1223,7 +1223,8 @@ static int pt_event_addr_filters_validate(struct list_head *filters) static void pt_event_addr_filters_sync(struct perf_event *event) { struct perf_addr_filters_head *head = perf_event_addr_filters(event); - unsigned long msr_a, msr_b, *offs = event->addr_filters_offs; + unsigned long msr_a, msr_b; + struct perf_addr_filter_range *fr = event->addr_filter_ranges; struct pt_filters *filters = event->hw.addr_filters; struct perf_addr_filter *filter; int range = 0; @@ -1232,12 +1233,12 @@ static void pt_event_addr_filters_sync(struct perf_event *event) return; list_for_each_entry(filter, &head->list, entry) { - if (filter->path.dentry && !offs[range]) { + if (filter->path.dentry && !fr[range].start) { msr_a = msr_b = 0; } else { /* apply the offset */ - msr_a = filter->offset + offs[range]; - msr_b = filter->size + msr_a - 1; + msr_a = fr[range].start; + msr_b = msr_a + fr[range].size - 1; } filters->filter[range].msr_a = msr_a; diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index 8c88bf0a1e5f..4d5a2b9f9d6a 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -433,15 +433,16 @@ static int etm_addr_filters_validate(struct list_head *filters) static void etm_addr_filters_sync(struct perf_event *event) { struct perf_addr_filters_head *head = perf_event_addr_filters(event); - unsigned long start, stop, *offs = event->addr_filters_offs; + unsigned long start, stop; + struct perf_addr_filter_range *fr = event->addr_filter_ranges; struct etm_filters *filters = event->hw.addr_filters; struct etm_filter *etm_filter; struct perf_addr_filter *filter; int i = 0; list_for_each_entry(filter, &head->list, entry) { - start = filter->offset + offs[i]; - stop = start + filter->size; + start = fr[i].start; + stop = start + fr[i].size; etm_filter = &filters->etm_filter[i]; switch (filter->action) { diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index d9c3610e0e25..6ebc72f65017 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -490,6 +490,11 @@ struct perf_addr_filters_head { unsigned int nr_file_filters; }; +struct perf_addr_filter_range { + unsigned long start; + unsigned long size; +}; + /** * enum perf_event_state - the states of an event: */ @@ -666,7 +671,7 @@ struct perf_event { /* address range filters */ struct perf_addr_filters_head addr_filters; /* vma address array for file-based filders */ - unsigned long *addr_filters_offs; + struct perf_addr_filter_range *addr_filter_ranges; unsigned long addr_filters_gen; void (*destroy)(struct perf_event *); diff --git a/kernel/events/core.c b/kernel/events/core.c index 2d89efc0a3e0..16609f6737da 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -2799,7 +2799,7 @@ static int perf_event_stop(struct perf_event *event, int restart) * * (p1) when userspace mappings change as a result of (1) or (2) or (3) below, * we update the addresses of corresponding vmas in - * event::addr_filters_offs array and bump the event::addr_filters_gen; + * event::addr_filter_ranges array and bump the event::addr_filters_gen; * (p2) when an event is scheduled in (pmu::add), it calls * perf_event_addr_filters_sync() which calls pmu::addr_filters_sync() * if the generation has changed since the previous call. @@ -4446,7 +4446,7 @@ static void _free_event(struct perf_event *event) perf_event_free_bpf_prog(event); perf_addr_filters_splice(event, NULL); - kfree(event->addr_filters_offs); + kfree(event->addr_filter_ranges); if (event->destroy) event->destroy(event); @@ -6687,7 +6687,8 @@ static void perf_event_addr_filters_exec(struct perf_event *event, void *data) raw_spin_lock_irqsave(&ifh->lock, flags); list_for_each_entry(filter, &ifh->list, entry) { if (filter->path.dentry) { - event->addr_filters_offs[count] = 0; + event->addr_filter_ranges[count].start = 0; + event->addr_filter_ranges[count].size = 0; restart++; } @@ -7367,28 +7368,47 @@ static bool perf_addr_filter_match(struct perf_addr_filter *filter, return true; } +static bool perf_addr_filter_vma_adjust(struct perf_addr_filter *filter, + struct vm_area_struct *vma, + struct perf_addr_filter_range *fr) +{ + unsigned long vma_size = vma->vm_end - vma->vm_start; + unsigned long off = vma->vm_pgoff << PAGE_SHIFT; + struct file *file = vma->vm_file; + + if (!perf_addr_filter_match(filter, file, off, vma_size)) + return false; + + if (filter->offset < off) { + fr->start = vma->vm_start; + fr->size = min(vma_size, filter->size - (off - filter->offset)); + } else { + fr->start = vma->vm_start + filter->offset - off; + fr->size = min(vma->vm_end - fr->start, filter->size); + } + + return true; +} + static void __perf_addr_filters_adjust(struct perf_event *event, void *data) { struct perf_addr_filters_head *ifh = perf_event_addr_filters(event); struct vm_area_struct *vma = data; - unsigned long off = vma->vm_pgoff << PAGE_SHIFT, flags; - struct file *file = vma->vm_file; struct perf_addr_filter *filter; unsigned int restart = 0, count = 0; + unsigned long flags; if (!has_addr_filter(event)) return; - if (!file) + if (!vma->vm_file) return; raw_spin_lock_irqsave(&ifh->lock, flags); list_for_each_entry(filter, &ifh->list, entry) { - if (perf_addr_filter_match(filter, file, off, - vma->vm_end - vma->vm_start)) { - event->addr_filters_offs[count] = vma->vm_start; + if (perf_addr_filter_vma_adjust(filter, vma, + &event->addr_filter_ranges[count])) restart++; - } count++; } @@ -8978,26 +8998,19 @@ static void perf_addr_filters_splice(struct perf_event *event, * @filter; if so, adjust filter's address range. * Called with mm::mmap_sem down for reading. */ -static unsigned long perf_addr_filter_apply(struct perf_addr_filter *filter, - struct mm_struct *mm) +static void perf_addr_filter_apply(struct perf_addr_filter *filter, + struct mm_struct *mm, + struct perf_addr_filter_range *fr) { struct vm_area_struct *vma; for (vma = mm->mmap; vma; vma = vma->vm_next) { - struct file *file = vma->vm_file; - unsigned long off = vma->vm_pgoff << PAGE_SHIFT; - unsigned long vma_size = vma->vm_end - vma->vm_start; - - if (!file) + if (!vma->vm_file) continue; - if (!perf_addr_filter_match(filter, file, off, vma_size)) - continue; - - return vma->vm_start; + if (perf_addr_filter_vma_adjust(filter, vma, fr)) + return; } - - return 0; } /* @@ -9031,15 +9044,15 @@ static void perf_event_addr_filters_apply(struct perf_event *event) raw_spin_lock_irqsave(&ifh->lock, flags); list_for_each_entry(filter, &ifh->list, entry) { - event->addr_filters_offs[count] = 0; + event->addr_filter_ranges[count].start = 0; + event->addr_filter_ranges[count].size = 0; /* * Adjust base offset if the filter is associated to a binary * that needs to be mapped: */ if (filter->path.dentry) - event->addr_filters_offs[count] = - perf_addr_filter_apply(filter, mm); + perf_addr_filter_apply(filter, mm, &event->addr_filter_ranges[count]); count++; } @@ -10305,10 +10318,10 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, goto err_pmu; if (has_addr_filter(event)) { - event->addr_filters_offs = kcalloc(pmu->nr_addr_filters, - sizeof(unsigned long), - GFP_KERNEL); - if (!event->addr_filters_offs) { + event->addr_filter_ranges = kcalloc(pmu->nr_addr_filters, + sizeof(struct perf_addr_filter_range), + GFP_KERNEL); + if (!event->addr_filter_ranges) { err = -ENOMEM; goto err_per_task; } @@ -10321,9 +10334,9 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, struct perf_addr_filters_head *ifh = perf_event_addr_filters(event); raw_spin_lock_irq(&ifh->lock); - memcpy(event->addr_filters_offs, - event->parent->addr_filters_offs, - pmu->nr_addr_filters * sizeof(unsigned long)); + memcpy(event->addr_filter_ranges, + event->parent->addr_filter_ranges, + pmu->nr_addr_filters * sizeof(struct perf_addr_filter_range)); raw_spin_unlock_irq(&ifh->lock); } @@ -10345,7 +10358,7 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, return event; err_addr_filters: - kfree(event->addr_filters_offs); + kfree(event->addr_filter_ranges); err_per_task: exclusive_event_destroy(event); -- cgit v1.3-14-g43fede