linux-stable/include/linux/trace_events.h

617 lines
19 KiB
C
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
/* SPDX-License-Identifier: GPL-2.0 */
tracing: Add and use generic set_trigger_filter() implementation Add a generic event_command.set_trigger_filter() op implementation and have the current set of trigger commands use it - this essentially gives them all support for filters. Syntactically, filters are supported by adding 'if <filter>' just after the command, in which case only events matching the filter will invoke the trigger. For example, to add a filter to an enable/disable_event command: echo 'enable_event:system:event if common_pid == 999' > \ .../othersys/otherevent/trigger The above command will only enable the system:event event if the common_pid field in the othersys:otherevent event is 999. As another example, to add a filter to a stacktrace command: echo 'stacktrace if common_pid == 999' > \ .../somesys/someevent/trigger The above command will only trigger a stacktrace if the common_pid field in the event is 999. The filter syntax is the same as that described in the 'Event filtering' section of Documentation/trace/events.txt. Because triggers can now use filters, the trigger-invoking logic needs to be moved in those cases - e.g. for ftrace_raw_event_calls, if a trigger has a filter associated with it, the trigger invocation now needs to happen after the { assign; } part of the call, in order for the trigger condition to be tested. There's still a SOFT_DISABLED-only check at the top of e.g. the ftrace_raw_events function, so when an event is soft disabled but not because of the presence of a trigger, the original SOFT_DISABLED behavior remains unchanged. There's also a bit of trickiness in that some triggers need to avoid being invoked while an event is currently in the process of being logged, since the trigger may itself log data into the trace buffer. Thus we make sure the current event is committed before invoking those triggers. To do that, we split the trigger invocation in two - the first part (event_triggers_call()) checks the filter using the current trace record; if a command has the post_trigger flag set, it sets a bit for itself in the return value, otherwise it directly invoks the trigger. Once all commands have been either invoked or set their return flag, event_triggers_call() returns. The current record is then either committed or discarded; if any commands have deferred their triggers, those commands are finally invoked following the close of the current event by event_triggers_post_call(). To simplify the above and make it more efficient, the TRIGGER_COND bit is introduced, which is set only if a soft-disabled trigger needs to use the log record for filter testing or needs to wait until the current log record is closed. The syscall event invocation code is also changed in analogous ways. Because event triggers need to be able to create and free filters, this also adds a couple external wrappers for the existing create_filter and free_filter functions, which are too generic to be made extern functions themselves. Link: http://lkml.kernel.org/r/7164930759d8719ef460357f143d995406e4eead.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:29 +00:00
#ifndef _LINUX_TRACE_EVENT_H
#define _LINUX_TRACE_EVENT_H
#include <linux/ring_buffer.h>
#include <linux/trace_seq.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
#include <linux/perf_event.h>
#include <linux/tracepoint.h>
struct trace_array;
struct trace_buffer;
struct tracer;
struct dentry;
tracing, perf: Implement BPF programs attached to kprobes BPF programs, attached to kprobes, provide a safe way to execute user-defined BPF byte-code programs without being able to crash or hang the kernel in any way. The BPF engine makes sure that such programs have a finite execution time and that they cannot break out of their sandbox. The user interface is to attach to a kprobe via the perf syscall: struct perf_event_attr attr = { .type = PERF_TYPE_TRACEPOINT, .config = event_id, ... }; event_fd = perf_event_open(&attr,...); ioctl(event_fd, PERF_EVENT_IOC_SET_BPF, prog_fd); 'prog_fd' is a file descriptor associated with BPF program previously loaded. 'event_id' is an ID of the kprobe created. Closing 'event_fd': close(event_fd); ... automatically detaches BPF program from it. BPF programs can call in-kernel helper functions to: - lookup/update/delete elements in maps - probe_read - wraper of probe_kernel_read() used to access any kernel data structures BPF programs receive 'struct pt_regs *' as an input ('struct pt_regs' is architecture dependent) and return 0 to ignore the event and 1 to store kprobe event into the ring buffer. Note, kprobes are a fundamentally _not_ a stable kernel ABI, so BPF programs attached to kprobes must be recompiled for every kernel version and user must supply correct LINUX_VERSION_CODE in attr.kern_version during bpf_prog_load() call. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Reviewed-by: Steven Rostedt <rostedt@goodmis.org> Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Arnaldo Carvalho de Melo <acme@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David S. Miller <davem@davemloft.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1427312966-8434-4-git-send-email-ast@plumgrid.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-25 19:49:20 +00:00
struct bpf_prog;
const char *trace_print_flags_seq(struct trace_seq *p, const char *delim,
unsigned long flags,
const struct trace_print_flags *flag_array);
const char *trace_print_symbols_seq(struct trace_seq *p, unsigned long val,
const struct trace_print_flags *symbol_array);
#if BITS_PER_LONG == 32
const char *trace_print_flags_seq_u64(struct trace_seq *p, const char *delim,
unsigned long long flags,
const struct trace_print_flags_u64 *flag_array);
const char *trace_print_symbols_seq_u64(struct trace_seq *p,
unsigned long long val,
const struct trace_print_flags_u64
*symbol_array);
#endif
const char *trace_print_bitmask_seq(struct trace_seq *p, void *bitmask_ptr,
unsigned int bitmask_size);
tracing: Add __bitmask() macro to trace events to cpumasks and other bitmasks Being able to show a cpumask of events can be useful as some events may affect only some CPUs. There is no standard way to record the cpumask and converting it to a string is rather expensive during the trace as traces happen in hotpaths. It would be better to record the raw event mask and be able to parse it at print time. The following macros were added for use with the TRACE_EVENT() macro: __bitmask() __assign_bitmask() __get_bitmask() To test this, I added this to the sched_migrate_task event, which looked like this: TRACE_EVENT(sched_migrate_task, TP_PROTO(struct task_struct *p, int dest_cpu, const struct cpumask *cpus), TP_ARGS(p, dest_cpu, cpus), TP_STRUCT__entry( __array( char, comm, TASK_COMM_LEN ) __field( pid_t, pid ) __field( int, prio ) __field( int, orig_cpu ) __field( int, dest_cpu ) __bitmask( cpumask, num_possible_cpus() ) ), TP_fast_assign( memcpy(__entry->comm, p->comm, TASK_COMM_LEN); __entry->pid = p->pid; __entry->prio = p->prio; __entry->orig_cpu = task_cpu(p); __entry->dest_cpu = dest_cpu; __assign_bitmask(cpumask, cpumask_bits(cpus), num_possible_cpus()); ), TP_printk("comm=%s pid=%d prio=%d orig_cpu=%d dest_cpu=%d cpumask=%s", __entry->comm, __entry->pid, __entry->prio, __entry->orig_cpu, __entry->dest_cpu, __get_bitmask(cpumask)) ); With the output of: ksmtuned-3613 [003] d..2 485.220508: sched_migrate_task: comm=ksmtuned pid=3615 prio=120 orig_cpu=3 dest_cpu=2 cpumask=00000000,0000000f migration/1-13 [001] d..5 485.221202: sched_migrate_task: comm=ksmtuned pid=3614 prio=120 orig_cpu=1 dest_cpu=0 cpumask=00000000,0000000f awk-3615 [002] d.H5 485.221747: sched_migrate_task: comm=rcu_preempt pid=7 prio=120 orig_cpu=0 dest_cpu=1 cpumask=00000000,000000ff migration/2-18 [002] d..5 485.222062: sched_migrate_task: comm=ksmtuned pid=3615 prio=120 orig_cpu=2 dest_cpu=3 cpumask=00000000,0000000f Link: http://lkml.kernel.org/r/1399377998-14870-6-git-send-email-javi.merino@arm.com Link: http://lkml.kernel.org/r/20140506132238.22e136d1@gandalf.local.home Suggested-by: Javi Merino <javi.merino@arm.com> Tested-by: Javi Merino <javi.merino@arm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2014-05-06 17:10:24 +00:00
const char *trace_print_hex_seq(struct trace_seq *p,
const unsigned char *buf, int len,
bool concatenate);
const char *trace_print_array_seq(struct trace_seq *p,
const void *buf, int count,
size_t el_size);
struct trace_iterator;
struct trace_event;
int trace_raw_output_prep(struct trace_iterator *iter,
struct trace_event *event);
/*
* The trace entry - the most basic unit of tracing. This is what
* is printed in the end as a single line in the trace output, such as:
*
* bash-15816 [01] 235.197585: idle_cpu <- irq_enter
*/
struct trace_entry {
unsigned short type;
unsigned char flags;
unsigned char preempt_count;
int pid;
};
#define TRACE_EVENT_TYPE_MAX \
((1 << (sizeof(((struct trace_entry *)0)->type) * 8)) - 1)
/*
* Trace iterator - used by printout routines who present trace
* results to users and which routines might sleep, etc:
*/
struct trace_iterator {
struct trace_array *tr;
struct tracer *trace;
struct trace_buffer *trace_buffer;
void *private;
int cpu_file;
struct mutex mutex;
struct ring_buffer_iter **buffer_iter;
unsigned long iter_flags;
/* trace_seq for __print_flags() and __print_symbolic() etc. */
struct trace_seq tmp_seq;
cpumask_var_t started;
/* it's true when current open file is snapshot */
bool snapshot;
/* The below is zeroed out in pipe_read */
struct trace_seq seq;
struct trace_entry *ent;
unsigned long lost_events;
tracing: Buffer the output of seq_file in case of filled buffer If the seq_read fills the buffer it will call s_start again on the next itertation with the same position. This causes a problem with the function_graph tracer because it consumes the iteration in order to determine leaf functions. What happens is that the iterator stores the entry, and the function graph plugin will look at the next entry. If that next entry is a return of the same function and task, then the function is a leaf and the function_graph plugin calls ring_buffer_read which moves the ring buffer iterator forward (the trace iterator still points to the function start entry). The copying of the trace_seq to the seq_file buffer will fail if the seq_file buffer is full. The seq_read will not show this entry. The next read by userspace will cause seq_read to again call s_start which will reuse the trace iterator entry (the function start entry). But the function return entry was already consumed. The function graph plugin will think that this entry is a nested function and not a leaf. To solve this, the trace code now checks the return status of the seq_printf (trace_print_seq). If the writing to the seq_file buffer fails, we set a flag in the iterator (leftover) and we do not reset the trace_seq buffer. On the next call to s_start, we check the leftover flag, and if it is set, we just reuse the trace_seq buffer and do not call into the plugin print functions. Before this patch: 2) | fput() { 2) | __fput() { 2) 0.550 us | inotify_inode_queue_event(); 2) | __fsnotify_parent() { 2) 0.540 us | inotify_dentry_parent_queue_event(); After the patch: 2) | fput() { 2) | __fput() { 2) 0.550 us | inotify_inode_queue_event(); 2) 0.548 us | __fsnotify_parent(); 2) 0.540 us | inotify_dentry_parent_queue_event(); [ Updated the patch to fix a missing return 0 from the trace_print_seq() stub when CONFIG_TRACING is disabled. Reported-by: Ingo Molnar <mingo@elte.hu> ] Reported-by: Jiri Olsa <jolsa@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2009-12-07 14:11:39 +00:00
int leftover;
int ent_size;
int cpu;
u64 ts;
loff_t pos;
long idx;
/* All new field here will be zeroed out in pipe_read */
};
tracing: Format non-nanosec times from tsc clock without a decimal point. With the addition of the "tsc" clock, formatting timestamps to look like fractional seconds is misleading. Mark clocks as either in nanoseconds or not, and format non-nanosecond timestamps as decimal integers. Tested: $ cd /sys/kernel/debug/tracing/ $ cat trace_clock [local] global tsc $ echo sched_switch > set_event $ echo 1 > tracing_on ; sleep 0.0005 ; echo 0 > tracing_on $ cat trace <idle>-0 [000] 6330.555552: sched_switch: prev_comm=swapper prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=bash next_pid=29964 next_prio=120 sleep-29964 [000] 6330.555628: sched_switch: prev_comm=bash prev_pid=29964 prev_prio=120 prev_state=S ==> next_comm=swapper next_pid=0 next_prio=120 ... $ echo 1 > options/latency-format $ cat trace <idle>-0 0 4104553247us+: sched_switch: prev_comm=swapper prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=bash next_pid=29964 next_prio=120 sleep-29964 0 4104553322us+: sched_switch: prev_comm=bash prev_pid=29964 prev_prio=120 prev_state=S ==> next_comm=swapper next_pid=0 next_prio=120 ... $ echo tsc > trace_clock $ cat trace $ echo 1 > tracing_on ; sleep 0.0005 ; echo 0 > tracing_on $ echo 0 > options/latency-format $ cat trace <idle>-0 [000] 16490053398357: sched_switch: prev_comm=swapper prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=bash next_pid=31128 next_prio=120 sleep-31128 [000] 16490053588518: sched_switch: prev_comm=bash prev_pid=31128 prev_prio=120 prev_state=S ==> next_comm=swapper next_pid=0 next_prio=120 ... echo 1 > options/latency-format $ cat trace <idle>-0 0 91557653238+: sched_switch: prev_comm=swapper prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=bash next_pid=31128 next_prio=120 sleep-31128 0 91557843399+: sched_switch: prev_comm=bash prev_pid=31128 prev_prio=120 prev_state=S ==> next_comm=swapper next_pid=0 next_prio=120 ... v2: Move arch-specific bits out of generic code. v4: Fix x86_32 build due to 64-bit division. Google-Bug-Id: 6980623 Link: http://lkml.kernel.org/r/1352837903-32191-2-git-send-email-dhsharp@google.com Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Signed-off-by: David Sharp <dhsharp@google.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2012-11-13 20:18:22 +00:00
enum trace_iter_flags {
TRACE_FILE_LAT_FMT = 1,
TRACE_FILE_ANNOTATE = 2,
TRACE_FILE_TIME_IN_NS = 4,
};
typedef enum print_line_t (*trace_print_func)(struct trace_iterator *iter,
int flags, struct trace_event *event);
struct trace_event_functions {
trace_print_func trace;
trace_print_func raw;
trace_print_func hex;
trace_print_func binary;
};
struct trace_event {
struct hlist_node node;
struct list_head list;
int type;
struct trace_event_functions *funcs;
};
extern int register_trace_event(struct trace_event *event);
extern int unregister_trace_event(struct trace_event *event);
/* Return values for print_line callback */
enum print_line_t {
TRACE_TYPE_PARTIAL_LINE = 0, /* Retry after flushing the seq */
TRACE_TYPE_HANDLED = 1,
TRACE_TYPE_UNHANDLED = 2, /* Relay to other output functions */
TRACE_TYPE_NO_CONSUME = 3 /* Handled but ask to not consume */
};
tracing: Move trace_handle_return() out of line Currently trace_handle_return() looks like this: static inline enum print_line_t trace_handle_return(struct trace_seq *s) { return trace_seq_has_overflowed(s) ? TRACE_TYPE_PARTIAL_LINE : TRACE_TYPE_HANDLED; } Where trace_seq_overflowed(s) is: static inline bool trace_seq_has_overflowed(struct trace_seq *s) { return s->full || seq_buf_has_overflowed(&s->seq); } And seq_buf_has_overflowed(&s->seq) is: static inline bool seq_buf_has_overflowed(struct seq_buf *s) { return s->len > s->size; } Making trace_handle_return() into: return (s->full || (s->seq->len > s->seq->size)) ? TRACE_TYPE_PARTIAL_LINE : TRACE_TYPE_HANDLED; One would think this is not an issue to keep as an inline. But because this is used in the TRACE_EVENT() macro, it is extended for every tracepoint in the system. Taking a look at a single tracepoint x86_irq_vector (was the first one I randomly chosen). As trace_handle_return is used in the TRACE_EVENT() macro of trace_raw_output_##call() we disassemble trace_raw_output_x86_irq_vector and do a diff: - is the original + is the out-of-line code I removed identical lines that were different just due to different addresses. --- /tmp/irq-vec-orig 2017-03-16 09:12:48.569384851 -0400 +++ /tmp/irq-vec-ool 2017-03-16 09:13:39.378153385 -0400 @@ -6,27 +6,23 @@ 53 push %rbx 48 89 fb mov %rdi,%rbx 4c 8b a7 c0 20 00 00 mov 0x20c0(%rdi),%r12 e8 f7 72 13 00 callq ffffffff81155c80 <trace_raw_output_prep> 83 f8 01 cmp $0x1,%eax 74 05 je ffffffff8101e993 <trace_raw_output_x86_irq_vector+0x23> 5b pop %rbx 41 5c pop %r12 5d pop %rbp c3 retq 41 8b 54 24 08 mov 0x8(%r12),%edx - 48 8d bb 98 10 00 00 lea 0x1098(%rbx),%rdi + 48 81 c3 98 10 00 00 add $0x1098,%rbx - 48 c7 c6 7b 8a a0 81 mov $0xffffffff81a08a7b,%rsi + 48 c7 c6 ab 8a a0 81 mov $0xffffffff81a08aab,%rsi - e8 c5 85 13 00 callq ffffffff81156f70 <trace_seq_printf> === here's the start of the main difference === + 48 89 df mov %rbx,%rdi + e8 62 7e 13 00 callq ffffffff81156810 <trace_seq_printf> - 8b 93 b8 20 00 00 mov 0x20b8(%rbx),%edx - 31 c0 xor %eax,%eax - 85 d2 test %edx,%edx - 75 11 jne ffffffff8101e9c8 <trace_raw_output_x86_irq_vector+0x58> - 48 8b 83 a8 20 00 00 mov 0x20a8(%rbx),%rax - 48 39 83 a0 20 00 00 cmp %rax,0x20a0(%rbx) - 0f 93 c0 setae %al + 48 89 df mov %rbx,%rdi + e8 4a c5 12 00 callq ffffffff8114af00 <trace_handle_return> 5b pop %rbx - 0f b6 c0 movzbl %al,%eax === end === 41 5c pop %r12 5d pop %rbp c3 retq If you notice, the original has 22 bytes of text more than the out of line version. As this is for every TRACE_EVENT() defined in the system, this can become quite large. text data bss dec hex filename 8690305 5450490 1298432 15439227 eb957b vmlinux-orig 8681725 5450490 1298432 15430647 eb73f7 vmlinux-handle This change has a total of 8580 bytes in savings. $ objdump -dr /tmp/vmlinux-orig | grep '^[0-9a-f]* <trace_raw_output' | wc -l 324 That's 324 tracepoints. But this does not include modules (which contain many more tracepoints). For an allyesconfig build: $ objdump -dr vmlinux-allyes-orig | grep '^[0-9a-f]* <trace_raw_output' | wc -l 1401 That's 1401 tracepoints giving us: text data bss dec hex filename 137920629 140221067 53264384 331406080 13c0db00 vmlinux-allyes-orig 137827709 140221067 53264384 331313160 13bf7008 vmlinux-allyes-handle 92920 bytes in savings!!! Link: http://lkml.kernel.org/r/20170315021431.13107-2-andi@firstfloor.org Reported-by: Andi Kleen <andi@firstfloor.org> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-03-16 15:01:06 +00:00
enum print_line_t trace_handle_return(struct trace_seq *s);
perf_counter: Fix/complete ftrace event records sampling This patch implements the kernel side support for ftrace event record sampling. A new counter sampling attribute is added: PERF_SAMPLE_TP_RECORD which requests ftrace events record sampling. In this case if a PERF_TYPE_TRACEPOINT counter is active and a tracepoint fires, we emit the tracepoint binary record to the perfcounter event buffer, as a sample. Result, after setting PERF_SAMPLE_TP_RECORD attribute from perf record: perf record -f -F 1 -a -e workqueue:workqueue_execution perf report -D 0x21e18 [0x48]: event: 9 . . ... raw event: size 72 bytes . 0000: 09 00 00 00 01 00 48 00 d0 c7 00 81 ff ff ff ff ......H........ . 0010: 0a 00 00 00 0a 00 00 00 21 00 00 00 00 00 00 00 ........!...... . 0020: 2b 00 01 02 0a 00 00 00 0a 00 00 00 65 76 65 6e +...........eve . 0030: 74 73 2f 31 00 00 00 00 00 00 00 00 0a 00 00 00 ts/1........... . 0040: e0 b1 31 81 ff ff ff ff ....... . 0x21e18 [0x48]: PERF_EVENT_SAMPLE (IP, 1): 10: 0xffffffff8100c7d0 period: 33 The raw ftrace binary record starts at offset 0020. Translation: struct trace_entry { type = 0x2b = 43; flags = 1; preempt_count = 2; pid = 0xa = 10; tgid = 0xa = 10; } thread_comm = "events/1" thread_pid = 0xa = 10; func = 0xffffffff8131b1e0 = flush_to_ldisc() What will come next? - Userspace support ('perf trace'), 'flight data recorder' mode for perf trace, etc. - The unconditional copy from the profiling callback brings some costs however if someone wants no such sampling to occur, and needs to be fixed in the future. For that we need to have an instant access to the perf counter attribute. This is a matter of a flag to add in the struct ftrace_event. - Take care of the events recursivity! Don't ever try to record a lock event for example, it seems some locking is used in the profiling fast path and lead to a tracing recursivity. That will be fixed using raw spinlock or recursivity protection. - [...] - Profit! :-) Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Gabriel Munteanu <eduard.munteanu@linux360.ro> Cc: Lai Jiangshan <laijs@cn.fujitsu.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-06 23:25:54 +00:00
void tracing_generic_entry_update(struct trace_entry *entry,
unsigned long flags,
int pc);
struct trace_event_file;
struct ring_buffer_event *
trace_event_buffer_lock_reserve(struct ring_buffer **current_buffer,
struct trace_event_file *trace_file,
int type, unsigned long len,
unsigned long flags, int pc);
#define TRACE_RECORD_CMDLINE BIT(0)
#define TRACE_RECORD_TGID BIT(1)
void tracing_record_taskinfo(struct task_struct *task, int flags);
void tracing_record_taskinfo_sched_switch(struct task_struct *prev,
struct task_struct *next, int flags);
void tracing_record_cmdline(struct task_struct *task);
void tracing_record_tgid(struct task_struct *task);
int trace_output_call(struct trace_iterator *iter, char *name, char *fmt, ...);
struct event_filter;
tracing: Remove per event trace registering This patch removes the register functions of TRACE_EVENT() to enable and disable tracepoints. The registering of a event is now down directly in the trace_events.c file. The tracepoint_probe_register() is now called directly. The prototypes are no longer type checked, but this should not be an issue since the tracepoints are created automatically by the macros. If a prototype is incorrect in the TRACE_EVENT() macro, then other macros will catch it. The trace_event_class structure now holds the probes to be called by the callbacks. This removes needing to have each event have a separate pointer for the probe. To handle kprobes and syscalls, since they register probes in a different manner, a "reg" field is added to the ftrace_event_class structure. If the "reg" field is assigned, then it will be called for enabling and disabling of the probe for either ftrace or perf. To let the reg function know what is happening, a new enum (trace_reg) is created that has the type of control that is needed. With this new rework, the 82 kernel events and 618 syscall events has their footprint dramatically lowered: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4914025 1088868 861512 6864405 68be15 vmlinux.class 4918492 1084612 861512 6864616 68bee8 vmlinux.tracepoint 4900252 1057412 861512 6819176 680d68 vmlinux.regs The size went from 6863829 to 6819176, that's a total of 44K in savings. With tracepoints being continuously added, this is critical that the footprint becomes minimal. v5: Added #ifdef CONFIG_PERF_EVENTS around a reference to perf specific structure in trace_events.c. v4: Fixed trace self tests to check probe because regfunc no longer exists. v3: Updated to handle void *data in beginning of probe parameters. Also added the tracepoint: check_trace_callback_type_##call(). v2: Changed the callback probes to pass void * and typecast the value within the function. Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-21 16:27:06 +00:00
enum trace_reg {
TRACE_REG_REGISTER,
TRACE_REG_UNREGISTER,
#ifdef CONFIG_PERF_EVENTS
tracing: Remove per event trace registering This patch removes the register functions of TRACE_EVENT() to enable and disable tracepoints. The registering of a event is now down directly in the trace_events.c file. The tracepoint_probe_register() is now called directly. The prototypes are no longer type checked, but this should not be an issue since the tracepoints are created automatically by the macros. If a prototype is incorrect in the TRACE_EVENT() macro, then other macros will catch it. The trace_event_class structure now holds the probes to be called by the callbacks. This removes needing to have each event have a separate pointer for the probe. To handle kprobes and syscalls, since they register probes in a different manner, a "reg" field is added to the ftrace_event_class structure. If the "reg" field is assigned, then it will be called for enabling and disabling of the probe for either ftrace or perf. To let the reg function know what is happening, a new enum (trace_reg) is created that has the type of control that is needed. With this new rework, the 82 kernel events and 618 syscall events has their footprint dramatically lowered: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4914025 1088868 861512 6864405 68be15 vmlinux.class 4918492 1084612 861512 6864616 68bee8 vmlinux.tracepoint 4900252 1057412 861512 6819176 680d68 vmlinux.regs The size went from 6863829 to 6819176, that's a total of 44K in savings. With tracepoints being continuously added, this is critical that the footprint becomes minimal. v5: Added #ifdef CONFIG_PERF_EVENTS around a reference to perf specific structure in trace_events.c. v4: Fixed trace self tests to check probe because regfunc no longer exists. v3: Updated to handle void *data in beginning of probe parameters. Also added the tracepoint: check_trace_callback_type_##call(). v2: Changed the callback probes to pass void * and typecast the value within the function. Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-21 16:27:06 +00:00
TRACE_REG_PERF_REGISTER,
TRACE_REG_PERF_UNREGISTER,
TRACE_REG_PERF_OPEN,
TRACE_REG_PERF_CLOSE,
/*
* These (ADD/DEL) use a 'boolean' return value, where 1 (true) means a
* custom action was taken and the default action is not to be
* performed.
*/
TRACE_REG_PERF_ADD,
TRACE_REG_PERF_DEL,
#endif
tracing: Remove per event trace registering This patch removes the register functions of TRACE_EVENT() to enable and disable tracepoints. The registering of a event is now down directly in the trace_events.c file. The tracepoint_probe_register() is now called directly. The prototypes are no longer type checked, but this should not be an issue since the tracepoints are created automatically by the macros. If a prototype is incorrect in the TRACE_EVENT() macro, then other macros will catch it. The trace_event_class structure now holds the probes to be called by the callbacks. This removes needing to have each event have a separate pointer for the probe. To handle kprobes and syscalls, since they register probes in a different manner, a "reg" field is added to the ftrace_event_class structure. If the "reg" field is assigned, then it will be called for enabling and disabling of the probe for either ftrace or perf. To let the reg function know what is happening, a new enum (trace_reg) is created that has the type of control that is needed. With this new rework, the 82 kernel events and 618 syscall events has their footprint dramatically lowered: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4914025 1088868 861512 6864405 68be15 vmlinux.class 4918492 1084612 861512 6864616 68bee8 vmlinux.tracepoint 4900252 1057412 861512 6819176 680d68 vmlinux.regs The size went from 6863829 to 6819176, that's a total of 44K in savings. With tracepoints being continuously added, this is critical that the footprint becomes minimal. v5: Added #ifdef CONFIG_PERF_EVENTS around a reference to perf specific structure in trace_events.c. v4: Fixed trace self tests to check probe because regfunc no longer exists. v3: Updated to handle void *data in beginning of probe parameters. Also added the tracepoint: check_trace_callback_type_##call(). v2: Changed the callback probes to pass void * and typecast the value within the function. Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-21 16:27:06 +00:00
};
struct trace_event_call;
tracing: Remove per event trace registering This patch removes the register functions of TRACE_EVENT() to enable and disable tracepoints. The registering of a event is now down directly in the trace_events.c file. The tracepoint_probe_register() is now called directly. The prototypes are no longer type checked, but this should not be an issue since the tracepoints are created automatically by the macros. If a prototype is incorrect in the TRACE_EVENT() macro, then other macros will catch it. The trace_event_class structure now holds the probes to be called by the callbacks. This removes needing to have each event have a separate pointer for the probe. To handle kprobes and syscalls, since they register probes in a different manner, a "reg" field is added to the ftrace_event_class structure. If the "reg" field is assigned, then it will be called for enabling and disabling of the probe for either ftrace or perf. To let the reg function know what is happening, a new enum (trace_reg) is created that has the type of control that is needed. With this new rework, the 82 kernel events and 618 syscall events has their footprint dramatically lowered: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4914025 1088868 861512 6864405 68be15 vmlinux.class 4918492 1084612 861512 6864616 68bee8 vmlinux.tracepoint 4900252 1057412 861512 6819176 680d68 vmlinux.regs The size went from 6863829 to 6819176, that's a total of 44K in savings. With tracepoints being continuously added, this is critical that the footprint becomes minimal. v5: Added #ifdef CONFIG_PERF_EVENTS around a reference to perf specific structure in trace_events.c. v4: Fixed trace self tests to check probe because regfunc no longer exists. v3: Updated to handle void *data in beginning of probe parameters. Also added the tracepoint: check_trace_callback_type_##call(). v2: Changed the callback probes to pass void * and typecast the value within the function. Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-21 16:27:06 +00:00
struct trace_event_class {
const char *system;
tracing: Remove per event trace registering This patch removes the register functions of TRACE_EVENT() to enable and disable tracepoints. The registering of a event is now down directly in the trace_events.c file. The tracepoint_probe_register() is now called directly. The prototypes are no longer type checked, but this should not be an issue since the tracepoints are created automatically by the macros. If a prototype is incorrect in the TRACE_EVENT() macro, then other macros will catch it. The trace_event_class structure now holds the probes to be called by the callbacks. This removes needing to have each event have a separate pointer for the probe. To handle kprobes and syscalls, since they register probes in a different manner, a "reg" field is added to the ftrace_event_class structure. If the "reg" field is assigned, then it will be called for enabling and disabling of the probe for either ftrace or perf. To let the reg function know what is happening, a new enum (trace_reg) is created that has the type of control that is needed. With this new rework, the 82 kernel events and 618 syscall events has their footprint dramatically lowered: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4914025 1088868 861512 6864405 68be15 vmlinux.class 4918492 1084612 861512 6864616 68bee8 vmlinux.tracepoint 4900252 1057412 861512 6819176 680d68 vmlinux.regs The size went from 6863829 to 6819176, that's a total of 44K in savings. With tracepoints being continuously added, this is critical that the footprint becomes minimal. v5: Added #ifdef CONFIG_PERF_EVENTS around a reference to perf specific structure in trace_events.c. v4: Fixed trace self tests to check probe because regfunc no longer exists. v3: Updated to handle void *data in beginning of probe parameters. Also added the tracepoint: check_trace_callback_type_##call(). v2: Changed the callback probes to pass void * and typecast the value within the function. Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-21 16:27:06 +00:00
void *probe;
#ifdef CONFIG_PERF_EVENTS
void *perf_probe;
#endif
int (*reg)(struct trace_event_call *event,
enum trace_reg type, void *data);
int (*define_fields)(struct trace_event_call *);
struct list_head *(*get_fields)(struct trace_event_call *);
tracing: Move fields from event to class structure Move the defined fields from the event to the class structure. Since the fields of the event are defined by the class they belong to, it makes sense to have the class hold the information instead of the individual events. The events of the same class would just hold duplicate information. After this change the size of the kernel dropped another 3K: text data bss dec hex filename 4913961 1088356 861512 6863829 68bbd5 vmlinux.orig 4900252 1057412 861512 6819176 680d68 vmlinux.regs 4900375 1053380 861512 6815267 67fe23 vmlinux.fields Although the text increased, this was mainly due to the C files having to adapt to the change. This is a constant increase, where new tracepoints will not increase the Text. But the big drop is in the data size (as well as needed allocations to hold the fields). This will give even more savings as more tracepoints are created. Note, if just TRACE_EVENT()s are used and not DECLARE_EVENT_CLASS() with several DEFINE_EVENT()s, then the savings will be lost. But we are pushing developers to consolidate events with DEFINE_EVENT() so this should not be an issue. The kprobes define a unique class to every new event, but are dynamic so it should not be a issue. The syscalls however have a single class but the fields for the individual events are different. The syscalls use a metadata to define the fields. I moved the fields list from the event to the metadata and added a "get_fields()" function to the class. This function is used to find the fields. For normal events and kprobes, get_fields() just returns a pointer to the fields list_head in the class. For syscall events, it returns the fields list_head in the metadata for the event. v2: Fixed the syscall fields. The syscall metadata needs a list of fields for both enter and exit. Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Masami Hiramatsu <mhiramat@redhat.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-22 14:35:55 +00:00
struct list_head fields;
int (*raw_init)(struct trace_event_call *);
};
extern int trace_event_reg(struct trace_event_call *event,
enum trace_reg type, void *data);
struct trace_event_buffer {
struct ring_buffer *buffer;
struct ring_buffer_event *event;
struct trace_event_file *trace_file;
void *entry;
unsigned long flags;
int pc;
};
void *trace_event_buffer_reserve(struct trace_event_buffer *fbuffer,
struct trace_event_file *trace_file,
unsigned long len);
void trace_event_buffer_commit(struct trace_event_buffer *fbuffer);
enum {
TRACE_EVENT_FL_FILTERED_BIT,
TRACE_EVENT_FL_CAP_ANY_BIT,
TRACE_EVENT_FL_NO_SET_FILTER_BIT,
TRACE_EVENT_FL_IGNORE_ENABLE_BIT,
TRACE_EVENT_FL_TRACEPOINT_BIT,
TRACE_EVENT_FL_KPROBE_BIT,
TRACE_EVENT_FL_UPROBE_BIT,
};
/*
* Event flags:
* FILTERED - The event has a filter attached
* CAP_ANY - Any user can enable for perf
* NO_SET_FILTER - Set when filter has error and is to be ignored
* IGNORE_ENABLE - For trace internal events, do not enable with debugfs file
* TRACEPOINT - Event is a tracepoint
* KPROBE - Event is a kprobe
* UPROBE - Event is a uprobe
*/
enum {
TRACE_EVENT_FL_FILTERED = (1 << TRACE_EVENT_FL_FILTERED_BIT),
TRACE_EVENT_FL_CAP_ANY = (1 << TRACE_EVENT_FL_CAP_ANY_BIT),
TRACE_EVENT_FL_NO_SET_FILTER = (1 << TRACE_EVENT_FL_NO_SET_FILTER_BIT),
TRACE_EVENT_FL_IGNORE_ENABLE = (1 << TRACE_EVENT_FL_IGNORE_ENABLE_BIT),
TRACE_EVENT_FL_TRACEPOINT = (1 << TRACE_EVENT_FL_TRACEPOINT_BIT),
TRACE_EVENT_FL_KPROBE = (1 << TRACE_EVENT_FL_KPROBE_BIT),
TRACE_EVENT_FL_UPROBE = (1 << TRACE_EVENT_FL_UPROBE_BIT),
};
#define TRACE_EVENT_FL_UKPROBE (TRACE_EVENT_FL_KPROBE | TRACE_EVENT_FL_UPROBE)
struct trace_event_call {
struct list_head list;
struct trace_event_class *class;
union {
char *name;
/* Set TRACE_EVENT_FL_TRACEPOINT flag when using "tp" */
struct tracepoint *tp;
};
struct trace_event event;
tracing: Add TRACE_DEFINE_ENUM() macro to map enums to their values Several tracepoints use the helper functions __print_symbolic() or __print_flags() and pass in enums that do the mapping between the binary data stored and the value to print. This works well for reading the ASCII trace files, but when the data is read via userspace tools such as perf and trace-cmd, the conversion of the binary value to a human string format is lost if an enum is used, as userspace does not have access to what the ENUM is. For example, the tracepoint trace_tlb_flush() has: __print_symbolic(REC->reason, { TLB_FLUSH_ON_TASK_SWITCH, "flush on task switch" }, { TLB_REMOTE_SHOOTDOWN, "remote shootdown" }, { TLB_LOCAL_SHOOTDOWN, "local shootdown" }, { TLB_LOCAL_MM_SHOOTDOWN, "local mm shootdown" }) Which maps the enum values to the strings they represent. But perf and trace-cmd do no know what value TLB_LOCAL_MM_SHOOTDOWN is, and would not be able to map it. With TRACE_DEFINE_ENUM(), developers can place these in the event header files and ftrace will convert the enums to their values: By adding: TRACE_DEFINE_ENUM(TLB_FLUSH_ON_TASK_SWITCH); TRACE_DEFINE_ENUM(TLB_REMOTE_SHOOTDOWN); TRACE_DEFINE_ENUM(TLB_LOCAL_SHOOTDOWN); TRACE_DEFINE_ENUM(TLB_LOCAL_MM_SHOOTDOWN); $ cat /sys/kernel/debug/tracing/events/tlb/tlb_flush/format [...] __print_symbolic(REC->reason, { 0, "flush on task switch" }, { 1, "remote shootdown" }, { 2, "local shootdown" }, { 3, "local mm shootdown" }) The above is what userspace expects to see, and tools do not need to be modified to parse them. Link: http://lkml.kernel.org/r/20150403013802.220157513@goodmis.org Cc: Guilherme Cox <cox@computer.org> Cc: Tony Luck <tony.luck@gmail.com> Cc: Xie XiuQi <xiexiuqi@huawei.com> Acked-by: Namhyung Kim <namhyung@kernel.org> Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Tested-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2015-03-24 21:58:09 +00:00
char *print_fmt;
struct event_filter *filter;
void *mod;
void *data;
/*
* bit 0: filter_active
* bit 1: allow trace by non root (cap any)
* bit 2: failed to apply filter
* bit 3: trace internal event (do not enable)
* bit 4: Event was enabled by module
tracing: Update event filters for multibuffer The trace event filters are still tied to event calls rather than event files, which means you don't get what you'd expect when using filters in the multibuffer case: Before: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 2048 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 Setting the filter in tracing/instances/test1/events shouldn't affect the same event in tracing/events as it does above. After: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 We'd like to just move the filter directly from ftrace_event_call to ftrace_event_file, but there are a couple cases that don't yet have multibuffer support and therefore have to continue using the current event_call-based filters. For those cases, a new USE_CALL_FILTER bit is added to the event_call flags, whose main purpose is to keep the old behavior for those cases until they can be updated with multibuffer support; at that point, the USE_CALL_FILTER flag (and the new associated call_filter_check_discard() function) can go away. The multibuffer support also made filter_current_check_discard() redundant, so this change removes that function as well and replaces it with filter_check_discard() (or call_filter_check_discard() as appropriate). Link: http://lkml.kernel.org/r/f16e9ce4270c62f46b2e966119225e1c3cca7e60.1382620672.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:34:17 +00:00
* bit 5: use call filter rather than file filter
* bit 6: Event is a tracepoint
*/
int flags; /* static flags of different events */
#ifdef CONFIG_PERF_EVENTS
int perf_refcount;
struct hlist_head __percpu *perf_events;
struct bpf_prog_array __rcu *prog_array;
int (*perf_perm)(struct trace_event_call *,
struct perf_event *);
#endif
};
#ifdef CONFIG_PERF_EVENTS
static inline bool bpf_prog_array_valid(struct trace_event_call *call)
{
/*
* This inline function checks whether call->prog_array
* is valid or not. The function is called in various places,
* outside rcu_read_lock/unlock, as a heuristic to speed up execution.
*
* If this function returns true, and later call->prog_array
* becomes false inside rcu_read_lock/unlock region,
* we bail out then. If this function return false,
* there is a risk that we might miss a few events if the checking
* were delayed until inside rcu_read_lock/unlock region and
* call->prog_array happened to become non-NULL then.
*
* Here, READ_ONCE() is used instead of rcu_access_pointer().
* rcu_access_pointer() requires the actual definition of
* "struct bpf_prog_array" while READ_ONCE() only needs
* a declaration of the same type.
*/
return !!READ_ONCE(call->prog_array);
}
#endif
static inline const char *
trace_event_name(struct trace_event_call *call)
{
if (call->flags & TRACE_EVENT_FL_TRACEPOINT)
return call->tp ? call->tp->name : NULL;
else
return call->name;
}
struct trace_array;
struct trace_subsystem_dir;
enum {
EVENT_FILE_FL_ENABLED_BIT,
EVENT_FILE_FL_RECORDED_CMD_BIT,
EVENT_FILE_FL_RECORDED_TGID_BIT,
EVENT_FILE_FL_FILTERED_BIT,
EVENT_FILE_FL_NO_SET_FILTER_BIT,
EVENT_FILE_FL_SOFT_MODE_BIT,
EVENT_FILE_FL_SOFT_DISABLED_BIT,
EVENT_FILE_FL_TRIGGER_MODE_BIT,
EVENT_FILE_FL_TRIGGER_COND_BIT,
EVENT_FILE_FL_PID_FILTER_BIT,
EVENT_FILE_FL_WAS_ENABLED_BIT,
};
/*
* Event file flags:
* ENABLED - The event is enabled
* RECORDED_CMD - The comms should be recorded at sched_switch
* RECORDED_TGID - The tgids should be recorded at sched_switch
tracing: Update event filters for multibuffer The trace event filters are still tied to event calls rather than event files, which means you don't get what you'd expect when using filters in the multibuffer case: Before: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 2048 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 Setting the filter in tracing/instances/test1/events shouldn't affect the same event in tracing/events as it does above. After: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 We'd like to just move the filter directly from ftrace_event_call to ftrace_event_file, but there are a couple cases that don't yet have multibuffer support and therefore have to continue using the current event_call-based filters. For those cases, a new USE_CALL_FILTER bit is added to the event_call flags, whose main purpose is to keep the old behavior for those cases until they can be updated with multibuffer support; at that point, the USE_CALL_FILTER flag (and the new associated call_filter_check_discard() function) can go away. The multibuffer support also made filter_current_check_discard() redundant, so this change removes that function as well and replaces it with filter_check_discard() (or call_filter_check_discard() as appropriate). Link: http://lkml.kernel.org/r/f16e9ce4270c62f46b2e966119225e1c3cca7e60.1382620672.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:34:17 +00:00
* FILTERED - The event has a filter attached
* NO_SET_FILTER - Set when filter has error and is to be ignored
* SOFT_MODE - The event is enabled/disabled by SOFT_DISABLED
* SOFT_DISABLED - When set, do not trace the event (even though its
* tracepoint may be enabled)
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
* TRIGGER_MODE - When set, invoke the triggers associated with the event
tracing: Add and use generic set_trigger_filter() implementation Add a generic event_command.set_trigger_filter() op implementation and have the current set of trigger commands use it - this essentially gives them all support for filters. Syntactically, filters are supported by adding 'if <filter>' just after the command, in which case only events matching the filter will invoke the trigger. For example, to add a filter to an enable/disable_event command: echo 'enable_event:system:event if common_pid == 999' > \ .../othersys/otherevent/trigger The above command will only enable the system:event event if the common_pid field in the othersys:otherevent event is 999. As another example, to add a filter to a stacktrace command: echo 'stacktrace if common_pid == 999' > \ .../somesys/someevent/trigger The above command will only trigger a stacktrace if the common_pid field in the event is 999. The filter syntax is the same as that described in the 'Event filtering' section of Documentation/trace/events.txt. Because triggers can now use filters, the trigger-invoking logic needs to be moved in those cases - e.g. for ftrace_raw_event_calls, if a trigger has a filter associated with it, the trigger invocation now needs to happen after the { assign; } part of the call, in order for the trigger condition to be tested. There's still a SOFT_DISABLED-only check at the top of e.g. the ftrace_raw_events function, so when an event is soft disabled but not because of the presence of a trigger, the original SOFT_DISABLED behavior remains unchanged. There's also a bit of trickiness in that some triggers need to avoid being invoked while an event is currently in the process of being logged, since the trigger may itself log data into the trace buffer. Thus we make sure the current event is committed before invoking those triggers. To do that, we split the trigger invocation in two - the first part (event_triggers_call()) checks the filter using the current trace record; if a command has the post_trigger flag set, it sets a bit for itself in the return value, otherwise it directly invoks the trigger. Once all commands have been either invoked or set their return flag, event_triggers_call() returns. The current record is then either committed or discarded; if any commands have deferred their triggers, those commands are finally invoked following the close of the current event by event_triggers_post_call(). To simplify the above and make it more efficient, the TRIGGER_COND bit is introduced, which is set only if a soft-disabled trigger needs to use the log record for filter testing or needs to wait until the current log record is closed. The syscall event invocation code is also changed in analogous ways. Because event triggers need to be able to create and free filters, this also adds a couple external wrappers for the existing create_filter and free_filter functions, which are too generic to be made extern functions themselves. Link: http://lkml.kernel.org/r/7164930759d8719ef460357f143d995406e4eead.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:29 +00:00
* TRIGGER_COND - When set, one or more triggers has an associated filter
* PID_FILTER - When set, the event is filtered based on pid
* WAS_ENABLED - Set when enabled to know to clear trace on module removal
*/
enum {
EVENT_FILE_FL_ENABLED = (1 << EVENT_FILE_FL_ENABLED_BIT),
EVENT_FILE_FL_RECORDED_CMD = (1 << EVENT_FILE_FL_RECORDED_CMD_BIT),
EVENT_FILE_FL_RECORDED_TGID = (1 << EVENT_FILE_FL_RECORDED_TGID_BIT),
EVENT_FILE_FL_FILTERED = (1 << EVENT_FILE_FL_FILTERED_BIT),
EVENT_FILE_FL_NO_SET_FILTER = (1 << EVENT_FILE_FL_NO_SET_FILTER_BIT),
EVENT_FILE_FL_SOFT_MODE = (1 << EVENT_FILE_FL_SOFT_MODE_BIT),
EVENT_FILE_FL_SOFT_DISABLED = (1 << EVENT_FILE_FL_SOFT_DISABLED_BIT),
EVENT_FILE_FL_TRIGGER_MODE = (1 << EVENT_FILE_FL_TRIGGER_MODE_BIT),
EVENT_FILE_FL_TRIGGER_COND = (1 << EVENT_FILE_FL_TRIGGER_COND_BIT),
EVENT_FILE_FL_PID_FILTER = (1 << EVENT_FILE_FL_PID_FILTER_BIT),
EVENT_FILE_FL_WAS_ENABLED = (1 << EVENT_FILE_FL_WAS_ENABLED_BIT),
};
struct trace_event_file {
struct list_head list;
struct trace_event_call *event_call;
struct event_filter __rcu *filter;
struct dentry *dir;
struct trace_array *tr;
struct trace_subsystem_dir *system;
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
struct list_head triggers;
/*
* 32 bit flags:
* bit 0: enabled
* bit 1: enabled cmd record
* bit 2: enable/disable with the soft disable bit
* bit 3: soft disabled
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
* bit 4: trigger enabled
*
* Note: The bits must be set atomically to prevent races
* from other writers. Reads of flags do not need to be in
* sync as they occur in critical sections. But the way flags
* is currently used, these changes do not affect the code
tracing: Comment the use of event_mutex with trace event flags The flags variable is protected by the event_mutex when modifying, but the event_mutex is not held when reading the variable. This is due to the fact that the reads occur in critical sections where taking a mutex (or even a spinlock) is not wanted. But the two flags that exist (enable and filter_active) have the code written as such to handle the reads to not need a lock. The enable flag is used just to know if the event is enabled or not and its use is always under the event_mutex. Whether or not the event is actually enabled is really determined by the tracepoint being registered. The flag is just a way to let the code know if the tracepoint is registered. The filter_active is different. It is read without the lock. If it is set, then the event probes jump to the filter code. There can be a slight mismatch between filters available and filter_active. If the flag is set but no filters are available, the code safely jumps to a filter nop. If the flag is not set and the filters are available, then the filters are skipped. This is acceptable since filters are usually set before tracing or they are set by humans, which would not notice the slight delay that this causes. v2: Fixed typo: "cacheing" -> "caching" Reported-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Cc: Tom Zanussi <tzanussi@gmail.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-05-14 14:19:13 +00:00
* except that when a change is made, it may have a slight
* delay in propagating the changes to other CPUs due to
* caching and such. Which is mostly OK ;-)
*/
unsigned long flags;
atomic_t sm_ref; /* soft-mode reference counter */
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
atomic_t tm_ref; /* trigger-mode reference counter */
};
#define __TRACE_EVENT_FLAGS(name, value) \
static int __init trace_init_flags_##name(void) \
{ \
event_##name.flags |= value; \
return 0; \
} \
early_initcall(trace_init_flags_##name);
#define __TRACE_EVENT_PERF_PERM(name, expr...) \
static int perf_perm_##name(struct trace_event_call *tp_event, \
struct perf_event *p_event) \
{ \
return ({ expr; }); \
} \
static int __init trace_init_perf_perm_##name(void) \
{ \
event_##name.perf_perm = &perf_perm_##name; \
return 0; \
} \
early_initcall(trace_init_perf_perm_##name);
#define PERF_MAX_TRACE_SIZE 2048
#define MAX_FILTER_STR_VAL 256 /* Should handle KSYM_SYMBOL_LEN */
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
enum event_trigger_type {
ETT_NONE = (0),
tracing: Add 'traceon' and 'traceoff' event trigger commands Add 'traceon' and 'traceoff' event_command commands. traceon and traceoff event triggers are added by the user via these commands in a similar way and using practically the same syntax as the analagous 'traceon' and 'traceoff' ftrace function commands, but instead of writing to the set_ftrace_filter file, the traceon and traceoff triggers are written to the per-event 'trigger' files: echo 'traceon' > .../tracing/events/somesys/someevent/trigger echo 'traceoff' > .../tracing/events/somesys/someevent/trigger The above command will turn tracing on or off whenever someevent is hit. This also adds a 'count' version that limits the number of times the command will be invoked: echo 'traceon:N' > .../tracing/events/somesys/someevent/trigger echo 'traceoff:N' > .../tracing/events/somesys/someevent/trigger Where N is the number of times the command will be invoked. The above commands will will turn tracing on or off whenever someevent is hit, but only N times. Some common register/unregister_trigger() implementations of the event_command reg()/unreg() callbacks are also provided, which add and remove trigger instances to the per-event list of triggers, and arm/disarm them as appropriate. event_trigger_callback() is a general-purpose event_command func() implementation that orchestrates command parsing and registration for most normal commands. Most event commands will use these, but some will override and possibly reuse them. The event_trigger_init(), event_trigger_free(), and event_trigger_print() functions are meant to be common implementations of the event_trigger_ops init(), free(), and print() ops, respectively. Most trigger_ops implementations will use these, but some will override and possibly reuse them. Link: http://lkml.kernel.org/r/00a52816703b98d2072947478dd6e2d70cde5197.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:25 +00:00
ETT_TRACE_ONOFF = (1 << 0),
tracing: Add 'snapshot' event trigger command Add 'snapshot' event_command. snapshot event triggers are added by the user via this command in a similar way and using practically the same syntax as the analogous 'snapshot' ftrace function command, but instead of writing to the set_ftrace_filter file, the snapshot event trigger is written to the per-event 'trigger' files: echo 'snapshot' > .../somesys/someevent/trigger The above command will turn on snapshots for someevent i.e. whenever someevent is hit, a snapshot will be done. This also adds a 'count' version that limits the number of times the command will be invoked: echo 'snapshot:N' > .../somesys/someevent/trigger Where N is the number of times the command will be invoked. The above command will snapshot N times for someevent i.e. whenever someevent is hit N times, a snapshot will be done. Also adds a new tracing_alloc_snapshot() function - the existing tracing_snapshot_alloc() function is a special version of tracing_snapshot() that also does the snapshot allocation - the snapshot triggers would like to be able to do just the allocation but not take a snapshot; the existing tracing_snapshot_alloc() in turn now also calls tracing_alloc_snapshot() underneath to do that allocation. Link: http://lkml.kernel.org/r/c9524dd07ce01f9dcbd59011290e0a8d5b47d7ad.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> [ fix up from kbuild test robot <fengguang.wu@intel.com report ] Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:26 +00:00
ETT_SNAPSHOT = (1 << 1),
ETT_STACKTRACE = (1 << 2),
tracing: Add 'enable_event' and 'disable_event' event trigger commands Add 'enable_event' and 'disable_event' event_command commands. enable_event and disable_event event triggers are added by the user via these commands in a similar way and using practically the same syntax as the analagous 'enable_event' and 'disable_event' ftrace function commands, but instead of writing to the set_ftrace_filter file, the enable_event and disable_event triggers are written to the per-event 'trigger' files: echo 'enable_event:system:event' > .../othersys/otherevent/trigger echo 'disable_event:system:event' > .../othersys/otherevent/trigger The above commands will enable or disable the 'system:event' trace events whenever the othersys:otherevent events are hit. This also adds a 'count' version that limits the number of times the command will be invoked: echo 'enable_event:system:event:N' > .../othersys/otherevent/trigger echo 'disable_event:system:event:N' > .../othersys/otherevent/trigger Where N is the number of times the command will be invoked. The above commands will will enable or disable the 'system:event' trace events whenever the othersys:otherevent events are hit, but only N times. This also makes the find_event_file() helper function extern, since it's useful to use from other places, such as the event triggers code, so make it accessible. Link: http://lkml.kernel.org/r/f825f3048c3f6b026ee37ae5825f9fc373451828.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:28 +00:00
ETT_EVENT_ENABLE = (1 << 3),
tracing: Add 'hist' event trigger command 'hist' triggers allow users to continually aggregate trace events, which can then be viewed afterwards by simply reading a 'hist' file containing the aggregation in a human-readable format. The basic idea is very simple and boils down to a mechanism whereby trace events, rather than being exhaustively dumped in raw form and viewed directly, are automatically 'compressed' into meaningful tables completely defined by the user. This is done strictly via single-line command-line commands and without the aid of any kind of programming language or interpreter. A surprising number of typical use cases can be accomplished by users via this simple mechanism. In fact, a large number of the tasks that users typically do using the more complicated script-based tracing tools, at least during the initial stages of an investigation, can be accomplished by simply specifying a set of keys and values to be used in the creation of a hash table. The Linux kernel trace event subsystem happens to provide an extensive list of keys and values ready-made for such a purpose in the form of the event format files associated with each trace event. By simply consulting the format file for field names of interest and by plugging them into the hist trigger command, users can create an endless number of useful aggregations to help with investigating various properties of the system. See Documentation/trace/events.txt for examples. hist triggers are implemented on top of the existing event trigger infrastructure, and as such are consistent with the existing triggers from a user's perspective as well. The basic syntax follows the existing trigger syntax. Users start an aggregation by writing a 'hist' trigger to the event of interest's trigger file: # echo hist:keys=xxx [ if filter] > event/trigger Once a hist trigger has been set up, by default it continually aggregates every matching event into a hash table using the event key and a value field named 'hitcount'. To view the aggregation at any point in time, simply read the 'hist' file in the same directory as the 'trigger' file: # cat event/hist The detailed syntax provides additional options for user control, and is described exhaustively in Documentation/trace/events.txt and in the virtual tracing/README file in the tracing subsystem. Link: http://lkml.kernel.org/r/72d263b5e1853fe9c314953b65833c3aa75479f2.1457029949.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Tested-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2016-03-03 18:54:42 +00:00
ETT_EVENT_HIST = (1 << 4),
ETT_HIST_ENABLE = (1 << 5),
tracing: Add basic event trigger framework Add a 'trigger' file for each trace event, enabling 'trace event triggers' to be set for trace events. 'trace event triggers' are patterned after the existing 'ftrace function triggers' implementation except that triggers are written to per-event 'trigger' files instead of to a single file such as the 'set_ftrace_filter' used for ftrace function triggers. The implementation is meant to be entirely separate from ftrace function triggers, in order to keep the respective implementations relatively simple and to allow them to diverge. The event trigger functionality is built on top of SOFT_DISABLE functionality. It adds a TRIGGER_MODE bit to the ftrace_event_file flags which is checked when any trace event fires. Triggers set for a particular event need to be checked regardless of whether that event is actually enabled or not - getting an event to fire even if it's not enabled is what's already implemented by SOFT_DISABLE mode, so trigger mode directly reuses that. Event trigger essentially inherit the soft disable logic in __ftrace_event_enable_disable() while adding a bit of logic and trigger reference counting via tm_ref on top of that in a new trace_event_trigger_enable_disable() function. Because the base __ftrace_event_enable_disable() code now needs to be invoked from outside trace_events.c, a wrapper is also added for those usages. The triggers for an event are actually invoked via a new function, event_triggers_call(), and code is also added to invoke them for ftrace_raw_event calls as well as syscall events. The main part of the patch creates a new trace_events_trigger.c file to contain the trace event triggers implementation. The standard open, read, and release file operations are implemented here. The open() implementation sets up for the various open modes of the 'trigger' file. It creates and attaches the trigger iterator and sets up the command parser. If opened for reading set up the trigger seq_ops. The read() implementation parses the event trigger written to the 'trigger' file, looks up the trigger command, and passes it along to that event_command's func() implementation for command-specific processing. The release() implementation does whatever cleanup is needed to release the 'trigger' file, like releasing the parser and trigger iterator, etc. A couple of functions for event command registration and unregistration are added, along with a list to add them to and a mutex to protect them, as well as an (initially empty) registration function to add the set of commands that will be added by future commits, and call to it from the trace event initialization code. also added are a couple trigger-specific data structures needed for these implementations such as a trigger iterator and a struct for trigger-specific data. A couple structs consisting mostly of function meant to be implemented in command-specific ways, event_command and event_trigger_ops, are used by the generic event trigger command implementations. They're being put into trace.h alongside the other trace_event data structures and functions, in the expectation that they'll be needed in several trace_event-related files such as trace_events_trigger.c and trace_events.c. The event_command.func() function is meant to be called by the trigger parsing code in order to add a trigger instance to the corresponding event. It essentially coordinates adding a live trigger instance to the event, and arming the triggering the event. Every event_command func() implementation essentially does the same thing for any command: - choose ops - use the value of param to choose either a number or count version of event_trigger_ops specific to the command - do the register or unregister of those ops - associate a filter, if specified, with the triggering event The reg() and unreg() ops allow command-specific implementations for event_trigger_op registration and unregistration, and the get_trigger_ops() op allows command-specific event_trigger_ops selection to be parameterized. When a trigger instance is added, the reg() op essentially adds that trigger to the triggering event and arms it, while unreg() does the opposite. The set_filter() function is used to associate a filter with the trigger - if the command doesn't specify a set_filter() implementation, the command will ignore filters. Each command has an associated trigger_type, which serves double duty, both as a unique identifier for the command as well as a value that can be used for setting a trigger mode bit during trigger invocation. The signature of func() adds a pointer to the event_command struct, used to invoke those functions, along with a command_data param that can be passed to the reg/unreg functions. This allows func() implementations to use command-specific blobs and supports code re-use. The event_trigger_ops.func() command corrsponds to the trigger 'probe' function that gets called when the triggering event is actually invoked. The other functions are used to list the trigger when needed, along with a couple mundane book-keeping functions. This also moves event_file_data() into trace.h so it can be used outside of trace_events.c. Link: http://lkml.kernel.org/r/316d95061accdee070aac8e5750afba0192fa5b9.1382622043.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Idea-by: Steve Rostedt <rostedt@goodmis.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:59:24 +00:00
};
extern int filter_match_preds(struct event_filter *filter, void *rec);
tracing: Update event filters for multibuffer The trace event filters are still tied to event calls rather than event files, which means you don't get what you'd expect when using filters in the multibuffer case: Before: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 2048 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 Setting the filter in tracing/instances/test1/events shouldn't affect the same event in tracing/events as it does above. After: # echo 'bytes_alloc > 8192' > /sys/kernel/debug/tracing/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # mkdir /sys/kernel/debug/tracing/instances/test1 # echo 'bytes_alloc > 2048' > /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter # cat /sys/kernel/debug/tracing/events/kmem/kmalloc/filter bytes_alloc > 8192 # cat /sys/kernel/debug/tracing/instances/test1/events/kmem/kmalloc/filter bytes_alloc > 2048 We'd like to just move the filter directly from ftrace_event_call to ftrace_event_file, but there are a couple cases that don't yet have multibuffer support and therefore have to continue using the current event_call-based filters. For those cases, a new USE_CALL_FILTER bit is added to the event_call flags, whose main purpose is to keep the old behavior for those cases until they can be updated with multibuffer support; at that point, the USE_CALL_FILTER flag (and the new associated call_filter_check_discard() function) can go away. The multibuffer support also made filter_current_check_discard() redundant, so this change removes that function as well and replaces it with filter_check_discard() (or call_filter_check_discard() as appropriate). Link: http://lkml.kernel.org/r/f16e9ce4270c62f46b2e966119225e1c3cca7e60.1382620672.git.tom.zanussi@linux.intel.com Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2013-10-24 13:34:17 +00:00
extern enum event_trigger_type
event_triggers_call(struct trace_event_file *file, void *rec,
struct ring_buffer_event *event);
extern void
event_triggers_post_call(struct trace_event_file *file,
enum event_trigger_type tt,
void *rec, struct ring_buffer_event *event);
bool trace_event_ignore_this_pid(struct trace_event_file *trace_file);
/**
* trace_trigger_soft_disabled - do triggers and test if soft disabled
* @file: The file pointer of the event to test
*
* If any triggers without filters are attached to this event, they
* will be called here. If the event is soft disabled and has no
* triggers that require testing the fields, it will return true,
* otherwise false.
*/
static inline bool
trace_trigger_soft_disabled(struct trace_event_file *file)
{
unsigned long eflags = file->flags;
if (!(eflags & EVENT_FILE_FL_TRIGGER_COND)) {
if (eflags & EVENT_FILE_FL_TRIGGER_MODE)
event_triggers_call(file, NULL, NULL);
if (eflags & EVENT_FILE_FL_SOFT_DISABLED)
return true;
if (eflags & EVENT_FILE_FL_PID_FILTER)
return trace_event_ignore_this_pid(file);
}
return false;
}
bpf: Use correct #ifdef controller for trace_call_bpf() Commit e1abf2cc8d5d80b41c4419368ec743ccadbb131e ("bpf: Fix the build on BPF_SYSCALL=y && !CONFIG_TRACING kernels, make it more configurable") updated the building condition of bpf_trace.o from CONFIG_BPF_SYSCALL to CONFIG_BPF_EVENTS, but the corresponding #ifdef controller in trace_events.h for trace_call_bpf() was not changed. Which, in theory, is incorrect. With current Kconfigs, we can create a .config with CONFIG_BPF_SYSCALL=y and CONFIG_BPF_EVENTS=n by unselecting CONFIG_KPROBE_EVENT and selecting CONFIG_BPF_SYSCALL. With these options, trace_call_bpf() will be defined as an extern function, but if anyone calls it a symbol missing error will be triggered since bpf_trace.o was not built. This patch changes the #ifdef controller for trace_call_bpf() from CONFIG_BPF_SYSCALL to CONFIG_BPF_EVENTS. I'll show its correctness: Before this patch: BPF_SYSCALL BPF_EVENTS trace_call_bpf bpf_trace.o y y normal compiled n n inline not compiled y n normal not compiled (incorrect) n y impossible (BPF_EVENTS depends on BPF_SYSCALL) After this patch: BPF_SYSCALL BPF_EVENTS trace_call_bpf bpf_trace.o y y normal compiled n n inline not compiled y n inline not compiled (fixed) n y impossible (BPF_EVENTS depends on BPF_SYSCALL) So this patch doesn't break anything. QED. Signed-off-by: Wang Nan <wangnan0@huawei.com> Cc: Alexei Starovoitov <ast@plumgrid.com> Cc: Brendan Gregg <brendan.d.gregg@gmail.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David Ahern <dsahern@gmail.com> Cc: He Kuang <hekuang@huawei.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kaixu Xia <xiakaixu@huawei.com> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Zefan Li <lizefan@huawei.com> Cc: pi3orama@163.com Link: http://lkml.kernel.org/r/1435716878-189507-2-git-send-email-wangnan0@huawei.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-07-01 02:13:49 +00:00
#ifdef CONFIG_BPF_EVENTS
unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx);
int perf_event_attach_bpf_prog(struct perf_event *event, struct bpf_prog *prog);
void perf_event_detach_bpf_prog(struct perf_event *event);
int perf_event_query_prog_array(struct perf_event *event, void __user *info);
bpf: introduce BPF_RAW_TRACEPOINT Introduce BPF_PROG_TYPE_RAW_TRACEPOINT bpf program type to access kernel internal arguments of the tracepoints in their raw form. >From bpf program point of view the access to the arguments look like: struct bpf_raw_tracepoint_args { __u64 args[0]; }; int bpf_prog(struct bpf_raw_tracepoint_args *ctx) { // program can read args[N] where N depends on tracepoint // and statically verified at program load+attach time } kprobe+bpf infrastructure allows programs access function arguments. This feature allows programs access raw tracepoint arguments. Similar to proposed 'dynamic ftrace events' there are no abi guarantees to what the tracepoints arguments are and what their meaning is. The program needs to type cast args properly and use bpf_probe_read() helper to access struct fields when argument is a pointer. For every tracepoint __bpf_trace_##call function is prepared. In assembler it looks like: (gdb) disassemble __bpf_trace_xdp_exception Dump of assembler code for function __bpf_trace_xdp_exception: 0xffffffff81132080 <+0>: mov %ecx,%ecx 0xffffffff81132082 <+2>: jmpq 0xffffffff811231f0 <bpf_trace_run3> where TRACE_EVENT(xdp_exception, TP_PROTO(const struct net_device *dev, const struct bpf_prog *xdp, u32 act), The above assembler snippet is casting 32-bit 'act' field into 'u64' to pass into bpf_trace_run3(), while 'dev' and 'xdp' args are passed as-is. All of ~500 of __bpf_trace_*() functions are only 5-10 byte long and in total this approach adds 7k bytes to .text. This approach gives the lowest possible overhead while calling trace_xdp_exception() from kernel C code and transitioning into bpf land. Since tracepoint+bpf are used at speeds of 1M+ events per second this is valuable optimization. The new BPF_RAW_TRACEPOINT_OPEN sys_bpf command is introduced that returns anon_inode FD of 'bpf-raw-tracepoint' object. The user space looks like: // load bpf prog with BPF_PROG_TYPE_RAW_TRACEPOINT type prog_fd = bpf_prog_load(...); // receive anon_inode fd for given bpf_raw_tracepoint with prog attached raw_tp_fd = bpf_raw_tracepoint_open("xdp_exception", prog_fd); Ctrl-C of tracing daemon or cmdline tool that uses this feature will automatically detach bpf program, unload it and unregister tracepoint probe. On the kernel side the __bpf_raw_tp_map section of pointers to tracepoint definition and to __bpf_trace_*() probe function is used to find a tracepoint with "xdp_exception" name and corresponding __bpf_trace_xdp_exception() probe function which are passed to tracepoint_probe_register() to connect probe with tracepoint. Addition of bpf_raw_tracepoint doesn't interfere with ftrace and perf tracepoint mechanisms. perf_event_open() can be used in parallel on the same tracepoint. Multiple bpf_raw_tracepoint_open("xdp_exception", prog_fd) are permitted. Each with its own bpf program. The kernel will execute all tracepoint probes and all attached bpf programs. In the future bpf_raw_tracepoints can be extended with query/introspection logic. __bpf_raw_tp_map section logic was contributed by Steven Rostedt Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-03-28 19:05:37 +00:00
int bpf_probe_register(struct bpf_raw_event_map *btp, struct bpf_prog *prog);
int bpf_probe_unregister(struct bpf_raw_event_map *btp, struct bpf_prog *prog);
struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name);
tracing, perf: Implement BPF programs attached to kprobes BPF programs, attached to kprobes, provide a safe way to execute user-defined BPF byte-code programs without being able to crash or hang the kernel in any way. The BPF engine makes sure that such programs have a finite execution time and that they cannot break out of their sandbox. The user interface is to attach to a kprobe via the perf syscall: struct perf_event_attr attr = { .type = PERF_TYPE_TRACEPOINT, .config = event_id, ... }; event_fd = perf_event_open(&attr,...); ioctl(event_fd, PERF_EVENT_IOC_SET_BPF, prog_fd); 'prog_fd' is a file descriptor associated with BPF program previously loaded. 'event_id' is an ID of the kprobe created. Closing 'event_fd': close(event_fd); ... automatically detaches BPF program from it. BPF programs can call in-kernel helper functions to: - lookup/update/delete elements in maps - probe_read - wraper of probe_kernel_read() used to access any kernel data structures BPF programs receive 'struct pt_regs *' as an input ('struct pt_regs' is architecture dependent) and return 0 to ignore the event and 1 to store kprobe event into the ring buffer. Note, kprobes are a fundamentally _not_ a stable kernel ABI, so BPF programs attached to kprobes must be recompiled for every kernel version and user must supply correct LINUX_VERSION_CODE in attr.kern_version during bpf_prog_load() call. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Reviewed-by: Steven Rostedt <rostedt@goodmis.org> Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Arnaldo Carvalho de Melo <acme@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David S. Miller <davem@davemloft.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1427312966-8434-4-git-send-email-ast@plumgrid.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-25 19:49:20 +00:00
#else
static inline unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx)
tracing, perf: Implement BPF programs attached to kprobes BPF programs, attached to kprobes, provide a safe way to execute user-defined BPF byte-code programs without being able to crash or hang the kernel in any way. The BPF engine makes sure that such programs have a finite execution time and that they cannot break out of their sandbox. The user interface is to attach to a kprobe via the perf syscall: struct perf_event_attr attr = { .type = PERF_TYPE_TRACEPOINT, .config = event_id, ... }; event_fd = perf_event_open(&attr,...); ioctl(event_fd, PERF_EVENT_IOC_SET_BPF, prog_fd); 'prog_fd' is a file descriptor associated with BPF program previously loaded. 'event_id' is an ID of the kprobe created. Closing 'event_fd': close(event_fd); ... automatically detaches BPF program from it. BPF programs can call in-kernel helper functions to: - lookup/update/delete elements in maps - probe_read - wraper of probe_kernel_read() used to access any kernel data structures BPF programs receive 'struct pt_regs *' as an input ('struct pt_regs' is architecture dependent) and return 0 to ignore the event and 1 to store kprobe event into the ring buffer. Note, kprobes are a fundamentally _not_ a stable kernel ABI, so BPF programs attached to kprobes must be recompiled for every kernel version and user must supply correct LINUX_VERSION_CODE in attr.kern_version during bpf_prog_load() call. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Reviewed-by: Steven Rostedt <rostedt@goodmis.org> Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Arnaldo Carvalho de Melo <acme@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David S. Miller <davem@davemloft.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1427312966-8434-4-git-send-email-ast@plumgrid.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-25 19:49:20 +00:00
{
return 1;
}
static inline int
perf_event_attach_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
{
return -EOPNOTSUPP;
}
static inline void perf_event_detach_bpf_prog(struct perf_event *event) { }
static inline int
perf_event_query_prog_array(struct perf_event *event, void __user *info)
{
return -EOPNOTSUPP;
}
bpf: introduce BPF_RAW_TRACEPOINT Introduce BPF_PROG_TYPE_RAW_TRACEPOINT bpf program type to access kernel internal arguments of the tracepoints in their raw form. >From bpf program point of view the access to the arguments look like: struct bpf_raw_tracepoint_args { __u64 args[0]; }; int bpf_prog(struct bpf_raw_tracepoint_args *ctx) { // program can read args[N] where N depends on tracepoint // and statically verified at program load+attach time } kprobe+bpf infrastructure allows programs access function arguments. This feature allows programs access raw tracepoint arguments. Similar to proposed 'dynamic ftrace events' there are no abi guarantees to what the tracepoints arguments are and what their meaning is. The program needs to type cast args properly and use bpf_probe_read() helper to access struct fields when argument is a pointer. For every tracepoint __bpf_trace_##call function is prepared. In assembler it looks like: (gdb) disassemble __bpf_trace_xdp_exception Dump of assembler code for function __bpf_trace_xdp_exception: 0xffffffff81132080 <+0>: mov %ecx,%ecx 0xffffffff81132082 <+2>: jmpq 0xffffffff811231f0 <bpf_trace_run3> where TRACE_EVENT(xdp_exception, TP_PROTO(const struct net_device *dev, const struct bpf_prog *xdp, u32 act), The above assembler snippet is casting 32-bit 'act' field into 'u64' to pass into bpf_trace_run3(), while 'dev' and 'xdp' args are passed as-is. All of ~500 of __bpf_trace_*() functions are only 5-10 byte long and in total this approach adds 7k bytes to .text. This approach gives the lowest possible overhead while calling trace_xdp_exception() from kernel C code and transitioning into bpf land. Since tracepoint+bpf are used at speeds of 1M+ events per second this is valuable optimization. The new BPF_RAW_TRACEPOINT_OPEN sys_bpf command is introduced that returns anon_inode FD of 'bpf-raw-tracepoint' object. The user space looks like: // load bpf prog with BPF_PROG_TYPE_RAW_TRACEPOINT type prog_fd = bpf_prog_load(...); // receive anon_inode fd for given bpf_raw_tracepoint with prog attached raw_tp_fd = bpf_raw_tracepoint_open("xdp_exception", prog_fd); Ctrl-C of tracing daemon or cmdline tool that uses this feature will automatically detach bpf program, unload it and unregister tracepoint probe. On the kernel side the __bpf_raw_tp_map section of pointers to tracepoint definition and to __bpf_trace_*() probe function is used to find a tracepoint with "xdp_exception" name and corresponding __bpf_trace_xdp_exception() probe function which are passed to tracepoint_probe_register() to connect probe with tracepoint. Addition of bpf_raw_tracepoint doesn't interfere with ftrace and perf tracepoint mechanisms. perf_event_open() can be used in parallel on the same tracepoint. Multiple bpf_raw_tracepoint_open("xdp_exception", prog_fd) are permitted. Each with its own bpf program. The kernel will execute all tracepoint probes and all attached bpf programs. In the future bpf_raw_tracepoints can be extended with query/introspection logic. __bpf_raw_tp_map section logic was contributed by Steven Rostedt Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-03-28 19:05:37 +00:00
static inline int bpf_probe_register(struct bpf_raw_event_map *btp, struct bpf_prog *p)
{
return -EOPNOTSUPP;
}
static inline int bpf_probe_unregister(struct bpf_raw_event_map *btp, struct bpf_prog *p)
{
return -EOPNOTSUPP;
}
static inline struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name)
{
return NULL;
}
tracing, perf: Implement BPF programs attached to kprobes BPF programs, attached to kprobes, provide a safe way to execute user-defined BPF byte-code programs without being able to crash or hang the kernel in any way. The BPF engine makes sure that such programs have a finite execution time and that they cannot break out of their sandbox. The user interface is to attach to a kprobe via the perf syscall: struct perf_event_attr attr = { .type = PERF_TYPE_TRACEPOINT, .config = event_id, ... }; event_fd = perf_event_open(&attr,...); ioctl(event_fd, PERF_EVENT_IOC_SET_BPF, prog_fd); 'prog_fd' is a file descriptor associated with BPF program previously loaded. 'event_id' is an ID of the kprobe created. Closing 'event_fd': close(event_fd); ... automatically detaches BPF program from it. BPF programs can call in-kernel helper functions to: - lookup/update/delete elements in maps - probe_read - wraper of probe_kernel_read() used to access any kernel data structures BPF programs receive 'struct pt_regs *' as an input ('struct pt_regs' is architecture dependent) and return 0 to ignore the event and 1 to store kprobe event into the ring buffer. Note, kprobes are a fundamentally _not_ a stable kernel ABI, so BPF programs attached to kprobes must be recompiled for every kernel version and user must supply correct LINUX_VERSION_CODE in attr.kern_version during bpf_prog_load() call. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Reviewed-by: Steven Rostedt <rostedt@goodmis.org> Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Arnaldo Carvalho de Melo <acme@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David S. Miller <davem@davemloft.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1427312966-8434-4-git-send-email-ast@plumgrid.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-25 19:49:20 +00:00
#endif
enum {
FILTER_OTHER = 0,
FILTER_STATIC_STRING,
FILTER_DYN_STRING,
FILTER_PTR_STRING,
FILTER_TRACE_FN,
FILTER_COMM,
FILTER_CPU,
};
extern int trace_event_raw_init(struct trace_event_call *call);
extern int trace_define_field(struct trace_event_call *call, const char *type,
const char *name, int offset, int size,
int is_signed, int filter_type);
extern int trace_add_event_call(struct trace_event_call *call);
extern int trace_remove_event_call(struct trace_event_call *call);
extern int trace_event_get_offsets(struct trace_event_call *call);
#define is_signed_type(type) (((type)(-1)) < (type)1)
int trace_set_clr_event(const char *system, const char *event, int set);
/*
* The double __builtin_constant_p is because gcc will give us an error
* if we try to allocate the static variable to fmt if it is not a
* constant. Even with the outer if statement optimizing out.
*/
#define event_trace_printk(ip, fmt, args...) \
do { \
__trace_printk_check_format(fmt, ##args); \
tracing_record_cmdline(current); \
if (__builtin_constant_p(fmt)) { \
static const char *trace_printk_fmt \
__attribute__((section("__trace_printk_fmt"))) = \
__builtin_constant_p(fmt) ? fmt : NULL; \
\
__trace_bprintk(ip, trace_printk_fmt, ##args); \
} else \
__trace_printk(ip, fmt, ##args); \
} while (0)
#ifdef CONFIG_PERF_EVENTS
struct perf_event;
perf: Take a hot regs snapshot for trace events We are taking a wrong regs snapshot when a trace event triggers. Either we use get_irq_regs(), which gives us the interrupted registers if we are in an interrupt, or we use task_pt_regs() which gives us the state before we entered the kernel, assuming we are lucky enough to be no kernel thread, in which case task_pt_regs() returns the initial set of regs when the kernel thread was started. What we want is different. We need a hot snapshot of the regs, so that we can get the instruction pointer to record in the sample, the frame pointer for the callchain, and some other things. Let's use the new perf_fetch_caller_regs() for that. Comparison with perf record -e lock: -R -a -f -g Before: perf [kernel] [k] __do_softirq | --- __do_softirq | |--55.16%-- __open | --44.84%-- __write_nocancel After: perf [kernel] [k] perf_tp_event | --- perf_tp_event | |--41.07%-- lock_acquire | | | |--39.36%-- _raw_spin_lock | | | | | |--7.81%-- hrtimer_interrupt | | | smp_apic_timer_interrupt | | | apic_timer_interrupt The old case was producing unreliable callchains. Now having right frame and instruction pointers, we have the trace we want. Also syscalls and kprobe events already have the right regs, let's use them instead of wasting a retrieval. v2: Follow the rename perf_save_regs() -> perf_fetch_caller_regs() Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Archs <linux-arch@vger.kernel.org>
2010-03-03 06:16:16 +00:00
DECLARE_PER_CPU(struct pt_regs, perf_trace_regs);
DECLARE_PER_CPU(int, bpf_kprobe_override);
perf: Take a hot regs snapshot for trace events We are taking a wrong regs snapshot when a trace event triggers. Either we use get_irq_regs(), which gives us the interrupted registers if we are in an interrupt, or we use task_pt_regs() which gives us the state before we entered the kernel, assuming we are lucky enough to be no kernel thread, in which case task_pt_regs() returns the initial set of regs when the kernel thread was started. What we want is different. We need a hot snapshot of the regs, so that we can get the instruction pointer to record in the sample, the frame pointer for the callchain, and some other things. Let's use the new perf_fetch_caller_regs() for that. Comparison with perf record -e lock: -R -a -f -g Before: perf [kernel] [k] __do_softirq | --- __do_softirq | |--55.16%-- __open | --44.84%-- __write_nocancel After: perf [kernel] [k] perf_tp_event | --- perf_tp_event | |--41.07%-- lock_acquire | | | |--39.36%-- _raw_spin_lock | | | | | |--7.81%-- hrtimer_interrupt | | | smp_apic_timer_interrupt | | | apic_timer_interrupt The old case was producing unreliable callchains. Now having right frame and instruction pointers, we have the trace we want. Also syscalls and kprobe events already have the right regs, let's use them instead of wasting a retrieval. v2: Follow the rename perf_save_regs() -> perf_fetch_caller_regs() Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Archs <linux-arch@vger.kernel.org>
2010-03-03 06:16:16 +00:00
extern int perf_trace_init(struct perf_event *event);
extern void perf_trace_destroy(struct perf_event *event);
2010-06-16 12:37:10 +00:00
extern int perf_trace_add(struct perf_event *event, int flags);
extern void perf_trace_del(struct perf_event *event, int flags);
#ifdef CONFIG_KPROBE_EVENTS
extern int perf_kprobe_init(struct perf_event *event, bool is_retprobe);
extern void perf_kprobe_destroy(struct perf_event *event);
#endif
#ifdef CONFIG_UPROBE_EVENTS
extern int perf_uprobe_init(struct perf_event *event, bool is_retprobe);
extern void perf_uprobe_destroy(struct perf_event *event);
#endif
extern int ftrace_profile_set_filter(struct perf_event *event, int event_id,
char *filter_str);
extern void ftrace_profile_free_filter(struct perf_event *event);
void perf_trace_buf_update(void *record, u16 type);
void *perf_trace_buf_alloc(int size, struct pt_regs **regs, int *rctxp);
bpf: introduce BPF_RAW_TRACEPOINT Introduce BPF_PROG_TYPE_RAW_TRACEPOINT bpf program type to access kernel internal arguments of the tracepoints in their raw form. >From bpf program point of view the access to the arguments look like: struct bpf_raw_tracepoint_args { __u64 args[0]; }; int bpf_prog(struct bpf_raw_tracepoint_args *ctx) { // program can read args[N] where N depends on tracepoint // and statically verified at program load+attach time } kprobe+bpf infrastructure allows programs access function arguments. This feature allows programs access raw tracepoint arguments. Similar to proposed 'dynamic ftrace events' there are no abi guarantees to what the tracepoints arguments are and what their meaning is. The program needs to type cast args properly and use bpf_probe_read() helper to access struct fields when argument is a pointer. For every tracepoint __bpf_trace_##call function is prepared. In assembler it looks like: (gdb) disassemble __bpf_trace_xdp_exception Dump of assembler code for function __bpf_trace_xdp_exception: 0xffffffff81132080 <+0>: mov %ecx,%ecx 0xffffffff81132082 <+2>: jmpq 0xffffffff811231f0 <bpf_trace_run3> where TRACE_EVENT(xdp_exception, TP_PROTO(const struct net_device *dev, const struct bpf_prog *xdp, u32 act), The above assembler snippet is casting 32-bit 'act' field into 'u64' to pass into bpf_trace_run3(), while 'dev' and 'xdp' args are passed as-is. All of ~500 of __bpf_trace_*() functions are only 5-10 byte long and in total this approach adds 7k bytes to .text. This approach gives the lowest possible overhead while calling trace_xdp_exception() from kernel C code and transitioning into bpf land. Since tracepoint+bpf are used at speeds of 1M+ events per second this is valuable optimization. The new BPF_RAW_TRACEPOINT_OPEN sys_bpf command is introduced that returns anon_inode FD of 'bpf-raw-tracepoint' object. The user space looks like: // load bpf prog with BPF_PROG_TYPE_RAW_TRACEPOINT type prog_fd = bpf_prog_load(...); // receive anon_inode fd for given bpf_raw_tracepoint with prog attached raw_tp_fd = bpf_raw_tracepoint_open("xdp_exception", prog_fd); Ctrl-C of tracing daemon or cmdline tool that uses this feature will automatically detach bpf program, unload it and unregister tracepoint probe. On the kernel side the __bpf_raw_tp_map section of pointers to tracepoint definition and to __bpf_trace_*() probe function is used to find a tracepoint with "xdp_exception" name and corresponding __bpf_trace_xdp_exception() probe function which are passed to tracepoint_probe_register() to connect probe with tracepoint. Addition of bpf_raw_tracepoint doesn't interfere with ftrace and perf tracepoint mechanisms. perf_event_open() can be used in parallel on the same tracepoint. Multiple bpf_raw_tracepoint_open("xdp_exception", prog_fd) are permitted. Each with its own bpf program. The kernel will execute all tracepoint probes and all attached bpf programs. In the future bpf_raw_tracepoints can be extended with query/introspection logic. __bpf_raw_tp_map section logic was contributed by Steven Rostedt Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-03-28 19:05:37 +00:00
void bpf_trace_run1(struct bpf_prog *prog, u64 arg1);
void bpf_trace_run2(struct bpf_prog *prog, u64 arg1, u64 arg2);
void bpf_trace_run3(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3);
void bpf_trace_run4(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4);
void bpf_trace_run5(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5);
void bpf_trace_run6(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6);
void bpf_trace_run7(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7);
void bpf_trace_run8(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7,
u64 arg8);
void bpf_trace_run9(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7,
u64 arg8, u64 arg9);
void bpf_trace_run10(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7,
u64 arg8, u64 arg9, u64 arg10);
void bpf_trace_run11(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7,
u64 arg8, u64 arg9, u64 arg10, u64 arg11);
void bpf_trace_run12(struct bpf_prog *prog, u64 arg1, u64 arg2,
u64 arg3, u64 arg4, u64 arg5, u64 arg6, u64 arg7,
u64 arg8, u64 arg9, u64 arg10, u64 arg11, u64 arg12);
void perf_trace_run_bpf_submit(void *raw_data, int size, int rctx,
struct trace_event_call *call, u64 count,
struct pt_regs *regs, struct hlist_head *head,
struct task_struct *task);
static inline void
perf_trace_buf_submit(void *raw_data, int size, int rctx, u16 type,
u64 count, struct pt_regs *regs, void *head,
struct task_struct *task)
{
perf_tp_event(type, count, raw_data, size, regs, head, rctx, task);
}
#endif
#endif /* _LINUX_TRACE_EVENT_H */