Message-Id: <1396017924-7754-14-git-send-email-jovi.zhangwei@gmail.com>
Date:	Fri, 28 Mar 2014 10:45:08 -0400
From:	Jovi Zhangwei <jovi.zhangwei@...il.com>
To:	Ingo Molnar <mingo@...hat.com>,
	Steven Rostedt <rostedt@...dmis.org>
Cc:	linux-kernel@...r.kernel.org,
	Masami Hiramatsu <masami.hiramatsu.pt@...achi.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Frederic Weisbecker <fweisbec@...il.com>,
	Andi Kleen <andi@...stfloor.org>,
	Jovi Zhangwei <jovi.zhangwei@...il.com>
Subject: [PATCH v2 13/29] ktap: add ring buffer handling code(kernel/trace/ktap/kp_transport.[c|h])

ktap transport functionality is based on ftrace ring buffer.

All contents generated by ktap runtime will send to userspace
through this ring buffer.

Actually the concept of ktap transport is similar with ftrace,
for example, it will write event binary data into ring buffer
when use call: 'print(argstr)' or 'print(stack())', and the content
will be parse in consumer side, not producer side.

Userspace reader thread will consume content in ring buffer through
'/sys/kernel/debug/ktap/trace_pipe_$pid'

A lot of code in this file is duplicated with kernel trace_output.c.

Exposed functions:
1). kp_transport_init
2). kp_transport_exit
3). __kp_bputs/__kp_puts/kp_printf
4). kp_transport_write
5). kp_transport_event_write
        write a event content into ring buffer, called by: 'print(argstr)'
6). kp_transport_print_kstack
        write a kernel stack into ring buffer, called by: 'print(stack())'

Signed-off-by: Jovi Zhangwei <jovi.zhangwei@...il.com>
---
 kernel/trace/ktap/kp_transport.c | 649 +++++++++++++++++++++++++++++++++++++++
 kernel/trace/ktap/kp_transport.h |  13 +
 2 files changed, 662 insertions(+)
 create mode 100644 kernel/trace/ktap/kp_transport.c
 create mode 100644 kernel/trace/ktap/kp_transport.h

diff --git a/kernel/trace/ktap/kp_transport.c b/kernel/trace/ktap/kp_transport.c
new file mode 100644
index 0000000..5baf586
--- /dev/null
+++ b/kernel/trace/ktap/kp_transport.c
@@ -0,0 +1,649 @@
+/*
+ * kp_transport.c - ktap transport functionality
+ *
+ * This file is part of ktap by Jovi Zhangwei.
+ *
+ * Copyright (C) 2012-2014 Jovi Zhangwei <jovi.zhangwei@...il.com>.
+ *
+ * ktap is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * ktap is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#include <linux/debugfs.h>
+#include <linux/ftrace_event.h>
+#include <linux/stacktrace.h>
+#include <linux/clocksource.h>
+#include <asm/uaccess.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/kallsyms.h>
+#include <uapi/ktap/ktap_types.h>
+#include "ktap.h"
+#include "kp_events.h"
+#include "kp_transport.h"
+
+struct ktap_trace_iterator {
+	struct ring_buffer	*buffer;
+	int			print_timestamp;
+	void			*private;
+
+	struct trace_iterator	iter;
+};
+
+enum ktap_trace_type {
+	__TRACE_FIRST_TYPE = 0,
+
+	TRACE_FN = 1, /* must be same as ftrace definition in kernel */
+	TRACE_PRINT,
+	TRACE_BPUTS,
+	TRACE_STACK,
+	TRACE_USER_STACK,
+
+	__TRACE_LAST_TYPE,
+};
+
+#define KTAP_TRACE_ITER(iter)	\
+	container_of(iter, struct ktap_trace_iterator, iter)
+
+static
+ssize_t _trace_seq_to_user(struct trace_seq *s, char __user *ubuf, size_t cnt)
+{
+	int len;
+	int ret;
+
+	if (!cnt)
+		return 0;
+
+	if (s->len <= s->readpos)
+		return -EBUSY;
+
+	len = s->len - s->readpos;
+	if (cnt > len)
+		cnt = len;
+	ret = copy_to_user(ubuf, s->buffer + s->readpos, cnt);
+	if (ret == cnt)
+		return -EFAULT;
+
+	cnt -= ret;
+
+	s->readpos += cnt;
+	return cnt;
+}
+
+int _trace_seq_puts(struct trace_seq *s, const char *str)
+{
+	int len = strlen(str);
+
+	if (s->full)
+		return 0;
+
+	if (len > ((PAGE_SIZE - 1) - s->len)) {
+		s->full = 1;
+		return 0;
+	}
+
+	memcpy(s->buffer + s->len, str, len);
+	s->len += len;
+
+	return len;
+}
+
+static int trace_empty(struct trace_iterator *iter)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	int cpu;
+
+	for_each_online_cpu(cpu) {
+		if (!ring_buffer_empty_cpu(ktap_iter->buffer, cpu))
+			return 0;
+	}
+
+	return 1;
+}
+
+static void trace_consume(struct trace_iterator *iter)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+
+	ring_buffer_consume(ktap_iter->buffer, iter->cpu, &iter->ts,
+			    &iter->lost_events);
+}
+
+unsigned long long ns2usecs(cycle_t nsec)
+{
+	nsec += 500;
+	do_div(nsec, 1000);
+	return nsec;
+}
+
+static int trace_print_timestamp(struct trace_iterator *iter)
+{
+	struct trace_seq *s = &iter->seq;
+	unsigned long long t;
+	unsigned long secs, usec_rem;
+
+	t = ns2usecs(iter->ts);
+	usec_rem = do_div(t, USEC_PER_SEC);
+	secs = (unsigned long)t;
+
+	return trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem);
+}
+
+/* todo: export kernel function ftrace_find_event in future, and make faster */
+static struct trace_event *(*ftrace_find_event)(int type);
+
+static enum print_line_t print_trace_fmt(struct trace_iterator *iter)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	struct trace_entry *entry = iter->ent;
+	struct trace_event *ev;
+
+	ev = ftrace_find_event(entry->type);
+
+	if (ktap_iter->print_timestamp && !trace_print_timestamp(iter))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	if (ev) {
+		int ret = ev->funcs->trace(iter, 0, ev);
+
+		/* overwrite '\n' at the ending */
+		iter->seq.buffer[iter->seq.len - 1] = '\0';
+		iter->seq.len--;
+		return ret;
+	}
+
+	return TRACE_TYPE_PARTIAL_LINE;
+}
+
+static enum print_line_t print_trace_stack(struct trace_iterator *iter)
+{
+	struct trace_entry *entry = iter->ent;
+	struct stack_trace trace;
+	char str[KSYM_SYMBOL_LEN];
+	int i;
+
+	trace.entries = (unsigned long *)(entry + 1);
+	trace.nr_entries = (iter->ent_size - sizeof(*entry)) /
+			   sizeof(unsigned long);
+
+	if (!_trace_seq_puts(&iter->seq, "<stack trace>\n"))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	for (i = 0; i < trace.nr_entries; i++) {
+		unsigned long p = trace.entries[i];
+
+		if (p == ULONG_MAX)
+			break;
+
+		sprint_symbol(str, p);
+		if (!trace_seq_printf(&iter->seq, " => %s\n", str))
+			return TRACE_TYPE_PARTIAL_LINE;
+	}
+
+	return TRACE_TYPE_HANDLED;
+}
+
+struct ktap_ftrace_entry {
+	struct trace_entry entry;
+	unsigned long ip;
+	unsigned long parent_ip;
+};
+
+static enum print_line_t print_trace_fn(struct trace_iterator *iter)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	struct ktap_ftrace_entry *field = (struct ktap_ftrace_entry *)iter->ent;
+	char str[KSYM_SYMBOL_LEN];
+
+	if (ktap_iter->print_timestamp && !trace_print_timestamp(iter))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	sprint_symbol(str, field->ip);
+	if (!_trace_seq_puts(&iter->seq, str))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	if (!_trace_seq_puts(&iter->seq, " <- "))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	sprint_symbol(str, field->parent_ip);
+	if (!_trace_seq_puts(&iter->seq, str))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	return TRACE_TYPE_HANDLED;
+}
+
+static enum print_line_t print_trace_bputs(struct trace_iterator *iter)
+{
+	if (!_trace_seq_puts(&iter->seq,
+			    (const char *)(*(unsigned long *)(iter->ent + 1))))
+		return TRACE_TYPE_PARTIAL_LINE;
+
+	return TRACE_TYPE_HANDLED;
+}
+
+static enum print_line_t print_trace_line(struct trace_iterator *iter)
+{
+	struct trace_entry *entry = iter->ent;
+	char *str = (char *)(entry + 1);
+
+	if (entry->type == TRACE_PRINT) {
+		if (!trace_seq_printf(&iter->seq, "%s", str))
+			return TRACE_TYPE_PARTIAL_LINE;
+
+		return TRACE_TYPE_HANDLED;
+	}
+
+	if (entry->type == TRACE_BPUTS)
+		return print_trace_bputs(iter);
+
+	if (entry->type == TRACE_STACK)
+		return print_trace_stack(iter);
+
+	if (entry->type == TRACE_FN)
+		return print_trace_fn(iter);
+
+	return print_trace_fmt(iter);
+}
+
+static struct trace_entry *
+peek_next_entry(struct trace_iterator *iter, int cpu, u64 *ts,
+		unsigned long *lost_events)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	struct ring_buffer_event *event;
+
+	event = ring_buffer_peek(ktap_iter->buffer, cpu, ts, lost_events);
+	if (event) {
+		iter->ent_size = ring_buffer_event_length(event);
+		return ring_buffer_event_data(event);
+	}
+
+	return NULL;
+}
+
+static struct trace_entry *
+__find_next_entry(struct trace_iterator *iter, int *ent_cpu,
+		  unsigned long *missing_events, u64 *ent_ts)
+{
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	struct ring_buffer *buffer = ktap_iter->buffer;
+	struct trace_entry *ent, *next = NULL;
+	unsigned long lost_events = 0, next_lost = 0;
+	u64 next_ts = 0, ts;
+	int next_cpu = -1;
+	int next_size = 0;
+	int cpu;
+
+	for_each_online_cpu(cpu) {
+		if (ring_buffer_empty_cpu(buffer, cpu))
+			continue;
+
+		ent = peek_next_entry(iter, cpu, &ts, &lost_events);
+		/*
+		 * Pick the entry with the smallest timestamp:
+		 */
+		if (ent && (!next || ts < next_ts)) {
+			next = ent;
+			next_cpu = cpu;
+			next_ts = ts;
+			next_lost = lost_events;
+			next_size = iter->ent_size;
+		}
+	}
+
+	iter->ent_size = next_size;
+
+	if (ent_cpu)
+		*ent_cpu = next_cpu;
+
+	if (ent_ts)
+		*ent_ts = next_ts;
+
+	if (missing_events)
+		*missing_events = next_lost;
+
+	return next;
+}
+
+/* Find the next real entry, and increment the iterator to the next entry */
+static void *trace_find_next_entry_inc(struct trace_iterator *iter)
+{
+	iter->ent = __find_next_entry(iter, &iter->cpu,
+				      &iter->lost_events, &iter->ts);
+	if (iter->ent)
+		iter->idx++;
+
+	return iter->ent ? iter : NULL;
+}
+
+static void poll_wait_pipe(void)
+{
+	set_current_state(TASK_INTERRUPTIBLE);
+	/* sleep for 100 msecs, and try again. */
+	schedule_timeout(HZ / 10);
+}
+
+static int tracing_wait_pipe(struct file *filp)
+{
+	struct trace_iterator *iter = filp->private_data;
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+	ktap_state_t *ks = ktap_iter->private;
+
+	while (trace_empty(iter)) {
+
+		if ((filp->f_flags & O_NONBLOCK)) {
+			return -EAGAIN;
+		}
+
+		mutex_unlock(&iter->mutex);
+
+		poll_wait_pipe();
+
+		mutex_lock(&iter->mutex);
+
+		if (G(ks)->wait_user && trace_empty(iter))
+			return -EINTR;
+	}
+
+	return 1;
+}
+
+static ssize_t
+tracing_read_pipe(struct file *filp, char __user *ubuf, size_t cnt,
+		  loff_t *ppos)
+{
+	struct trace_iterator *iter = filp->private_data;
+	ssize_t sret;
+
+	/* return any leftover data */
+	sret = _trace_seq_to_user(&iter->seq, ubuf, cnt);
+	if (sret != -EBUSY)
+		return sret;
+	/*
+	 * Avoid more than one consumer on a single file descriptor
+	 * This is just a matter of traces coherency, the ring buffer itself
+	 * is protected.
+	 */
+	mutex_lock(&iter->mutex);
+
+waitagain:
+	sret = tracing_wait_pipe(filp);
+	if (sret <= 0)
+		goto out;
+
+	/* stop when tracing is finished */
+	if (trace_empty(iter)) {
+		sret = 0;
+		goto out;
+	}
+
+	if (cnt >= PAGE_SIZE)
+		cnt = PAGE_SIZE - 1;
+
+	/* reset all but tr, trace, and overruns */
+	memset(&iter->seq, 0,
+	       sizeof(struct trace_iterator) -
+	       offsetof(struct trace_iterator, seq));
+	iter->pos = -1;
+
+	while (trace_find_next_entry_inc(iter) != NULL) {
+		enum print_line_t ret;
+		int len = iter->seq.len;
+
+		ret = print_trace_line(iter);
+		if (ret == TRACE_TYPE_PARTIAL_LINE) {
+			/* don't print partial lines */
+			iter->seq.len = len;
+			break;
+		}
+		if (ret != TRACE_TYPE_NO_CONSUME)
+			trace_consume(iter);
+
+		if (iter->seq.len >= cnt)
+			break;
+
+		/*
+		 * Setting the full flag means we reached the trace_seq buffer
+		 * size and we should leave by partial output condition above.
+		 * One of the trace_seq_* functions is not used properly.
+		 */
+		WARN_ONCE(iter->seq.full, "full flag set for trace type %d",
+			  iter->ent->type);
+	}
+
+	/* Now copy what we have to the user */
+	sret = _trace_seq_to_user(&iter->seq, ubuf, cnt);
+	if (iter->seq.readpos >= iter->seq.len)
+		trace_seq_init(&iter->seq);
+
+	/*
+	 * If there was nothing to send to user, in spite of consuming trace
+	 * entries, go back to wait for more entries.
+	 */
+	if (sret == -EBUSY)
+		goto waitagain;
+
+out:
+	mutex_unlock(&iter->mutex);
+
+	return sret;
+}
+
+static int tracing_open_pipe(struct inode *inode, struct file *filp)
+{
+	struct ktap_trace_iterator *ktap_iter;
+	ktap_state_t *ks = inode->i_private;
+
+	/* create a buffer to store the information to pass to userspace */
+	ktap_iter = kzalloc(sizeof(*ktap_iter), GFP_KERNEL);
+	if (!ktap_iter)
+		return -ENOMEM;
+
+	ktap_iter->private = ks;
+	ktap_iter->buffer = G(ks)->buffer;
+	ktap_iter->print_timestamp = G(ks)->parm->print_timestamp;
+	mutex_init(&ktap_iter->iter.mutex);
+	filp->private_data = &ktap_iter->iter;
+
+	nonseekable_open(inode, filp);
+
+	return 0;
+}
+
+static int tracing_release_pipe(struct inode *inode, struct file *file)
+{
+	struct trace_iterator *iter = file->private_data;
+	struct ktap_trace_iterator *ktap_iter = KTAP_TRACE_ITER(iter);
+
+	mutex_destroy(&iter->mutex);
+	kfree(ktap_iter);
+	return 0;
+}
+
+static const struct file_operations tracing_pipe_fops = {
+	.open		= tracing_open_pipe,
+	.read		= tracing_read_pipe,
+	.splice_read	= NULL,
+	.release	= tracing_release_pipe,
+	.llseek		= no_llseek,
+};
+
+/*
+ * preempt disabled in ring_buffer_lock_reserve
+ *
+ * The implementation is similar with funtion __ftrace_trace_stack.
+ */
+void kp_transport_print_kstack(ktap_state_t *ks, uint16_t depth, uint16_t skip)
+{
+	struct ring_buffer *buffer = G(ks)->buffer;
+	struct ring_buffer_event *event;
+	struct trace_entry *entry;
+	int size;
+
+	size = depth * sizeof(unsigned long);
+	event = ring_buffer_lock_reserve(buffer, sizeof(*entry) + size);
+	if (!event) {
+		KTAP_STATS(ks)->events_missed += 1;
+		return;
+	} else {
+		struct stack_trace trace;
+
+		entry = ring_buffer_event_data(event);
+		tracing_generic_entry_update(entry, 0, 0);
+		entry->type = TRACE_STACK;
+
+		trace.nr_entries = 0;
+		trace.skip = skip;
+		trace.max_entries = depth;
+		trace.entries = (unsigned long *)(entry + 1);
+		save_stack_trace(&trace);
+
+		ring_buffer_unlock_commit(buffer, event);
+	}
+}
+
+void kp_transport_event_write(ktap_state_t *ks, struct ktap_event_data *e)
+{
+	struct ring_buffer *buffer = G(ks)->buffer;
+	struct ring_buffer_event *event;
+	struct trace_entry *ev_entry = e->data->raw->data;
+	struct trace_entry *entry;
+	int entry_size = e->data->raw->size;
+
+	event = ring_buffer_lock_reserve(buffer, entry_size +
+					 sizeof(struct ftrace_event_call *));
+	if (!event) {
+		KTAP_STATS(ks)->events_missed += 1;
+		return;
+	} else {
+		entry = ring_buffer_event_data(event);
+
+		memcpy(entry, ev_entry, entry_size);
+
+		ring_buffer_unlock_commit(buffer, event);
+	}
+}
+
+void kp_transport_write(ktap_state_t *ks, const void *data, size_t length)
+{
+	struct ring_buffer *buffer = G(ks)->buffer;
+	struct ring_buffer_event *event;
+	struct trace_entry *entry;
+	int size;
+
+	size = sizeof(struct trace_entry) + length;
+
+	event = ring_buffer_lock_reserve(buffer, size);
+	if (!event) {
+		KTAP_STATS(ks)->events_missed += 1;
+		return;
+	} else {
+		entry = ring_buffer_event_data(event);
+
+		tracing_generic_entry_update(entry, 0, 0);
+		entry->type = TRACE_PRINT;
+		memcpy(entry + 1, data, length);
+
+		ring_buffer_unlock_commit(buffer, event);
+	}
+}
+
+/* general print function */
+void kp_printf(ktap_state_t *ks, const char *fmt, ...)
+{
+	char buff[1024];
+	va_list args;
+	int len;
+
+	va_start(args, fmt);
+	len = vscnprintf(buff, 1024, fmt, args);
+	va_end(args);
+
+	buff[len] = '\0';
+	kp_transport_write(ks, buff, len + 1);
+}
+
+void __kp_puts(ktap_state_t *ks, const char *str)
+{
+	kp_transport_write(ks, str, strlen(str) + 1);
+}
+
+void __kp_bputs(ktap_state_t *ks, const char *str)
+{
+	struct ring_buffer *buffer = G(ks)->buffer;
+	struct ring_buffer_event *event;
+	struct trace_entry *entry;
+	int size;
+
+	size = sizeof(struct trace_entry) + sizeof(unsigned long *);
+
+	event = ring_buffer_lock_reserve(buffer, size);
+	if (!event) {
+		KTAP_STATS(ks)->events_missed += 1;
+		return;
+	} else {
+		entry = ring_buffer_event_data(event);
+
+		tracing_generic_entry_update(entry, 0, 0);
+		entry->type = TRACE_BPUTS;
+		*(unsigned long *)(entry + 1) = (unsigned long)str;
+
+		ring_buffer_unlock_commit(buffer, event);
+	}
+}
+
+void kp_transport_exit(ktap_state_t *ks)
+{
+	if (G(ks)->buffer)
+		ring_buffer_free(G(ks)->buffer);
+	debugfs_remove(G(ks)->trace_pipe_dentry);
+}
+
+#define TRACE_BUF_SIZE_DEFAULT	1441792UL /* 16384 * 88 (sizeof(entry)) */
+
+int kp_transport_init(ktap_state_t *ks, struct dentry *dir)
+{
+	struct ring_buffer *buffer;
+	struct dentry *dentry;
+	char filename[32] = {0};
+
+#ifdef CONFIG_PPC64
+	ftrace_find_event = (void *)kallsyms_lookup_name(".ftrace_find_event");
+#else
+	ftrace_find_event = (void *)kallsyms_lookup_name("ftrace_find_event");
+#endif
+	if (!ftrace_find_event) {
+		printk("ktap: cannot lookup ftrace_find_event in kallsyms\n");
+		return -EINVAL;
+	}
+
+	buffer = ring_buffer_alloc(TRACE_BUF_SIZE_DEFAULT, RB_FL_OVERWRITE);
+	if (!buffer)
+		return -ENOMEM;
+
+	sprintf(filename, "trace_pipe_%d", (int)task_tgid_vnr(current));
+
+	dentry = debugfs_create_file(filename, 0444, dir,
+				     ks, &tracing_pipe_fops);
+	if (!dentry) {
+		pr_err("ktapvm: cannot create trace_pipe file in debugfs\n");
+		ring_buffer_free(buffer);
+		return -1;
+	}
+
+	G(ks)->buffer = buffer;
+	G(ks)->trace_pipe_dentry = dentry;
+
+	return 0;
+}
+
diff --git a/kernel/trace/ktap/kp_transport.h b/kernel/trace/ktap/kp_transport.h
new file mode 100644
index 0000000..bc7c892
--- /dev/null
+++ b/kernel/trace/ktap/kp_transport.h
@@ -0,0 +1,13 @@
+#ifndef __KTAP_TRANSPORT_H__
+#define __KTAP_TRANSPORT_H__
+
+void kp_transport_write(ktap_state_t *ks, const void *data, size_t length);
+void kp_transport_event_write(ktap_state_t *ks, struct ktap_event_data *e);
+void kp_transport_print_kstack(ktap_state_t *ks, uint16_t depth, uint16_t skip);
+void *kp_transport_reserve(ktap_state_t *ks, size_t length);
+void kp_transport_exit(ktap_state_t *ks);
+int kp_transport_init(ktap_state_t *ks, struct dentry *dir);
+
+int _trace_seq_puts(struct trace_seq *s, const char *str);
+
+#endif /* __KTAP_TRANSPORT_H__ */
-- 
1.8.1.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists