My last posting was mangled by my mailer. I hope this one is better.
Also corrected Randy's concerns.
Please see previous posting for more information:
http://lkml.org/lkml/2007/9/19/4 (PATCH 0/2)
Note: this patch requires "[Patch 2/2] Relay reset consumed" is applied.
---------<start of patch>------------
Trace - Provides tracing primitives
Signed-off-by: Tom Zanussi <[email protected]>
Signed-off-by: Martin Hunt <[email protected]>
Signed-off-by: David Wilder <[email protected]>
---
Documentation/trace/src/Makefile | 7 +
Documentation/trace/src/README | 18 +
Documentation/trace/src/fork_trace.c | 119 +++++++
Documentation/trace/trace.txt | 164 ++++++++++
include/linux/trace.h | 99 ++++++
lib/Kconfig | 9 +
lib/Makefile | 2 +
lib/trace.c | 563 ++++++++++++++++++++++++++++++++++
8 files changed, 981 insertions(+), 0 deletions(-)
diff --git a/Documentation/trace/src/Makefile b/Documentation/trace/src/Makefile
new file mode 100644
index 0000000..9ee4c72
--- /dev/null
+++ b/Documentation/trace/src/Makefile
@@ -0,0 +1,7 @@
+obj-m := fork_trace.o
+KDIR := /lib/modules/$(shell uname -r)/build
+PWD := $(shell pwd)
+default:
+ $(MAKE) -C $(KDIR) SUBDIRS=$(PWD) modules
+clean:
+ rm -f *.mod.c *.ko *.o
diff --git a/Documentation/trace/src/README b/Documentation/trace/src/README
new file mode 100644
index 0000000..f538491
--- /dev/null
+++ b/Documentation/trace/src/README
@@ -0,0 +1,18 @@
+This small sample module creates a trace channel. It places a kprobe
+on the function do_fork(). The value of current->pid is written to
+the trace channel each time the kprobe is hit..
+
+How to run the example:
+$ mount -t debugfs /debug
+$ make
+$ insmod fork_trace.ko
+
+To view the data produced by the module:
+$ cat /debug/trace_example/do_fork/trace0
+
+Remove the module.
+$ rmmod fork_trace
+
+The function trace_cleanup() is called when the module
+is removed. This will cause the TRACE channel to be destroyed and the
+corresponding files to disappear from the debug file system.
diff --git a/Documentation/trace/src/fork_trace.c b/Documentation/trace/src/fork_trace.c
new file mode 100644
index 0000000..8a7bc55
--- /dev/null
+++ b/Documentation/trace/src/fork_trace.c
@@ -0,0 +1,119 @@
+/*
+ * An example of using trace in a kprobes module
+ *
+ * Copyright (C) 2007 IBM Inc.
+ *
+ * David Wilder <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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/kernel.h>
+#include <linux/module.h>
+#include <linux/kprobes.h>
+#include <linux/trace.h>
+
+#define USE_GLOBAL_BUFFERS 1
+#define USE_FLIGHT 1
+
+#define PROBE_POINT "do_fork"
+
+static struct kprobe kp;
+static struct trace_info *kprobes_trace;
+
+#ifdef USE_GLOBAL_BUFFERS
+static DEFINE_SPINLOCK(trace_lock);
+#endif
+
+/*
+ * Send formatted trace data to trace channel.
+ * @note Preemption must be disabled to use this.
+ */
+static void trace_printf(struct trace_info *trace, const char *format, ...)
+{
+ va_list ap, aq;
+ char *record;
+ unsigned long flags;
+ int len;
+
+ if (!trace)
+ return;
+
+#ifdef USE_GLOBAL_BUFFERS
+ spin_lock_irqsave(&trace_lock, flags);
+#endif
+ if (trace_running(trace)) {
+ va_start(ap, format);
+ va_copy(aq, ap);
+ len = vsnprintf(NULL, 0, format, aq);
+ va_end(aq);
+ record = relay_reserve(trace->rchan, ++len);
+ if (record)
+ vsnprintf(record, len, format, ap);
+ va_end(ap);
+ }
+#ifdef USE_GLOBAL_BUFFERS
+ spin_unlock_irqrestore(&trace_lock, flags);
+#endif
+}
+
+static int handler_pre(struct kprobe *p, struct pt_regs *regs)
+{
+ rcu_read_lock();
+ trace_printf(kprobes_trace, "%d\n", current->pid);
+ rcu_read_unlock();
+ return 0;
+}
+
+int init_module(void)
+{
+ int ret;
+ u32 flags = 0;
+
+#ifdef USE_GLOBAL_BUFFERS
+ flags |= TRACE_GLOBAL_CHANNEL;
+#endif
+
+#ifdef USE_FLIGHT
+ flags |= TRACE_FLIGHT_CHANNEL;
+#endif
+
+ /* setup the trace */
+ kprobes_trace = trace_setup("trace_example", PROBE_POINT,
+ 1024, 8, flags);
+ if (IS_ERR(kprobes_trace))
+ return PTR_ERR(kprobes_trace);
+
+ trace_start(kprobes_trace);
+
+ /* setup the kprobe */
+ kp.pre_handler = handler_pre;
+ kp.post_handler = NULL;
+ kp.fault_handler = NULL;
+ kp.symbol_name = PROBE_POINT;
+ ret = register_kprobe(&kp);
+ if (ret) {
+ printk(KERN_ERR "fork_trace: register_kprobe failed\n");
+ return ret;
+ }
+ return 0;
+}
+
+void cleanup_module(void)
+{
+ unregister_kprobe(&kp);
+ trace_stop(kprobes_trace);
+ trace_cleanup(kprobes_trace);
+}
+MODULE_LICENSE("GPL");
diff --git a/Documentation/trace/trace.txt b/Documentation/trace/trace.txt
new file mode 100644
index 0000000..fc9fedb
--- /dev/null
+++ b/Documentation/trace/trace.txt
@@ -0,0 +1,164 @@
+Trace Setup and Control
+=======================
+In the kernel, the trace interface provides a simple mechanism for
+starting and managing data channels (traces) to user space. The
+trace interface builds on the relay interface. For a complete
+description of the relay interface, please see:
+Documentation/filesystems/relay.txt.
+
+The trace interface provides a single layer in a complete tracing
+application. Trace provides a kernel API that can be used for the setup
+and control of tracing channels. User of trace must provide a data layer
+responsible for formatting and writing data into the trace channels.
+
+A layered approach to tracing
+=============================
+A complete kernel tracing application consists of a data provider and
+a data consumer. Both provider and consumer contain three layers; each
+layer works in tandem with the corresponding layer in the opposite side.
+The layers are represented in the following diagram.
+
+Provider Data layer
+ Formats raw trace data and provides data-related service.
+ For example, adding timestamps used by consumer to sort data.
+
+Provider Control layer
+ Provided by the trace interface, this layer creates trace channels
+ and informs the data layer and consumer of the current state
+ of the trace channels.
+
+Provider Buffering layer
+ Provided by relay. This layer buffers data in the
+ kernel for consumption by the consumer's buffer
+ layer.
+
+Provider (in-kernel facility)
+-----------------------------------------------------------------------------
+Consumer (user application)
+
+
+Consumer Buffer layer
+ Reads/consumes data from the provider's data buffers.
+
+Consumer Control layer
+ Communicates to the provider's control layer to control the state
+ of the trace channels.
+
+Consumer Data layer
+ Sorts and formats data as provided by the provider's data layer.
+
+The provider is coded as a kernel facility. The consumer is coded as
+a user application.
+
+
+Trace - Features
+================
+Trace exploits services and features provided by relay. These features
+are:
+- The creation and destruction of relay channels.
+- Buffer management. Overwrite or non-overwrite modes can be selected
+ as well as global or per-CPU buffering.
+
+Overwrite mode can be called "flight recorder mode". Flight recorder
+mode is selected by setting the TRACE_FLIGHT_CHANNEL flag when
+creating trace channels. In flight mode when a tracing buffer is
+full, the oldest records in the buffer will be discarded to make room
+as new records arrive. In the default non-overwrite mode, new records
+may be written only if the buffer has room. In either case, to
+prevent data loss, a user space reader must keep the buffers
+drained. Trace provides a means to detect the number of records that
+have been dropped due to a buffer-full condition (non-overwrite mode
+only).
+
+When per-CPU buffers are used, relay creates one debugfs file for each
+running CPU. The user-space consumer of the data is responsible for
+reading the per-CPU buffers and collating the records presumably using
+a time stamp or sequence number included in the trace records. The
+use of global buffers eliminates this extra work of sequencing
+records; however the provider's data layer must hold a lock when
+writing records. The lock prevents writers running on different CPUs
+from overwriting each other's data. However, buffering may be slower
+because writes to the buffer are serialized. Global buffering is
+selected by setting the TRACE_GLOBAL_CHANNEL flag when creating trace
+channels.
+
+Trace User Interface
+===================
+When a trace channel is created and started, the following
+directories and files are created in the root of the mounted debugfs.
+
+/debug (root of the debugfs)
+ /<trace-root-dir>
+ /<trace-name>
+ trace0...traceN-1 Per-CPU trace data, one
+ file per CPU.
+
+ state Start or stop tracing by
+ by writing the strings
+ "start" or "stop" to this
+ file. Read the file to get the
+ current state.
+
+ dropped The number of records dropped
+ due to a full-buffer condition,
+ for non-TRACE_FLIGHT_CHANNELs
+ only.
+
+ rewind Trigger a rewind by writing
+ to this file. i.e. start
+ next read at the beginning
+ again. Only available for
+ TRACE_FLIGHT_CHANNELS.
+
+
+ nr_sub Number of sub-buffers
+ in the channel.
+
+ sub_size Size of sub-buffers in
+ the channnel.
+
+Trace data is gathered from the trace[0...N] files using one of the
+available interfaces provided by relay.
+
+When using the read(2) interface, as data is read it is marked as
+consumed by the relay subsystem. Therefore, subsequent reads will
+only return unconsumed data.
+
+Trace Kernel API
+===============
+An overview of the trace Kernel API is now given. More details of the
+API can be found in linux/trace.h.
+
+The steps a kernel data provider takes to utilize the trace interface are:
+1) Set up a trace channel - trace_setup()
+2) Start the trace channel - trace_start()
+3) Write one or more trace records into the channel (using the relay API).
+
+ Important: When writing a trace record the provider must insure that
+ preemption is disabled and that trace state is set to "running". A
+ typical function used to write records into a trace channel should
+ follow the following semantics:
+
+ rcu_read_lock(); // disables preemption
+ if (trace_running(trace)){
+ relay_write(....); // use any available relay data
+ // function
+ }
+ rcu_read_unlock(); // enables preemption
+
+4) Stop and start tracing as desired - trace_start()/trace_stop()
+5) Destroy the trace channel and underlying relay channel -
+ trace_cleanup().
+
+Trace Example
+=============
+See Documentation/trace/src for an example of using trace.
+
+Credits
+=======
+Trace is adapted from blktrace authored by Jens Axboe ([email protected]).
+
+Major contributions were made by:
+Tom Zanussi <[email protected]>
+Martin Hunt <[email protected]>
+David Wilder <[email protected]>
diff --git a/include/linux/trace.h b/include/linux/trace.h
new file mode 100644
index 0000000..a3a10c9
--- /dev/null
+++ b/include/linux/trace.h
@@ -0,0 +1,99 @@
+/*
+ * TRACE defines and function prototypes
+ *
+ * Copyright (C) 2007 IBM Inc.
+ *
+ * Tom Zanussi <[email protected]>
+ * Martin Hunt <[email protected]>
+ * David Wilder <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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
+ *
+ */
+#ifndef _LINUX_TRACE_H
+#define _LINUX_TRACE_H
+
+#include <linux/relay.h>
+
+/*
+ * TRACE channel flags
+ */
+#define TRACE_GLOBAL_CHANNEL 0x01
+#define TRACE_FLIGHT_CHANNEL 0x02
+#define TRACE_DISABLE_STATE 0x04
+
+enum trace_state {
+ TRACE_SETUP,
+ TRACE_RUNNING,
+ TRACE_STOPPED,
+};
+
+#define TRACE_ROOT_NAME_SIZE 64 /* Max root dir identifier */
+#define TRACE_NAME_SIZE 64 /* Max trace identifier */
+
+/*
+ * Global root user information
+ */
+struct trace_root {
+ struct list_head list;
+ char name[TRACE_ROOT_NAME_SIZE];
+ struct dentry *root;
+ unsigned int users;
+};
+
+/*
+ * Client information
+ */
+struct trace_info {
+ struct mutex state_mutex; /* Used to protect state changes */
+ enum trace_state state;
+ struct dentry *state_file;
+ struct rchan *rchan;
+ struct dentry *dir;
+ struct dentry *dropped_file;
+ struct dentry *reset_consumed_file;
+ struct dentry *nr_sub_file;
+ struct dentry *sub_size_file;
+ atomic_t dropped;
+ struct trace_root *root;
+ void *private_data;
+ unsigned int flags;
+ unsigned int buf_size;
+ unsigned int buf_nr;
+};
+
+#ifdef CONFIG_TRACE
+static inline int trace_running(struct trace_info *trace)
+{
+ return trace->state == TRACE_RUNNING;
+}
+struct trace_info *trace_setup(const char *root, const char *name,
+ u32 buf_size, u32 buf_nr, u32 flags);
+int trace_start(struct trace_info *trace);
+int trace_stop(struct trace_info *trace);
+void trace_cleanup(struct trace_info *trace);
+#else
+static inline struct trace_info *trace_setup(const char *root,
+ const char *name, u32 buf_size,
+ u32 buf_nr, u32 flags)
+{
+ return NULL;
+}
+static inline int trace_start(struct trace_info *trace) { return -EINVAL; }
+static inline int trace_stop(struct trace_info *trace) { return -EINVAL; }
+static inline int trace_running(struct trace_info *trace) { return 0; }
+static inline void trace_cleanup(struct trace_info *trace) {}
+#endif
+
+#endif
diff --git a/lib/Kconfig b/lib/Kconfig
index ba3d104..ad19f87 100644
--- a/lib/Kconfig
+++ b/lib/Kconfig
@@ -141,4 +141,13 @@ config HAS_DMA
config CHECK_SIGNATURE
bool
+config TRACE
+ bool "Trace setup and control"
+ depends on RELAY && DEBUG_FS
+ help
+ This option provides support for the setup, teardown and control
+ of tracing channels from kernel code. It also provides trace
+ information and control to userspace via a set of debugfs control
+ files. If unsure, say N.
+
endmenu
diff --git a/lib/Makefile b/lib/Makefile
index 76d6619..fa00ea1 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -70,6 +70,8 @@ lib-$(CONFIG_GENERIC_BUG) += bug.o
obj-$(CONFIG_PROFILE_LIKELY) += likely_prof.o
+obj-$(CONFIG_TRACE) += trace.o
+
hostprogs-y := gen_crc32table
clean-files := crc32table.h
diff --git a/lib/trace.c b/lib/trace.c
new file mode 100644
index 0000000..a760597
--- /dev/null
+++ b/lib/trace.c
@@ -0,0 +1,563 @@
+/*
+ * Based on blktrace code, Copyright (C) 2006 Jens Axboe <[email protected]>
+ * Moved to utt.c by Tom Zanussi <[email protected]>, 2006
+ * Additional contributions by:
+ * Martin Hunt <[email protected]>, 2007
+ * David Wilder <[email protected]>, 2007
+ * Renamed to trace <dwilder.ibm.com>, 2007
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/mutex.h>
+#include <linux/debugfs.h>
+#include <linux/trace.h>
+
+static LIST_HEAD(trace_roots);
+static DEFINE_MUTEX(trace_mutex);
+
+static int state_open(struct inode *inode, struct file *filp)
+{
+ filp->private_data = inode->i_private;
+ return 0;
+}
+
+static ssize_t state_read(struct file *filp, char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ struct trace_info *trace = filp->private_data;
+ char *buf = "trace not started\n";
+
+ if (trace->state == TRACE_STOPPED)
+ buf = "stopped\n";
+ else if (trace->state == TRACE_RUNNING)
+ buf = "running\n";
+ return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf));
+}
+
+static ssize_t state_write(struct file *filp, const char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ struct trace_info *trace = filp->private_data;
+ char buf[16];
+ int ret;
+
+ if (trace->flags & TRACE_DISABLE_STATE)
+ return -EINVAL;
+
+ if (count > sizeof(buf) - 1)
+ return -EINVAL;
+
+ if (copy_from_user(buf, buffer, count))
+ return -EFAULT;
+
+ buf[count] = '\0';
+
+ if (strncmp(buf, "start", strlen("start")) == 0) {
+ ret = trace_start(trace);
+ if (ret)
+ return ret;
+ } else if (strncmp(buf, "stop", strlen("stop")) == 0)
+ trace_stop(trace);
+ else
+ return -EINVAL;
+
+ return count;
+}
+
+static struct file_operations state_fops = {
+ .owner = THIS_MODULE,
+ .open = state_open,
+ .read = state_read,
+ .write = state_write,
+};
+
+static void remove_root(struct trace_info *trace)
+{
+ if (trace->root->root && simple_empty(trace->root->root)) {
+ debugfs_remove(trace->root->root);
+ list_del(&trace->root->list);
+ kfree(trace->root);
+ trace->root = NULL;
+ }
+}
+
+static void remove_tree(struct trace_info *trace)
+{
+ mutex_lock(&trace_mutex);
+ debugfs_remove(trace->dir);
+
+ if (trace->root) {
+ if (--trace->root->users == 0)
+ remove_root(trace);
+ }
+
+ mutex_unlock(&trace_mutex);
+}
+
+/*
+ * Creates the trace_root if it's not found.
+ */
+static struct trace_root *lookup_root(const char *root)
+{
+ struct list_head *pos;
+ struct trace_root *r;
+
+ list_for_each(pos, &trace_roots) {
+ r = list_entry(pos, struct trace_root, list);
+ if (!strcmp(r->name, root))
+ return r;
+ }
+
+ r = kzalloc(sizeof(struct trace_root), GFP_KERNEL);
+ if (!r)
+ return ERR_PTR(-ENOMEM);
+
+ strlcpy(r->name, root, sizeof(r->name));
+
+ r->root = debugfs_create_dir(root, NULL);
+ if (IS_ERR(r->root))
+ r->root = NULL;
+ else
+ list_add(&r->list, &trace_roots);
+
+ return r;
+}
+
+static struct dentry *create_tree(struct trace_info *trace, const char *root,
+ const char *name)
+{
+ struct dentry *dir = NULL;
+
+ if (root == NULL || name == NULL)
+ return ERR_PTR(-EINVAL);
+
+ mutex_lock(&trace_mutex);
+
+ trace->root = lookup_root(root);
+ if (IS_ERR(trace->root)) {
+ trace->root = NULL;
+ goto err;
+ }
+
+ dir = debugfs_create_dir(name, trace->root->root);
+ if (IS_ERR(dir))
+ remove_root(trace);
+ else
+ trace->root->users++;
+
+err:
+ mutex_unlock(&trace_mutex);
+ return dir;
+}
+
+static int dropped_open(struct inode *inode, struct file *filp)
+{
+ filp->private_data = inode->i_private;
+
+ return 0;
+}
+
+static ssize_t dropped_read(struct file *filp, char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ struct trace_info *trace = filp->private_data;
+ char buf[16];
+
+ snprintf(buf, sizeof(buf), "%u\n", atomic_read(&trace->dropped));
+
+ return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf));
+}
+
+static struct file_operations dropped_fops = {
+ .owner = THIS_MODULE,
+ .open = dropped_open,
+ .read = dropped_read,
+};
+
+static int reset_consumed_open(struct inode *inode, struct file *filp)
+{
+ filp->private_data = inode->i_private;
+
+ return 0;
+}
+
+static ssize_t reset_consumed_write(struct file *filp,
+ const char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ int ret = count;
+ struct trace_info *trace = filp->private_data;
+
+ mutex_lock(&trace->state_mutex);
+ switch (trace->state) {
+ case TRACE_RUNNING:
+ trace->state = TRACE_STOPPED;
+ synchronize_rcu();
+ relay_flush(trace->rchan);
+ relay_reset_consumed(trace->rchan);
+ trace->state = TRACE_RUNNING;
+ break;
+ case TRACE_STOPPED:
+ relay_reset_consumed(trace->rchan);
+ break;
+ default:
+ ret = -EINVAL;
+ }
+ mutex_unlock(&trace->state_mutex);
+ return ret;
+}
+
+static struct file_operations reset_consumed_fops = {
+ .owner = THIS_MODULE,
+ .open = reset_consumed_open,
+ .write = reset_consumed_write
+};
+
+static int sub_size_open(struct inode *inode, struct file *filp)
+{
+ filp->private_data = inode->i_private;
+
+ return 0;
+}
+
+static ssize_t sub_size_read(struct file *filp, char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ struct trace_info *trace = filp->private_data;
+ char buf[32];
+
+ snprintf(buf, sizeof(buf), "%zu\n", trace->rchan->subbuf_size);
+
+ return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf));
+}
+
+static struct file_operations sub_size_fops = {
+ .owner = THIS_MODULE,
+ .open = sub_size_open,
+ .read = sub_size_read,
+};
+
+static int nr_sub_open(struct inode *inode, struct file *filp)
+{
+ filp->private_data = inode->i_private;
+ return 0;
+}
+
+static ssize_t nr_sub_read(struct file *filp, char __user *buffer,
+ size_t count, loff_t *ppos)
+{
+ struct trace_info *trace = filp->private_data;
+ char buf[32];
+
+ snprintf(buf, sizeof(buf), "%zu\n", trace->rchan->n_subbufs);
+
+ return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf));
+}
+
+static struct file_operations nr_sub_fops = {
+ .owner = THIS_MODULE,
+ .open = nr_sub_open,
+ .read = nr_sub_read,
+};
+
+/*
+ * Keep track of how many times we encountered a full subbuffer, to aid
+ * the user space app in telling how many lost events there were.
+ */
+static int subbuf_start_callback(struct rchan_buf *buf, void *subbuf,
+ void *prev_subbuf, size_t prev_padding)
+{
+ struct trace_info *trace = buf->chan->private_data;
+
+ if (trace->flags & TRACE_FLIGHT_CHANNEL)
+ return 1;
+
+ if (!relay_buf_full(buf))
+ return 1;
+
+ atomic_inc(&trace->dropped);
+
+ return 0;
+}
+
+static int remove_buf_file_callback(struct dentry *dentry)
+{
+ debugfs_remove(dentry);
+
+ return 0;
+}
+
+static struct dentry *create_buf_file_callback(const char *filename,
+ struct dentry *parent, int mode,
+ struct rchan_buf *buf,
+ int *is_global)
+{
+ return debugfs_create_file(filename, mode, parent, buf,
+ &relay_file_operations);
+}
+
+static struct dentry *create_global_buf_file_callback(const char *filename,
+ struct dentry *parent,
+ int mode,
+ struct rchan_buf *buf,
+ int *is_global)
+{
+ *is_global = 1;
+
+ return debugfs_create_file(filename, mode, parent, buf,
+ &relay_file_operations);
+}
+
+static struct rchan_callbacks relay_callbacks = {
+ .subbuf_start = subbuf_start_callback,
+ .create_buf_file = create_buf_file_callback,
+ .remove_buf_file = remove_buf_file_callback,
+};
+static struct rchan_callbacks relay_callbacks_global = {
+ .subbuf_start = subbuf_start_callback,
+ .create_buf_file = create_global_buf_file_callback,
+ .remove_buf_file = remove_buf_file_callback,
+};
+
+static void remove_controls(struct trace_info *trace)
+{
+ debugfs_remove(trace->state_file);
+ debugfs_remove(trace->dropped_file);
+ debugfs_remove(trace->reset_consumed_file);
+ debugfs_remove(trace->nr_sub_file);
+ debugfs_remove(trace->sub_size_file);
+ remove_tree(trace);
+}
+
+/*
+ * Setup controls for tracing.
+ */
+static struct trace_info *setup_controls(const char *root,
+ const char *name, u32 flags)
+{
+ struct trace_info *trace;
+ long ret;
+
+ trace = kzalloc(sizeof(*trace), GFP_KERNEL);
+ if (!trace) {
+ ret = -ENOMEM;
+ goto err;
+ }
+
+ trace->dir = create_tree(trace, root, name);
+ if (IS_ERR(trace->dir)) {
+ ret = PTR_ERR(trace->dir);
+ trace->dir = NULL;
+ goto err;
+ }
+
+ trace->state_file = debugfs_create_file("state", 0444, trace->dir,
+ trace, &state_fops);
+ if (IS_ERR(trace->state_file)) {
+ ret = PTR_ERR(trace->state_file);
+ trace->state_file = NULL;
+ goto err;
+ }
+
+ if (!flags & TRACE_FLIGHT_CHANNEL) {
+ trace->dropped_file = debugfs_create_file("dropped", 0444,
+ trace->dir, trace,
+ &dropped_fops);
+ if (IS_ERR(trace->dropped_file)) {
+ ret = PTR_ERR(trace->dropped_file);
+ trace->dropped_file = NULL;
+ goto err;
+ }
+ }
+
+ if (flags & TRACE_FLIGHT_CHANNEL) {
+ trace->reset_consumed_file = debugfs_create_file("rewind", 0444,
+ trace->dir, trace,
+ &reset_consumed_fops);
+ if (IS_ERR(trace->reset_consumed_file)) {
+ ret = PTR_ERR(trace->reset_consumed_file);
+ trace->reset_consumed_file = NULL;
+ goto err;
+ }
+ }
+
+ trace->nr_sub_file = debugfs_create_file("nr_sub", 0444,
+ trace->dir, trace,
+ &nr_sub_fops);
+ if (IS_ERR(trace->nr_sub_file)) {
+ ret = PTR_ERR(trace->nr_sub_file);
+ trace->nr_sub_file = NULL;
+ goto err;
+ }
+
+ trace->sub_size_file = debugfs_create_file("sub_size", 0444,
+ trace->dir, trace,
+ &sub_size_fops);
+ if (IS_ERR(trace->sub_size_file)) {
+ ret = PTR_ERR(trace->sub_size_file);
+ trace->sub_size_file = NULL;
+ goto err;
+ }
+
+ return trace;
+err:
+ if (trace) {
+ remove_controls(trace);
+ kfree(trace);
+ }
+
+ return ERR_PTR(ret);
+}
+
+static int trace_setup_channel(struct trace_info *trace, u32 buf_size,
+ u32 buf_nr, u32 flags)
+{
+ if (!buf_size || !buf_nr)
+ return -EINVAL;
+
+ if (flags & TRACE_GLOBAL_CHANNEL)
+ trace->rchan = relay_open("trace", trace->dir, buf_size,
+ buf_nr, &relay_callbacks_global,
+ trace);
+ else
+ trace->rchan = relay_open("trace", trace->dir, buf_size,
+ buf_nr, &relay_callbacks, trace);
+
+ if (!trace->rchan)
+ return -ENOMEM;
+
+ trace->flags = flags;
+ trace->state = TRACE_SETUP;
+
+ return 0;
+}
+
+/**
+ * trace_setup - create a new trace trace handle
+ * @root: The root directory name to place trace directories.
+ * @name: Trace directory name, created in @root
+ * @buf_size: size of the relay sub-buffers
+ * @buf_nr: number of relay sub-buffers
+ * @flags: Option selection (see trace channel flags definitions)
+ *
+ * returns a trace_info handle or NULL, if setup failed.
+ *
+ * The @root is created (if needed) in the root of the debugfs.
+ * The default values when flags=0 are: use per-CPU buffering,
+ * use non-overwrite mode. See Documentation/trace.txt for details.
+ */
+struct trace_info *trace_setup(const char *root, const char *name,
+ u32 buf_size, u32 buf_nr, u32 flags)
+{
+ struct trace_info *trace;
+
+ trace = setup_controls(root, name, flags);
+ if (IS_ERR(trace))
+ return trace;
+
+ trace->buf_size = buf_size;
+ trace->buf_nr = buf_nr;
+ trace->flags = flags;
+ mutex_init(&trace->state_mutex);
+ trace->state = TRACE_SETUP;
+
+ return trace;
+}
+EXPORT_SYMBOL_GPL(trace_setup);
+
+/**
+ * trace_start - start tracing
+ * @trace: trace handle to start.
+ *
+ * returns 0 if successful.
+ */
+int trace_start(struct trace_info *trace)
+{
+ /*
+ * For starting a trace, we can transition from a setup or stopped
+ * trace.
+ */
+ if (trace->state == TRACE_RUNNING)
+ return -EINVAL;
+
+ mutex_lock(&trace->state_mutex);
+ if (trace->state == TRACE_SETUP) {
+ int ret;
+
+ ret = trace_setup_channel(trace, trace->buf_size,
+ trace->buf_nr, trace->flags);
+ if (ret) {
+ mutex_unlock(&trace->state_mutex);
+ return ret;
+ }
+ }
+
+ trace->state = TRACE_RUNNING;
+ mutex_unlock(&trace->state_mutex);
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(trace_start);
+
+/**
+ * trace_stop - stop tracing
+ * @trace: trace handle to stop.
+ *
+ */
+int trace_stop(struct trace_info *trace)
+{
+ int ret = -EINVAL;
+
+ /*
+ * For stopping a trace, the state must be running
+ */
+ mutex_lock(&trace->state_mutex);
+ if (trace->state == TRACE_RUNNING) {
+ trace->state = TRACE_STOPPED;
+ /*
+ * wait for all cpus to see the change in
+ * state before continuing
+ */
+ synchronize_sched();
+ relay_flush(trace->rchan);
+ ret = 0;
+ }
+ mutex_unlock(&trace->state_mutex);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(trace_stop);
+
+static void trace_cleanup_channel(struct trace_info *trace)
+{
+ trace_stop(trace);
+ relay_close(trace->rchan);
+ trace->rchan = NULL;
+}
+
+/**
+ * trace_cleanup - destroys the trace channel, control files and dir
+ * @trace: trace handle to cleanup
+ */
+void trace_cleanup(struct trace_info *trace)
+{
+ trace_cleanup_channel(trace);
+ remove_controls(trace);
+ kfree(trace);
+}
+EXPORT_SYMBOL_GPL(trace_cleanup);
On Fri, Sep 21, 2007 at 09:23:28PM -0700, David J. Wilder wrote:
> My last posting was mangled by my mailer. I hope this one is better.
> Also corrected Randy's concerns.
>
> Please see previous posting for more information:
> http://lkml.org/lkml/2007/9/19/4 (PATCH 0/2)
>
> Note: this patch requires "[Patch 2/2] Relay reset consumed" is applied.
>
> ---------<start of patch>------------
> Trace - Provides tracing primitives
>
> Signed-off-by: Tom Zanussi <[email protected]>
> Signed-off-by: Martin Hunt <[email protected]>
> Signed-off-by: David Wilder <[email protected]>
> ---
> Documentation/trace/src/Makefile | 7 +
> Documentation/trace/src/README | 18 +
> Documentation/trace/src/fork_trace.c | 119 +++++++
> Documentation/trace/trace.txt | 164 ++++++++++
NACK, don't put code into Documentation/. Put it into kernel as it's
actually useful kernel code. And add clone,exec and exit while you're at it.
Christoph Hellwig wrote:
> On Fri, Sep 21, 2007 at 09:23:28PM -0700, David J. Wilder wrote:
>> My last posting was mangled by my mailer. I hope this one is better.
>> Also corrected Randy's concerns.
>>
>> Please see previous posting for more information:
>> http://lkml.org/lkml/2007/9/19/4 (PATCH 0/2)
>>
>> Note: this patch requires "[Patch 2/2] Relay reset consumed" is applied.
>>
>> ---------<start of patch>------------
>> Trace - Provides tracing primitives
>>
>> Signed-off-by: Tom Zanussi <[email protected]>
>> Signed-off-by: Martin Hunt <[email protected]>
>> Signed-off-by: David Wilder <[email protected]>
>> ---
>> Documentation/trace/src/Makefile | 7 +
>> Documentation/trace/src/README | 18 +
>> Documentation/trace/src/fork_trace.c | 119 +++++++
>> Documentation/trace/trace.txt | 164 ++++++++++
>
> NACK, don't put code into Documentation/. Put it into kernel as it's
> actually useful kernel code.
Are you suggesting moving the example code into kernel? Or complaining
about example code in /Documentation?
And add clone,exec and exit while you're at it.
Hu? A syscall tracer sounds like a nice idea but that is not what I am
trying to accomplish. I will let Systemtap handle that.
On Mon, Sep 24, 2007 at 08:38:34AM -0700, David Wilder wrote:
> >NACK, don't put code into Documentation/. Put it into kernel as it's
> >actually useful kernel code.
>
> Are you suggesting moving the example code into kernel? Or complaining
> about example code in /Documentation?
Both. example code should be integrated with the build system so it
gets built.
> And add clone,exec and exit while you're at it.
>
> Hu? A syscall tracer sounds like a nice idea but that is not what I am
> trying to accomplish. I will let Systemtap handle that.
Systemtap doesn't help anyone as it's not in the tree. I haven't even
asked you to provide a full system call tracing modulem but provide at
least one that's useful for a certain use-case (looking at processes)
instead of almost useless code.
On Mon, 24 Sep 2007 16:41:14 +0100 Christoph Hellwig wrote:
> On Mon, Sep 24, 2007 at 08:38:34AM -0700, David Wilder wrote:
> > >NACK, don't put code into Documentation/. Put it into kernel as it's
> > >actually useful kernel code.
> >
> > Are you suggesting moving the example code into kernel? Or complaining
> > about example code in /Documentation?
>
> Both. example code should be integrated with the build system so it
> gets built.
How about if we build the sample files in the Documentation/ tree?
with added Kconfig, Makefiles etc.
> > And add clone,exec and exit while you're at it.
> >
> > Hu? A syscall tracer sounds like a nice idea but that is not what I am
> > trying to accomplish. I will let Systemtap handle that.
>
> Systemtap doesn't help anyone as it's not in the tree. I haven't even
> asked you to provide a full system call tracing modulem but provide at
> least one that's useful for a certain use-case (looking at processes)
> instead of almost useless code.
---
~Randy
Phaedrus says that Quality is about caring.
Christoph Hellwig wrote:
> On Mon, Sep 24, 2007 at 08:38:34AM -0700, David Wilder wrote:
>>> NACK, don't put code into Documentation/. Put it into kernel as it's
>>> actually useful kernel code.
>> Are you suggesting moving the example code into kernel? Or complaining
>> about example code in /Documentation?
>
> Both. example code should be integrated with the build system so it
> gets built.
I agree, but I have not seen this done before, can you point me at an
example of how to structure this?
>
>> And add clone,exec and exit while you're at it.
>>
>> Hu? A syscall tracer sounds like a nice idea but that is not what I am
>> trying to accomplish. I will let Systemtap handle that.
>
> Systemtap doesn't help anyone as it's not in the tree. I haven't even
> asked you to provide a full system call tracing modulem but provide at
> least one that's useful for a certain use-case (looking at processes)
> instead of almost useless code.
>
I don't have a problem adding additional trace points to the example
code. However, anyone trying to use the code for some real purpose will
want to tweak the code based on their needs, at a minimum to select what
data to trace. I don't think we gain much by adding more to the example
other than to make it more complicated. I am strong believer in keeping
example code as simple as possible.
If you are suggesting adding a separate feature for process tracing (not
just an example) that is a good idea also. But is should be a separate
patch, not part of the trace patch.