2009-03-11 10:51:26 +01:00
|
|
|
/*
|
|
|
|
* Monitoring code for network dropped packet alerts
|
|
|
|
*
|
|
|
|
* Copyright (C) 2009 Neil Horman <nhorman@tuxdriver.com>
|
|
|
|
*/
|
|
|
|
|
2012-05-16 21:58:40 +02:00
|
|
|
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
#include <linux/netdevice.h>
|
|
|
|
#include <linux/etherdevice.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/if_arp.h>
|
|
|
|
#include <linux/inetdevice.h>
|
|
|
|
#include <linux/inet.h>
|
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/netpoll.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/delay.h>
|
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/workqueue.h>
|
|
|
|
#include <linux/netlink.h>
|
|
|
|
#include <linux/net_dropmon.h>
|
|
|
|
#include <linux/percpu.h>
|
|
|
|
#include <linux/timer.h>
|
|
|
|
#include <linux/bitops.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 09:04:11 +01:00
|
|
|
#include <linux/slab.h>
|
2012-05-17 12:04:00 +02:00
|
|
|
#include <linux/module.h>
|
2009-03-11 10:51:26 +01:00
|
|
|
#include <net/genetlink.h>
|
2009-05-21 09:36:08 +02:00
|
|
|
#include <net/netevent.h>
|
2009-03-11 10:51:26 +01:00
|
|
|
|
2009-04-15 01:39:12 +02:00
|
|
|
#include <trace/events/skb.h>
|
2009-06-15 12:02:23 +02:00
|
|
|
#include <trace/events/napi.h>
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
#include <asm/unaligned.h>
|
|
|
|
|
|
|
|
#define TRACE_ON 1
|
|
|
|
#define TRACE_OFF 0
|
|
|
|
|
|
|
|
static void send_dm_alert(struct work_struct *unused);
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Globals, our netlink socket pointer
|
|
|
|
* and the work handle that will send up
|
|
|
|
* netlink alerts
|
|
|
|
*/
|
2009-05-21 09:36:08 +02:00
|
|
|
static int trace_state = TRACE_OFF;
|
2012-04-27 12:11:48 +02:00
|
|
|
static DEFINE_MUTEX(trace_state_mutex);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
struct per_cpu_dm_data {
|
|
|
|
struct work_struct dm_alert_work;
|
2012-04-27 12:11:49 +02:00
|
|
|
struct sk_buff __rcu *skb;
|
2009-03-11 10:51:26 +01:00
|
|
|
atomic_t dm_hit_count;
|
|
|
|
struct timer_list send_timer;
|
drop_monitor: prevent init path from scheduling on the wrong cpu
I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error. drop monitor maintains a per cpu structure,
that gets initalized from a single cpu. Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself . Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu. If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.
This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from. In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-01 10:18:02 +02:00
|
|
|
int cpu;
|
2009-03-11 10:51:26 +01:00
|
|
|
};
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
struct dm_hw_stat_delta {
|
|
|
|
struct net_device *dev;
|
2009-09-02 23:37:45 +02:00
|
|
|
unsigned long last_rx;
|
2009-05-21 09:36:08 +02:00
|
|
|
struct list_head list;
|
|
|
|
struct rcu_head rcu;
|
|
|
|
unsigned long last_drop_val;
|
|
|
|
};
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
static struct genl_family net_drop_monitor_family = {
|
|
|
|
.id = GENL_ID_GENERATE,
|
|
|
|
.hdrsize = 0,
|
|
|
|
.name = "NET_DM",
|
2009-04-27 12:17:31 +02:00
|
|
|
.version = 2,
|
2009-03-11 10:51:26 +01:00
|
|
|
.maxattr = NET_DM_CMD_MAX,
|
|
|
|
};
|
|
|
|
|
|
|
|
static DEFINE_PER_CPU(struct per_cpu_dm_data, dm_cpu_data);
|
|
|
|
|
|
|
|
static int dm_hit_limit = 64;
|
|
|
|
static int dm_delay = 1;
|
2009-05-21 09:36:08 +02:00
|
|
|
static unsigned long dm_hw_check_delta = 2*HZ;
|
|
|
|
static LIST_HEAD(hw_stats_list);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
static void reset_per_cpu_data(struct per_cpu_dm_data *data)
|
|
|
|
{
|
|
|
|
size_t al;
|
|
|
|
struct net_dm_alert_msg *msg;
|
2009-04-27 12:17:31 +02:00
|
|
|
struct nlattr *nla;
|
2012-04-27 12:11:49 +02:00
|
|
|
struct sk_buff *skb;
|
|
|
|
struct sk_buff *oskb = rcu_dereference_protected(data->skb, 1);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
al = sizeof(struct net_dm_alert_msg);
|
|
|
|
al += dm_hit_limit * sizeof(struct net_dm_drop_point);
|
2009-04-27 12:17:31 +02:00
|
|
|
al += sizeof(struct nlattr);
|
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
skb = genlmsg_new(al, GFP_KERNEL);
|
|
|
|
|
|
|
|
if (skb) {
|
|
|
|
genlmsg_put(skb, 0, 0, &net_drop_monitor_family,
|
|
|
|
0, NET_DM_CMD_ALERT);
|
|
|
|
nla = nla_reserve(skb, NLA_UNSPEC,
|
|
|
|
sizeof(struct net_dm_alert_msg));
|
|
|
|
msg = nla_data(nla);
|
|
|
|
memset(msg, 0, al);
|
drop_monitor: prevent init path from scheduling on the wrong cpu
I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error. drop monitor maintains a per cpu structure,
that gets initalized from a single cpu. Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself . Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu. If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.
This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from. In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-01 10:18:02 +02:00
|
|
|
} else
|
|
|
|
schedule_work_on(data->cpu, &data->dm_alert_work);
|
2012-04-27 12:11:49 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't need to lock this, since we are guaranteed to only
|
|
|
|
* run this on a single cpu at a time.
|
|
|
|
* Note also that we only update data->skb if the old and new skb
|
|
|
|
* pointers don't match. This ensures that we don't continually call
|
|
|
|
* synchornize_rcu if we repeatedly fail to alloc a new netlink message.
|
|
|
|
*/
|
|
|
|
if (skb != oskb) {
|
|
|
|
rcu_assign_pointer(data->skb, skb);
|
|
|
|
|
|
|
|
synchronize_rcu();
|
|
|
|
|
|
|
|
atomic_set(&data->dm_hit_count, dm_hit_limit);
|
|
|
|
}
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static void send_dm_alert(struct work_struct *unused)
|
|
|
|
{
|
|
|
|
struct sk_buff *skb;
|
2012-04-27 12:11:49 +02:00
|
|
|
struct per_cpu_dm_data *data = &get_cpu_var(dm_cpu_data);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
drop_monitor: prevent init path from scheduling on the wrong cpu
I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error. drop monitor maintains a per cpu structure,
that gets initalized from a single cpu. Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself . Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu. If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.
This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from. In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-01 10:18:02 +02:00
|
|
|
WARN_ON_ONCE(data->cpu != smp_processor_id());
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
/*
|
|
|
|
* Grab the skb we're about to send
|
|
|
|
*/
|
2012-04-27 12:11:49 +02:00
|
|
|
skb = rcu_dereference_protected(data->skb, 1);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Replace it with a new one
|
|
|
|
*/
|
|
|
|
reset_per_cpu_data(data);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Ship it!
|
|
|
|
*/
|
2012-04-27 12:11:49 +02:00
|
|
|
if (skb)
|
|
|
|
genlmsg_multicast(skb, 0, NET_DM_GRP_ALERT, GFP_KERNEL);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
put_cpu_var(dm_cpu_data);
|
2009-03-11 10:51:26 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is the timer function to delay the sending of an alert
|
|
|
|
* in the event that more drops will arrive during the
|
|
|
|
* hysteresis period. Note that it operates under the timer interrupt
|
|
|
|
* so we don't need to disable preemption here
|
|
|
|
*/
|
|
|
|
static void sched_send_work(unsigned long unused)
|
|
|
|
{
|
2012-04-27 12:11:49 +02:00
|
|
|
struct per_cpu_dm_data *data = &get_cpu_var(dm_cpu_data);
|
|
|
|
|
|
|
|
schedule_work_on(smp_processor_id(), &data->dm_alert_work);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
put_cpu_var(dm_cpu_data);
|
2009-03-11 10:51:26 +01:00
|
|
|
}
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
static void trace_drop_common(struct sk_buff *skb, void *location)
|
2009-03-11 10:51:26 +01:00
|
|
|
{
|
|
|
|
struct net_dm_alert_msg *msg;
|
|
|
|
struct nlmsghdr *nlh;
|
2009-04-27 12:17:31 +02:00
|
|
|
struct nlattr *nla;
|
2009-03-11 10:51:26 +01:00
|
|
|
int i;
|
2012-04-27 12:11:49 +02:00
|
|
|
struct sk_buff *dskb;
|
|
|
|
struct per_cpu_dm_data *data = &get_cpu_var(dm_cpu_data);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
rcu_read_lock();
|
|
|
|
dskb = rcu_dereference(data->skb);
|
|
|
|
|
|
|
|
if (!dskb)
|
|
|
|
goto out;
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
if (!atomic_add_unless(&data->dm_hit_count, -1, 0)) {
|
|
|
|
/*
|
|
|
|
* we're already at zero, discard this hit
|
|
|
|
*/
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
nlh = (struct nlmsghdr *)dskb->data;
|
2009-04-27 12:17:31 +02:00
|
|
|
nla = genlmsg_data(nlmsg_data(nlh));
|
|
|
|
msg = nla_data(nla);
|
2009-03-11 10:51:26 +01:00
|
|
|
for (i = 0; i < msg->entries; i++) {
|
|
|
|
if (!memcmp(&location, msg->points[i].pc, sizeof(void *))) {
|
|
|
|
msg->points[i].count++;
|
2012-04-19 09:16:21 +02:00
|
|
|
atomic_inc(&data->dm_hit_count);
|
2009-03-11 10:51:26 +01:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to create a new entry
|
|
|
|
*/
|
2012-04-27 12:11:49 +02:00
|
|
|
__nla_reserve_nohdr(dskb, sizeof(struct net_dm_drop_point));
|
2009-04-27 12:17:31 +02:00
|
|
|
nla->nla_len += NLA_ALIGN(sizeof(struct net_dm_drop_point));
|
2009-03-11 10:51:26 +01:00
|
|
|
memcpy(msg->points[msg->entries].pc, &location, sizeof(void *));
|
|
|
|
msg->points[msg->entries].count = 1;
|
|
|
|
msg->entries++;
|
|
|
|
|
|
|
|
if (!timer_pending(&data->send_timer)) {
|
|
|
|
data->send_timer.expires = jiffies + dm_delay * HZ;
|
|
|
|
add_timer_on(&data->send_timer, smp_processor_id());
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2012-04-27 12:11:49 +02:00
|
|
|
rcu_read_unlock();
|
|
|
|
put_cpu_var(dm_cpu_data);
|
2009-03-11 10:51:26 +01:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
tracing: Let tracepoints have data passed to tracepoint callbacks
This patch adds data to be passed to tracepoint callbacks.
The created functions from DECLARE_TRACE() now need a mandatory data
parameter. For example:
DECLARE_TRACE(mytracepoint, int value, value)
Will create the register function:
int register_trace_mytracepoint((void(*)(void *data, int value))probe,
void *data);
As the first argument, all callbacks (probes) must take a (void *data)
parameter. So a callback for the above tracepoint will look like:
void myprobe(void *data, int value)
{
}
The callback may choose to ignore the data parameter.
This change allows callbacks to register a private data pointer along
with the function probe.
void mycallback(void *data, int value);
register_trace_mytracepoint(mycallback, mydata);
Then the mycallback() will receive the "mydata" as the first parameter
before the args.
A more detailed example:
DECLARE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
/* In the C file */
DEFINE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
[...]
trace_mytracepoint(status);
/* In a file registering this tracepoint */
int my_callback(void *data, int status)
{
struct my_struct my_data = data;
[...]
}
[...]
my_data = kmalloc(sizeof(*my_data), GFP_KERNEL);
init_my_data(my_data);
register_trace_mytracepoint(my_callback, my_data);
The same callback can also be registered to the same tracepoint as long
as the data registered is different. Note, the data must also be used
to unregister the callback:
unregister_trace_mytracepoint(my_callback, my_data);
Because of the data parameter, tracepoints declared this way can not have
no args. That is:
DECLARE_TRACE(mytracepoint, TP_PROTO(void), TP_ARGS());
will cause an error.
If no arguments are needed, a new macro can be used instead:
DECLARE_TRACE_NOARGS(mytracepoint);
Since there are no arguments, the proto and args fields are left out.
This is part of a series to make the tracepoint footprint smaller:
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
Again, this patch also increases the size of the kernel, but
lays the ground work for decreasing it.
v5: Fixed net/core/drop_monitor.c to handle these updates.
v4: Moved the DECLARE_TRACE() DECLARE_TRACE_NOARGS out of the
#ifdef CONFIG_TRACE_POINTS, since the two are the same in both
cases. The __DECLARE_TRACE() is what changes.
Thanks to Frederic Weisbecker for pointing this out.
v3: Made all register_* functions require data to be passed and
all callbacks to take a void * parameter as its first argument.
This makes the calling functions comply with C standards.
Also added more comments to the modifications of DECLARE_TRACE().
v2: Made the DECLARE_TRACE() have the ability to pass arguments
and added a new DECLARE_TRACE_NOARGS() for tracepoints that
do not need any arguments.
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-20 23:04:50 +02:00
|
|
|
static void trace_kfree_skb_hit(void *ignore, struct sk_buff *skb, void *location)
|
2009-05-21 09:36:08 +02:00
|
|
|
{
|
|
|
|
trace_drop_common(skb, location);
|
|
|
|
}
|
|
|
|
|
tracing: Let tracepoints have data passed to tracepoint callbacks
This patch adds data to be passed to tracepoint callbacks.
The created functions from DECLARE_TRACE() now need a mandatory data
parameter. For example:
DECLARE_TRACE(mytracepoint, int value, value)
Will create the register function:
int register_trace_mytracepoint((void(*)(void *data, int value))probe,
void *data);
As the first argument, all callbacks (probes) must take a (void *data)
parameter. So a callback for the above tracepoint will look like:
void myprobe(void *data, int value)
{
}
The callback may choose to ignore the data parameter.
This change allows callbacks to register a private data pointer along
with the function probe.
void mycallback(void *data, int value);
register_trace_mytracepoint(mycallback, mydata);
Then the mycallback() will receive the "mydata" as the first parameter
before the args.
A more detailed example:
DECLARE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
/* In the C file */
DEFINE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
[...]
trace_mytracepoint(status);
/* In a file registering this tracepoint */
int my_callback(void *data, int status)
{
struct my_struct my_data = data;
[...]
}
[...]
my_data = kmalloc(sizeof(*my_data), GFP_KERNEL);
init_my_data(my_data);
register_trace_mytracepoint(my_callback, my_data);
The same callback can also be registered to the same tracepoint as long
as the data registered is different. Note, the data must also be used
to unregister the callback:
unregister_trace_mytracepoint(my_callback, my_data);
Because of the data parameter, tracepoints declared this way can not have
no args. That is:
DECLARE_TRACE(mytracepoint, TP_PROTO(void), TP_ARGS());
will cause an error.
If no arguments are needed, a new macro can be used instead:
DECLARE_TRACE_NOARGS(mytracepoint);
Since there are no arguments, the proto and args fields are left out.
This is part of a series to make the tracepoint footprint smaller:
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
Again, this patch also increases the size of the kernel, but
lays the ground work for decreasing it.
v5: Fixed net/core/drop_monitor.c to handle these updates.
v4: Moved the DECLARE_TRACE() DECLARE_TRACE_NOARGS out of the
#ifdef CONFIG_TRACE_POINTS, since the two are the same in both
cases. The __DECLARE_TRACE() is what changes.
Thanks to Frederic Weisbecker for pointing this out.
v3: Made all register_* functions require data to be passed and
all callbacks to take a void * parameter as its first argument.
This makes the calling functions comply with C standards.
Also added more comments to the modifications of DECLARE_TRACE().
v2: Made the DECLARE_TRACE() have the ability to pass arguments
and added a new DECLARE_TRACE_NOARGS() for tracepoints that
do not need any arguments.
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-20 23:04:50 +02:00
|
|
|
static void trace_napi_poll_hit(void *ignore, struct napi_struct *napi)
|
2009-05-21 09:36:08 +02:00
|
|
|
{
|
|
|
|
struct dm_hw_stat_delta *new_stat;
|
|
|
|
|
|
|
|
/*
|
2009-09-02 23:37:45 +02:00
|
|
|
* Don't check napi structures with no associated device
|
2009-05-21 09:36:08 +02:00
|
|
|
*/
|
2009-09-02 23:37:45 +02:00
|
|
|
if (!napi->dev)
|
2009-05-21 09:36:08 +02:00
|
|
|
return;
|
|
|
|
|
|
|
|
rcu_read_lock();
|
|
|
|
list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {
|
2009-09-02 23:37:45 +02:00
|
|
|
/*
|
|
|
|
* only add a note to our monitor buffer if:
|
|
|
|
* 1) this is the dev we received on
|
|
|
|
* 2) its after the last_rx delta
|
|
|
|
* 3) our rx_dropped count has gone up
|
|
|
|
*/
|
2009-05-21 09:36:08 +02:00
|
|
|
if ((new_stat->dev == napi->dev) &&
|
2009-09-02 23:37:45 +02:00
|
|
|
(time_after(jiffies, new_stat->last_rx + dm_hw_check_delta)) &&
|
2009-05-21 09:36:08 +02:00
|
|
|
(napi->dev->stats.rx_dropped != new_stat->last_drop_val)) {
|
|
|
|
trace_drop_common(NULL, NULL);
|
|
|
|
new_stat->last_drop_val = napi->dev->stats.rx_dropped;
|
2009-09-02 23:37:45 +02:00
|
|
|
new_stat->last_rx = jiffies;
|
2009-05-21 09:36:08 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
rcu_read_unlock();
|
|
|
|
}
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
static int set_all_monitor_traces(int state)
|
|
|
|
{
|
|
|
|
int rc = 0;
|
2009-05-21 09:36:08 +02:00
|
|
|
struct dm_hw_stat_delta *new_stat = NULL;
|
|
|
|
struct dm_hw_stat_delta *temp;
|
|
|
|
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_lock(&trace_state_mutex);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
2010-07-20 06:52:09 +02:00
|
|
|
if (state == trace_state) {
|
|
|
|
rc = -EAGAIN;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
switch (state) {
|
|
|
|
case TRACE_ON:
|
2012-05-17 12:04:00 +02:00
|
|
|
if (!try_module_get(THIS_MODULE)) {
|
|
|
|
rc = -ENODEV;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
tracing: Let tracepoints have data passed to tracepoint callbacks
This patch adds data to be passed to tracepoint callbacks.
The created functions from DECLARE_TRACE() now need a mandatory data
parameter. For example:
DECLARE_TRACE(mytracepoint, int value, value)
Will create the register function:
int register_trace_mytracepoint((void(*)(void *data, int value))probe,
void *data);
As the first argument, all callbacks (probes) must take a (void *data)
parameter. So a callback for the above tracepoint will look like:
void myprobe(void *data, int value)
{
}
The callback may choose to ignore the data parameter.
This change allows callbacks to register a private data pointer along
with the function probe.
void mycallback(void *data, int value);
register_trace_mytracepoint(mycallback, mydata);
Then the mycallback() will receive the "mydata" as the first parameter
before the args.
A more detailed example:
DECLARE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
/* In the C file */
DEFINE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
[...]
trace_mytracepoint(status);
/* In a file registering this tracepoint */
int my_callback(void *data, int status)
{
struct my_struct my_data = data;
[...]
}
[...]
my_data = kmalloc(sizeof(*my_data), GFP_KERNEL);
init_my_data(my_data);
register_trace_mytracepoint(my_callback, my_data);
The same callback can also be registered to the same tracepoint as long
as the data registered is different. Note, the data must also be used
to unregister the callback:
unregister_trace_mytracepoint(my_callback, my_data);
Because of the data parameter, tracepoints declared this way can not have
no args. That is:
DECLARE_TRACE(mytracepoint, TP_PROTO(void), TP_ARGS());
will cause an error.
If no arguments are needed, a new macro can be used instead:
DECLARE_TRACE_NOARGS(mytracepoint);
Since there are no arguments, the proto and args fields are left out.
This is part of a series to make the tracepoint footprint smaller:
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
Again, this patch also increases the size of the kernel, but
lays the ground work for decreasing it.
v5: Fixed net/core/drop_monitor.c to handle these updates.
v4: Moved the DECLARE_TRACE() DECLARE_TRACE_NOARGS out of the
#ifdef CONFIG_TRACE_POINTS, since the two are the same in both
cases. The __DECLARE_TRACE() is what changes.
Thanks to Frederic Weisbecker for pointing this out.
v3: Made all register_* functions require data to be passed and
all callbacks to take a void * parameter as its first argument.
This makes the calling functions comply with C standards.
Also added more comments to the modifications of DECLARE_TRACE().
v2: Made the DECLARE_TRACE() have the ability to pass arguments
and added a new DECLARE_TRACE_NOARGS() for tracepoints that
do not need any arguments.
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-20 23:04:50 +02:00
|
|
|
rc |= register_trace_kfree_skb(trace_kfree_skb_hit, NULL);
|
|
|
|
rc |= register_trace_napi_poll(trace_napi_poll_hit, NULL);
|
2009-03-11 10:51:26 +01:00
|
|
|
break;
|
2012-05-17 12:04:00 +02:00
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
case TRACE_OFF:
|
tracing: Let tracepoints have data passed to tracepoint callbacks
This patch adds data to be passed to tracepoint callbacks.
The created functions from DECLARE_TRACE() now need a mandatory data
parameter. For example:
DECLARE_TRACE(mytracepoint, int value, value)
Will create the register function:
int register_trace_mytracepoint((void(*)(void *data, int value))probe,
void *data);
As the first argument, all callbacks (probes) must take a (void *data)
parameter. So a callback for the above tracepoint will look like:
void myprobe(void *data, int value)
{
}
The callback may choose to ignore the data parameter.
This change allows callbacks to register a private data pointer along
with the function probe.
void mycallback(void *data, int value);
register_trace_mytracepoint(mycallback, mydata);
Then the mycallback() will receive the "mydata" as the first parameter
before the args.
A more detailed example:
DECLARE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
/* In the C file */
DEFINE_TRACE(mytracepoint, TP_PROTO(int status), TP_ARGS(status));
[...]
trace_mytracepoint(status);
/* In a file registering this tracepoint */
int my_callback(void *data, int status)
{
struct my_struct my_data = data;
[...]
}
[...]
my_data = kmalloc(sizeof(*my_data), GFP_KERNEL);
init_my_data(my_data);
register_trace_mytracepoint(my_callback, my_data);
The same callback can also be registered to the same tracepoint as long
as the data registered is different. Note, the data must also be used
to unregister the callback:
unregister_trace_mytracepoint(my_callback, my_data);
Because of the data parameter, tracepoints declared this way can not have
no args. That is:
DECLARE_TRACE(mytracepoint, TP_PROTO(void), TP_ARGS());
will cause an error.
If no arguments are needed, a new macro can be used instead:
DECLARE_TRACE_NOARGS(mytracepoint);
Since there are no arguments, the proto and args fields are left out.
This is part of a series to make the tracepoint footprint smaller:
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
Again, this patch also increases the size of the kernel, but
lays the ground work for decreasing it.
v5: Fixed net/core/drop_monitor.c to handle these updates.
v4: Moved the DECLARE_TRACE() DECLARE_TRACE_NOARGS out of the
#ifdef CONFIG_TRACE_POINTS, since the two are the same in both
cases. The __DECLARE_TRACE() is what changes.
Thanks to Frederic Weisbecker for pointing this out.
v3: Made all register_* functions require data to be passed and
all callbacks to take a void * parameter as its first argument.
This makes the calling functions comply with C standards.
Also added more comments to the modifications of DECLARE_TRACE().
v2: Made the DECLARE_TRACE() have the ability to pass arguments
and added a new DECLARE_TRACE_NOARGS() for tracepoints that
do not need any arguments.
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2010-04-20 23:04:50 +02:00
|
|
|
rc |= unregister_trace_kfree_skb(trace_kfree_skb_hit, NULL);
|
|
|
|
rc |= unregister_trace_napi_poll(trace_napi_poll_hit, NULL);
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
tracepoint_synchronize_unregister();
|
2009-05-21 09:36:08 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Clean the device list
|
|
|
|
*/
|
|
|
|
list_for_each_entry_safe(new_stat, temp, &hw_stats_list, list) {
|
|
|
|
if (new_stat->dev == NULL) {
|
|
|
|
list_del_rcu(&new_stat->list);
|
2011-03-18 04:39:43 +01:00
|
|
|
kfree_rcu(new_stat, rcu);
|
2009-05-21 09:36:08 +02:00
|
|
|
}
|
|
|
|
}
|
2012-05-17 12:04:00 +02:00
|
|
|
|
|
|
|
module_put(THIS_MODULE);
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
rc = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
if (!rc)
|
|
|
|
trace_state = state;
|
2010-07-20 06:52:09 +02:00
|
|
|
else
|
|
|
|
rc = -EINPROGRESS;
|
2009-05-21 09:36:08 +02:00
|
|
|
|
2010-07-20 06:52:09 +02:00
|
|
|
out_unlock:
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_unlock(&trace_state_mutex);
|
2009-05-21 09:36:08 +02:00
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int net_dm_cmd_config(struct sk_buff *skb,
|
|
|
|
struct genl_info *info)
|
|
|
|
{
|
|
|
|
return -ENOTSUPP;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int net_dm_cmd_trace(struct sk_buff *skb,
|
|
|
|
struct genl_info *info)
|
|
|
|
{
|
|
|
|
switch (info->genlhdr->cmd) {
|
|
|
|
case NET_DM_CMD_START:
|
|
|
|
return set_all_monitor_traces(TRACE_ON);
|
|
|
|
break;
|
|
|
|
case NET_DM_CMD_STOP:
|
|
|
|
return set_all_monitor_traces(TRACE_OFF);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return -ENOTSUPP;
|
|
|
|
}
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
static int dropmon_net_event(struct notifier_block *ev_block,
|
|
|
|
unsigned long event, void *ptr)
|
|
|
|
{
|
|
|
|
struct net_device *dev = ptr;
|
|
|
|
struct dm_hw_stat_delta *new_stat = NULL;
|
|
|
|
struct dm_hw_stat_delta *tmp;
|
|
|
|
|
|
|
|
switch (event) {
|
|
|
|
case NETDEV_REGISTER:
|
|
|
|
new_stat = kzalloc(sizeof(struct dm_hw_stat_delta), GFP_KERNEL);
|
|
|
|
|
|
|
|
if (!new_stat)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
new_stat->dev = dev;
|
2009-09-02 23:37:45 +02:00
|
|
|
new_stat->last_rx = jiffies;
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_lock(&trace_state_mutex);
|
2009-05-21 09:36:08 +02:00
|
|
|
list_add_rcu(&new_stat->list, &hw_stats_list);
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_unlock(&trace_state_mutex);
|
2009-05-21 09:36:08 +02:00
|
|
|
break;
|
|
|
|
case NETDEV_UNREGISTER:
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_lock(&trace_state_mutex);
|
2009-05-21 09:36:08 +02:00
|
|
|
list_for_each_entry_safe(new_stat, tmp, &hw_stats_list, list) {
|
|
|
|
if (new_stat->dev == dev) {
|
|
|
|
new_stat->dev = NULL;
|
|
|
|
if (trace_state == TRACE_OFF) {
|
|
|
|
list_del_rcu(&new_stat->list);
|
2011-03-18 04:39:43 +01:00
|
|
|
kfree_rcu(new_stat, rcu);
|
2009-05-21 09:36:08 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2012-04-27 12:11:48 +02:00
|
|
|
mutex_unlock(&trace_state_mutex);
|
2009-05-21 09:36:08 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
out:
|
|
|
|
return NOTIFY_DONE;
|
|
|
|
}
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
static struct genl_ops dropmon_ops[] = {
|
|
|
|
{
|
|
|
|
.cmd = NET_DM_CMD_CONFIG,
|
|
|
|
.doit = net_dm_cmd_config,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.cmd = NET_DM_CMD_START,
|
|
|
|
.doit = net_dm_cmd_trace,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.cmd = NET_DM_CMD_STOP,
|
|
|
|
.doit = net_dm_cmd_trace,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
static struct notifier_block dropmon_net_notifier = {
|
|
|
|
.notifier_call = dropmon_net_event
|
|
|
|
};
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
static int __init init_net_drop_monitor(void)
|
|
|
|
{
|
|
|
|
struct per_cpu_dm_data *data;
|
2010-07-27 05:59:42 +02:00
|
|
|
int cpu, rc;
|
|
|
|
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_info("Initializing network drop monitor service\n");
|
2009-03-11 10:51:26 +01:00
|
|
|
|
|
|
|
if (sizeof(void *) > 8) {
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_err("Unable to store program counters on this arch, Drop monitor failed\n");
|
2009-03-11 10:51:26 +01:00
|
|
|
return -ENOSPC;
|
|
|
|
}
|
|
|
|
|
2010-07-27 05:59:42 +02:00
|
|
|
rc = genl_register_family_with_ops(&net_drop_monitor_family,
|
|
|
|
dropmon_ops,
|
|
|
|
ARRAY_SIZE(dropmon_ops));
|
|
|
|
if (rc) {
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_err("Could not create drop monitor netlink family\n");
|
2010-07-27 05:59:42 +02:00
|
|
|
return rc;
|
2009-03-11 10:51:26 +01:00
|
|
|
}
|
|
|
|
|
2009-05-21 09:36:08 +02:00
|
|
|
rc = register_netdevice_notifier(&dropmon_net_notifier);
|
|
|
|
if (rc < 0) {
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_crit("Failed to register netdevice notifier\n");
|
2009-05-21 09:36:08 +02:00
|
|
|
goto out_unreg;
|
|
|
|
}
|
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
rc = 0;
|
|
|
|
|
2012-05-17 12:04:00 +02:00
|
|
|
for_each_possible_cpu(cpu) {
|
2009-03-11 10:51:26 +01:00
|
|
|
data = &per_cpu(dm_cpu_data, cpu);
|
drop_monitor: prevent init path from scheduling on the wrong cpu
I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error. drop monitor maintains a per cpu structure,
that gets initalized from a single cpu. Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself . Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu. If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.
This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from. In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-01 10:18:02 +02:00
|
|
|
data->cpu = cpu;
|
2009-03-11 10:51:26 +01:00
|
|
|
INIT_WORK(&data->dm_alert_work, send_dm_alert);
|
|
|
|
init_timer(&data->send_timer);
|
|
|
|
data->send_timer.data = cpu;
|
|
|
|
data->send_timer.function = sched_send_work;
|
drop_monitor: prevent init path from scheduling on the wrong cpu
I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error. drop monitor maintains a per cpu structure,
that gets initalized from a single cpu. Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself . Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu. If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.
This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from. In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-01 10:18:02 +02:00
|
|
|
reset_per_cpu_data(data);
|
2009-03-11 10:51:26 +01:00
|
|
|
}
|
2009-05-21 09:36:08 +02:00
|
|
|
|
2012-04-27 12:11:49 +02:00
|
|
|
|
2009-03-11 10:51:26 +01:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
out_unreg:
|
|
|
|
genl_unregister_family(&net_drop_monitor_family);
|
|
|
|
out:
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2012-05-17 12:04:00 +02:00
|
|
|
static void exit_net_drop_monitor(void)
|
|
|
|
{
|
|
|
|
struct per_cpu_dm_data *data;
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
BUG_ON(unregister_netdevice_notifier(&dropmon_net_notifier));
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Because of the module_get/put we do in the trace state change path
|
|
|
|
* we are guarnateed not to have any current users when we get here
|
|
|
|
* all we need to do is make sure that we don't have any running timers
|
|
|
|
* or pending schedule calls
|
|
|
|
*/
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu) {
|
|
|
|
data = &per_cpu(dm_cpu_data, cpu);
|
|
|
|
del_timer_sync(&data->send_timer);
|
|
|
|
cancel_work_sync(&data->dm_alert_work);
|
|
|
|
/*
|
|
|
|
* At this point, we should have exclusive access
|
|
|
|
* to this struct and can free the skb inside it
|
|
|
|
*/
|
|
|
|
kfree_skb(data->skb);
|
|
|
|
}
|
|
|
|
|
|
|
|
BUG_ON(genl_unregister_family(&net_drop_monitor_family));
|
|
|
|
}
|
|
|
|
|
|
|
|
module_init(init_net_drop_monitor);
|
|
|
|
module_exit(exit_net_drop_monitor);
|
|
|
|
|
|
|
|
MODULE_LICENSE("GPL v2");
|
|
|
|
MODULE_AUTHOR("Neil Horman <nhorman@tuxdriver.com>");
|
2012-05-29 11:30:42 +02:00
|
|
|
MODULE_ALIAS_GENL_FAMILY("NET_DM");
|