diff options
author | Li Zefan <lizf@cn.fujitsu.com> | 2009-06-09 01:43:05 -0400 |
---|---|---|
committer | Steven Rostedt <rostedt@goodmis.org> | 2009-06-09 12:34:23 -0400 |
commit | 55782138e47d9baf2f7d3a7af9e7cf42adf72c56 (patch) | |
tree | c7ccabae20e27bbeb08b69a358e8b86c98d1d9f3 | |
parent | f57a8a1911342265e7acdc190333c4e9235a6632 (diff) |
tracing/events: convert block trace points to TRACE_EVENT()
TRACE_EVENT is a more generic way to define tracepoints. Doing so adds
these new capabilities to this tracepoint:
- zero-copy and per-cpu splice() tracing
- binary tracing without printf overhead
- structured logging records exposed under /debug/tracing/events
- trace events embedded in function tracer output and other plugins
- user-defined, per tracepoint filter expressions
...
Cons:
- no dev_t info for the output of plug, unplug_timer and unplug_io events.
no dev_t info for getrq and sleeprq events if bio == NULL.
no dev_t info for rq_abort,...,rq_requeue events if rq->rq_disk == NULL.
This is mainly because we can't get the deivce from a request queue.
But this may change in the future.
- A packet command is converted to a string in TP_assign, not TP_print.
While blktrace do the convertion just before output.
Since pc requests should be rather rare, this is not a big issue.
- In blktrace, an event can have 2 different print formats, but a TRACE_EVENT
has a unique format, which means we have some unused data in a trace entry.
The overhead is minimized by using __dynamic_array() instead of __array().
I've benchmarked the ioctl blktrace vs the splice based TRACE_EVENT tracing:
dd dd + ioctl blktrace dd + TRACE_EVENT (splice)
1 7.36s, 42.7 MB/s 7.50s, 42.0 MB/s 7.41s, 42.5 MB/s
2 7.43s, 42.3 MB/s 7.48s, 42.1 MB/s 7.43s, 42.4 MB/s
3 7.38s, 42.6 MB/s 7.45s, 42.2 MB/s 7.41s, 42.5 MB/s
So the overhead of tracing is very small, and no regression when using
those trace events vs blktrace.
And the binary output of TRACE_EVENT is much smaller than blktrace:
# ls -l -h
-rw-r--r-- 1 root root 8.8M 06-09 13:24 sda.blktrace.0
-rw-r--r-- 1 root root 195K 06-09 13:24 sda.blktrace.1
-rw-r--r-- 1 root root 2.7M 06-09 13:25 trace_splice.out
Following are some comparisons between TRACE_EVENT and blktrace:
plug:
kjournald-480 [000] 303.084981: block_plug: [kjournald]
kjournald-480 [000] 303.084981: 8,0 P N [kjournald]
unplug_io:
kblockd/0-118 [000] 300.052973: block_unplug_io: [kblockd/0] 1
kblockd/0-118 [000] 300.052974: 8,0 U N [kblockd/0] 1
remap:
kjournald-480 [000] 303.085042: block_remap: 8,0 W 102736992 + 8 <- (8,8) 33384
kjournald-480 [000] 303.085043: 8,0 A W 102736992 + 8 <- (8,8) 33384
bio_backmerge:
kjournald-480 [000] 303.085086: block_bio_backmerge: 8,0 W 102737032 + 8 [kjournald]
kjournald-480 [000] 303.085086: 8,0 M W 102737032 + 8 [kjournald]
getrq:
kjournald-480 [000] 303.084974: block_getrq: 8,0 W 102736984 + 8 [kjournald]
kjournald-480 [000] 303.084975: 8,0 G W 102736984 + 8 [kjournald]
bash-2066 [001] 1072.953770: 8,0 G N [bash]
bash-2066 [001] 1072.953773: block_getrq: 0,0 N 0 + 0 [bash]
rq_complete:
konsole-2065 [001] 300.053184: block_rq_complete: 8,0 W () 103669040 + 16 [0]
konsole-2065 [001] 300.053191: 8,0 C W 103669040 + 16 [0]
ksoftirqd/1-7 [001] 1072.953811: 8,0 C N (5a 00 08 00 00 00 00 00 24 00) [0]
ksoftirqd/1-7 [001] 1072.953813: block_rq_complete: 0,0 N (5a 00 08 00 00 00 00 00 24 00) 0 + 0 [0]
rq_insert:
kjournald-480 [000] 303.084985: block_rq_insert: 8,0 W 0 () 102736984 + 8 [kjournald]
kjournald-480 [000] 303.084986: 8,0 I W 102736984 + 8 [kjournald]
Changelog from v2 -> v3:
- use the newly introduced __dynamic_array().
Changelog from v1 -> v2:
- use __string() instead of __array() to minimize the memory required
to store hex dump of rq->cmd().
- support large pc requests.
- add missing blk_fill_rwbs_rq() in block_rq_requeue TRACE_EVENT.
- some cleanups.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4A2DF669.5070905@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
-rw-r--r-- | block/blk-core.c | 16 | ||||
-rw-r--r-- | block/elevator.c | 8 | ||||
-rw-r--r-- | drivers/md/dm.c | 5 | ||||
-rw-r--r-- | fs/bio.c | 3 | ||||
-rw-r--r-- | include/linux/blktrace_api.h | 13 | ||||
-rw-r--r-- | include/trace/block.h | 76 | ||||
-rw-r--r-- | include/trace/events/block.h | 483 | ||||
-rw-r--r-- | kernel/trace/Makefile | 5 | ||||
-rw-r--r-- | kernel/trace/blktrace.c | 78 | ||||
-rw-r--r-- | mm/bounce.c | 5 |
10 files changed, 588 insertions, 104 deletions
diff --git a/block/blk-core.c b/block/blk-core.c index 1306de9cce04..9475bf99b891 100644 --- a/block/blk-core.c +++ b/block/blk-core.c | |||
@@ -28,22 +28,14 @@ | |||
28 | #include <linux/task_io_accounting_ops.h> | 28 | #include <linux/task_io_accounting_ops.h> |
29 | #include <linux/blktrace_api.h> | 29 | #include <linux/blktrace_api.h> |
30 | #include <linux/fault-inject.h> | 30 | #include <linux/fault-inject.h> |
31 | #include <trace/block.h> | 31 | |
32 | #define CREATE_TRACE_POINTS | ||
33 | #include <trace/events/block.h> | ||
32 | 34 | ||
33 | #include "blk.h" | 35 | #include "blk.h" |
34 | 36 | ||
35 | DEFINE_TRACE(block_plug); | ||
36 | DEFINE_TRACE(block_unplug_io); | ||
37 | DEFINE_TRACE(block_unplug_timer); | ||
38 | DEFINE_TRACE(block_getrq); | ||
39 | DEFINE_TRACE(block_sleeprq); | ||
40 | DEFINE_TRACE(block_rq_requeue); | ||
41 | DEFINE_TRACE(block_bio_backmerge); | ||
42 | DEFINE_TRACE(block_bio_frontmerge); | ||
43 | DEFINE_TRACE(block_bio_queue); | ||
44 | DEFINE_TRACE(block_rq_complete); | ||
45 | DEFINE_TRACE(block_remap); /* Also used in drivers/md/dm.c */ | ||
46 | EXPORT_TRACEPOINT_SYMBOL_GPL(block_remap); | 37 | EXPORT_TRACEPOINT_SYMBOL_GPL(block_remap); |
38 | EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete); | ||
47 | 39 | ||
48 | static int __make_request(struct request_queue *q, struct bio *bio); | 40 | static int __make_request(struct request_queue *q, struct bio *bio); |
49 | 41 | ||
diff --git a/block/elevator.c b/block/elevator.c index 7073a9072577..e220f0c543e3 100644 --- a/block/elevator.c +++ b/block/elevator.c | |||
@@ -33,17 +33,16 @@ | |||
33 | #include <linux/compiler.h> | 33 | #include <linux/compiler.h> |
34 | #include <linux/delay.h> | 34 | #include <linux/delay.h> |
35 | #include <linux/blktrace_api.h> | 35 | #include <linux/blktrace_api.h> |
36 | #include <trace/block.h> | ||
37 | #include <linux/hash.h> | 36 | #include <linux/hash.h> |
38 | #include <linux/uaccess.h> | 37 | #include <linux/uaccess.h> |
39 | 38 | ||
39 | #include <trace/events/block.h> | ||
40 | |||
40 | #include "blk.h" | 41 | #include "blk.h" |
41 | 42 | ||
42 | static DEFINE_SPINLOCK(elv_list_lock); | 43 | static DEFINE_SPINLOCK(elv_list_lock); |
43 | static LIST_HEAD(elv_list); | 44 | static LIST_HEAD(elv_list); |
44 | 45 | ||
45 | DEFINE_TRACE(block_rq_abort); | ||
46 | |||
47 | /* | 46 | /* |
48 | * Merge hash stuff. | 47 | * Merge hash stuff. |
49 | */ | 48 | */ |
@@ -55,9 +54,6 @@ static const int elv_hash_shift = 6; | |||
55 | #define rq_hash_key(rq) ((rq)->sector + (rq)->nr_sectors) | 54 | #define rq_hash_key(rq) ((rq)->sector + (rq)->nr_sectors) |
56 | #define ELV_ON_HASH(rq) (!hlist_unhashed(&(rq)->hash)) | 55 | #define ELV_ON_HASH(rq) (!hlist_unhashed(&(rq)->hash)) |
57 | 56 | ||
58 | DEFINE_TRACE(block_rq_insert); | ||
59 | DEFINE_TRACE(block_rq_issue); | ||
60 | |||
61 | /* | 57 | /* |
62 | * Query io scheduler to see if the current process issuing bio may be | 58 | * Query io scheduler to see if the current process issuing bio may be |
63 | * merged with rq. | 59 | * merged with rq. |
diff --git a/drivers/md/dm.c b/drivers/md/dm.c index e2ee4a79ea2c..3fd8b1e65483 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c | |||
@@ -20,7 +20,8 @@ | |||
20 | #include <linux/idr.h> | 20 | #include <linux/idr.h> |
21 | #include <linux/hdreg.h> | 21 | #include <linux/hdreg.h> |
22 | #include <linux/blktrace_api.h> | 22 | #include <linux/blktrace_api.h> |
23 | #include <trace/block.h> | 23 | |
24 | #include <trace/events/block.h> | ||
24 | 25 | ||
25 | #define DM_MSG_PREFIX "core" | 26 | #define DM_MSG_PREFIX "core" |
26 | 27 | ||
@@ -53,8 +54,6 @@ struct dm_target_io { | |||
53 | union map_info info; | 54 | union map_info info; |
54 | }; | 55 | }; |
55 | 56 | ||
56 | DEFINE_TRACE(block_bio_complete); | ||
57 | |||
58 | /* | 57 | /* |
59 | * For request-based dm. | 58 | * For request-based dm. |
60 | * One of these is allocated per request. | 59 | * One of these is allocated per request. |
@@ -26,10 +26,9 @@ | |||
26 | #include <linux/mempool.h> | 26 | #include <linux/mempool.h> |
27 | #include <linux/workqueue.h> | 27 | #include <linux/workqueue.h> |
28 | #include <linux/blktrace_api.h> | 28 | #include <linux/blktrace_api.h> |
29 | #include <trace/block.h> | ||
30 | #include <scsi/sg.h> /* for struct sg_iovec */ | 29 | #include <scsi/sg.h> /* for struct sg_iovec */ |
31 | 30 | ||
32 | DEFINE_TRACE(block_split); | 31 | #include <trace/events/block.h> |
33 | 32 | ||
34 | /* | 33 | /* |
35 | * Test patch to inline a certain number of bi_io_vec's inside the bio | 34 | * Test patch to inline a certain number of bi_io_vec's inside the bio |
diff --git a/include/linux/blktrace_api.h b/include/linux/blktrace_api.h index 82b4636030e9..c7ec31dd04c9 100644 --- a/include/linux/blktrace_api.h +++ b/include/linux/blktrace_api.h | |||
@@ -218,5 +218,18 @@ static inline int blk_trace_init_sysfs(struct device *dev) | |||
218 | 218 | ||
219 | #endif /* CONFIG_BLK_DEV_IO_TRACE */ | 219 | #endif /* CONFIG_BLK_DEV_IO_TRACE */ |
220 | 220 | ||
221 | #ifdef CONFIG_EVENT_TRACING | ||
222 | |||
223 | static inline int blk_cmd_buf_len(struct request *rq) | ||
224 | { | ||
225 | return blk_pc_request(rq) ? rq->cmd_len * 3 : 1; | ||
226 | } | ||
227 | |||
228 | extern void blk_dump_cmd(char *buf, struct request *rq); | ||
229 | extern void blk_fill_rwbs(char *rwbs, u32 rw, int bytes); | ||
230 | extern void blk_fill_rwbs_rq(char *rwbs, struct request *rq); | ||
231 | |||
232 | #endif /* CONFIG_EVENT_TRACING */ | ||
233 | |||
221 | #endif /* __KERNEL__ */ | 234 | #endif /* __KERNEL__ */ |
222 | #endif | 235 | #endif |
diff --git a/include/trace/block.h b/include/trace/block.h deleted file mode 100644 index 5b12efa096b6..000000000000 --- a/include/trace/block.h +++ /dev/null | |||
@@ -1,76 +0,0 @@ | |||
1 | #ifndef _TRACE_BLOCK_H | ||
2 | #define _TRACE_BLOCK_H | ||
3 | |||
4 | #include <linux/blkdev.h> | ||
5 | #include <linux/tracepoint.h> | ||
6 | |||
7 | DECLARE_TRACE(block_rq_abort, | ||
8 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
9 | TP_ARGS(q, rq)); | ||
10 | |||
11 | DECLARE_TRACE(block_rq_insert, | ||
12 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
13 | TP_ARGS(q, rq)); | ||
14 | |||
15 | DECLARE_TRACE(block_rq_issue, | ||
16 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
17 | TP_ARGS(q, rq)); | ||
18 | |||
19 | DECLARE_TRACE(block_rq_requeue, | ||
20 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
21 | TP_ARGS(q, rq)); | ||
22 | |||
23 | DECLARE_TRACE(block_rq_complete, | ||
24 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
25 | TP_ARGS(q, rq)); | ||
26 | |||
27 | DECLARE_TRACE(block_bio_bounce, | ||
28 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
29 | TP_ARGS(q, bio)); | ||
30 | |||
31 | DECLARE_TRACE(block_bio_complete, | ||
32 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
33 | TP_ARGS(q, bio)); | ||
34 | |||
35 | DECLARE_TRACE(block_bio_backmerge, | ||
36 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
37 | TP_ARGS(q, bio)); | ||
38 | |||
39 | DECLARE_TRACE(block_bio_frontmerge, | ||
40 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
41 | TP_ARGS(q, bio)); | ||
42 | |||
43 | DECLARE_TRACE(block_bio_queue, | ||
44 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
45 | TP_ARGS(q, bio)); | ||
46 | |||
47 | DECLARE_TRACE(block_getrq, | ||
48 | TP_PROTO(struct request_queue *q, struct bio *bio, int rw), | ||
49 | TP_ARGS(q, bio, rw)); | ||
50 | |||
51 | DECLARE_TRACE(block_sleeprq, | ||
52 | TP_PROTO(struct request_queue *q, struct bio *bio, int rw), | ||
53 | TP_ARGS(q, bio, rw)); | ||
54 | |||
55 | DECLARE_TRACE(block_plug, | ||
56 | TP_PROTO(struct request_queue *q), | ||
57 | TP_ARGS(q)); | ||
58 | |||
59 | DECLARE_TRACE(block_unplug_timer, | ||
60 | TP_PROTO(struct request_queue *q), | ||
61 | TP_ARGS(q)); | ||
62 | |||
63 | DECLARE_TRACE(block_unplug_io, | ||
64 | TP_PROTO(struct request_queue *q), | ||
65 | TP_ARGS(q)); | ||
66 | |||
67 | DECLARE_TRACE(block_split, | ||
68 | TP_PROTO(struct request_queue *q, struct bio *bio, unsigned int pdu), | ||
69 | TP_ARGS(q, bio, pdu)); | ||
70 | |||
71 | DECLARE_TRACE(block_remap, | ||
72 | TP_PROTO(struct request_queue *q, struct bio *bio, dev_t dev, | ||
73 | sector_t from), | ||
74 | TP_ARGS(q, bio, dev, from)); | ||
75 | |||
76 | #endif | ||
diff --git a/include/trace/events/block.h b/include/trace/events/block.h new file mode 100644 index 000000000000..a99d1e565bb0 --- /dev/null +++ b/include/trace/events/block.h | |||
@@ -0,0 +1,483 @@ | |||
1 | #if !defined(_TRACE_BLOCK_H) || defined(TRACE_HEADER_MULTI_READ) | ||
2 | #define _TRACE_BLOCK_H | ||
3 | |||
4 | #include <linux/blktrace_api.h> | ||
5 | #include <linux/blkdev.h> | ||
6 | #include <linux/tracepoint.h> | ||
7 | |||
8 | #undef TRACE_SYSTEM | ||
9 | #define TRACE_SYSTEM block | ||
10 | |||
11 | TRACE_EVENT(block_rq_abort, | ||
12 | |||
13 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
14 | |||
15 | TP_ARGS(q, rq), | ||
16 | |||
17 | TP_STRUCT__entry( | ||
18 | __field( dev_t, dev ) | ||
19 | __field( sector_t, sector ) | ||
20 | __field( unsigned int, nr_sector ) | ||
21 | __field( int, errors ) | ||
22 | __array( char, rwbs, 6 ) | ||
23 | __dynamic_array( char, cmd, blk_cmd_buf_len(rq) ) | ||
24 | ), | ||
25 | |||
26 | TP_fast_assign( | ||
27 | __entry->dev = rq->rq_disk ? disk_devt(rq->rq_disk) : 0; | ||
28 | __entry->sector = blk_pc_request(rq) ? 0 : rq->hard_sector; | ||
29 | __entry->nr_sector = blk_pc_request(rq) ? | ||
30 | 0 : rq->hard_nr_sectors; | ||
31 | __entry->errors = rq->errors; | ||
32 | |||
33 | blk_fill_rwbs_rq(__entry->rwbs, rq); | ||
34 | blk_dump_cmd(__get_str(cmd), rq); | ||
35 | ), | ||
36 | |||
37 | TP_printk("%d,%d %s (%s) %llu + %u [%d]", | ||
38 | MAJOR(__entry->dev), MINOR(__entry->dev), | ||
39 | __entry->rwbs, __get_str(cmd), | ||
40 | __entry->sector, __entry->nr_sector, __entry->errors) | ||
41 | ); | ||
42 | |||
43 | TRACE_EVENT(block_rq_insert, | ||
44 | |||
45 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
46 | |||
47 | TP_ARGS(q, rq), | ||
48 | |||
49 | TP_STRUCT__entry( | ||
50 | __field( dev_t, dev ) | ||
51 | __field( sector_t, sector ) | ||
52 | __field( unsigned int, nr_sector ) | ||
53 | __field( unsigned int, bytes ) | ||
54 | __array( char, rwbs, 6 ) | ||
55 | __array( char, comm, TASK_COMM_LEN ) | ||
56 | __dynamic_array( char, cmd, blk_cmd_buf_len(rq) ) | ||
57 | ), | ||
58 | |||
59 | TP_fast_assign( | ||
60 | __entry->dev = rq->rq_disk ? disk_devt(rq->rq_disk) : 0; | ||
61 | __entry->sector = blk_pc_request(rq) ? 0 : rq->hard_sector; | ||
62 | __entry->nr_sector = blk_pc_request(rq) ? | ||
63 | 0 : rq->hard_nr_sectors; | ||
64 | __entry->bytes = blk_pc_request(rq) ? rq->data_len : 0; | ||
65 | |||
66 | blk_fill_rwbs_rq(__entry->rwbs, rq); | ||
67 | blk_dump_cmd(__get_str(cmd), rq); | ||
68 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
69 | ), | ||
70 | |||
71 | TP_printk("%d,%d %s %u (%s) %llu + %u [%s]", | ||
72 | MAJOR(__entry->dev), MINOR(__entry->dev), | ||
73 | __entry->rwbs, __entry->bytes, __get_str(cmd), | ||
74 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
75 | ); | ||
76 | |||
77 | TRACE_EVENT(block_rq_issue, | ||
78 | |||
79 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
80 | |||
81 | TP_ARGS(q, rq), | ||
82 | |||
83 | TP_STRUCT__entry( | ||
84 | __field( dev_t, dev ) | ||
85 | __field( sector_t, sector ) | ||
86 | __field( unsigned int, nr_sector ) | ||
87 | __field( unsigned int, bytes ) | ||
88 | __array( char, rwbs, 6 ) | ||
89 | __array( char, comm, TASK_COMM_LEN ) | ||
90 | __dynamic_array( char, cmd, blk_cmd_buf_len(rq) ) | ||
91 | ), | ||
92 | |||
93 | TP_fast_assign( | ||
94 | __entry->dev = rq->rq_disk ? disk_devt(rq->rq_disk) : 0; | ||
95 | __entry->sector = blk_pc_request(rq) ? 0 : rq->hard_sector; | ||
96 | __entry->nr_sector = blk_pc_request(rq) ? | ||
97 | 0 : rq->hard_nr_sectors; | ||
98 | __entry->bytes = blk_pc_request(rq) ? rq->data_len : 0; | ||
99 | |||
100 | blk_fill_rwbs_rq(__entry->rwbs, rq); | ||
101 | blk_dump_cmd(__get_str(cmd), rq); | ||
102 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
103 | ), | ||
104 | |||
105 | TP_printk("%d,%d %s %u (%s) %llu + %u [%s]", | ||
106 | MAJOR(__entry->dev), MINOR(__entry->dev), | ||
107 | __entry->rwbs, __entry->bytes, __get_str(cmd), | ||
108 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
109 | ); | ||
110 | |||
111 | TRACE_EVENT(block_rq_requeue, | ||
112 | |||
113 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
114 | |||
115 | TP_ARGS(q, rq), | ||
116 | |||
117 | TP_STRUCT__entry( | ||
118 | __field( dev_t, dev ) | ||
119 | __field( sector_t, sector ) | ||
120 | __field( unsigned int, nr_sector ) | ||
121 | __field( int, errors ) | ||
122 | __array( char, rwbs, 6 ) | ||
123 | __dynamic_array( char, cmd, blk_cmd_buf_len(rq) ) | ||
124 | ), | ||
125 | |||
126 | TP_fast_assign( | ||
127 | __entry->dev = rq->rq_disk ? disk_devt(rq->rq_disk) : 0; | ||
128 | __entry->sector = blk_pc_request(rq) ? 0 : rq->hard_sector; | ||
129 | __entry->nr_sector = blk_pc_request(rq) ? | ||
130 | 0 : rq->hard_nr_sectors; | ||
131 | __entry->errors = rq->errors; | ||
132 | |||
133 | blk_fill_rwbs_rq(__entry->rwbs, rq); | ||
134 | blk_dump_cmd(__get_str(cmd), rq); | ||
135 | ), | ||
136 | |||
137 | TP_printk("%d,%d %s (%s) %llu + %u [%d]", | ||
138 | MAJOR(__entry->dev), MINOR(__entry->dev), | ||
139 | __entry->rwbs, __get_str(cmd), | ||
140 | __entry->sector, __entry->nr_sector, __entry->errors) | ||
141 | ); | ||
142 | |||
143 | TRACE_EVENT(block_rq_complete, | ||
144 | |||
145 | TP_PROTO(struct request_queue *q, struct request *rq), | ||
146 | |||
147 | TP_ARGS(q, rq), | ||
148 | |||
149 | TP_STRUCT__entry( | ||
150 | __field( dev_t, dev ) | ||
151 | __field( sector_t, sector ) | ||
152 | __field( unsigned int, nr_sector ) | ||
153 | __field( int, errors ) | ||
154 | __array( char, rwbs, 6 ) | ||
155 | __dynamic_array( char, cmd, blk_cmd_buf_len(rq) ) | ||
156 | ), | ||
157 | |||
158 | TP_fast_assign( | ||
159 | __entry->dev = rq->rq_disk ? disk_devt(rq->rq_disk) : 0; | ||
160 | __entry->sector = blk_pc_request(rq) ? 0 : rq->hard_sector; | ||
161 | __entry->nr_sector = blk_pc_request(rq) ? | ||
162 | 0 : rq->hard_nr_sectors; | ||
163 | __entry->errors = rq->errors; | ||
164 | |||
165 | blk_fill_rwbs_rq(__entry->rwbs, rq); | ||
166 | blk_dump_cmd(__get_str(cmd), rq); | ||
167 | ), | ||
168 | |||
169 | TP_printk("%d,%d %s (%s) %llu + %u [%d]", | ||
170 | MAJOR(__entry->dev), MINOR(__entry->dev), | ||
171 | __entry->rwbs, __get_str(cmd), | ||
172 | __entry->sector, __entry->nr_sector, __entry->errors) | ||
173 | ); | ||
174 | TRACE_EVENT(block_bio_bounce, | ||
175 | |||
176 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
177 | |||
178 | TP_ARGS(q, bio), | ||
179 | |||
180 | TP_STRUCT__entry( | ||
181 | __field( dev_t, dev ) | ||
182 | __field( sector_t, sector ) | ||
183 | __field( unsigned int, nr_sector ) | ||
184 | __array( char, rwbs, 6 ) | ||
185 | __array( char, comm, TASK_COMM_LEN ) | ||
186 | ), | ||
187 | |||
188 | TP_fast_assign( | ||
189 | __entry->dev = bio->bi_bdev->bd_dev; | ||
190 | __entry->sector = bio->bi_sector; | ||
191 | __entry->nr_sector = bio->bi_size >> 9; | ||
192 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
193 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
194 | ), | ||
195 | |||
196 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
197 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
198 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
199 | ); | ||
200 | |||
201 | TRACE_EVENT(block_bio_complete, | ||
202 | |||
203 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
204 | |||
205 | TP_ARGS(q, bio), | ||
206 | |||
207 | TP_STRUCT__entry( | ||
208 | __field( dev_t, dev ) | ||
209 | __field( sector_t, sector ) | ||
210 | __field( unsigned, nr_sector ) | ||
211 | __field( int, error ) | ||
212 | __array( char, rwbs, 6 ) | ||
213 | ), | ||
214 | |||
215 | TP_fast_assign( | ||
216 | __entry->dev = bio->bi_bdev->bd_dev; | ||
217 | __entry->sector = bio->bi_sector; | ||
218 | __entry->nr_sector = bio->bi_size >> 9; | ||
219 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
220 | ), | ||
221 | |||
222 | TP_printk("%d,%d %s %llu + %u [%d]", | ||
223 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
224 | __entry->sector, __entry->nr_sector, __entry->error) | ||
225 | ); | ||
226 | |||
227 | TRACE_EVENT(block_bio_backmerge, | ||
228 | |||
229 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
230 | |||
231 | TP_ARGS(q, bio), | ||
232 | |||
233 | TP_STRUCT__entry( | ||
234 | __field( dev_t, dev ) | ||
235 | __field( sector_t, sector ) | ||
236 | __field( unsigned int, nr_sector ) | ||
237 | __array( char, rwbs, 6 ) | ||
238 | __array( char, comm, TASK_COMM_LEN ) | ||
239 | ), | ||
240 | |||
241 | TP_fast_assign( | ||
242 | __entry->dev = bio->bi_bdev->bd_dev; | ||
243 | __entry->sector = bio->bi_sector; | ||
244 | __entry->nr_sector = bio->bi_size >> 9; | ||
245 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
246 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
247 | ), | ||
248 | |||
249 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
250 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
251 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
252 | ); | ||
253 | |||
254 | TRACE_EVENT(block_bio_frontmerge, | ||
255 | |||
256 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
257 | |||
258 | TP_ARGS(q, bio), | ||
259 | |||
260 | TP_STRUCT__entry( | ||
261 | __field( dev_t, dev ) | ||
262 | __field( sector_t, sector ) | ||
263 | __field( unsigned, nr_sector ) | ||
264 | __array( char, rwbs, 6 ) | ||
265 | __array( char, comm, TASK_COMM_LEN ) | ||
266 | ), | ||
267 | |||
268 | TP_fast_assign( | ||
269 | __entry->dev = bio->bi_bdev->bd_dev; | ||
270 | __entry->sector = bio->bi_sector; | ||
271 | __entry->nr_sector = bio->bi_size >> 9; | ||
272 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
273 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
274 | ), | ||
275 | |||
276 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
277 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
278 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
279 | ); | ||
280 | |||
281 | TRACE_EVENT(block_bio_queue, | ||
282 | |||
283 | TP_PROTO(struct request_queue *q, struct bio *bio), | ||
284 | |||
285 | TP_ARGS(q, bio), | ||
286 | |||
287 | TP_STRUCT__entry( | ||
288 | __field( dev_t, dev ) | ||
289 | __field( sector_t, sector ) | ||
290 | __field( unsigned int, nr_sector ) | ||
291 | __array( char, rwbs, 6 ) | ||
292 | __array( char, comm, TASK_COMM_LEN ) | ||
293 | ), | ||
294 | |||
295 | TP_fast_assign( | ||
296 | __entry->dev = bio->bi_bdev->bd_dev; | ||
297 | __entry->sector = bio->bi_sector; | ||
298 | __entry->nr_sector = bio->bi_size >> 9; | ||
299 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
300 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
301 | ), | ||
302 | |||
303 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
304 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
305 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
306 | ); | ||
307 | |||
308 | TRACE_EVENT(block_getrq, | ||
309 | |||
310 | TP_PROTO(struct request_queue *q, struct bio *bio, int rw), | ||
311 | |||
312 | TP_ARGS(q, bio, rw), | ||
313 | |||
314 | TP_STRUCT__entry( | ||
315 | __field( dev_t, dev ) | ||
316 | __field( sector_t, sector ) | ||
317 | __field( unsigned int, nr_sector ) | ||
318 | __array( char, rwbs, 6 ) | ||
319 | __array( char, comm, TASK_COMM_LEN ) | ||
320 | ), | ||
321 | |||
322 | TP_fast_assign( | ||
323 | __entry->dev = bio ? bio->bi_bdev->bd_dev : 0; | ||
324 | __entry->sector = bio ? bio->bi_sector : 0; | ||
325 | __entry->nr_sector = bio ? bio->bi_size >> 9 : 0; | ||
326 | blk_fill_rwbs(__entry->rwbs, | ||
327 | bio ? bio->bi_rw : 0, __entry->nr_sector); | ||
328 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
329 | ), | ||
330 | |||
331 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
332 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
333 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
334 | ); | ||
335 | |||
336 | TRACE_EVENT(block_sleeprq, | ||
337 | |||
338 | TP_PROTO(struct request_queue *q, struct bio *bio, int rw), | ||
339 | |||
340 | TP_ARGS(q, bio, rw), | ||
341 | |||
342 | TP_STRUCT__entry( | ||
343 | __field( dev_t, dev ) | ||
344 | __field( sector_t, sector ) | ||
345 | __field( unsigned int, nr_sector ) | ||
346 | __array( char, rwbs, 6 ) | ||
347 | __array( char, comm, TASK_COMM_LEN ) | ||
348 | ), | ||
349 | |||
350 | TP_fast_assign( | ||
351 | __entry->dev = bio ? bio->bi_bdev->bd_dev : 0; | ||
352 | __entry->sector = bio ? bio->bi_sector : 0; | ||
353 | __entry->nr_sector = bio ? bio->bi_size >> 9 : 0; | ||
354 | blk_fill_rwbs(__entry->rwbs, | ||
355 | bio ? bio->bi_rw : 0, __entry->nr_sector); | ||
356 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
357 | ), | ||
358 | |||
359 | TP_printk("%d,%d %s %llu + %u [%s]", | ||
360 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
361 | __entry->sector, __entry->nr_sector, __entry->comm) | ||
362 | ); | ||
363 | |||
364 | TRACE_EVENT(block_plug, | ||
365 | |||
366 | TP_PROTO(struct request_queue *q), | ||
367 | |||
368 | TP_ARGS(q), | ||
369 | |||
370 | TP_STRUCT__entry( | ||
371 | __array( char, comm, TASK_COMM_LEN ) | ||
372 | ), | ||
373 | |||
374 | TP_fast_assign( | ||
375 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
376 | ), | ||
377 | |||
378 | TP_printk("[%s]", __entry->comm) | ||
379 | ); | ||
380 | |||
381 | TRACE_EVENT(block_unplug_timer, | ||
382 | |||
383 | TP_PROTO(struct request_queue *q), | ||
384 | |||
385 | TP_ARGS(q), | ||
386 | |||
387 | TP_STRUCT__entry( | ||
388 | __field( int, nr_rq ) | ||
389 | __array( char, comm, TASK_COMM_LEN ) | ||
390 | ), | ||
391 | |||
392 | TP_fast_assign( | ||
393 | __entry->nr_rq = q->rq.count[READ] + q->rq.count[WRITE]; | ||
394 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
395 | ), | ||
396 | |||
397 | TP_printk("[%s] %d", __entry->comm, __entry->nr_rq) | ||
398 | ); | ||
399 | |||
400 | TRACE_EVENT(block_unplug_io, | ||
401 | |||
402 | TP_PROTO(struct request_queue *q), | ||
403 | |||
404 | TP_ARGS(q), | ||
405 | |||
406 | TP_STRUCT__entry( | ||
407 | __field( int, nr_rq ) | ||
408 | __array( char, comm, TASK_COMM_LEN ) | ||
409 | ), | ||
410 | |||
411 | TP_fast_assign( | ||
412 | __entry->nr_rq = q->rq.count[READ] + q->rq.count[WRITE]; | ||
413 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
414 | ), | ||
415 | |||
416 | TP_printk("[%s] %d", __entry->comm, __entry->nr_rq) | ||
417 | ); | ||
418 | |||
419 | TRACE_EVENT(block_split, | ||
420 | |||
421 | TP_PROTO(struct request_queue *q, struct bio *bio, | ||
422 | unsigned int new_sector), | ||
423 | |||
424 | TP_ARGS(q, bio, new_sector), | ||
425 | |||
426 | TP_STRUCT__entry( | ||
427 | __field( dev_t, dev ) | ||
428 | __field( sector_t, sector ) | ||
429 | __field( sector_t, new_sector ) | ||
430 | __array( char, rwbs, 6 ) | ||
431 | __array( char, comm, TASK_COMM_LEN ) | ||
432 | ), | ||
433 | |||
434 | TP_fast_assign( | ||
435 | __entry->dev = bio->bi_bdev->bd_dev; | ||
436 | __entry->sector = bio->bi_sector; | ||
437 | __entry->new_sector = new_sector; | ||
438 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
439 | memcpy(__entry->comm, current->comm, TASK_COMM_LEN); | ||
440 | ), | ||
441 | |||
442 | TP_printk("%d,%d %s %llu / %llu [%s]", | ||
443 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
444 | __entry->sector, __entry->new_sector, __entry->comm) | ||
445 | ); | ||
446 | |||
447 | TRACE_EVENT(block_remap, | ||
448 | |||
449 | TP_PROTO(struct request_queue *q, struct bio *bio, dev_t dev, | ||
450 | sector_t from), | ||
451 | |||
452 | TP_ARGS(q, bio, dev, from), | ||
453 | |||
454 | TP_STRUCT__entry( | ||
455 | __field( dev_t, dev ) | ||
456 | __field( sector_t, sector ) | ||
457 | __field( unsigned int, nr_sector ) | ||
458 | __field( dev_t, old_dev ) | ||
459 | __field( sector_t, old_sector ) | ||
460 | __array( char, rwbs, 6 ) | ||
461 | ), | ||
462 | |||
463 | TP_fast_assign( | ||
464 | __entry->dev = bio->bi_bdev->bd_dev; | ||
465 | __entry->sector = bio->bi_sector; | ||
466 | __entry->nr_sector = bio->bi_size >> 9; | ||
467 | __entry->old_dev = dev; | ||
468 | __entry->old_sector = from; | ||
469 | blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size); | ||
470 | ), | ||
471 | |||
472 | TP_printk("%d,%d %s %llu + %u <- (%d,%d) %llu", | ||
473 | MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs, | ||
474 | __entry->sector, __entry->nr_sector, | ||
475 | MAJOR(__entry->old_dev), MINOR(__entry->old_dev), | ||
476 | __entry->old_sector) | ||
477 | ); | ||
478 | |||
479 | #endif /* _TRACE_BLOCK_H */ | ||
480 | |||
481 | /* This part must be outside protection */ | ||
482 | #include <trace/define_trace.h> | ||
483 | |||
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 06b85850fab4..844164dca90a 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile | |||
@@ -45,7 +45,10 @@ obj-$(CONFIG_HW_BRANCH_TRACER) += trace_hw_branches.o | |||
45 | obj-$(CONFIG_POWER_TRACER) += trace_power.o | 45 | obj-$(CONFIG_POWER_TRACER) += trace_power.o |
46 | obj-$(CONFIG_KMEMTRACE) += kmemtrace.o | 46 | obj-$(CONFIG_KMEMTRACE) += kmemtrace.o |
47 | obj-$(CONFIG_WORKQUEUE_TRACER) += trace_workqueue.o | 47 | obj-$(CONFIG_WORKQUEUE_TRACER) += trace_workqueue.o |
48 | obj-$(CONFIG_BLK_DEV_IO_TRACE) += blktrace.o | 48 | obj-$(CONFIG_BLK_DEV_IO_TRACE) += blktrace.o |
49 | ifeq ($(CONFIG_BLOCK),y) | ||
50 | obj-$(CONFIG_EVENT_TRACING) += blktrace.o | ||
51 | endif | ||
49 | obj-$(CONFIG_EVENT_TRACING) += trace_events.o | 52 | obj-$(CONFIG_EVENT_TRACING) += trace_events.o |
50 | obj-$(CONFIG_EVENT_TRACING) += trace_export.o | 53 | obj-$(CONFIG_EVENT_TRACING) += trace_export.o |
51 | obj-$(CONFIG_FTRACE_SYSCALLS) += trace_syscalls.o | 54 | obj-$(CONFIG_FTRACE_SYSCALLS) += trace_syscalls.o |
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index e3abf55bc8e5..7bd6a9893c24 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c | |||
@@ -23,10 +23,14 @@ | |||
23 | #include <linux/mutex.h> | 23 | #include <linux/mutex.h> |
24 | #include <linux/debugfs.h> | 24 | #include <linux/debugfs.h> |
25 | #include <linux/time.h> | 25 | #include <linux/time.h> |
26 | #include <trace/block.h> | ||
27 | #include <linux/uaccess.h> | 26 | #include <linux/uaccess.h> |
27 | |||
28 | #include <trace/events/block.h> | ||
29 | |||
28 | #include "trace_output.h" | 30 | #include "trace_output.h" |
29 | 31 | ||
32 | #ifdef CONFIG_BLK_DEV_IO_TRACE | ||
33 | |||
30 | static unsigned int blktrace_seq __read_mostly = 1; | 34 | static unsigned int blktrace_seq __read_mostly = 1; |
31 | 35 | ||
32 | static struct trace_array *blk_tr; | 36 | static struct trace_array *blk_tr; |
@@ -1658,3 +1662,75 @@ int blk_trace_init_sysfs(struct device *dev) | |||
1658 | return sysfs_create_group(&dev->kobj, &blk_trace_attr_group); | 1662 | return sysfs_create_group(&dev->kobj, &blk_trace_attr_group); |
1659 | } | 1663 | } |
1660 | 1664 | ||
1665 | #endif /* CONFIG_BLK_DEV_IO_TRACE */ | ||
1666 | |||
1667 | #ifdef CONFIG_EVENT_TRACING | ||
1668 | |||
1669 | void blk_dump_cmd(char *buf, struct request *rq) | ||
1670 | { | ||
1671 | int i, end; | ||
1672 | int len = rq->cmd_len; | ||
1673 | unsigned char *cmd = rq->cmd; | ||
1674 | |||
1675 | if (!blk_pc_request(rq)) { | ||
1676 | buf[0] = '\0'; | ||
1677 | return; | ||
1678 | } | ||
1679 | |||
1680 | for (end = len - 1; end >= 0; end--) | ||
1681 | if (cmd[end]) | ||
1682 | break; | ||
1683 | end++; | ||
1684 | |||
1685 | for (i = 0; i < len; i++) { | ||
1686 | buf += sprintf(buf, "%s%02x", i == 0 ? "" : " ", cmd[i]); | ||
1687 | if (i == end && end != len - 1) { | ||
1688 | sprintf(buf, " .."); | ||
1689 | break; | ||
1690 | } | ||
1691 | } | ||
1692 | } | ||
1693 | |||
1694 | void blk_fill_rwbs(char *rwbs, u32 rw, int bytes) | ||
1695 | { | ||
1696 | int i = 0; | ||
1697 | |||
1698 | if (rw & WRITE) | ||
1699 | rwbs[i++] = 'W'; | ||
1700 | else if (rw & 1 << BIO_RW_DISCARD) | ||
1701 | rwbs[i++] = 'D'; | ||
1702 | else if (bytes) | ||
1703 | rwbs[i++] = 'R'; | ||
1704 | else | ||
1705 | rwbs[i++] = 'N'; | ||
1706 | |||
1707 | if (rw & 1 << BIO_RW_AHEAD) | ||
1708 | rwbs[i++] = 'A'; | ||
1709 | if (rw & 1 << BIO_RW_BARRIER) | ||
1710 | rwbs[i++] = 'B'; | ||
1711 | if (rw & 1 << BIO_RW_SYNCIO) | ||
1712 | rwbs[i++] = 'S'; | ||
1713 | if (rw & 1 << BIO_RW_META) | ||
1714 | rwbs[i++] = 'M'; | ||
1715 | |||
1716 | rwbs[i] = '\0'; | ||
1717 | } | ||
1718 | |||
1719 | void blk_fill_rwbs_rq(char *rwbs, struct request *rq) | ||
1720 | { | ||
1721 | int rw = rq->cmd_flags & 0x03; | ||
1722 | int bytes; | ||
1723 | |||
1724 | if (blk_discard_rq(rq)) | ||
1725 | rw |= (1 << BIO_RW_DISCARD); | ||
1726 | |||
1727 | if (blk_pc_request(rq)) | ||
1728 | bytes = rq->data_len; | ||
1729 | else | ||
1730 | bytes = rq->hard_nr_sectors << 9; | ||
1731 | |||
1732 | blk_fill_rwbs(rwbs, rw, bytes); | ||
1733 | } | ||
1734 | |||
1735 | #endif /* CONFIG_EVENT_TRACING */ | ||
1736 | |||
diff --git a/mm/bounce.c b/mm/bounce.c index e590272fe7a8..65f5e17e411a 100644 --- a/mm/bounce.c +++ b/mm/bounce.c | |||
@@ -14,16 +14,15 @@ | |||
14 | #include <linux/hash.h> | 14 | #include <linux/hash.h> |
15 | #include <linux/highmem.h> | 15 | #include <linux/highmem.h> |
16 | #include <linux/blktrace_api.h> | 16 | #include <linux/blktrace_api.h> |
17 | #include <trace/block.h> | ||
18 | #include <asm/tlbflush.h> | 17 | #include <asm/tlbflush.h> |
19 | 18 | ||
19 | #include <trace/events/block.h> | ||
20 | |||
20 | #define POOL_SIZE 64 | 21 | #define POOL_SIZE 64 |
21 | #define ISA_POOL_SIZE 16 | 22 | #define ISA_POOL_SIZE 16 |
22 | 23 | ||
23 | static mempool_t *page_pool, *isa_page_pool; | 24 | static mempool_t *page_pool, *isa_page_pool; |
24 | 25 | ||
25 | DEFINE_TRACE(block_bio_bounce); | ||
26 | |||
27 | #ifdef CONFIG_HIGHMEM | 26 | #ifdef CONFIG_HIGHMEM |
28 | static __init int init_emergency_pool(void) | 27 | static __init int init_emergency_pool(void) |
29 | { | 28 | { |