1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
|
#!/bin/sh
# This function just re-writes the timestamp of the strace entries to be
# seconds.usecs since boot. To match the timestamping of ftrace (so we can
# merge them later).
process_strace()
{
strace=$1
# parse in data/system/vendor and parse out /sys/devices/system/...
egrep '\/system\/|\/data\/|\/vendor\/' $strace | egrep -v '\/sys\/devices\/system\/' > bar
fgrep -v '= -1' bar > foo
mv foo bar
# begin_time is seconds since epoch
begin_time=`cat trace.begin`
# replace seconds since epoch with SECONDS SINCE BOOT in the
# strace files
awk -v begin="$begin_time" '{ printf "%f strace ", $1 - begin; $1=""; print $0}' bar > $2
rm bar
}
#
# This function processes the ftrace file, removing the fields that we don't care
# about, breaks up the ftrace file into one file per pid.
# Input : One single fstrace file.
# Output : Multiple fstrace.pid files.
prep_fstrace()
{
# Remove leading junk
fgrep f2fs_data $1 | sed 's/^.* \[.*\] //' | sed s/://g | sed s/,//g > foo
sed 's/f2fs_dataread_start/read/' foo > bar1
mv bar1 bar
# First column is timestamp SECONDS SINCE BOOT
awk '{ print $2, "ftrace", $3, $5, $7, $9, $13 }' bar > foo
rm bar
# Get all the uniq pids
awk '{print $7}' foo | sort | uniq > pidlist
for i in `cat pidlist`
do
awk -v pid=$i '{ if (pid == $7) print $0}' foo > fstrace.$i
done
rm pidlist
rm foo
}
# Merge straces and ftraces.
# The goal here is to catch mmap'ed IO (reads) that won't be in the
# strace file. The algorithm is to look for mmaps in the strace file,
# use the files tha are mmap'ed to search in the ftraces to pick up
# tracepoints from there, and merge those with the straces.
# The output of this function is a set of parsed_input_trace.<pid>
# files, that can then be compiled into .wl files
merge_compile()
{
for stracefile in trace.*
do
if [ $stracefile == trace.begin ] || [ $stracefile == trace.tar ];
then
continue
fi
# Get the pid from the strace filename (pid is the extension)
pid=${stracefile##*.}
process_strace $stracefile foo.$pid
if ! [ -s foo.$pid ]; then
rm foo.$pid
continue
fi
#
# If we have matching strace and ftrace files, then look for mmaps in
# the strace pluck the corresponding entries for the mmap (mmaped IO)
# from the ftrace and merge them into the strace
#
if [ -f fstrace.$pid ]; then
fgrep mmap foo.$pid > bar
if [ -s bar ]; then
# Get all the unique mmap'ed filenames from the strace
awk '{ print $7 }' bar | sed 's/^[^<]*<//g' | sed 's/>,//g' > mapped_files
# Pluck all the lines from the ftrace corresponding to the mmaps
cat /dev/null > footemp
for j in `sort mapped_files | uniq`
do
# Merge the readpage(s) traces from the ftrace into strace
# for this mmaped file.
grep -w $j fstrace.$pid > foobar
if [ $? == 0 ]; then
sort foo.$pid foobar >> footemp
fi
rm foobar
done
rm mapped_files
if [ -s footemp ]; then
mv footemp parsed_input_trace.$pid
else
mv foo.$pid parsed_input_trace.$pid
fi
else
mv foo.$pid parsed_input_trace.$pid
fi
rm bar
else
mv foo.$pid parsed_input_trace.$pid
fi
echo compiling parsed_input_trace.$pid
compile_ioshark parsed_input_trace.$pid $pid.wl
rm parsed_input_trace.$pid
rm -f foo.$pid
done
}
# main() starts here
rm -f *.wl
rm -f parsed*
rm -f ioshark_filenames
# Pre-process the ftrace file
prep_fstrace fstrace
# Merge the ftrace file(s) with the strace files
merge_compile
# tar up the .wl files just created
tar cf wl-test.tar ioshark_filenames *.wl
|
