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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
1 - LOOPBACK DRIVER
The driver implements the main logic of the loopback test and provides
sysfs files to configure the test and retrieve the results.
A user could run a test without the need of the test application given
that he understands the sysfs interface of the loopback driver.
The loopback kernel driver needs to be loaded and at least one module
with the loopback feature enabled must be present for the sysfs files to be
created and for the loopback test application to be able to run.
To load the module:
# modprobe gb-loopback
When the module is probed, New files are available on the sysfs
directory of the detected loopback device.
(typically under "/sys/bus/graybus/devices").
Here is a short summary of the sysfs interface files that should be visible:
* Loopback Configuration Files:
async - Use asynchronous operations.
iteration_max - Number of tests iterations to perform.
size - payload size of the transfer.
timeout - The number of microseconds to give an individual
asynchronous request before timing out.
us_wait - Time to wait between 2 messages
type - By writing the test type to this file, the test starts.
Valid tests are:
0 stop the test
2 - ping
3 - transfer
4 - sink
* Loopback feedback files:
error - number of errors that have occurred.
iteration_count - Number of iterations performed.
requests_completed - Number of requests successfully completed.
requests_timedout - Number of requests that have timed out.
timeout_max - Max allowed timeout
timeout_min - Min allowed timeout.
* Loopback result files:
apbridge_unipro_latency_avg
apbridge_unipro_latency_max
apbridge_unipro_latency_min
gpbridge_firmware_latency_avg
gpbridge_firmware_latency_max
gpbridge_firmware_latency_min
requests_per_second_avg
requests_per_second_max
requests_per_second_min
latency_avg
latency_max
latency_min
throughput_avg
throughput_max
throughput_min
2 - LOOPBACK TEST APPLICATION
The loopback test application manages and formats the results provided by
the loopback kernel module. The purpose of this application
is to:
- Start and manage multiple loopback device tests concurrently.
- Calculate the aggregate results for multiple devices.
- Gather and format test results (csv or human readable).
The best way to get up to date usage information for the application is
usually to pass the "-h" parameter.
Here is the summary of the available options:
Mandatory arguments
-t must be one of the test names - sink, transfer or ping
-i iteration count - the number of iterations to run the test over
Optional arguments
-S sysfs location - location for greybus 'endo' entries default /sys/bus/greybus/devices/
-D debugfs location - location for loopback debugfs entries default /sys/kernel/debug/gb_loopback/
-s size of data packet to send during test - defaults to zero
-m mask - a bit mask of connections to include example: -m 8 = 4th connection -m 9 = 1st and 4th connection etc
default is zero which means broadcast to all connections
-v verbose output
-d debug output
-r raw data output - when specified the full list of latency values are included in the output CSV
-p porcelain - when specified printout is in a user-friendly non-CSV format. This option suppresses writing to CSV file
-a aggregate - show aggregation of all enabled devies
-l list found loopback devices and exit.
-x Async - Enable async transfers.
-o Timeout - Timeout in microseconds for async operations.
3 - REAL WORLD EXAMPLE USAGES
3.1 - Using the driver sysfs files to run a test on a single device:
* Run a 1000 transfers of a 100 byte packet. Each transfer is started only
after the previous one finished successfully:
echo 0 > /sys/bus/greybus/devices/1-2.17/type
echo 0 > /sys/bus/greybus/devices/1-2.17/async
echo 2000 > /sys/bus/greybus/devices/1-2.17/us_wait
echo 100 > /sys/bus/greybus/devices/1-2.17/size
echo 1000 > /sys/bus/greybus/devices/1-2.17/iteration_max
echo 0 > /sys/bus/greybus/devices/1-2.17/mask
echo 200000 > /sys/bus/greybus/devices/1-2.17/timeout
echo 3 > /sys/bus/greybus/devices/1-2.17/type
* Run a 1000 transfers of a 100 byte packet. Transfers are started without
waiting for the previous one to finish:
echo 0 > /sys/bus/greybus/devices/1-2.17/type
echo 3 > /sys/bus/greybus/devices/1-2.17/async
echo 0 > /sys/bus/greybus/devices/1-2.17/us_wait
echo 100 > /sys/bus/greybus/devices/1-2.17/size
echo 1000 > /sys/bus/greybus/devices/1-2.17/iteration_max
echo 0 > /sys/bus/greybus/devices/1-2.17/mask
echo 200000 > /sys/bus/greybus/devices/1-2.17/timeout
echo 3 > /sys/bus/greybus/devices/1-2.17/type
* Read the results from sysfs:
cat /sys/bus/greybus/devices/1-2.17/requests_per_second_min
cat /sys/bus/greybus/devices/1-2.17/requests_per_second_max
cat /sys/bus/greybus/devices/1-2.17/requests_per_second_avg
cat /sys/bus/greybus/devices/1-2.17/latency_min
cat /sys/bus/greybus/devices/1-2.17/latency_max
cat /sys/bus/greybus/devices/1-2.17/latency_avg
cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_min
cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_max
cat /sys/bus/greybus/devices/1-2.17/apbridge_unipro_latency_avg
cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_min
cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_max
cat /sys/bus/greybus/devices/1-2.17/gpbridge_firmware_latency_avg
cat /sys/bus/greybus/devices/1-2.17/error
cat /sys/bus/greybus/devices/1-2.17/requests_completed
cat /sys/bus/greybus/devices/1-2.17/requests_timedout
3.2 - using the test application:
* Run a transfer test 10 iterations of size 100 bytes on all available devices
#/loopback_test -t transfer -i 10 -s 100
1970-1-1 0:10:7,transfer,1-4.17,100,10,0,443,509,471.700012,66,1963,2256,2124.600098,293,102776,118088,109318.898438,15312,1620,1998,1894.099976,378,56,57,56.799999,1
1970-1-1 0:10:7,transfer,1-5.17,100,10,0,399,542,463.399994,143,1845,2505,2175.800049,660,92568,125744,107393.296875,33176,1469,2305,1806.500000,836,56,57,56.799999,1
* Show the aggregate results of both devices. ("-a")
#/loopback_test -t transfer -i 10 -s 100 -a
1970-1-1 0:10:35,transfer,1-4.17,100,10,0,448,580,494.100006,132,1722,2230,2039.400024,508,103936,134560,114515.703125,30624,1513,1980,1806.900024,467,56,57,57.299999,1
1970-1-1 0:10:35,transfer,1-5.17,100,10,0,383,558,478.600006,175,1791,2606,2115.199951,815,88856,129456,110919.703125,40600,1457,2246,1773.599976,789,56,57,57.099998,1
1970-1-1 0:10:35,transfer,aggregate,100,10,0,383,580,486.000000,197,1722,2606,2077.000000,884,88856,134560,112717.000000,45704,1457,2246,1789.000000,789,56,57,57.000000,1
* Example usage of the mask option to select which devices will
run the test (1st, 2nd, or both devices):
# /loopback_test -t transfer -i 10 -s 100 -m 1
1970-1-1 0:11:56,transfer,1-4.17,100,10,0,514,558,544.900024,44,1791,1943,1836.599976,152,119248,129456,126301.296875,10208,1600,1001609,101613.601562,1000009,56,57,56.900002,1
# /loopback_test -t transfer -i 10 -s 100 -m 2
1970-1-1 0:12:0,transfer,1-5.17,100,10,0,468,554,539.000000,86,1804,2134,1859.500000,330,108576,128528,124932.500000,19952,1606,1626,1619.300049,20,56,57,57.400002,1
# /loopback_test -t transfer -i 10 -s 100 -m 3
1970-1-1 0:12:3,transfer,1-4.17,100,10,0,432,510,469.399994,78,1959,2313,2135.800049,354,100224,118320,108785.296875,18096,1610,2024,1893.500000,414,56,57,57.200001,1
1970-1-1 0:12:3,transfer,1-5.17,100,10,0,404,542,468.799988,138,1843,2472,2152.500000,629,93728,125744,108646.101562,32016,1504,2247,1853.099976,743,56,57,57.099998,1
* Show output in human readable format ("-p")
# /loopback_test -t transfer -i 10 -s 100 -m 3 -p
1970-1-1 0:12:37
test: transfer
path: 1-4.17
size: 100
iterations: 10
errors: 0
async: Disabled
requests per-sec: min=390, max=547, average=469.299988, jitter=157
ap-throughput B/s: min=90480 max=126904 average=108762.101562 jitter=36424
ap-latency usec: min=1826 max=2560 average=2146.000000 jitter=734
apbridge-latency usec: min=1620 max=1982 average=1882.099976 jitter=362
gpbridge-latency usec: min=56 max=57 average=57.099998 jitter=1
1970-1-1 0:12:37
test: transfer
path: 1-5.17
size: 100
iterations: 10
errors: 0
async: Disabled
requests per-sec: min=397, max=538, average=461.700012, jitter=141
ap-throughput B/s: min=92104 max=124816 average=106998.898438 jitter=32712
ap-latency usec: min=1856 max=2514 average=2185.699951 jitter=658
apbridge-latency usec: min=1460 max=2296 average=1828.599976 jitter=836
gpbridge-latency usec: min=56 max=57 average=57.099998 jitter=1
|
