README.md 11.7 KB
Newer Older
Grant Likely's avatar
Grant Likely committed
1
# EDK2 SCT Results Parser
2

Vincent Stehlé's avatar
Vincent Stehlé committed
3
This is an external parser script for [UEFI SCT]. (WIP)
4

Vincent Stehlé's avatar
Vincent Stehlé committed
5
6
It's designed to read a `.ekl` results log from an [UEFI SCT] run, and a
generated `.seq` from [UEFI SCT] configurator.
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
7

8
It will proceed to generate a Markdown file listing number of failures, passes, each test from the sequence file set that was Silently dropped, and a list of all failures and warnings.
9

Vincent Stehlé's avatar
Vincent Stehlé committed
10
[UEFI SCT]: https://uefi.org/testtools
11

Grant Likely's avatar
Grant Likely committed
12
13
14
15
16
17
18
19
## Quick Start

If you're using this tool to analyze EBBR test results, use the following
command. The parsed report can be found in `result.md`.

``` {.sh}
$ ./parser.py --config EBBR.yaml \
		</path/to/sct_results/Overall/Summary.ekl> \
20
		contrib/v21.07_0.9_BETA/EBBR.seq
Grant Likely's avatar
Grant Likely committed
21
22
23
24
INFO apply_rules: Updated 200 test(s) out of 12206 after applying 124 rule(s)
INFO main: 0 dropped(s), 1 failure(s), 93 ignored(s), 106 known u-boot limitation(s), 12006 pass(s), 0 warning(s)
```

25
## Usage
26
27
28
29
30
31
32

Usage to generate a `result.md` is such:

``` {.sh}
$ python3 parser.py <log_file.ekl> <seq_file.seq>
```

33
The output filename can be specified with the `--md` option:
34
35
36
37

``` {.sh}
$ ./parser.py --md out.md ...
```
Jeffrey Booher-Kaeding's avatar
Jeffrey Booher-Kaeding committed
38

Vincent Stehlé's avatar
Vincent Stehlé committed
39
An online help is available with the `-h` option.
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
40

41
42
43
44
45
46
The generated `result md` can be easily converted to HTML using `pandoc` with:

``` {.sh}
$ pandoc -oresult.html result.md
```

47
### Custom search
Vincent Stehlé's avatar
Vincent Stehlé committed
48
For a custom Key:value search, the next two arguments *MUST be included together.* The program will search and display files that met that constraint, without the crosscheck, and display the names, guid, and key:value to the command line. `python3 parser.py <file.ekl> <file.seq> <search key> <search value>`
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
49

Vincent Stehlé's avatar
Vincent Stehlé committed
50
you can use the `test_dict` below to see available keys.
51

Vincent Stehlé's avatar
Vincent Stehlé committed
52
### Sorting data
53

Vincent Stehlé's avatar
Vincent Stehlé committed
54
55
56
It is possible to sort the tests data before output using
the `--sort <key1,key2,...>` option.
Sorting the test data helps when comparing results with diff.
57

Vincent Stehlé's avatar
Vincent Stehlé committed
58
Example command:
59

Vincent Stehlé's avatar
Vincent Stehlé committed
60
61
62
63
``` {.sh}
$ ./parser.py --sort \
      'group,descr,set guid,test set,sub set,guid,name,log' ...
```
64

65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
### Filtering data

The `--filter` option allows to specify a python3 expression, which is used as a
filter. The expression is evaluated for each test; if it evaluates to True the
test is kept, otherwise it is omitted. The expression has access to the test
as dict "x".

Example command, which keeps only the failed tests:

``` {.sh}
$ ./parser.py --filter "x['result'] == 'FAILURE'" ...
```

Filtering takes place after the configuration rules, which are described below.

80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
This filtering mechanism can also be (ab)used to transform tests results.

Example command, which adds a "comment" field (and keeps all the tests):

``` {.sh}
$ ./parser.py \
      --filter "x.update({'comment': 'Transformed'}) or True" ...
```

Example command, which removes filenames prefixes in the "log" field (and keeps
all the tests):

``` {.sh}
$ ./parser.py \
      --filter "x.update({'log': re.sub(r'/.*/', '', x['log'])}) \
                or True" ...
```

98
99
100
101
102
103
104
105
106
107
108
### Keeping only certain fields

Except for the markdown and the config template formats, it is possible to
specify which tests data fields to actually write using the `--fields` option.

Example command, suitable for triaging:

``` {.sh}
$ ./parser.py --fields 'result,sub set,descr,name,log' ...
```

109
The csv format and printing to stdout can retain the fields order.
110

111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
### Collapsing duplicates

It is possible to "collapse" duplicate tests data into a single entry using the
`--uniq` option, much similar in principle to the UNIX `uniq -c` command.

This step happens after tests and fields filtering, and it adds a "count" field.

Example command, suitable for triaging:

``` {.sh}
$ ./parser.py \
      --filter "x.update({'log': re.sub(r'/.*/', '', x['log'])}) \
                or x['result'] != 'PASS'" \
      --fields 'count,result,sub set,descr,name,log' --uniq ...
```

127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
### Printing a summary

It is possible to print results to stdout using the `--print` option. This is
more useful when only few fields are printed. Example command:

Example printing command:

``` {.sh}
$ ./parser.py --fields 'result,name' --print ...
```

More condensed summaries can be obtained with further filtering.

Example summary command:

``` {.sh}
$ ./parser.py \
      --filter "x.update({'log': re.sub(r'/.*/', '', x['log'])}) \
                or x['result'] in ['FAILURE', 'WARNING']" \
      --fields 'count,result,name' --uniq --print ...
```

149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
### Re-reading markdown results

It is possible to re-read a previously generated markdown results file with the
`--input-md` option. This can be useful to perform further processing on the
tests.

Example command to read a previously generated markdown:

``` {.sh}
$ ./parser.py --input-md 'result.md' ...
```

* By default an output markdown is still generated, except in the case where the
  input and output markdown have the same filename.
* The generated markdown results do not contain the "passed" tests. They can
  therefore not be re-read.

Vincent Stehlé's avatar
Vincent Stehlé committed
166
167
168
169
170
171
172
## Configuration file

It is possible to use a configuration file with command line option `--config
<filename>`.
This configuration file describes operations to perform on the tests results,
such as marking tests as false positives or waiving failures.

Vincent Stehlé's avatar
Vincent Stehlé committed
173
Example command for [EBBR]:
Vincent Stehlé's avatar
Vincent Stehlé committed
174
175
176
177
178

``` {.sh}
$ ./parser.py --config EBBR.yaml /path/to/Summary.ekl EBBR.seq ...
```

Vincent Stehlé's avatar
Vincent Stehlé committed
179
180
181
182
You need to install the [PyYAML] module for this to work.

[EBBR]: https://github.com/ARM-software/ebbr
[PyYAML]: https://github.com/yaml/pyyaml
Vincent Stehlé's avatar
Vincent Stehlé committed
183
184
185

### Configuration file format

Vincent Stehlé's avatar
Vincent Stehlé committed
186
The configuration file is in [YAML] format. It contains a list of rules:
Vincent Stehlé's avatar
Vincent Stehlé committed
187
188
189
190
191
192
193
194
195
196
197
198
199
200

``` {.yaml}
- rule: name/description (optional)
  criteria:
    key1: value1
    key2: value2
    ...
  update:
    key3: value3
    key4: value4
    ...
- rule...
```

Vincent Stehlé's avatar
Vincent Stehlé committed
201
202
[YAML]: https://yaml.org

Vincent Stehlé's avatar
Vincent Stehlé committed
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
### Rule processing

The rules will be applied to each test one by one in the following manner:

* An attempt is made at matching all the keys/values of the rule's 'criteria'
  dict to the keys/values of the test dict. Matching test and criteria is done
  with a "relaxed" comparison (more below).
  - If there is no match, processing moves on to the next rule.
  - If there is a match:
    1. The test dict is updated with the 'update' dict of the rule.
    2. An 'Updated by' key is set in the test dict to the rule name.
    3. Finally, no more rule is applied to that test.

A test value and a criteria value match if the criteria value string is present
anywhere in the test value string.
For example, the test value "abcde" matches the criteria value "cd".

You can use `--debug` to see more details about which rules are applied to the
tests.

### Sample

225
226
In the folder `sample`, a `sample.yaml` configuration file is provided as
example, to use with the `sample.ekl` and `sample.seq` files.
Vincent Stehlé's avatar
Vincent Stehlé committed
227
228
229
230

Try it with:

``` {.sh}
231
$ ./parser.py --config sample/sample.yaml sample/sample.ekl sample/sample.seq
Vincent Stehlé's avatar
Vincent Stehlé committed
232
233
```

234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
### Generating a configuration template

To ease the writing of yaml configurations, there is a `--template` option to
generate a configuration "template" from the results of a run:

``` {.sh}
$ ./parser.py --template template.yaml ...
```

This generated configuration can then be further edited manually.

* Tests with result "PASS" are omitted.
* The following tests fields are omitted from the generated rule "criteria":
  "iteration", "start date" and "start time".
* The generated rule "criteria" "log" field is filtered to remove the leading
  path before C filename.

251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
### EBBR configuration

The `EBBR.yaml` file is a configuration file meant for [EBBR] testing. It can
override the result of some tests with the following ones:

-------------------------------------------------------------------------------
                   Result  Description
-------------------------  ----------------------------------------------------
                `IGNORED`  False-positive test failure, not mandated by [EBBR]
                           and too fine-grained to be removed from the
                           `EBBR.seq` sequence file.

`KNOWN U-BOOT LIMITATION`  Genuine bugs, which much ultimately be fixed.
                           We know about them; they are due to U-Boot FAT
                           filesystem implementation limitations and they do
                           not prevent an OS to boot.
267
268
269
270

   `KNOWN ACS LIMITATION` Genuine bugs, which are fixed in a more recent version
                          of the ACS or which must ultimately be fixed and which
                          we know about.
271
272
273
274
-------------------------------------------------------------------------------

Some of the rules just add a `comments` field with some help text.

Jeffrey Booher-Kaeding's avatar
Jeffrey Booher-Kaeding committed
275
276
## Notes
### Known Issues:
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
277
* "comment" is currently not implemented, as formatting is not currently consistent, should reflect the comments from the test.
Vincent Stehlé's avatar
Vincent Stehlé committed
278
* some [UEFI SCT] tests have shared GUIDs,
279
* some lines in ekl file follow Different naming Conventions
Vincent Stehlé's avatar
Vincent Stehlé committed
280
* some tests in the sequence file are not strongly Associated with the test spec.
281

282
283
284
285
### Documentation

It is possible to convert this `README.md` into `README.pdf` with pandoc using
`make doc`. See `make help`.
286

Vincent Stehlé's avatar
Vincent Stehlé committed
287
288
### Sanity checks

289
290
291
292
293
294
295
296
297
298
299
300
To perform sanity checks, run `make check`. It runs a number of checkers and
reports errors:

-------------------------------
      Checker  Target
-------------  ----------------
     `flake8`  Python scripts.
   `yamllint`  [YAML] files.
 `shellcheck`  Shell scripts.
-------------------------------

See `make help`.
Vincent Stehlé's avatar
Vincent Stehlé committed
301

302
### db structure:
303

Vincent Stehlé's avatar
Vincent Stehlé committed
304
``` {.python}
305
306
307
308
tests = [
    test_dict,
    est_dict2...
]
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
309

310
311
312
313
test_dict = {
   "name": "some test",
   "result": "pass/fail",
   "group": "some group",
Vincent Stehlé's avatar
Vincent Stehlé committed
314
   "test set": "some set",
315
316
317
   "sub set": "some subset",
   "set guid": "XXXXXX",
   "guid": "XXXXXX",
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
318
   "comment": "some comment",
319
320
321
322
   "log": "full log output"
}


Vincent Stehlé's avatar
Vincent Stehlé committed
323
seqs = {
324
325
326
327
328
329
330
331
332
333
    <guid> : seq_dict
    <guid2> : seq_dict2...
}

seq_dict = {
                "name": "set name",
                "guid": "set guid",
                "Iteration": "some hex/num of how many times to run",
                "rev": "some hex/numb",
                "Order": "some hex/num"
Jeff Booher-Kaeding's avatar
V1.0?    
Jeff Booher-Kaeding committed
334
}
335
```
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366

#### Spurious tests

Spurious tests are tests, which were run according to the log file but were not
meant to be run according to the sequence file.

We force the "result" fields of those tests to "SPURIOUS".

#### Dropped tests sets

Dropped tests sets are the tests sets, which were were meant to be run according
to the sequence file but for which no test have been run according to the log
file.

We create artificial tests entries for those dropped tests sets, with the
"result" fields set to "DROPPED". We convert some fields coming from the
sequence file, and auto-generate others:

``` {.python}
dropped_test_dict = {
   "name": "",
   "result": "DROPPED",
   "group": "Unknown",
   "test set": "",
   "sub set": <name from sequence file>,
   "set guid": <guid from sequence file>,
   "revision": <rev from sequence file>,
   "guid": "",
   "log": ""
}
```
367
368
369
370
371
372
373
374

#### Skipped tests sets

Skipped tests sets are the tests sets, which were considered but had zero of
their test run according to the log file.

We create artificial tests entries for those dropped tests sets, with the
"result" fields set to "SKIPPED".
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390

## Contributed files

A few contributed files are stored in sub-folders under `contrib` for
convenience:

-------------------------------------------------------------------------------
           Sub-folder  Contents
---------------------  --------------------------------------------------------
 `v21.05_0.8_BETA-0/`  EBBR sequence file from [ACS-IR v21.05_0.8_BETA-0].

   `v21.07_0.9_BETA/`  EBBR sequence files from [ACS-IR v21.07_0.9_BETA].
-------------------------------------------------------------------------------

[ACS-IR v21.05_0.8_BETA-0]: https://github.com/ARM-software/arm-systemready/tree/main/IR/prebuilt_images/v21.05_0.8_BETA-0
[ACS-IR v21.07_0.9_BETA]: https://github.com/ARM-software/arm-systemready/tree/main/IR/prebuilt_images/v21.07_0.9_BETA