]>
Commit | Line | Data |
---|---|---|
fc3fe1c2 SG |
1 | # Copyright (c) 2013 The Chromium OS Authors. |
2 | # | |
3 | # See file CREDITS for list of people who contributed to this | |
4 | # project. | |
5 | # | |
6 | # This program is free software; you can redistribute it and/or | |
7 | # modify it under the terms of the GNU General Public License as | |
8 | # published by the Free Software Foundation; either version 2 of | |
9 | # the License, or (at your option) any later version. | |
10 | # | |
11 | # This program is distributed in the hope that it will be useful, | |
12 | # but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | # GNU General Public License for more details. | |
15 | # | |
16 | # You should have received a copy of the GNU General Public License | |
17 | # along with this program; if not, write to the Free Software | |
18 | # Foundation, Inc., 59 Temple Place, Suite 330, Boston, | |
19 | # MA 02111-1307 USA | |
20 | # | |
21 | ||
22 | What is this? | |
23 | ============= | |
24 | ||
25 | This tool handles building U-Boot to check that you have not broken it | |
26 | with your patch series. It can build each individual commit and report | |
27 | which boards fail on which commits, and which errors come up. It aims | |
28 | to make full use of multi-processor machines. | |
29 | ||
30 | A key feature of buildman is its output summary, which allows warnings, | |
31 | errors or image size increases in a particular commit or board to be | |
32 | quickly identified and the offending commit pinpointed. This can be a big | |
33 | help for anyone working with >10 patches at a time. | |
34 | ||
35 | ||
36 | Caveats | |
37 | ======= | |
38 | ||
39 | Buildman is still in its infancy. It is already a very useful tool, but | |
40 | expect to find problems and send patches. | |
41 | ||
42 | Buildman can be stopped and restarted, in which case it will continue | |
43 | where it left off. This should happen cleanly and without side-effects. | |
44 | If not, it is a bug, for which a patch would be welcome. | |
45 | ||
46 | Buildman gets so tied up in its work that it can ignore the outside world. | |
47 | You may need to press Ctrl-C several times to quit it. Also it will print | |
48 | out various exceptions when stopped. | |
49 | ||
50 | ||
51 | Theory of Operation | |
52 | =================== | |
53 | ||
54 | (please read this section in full twice or you will be perpetually confused) | |
55 | ||
56 | Buildman is a builder. It is not make, although it runs make. It does not | |
57 | produce any useful output on the terminal while building, except for | |
58 | progress information. All the output (errors, warnings and binaries if you | |
59 | are ask for them) is stored in output directories, which you can look at | |
60 | while the build is progressing, or when it is finished. | |
61 | ||
62 | Buildman produces a concise summary of which boards succeeded and failed. | |
63 | It shows which commit introduced which board failure using a simple | |
64 | red/green colour coding. Full error information can be requested, in which | |
65 | case it is de-duped and displayed against the commit that introduced the | |
66 | error. An example workflow is below. | |
67 | ||
68 | Buildman stores image size information and can report changes in image size | |
69 | from commit to commit. An example of this is below. | |
70 | ||
71 | Buildman starts multiple threads, and each thread builds for one board at | |
72 | a time. A thread starts at the first commit, configures the source for your | |
73 | board and builds it. Then it checks out the next commit and does an | |
74 | incremental build. Eventually the thread reaches the last commit and stops. | |
75 | If errors or warnings are found along the way, the thread will reconfigure | |
76 | after every commit, and your build will be very slow. This is because a | |
77 | file that produces just a warning would not normally be rebuilt in an | |
78 | incremental build. | |
79 | ||
80 | Buildman works in an entirely separate place from your U-Boot repository. | |
81 | It creates a separate working directory for each thread, and puts the | |
82 | output files in the working directory, organised by commit name and board | |
83 | name, in a two-level hierarchy. | |
84 | ||
85 | Buildman is invoked in your U-Boot directory, the one with the .git | |
86 | directory. It clones this repository into a copy for each thread, and the | |
87 | threads do not affect the state of your git repository. Any checkouts done | |
88 | by the thread affect only the working directory for that thread. | |
89 | ||
90 | Buildman automatically selects the correct toolchain for each board. You | |
91 | must supply suitable toolchains, but buildman takes care of selecting the | |
92 | right one. | |
93 | ||
94 | Buildman always builds a branch, and always builds the upstream commit as | |
95 | well, for comparison. It cannot build individual commits at present, unless | |
96 | (maybe) you point it at an empty branch. Put all your commits in a branch, | |
97 | set the branch's upstream to a valid value, and all will be well. Otherwise | |
98 | buildman will perform random actions. Use -n to check what the random | |
99 | actions might be. | |
100 | ||
101 | Buildman is optimised for building many commits at once, for many boards. | |
102 | On multi-core machines, Buildman is fast because it uses most of the | |
103 | available CPU power. When it gets to the end, or if you are building just | |
104 | a few commits or boards, it will be pretty slow. As a tip, if you don't | |
105 | plan to use your machine for anything else, you can use -T to increase the | |
106 | number of threads beyond the default. | |
107 | ||
108 | Buildman lets you build all boards, or a subset. Specify the subset using | |
109 | the board name, architecture name, SOC name, or anything else in the | |
110 | boards.cfg file. So 'at91' will build all AT91 boards (arm), powerpc will | |
111 | build all PowerPC boards. | |
112 | ||
113 | Buildman does not store intermediate object files. It optionally copies | |
114 | the binary output into a directory when a build is successful. Size | |
115 | information is always recorded. It needs a fair bit of disk space to work, | |
116 | typically 250MB per thread. | |
117 | ||
118 | ||
119 | Setting up | |
120 | ========== | |
121 | ||
122 | 1. Get the U-Boot source. You probably already have it, but if not these | |
123 | steps should get you started with a repo and some commits for testing. | |
124 | ||
125 | $ cd /path/to/u-boot | |
126 | $ git clone git://git.denx.de/u-boot.git . | |
127 | $ git checkout -b my-branch origin/master | |
128 | $ # Add some commits to the branch, reading for testing | |
129 | ||
130 | 2. Create ~/.buildman to tell buildman where to find tool chains. As an | |
131 | example: | |
132 | ||
133 | # Buildman settings file | |
134 | ||
135 | [toolchain] | |
136 | root: / | |
137 | rest: /toolchains/* | |
138 | eldk: /opt/eldk-4.2 | |
139 | ||
140 | [toolchain-alias] | |
141 | x86: i386 | |
142 | blackfin: bfin | |
143 | sh: sh4 | |
144 | nds32: nds32le | |
145 | openrisc: or32 | |
146 | ||
147 | ||
148 | This selects the available toolchain paths. Add the base directory for | |
149 | each of your toolchains here. Buildman will search inside these directories | |
150 | and also in any '/usr' and '/usr/bin' subdirectories. | |
151 | ||
152 | Make sure the tags (here root: rest: and eldk:) are unique. | |
153 | ||
154 | The toolchain-alias section indicates that the i386 toolchain should be used | |
155 | to build x86 commits. | |
156 | ||
157 | ||
158 | 2. Check the available toolchains | |
159 | ||
160 | Run this check to make sure that you have a toolchain for every architecture. | |
161 | ||
162 | $ ./tools/buildman/buildman --list-tool-chains | |
163 | Scanning for tool chains | |
164 | - scanning path '/' | |
165 | - looking in '/.' | |
166 | - looking in '/bin' | |
167 | - looking in '/usr/bin' | |
168 | - found '/usr/bin/gcc' | |
169 | Tool chain test: OK | |
170 | - found '/usr/bin/c89-gcc' | |
171 | Tool chain test: OK | |
172 | - found '/usr/bin/c99-gcc' | |
173 | Tool chain test: OK | |
174 | - found '/usr/bin/x86_64-linux-gnu-gcc' | |
175 | Tool chain test: OK | |
176 | - scanning path '/toolchains/powerpc-linux' | |
177 | - looking in '/toolchains/powerpc-linux/.' | |
178 | - looking in '/toolchains/powerpc-linux/bin' | |
179 | - found '/toolchains/powerpc-linux/bin/powerpc-linux-gcc' | |
180 | Tool chain test: OK | |
181 | - looking in '/toolchains/powerpc-linux/usr/bin' | |
182 | - scanning path '/toolchains/nds32le-linux-glibc-v1f' | |
183 | - looking in '/toolchains/nds32le-linux-glibc-v1f/.' | |
184 | - looking in '/toolchains/nds32le-linux-glibc-v1f/bin' | |
185 | - found '/toolchains/nds32le-linux-glibc-v1f/bin/nds32le-linux-gcc' | |
186 | Tool chain test: OK | |
187 | - looking in '/toolchains/nds32le-linux-glibc-v1f/usr/bin' | |
188 | - scanning path '/toolchains/nios2' | |
189 | - looking in '/toolchains/nios2/.' | |
190 | - looking in '/toolchains/nios2/bin' | |
191 | - found '/toolchains/nios2/bin/nios2-linux-gcc' | |
192 | Tool chain test: OK | |
193 | - found '/toolchains/nios2/bin/nios2-linux-uclibc-gcc' | |
194 | Tool chain test: OK | |
195 | - looking in '/toolchains/nios2/usr/bin' | |
196 | - found '/toolchains/nios2/usr/bin/nios2-linux-gcc' | |
197 | Tool chain test: OK | |
198 | - found '/toolchains/nios2/usr/bin/nios2-linux-uclibc-gcc' | |
199 | Tool chain test: OK | |
200 | - scanning path '/toolchains/microblaze-unknown-linux-gnu' | |
201 | - looking in '/toolchains/microblaze-unknown-linux-gnu/.' | |
202 | - looking in '/toolchains/microblaze-unknown-linux-gnu/bin' | |
203 | - found '/toolchains/microblaze-unknown-linux-gnu/bin/microblaze-unknown-linux-gnu-gcc' | |
204 | Tool chain test: OK | |
205 | - found '/toolchains/microblaze-unknown-linux-gnu/bin/mb-linux-gcc' | |
206 | Tool chain test: OK | |
207 | - looking in '/toolchains/microblaze-unknown-linux-gnu/usr/bin' | |
208 | - scanning path '/toolchains/mips-linux' | |
209 | - looking in '/toolchains/mips-linux/.' | |
210 | - looking in '/toolchains/mips-linux/bin' | |
211 | - found '/toolchains/mips-linux/bin/mips-linux-gcc' | |
212 | Tool chain test: OK | |
213 | - looking in '/toolchains/mips-linux/usr/bin' | |
214 | - scanning path '/toolchains/old' | |
215 | - looking in '/toolchains/old/.' | |
216 | - looking in '/toolchains/old/bin' | |
217 | - looking in '/toolchains/old/usr/bin' | |
218 | - scanning path '/toolchains/i386-linux' | |
219 | - looking in '/toolchains/i386-linux/.' | |
220 | - looking in '/toolchains/i386-linux/bin' | |
221 | - found '/toolchains/i386-linux/bin/i386-linux-gcc' | |
222 | Tool chain test: OK | |
223 | - looking in '/toolchains/i386-linux/usr/bin' | |
224 | - scanning path '/toolchains/bfin-uclinux' | |
225 | - looking in '/toolchains/bfin-uclinux/.' | |
226 | - looking in '/toolchains/bfin-uclinux/bin' | |
227 | - found '/toolchains/bfin-uclinux/bin/bfin-uclinux-gcc' | |
228 | Tool chain test: OK | |
229 | - looking in '/toolchains/bfin-uclinux/usr/bin' | |
230 | - scanning path '/toolchains/sparc-elf' | |
231 | - looking in '/toolchains/sparc-elf/.' | |
232 | - looking in '/toolchains/sparc-elf/bin' | |
233 | - found '/toolchains/sparc-elf/bin/sparc-elf-gcc' | |
234 | Tool chain test: OK | |
235 | - looking in '/toolchains/sparc-elf/usr/bin' | |
236 | - scanning path '/toolchains/arm-2010q1' | |
237 | - looking in '/toolchains/arm-2010q1/.' | |
238 | - looking in '/toolchains/arm-2010q1/bin' | |
239 | - found '/toolchains/arm-2010q1/bin/arm-none-linux-gnueabi-gcc' | |
240 | Tool chain test: OK | |
241 | - looking in '/toolchains/arm-2010q1/usr/bin' | |
242 | - scanning path '/toolchains/from' | |
243 | - looking in '/toolchains/from/.' | |
244 | - looking in '/toolchains/from/bin' | |
245 | - looking in '/toolchains/from/usr/bin' | |
246 | - scanning path '/toolchains/sh4-gentoo-linux-gnu' | |
247 | - looking in '/toolchains/sh4-gentoo-linux-gnu/.' | |
248 | - looking in '/toolchains/sh4-gentoo-linux-gnu/bin' | |
249 | - found '/toolchains/sh4-gentoo-linux-gnu/bin/sh4-gentoo-linux-gnu-gcc' | |
250 | Tool chain test: OK | |
251 | - looking in '/toolchains/sh4-gentoo-linux-gnu/usr/bin' | |
252 | - scanning path '/toolchains/avr32-linux' | |
253 | - looking in '/toolchains/avr32-linux/.' | |
254 | - looking in '/toolchains/avr32-linux/bin' | |
255 | - found '/toolchains/avr32-linux/bin/avr32-gcc' | |
256 | Tool chain test: OK | |
257 | - looking in '/toolchains/avr32-linux/usr/bin' | |
258 | - scanning path '/toolchains/m68k-linux' | |
259 | - looking in '/toolchains/m68k-linux/.' | |
260 | - looking in '/toolchains/m68k-linux/bin' | |
261 | - found '/toolchains/m68k-linux/bin/m68k-linux-gcc' | |
262 | Tool chain test: OK | |
263 | - looking in '/toolchains/m68k-linux/usr/bin' | |
264 | List of available toolchains (17): | |
265 | arm : /toolchains/arm-2010q1/bin/arm-none-linux-gnueabi-gcc | |
266 | avr32 : /toolchains/avr32-linux/bin/avr32-gcc | |
267 | bfin : /toolchains/bfin-uclinux/bin/bfin-uclinux-gcc | |
268 | c89 : /usr/bin/c89-gcc | |
269 | c99 : /usr/bin/c99-gcc | |
270 | i386 : /toolchains/i386-linux/bin/i386-linux-gcc | |
271 | m68k : /toolchains/m68k-linux/bin/m68k-linux-gcc | |
272 | mb : /toolchains/microblaze-unknown-linux-gnu/bin/mb-linux-gcc | |
273 | microblaze: /toolchains/microblaze-unknown-linux-gnu/bin/microblaze-unknown-linux-gnu-gcc | |
274 | mips : /toolchains/mips-linux/bin/mips-linux-gcc | |
275 | nds32le : /toolchains/nds32le-linux-glibc-v1f/bin/nds32le-linux-gcc | |
276 | nios2 : /toolchains/nios2/bin/nios2-linux-gcc | |
277 | powerpc : /toolchains/powerpc-linux/bin/powerpc-linux-gcc | |
278 | sandbox : /usr/bin/gcc | |
279 | sh4 : /toolchains/sh4-gentoo-linux-gnu/bin/sh4-gentoo-linux-gnu-gcc | |
280 | sparc : /toolchains/sparc-elf/bin/sparc-elf-gcc | |
281 | x86_64 : /usr/bin/x86_64-linux-gnu-gcc | |
282 | ||
283 | ||
284 | You can see that everything is covered, even some strange ones that won't | |
285 | be used (c88 and c99). This is a feature. | |
286 | ||
287 | ||
288 | How to run it | |
289 | ============= | |
290 | ||
291 | First do a dry run using the -n flag: (replace <branch> with a real, local | |
292 | branch with a valid upstream) | |
293 | ||
294 | $ ./tools/buildman/buildman -b <branch> -n | |
295 | ||
296 | If it can't detect the upstream branch, try checking out the branch, and | |
297 | doing something like 'git branch --set-upstream <branch> upstream/master' | |
298 | or something similar. | |
299 | ||
300 | As an exmmple: | |
301 | ||
302 | Dry run, so not doing much. But I would do this: | |
303 | ||
304 | Building 18 commits for 1059 boards (4 threads, 1 job per thread) | |
305 | Build directory: ../lcd9b | |
306 | 5bb3505 Merge branch 'master' of git://git.denx.de/u-boot-arm | |
307 | c18f1b4 tegra: Use const for pinmux_config_pingroup/table() | |
308 | 2f043ae tegra: Add display support to funcmux | |
309 | e349900 tegra: fdt: Add pwm binding and node | |
310 | 424a5f0 tegra: fdt: Add LCD definitions for Tegra | |
311 | 0636ccf tegra: Add support for PWM | |
312 | a994fe7 tegra: Add SOC support for display/lcd | |
313 | fcd7350 tegra: Add LCD driver | |
314 | 4d46e9d tegra: Add LCD support to Nvidia boards | |
315 | 991bd48 arm: Add control over cachability of memory regions | |
316 | 54e8019 lcd: Add CONFIG_LCD_ALIGNMENT to select frame buffer alignment | |
317 | d92aff7 lcd: Add support for flushing LCD fb from dcache after update | |
318 | dbd0677 tegra: Align LCD frame buffer to section boundary | |
319 | 0cff9b8 tegra: Support control of cache settings for LCD | |
320 | 9c56900 tegra: fdt: Add LCD definitions for Seaboard | |
321 | 5cc29db lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console | |
322 | cac5a23 tegra: Enable display/lcd support on Seaboard | |
323 | 49ff541 wip | |
324 | ||
325 | Total boards to build for each commit: 1059 | |
326 | ||
327 | This shows that it will build all 1059 boards, using 4 threads (because | |
328 | we have a 4-core CPU). Each thread will run with -j1, meaning that each | |
329 | make job will use a single CPU. The list of commits to be built helps you | |
330 | confirm that things look about right. Notice that buildman has chosen a | |
331 | 'base' directory for you, immediately above your source tree. | |
332 | ||
333 | Buildman works entirely inside the base directory, here ../lcd9b, | |
334 | creating a working directory for each thread, and creating output | |
335 | directories for each commit and board. | |
336 | ||
337 | ||
338 | Suggested Workflow | |
339 | ================== | |
340 | ||
341 | To run the build for real, take off the -n: | |
342 | ||
343 | $ ./tools/buildman/buildman -b <branch> | |
344 | ||
345 | Buildman will set up some working directories, and get started. After a | |
346 | minute or so it will settle down to a steady pace, with a display like this: | |
347 | ||
348 | Building 18 commits for 1059 boards (4 threads, 1 job per thread) | |
349 | 528 36 124 /19062 1:13:30 : SIMPC8313_SP | |
350 | ||
351 | This means that it is building 19062 board/commit combinations. So far it | |
352 | has managed to succesfully build 528. Another 36 have built with warnings, | |
353 | and 124 more didn't build at all. Buildman expects to complete the process | |
354 | in an hour and 15 minutes. Use this time to buy a faster computer. | |
355 | ||
356 | ||
357 | To find out how the build went, ask for a summary with -s. You can do this | |
358 | either before the build completes (presumably in another terminal) or or | |
359 | afterwards. Let's work through an example of how this is used: | |
360 | ||
361 | $ ./tools/buildman/buildman -b lcd9b -s | |
362 | ... | |
363 | 01: Merge branch 'master' of git://git.denx.de/u-boot-arm | |
364 | powerpc: + galaxy5200_LOWBOOT | |
365 | 02: tegra: Use const for pinmux_config_pingroup/table() | |
366 | 03: tegra: Add display support to funcmux | |
367 | 04: tegra: fdt: Add pwm binding and node | |
368 | 05: tegra: fdt: Add LCD definitions for Tegra | |
369 | 06: tegra: Add support for PWM | |
370 | 07: tegra: Add SOC support for display/lcd | |
371 | 08: tegra: Add LCD driver | |
372 | 09: tegra: Add LCD support to Nvidia boards | |
373 | 10: arm: Add control over cachability of memory regions | |
374 | 11: lcd: Add CONFIG_LCD_ALIGNMENT to select frame buffer alignment | |
375 | 12: lcd: Add support for flushing LCD fb from dcache after update | |
376 | arm: + lubbock | |
377 | 13: tegra: Align LCD frame buffer to section boundary | |
378 | 14: tegra: Support control of cache settings for LCD | |
379 | 15: tegra: fdt: Add LCD definitions for Seaboard | |
380 | 16: lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console | |
381 | 17: tegra: Enable display/lcd support on Seaboard | |
382 | 18: wip | |
383 | ||
384 | This shows which commits have succeeded and which have failed. In this case | |
385 | the build is still in progress so many boards are not built yet (use -u to | |
386 | see which ones). But still we can see a few failures. The galaxy5200_LOWBOOT | |
387 | never builds correctly. This could be a problem with our toolchain, or it | |
388 | could be a bug in the upstream. The good news is that we probably don't need | |
389 | to blame our commits. The bad news is it isn't tested on that board. | |
390 | ||
391 | Commit 12 broke lubbock. That's what the '+ lubbock' means. The failure | |
392 | is never fixed by a later commit, or you would see lubbock again, in green, | |
393 | without the +. | |
394 | ||
395 | To see the actual error: | |
396 | ||
397 | $ ./tools/buildman/buildman -b <branch> -se lubbock | |
398 | ... | |
399 | 12: lcd: Add support for flushing LCD fb from dcache after update | |
400 | arm: + lubbock | |
401 | +common/libcommon.o: In function `lcd_sync': | |
402 | +/u-boot/lcd9b/.bm-work/00/common/lcd.c:120: undefined reference to `flush_dcache_range' | |
403 | +arm-none-linux-gnueabi-ld: BFD (Sourcery G++ Lite 2010q1-202) 2.19.51.20090709 assertion fail /scratch/julian/2010q1-release-linux-lite/obj/binutils-src-2010q1-202-arm-none-linux-gnueabi-i686-pc-linux-gnu/bfd/elf32-arm.c:12572 | |
404 | +make: *** [/u-boot/lcd9b/.bm-work/00/build/u-boot] Error 139 | |
405 | 13: tegra: Align LCD frame buffer to section boundary | |
406 | 14: tegra: Support control of cache settings for LCD | |
407 | 15: tegra: fdt: Add LCD definitions for Seaboard | |
408 | 16: lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console | |
409 | -/u-boot/lcd9b/.bm-work/00/common/lcd.c:120: undefined reference to `flush_dcache_range' | |
410 | +/u-boot/lcd9b/.bm-work/00/common/lcd.c:125: undefined reference to `flush_dcache_range' | |
411 | 17: tegra: Enable display/lcd support on Seaboard | |
412 | 18: wip | |
413 | ||
414 | So the problem is in lcd.c, due to missing cache operations. This information | |
415 | should be enough to work out what that commit is doing to break these | |
416 | boards. (In this case pxa did not have cache operations defined). | |
417 | ||
418 | If you see error lines marked with - that means that the errors were fixed | |
419 | by that commit. Sometimes commits can be in the wrong order, so that a | |
420 | breakage is introduced for a few commits and fixed by later commits. This | |
421 | shows up clearly with buildman. You can then reorder the commits and try | |
422 | again. | |
423 | ||
424 | At commit 16, the error moves - you can see that the old error at line 120 | |
425 | is fixed, but there is a new one at line 126. This is probably only because | |
426 | we added some code and moved the broken line futher down the file. | |
427 | ||
428 | If many boards have the same error, then -e will display the error only | |
429 | once. This makes the output as concise as possible. | |
430 | ||
431 | The full build output in this case is available in: | |
432 | ||
433 | ../lcd9b/12_of_18_gd92aff7_lcd--Add-support-for/lubbock/ | |
434 | ||
435 | done: Indicates the build was done, and holds the return code from make. | |
436 | This is 0 for a good build, typically 2 for a failure. | |
437 | ||
438 | err: Output from stderr, if any. Errors and warnings appear here. | |
439 | ||
440 | log: Output from stdout. Normally there isn't any since buildman runs | |
441 | in silent mode for now. | |
442 | ||
443 | toolchain: Shows information about the toolchain used for the build. | |
444 | ||
445 | sizes: Shows image size information. | |
446 | ||
447 | It is possible to get the build output there also. Use the -k option for | |
448 | this. In that case you will also see some output files, like: | |
449 | ||
450 | System.map toolchain u-boot u-boot.bin u-boot.map autoconf.mk | |
451 | (also SPL versions u-boot-spl and u-boot-spl.bin if available) | |
452 | ||
453 | ||
454 | Checking Image Sizes | |
455 | ==================== | |
456 | ||
457 | A key requirement for U-Boot is that you keep code/data size to a minimum. | |
458 | Where a new feature increases this noticeably it should normally be put | |
459 | behind a CONFIG flag so that boards can leave it off and keep the image | |
460 | size more or less the same with each new release. | |
461 | ||
462 | To check the impact of your commits on image size, use -S. For example: | |
463 | ||
464 | $ ./tools/buildman/buildman -b us-x86 -sS | |
465 | Summary of 10 commits for 1066 boards (4 threads, 1 job per thread) | |
466 | 01: MAKEALL: add support for per architecture toolchains | |
467 | 02: x86: Add function to get top of usable ram | |
468 | x86: (for 1/3 boards) text -272.0 rodata +41.0 | |
469 | 03: x86: Add basic cache operations | |
470 | 04: x86: Permit bootstage and timer data to be used prior to relocation | |
471 | x86: (for 1/3 boards) data +16.0 | |
472 | 05: x86: Add an __end symbol to signal the end of the U-Boot binary | |
473 | x86: (for 1/3 boards) text +76.0 | |
474 | 06: x86: Rearrange the output input to remove BSS | |
475 | x86: (for 1/3 boards) bss -2140.0 | |
476 | 07: x86: Support relocation of FDT on start-up | |
477 | x86: + coreboot-x86 | |
478 | 08: x86: Add error checking to x86 relocation code | |
479 | 09: x86: Adjust link device tree include file | |
480 | 10: x86: Enable CONFIG_OF_CONTROL on coreboot | |
481 | ||
482 | ||
483 | You can see that image size only changed on x86, which is good because this | |
484 | series is not supposed to change any other board. From commit 7 onwards the | |
485 | build fails so we don't get code size numbers. The numbers are fractional | |
486 | because they are an average of all boards for that architecture. The | |
487 | intention is to allow you to quickly find image size problems introduced by | |
488 | your commits. | |
489 | ||
490 | Note that the 'text' region and 'rodata' are split out. You should add the | |
491 | two together to get the total read-only size (reported as the first column | |
492 | in the output from binutil's 'size' utility). | |
493 | ||
494 | A useful option is --step which lets you skip some commits. For example | |
495 | --step 2 will show the image sizes for only every 2nd commit (so it will | |
496 | compare the image sizes of the 1st, 3rd, 5th... commits). You can also use | |
497 | --step 0 which will compare only the first and last commits. This is useful | |
498 | for an overview of how your entire series affects code size. | |
499 | ||
500 | You can also use -d to see a detailed size breakdown for each board. This | |
501 | list is sorted in order from largest growth to largest reduction. | |
502 | ||
503 | It is possible to go a little further with the -B option (--bloat). This | |
504 | shows where U-Boot has bloted, breaking the size change down to the function | |
505 | level. Example output is below: | |
506 | ||
507 | $ ./tools/buildman/buildman -b us-mem4 -sSdB | |
508 | ... | |
509 | 19: Roll crc32 into hash infrastructure | |
510 | arm: (for 10/10 boards) all -143.4 bss +1.2 data -4.8 rodata -48.2 text -91.6 | |
511 | paz00 : all +23 bss -4 rodata -29 text +56 | |
512 | u-boot: add: 1/0, grow: 3/-2 bytes: 168/-104 (64) | |
513 | function old new delta | |
514 | hash_command 80 160 +80 | |
515 | crc32_wd_buf - 56 +56 | |
516 | ext4fs_read_file 540 568 +28 | |
517 | insert_var_value_sub 688 692 +4 | |
518 | run_list_real 1996 1992 -4 | |
519 | do_mem_crc 168 68 -100 | |
520 | trimslice : all -9 bss +16 rodata -29 text +4 | |
521 | u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12) | |
522 | function old new delta | |
523 | hash_command 80 160 +80 | |
524 | crc32_wd_buf - 56 +56 | |
525 | ext4fs_iterate_dir 672 668 -4 | |
526 | ext4fs_read_file 568 548 -20 | |
527 | do_mem_crc 168 68 -100 | |
528 | whistler : all -9 bss +16 rodata -29 text +4 | |
529 | u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12) | |
530 | function old new delta | |
531 | hash_command 80 160 +80 | |
532 | crc32_wd_buf - 56 +56 | |
533 | ext4fs_iterate_dir 672 668 -4 | |
534 | ext4fs_read_file 568 548 -20 | |
535 | do_mem_crc 168 68 -100 | |
536 | seaboard : all -9 bss -28 rodata -29 text +48 | |
537 | u-boot: add: 1/0, grow: 3/-2 bytes: 160/-104 (56) | |
538 | function old new delta | |
539 | hash_command 80 160 +80 | |
540 | crc32_wd_buf - 56 +56 | |
541 | ext4fs_read_file 548 568 +20 | |
542 | run_list_real 1996 2000 +4 | |
543 | do_nandboot 760 756 -4 | |
544 | do_mem_crc 168 68 -100 | |
545 | colibri_t20_iris: all -9 rodata -29 text +20 | |
546 | u-boot: add: 1/0, grow: 2/-3 bytes: 140/-112 (28) | |
547 | function old new delta | |
548 | hash_command 80 160 +80 | |
549 | crc32_wd_buf - 56 +56 | |
550 | read_abs_bbt 204 208 +4 | |
551 | do_nandboot 760 756 -4 | |
552 | ext4fs_read_file 576 568 -8 | |
553 | do_mem_crc 168 68 -100 | |
554 | ventana : all -37 bss -12 rodata -29 text +4 | |
555 | u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12) | |
556 | function old new delta | |
557 | hash_command 80 160 +80 | |
558 | crc32_wd_buf - 56 +56 | |
559 | ext4fs_iterate_dir 672 668 -4 | |
560 | ext4fs_read_file 568 548 -20 | |
561 | do_mem_crc 168 68 -100 | |
562 | harmony : all -37 bss -16 rodata -29 text +8 | |
563 | u-boot: add: 1/0, grow: 2/-3 bytes: 140/-124 (16) | |
564 | function old new delta | |
565 | hash_command 80 160 +80 | |
566 | crc32_wd_buf - 56 +56 | |
567 | nand_write_oob_syndrome 428 432 +4 | |
568 | ext4fs_iterate_dir 672 668 -4 | |
569 | ext4fs_read_file 568 548 -20 | |
570 | do_mem_crc 168 68 -100 | |
571 | medcom-wide : all -417 bss +28 data -16 rodata -93 text -336 | |
572 | u-boot: add: 1/-1, grow: 1/-2 bytes: 88/-376 (-288) | |
573 | function old new delta | |
574 | crc32_wd_buf - 56 +56 | |
575 | do_fat_read_at 2872 2904 +32 | |
576 | hash_algo 16 - -16 | |
577 | do_mem_crc 168 68 -100 | |
578 | hash_command 420 160 -260 | |
579 | tec : all -449 bss -4 data -16 rodata -93 text -336 | |
580 | u-boot: add: 1/-1, grow: 1/-2 bytes: 88/-376 (-288) | |
581 | function old new delta | |
582 | crc32_wd_buf - 56 +56 | |
583 | do_fat_read_at 2872 2904 +32 | |
584 | hash_algo 16 - -16 | |
585 | do_mem_crc 168 68 -100 | |
586 | hash_command 420 160 -260 | |
587 | plutux : all -481 bss +16 data -16 rodata -93 text -388 | |
588 | u-boot: add: 1/-1, grow: 1/-3 bytes: 68/-408 (-340) | |
589 | function old new delta | |
590 | crc32_wd_buf - 56 +56 | |
591 | do_load_serial_bin 1688 1700 +12 | |
592 | hash_algo 16 - -16 | |
593 | do_fat_read_at 2904 2872 -32 | |
594 | do_mem_crc 168 68 -100 | |
595 | hash_command 420 160 -260 | |
596 | powerpc: (for 5/5 boards) all +37.4 data -3.2 rodata -41.8 text +82.4 | |
597 | MPC8610HPCD : all +55 rodata -29 text +84 | |
598 | u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80) | |
599 | function old new delta | |
600 | hash_command - 176 +176 | |
601 | do_mem_crc 184 88 -96 | |
602 | MPC8641HPCN : all +55 rodata -29 text +84 | |
603 | u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80) | |
604 | function old new delta | |
605 | hash_command - 176 +176 | |
606 | do_mem_crc 184 88 -96 | |
607 | MPC8641HPCN_36BIT: all +55 rodata -29 text +84 | |
608 | u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80) | |
609 | function old new delta | |
610 | hash_command - 176 +176 | |
611 | do_mem_crc 184 88 -96 | |
612 | sbc8641d : all +55 rodata -29 text +84 | |
613 | u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80) | |
614 | function old new delta | |
615 | hash_command - 176 +176 | |
616 | do_mem_crc 184 88 -96 | |
617 | xpedite517x : all -33 data -16 rodata -93 text +76 | |
618 | u-boot: add: 1/-1, grow: 0/-1 bytes: 176/-112 (64) | |
619 | function old new delta | |
620 | hash_command - 176 +176 | |
621 | hash_algo 16 - -16 | |
622 | do_mem_crc 184 88 -96 | |
623 | ... | |
624 | ||
625 | ||
626 | This shows that commit 19 has increased text size for arm (although only one | |
627 | board was built) and by 96 bytes for powerpc. This increase was offset in both | |
628 | cases by reductions in rodata and data/bss. | |
629 | ||
630 | Shown below the summary lines is the sizes for each board. Below each board | |
631 | is the sizes for each function. This information starts with: | |
632 | ||
633 | add - number of functions added / removed | |
634 | grow - number of functions which grew / shrunk | |
635 | bytes - number of bytes of code added to / removed from all functions, | |
636 | plus the total byte change in brackets | |
637 | ||
638 | The change seems to be that hash_command() has increased by more than the | |
639 | do_mem_crc() function has decreased. The function sizes typically add up to | |
640 | roughly the text area size, but note that every read-only section except | |
641 | rodata is included in 'text', so the function total does not exactly | |
642 | correspond. | |
643 | ||
644 | It is common when refactoring code for the rodata to decrease as the text size | |
645 | increases, and vice versa. | |
646 | ||
647 | ||
648 | Other options | |
649 | ============= | |
650 | ||
651 | Buildman has various other command line options. Try --help to see them. | |
652 | ||
653 | ||
654 | TODO | |
655 | ==== | |
656 | ||
657 | This has mostly be written in my spare time as a response to my difficulties | |
658 | in testing large series of patches. Apart from tidying up there is quite a | |
659 | bit of scope for improvement. Things like better error diffs, easier access | |
660 | to log files, error display while building. Also it would be nice it buildman | |
661 | could 'hunt' for problems, perhaps by building a few boards for each arch, | |
662 | or checking commits for changed files and building only boards which use | |
663 | those files. | |
664 | ||
665 | ||
666 | Credits | |
667 | ======= | |
668 | ||
669 | Thanks to Grant Grundler <grundler@chromium.org> for his ideas for improving | |
670 | the build speed by building all commits for a board instead of the other | |
671 | way around. | |
672 | ||
673 | ||
674 | ||
675 | Simon Glass | |
676 | sjg@chromium.org | |
677 | Halloween 2012 | |
678 | Updated 12-12-12 | |
679 | Updated 23-02-13 |