blob: 8c5f8610d02f8c4c3145904bae05bb6bcd71f2c2 [file] [log] [blame]
Simon Glassfc3fe1c2013-04-03 11:07:16 +00001# Copyright (c) 2013 The Chromium OS Authors.
2#
Wolfgang Denk1a459662013-07-08 09:37:19 +02003# SPDX-License-Identifier: GPL-2.0+
Simon Glassfc3fe1c2013-04-03 11:07:16 +00004#
5
Simon Glass6eede342014-08-09 15:32:58 -06006(Please read 'How to change from MAKEALL' if you are used to that tool)
7
Simon Glassc8d73932016-07-27 20:33:06 -06008Quick-start
9===========
10
11If you just want to quickly set up buildman so you can build something (for
12example Raspberry Pi 2):
13
14 cd /path/to/u-boot
15 PATH=$PATH:`pwd`/tools/buildman
16 buildman --fetch-arch arm
17 buildman -k rpi_2
18 ls ../current/rpi_2
19 # u-boot.bin is the output image
20
21
Simon Glassfc3fe1c2013-04-03 11:07:16 +000022What is this?
23=============
24
25This tool handles building U-Boot to check that you have not broken it
26with your patch series. It can build each individual commit and report
27which boards fail on which commits, and which errors come up. It aims
28to make full use of multi-processor machines.
29
30A key feature of buildman is its output summary, which allows warnings,
31errors or image size increases in a particular commit or board to be
32quickly identified and the offending commit pinpointed. This can be a big
33help for anyone working with >10 patches at a time.
34
35
36Caveats
37=======
38
Simon Glassfc3fe1c2013-04-03 11:07:16 +000039Buildman can be stopped and restarted, in which case it will continue
40where it left off. This should happen cleanly and without side-effects.
41If not, it is a bug, for which a patch would be welcome.
42
43Buildman gets so tied up in its work that it can ignore the outside world.
44You may need to press Ctrl-C several times to quit it. Also it will print
Simon Glass8ea42102016-07-27 20:32:58 -060045out various exceptions when stopped. You may have to kill it since the
46Ctrl-C handling is somewhat broken.
Simon Glassfc3fe1c2013-04-03 11:07:16 +000047
48
49Theory of Operation
50===================
51
52(please read this section in full twice or you will be perpetually confused)
53
54Buildman is a builder. It is not make, although it runs make. It does not
55produce any useful output on the terminal while building, except for
Simon Glasse5a0e5d2014-08-09 15:33:03 -060056progress information (except with -v, see below). All the output (errors,
Dirk Behme3e1ded12014-12-23 07:41:26 +010057warnings and binaries if you ask for them) is stored in output
Simon Glasse5a0e5d2014-08-09 15:33:03 -060058directories, which you can look at while the build is progressing, or when
59it is finished.
Simon Glassfc3fe1c2013-04-03 11:07:16 +000060
Simon Glass8ea42102016-07-27 20:32:58 -060061Buildman is designed to build entire git branches, i.e. muliple commits. It
62can be run repeatedly on the same branch. In this case it will automatically
63rebuild commits which have changed (and remove its old results for that
64commit). It is possible to build a branch for one board, then later build it
65for another board. If you want buildman to re-build a commit it has already
66built (e.g. because of a toolchain update), use the -f flag.
67
Simon Glassfc3fe1c2013-04-03 11:07:16 +000068Buildman produces a concise summary of which boards succeeded and failed.
69It shows which commit introduced which board failure using a simple
70red/green colour coding. Full error information can be requested, in which
71case it is de-duped and displayed against the commit that introduced the
72error. An example workflow is below.
73
74Buildman stores image size information and can report changes in image size
75from commit to commit. An example of this is below.
76
77Buildman starts multiple threads, and each thread builds for one board at
78a time. A thread starts at the first commit, configures the source for your
79board and builds it. Then it checks out the next commit and does an
80incremental build. Eventually the thread reaches the last commit and stops.
81If errors or warnings are found along the way, the thread will reconfigure
82after every commit, and your build will be very slow. This is because a
83file that produces just a warning would not normally be rebuilt in an
84incremental build.
85
86Buildman works in an entirely separate place from your U-Boot repository.
87It creates a separate working directory for each thread, and puts the
88output files in the working directory, organised by commit name and board
89name, in a two-level hierarchy.
90
91Buildman is invoked in your U-Boot directory, the one with the .git
92directory. It clones this repository into a copy for each thread, and the
93threads do not affect the state of your git repository. Any checkouts done
94by the thread affect only the working directory for that thread.
95
Simon Glasscec83c32014-08-09 15:32:57 -060096Buildman automatically selects the correct tool chain for each board. You
97must supply suitable tool chains, but buildman takes care of selecting the
Simon Glassfc3fe1c2013-04-03 11:07:16 +000098right one.
99
Simon Glasse5a0e5d2014-08-09 15:33:03 -0600100Buildman generally builds a branch (with the -b flag), and in this case
101builds the upstream commit as well, for comparison. It cannot build
102individual commits at present, unless (maybe) you point it at an empty
103branch. Put all your commits in a branch, set the branch's upstream to a
104valid value, and all will be well. Otherwise buildman will perform random
105actions. Use -n to check what the random actions might be.
106
Simon Glass1d8104f2014-10-16 01:05:56 -0600107If you just want to build the current source tree, leave off the -b flag
108and add -e. This will display results and errors as they happen. You can
109still look at them later using -se. Note that buildman will assume that the
110source has changed, and will build all specified boards in this case.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000111
112Buildman is optimised for building many commits at once, for many boards.
113On multi-core machines, Buildman is fast because it uses most of the
114available CPU power. When it gets to the end, or if you are building just
115a few commits or boards, it will be pretty slow. As a tip, if you don't
116plan to use your machine for anything else, you can use -T to increase the
117number of threads beyond the default.
118
Stephen Warren8426d8b2013-10-10 10:00:20 -0600119Buildman lets you build all boards, or a subset. Specify the subset by passing
120command-line arguments that list the desired board name, architecture name,
121SOC name, or anything else in the boards.cfg file. Multiple arguments are
122allowed. Each argument will be interpreted as a regular expression, so
123behaviour is a superset of exact or substring matching. Examples are:
124
125* 'tegra20' All boards with a Tegra20 SoC
126* 'tegra' All boards with any Tegra Soc (Tegra20, Tegra30, Tegra114...)
127* '^tegra[23]0$' All boards with either Tegra20 or Tegra30 SoC
128* 'powerpc' All PowerPC boards
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000129
Simon Glass6131bea2014-08-09 15:33:08 -0600130While the default is to OR the terms together, you can also make use of
131the '&' operator to limit the selection:
132
133* 'freescale & arm sandbox' All Freescale boards with ARM architecture,
134 plus sandbox
135
Simon Glass3cf4ae62014-08-28 09:43:41 -0600136You can also use -x to specifically exclude some boards. For example:
137
138 buildmand arm -x nvidia,freescale,.*ball$
139
140means to build all arm boards except nvidia, freescale and anything ending
141with 'ball'.
142
Dirk Behme3e1ded12014-12-23 07:41:26 +0100143It is convenient to use the -n option to see what will be built based on
Simon Glass6131bea2014-08-09 15:33:08 -0600144the subset given.
145
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000146Buildman does not store intermediate object files. It optionally copies
147the binary output into a directory when a build is successful. Size
148information is always recorded. It needs a fair bit of disk space to work,
149typically 250MB per thread.
150
151
152Setting up
153==========
154
1551. Get the U-Boot source. You probably already have it, but if not these
156steps should get you started with a repo and some commits for testing.
157
158$ cd /path/to/u-boot
159$ git clone git://git.denx.de/u-boot.git .
160$ git checkout -b my-branch origin/master
161$ # Add some commits to the branch, reading for testing
162
Simon Glass62005342014-12-01 17:34:02 -07001632. Create ~/.buildman to tell buildman where to find tool chains (see 'The
164.buildman file' later for details). As an example:
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000165
166# Buildman settings file
167
168[toolchain]
169root: /
170rest: /toolchains/*
171eldk: /opt/eldk-4.2
Simon Glasse9569472014-08-09 15:33:07 -0600172arm: /opt/linaro/gcc-linaro-arm-linux-gnueabihf-4.8-2013.08_linux
173aarch64: /opt/linaro/gcc-linaro-aarch64-none-elf-4.8-2013.10_linux
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000174
175[toolchain-alias]
176x86: i386
177blackfin: bfin
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000178nds32: nds32le
Bin Menge8aebc42016-02-21 21:18:02 -0800179openrisc: or1k
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000180
181
182This selects the available toolchain paths. Add the base directory for
183each of your toolchains here. Buildman will search inside these directories
184and also in any '/usr' and '/usr/bin' subdirectories.
185
186Make sure the tags (here root: rest: and eldk:) are unique.
187
188The toolchain-alias section indicates that the i386 toolchain should be used
189to build x86 commits.
190
Simon Glass17bce662016-03-12 18:50:32 -0700191Note that you can also specific exactly toolchain prefixes if you like:
192
193[toolchain-prefix]
194arm: /opt/arm-eabi-4.6/bin/arm-eabi-
195
196or even:
197
198[toolchain-prefix]
199arm: /opt/arm-eabi-4.6/bin/arm-eabi-gcc
200
201This tells buildman that you want to use this exact toolchain for the arm
202architecture. This will override any toolchains found by searching using the
203[toolchain] settings.
204
205Since the toolchain prefix is an explicit request, buildman will report an
206error if a toolchain is not found with that prefix. The current PATH will be
207searched, so it is possible to use:
208
209[toolchain-prefix]
210arm: arm-none-eabi-
211
212and buildman will find arm-none-eabi-gcc in /usr/bin if you have it installed.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000213
Simon Glass34699692014-12-01 17:34:01 -07002143. Make sure you have the require Python pre-requisites
215
Simon Glass827e37b2014-12-01 17:34:06 -0700216Buildman uses multiprocessing, Queue, shutil, StringIO, ConfigParser and
217urllib2. These should normally be available, but if you get an error like
218this then you will need to obtain those modules:
Simon Glass34699692014-12-01 17:34:01 -0700219
220 ImportError: No module named multiprocessing
221
222
2234. Check the available toolchains
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000224
225Run this check to make sure that you have a toolchain for every architecture.
226
227$ ./tools/buildman/buildman --list-tool-chains
228Scanning for tool chains
Simon Glass17bce662016-03-12 18:50:32 -0700229 - scanning prefix '/opt/gcc-4.6.3-nolibc/x86_64-linux/bin/x86_64-linux-'
230Tool chain test: OK, arch='x86', priority 1
231 - scanning prefix '/opt/arm-eabi-4.6/bin/arm-eabi-'
232Tool chain test: OK, arch='arm', priority 1
233 - scanning path '/toolchains/gcc-4.9.0-nolibc/i386-linux'
234 - looking in '/toolchains/gcc-4.9.0-nolibc/i386-linux/.'
235 - looking in '/toolchains/gcc-4.9.0-nolibc/i386-linux/bin'
236 - found '/toolchains/gcc-4.9.0-nolibc/i386-linux/bin/i386-linux-gcc'
237 - looking in '/toolchains/gcc-4.9.0-nolibc/i386-linux/usr/bin'
238Tool chain test: OK, arch='i386', priority 4
239 - scanning path '/toolchains/gcc-4.9.0-nolibc/aarch64-linux'
240 - looking in '/toolchains/gcc-4.9.0-nolibc/aarch64-linux/.'
241 - looking in '/toolchains/gcc-4.9.0-nolibc/aarch64-linux/bin'
242 - found '/toolchains/gcc-4.9.0-nolibc/aarch64-linux/bin/aarch64-linux-gcc'
243 - looking in '/toolchains/gcc-4.9.0-nolibc/aarch64-linux/usr/bin'
244Tool chain test: OK, arch='aarch64', priority 4
245 - scanning path '/toolchains/gcc-4.9.0-nolibc/microblaze-linux'
246 - looking in '/toolchains/gcc-4.9.0-nolibc/microblaze-linux/.'
247 - looking in '/toolchains/gcc-4.9.0-nolibc/microblaze-linux/bin'
248 - found '/toolchains/gcc-4.9.0-nolibc/microblaze-linux/bin/microblaze-linux-gcc'
249 - looking in '/toolchains/gcc-4.9.0-nolibc/microblaze-linux/usr/bin'
250Tool chain test: OK, arch='microblaze', priority 4
251 - scanning path '/toolchains/gcc-4.9.0-nolibc/mips64-linux'
252 - looking in '/toolchains/gcc-4.9.0-nolibc/mips64-linux/.'
253 - looking in '/toolchains/gcc-4.9.0-nolibc/mips64-linux/bin'
254 - found '/toolchains/gcc-4.9.0-nolibc/mips64-linux/bin/mips64-linux-gcc'
255 - looking in '/toolchains/gcc-4.9.0-nolibc/mips64-linux/usr/bin'
256Tool chain test: OK, arch='mips64', priority 4
257 - scanning path '/toolchains/gcc-4.9.0-nolibc/sparc64-linux'
258 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc64-linux/.'
259 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc64-linux/bin'
260 - found '/toolchains/gcc-4.9.0-nolibc/sparc64-linux/bin/sparc64-linux-gcc'
261 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc64-linux/usr/bin'
262Tool chain test: OK, arch='sparc64', priority 4
263 - scanning path '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi'
264 - looking in '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi/.'
265 - looking in '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi/bin'
266 - found '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi/bin/arm-unknown-linux-gnueabi-gcc'
267 - looking in '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi/usr/bin'
268Tool chain test: OK, arch='arm', priority 3
269Toolchain '/toolchains/gcc-4.9.0-nolibc/arm-unknown-linux-gnueabi/bin/arm-unknown-linux-gnueabi-gcc' at priority 3 will be ignored because another toolchain for arch 'arm' has priority 1
270 - scanning path '/toolchains/gcc-4.9.0-nolibc/sparc-linux'
271 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc-linux/.'
272 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc-linux/bin'
273 - found '/toolchains/gcc-4.9.0-nolibc/sparc-linux/bin/sparc-linux-gcc'
274 - looking in '/toolchains/gcc-4.9.0-nolibc/sparc-linux/usr/bin'
275Tool chain test: OK, arch='sparc', priority 4
276 - scanning path '/toolchains/gcc-4.9.0-nolibc/mips-linux'
277 - looking in '/toolchains/gcc-4.9.0-nolibc/mips-linux/.'
278 - looking in '/toolchains/gcc-4.9.0-nolibc/mips-linux/bin'
279 - found '/toolchains/gcc-4.9.0-nolibc/mips-linux/bin/mips-linux-gcc'
280 - looking in '/toolchains/gcc-4.9.0-nolibc/mips-linux/usr/bin'
281Tool chain test: OK, arch='mips', priority 4
282 - scanning path '/toolchains/gcc-4.9.0-nolibc/x86_64-linux'
283 - looking in '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/.'
284 - looking in '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/bin'
285 - found '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/bin/x86_64-linux-gcc'
286 - found '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/bin/x86_64-linux-x86_64-linux-gcc'
287 - looking in '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/usr/bin'
288Tool chain test: OK, arch='x86_64', priority 4
289Tool chain test: OK, arch='x86_64', priority 4
290Toolchain '/toolchains/gcc-4.9.0-nolibc/x86_64-linux/bin/x86_64-linux-x86_64-linux-gcc' at priority 4 will be ignored because another toolchain for arch 'x86_64' has priority 4
291 - scanning path '/toolchains/gcc-4.9.0-nolibc/m68k-linux'
292 - looking in '/toolchains/gcc-4.9.0-nolibc/m68k-linux/.'
293 - looking in '/toolchains/gcc-4.9.0-nolibc/m68k-linux/bin'
294 - found '/toolchains/gcc-4.9.0-nolibc/m68k-linux/bin/m68k-linux-gcc'
295 - looking in '/toolchains/gcc-4.9.0-nolibc/m68k-linux/usr/bin'
296Tool chain test: OK, arch='m68k', priority 4
297 - scanning path '/toolchains/gcc-4.9.0-nolibc/powerpc-linux'
298 - looking in '/toolchains/gcc-4.9.0-nolibc/powerpc-linux/.'
299 - looking in '/toolchains/gcc-4.9.0-nolibc/powerpc-linux/bin'
300 - found '/toolchains/gcc-4.9.0-nolibc/powerpc-linux/bin/powerpc-linux-gcc'
301 - looking in '/toolchains/gcc-4.9.0-nolibc/powerpc-linux/usr/bin'
302Tool chain test: OK, arch='powerpc', priority 4
303 - scanning path '/toolchains/gcc-4.6.3-nolibc/bfin-uclinux'
304 - looking in '/toolchains/gcc-4.6.3-nolibc/bfin-uclinux/.'
305 - looking in '/toolchains/gcc-4.6.3-nolibc/bfin-uclinux/bin'
306 - found '/toolchains/gcc-4.6.3-nolibc/bfin-uclinux/bin/bfin-uclinux-gcc'
307 - looking in '/toolchains/gcc-4.6.3-nolibc/bfin-uclinux/usr/bin'
308Tool chain test: OK, arch='bfin', priority 6
309 - scanning path '/toolchains/gcc-4.6.3-nolibc/sparc-linux'
310 - looking in '/toolchains/gcc-4.6.3-nolibc/sparc-linux/.'
311 - looking in '/toolchains/gcc-4.6.3-nolibc/sparc-linux/bin'
312 - found '/toolchains/gcc-4.6.3-nolibc/sparc-linux/bin/sparc-linux-gcc'
313 - looking in '/toolchains/gcc-4.6.3-nolibc/sparc-linux/usr/bin'
314Tool chain test: OK, arch='sparc', priority 4
315Toolchain '/toolchains/gcc-4.6.3-nolibc/sparc-linux/bin/sparc-linux-gcc' at priority 4 will be ignored because another toolchain for arch 'sparc' has priority 4
316 - scanning path '/toolchains/gcc-4.6.3-nolibc/mips-linux'
317 - looking in '/toolchains/gcc-4.6.3-nolibc/mips-linux/.'
318 - looking in '/toolchains/gcc-4.6.3-nolibc/mips-linux/bin'
319 - found '/toolchains/gcc-4.6.3-nolibc/mips-linux/bin/mips-linux-gcc'
320 - looking in '/toolchains/gcc-4.6.3-nolibc/mips-linux/usr/bin'
321Tool chain test: OK, arch='mips', priority 4
322Toolchain '/toolchains/gcc-4.6.3-nolibc/mips-linux/bin/mips-linux-gcc' at priority 4 will be ignored because another toolchain for arch 'mips' has priority 4
323 - scanning path '/toolchains/gcc-4.6.3-nolibc/m68k-linux'
324 - looking in '/toolchains/gcc-4.6.3-nolibc/m68k-linux/.'
325 - looking in '/toolchains/gcc-4.6.3-nolibc/m68k-linux/bin'
326 - found '/toolchains/gcc-4.6.3-nolibc/m68k-linux/bin/m68k-linux-gcc'
327 - looking in '/toolchains/gcc-4.6.3-nolibc/m68k-linux/usr/bin'
328Tool chain test: OK, arch='m68k', priority 4
329Toolchain '/toolchains/gcc-4.6.3-nolibc/m68k-linux/bin/m68k-linux-gcc' at priority 4 will be ignored because another toolchain for arch 'm68k' has priority 4
330 - scanning path '/toolchains/gcc-4.6.3-nolibc/powerpc-linux'
331 - looking in '/toolchains/gcc-4.6.3-nolibc/powerpc-linux/.'
332 - looking in '/toolchains/gcc-4.6.3-nolibc/powerpc-linux/bin'
333 - found '/toolchains/gcc-4.6.3-nolibc/powerpc-linux/bin/powerpc-linux-gcc'
334 - looking in '/toolchains/gcc-4.6.3-nolibc/powerpc-linux/usr/bin'
335Tool chain test: OK, arch='powerpc', priority 4
336Tool chain test: OK, arch='or32', priority 4
337 - scanning path '/toolchains/gcc-4.2.4-nolibc/avr32-linux'
338 - looking in '/toolchains/gcc-4.2.4-nolibc/avr32-linux/.'
339 - looking in '/toolchains/gcc-4.2.4-nolibc/avr32-linux/bin'
340 - found '/toolchains/gcc-4.2.4-nolibc/avr32-linux/bin/avr32-linux-gcc'
341 - looking in '/toolchains/gcc-4.2.4-nolibc/avr32-linux/usr/bin'
342Tool chain test: OK, arch='avr32', priority 4
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000343 - scanning path '/'
344 - looking in '/.'
345 - looking in '/bin'
346 - looking in '/usr/bin'
Simon Glass17bce662016-03-12 18:50:32 -0700347 - found '/usr/bin/i586-mingw32msvc-gcc'
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000348 - found '/usr/bin/c89-gcc'
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000349 - found '/usr/bin/x86_64-linux-gnu-gcc'
Simon Glass17bce662016-03-12 18:50:32 -0700350 - found '/usr/bin/gcc'
351 - found '/usr/bin/c99-gcc'
352 - found '/usr/bin/arm-linux-gnueabi-gcc'
353 - found '/usr/bin/aarch64-linux-gnu-gcc'
354 - found '/usr/bin/winegcc'
355 - found '/usr/bin/arm-linux-gnueabihf-gcc'
356Tool chain test: OK, arch='i586', priority 11
357Tool chain test: OK, arch='c89', priority 11
358Tool chain test: OK, arch='x86_64', priority 4
359Toolchain '/usr/bin/x86_64-linux-gnu-gcc' at priority 4 will be ignored because another toolchain for arch 'x86_64' has priority 4
360Tool chain test: OK, arch='sandbox', priority 11
361Tool chain test: OK, arch='c99', priority 11
362Tool chain test: OK, arch='arm', priority 4
363Toolchain '/usr/bin/arm-linux-gnueabi-gcc' at priority 4 will be ignored because another toolchain for arch 'arm' has priority 1
364Tool chain test: OK, arch='aarch64', priority 4
365Toolchain '/usr/bin/aarch64-linux-gnu-gcc' at priority 4 will be ignored because another toolchain for arch 'aarch64' has priority 4
366Tool chain test: OK, arch='sandbox', priority 11
367Toolchain '/usr/bin/winegcc' at priority 11 will be ignored because another toolchain for arch 'sandbox' has priority 11
368Tool chain test: OK, arch='arm', priority 4
369Toolchain '/usr/bin/arm-linux-gnueabihf-gcc' at priority 4 will be ignored because another toolchain for arch 'arm' has priority 1
370List of available toolchains (34):
371aarch64 : /toolchains/gcc-4.9.0-nolibc/aarch64-linux/bin/aarch64-linux-gcc
372alpha : /toolchains/gcc-4.9.0-nolibc/alpha-linux/bin/alpha-linux-gcc
373am33_2.0 : /toolchains/gcc-4.9.0-nolibc/am33_2.0-linux/bin/am33_2.0-linux-gcc
374arm : /opt/arm-eabi-4.6/bin/arm-eabi-gcc
375avr32 : /toolchains/gcc-4.2.4-nolibc/avr32-linux/bin/avr32-linux-gcc
376bfin : /toolchains/gcc-4.6.3-nolibc/bfin-uclinux/bin/bfin-uclinux-gcc
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000377c89 : /usr/bin/c89-gcc
378c99 : /usr/bin/c99-gcc
Simon Glass17bce662016-03-12 18:50:32 -0700379frv : /toolchains/gcc-4.9.0-nolibc/frv-linux/bin/frv-linux-gcc
380h8300 : /toolchains/gcc-4.9.0-nolibc/h8300-elf/bin/h8300-elf-gcc
381hppa : /toolchains/gcc-4.9.0-nolibc/hppa-linux/bin/hppa-linux-gcc
382hppa64 : /toolchains/gcc-4.9.0-nolibc/hppa64-linux/bin/hppa64-linux-gcc
383i386 : /toolchains/gcc-4.9.0-nolibc/i386-linux/bin/i386-linux-gcc
384i586 : /usr/bin/i586-mingw32msvc-gcc
385ia64 : /toolchains/gcc-4.9.0-nolibc/ia64-linux/bin/ia64-linux-gcc
386m32r : /toolchains/gcc-4.9.0-nolibc/m32r-linux/bin/m32r-linux-gcc
387m68k : /toolchains/gcc-4.9.0-nolibc/m68k-linux/bin/m68k-linux-gcc
388microblaze: /toolchains/gcc-4.9.0-nolibc/microblaze-linux/bin/microblaze-linux-gcc
389mips : /toolchains/gcc-4.9.0-nolibc/mips-linux/bin/mips-linux-gcc
390mips64 : /toolchains/gcc-4.9.0-nolibc/mips64-linux/bin/mips64-linux-gcc
391or32 : /toolchains/gcc-4.5.1-nolibc/or32-linux/bin/or32-linux-gcc
392powerpc : /toolchains/gcc-4.9.0-nolibc/powerpc-linux/bin/powerpc-linux-gcc
393powerpc64 : /toolchains/gcc-4.9.0-nolibc/powerpc64-linux/bin/powerpc64-linux-gcc
394ppc64le : /toolchains/gcc-4.9.0-nolibc/ppc64le-linux/bin/ppc64le-linux-gcc
395s390x : /toolchains/gcc-4.9.0-nolibc/s390x-linux/bin/s390x-linux-gcc
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000396sandbox : /usr/bin/gcc
Simon Glass17bce662016-03-12 18:50:32 -0700397sh4 : /toolchains/gcc-4.6.3-nolibc/sh4-linux/bin/sh4-linux-gcc
398sparc : /toolchains/gcc-4.9.0-nolibc/sparc-linux/bin/sparc-linux-gcc
399sparc64 : /toolchains/gcc-4.9.0-nolibc/sparc64-linux/bin/sparc64-linux-gcc
400tilegx : /toolchains/gcc-4.6.2-nolibc/tilegx-linux/bin/tilegx-linux-gcc
401x86 : /opt/gcc-4.6.3-nolibc/x86_64-linux/bin/x86_64-linux-gcc
402x86_64 : /toolchains/gcc-4.9.0-nolibc/x86_64-linux/bin/x86_64-linux-gcc
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000403
404
405You can see that everything is covered, even some strange ones that won't
406be used (c88 and c99). This is a feature.
407
408
Simon Glass827e37b2014-12-01 17:34:06 -07004095. Install new toolchains if needed
410
411You can download toolchains and update the [toolchain] section of the
412settings file to find them.
413
414To make this easier, buildman can automatically download and install
415toolchains from kernel.org. First list the available architectures:
416
Bin Meng9f244b22015-07-16 19:43:46 -0700417$ ./tools/buildman/buildman --fetch-arch list
Simon Glass827e37b2014-12-01 17:34:06 -0700418Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.6.3/
419Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.6.2/
420Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.5.1/
421Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.2.4/
422Available architectures: alpha am33_2.0 arm avr32 bfin cris crisv32 frv h8300
423hppa hppa64 i386 ia64 m32r m68k mips mips64 or32 powerpc powerpc64 s390x sh4
424sparc sparc64 tilegx x86_64 xtensa
425
426Then pick one and download it:
427
Bin Meng9f244b22015-07-16 19:43:46 -0700428$ ./tools/buildman/buildman --fetch-arch or32
Simon Glass827e37b2014-12-01 17:34:06 -0700429Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.6.3/
430Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.6.2/
431Checking: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.5.1/
432Downloading: https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.5.1//x86_64-gcc-4.5.1-nolibc_or32-linux.tar.xz
433Unpacking to: /home/sjg/.buildman-toolchains
434Testing
435 - looking in '/home/sjg/.buildman-toolchains/gcc-4.5.1-nolibc/or32-linux/.'
436 - looking in '/home/sjg/.buildman-toolchains/gcc-4.5.1-nolibc/or32-linux/bin'
437 - found '/home/sjg/.buildman-toolchains/gcc-4.5.1-nolibc/or32-linux/bin/or32-linux-gcc'
438Tool chain test: OK
439
Thomas Chou89515232015-11-12 09:29:09 +0800440Or download them all from kernel.org and move them to /toolchains directory,
441
Simon Glass8ea42102016-07-27 20:32:58 -0600442$ ./tools/buildman/buildman --fetch-arch all
Thomas Chou89515232015-11-12 09:29:09 +0800443$ sudo mkdir -p /toolchains
444$ sudo mv ~/.buildman-toolchains/*/* /toolchains/
445
446For those not available from kernel.org, download from the following links.
447
448arc: https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/
449 arc_gnu_2015.06_prebuilt_uclibc_le_archs_linux_install.tar.gz
450blackfin: http://sourceforge.net/projects/adi-toolchain/files/
451 blackfin-toolchain-elf-gcc-4.5-2014R1_45-RC2.x86_64.tar.bz2
452nds32: http://osdk.andestech.com/packages/
453 nds32le-linux-glibc-v1.tgz
454nios2: http://sourcery.mentor.com/public/gnu_toolchain/nios2-linux-gnu/
455 sourceryg++-2015.11-27-nios2-linux-gnu-i686-pc-linux-gnu.tar.bz2
456sh: http://sourcery.mentor.com/public/gnu_toolchain/sh-linux-gnu/
457 renesas-4.4-200-sh-linux-gnu-i686-pc-linux-gnu.tar.bz2
458
Simon Glass8ea42102016-07-27 20:32:58 -0600459Note openrisc kernel.org toolchain is out of date. Download the latest one from
460http://opencores.org/or1k/OpenRISC_GNU_tool_chain#Prebuilt_versions - eg:
Bin Menge8aebc42016-02-21 21:18:02 -0800461ftp://ocuser:ocuser@openrisc.opencores.org/toolchain/gcc-or1k-elf-4.8.1-x86.tar.bz2.
462
Simon Glass827e37b2014-12-01 17:34:06 -0700463Buildman should now be set up to use your new toolchain.
464
465At the time of writing, U-Boot has these architectures:
466
467 arc, arm, avr32, blackfin, m68k, microblaze, mips, nds32, nios2, openrisc
468 powerpc, sandbox, sh, sparc, x86
469
Michal Simek12462312015-04-20 11:46:24 +0200470Of these, only arc and nds32 are not available at kernel.org..
Simon Glass827e37b2014-12-01 17:34:06 -0700471
472
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000473How to run it
474=============
475
476First do a dry run using the -n flag: (replace <branch> with a real, local
477branch with a valid upstream)
478
479$ ./tools/buildman/buildman -b <branch> -n
480
481If it can't detect the upstream branch, try checking out the branch, and
Simon Glass2a9e2c62014-12-01 17:33:54 -0700482doing something like 'git branch --set-upstream-to upstream/master'
483or something similar. Buildman will try to guess a suitable upstream branch
484if it can't find one (you will see a message like" Guessing upstream as ...).
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000485
Simon Glasscec83c32014-08-09 15:32:57 -0600486As an example:
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000487
488Dry run, so not doing much. But I would do this:
489
490Building 18 commits for 1059 boards (4 threads, 1 job per thread)
491Build directory: ../lcd9b
492 5bb3505 Merge branch 'master' of git://git.denx.de/u-boot-arm
493 c18f1b4 tegra: Use const for pinmux_config_pingroup/table()
494 2f043ae tegra: Add display support to funcmux
495 e349900 tegra: fdt: Add pwm binding and node
496 424a5f0 tegra: fdt: Add LCD definitions for Tegra
497 0636ccf tegra: Add support for PWM
498 a994fe7 tegra: Add SOC support for display/lcd
499 fcd7350 tegra: Add LCD driver
500 4d46e9d tegra: Add LCD support to Nvidia boards
501 991bd48 arm: Add control over cachability of memory regions
502 54e8019 lcd: Add CONFIG_LCD_ALIGNMENT to select frame buffer alignment
503 d92aff7 lcd: Add support for flushing LCD fb from dcache after update
504 dbd0677 tegra: Align LCD frame buffer to section boundary
505 0cff9b8 tegra: Support control of cache settings for LCD
506 9c56900 tegra: fdt: Add LCD definitions for Seaboard
507 5cc29db lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console
508 cac5a23 tegra: Enable display/lcd support on Seaboard
509 49ff541 wip
510
511Total boards to build for each commit: 1059
512
513This shows that it will build all 1059 boards, using 4 threads (because
514we have a 4-core CPU). Each thread will run with -j1, meaning that each
515make job will use a single CPU. The list of commits to be built helps you
516confirm that things look about right. Notice that buildman has chosen a
517'base' directory for you, immediately above your source tree.
518
519Buildman works entirely inside the base directory, here ../lcd9b,
520creating a working directory for each thread, and creating output
521directories for each commit and board.
522
523
524Suggested Workflow
525==================
526
527To run the build for real, take off the -n:
528
529$ ./tools/buildman/buildman -b <branch>
530
531Buildman will set up some working directories, and get started. After a
532minute or so it will settle down to a steady pace, with a display like this:
533
534Building 18 commits for 1059 boards (4 threads, 1 job per thread)
535 528 36 124 /19062 1:13:30 : SIMPC8313_SP
536
537This means that it is building 19062 board/commit combinations. So far it
Simon Glasscec83c32014-08-09 15:32:57 -0600538has managed to successfully build 528. Another 36 have built with warnings,
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000539and 124 more didn't build at all. Buildman expects to complete the process
Simon Glass8ea42102016-07-27 20:32:58 -0600540in around an hour and a quarter. Use this time to buy a faster computer.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000541
542
543To find out how the build went, ask for a summary with -s. You can do this
Dirk Behme3e1ded12014-12-23 07:41:26 +0100544either before the build completes (presumably in another terminal) or
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000545afterwards. Let's work through an example of how this is used:
546
547$ ./tools/buildman/buildman -b lcd9b -s
548...
54901: Merge branch 'master' of git://git.denx.de/u-boot-arm
550 powerpc: + galaxy5200_LOWBOOT
55102: tegra: Use const for pinmux_config_pingroup/table()
55203: tegra: Add display support to funcmux
55304: tegra: fdt: Add pwm binding and node
55405: tegra: fdt: Add LCD definitions for Tegra
55506: tegra: Add support for PWM
55607: tegra: Add SOC support for display/lcd
55708: tegra: Add LCD driver
55809: tegra: Add LCD support to Nvidia boards
55910: arm: Add control over cachability of memory regions
56011: lcd: Add CONFIG_LCD_ALIGNMENT to select frame buffer alignment
56112: lcd: Add support for flushing LCD fb from dcache after update
562 arm: + lubbock
56313: tegra: Align LCD frame buffer to section boundary
56414: tegra: Support control of cache settings for LCD
56515: tegra: fdt: Add LCD definitions for Seaboard
56616: lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console
56717: tegra: Enable display/lcd support on Seaboard
56818: wip
569
570This shows which commits have succeeded and which have failed. In this case
571the build is still in progress so many boards are not built yet (use -u to
572see which ones). But still we can see a few failures. The galaxy5200_LOWBOOT
573never builds correctly. This could be a problem with our toolchain, or it
574could be a bug in the upstream. The good news is that we probably don't need
Simon Glass8ea42102016-07-27 20:32:58 -0600575to blame our commits. The bad news is that our commits are not tested on that
576board.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000577
578Commit 12 broke lubbock. That's what the '+ lubbock' means. The failure
579is never fixed by a later commit, or you would see lubbock again, in green,
580without the +.
581
582To see the actual error:
583
584$ ./tools/buildman/buildman -b <branch> -se lubbock
585...
58612: lcd: Add support for flushing LCD fb from dcache after update
587 arm: + lubbock
588+common/libcommon.o: In function `lcd_sync':
589+/u-boot/lcd9b/.bm-work/00/common/lcd.c:120: undefined reference to `flush_dcache_range'
590+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
591+make: *** [/u-boot/lcd9b/.bm-work/00/build/u-boot] Error 139
59213: tegra: Align LCD frame buffer to section boundary
59314: tegra: Support control of cache settings for LCD
59415: tegra: fdt: Add LCD definitions for Seaboard
59516: lcd: Add CONFIG_CONSOLE_SCROLL_LINES option to speed console
596-/u-boot/lcd9b/.bm-work/00/common/lcd.c:120: undefined reference to `flush_dcache_range'
597+/u-boot/lcd9b/.bm-work/00/common/lcd.c:125: undefined reference to `flush_dcache_range'
59817: tegra: Enable display/lcd support on Seaboard
59918: wip
600
601So the problem is in lcd.c, due to missing cache operations. This information
602should be enough to work out what that commit is doing to break these
603boards. (In this case pxa did not have cache operations defined).
604
Simon Glass8ea42102016-07-27 20:32:58 -0600605If you see error lines marked with '-', that means that the errors were fixed
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000606by that commit. Sometimes commits can be in the wrong order, so that a
607breakage is introduced for a few commits and fixed by later commits. This
608shows up clearly with buildman. You can then reorder the commits and try
609again.
610
Simon Glass8ea42102016-07-27 20:32:58 -0600611At commit 16, the error moves: you can see that the old error at line 120
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000612is fixed, but there is a new one at line 126. This is probably only because
Dirk Behme3e1ded12014-12-23 07:41:26 +0100613we added some code and moved the broken line further down the file.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000614
615If many boards have the same error, then -e will display the error only
Simon Glassed966652014-08-28 09:43:43 -0600616once. This makes the output as concise as possible. To see which boards have
Simon Glass8ea42102016-07-27 20:32:58 -0600617each error, use -l. So it is safe to omit the board name - you will not get
618lots of repeated output for every board.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000619
Simon Glasse30965d2014-08-28 09:43:44 -0600620Buildman tries to distinguish warnings from errors, and shows warning lines
621separately with a 'w' prefix.
622
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000623The full build output in this case is available in:
624
625../lcd9b/12_of_18_gd92aff7_lcd--Add-support-for/lubbock/
626
627 done: Indicates the build was done, and holds the return code from make.
628 This is 0 for a good build, typically 2 for a failure.
629
630 err: Output from stderr, if any. Errors and warnings appear here.
631
632 log: Output from stdout. Normally there isn't any since buildman runs
Simon Glassc81d0d22016-03-12 18:50:33 -0700633 in silent mode. Use -V to force a verbose build (this passes V=1
634 to 'make')
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000635
636 toolchain: Shows information about the toolchain used for the build.
637
638 sizes: Shows image size information.
639
Simon Glass8ea42102016-07-27 20:32:58 -0600640It is possible to get the build binary output there also. Use the -k option
641for this. In that case you will also see some output files, like:
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000642
643 System.map toolchain u-boot u-boot.bin u-boot.map autoconf.mk
644 (also SPL versions u-boot-spl and u-boot-spl.bin if available)
645
646
647Checking Image Sizes
648====================
649
650A key requirement for U-Boot is that you keep code/data size to a minimum.
651Where a new feature increases this noticeably it should normally be put
Simon Glass8ea42102016-07-27 20:32:58 -0600652behind a CONFIG flag so that boards can leave it disabled and keep the image
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000653size more or less the same with each new release.
654
655To check the impact of your commits on image size, use -S. For example:
656
657$ ./tools/buildman/buildman -b us-x86 -sS
658Summary of 10 commits for 1066 boards (4 threads, 1 job per thread)
65901: MAKEALL: add support for per architecture toolchains
66002: x86: Add function to get top of usable ram
661 x86: (for 1/3 boards) text -272.0 rodata +41.0
66203: x86: Add basic cache operations
66304: x86: Permit bootstage and timer data to be used prior to relocation
664 x86: (for 1/3 boards) data +16.0
66505: x86: Add an __end symbol to signal the end of the U-Boot binary
666 x86: (for 1/3 boards) text +76.0
66706: x86: Rearrange the output input to remove BSS
668 x86: (for 1/3 boards) bss -2140.0
66907: x86: Support relocation of FDT on start-up
670 x86: + coreboot-x86
67108: x86: Add error checking to x86 relocation code
67209: x86: Adjust link device tree include file
67310: x86: Enable CONFIG_OF_CONTROL on coreboot
674
675
676You can see that image size only changed on x86, which is good because this
677series is not supposed to change any other board. From commit 7 onwards the
678build fails so we don't get code size numbers. The numbers are fractional
679because they are an average of all boards for that architecture. The
680intention is to allow you to quickly find image size problems introduced by
681your commits.
682
683Note that the 'text' region and 'rodata' are split out. You should add the
684two together to get the total read-only size (reported as the first column
685in the output from binutil's 'size' utility).
686
687A useful option is --step which lets you skip some commits. For example
688--step 2 will show the image sizes for only every 2nd commit (so it will
689compare the image sizes of the 1st, 3rd, 5th... commits). You can also use
690--step 0 which will compare only the first and last commits. This is useful
Simon Glass8ea42102016-07-27 20:32:58 -0600691for an overview of how your entire series affects code size. It will build
692only the upstream commit and your final branch commit.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000693
694You can also use -d to see a detailed size breakdown for each board. This
695list is sorted in order from largest growth to largest reduction.
696
Simon Glass8ea42102016-07-27 20:32:58 -0600697It is even possible to go a little further with the -B option (--bloat). This
Simon Glasscec83c32014-08-09 15:32:57 -0600698shows where U-Boot has bloated, breaking the size change down to the function
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000699level. Example output is below:
700
701$ ./tools/buildman/buildman -b us-mem4 -sSdB
702...
70319: Roll crc32 into hash infrastructure
704 arm: (for 10/10 boards) all -143.4 bss +1.2 data -4.8 rodata -48.2 text -91.6
705 paz00 : all +23 bss -4 rodata -29 text +56
706 u-boot: add: 1/0, grow: 3/-2 bytes: 168/-104 (64)
707 function old new delta
708 hash_command 80 160 +80
709 crc32_wd_buf - 56 +56
710 ext4fs_read_file 540 568 +28
711 insert_var_value_sub 688 692 +4
712 run_list_real 1996 1992 -4
713 do_mem_crc 168 68 -100
714 trimslice : all -9 bss +16 rodata -29 text +4
715 u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12)
716 function old new delta
717 hash_command 80 160 +80
718 crc32_wd_buf - 56 +56
719 ext4fs_iterate_dir 672 668 -4
720 ext4fs_read_file 568 548 -20
721 do_mem_crc 168 68 -100
722 whistler : all -9 bss +16 rodata -29 text +4
723 u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12)
724 function old new delta
725 hash_command 80 160 +80
726 crc32_wd_buf - 56 +56
727 ext4fs_iterate_dir 672 668 -4
728 ext4fs_read_file 568 548 -20
729 do_mem_crc 168 68 -100
730 seaboard : all -9 bss -28 rodata -29 text +48
731 u-boot: add: 1/0, grow: 3/-2 bytes: 160/-104 (56)
732 function old new delta
733 hash_command 80 160 +80
734 crc32_wd_buf - 56 +56
735 ext4fs_read_file 548 568 +20
736 run_list_real 1996 2000 +4
737 do_nandboot 760 756 -4
738 do_mem_crc 168 68 -100
Marcel Ziswilere57c6e52015-03-26 01:31:54 +0100739 colibri_t20 : all -9 rodata -29 text +20
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000740 u-boot: add: 1/0, grow: 2/-3 bytes: 140/-112 (28)
741 function old new delta
742 hash_command 80 160 +80
743 crc32_wd_buf - 56 +56
744 read_abs_bbt 204 208 +4
745 do_nandboot 760 756 -4
746 ext4fs_read_file 576 568 -8
747 do_mem_crc 168 68 -100
748 ventana : all -37 bss -12 rodata -29 text +4
749 u-boot: add: 1/0, grow: 1/-3 bytes: 136/-124 (12)
750 function old new delta
751 hash_command 80 160 +80
752 crc32_wd_buf - 56 +56
753 ext4fs_iterate_dir 672 668 -4
754 ext4fs_read_file 568 548 -20
755 do_mem_crc 168 68 -100
756 harmony : all -37 bss -16 rodata -29 text +8
757 u-boot: add: 1/0, grow: 2/-3 bytes: 140/-124 (16)
758 function old new delta
759 hash_command 80 160 +80
760 crc32_wd_buf - 56 +56
761 nand_write_oob_syndrome 428 432 +4
762 ext4fs_iterate_dir 672 668 -4
763 ext4fs_read_file 568 548 -20
764 do_mem_crc 168 68 -100
765 medcom-wide : all -417 bss +28 data -16 rodata -93 text -336
766 u-boot: add: 1/-1, grow: 1/-2 bytes: 88/-376 (-288)
767 function old new delta
768 crc32_wd_buf - 56 +56
769 do_fat_read_at 2872 2904 +32
770 hash_algo 16 - -16
771 do_mem_crc 168 68 -100
772 hash_command 420 160 -260
773 tec : all -449 bss -4 data -16 rodata -93 text -336
774 u-boot: add: 1/-1, grow: 1/-2 bytes: 88/-376 (-288)
775 function old new delta
776 crc32_wd_buf - 56 +56
777 do_fat_read_at 2872 2904 +32
778 hash_algo 16 - -16
779 do_mem_crc 168 68 -100
780 hash_command 420 160 -260
781 plutux : all -481 bss +16 data -16 rodata -93 text -388
782 u-boot: add: 1/-1, grow: 1/-3 bytes: 68/-408 (-340)
783 function old new delta
784 crc32_wd_buf - 56 +56
785 do_load_serial_bin 1688 1700 +12
786 hash_algo 16 - -16
787 do_fat_read_at 2904 2872 -32
788 do_mem_crc 168 68 -100
789 hash_command 420 160 -260
790 powerpc: (for 5/5 boards) all +37.4 data -3.2 rodata -41.8 text +82.4
791 MPC8610HPCD : all +55 rodata -29 text +84
792 u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80)
793 function old new delta
794 hash_command - 176 +176
795 do_mem_crc 184 88 -96
796 MPC8641HPCN : all +55 rodata -29 text +84
797 u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80)
798 function old new delta
799 hash_command - 176 +176
800 do_mem_crc 184 88 -96
801 MPC8641HPCN_36BIT: all +55 rodata -29 text +84
802 u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80)
803 function old new delta
804 hash_command - 176 +176
805 do_mem_crc 184 88 -96
806 sbc8641d : all +55 rodata -29 text +84
807 u-boot: add: 1/0, grow: 0/-1 bytes: 176/-96 (80)
808 function old new delta
809 hash_command - 176 +176
810 do_mem_crc 184 88 -96
811 xpedite517x : all -33 data -16 rodata -93 text +76
812 u-boot: add: 1/-1, grow: 0/-1 bytes: 176/-112 (64)
813 function old new delta
814 hash_command - 176 +176
815 hash_algo 16 - -16
816 do_mem_crc 184 88 -96
817...
818
819
Simon Glass8ea42102016-07-27 20:32:58 -0600820This shows that commit 19 has reduced codesize for arm slightly and increased
821it for powerpc. This increase was offset in by reductions in rodata and
822data/bss.
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000823
Dirk Behme3e1ded12014-12-23 07:41:26 +0100824Shown below the summary lines are the sizes for each board. Below each board
825are the sizes for each function. This information starts with:
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000826
827 add - number of functions added / removed
828 grow - number of functions which grew / shrunk
829 bytes - number of bytes of code added to / removed from all functions,
830 plus the total byte change in brackets
831
832The change seems to be that hash_command() has increased by more than the
833do_mem_crc() function has decreased. The function sizes typically add up to
834roughly the text area size, but note that every read-only section except
835rodata is included in 'text', so the function total does not exactly
836correspond.
837
838It is common when refactoring code for the rodata to decrease as the text size
839increases, and vice versa.
840
841
Simon Glass62005342014-12-01 17:34:02 -0700842The .buildman file
843==================
Simon Glass4281ad82013-09-23 17:35:17 -0600844
Simon Glass62005342014-12-01 17:34:02 -0700845The .buildman file provides information about the available toolchains and
846also allows build flags to be passed to 'make'. It consists of several
847sections, with the section name in square brackets. Within each section are
848a set of (tag, value) pairs.
Simon Glass4281ad82013-09-23 17:35:17 -0600849
Simon Glass62005342014-12-01 17:34:02 -0700850'[toolchain]' section
Simon Glass4281ad82013-09-23 17:35:17 -0600851
Simon Glass62005342014-12-01 17:34:02 -0700852 This lists the available toolchains. The tag here doesn't matter, but
853 make sure it is unique. The value is the path to the toolchain. Buildman
854 will look in that path for a file ending in 'gcc'. It will then execute
855 it to check that it is a C compiler, passing only the --version flag to
856 it. If the return code is 0, buildman assumes that it is a valid C
857 compiler. It uses the first part of the name as the architecture and
858 strips off the last part when setting the CROSS_COMPILE environment
859 variable (parts are delimited with a hyphen).
Simon Glass4281ad82013-09-23 17:35:17 -0600860
Simon Glass62005342014-12-01 17:34:02 -0700861 For example powerpc-linux-gcc will be noted as a toolchain for 'powerpc'
862 and CROSS_COMPILE will be set to powerpc-linux- when using it.
863
864'[toolchain-alias]' section
865
866 This converts toolchain architecture names to U-Boot names. For example,
867 if an x86 toolchains is called i386-linux-gcc it will not normally be
Simon Glass9b83bfd2014-12-01 17:34:05 -0700868 used for architecture 'x86'. Adding 'x86: i386 x86_64' to this section
869 will tell buildman that the i386 and x86_64 toolchains can be used for
870 the x86 architecture.
Simon Glass62005342014-12-01 17:34:02 -0700871
872'[make-flags]' section
873
874 U-Boot's build system supports a few flags (such as BUILD_TAG) which
875 affect the build product. These flags can be specified in the buildman
876 settings file. They can also be useful when building U-Boot against other
877 open source software.
878
879 [make-flags]
880 at91-boards=ENABLE_AT91_TEST=1
881 snapper9260=${at91-boards} BUILD_TAG=442
882 snapper9g45=${at91-boards} BUILD_TAG=443
883
884 This will use 'make ENABLE_AT91_TEST=1 BUILD_TAG=442' for snapper9260
885 and 'make ENABLE_AT91_TEST=1 BUILD_TAG=443' for snapper9g45. A special
886 variable ${target} is available to access the target name (snapper9260
887 and snapper9g20 in this case). Variables are resolved recursively. Note
888 that variables can only contain the characters A-Z, a-z, 0-9, hyphen (-)
889 and underscore (_).
890
891 It is expected that any variables added are dealt with in U-Boot's
892 config.mk file and documented in the README.
893
894 Note that you can pass ad-hoc options to the build using environment
895 variables, for example:
896
897 SOME_OPTION=1234 ./tools/buildman/buildman my_board
Simon Glass4281ad82013-09-23 17:35:17 -0600898
899
Simon Glasse5a0e5d2014-08-09 15:33:03 -0600900Quick Sanity Check
901==================
902
903If you have made changes and want to do a quick sanity check of the
Simon Glass1d8104f2014-10-16 01:05:56 -0600904currently checked-out source, run buildman without the -b flag. This will
905build the selected boards and display build status as it runs (i.e. -v is
906enabled automatically). Use -e to see errors/warnings as well.
Simon Glasse5a0e5d2014-08-09 15:33:03 -0600907
908
Simon Glass5abab202014-12-01 17:33:57 -0700909Building Ranges
910===============
911
912You can build a range of commits by specifying a range instead of a branch
913when using the -b flag. For example:
914
915 upstream/master..us-buildman
916
917will build commits in us-buildman that are not in upstream/master.
918
919
Stephen Warrenf79f1e02016-04-11 10:48:44 -0600920Building Faster
921===============
922
923By default, buildman executes 'make mrproper' prior to building the first
924commit for each board. This causes everything to be built from scratch. If you
925trust the build system's incremental build capabilities, you can pass the -I
926flag to skip the 'make mproper' invocation, which will reduce the amount of
927work 'make' does, and hence speed up the build. This flag will speed up any
928buildman invocation, since it reduces the amount of work done on any build.
929
930One possible application of buildman is as part of a continual edit, build,
931edit, build, ... cycle; repeatedly applying buildman to the same change or
932series of changes while making small incremental modifications to the source
933each time. This provides quick feedback regarding the correctness of recent
934modifications. In this scenario, buildman's default choice of build directory
935causes more build work to be performed than strictly necessary.
936
937By default, each buildman thread uses a single directory for all builds. When a
938thread builds multiple boards, the configuration built in this directory will
939cycle through various different configurations, one per board built by the
940thread. Variations in the configuration will force a rebuild of affected source
941files when a thread switches between boards. Ideally, such buildman-induced
942rebuilds would not happen, thus allowing the build to operate as efficiently as
943the build system and source changes allow. buildman's -P flag may be used to
944enable this; -P causes each board to be built in a separate (board-specific)
945directory, thus avoiding any buildman-induced configuration changes in any
946build directory.
947
948U-Boot's build system embeds information such as a build timestamp into the
949final binary. This information varies each time U-Boot is built. This causes
950various files to be rebuilt even if no source changes are made, which in turn
951requires that the final U-Boot binary be re-linked. This unnecessary work can
952be avoided by turning off the timestamp feature. This can be achieved by
953setting the SOURCE_DATE_EPOCH environment variable to 0.
954
955Combining all of these options together yields the command-line shown below.
956This will provide the quickest possible feedback regarding the current content
957of the source tree, thus allowing rapid tested evolution of the code.
958
959 SOURCE_DATE_EPOCH=0 ./tools/buildman/buildman -I -P tegra
960
961
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000962Other options
963=============
964
965Buildman has various other command line options. Try --help to see them.
966
Simon Glass2c3deb92014-08-28 09:43:39 -0600967When doing builds, Buildman's return code will reflect the overall result:
968
969 0 (success) No errors or warnings found
970 128 Errors found
971 129 Warnings found
972
Simon Glassfc3fe1c2013-04-03 11:07:16 +0000973
Simon Glass6eede342014-08-09 15:32:58 -0600974How to change from MAKEALL
975==========================
976
977Buildman includes most of the features of MAKEALL and is generally faster
978and easier to use. In particular it builds entire branches: if a particular
979commit introduces an error in a particular board, buildman can easily show
980you this, even if a later commit fixes that error.
981
982The reasons to deprecate MAKEALL are:
983- We don't want to maintain two build systems
984- Buildman is typically faster
985- Buildman has a lot more features
986
987But still, many people will be sad to lose MAKEALL. If you are used to
988MAKEALL, here are a few pointers.
989
990First you need to set up your tool chains - see the 'Setting up' section
991for details. Once you have your required toolchain(s) detected then you are
992ready to go.
993
Simon Glasse5a0e5d2014-08-09 15:33:03 -0600994To build the current source tree, run buildman without a -b flag:
995
996 ./tools/buildman/buildman <list of things to build>
997
998This will build the current source tree for the given boards and display
999the results and errors.
1000
1001However buildman usually works on entire branches, and for that you must
1002specify a board flag:
Simon Glass6eede342014-08-09 15:32:58 -06001003
1004 ./tools/buildman/buildman -b <branch_name> <list of things to build>
1005
1006followed by (afterwards, or perhaps concurrently in another terminal):
1007
1008 ./tools/buildman/buildman -b <branch_name> -s <list of things to build>
1009
1010to see the results of the build. Rather than showing you all the output,
1011buildman just shows a summary, with red indicating that a commit introduced
1012an error and green indicating that a commit fixed an error. Use the -e
Simon Glassed966652014-08-28 09:43:43 -06001013flag to see the full errors and -l to see which boards caused which errors.
Simon Glass6eede342014-08-09 15:32:58 -06001014
Simon Glasse5a0e5d2014-08-09 15:33:03 -06001015If you really want to see build results as they happen, use -v when doing a
Simon Glass1d8104f2014-10-16 01:05:56 -06001016build (and -e to see the errors/warnings too).
Simon Glasse5a0e5d2014-08-09 15:33:03 -06001017
Simon Glass6eede342014-08-09 15:32:58 -06001018You don't need to stick around on that branch while buildman is running. It
1019checks out its own copy of the source code, so you can change branches,
1020add commits, etc. without affecting the build in progress.
1021
1022The <list of things to build> can include board names, architectures or the
1023like. There are no flags to disambiguate since ambiguities are rare. Using
1024the examples from MAKEALL:
1025
1026Examples:
1027 - build all Power Architecture boards:
1028 MAKEALL -a powerpc
1029 MAKEALL --arch powerpc
1030 MAKEALL powerpc
1031 ** buildman -b <branch> powerpc
1032 - build all PowerPC boards manufactured by vendor "esd":
1033 MAKEALL -a powerpc -v esd
1034 ** buildman -b <branch> esd
1035 - build all PowerPC boards manufactured either by "keymile" or "siemens":
1036 MAKEALL -a powerpc -v keymile -v siemens
1037 ** buildman -b <branch> keymile siemens
1038 - build all Freescale boards with MPC83xx CPUs, plus all 4xx boards:
1039 MAKEALL -c mpc83xx -v freescale 4xx
1040 ** buildman -b <branch> mpc83xx freescale 4xx
1041
1042Buildman automatically tries to use all the CPUs in your machine. If you
1043are building a lot of boards it will use one thread for every CPU core
1044it detects in your machine. This is like MAKEALL's BUILD_NBUILDS option.
1045You can use the -T flag to change the number of threads. If you are only
1046building a few boards, buildman will automatically run make with the -j
1047flag to increase the number of concurrent make tasks. It isn't normally
1048that helpful to fiddle with this option, but if you use the BUILD_NCPUS
1049option in MAKEALL then -j is the equivalent in buildman.
1050
1051Buildman puts its output in ../<branch_name> by default but you can change
1052this with the -o option. Buildman normally does out-of-tree builds: use -i
1053to disable that if you really want to. But be careful that once you have
1054used -i you pollute buildman's copies of the source tree, and you will need
1055to remove the build directory (normally ../<branch_name>) to run buildman
1056in normal mode (without -i).
1057
1058Buildman doesn't keep the output result normally, but use the -k option to
1059do this.
1060
1061Please read 'Theory of Operation' a few times as it will make a lot of
1062things clearer.
1063
1064Some options you might like are:
1065
1066 -B shows which functions are growing/shrinking in which commit - great
1067 for finding code bloat.
1068 -S shows image sizes for each commit (just an overall summary)
1069 -u shows boards that you haven't built yet
1070 --step 0 will build just the upstream commit and the last commit of your
1071 branch. This is often a quick sanity check that your branch doesn't
1072 break anything. But note this does not check bisectability!
1073
1074
Simon Glassfc3fe1c2013-04-03 11:07:16 +00001075TODO
1076====
1077
1078This has mostly be written in my spare time as a response to my difficulties
1079in testing large series of patches. Apart from tidying up there is quite a
Simon Glass1d8104f2014-10-16 01:05:56 -06001080bit of scope for improvement. Things like better error diffs and easier
Dirk Behme3e1ded12014-12-23 07:41:26 +01001081access to log files. Also it would be nice if buildman could 'hunt' for
Simon Glass1d8104f2014-10-16 01:05:56 -06001082problems, perhaps by building a few boards for each arch, or checking
1083commits for changed files and building only boards which use those files.
Simon Glassfc3fe1c2013-04-03 11:07:16 +00001084
Simon Glass8ea42102016-07-27 20:32:58 -06001085A specific problem to fix is that Ctrl-C does not exit buildman cleanly when
1086multiple builder threads are active.
Simon Glassfc3fe1c2013-04-03 11:07:16 +00001087
1088Credits
1089=======
1090
1091Thanks to Grant Grundler <grundler@chromium.org> for his ideas for improving
1092the build speed by building all commits for a board instead of the other
1093way around.
1094
1095
Simon Glassfc3fe1c2013-04-03 11:07:16 +00001096Simon Glass
1097sjg@chromium.org
1098Halloween 2012
1099Updated 12-12-12
1100Updated 23-02-13