gitlab: Simplify the test script
cover gitlab: Simplify the test script
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directory
It also
- moves test.py over to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docs
Simon Glass (20):
sandbox: Add documentation about required/useful packages
main: Drop show_boot_progress() prototype
buildman: Document the members of BuilderJob
bulidman: Add support for a simple build
gitlab: Use the -w option for sandbox_spl
azure: Use the -w option for sandbox_spl
gitlab: Use the --board buildman flag
gitlab: Drop the BUILDMAN variable
buildman: Update help for -d
gitlab: Drop the buildman -d flag
gitlab: Drop unnecessary if..fi
gitlab: Use -w flag for all builds
gitlab: Use bash to avoid needing a_test_which_does_not_exist
buildman: Allow ignoring warnings in the return code
gitlab: Use the buildman -W flag
gitlab: Enable test_handoff
buildman: Be more selective about which directories to remove
buildman: Allow building within a subdir of the current dir test/py: Use buildman to build U-Boot
gitlab: Simplify the exit code for test.py
.azure-pipelines.yml | 4 +- .gitlab-ci.yml | 81 +++++++++------------------------ common/main.c | 5 -- doc/arch/sandbox.rst | 10 ++++ test/py/conftest.py | 16 +++---- tools/buildman/README | 13 +++++- tools/buildman/builder.py | 51 ++++++++++++++++----- tools/buildman/builderthread.py | 34 ++++++++++---- tools/buildman/cmdline.py | 6 ++- tools/buildman/control.py | 35 ++++---------- tools/buildman/func_test.py | 46 ++++++++++++++----- tools/buildman/test.py | 20 ++++++++ 12 files changed, 186 insertions(+), 135 deletions(-)
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 1/20 sandbox: Add documentation about required/useful packages
Quite a few packages are used by sandbox or tools. Add a list of these to help people setting up for the first time.
---
doc/arch/sandbox.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)
doc/arch/sandbox.rst +10 -0
@@ -34,6 +34,16 @@ integers can only be built on 64-bit hosts. Note that standalone/API support is not available at present. +Prerequisites +------------- + +Here are some packages that are worth installing if you are doing sandbox or +tools development in U-Boot: + + python3-pytest lzma lzma-alone lz4 python3 python3-virtualenv + libssl1.0-dev + + Basic Operation ---------------
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 2/20 main: Drop show_boot_progress() prototype
In reply to Simon GlassThis is defined in bootstage.h and is not called in this file anyway. Drop it.
---
common/main.c | 5 ----- 1 file changed, 5 deletions(-)
common/main.c +0 -5
@@ -15,11 +15,6 @@#include <init.h> #include <version.h> -/* -* Board-specific Platform code can reimplement show_boot_progress () if needed -*/ -__weak void show_boot_progress (int val) {} -static void run_preboot_environment_command (void ) {char *p;
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 3/20 buildman: Document the members of BuilderJob
In reply to Simon GlassThis class has a few more members now. Add documentation for them and fix a nit in the 'commits' comment.
---
tools/buildman/builderthread.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
tools/buildman/builderthread.py +5 -1
@@ -39,11 +39,15 @@ class BuilderJob: Members: board: Board object to build - commits: List of commit options to build. + commits: List of Commit objects to build + keep_outputs: True to save build output files + step: 1 to process every commit, n to process every nth commit """ def __init__(self): self.board = None self.commits = [] + self.keep_outputs = False + self.step = 1 class ResultThread(threading.Thread):
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 4/20 bulidman: Add support for a simple build
In reply to Simon GlassIt is useful to run a simple build and put all the output in a single directory. Add a -w option to support this.
---
tools/buildman/README | 11 ++++++++++ tools/buildman/builder.py | 15 +++++++++++-- tools/buildman/builderthread.py | 28 ++++++++++++++++------- tools/buildman/cmdline.py | 2 ++ tools/buildman/control.py | 10 ++++++++- tools/buildman/func_test.py | 39 +++++++++++++++++++++++++++++---- 6 files changed, 90 insertions(+), 15 deletions(-)
tools/buildman/README +11 -0
@@ -1056,6 +1056,17 @@ toolchain. For example: buildman -O clang-7 --board sandbox +Doing a simple build +==================== + +In some cases you just want to build a single board and get the full output, use +the -w option, for example: + + buildman -o /tmp/build --board sandbox -w + +This will write the full build into /tmp/build including object files. + + Other options =============
tools/buildman/builder.py +13 -2
@@ -174,6 +174,8 @@ class Builder: in_tree: Build U-Boot in-tree instead of specifying an output directory separate from the source code. This option is really only useful for testing in-tree builds. + work_in_output: Use the output directory as the work directory and + don't write to a separate output directory. Private members: _base_board_dict: Last-summarised Dict of boards
@@ -224,7 +226,7 @@ class Builder: no_subdirs=False, full_path=False, verbose_build=False, incremental=False, per_board_out_dir=False, config_only=False, squash_config_y=False, - warnings_as_errors=False): + warnings_as_errors=False, work_in_output=False): """Create a new Builder object Args:
@@ -250,10 +252,15 @@ class Builder: config_only: Only configure each build, don't build it squash_config_y: Convert CONFIG options with the value 'y' to '1' warnings_as_errors: Treat all compiler warnings as errors + work_in_output: Use the output directory as the work directory and + don't write to a separate output directory. """ self.toolchains = toolchains self.base_dir = base_dir - self._working_dir = os.path.join(base_dir, '.bm-work') + if work_in_output: + self._working_dir = base_dir + else: + self._working_dir = os.path.join(base_dir, '.bm-work') self.threads = [] self.do_make = self.Make self.gnu_make = gnu_make
@@ -280,6 +287,7 @@ class Builder: self.config_only = config_only self.squash_config_y = squash_config_y self.config_filenames = BASE_CONFIG_FILENAMES + self.work_in_output = work_in_output if not self.squash_config_y: self.config_filenames += EXTRA_CONFIG_FILENAMES
@@ -1474,6 +1482,8 @@ class Builder: Args: thread_num: Number of thread to check. """ + if self.work_in_output: + return self._working_dir return os.path.join(self._working_dir, '%02d' % thread_num) def _PrepareThread(self, thread_num, setup_git):
@@ -1571,6 +1581,7 @@ class Builder: job.board = brd job.commits = commits job.keep_outputs = keep_outputs + job.work_in_output = self.work_in_output job.step = self._step self.queue.put(job)
tools/buildman/builderthread.py +20 -8
@@ -42,12 +42,15 @@ class BuilderJob: commits: List of Commit objects to build keep_outputs: True to save build output files step: 1 to process every commit, n to process every nth commit + work_in_output: Use the output directory as the work directory and + don't write to a separate output directory. """ def __init__(self): self.board = None self.commits = [] self.keep_outputs = False self.step = 1 + self.work_in_output = False class ResultThread(threading.Thread):
@@ -118,7 +121,7 @@ class BuilderThread(threading.Thread): **kwargs) def RunCommit(self, commit_upto, brd, work_dir, do_config, config_only, - force_build, force_build_failures): + force_build, force_build_failures, work_in_output): """Build a particular commit. If the build is already done, and we are not forcing a build, we skip
@@ -133,6 +136,8 @@ class BuilderThread(threading.Thread): force_build: Force a build even if one was previously done force_build_failures: Force a bulid if the previous result showed failure + work_in_output: Use the output directory as the work directory and + don't write to a separate output directory. Returns: tuple containing:
@@ -143,7 +148,7 @@ class BuilderThread(threading.Thread): # self.Make() below, in the event that we do a build. result = command.CommandResult() result.return_code = 0 - if self.builder.in_tree: + if work_in_output or self.builder.in_tree: out_dir = work_dir else: if self.per_board_out_dir:
@@ -265,14 +270,18 @@ class BuilderThread(threading.Thread): result.out_dir = out_dir return result, do_config - def _WriteResult(self, result, keep_outputs): + def _WriteResult(self, result, keep_outputs, work_in_output): """Write a built result to the output directory. Args: result: CommandResult object containing result to write keep_outputs: True to store the output binaries, False to delete them + work_in_output: Use the output directory as the work directory and + don't write to a separate output directory. """ + if work_in_output: + return # Fatal error if result.return_code < 0: return
@@ -434,7 +443,8 @@ class BuilderThread(threading.Thread): result, request_config = self.RunCommit(commit_upto, brd, work_dir, do_config, self.builder.config_only, force_build or self.builder.force_build, - self.builder.force_build_failures) + self.builder.force_build_failures, + work_in_output=job.work_in_output) failed = result.return_code or result.stderr did_config = do_config if failed and not do_config:
@@ -442,7 +452,8 @@ class BuilderThread(threading.Thread): # with a reconfig. if self.builder.force_config_on_failure: result, request_config = self.RunCommit(commit_upto, - brd, work_dir, True, False, True, False) + brd, work_dir, True, False, True, False, + work_in_output=job.work_in_output) did_config = True if not self.builder.force_reconfig: do_config = request_config
@@ -481,15 +492,16 @@ class BuilderThread(threading.Thread): raise ValueError('Interrupt') # We have the build results, so output the result - self._WriteResult(result, job.keep_outputs) + self._WriteResult(result, job.keep_outputs, job.work_in_output) self.builder.out_queue.put(result) else: # Just build the currently checked-out build result, request_config = self.RunCommit(None, brd, work_dir, True, self.builder.config_only, True, - self.builder.force_build_failures) + self.builder.force_build_failures, + work_in_output=job.work_in_output) result.commit_upto = 0 - self._WriteResult(result, job.keep_outputs) + self._WriteResult(result, job.keep_outputs, job.work_in_output) self.builder.out_queue.put(result) def run(self):
tools/buildman/cmdline.py +2 -0
@@ -106,6 +106,8 @@ def ParseArgs(): default=False, help='Show build results while the build progresses') parser.add_option('-V', '--verbose-build', action='store_true', default=False, help='Run make with V=1, logging all output') + parser.add_option('-w', '--work-in-output', action='store_true', + default=False, help='Use the output directory as the work directory') parser.add_option('-x', '--exclude', dest='exclude', type='string', action='append', help='Specify a list of boards to exclude, separated by comma')
tools/buildman/control.py +9 -1
@@ -263,6 +263,13 @@ def DoBuildman(options, args, toolchains=None, make_func=None, boards=None, str = ("No commits found to process in branch '%s': " "set branch's upstream or use -c flag" % options.branch) sys.exit(col.Color(col.RED, str)) + if options.work_in_output: + if len(selected) != 1: + sys.exit(col.Color(col.RED, + '-w can only be used with a single board')) + if count != 1: + sys.exit(col.Color(col.RED, + '-w can only be used with a single commit')) # Read the metadata from the commits. First look@the upstream commit, # then the ones in the branch. We would like to do something like
@@ -334,7 +341,8 @@ def DoBuildman(options, args, toolchains=None, make_func=None, boards=None, per_board_out_dir=options.per_board_out_dir, config_only=options.config_only, squash_config_y=not options.preserve_config_y, - warnings_as_errors=options.warnings_as_errors) + warnings_as_errors=options.warnings_as_errors, + work_in_output=options.work_in_output) builder.force_config_on_failure = not options.quick if make_func: builder.do_make = make_func
tools/buildman/func_test.py +35 -4
@@ -16,6 +16,7 @@ import control import gitutil import terminal import toolchain +import tools settings_data = ''' # Buildman settings file
@@ -208,7 +209,7 @@ class TestFunctional(unittest.TestCase): def tearDown(self): shutil.rmtree(self._base_dir) - shutil.rmtree(self._output_dir) + #shutil.rmtree(self._output_dir) def setupToolchains(self): self._toolchains = toolchain.Toolchains()
@@ -218,12 +219,12 @@ class TestFunctional(unittest.TestCase): return command.RunPipe([[self._buildman_pathname] + list(args)], capture=True, capture_stderr=True) - def _RunControl(self, *args, **kwargs): + def _RunControl(self, *args, clean_dir=False, boards=None): sys.argv = [sys.argv[0]] + list(args) options, args = cmdline.ParseArgs() result = control.DoBuildman(options, args, toolchains=self._toolchains, - make_func=self._HandleMake, boards=self._boards, - clean_dir=kwargs.get('clean_dir', True)) + make_func=self._HandleMake, boards=boards or self._boards, + clean_dir=clean_dir) self._builder = control.builder return result
@@ -397,6 +398,12 @@ class TestFunctional(unittest.TestCase): combined='Test configuration complete') elif stage == 'build': stderr = '' + out_dir = '' + for arg in args: + if arg.startswith('O='): + out_dir = arg[2:] + fname = os.path.join(cwd or '', out_dir, 'u-boot') + tools.WriteFile(fname, b'U-Boot') if type(commit) is not str: stderr = self._error.get((brd.target, commit.sequence)) if stderr:
@@ -535,3 +542,27 @@ class TestFunctional(unittest.TestCase): with self.assertRaises(SystemExit): self._RunControl('-b', self._test_branch, '-o', os.path.join(os.getcwd(), 'test')) + + def testWorkInOutput(self): + """Test the -w option which should write directly to the output dir""" + board_list = board.Boards() + board_list.AddBoard(board.Board(*boards[0])) + self._RunControl('-o', self._output_dir, '-w', clean_dir=False, + boards=board_list) + self.assertTrue( + os.path.exists(os.path.join(self._output_dir, 'u-boot'))) + + def testWorkInOutputFail(self): + """Test the -w option failures""" + with self.assertRaises(SystemExit) as e: + self._RunControl('-o', self._output_dir, '-w', clean_dir=False) + self.assertIn("single board", str(e.exception)) + self.assertFalse( + os.path.exists(os.path.join(self._output_dir, 'u-boot'))) + + board_list = board.Boards() + board_list.AddBoard(board.Board(*boards[0])) + with self.assertRaises(SystemExit) as e: + self._RunControl('-b', self._test_branch, '-o', self._output_dir, + '-w', clean_dir=False, boards=board_list) + self.assertIn("single commit", str(e.exception))
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 5/20 gitlab: Use the -w option for sandbox_spl
In reply to Simon GlassAvoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
---
.gitlab-ci.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
.gitlab-ci.yml +2 -2
@@ -167,10 +167,10 @@ Run binman, buildman, dtoc and patman testsuites: virtualenv -p /usr/bin/python3 /tmp/venv; . /tmp/venv/bin/activate; pip install pyelftools; - export UBOOT_TRAVIS_BUILD_DIR=/tmp/.bm-work/sandbox_spl; + export UBOOT_TRAVIS_BUILD_DIR=/tmp/sandbox_spl; export PYTHONPATH="${UBOOT_TRAVIS_BUILD_DIR}/scripts/dtc/pylibfdt"; export PATH="${UBOOT_TRAVIS_BUILD_DIR}/scripts/dtc:${PATH}"; - ./tools/buildman/buildman -o /tmp -P sandbox_spl; + ./tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w sandbox_spl; ./tools/binman/binman --toolpath ${UBOOT_TRAVIS_BUILD_DIR}/tools test; ./tools/buildman/buildman -t; ./tools/dtoc/dtoc -t;
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 5/20 gitlab: Use the -w option for sandbox_spl
Quoted 8 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
What I'd like to see is GitLab / Azure / Travis changing in-sync with these steps, rather than one commit for each (and Travis getting out of sync). Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/c40d1dd6/attachment.sig>
PATCH 6/20 azure: Use the -w option for sandbox_spl
In reply to Simon GlassAvoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
---
.azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
.azure-pipelines.yml +2 -2
@@ -136,10 +136,10 @@ jobs: virtualenv -p /usr/bin/python3 /tmp/venv . /tmp/venv/bin/activate pip install pyelftools - export UBOOT_TRAVIS_BUILD_DIR=/tmp/.bm-work/sandbox_spl + export UBOOT_TRAVIS_BUILD_DIR=/tmp/sandbox_spl export PYTHONPATH=${UBOOT_TRAVIS_BUILD_DIR}/scripts/dtc/pylibfdt export PATH=${UBOOT_TRAVIS_BUILD_DIR}/scripts/dtc:${PATH} - ./tools/buildman/buildman -o /tmp -P sandbox_spl + ./tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w sandbox_spl ./tools/binman/binman --toolpath ${UBOOT_TRAVIS_BUILD_DIR}/tools test ./tools/buildman/buildman -t ./tools/dtoc/dtoc -t
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 7/20 gitlab: Use the --board buildman flag
In reply to Simon GlassThe current method of selecting the board to build is a bit error-prone, e.g. with "^sandbox$" it actually builds 5 boards (all of those in the sandbox architecture).
Use the (newish) --board flag instead, to get the same result.
---
.gitlab-ci.yml | 47 ++++++++++++++++++++++++----------------------- 1 file changed, 24 insertions(+), 23 deletions(-)
.gitlab-ci.yml +24 -23
@@ -31,9 +31,10 @@ stages: # use clang only do one configuration. - if [[ "${BUILDMAN}" != "" ]]; then ret=0; - tools/buildman/buildman -o /tmp -P -E ${BUILDMAN} ${OVERRIDE}|| ret=$?; + tools/buildman/buildman -o /tmp -P -E --board ${BUILDMAN} ${OVERRIDE} + || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -sdeP ${BUILDMAN}; + tools/buildman/buildman -o /tmp -sdeP --board ${BUILDMAN}; exit $ret; fi; fi
@@ -181,14 +182,14 @@ sandbox test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox" - BUILDMAN: "^sandbox$" + BUILDMAN: "sandbox" <<: *buildman_and_testpy_dfn sandbox with clang test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox" - BUILDMAN: "^sandbox$" + BUILDMAN: "sandbox" OVERRIDE: "-O clang-7" <<: *buildman_and_testpy_dfn
@@ -196,7 +197,7 @@ sandbox_spl test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox_spl" - BUILDMAN: "^sandbox_spl$" + BUILDMAN: "sandbox_spl" TEST_PY_TEST_SPEC: "test_ofplatdata" <<: *buildman_and_testpy_dfn
@@ -205,14 +206,14 @@ evb-ast2500 test.py: variables: TEST_PY_BD: "evb-ast2500" TEST_PY_ID: "--id qemu" - BUILDMAN: "^evb-ast2500$" + BUILDMAN: "evb-ast2500" <<: *buildman_and_testpy_dfn sandbox_flattree test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox_flattree" - BUILDMAN: "^sandbox_flattree$" + BUILDMAN: "sandbox_flattree" <<: *buildman_and_testpy_dfn vexpress_ca15_tc2 test.py:
@@ -220,7 +221,7 @@ vexpress_ca15_tc2 test.py: variables: TEST_PY_BD: "vexpress_ca15_tc2" TEST_PY_ID: "--id qemu" - BUILDMAN: "^vexpress_ca15_tc2$" + BUILDMAN: "vexpress_ca15_tc2" <<: *buildman_and_testpy_dfn vexpress_ca9x4 test.py:
@@ -228,7 +229,7 @@ vexpress_ca9x4 test.py: variables: TEST_PY_BD: "vexpress_ca9x4" TEST_PY_ID: "--id qemu" - BUILDMAN: "^vexpress_ca9x4$" + BUILDMAN: "vexpress_ca9x4" <<: *buildman_and_testpy_dfn integratorcp_cm926ejs test.py:
@@ -237,7 +238,7 @@ integratorcp_cm926ejs test.py: TEST_PY_BD: "integratorcp_cm926ejs" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "^integratorcp_cm926ejs$" + BUILDMAN: "integratorcp_cm926ejs" <<: *buildman_and_testpy_dfn qemu_arm test.py:
@@ -245,7 +246,7 @@ qemu_arm test.py: variables: TEST_PY_BD: "qemu_arm" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_arm$" + BUILDMAN: "qemu_arm" <<: *buildman_and_testpy_dfn qemu_arm64 test.py:
@@ -253,7 +254,7 @@ qemu_arm64 test.py: variables: TEST_PY_BD: "qemu_arm64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_arm64$" + BUILDMAN: "qemu_arm64" <<: *buildman_and_testpy_dfn qemu_mips test.py:
@@ -261,7 +262,7 @@ qemu_mips test.py: variables: TEST_PY_BD: "qemu_mips" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_mips$" + BUILDMAN: "qemu_mips" <<: *buildman_and_testpy_dfn qemu_mipsel test.py:
@@ -269,7 +270,7 @@ qemu_mipsel test.py: variables: TEST_PY_BD: "qemu_mipsel" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_mipsel$" + BUILDMAN: "qemu_mipsel" <<: *buildman_and_testpy_dfn qemu_mips64 test.py:
@@ -277,7 +278,7 @@ qemu_mips64 test.py: variables: TEST_PY_BD: "qemu_mips64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_mips64$" + BUILDMAN: "qemu_mips64" <<: *buildman_and_testpy_dfn qemu_mips64el test.py:
@@ -285,7 +286,7 @@ qemu_mips64el test.py: variables: TEST_PY_BD: "qemu_mips64el" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu_mips64el$" + BUILDMAN: "qemu_mips64el" <<: *buildman_and_testpy_dfn qemu-ppce500 test.py:
@@ -293,7 +294,7 @@ qemu-ppce500 test.py: variables: TEST_PY_BD: "qemu-ppce500" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu-ppce500$" + BUILDMAN: "qemu-ppce500" <<: *buildman_and_testpy_dfn qemu-riscv64 test.py:
@@ -301,7 +302,7 @@ qemu-riscv64 test.py: variables: TEST_PY_BD: "qemu-riscv64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu-riscv64$" + BUILDMAN: "qemu-riscv64" <<: *buildman_and_testpy_dfn qemu-x86 test.py:
@@ -309,7 +310,7 @@ qemu-x86 test.py: variables: TEST_PY_BD: "qemu-x86" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu-x86$" + BUILDMAN: "qemu-x86" <<: *buildman_and_testpy_dfn qemu-x86_64 test.py:
@@ -317,7 +318,7 @@ qemu-x86_64 test.py: variables: TEST_PY_BD: "qemu-x86_64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "^qemu-x86_64$" + BUILDMAN: "qemu-x86_64" <<: *buildman_and_testpy_dfn zynq_zc702 test.py:
@@ -326,7 +327,7 @@ zynq_zc702 test.py: TEST_PY_BD: "zynq_zc702" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "^zynq_zc702$" + BUILDMAN: "zynq_zc702" <<: *buildman_and_testpy_dfn xilinx_versal_virt test.py:
@@ -335,7 +336,7 @@ xilinx_versal_virt test.py: TEST_PY_BD: "xilinx_versal_virt" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "^xilinx_versal_virt$" + BUILDMAN: "xilinx_versal_virt" <<: *buildman_and_testpy_dfn xtfpga test.py:
@@ -344,5 +345,5 @@ xtfpga test.py: TEST_PY_BD: "xtfpga" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "^xtfpga$" + BUILDMAN: "xtfpga" <<: *buildman_and_testpy_dfn
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 7/20 gitlab: Use the --board buildman flag
Quoted 3 lines · Simon Glass
The current method of selecting the board to build is a bit error-prone, e.g. with "^sandbox$" it actually builds 5 boards (all of those in the sandbox architecture).
OK, hold up. You've reminded me of something that's been broken for a while and I forgot to follow up on. Which is that at some point we stopped applying regex to either only the defconfig name, or stopped checking that as the first one. The example in question used to only build sandbox_defconfig, but now (and for some time) has built all of arch or soc sandbox.
That said, I'm not sure that at the end of the day I'm opposed to the --board flag instead.
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/cfc534d3/attachment.sig>
PATCH 7/20 gitlab: Use the --board buildman flag
Hi Tom,
Quoted 7 lines · Tom Rini
Quoted 3 lines · Simon Glass
The current method of selecting the board to build is a bit error-prone, e.g. with "^sandbox$" it actually builds 5 boards (all of those in the sandbox architecture).
Quoted 6 lines
OK, hold up. You've reminded me of something that's been broken for a while and I forgot to follow up on. Which is that at some point we stopped applying regex to either only the defconfig name, or stopped checking that as the first one. The example in question used to only build
sandbox_defconfig, but now (and for some time) has built all of arch or soc sandbox.
Are you sure we even used just the defconfig name? I don't recall that. Do you know when this was?
Since 'sandbox' is an arch name this will build anything sandbox board, and I think that is intended.
Quoted 3 lines
That said, I'm not sure that at the end of the day I'm opposed to the --board flag instead.
OK.
Regards,
Simon
PATCH 8/20 gitlab: Drop the BUILDMAN variable
In reply to Simon GlassThis is not needed now since we use the same name as the pytests.
---
.gitlab-ci.yml | 27 +++------------------------ 1 file changed, 3 insertions(+), 24 deletions(-)
.gitlab-ci.yml +3 -24
@@ -29,12 +29,12 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. - - if [[ "${BUILDMAN}" != "" ]]; then + - if [[ "${TEST_PY_BD}" != "" ]]; then ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${BUILDMAN} ${OVERRIDE} + tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -sdeP --board ${BUILDMAN}; + tools/buildman/buildman -o /tmp -sdeP --board ${TEST_PY_BD}; exit $ret; fi; fi
@@ -182,14 +182,12 @@ sandbox test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox" - BUILDMAN: "sandbox" <<: *buildman_and_testpy_dfn sandbox with clang test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox" - BUILDMAN: "sandbox" OVERRIDE: "-O clang-7" <<: *buildman_and_testpy_dfn
@@ -197,7 +195,6 @@ sandbox_spl test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox_spl" - BUILDMAN: "sandbox_spl" TEST_PY_TEST_SPEC: "test_ofplatdata" <<: *buildman_and_testpy_dfn
@@ -206,14 +203,12 @@ evb-ast2500 test.py: variables: TEST_PY_BD: "evb-ast2500" TEST_PY_ID: "--id qemu" - BUILDMAN: "evb-ast2500" <<: *buildman_and_testpy_dfn sandbox_flattree test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox_flattree" - BUILDMAN: "sandbox_flattree" <<: *buildman_and_testpy_dfn vexpress_ca15_tc2 test.py:
@@ -221,7 +216,6 @@ vexpress_ca15_tc2 test.py: variables: TEST_PY_BD: "vexpress_ca15_tc2" TEST_PY_ID: "--id qemu" - BUILDMAN: "vexpress_ca15_tc2" <<: *buildman_and_testpy_dfn vexpress_ca9x4 test.py:
@@ -229,7 +223,6 @@ vexpress_ca9x4 test.py: variables: TEST_PY_BD: "vexpress_ca9x4" TEST_PY_ID: "--id qemu" - BUILDMAN: "vexpress_ca9x4" <<: *buildman_and_testpy_dfn integratorcp_cm926ejs test.py:
@@ -238,7 +231,6 @@ integratorcp_cm926ejs test.py: TEST_PY_BD: "integratorcp_cm926ejs" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "integratorcp_cm926ejs" <<: *buildman_and_testpy_dfn qemu_arm test.py:
@@ -246,7 +238,6 @@ qemu_arm test.py: variables: TEST_PY_BD: "qemu_arm" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_arm" <<: *buildman_and_testpy_dfn qemu_arm64 test.py:
@@ -254,7 +245,6 @@ qemu_arm64 test.py: variables: TEST_PY_BD: "qemu_arm64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_arm64" <<: *buildman_and_testpy_dfn qemu_mips test.py:
@@ -262,7 +252,6 @@ qemu_mips test.py: variables: TEST_PY_BD: "qemu_mips" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_mips" <<: *buildman_and_testpy_dfn qemu_mipsel test.py:
@@ -270,7 +259,6 @@ qemu_mipsel test.py: variables: TEST_PY_BD: "qemu_mipsel" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_mipsel" <<: *buildman_and_testpy_dfn qemu_mips64 test.py:
@@ -278,7 +266,6 @@ qemu_mips64 test.py: variables: TEST_PY_BD: "qemu_mips64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_mips64" <<: *buildman_and_testpy_dfn qemu_mips64el test.py:
@@ -286,7 +273,6 @@ qemu_mips64el test.py: variables: TEST_PY_BD: "qemu_mips64el" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu_mips64el" <<: *buildman_and_testpy_dfn qemu-ppce500 test.py:
@@ -294,7 +280,6 @@ qemu-ppce500 test.py: variables: TEST_PY_BD: "qemu-ppce500" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu-ppce500" <<: *buildman_and_testpy_dfn qemu-riscv64 test.py:
@@ -302,7 +287,6 @@ qemu-riscv64 test.py: variables: TEST_PY_BD: "qemu-riscv64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu-riscv64" <<: *buildman_and_testpy_dfn qemu-x86 test.py:
@@ -310,7 +294,6 @@ qemu-x86 test.py: variables: TEST_PY_BD: "qemu-x86" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu-x86" <<: *buildman_and_testpy_dfn qemu-x86_64 test.py:
@@ -318,7 +301,6 @@ qemu-x86_64 test.py: variables: TEST_PY_BD: "qemu-x86_64" TEST_PY_TEST_SPEC: "not sleep" - BUILDMAN: "qemu-x86_64" <<: *buildman_and_testpy_dfn zynq_zc702 test.py:
@@ -327,7 +309,6 @@ zynq_zc702 test.py: TEST_PY_BD: "zynq_zc702" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "zynq_zc702" <<: *buildman_and_testpy_dfn xilinx_versal_virt test.py:
@@ -336,7 +317,6 @@ xilinx_versal_virt test.py: TEST_PY_BD: "xilinx_versal_virt" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "xilinx_versal_virt" <<: *buildman_and_testpy_dfn xtfpga test.py:
@@ -345,5 +325,4 @@ xtfpga test.py: TEST_PY_BD: "xtfpga" TEST_PY_TEST_SPEC: "not sleep" TEST_PY_ID: "--id qemu" - BUILDMAN: "xtfpga" <<: *buildman_and_testpy_dfn
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 9/20 buildman: Update help for -d
In reply to Simon GlassThis help is a bit ambiguous. It only does anything if showing size changes with -S. Update the help and the function comments.
---
tools/buildman/builder.py | 6 +++--- tools/buildman/cmdline.py | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)
tools/buildman/builder.py +3 -3
@@ -337,7 +337,7 @@ class Builder: show_errors: True to show summarised error/warning info show_sizes: Show size deltas - show_detail: Show detail for each board + show_detail: Show size delta detail for each board if show_sizes show_bloat: Show detail for each function list_error_boards: Show the boards which caused each error/warning show_config: Show config deltas
@@ -1000,7 +1000,7 @@ class Builder: board.target board_dict: Dict containing boards for which we built this commit, keyed by board.target. The value is an Outcome object. - show_detail: Show detail for each board + show_detail: Show size delta detail for each board show_bloat: Show detail for each function """ arch_list = {}
@@ -1117,7 +1117,7 @@ class Builder: environment: Dictionary keyed by environment variable, Each value is the value of environment variable. show_sizes: Show image size deltas - show_detail: Show detail for each board + show_detail: Show size delta detail for each board if show_sizes show_bloat: Show detail for each function show_config: Show config changes show_environment: Show environment changes
tools/buildman/cmdline.py +1 -1
@@ -31,7 +31,7 @@ def ParseArgs(): help='Reconfigure for every commit (disable incremental build)') parser.add_option('-d', '--detail', dest='show_detail', action='store_true', default=False, - help='Show detailed information for each board in summary') + help='Show detailed size delta for each board in the -S summary') parser.add_option('-D', '--config-only', action='store_true', default=False, help="Don't build, just configure each commit") parser.add_option('-e', '--show_errors', action='store_true',
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 10/20 gitlab: Drop the buildman -d flag
In reply to Simon GlassThis has no effect since -S is not given also. Drop it.
---
.gitlab-ci.yml | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
.gitlab-ci.yml +5 -5
@@ -34,7 +34,7 @@ stages: tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -sdeP --board ${TEST_PY_BD}; + tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fi; fi
@@ -65,7 +65,7 @@ build all 32bit ARM platforms: - ret=0; ./tools/buildman/buildman -o /tmp -P -E arm -x aarch64 || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - ./tools/buildman/buildman -o /tmp -sdeP; + ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -79,7 +79,7 @@ build all 64bit ARM platforms: - ret=0; ./tools/buildman/buildman -o /tmp -P -E aarch64 || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - ./tools/buildman/buildman -o /tmp -sdeP; + ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -90,7 +90,7 @@ build all PowerPC platforms: - ret=0; ./tools/buildman/buildman -o /tmp -P -E powerpc || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - ./tools/buildman/buildman -o /tmp -sdeP; + ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -101,7 +101,7 @@ build all other platforms: - ret=0; ./tools/buildman/buildman -o /tmp -P -E -x arm,powerpc || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - ./tools/buildman/buildman -o /tmp -sdeP; + ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 11/20 gitlab: Drop unnecessary if..fi
In reply to Simon GlassSince TEST_PY_BD is always defined we can drop this check.
---
.gitlab-ci.yml | 30 +++++++++++++----------------- 1 file changed, 13 insertions(+), 17 deletions(-)
.gitlab-ci.yml +13 -17
@@ -29,14 +29,12 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. - - if [[ "${TEST_PY_BD}" != "" ]]; then - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; - exit $ret; - fi; + - ret=0; + tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} + || ret=$?; + if [[ $ret -ne 0 && $ret -ne 129 ]]; then + tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; + exit $ret; fi # "not a_test_which_does_not_exist" is a dummy -k parameter which will # never prevent any test from running. That way, we can always pass
@@ -48,15 +46,13 @@ stages: - export UBOOT_TRAVIS_BUILD_DIR=/tmp/.bm-work/${TEST_PY_BD}; export PATH=/opt/qemu/bin:/tmp/uboot-test-hooks/bin:${PATH}; export PYTHONPATH=/tmp/uboot-test-hooks/py/travis-ci; - if [[ "${TEST_PY_BD}" != "" ]]; then - ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} - -k "${TEST_PY_TEST_SPEC:-not a_test_which_does_not_exist}" - --build-dir "$UBOOT_TRAVIS_BUILD_DIR"; - ret=$?; - if [[ $ret -ne 0 ]]; then - exit $ret; - fi; - fi; + ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} + -k "${TEST_PY_TEST_SPEC:-not a_test_which_does_not_exist}" + --build-dir "$UBOOT_TRAVIS_BUILD_DIR"; + ret=$?; + if [[ $ret -ne 0 ]]; then + exit $ret; + fi build all 32bit ARM platforms: tags: [ 'all' ]
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 12/20 gitlab: Use -w flag for all builds
In reply to Simon GlassAvoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
.gitlab-ci.yml +4 -5
@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fi # "not a_test_which_does_not_exist" is a dummy -k parameter which will
@@ -43,8 +43,7 @@ stages: - virtualenv -p /usr/bin/python3 /tmp/venv - . /tmp/venv/bin/activate - pip install -r test/py/requirements.txt - - export UBOOT_TRAVIS_BUILD_DIR=/tmp/.bm-work/${TEST_PY_BD}; - export PATH=/opt/qemu/bin:/tmp/uboot-test-hooks/bin:${PATH}; + - export PATH=/opt/qemu/bin:/tmp/uboot-test-hooks/bin:${PATH}; export PYTHONPATH=/tmp/uboot-test-hooks/py/travis-ci; ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} -k "${TEST_PY_TEST_SPEC:-not a_test_which_does_not_exist}"
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 12/20 gitlab: Use -w flag for all builds
Quoted 22 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +0
-0@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration.
.gitlab-ci.yml stages: · line 38
Quoted 10 lines
.gitlab-ci.yml +3
-3+ - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fi
The repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/0668d69c/attachment.sig>
PATCH 12/20 gitlab: Use -w flag for all builds
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;
Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Regards,
Simon
PATCH 12/20 gitlab: Use -w flag for all builds
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..
At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200315/67f4f870/attachment.sig>
PATCH 12/20 gitlab: Use -w flag for all builds
Hi Tom,
Quoted 47 lines · Tom Rini
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Quoted 7 lines
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Yes I see. But in this case we are only building a single board so there should be no difference.
BTW it looks like we are not using the -l flag for the 'multiple build' case, so we shouldn't see the (BOARD1,BOARD2) thing.
Regards,
Simon
PATCH 12/20 gitlab: Use -w flag for all builds
Quoted 55 lines · Simon Glass
Hi Tom,
Quoted 52 lines · Tom Rini
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..
Quoted 8 lines
Quoted 2 lines
At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Yes I see. But in this case we are only building a single board so there should be no difference.
BTW it looks like we are not using the -l flag for the 'multiple build' case, so we shouldn't see the (BOARD1,BOARD2) thing.
Perhaps this is a good example of why I'm asking for the commit message to be clearer about the tests being changed :) But isn't this also the same build area as the general builds?
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200315/c03855e0/attachment.sig>
PATCH 12/20 gitlab: Use -w flag for all builds
Hi Tom,
Quoted 63 lines · Tom Rini
Quoted 61 lines · Simon Glass
Hi Tom,
Quoted 47 lines · Tom Rini
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Quoted 7 lines
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Yes I see. But in this case we are only building a single board so there should be no difference.
Quoted 6 lines
Quoted 2 lines
BTW it looks like we are not using the -l flag for the 'multiple build' case, so we shouldn't see the (BOARD1,BOARD2) thing.
Perhaps this is a good example of why I'm asking for the commit message to be clearer about the tests being changed :) But isn't this also the same build area as the general builds?
Yes I'll update the commit subjects.
This is the buildman_and_testpy_template section, which is only used to build a single board, so far as I can tell.
Regards,
Simon
PATCH 12/20 gitlab: Use -w flag for all builds
Quoted 69 lines · Simon Glass
Hi Tom,
Quoted 66 lines · Tom Rini
Quoted 63 lines · Simon Glass
Hi Tom,
Quoted 47 lines · Tom Rini
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Quoted 7 lines
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Yes I see. But in this case we are only building a single board so there should be no difference.
BTW it looks like we are not using the -l flag for the 'multiple build' case, so we shouldn't see the (BOARD1,BOARD2) thing.
Quoted 8 lines
Quoted 3 lines
Perhaps this is a good example of why I'm asking for the commit message to be clearer about the tests being changed :) But isn't this also the same build area as the general builds?
Yes I'll update the commit subjects.
This is the
buildman_and_testpy_templatesection, which is only used to build a single board, so far as I can tell.
OK then yes, the cases where we are building a single board it would make sense to not do the second build. Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200315/b6d05483/attachment.sig>
PATCH 12/20 gitlab: Use -w flag for all builds
Hi Tom,
Quoted 75 lines · Tom Rini
Quoted 73 lines · Simon Glass
Hi Tom,
Quoted 69 lines · Tom Rini
Quoted 55 lines · Simon Glass
Hi Tom,
Quoted 52 lines · Tom Rini
Quoted 42 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 28 lines · Simon Glass
Avoid needing to know about the internal .bm-work directory, by passing the -w flag to buildman.
Also drop the repeated call to buildman since the first one should show all the expected output. We only need to use -s if we are building multiple boards and want the errors to be coalesced. In this case we are only building a single board.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-).gitlab-ci.yml +3
-2@@ -29,11 +29,11 @@ stages: script: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. + - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - ret=0; - tools/buildman/buildman -o /tmp -P -E --board ${TEST_PY_BD} ${OVERRIDE} - || ret=$?; + tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E + --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?;Quoted 8 lines
.gitlab-ci.yml stages: · line 38
Quoted 4 lines
.gitlab-ci.yml +0
-1if [[ $ret -ne 0 && $ret -ne 129 ]]; then - tools/buildman/buildman -o /tmp -seP --board ${TEST_PY_BD}; exit $ret; fiThe repeated call is so that when we have a CI error from buildman the error is at the bottom of the output and we don't have to hunt for it, so I'm not sure this is a developer-friendly change.
Quoted 2 lines
I don't quite get this, since the two buildman calls are one after the other. What difference do you see in the output?
Instead of having the errors be throughout the page we see something like:
arch: +BOARD1 BOARD2
+(BOARD1,BOARD2) error..Quoted 8 lines
Quoted 2 lines
At the bottom of the page. So you open the failed build link, hit "End" and there's where to find what to fix.
Yes I see. But in this case we are only building a single board so there should be no difference.
BTW it looks like we are not using the -l flag for the 'multiple build' case, so we shouldn't see the (BOARD1,BOARD2) thing.
Perhaps this is a good example of why I'm asking for the commit message to be clearer about the tests being changed :) But isn't this also the same build area as the general builds?
Quoted 7 lines
Quoted 4 lines
Yes I'll update the commit subjects.
This is the
buildman_and_testpy_templatesection, which is only used to build a single board, so far as I can tell.OK then yes, the cases where we are building a single board it would make sense to not do the second build. Thanks!
OK.
Also I think I got my 'kea' gitlab runner working. Do you want to try to add it to the group runners?
Regards,
Simon
PATCH 13/20 gitlab: Use bash to avoid needing a_test_which_does_not_exist
In reply to Simon GlassBash allows for variables to expand only if non-empty:
$ var=test
$ echo ${var:+"$var"}
test
$ echo ${var:+"-k $var"}
-k test
$ var=
$ echo ${var:+"-k $var"}Use this feature to avoid the workaround.
---
.gitlab-ci.yml | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-)
.gitlab-ci.yml +1 -5
@@ -36,17 +36,13 @@ stages: if [[ $ret -ne 0 && $ret -ne 129 ]]; then exit $ret; fi - # "not a_test_which_does_not_exist" is a dummy -k parameter which will - # never prevent any test from running. That way, we can always pass - # "-k something" even when $TEST_PY_TEST_SPEC doesnt need a custom - # value. - virtualenv -p /usr/bin/python3 /tmp/venv - . /tmp/venv/bin/activate - pip install -r test/py/requirements.txt - export PATH=/opt/qemu/bin:/tmp/uboot-test-hooks/bin:${PATH}; export PYTHONPATH=/tmp/uboot-test-hooks/py/travis-ci; ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} - -k "${TEST_PY_TEST_SPEC:-not a_test_which_does_not_exist}" + ${TEST_PY_TEST_SPEC:+"-k ${TEST_PY_TEST_SPEC}"} --build-dir "$UBOOT_TRAVIS_BUILD_DIR"; ret=$?; if [[ $ret -ne 0 ]]; then
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 13/20 gitlab: Use bash to avoid needing a_test_which_does_not_exist
Quoted 31 lines · Simon Glass
Bash allows for variables to expand only if non-empty:
$ var=test $ echo ${var:+"$var"} test $ echo ${var:+"-k $var"} -k test $ var= $ echo ${var:+"-k $var"}Use this feature to avoid the workaround.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-).gitlab-ci.yml +0
-4@@ -36,17 +36,13 @@ stages: if [[ $ret -ne 0 && $ret -ne 129 ]]; then exit $ret; fi - # "not a_test_which_does_not_exist" is a dummy -k parameter which will - # never prevent any test from running. That way, we can always pass - # "-k something" even when $TEST_PY_TEST_SPEC doesnt need a custom - # value. - virtualenv -p /usr/bin/python3 /tmp/venv
.gitlab-ci.yml stages: · line 48
Quoted 10 lines
.gitlab-ci.yml +1
-1- . /tmp/venv/bin/activate - pip install -r test/py/requirements.txt - export PATH=/opt/qemu/bin:/tmp/uboot-test-hooks/bin:${PATH}; export PYTHONPATH=/tmp/uboot-test-hooks/py/travis-ci; ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} - -k "${TEST_PY_TEST_SPEC:-not a_test_which_does_not_exist}" + ${TEST_PY_TEST_SPEC:+"-k ${TEST_PY_TEST_SPEC}"} --build-dir "$UBOOT_TRAVIS_BUILD_DIR"; ret=$?; if [[ $ret -ne 0 ]]; then
Please change the comment to note that other less-than-obvious bash feature being used here, thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/c59dd63c/attachment.sig>
PATCH 14/20 buildman: Allow ignoring warnings in the return code
In reply to Simon GlassSometimes we don't want to get an error with warnings. Add a -W option to support this. If buildman detects warnings (and no error) it will return an exit code of 0 (success).
---
tools/buildman/README | 2 +- tools/buildman/cmdline.py | 2 ++ tools/buildman/control.py | 2 +- 3 files changed, 4 insertions(+), 2 deletions(-)
tools/buildman/README +1 -1
@@ -1079,7 +1079,7 @@ When doing builds, Buildman's return code will reflect the overall result: 0 (success) No errors or warnings found 128 Errors found - 129 Warnings found + 129 Warnings found (use -W to return 0 in this case) How to change from MAKEALL
tools/buildman/cmdline.py +2 -0
@@ -108,6 +108,8 @@ def ParseArgs(): default=False, help='Run make with V=1, logging all output') parser.add_option('-w', '--work-in-output', action='store_true', default=False, help='Use the output directory as the work directory') + parser.add_option('-W', '--ignore-warnings', action='store_true', + default=False, help='Return success even if there are warnings') parser.add_option('-x', '--exclude', dest='exclude', type='string', action='append', help='Specify a list of boards to exclude, separated by comma')
tools/buildman/control.py +1 -1
@@ -386,6 +386,6 @@ def DoBuildman(options, args, toolchains=None, make_func=None, boards=None, options.keep_outputs, options.verbose) if fail: return 128 - elif warned: + elif warned and not options.ignore_warnings: return 129 return 0
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 15/20 gitlab: Use the buildman -W flag
In reply to Simon GlassIt doesn't seem to make sense to tell buildman to report warning as errors (thus ensuring there are no warnings) and then ignore the warnings.
We can simplify the logic now that we can tell buildman to ignore warnings.
---
.gitlab-ci.yml | 24 ++++++++++-------------- 1 file changed, 10 insertions(+), 14 deletions(-)
.gitlab-ci.yml +10 -14
@@ -30,12 +30,8 @@ stages: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - - ret=0; - tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E - --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then - exit $ret; - fi + - tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -W + --board ${TEST_PY_BD} ${OVERRIDE} - virtualenv -p /usr/bin/python3 /tmp/venv - . /tmp/venv/bin/activate - pip install -r test/py/requirements.txt
@@ -54,8 +50,8 @@ build all 32bit ARM platforms: stage: world build script: - ret=0; - ./tools/buildman/buildman -o /tmp -P -E arm -x aarch64 || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then + ./tools/buildman/buildman -o /tmp -PW arm -x aarch64 || ret=$?; + if [[ $ret -ne 0 ]]; then ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -68,8 +64,8 @@ build all 64bit ARM platforms: - . /tmp/venv/bin/activate - pip install pyelftools - ret=0; - ./tools/buildman/buildman -o /tmp -P -E aarch64 || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then + ./tools/buildman/buildman -o /tmp -PW aarch64 || ret=$?; + if [[ $ret -ne 0 ]]; then ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -79,8 +75,8 @@ build all PowerPC platforms: stage: world build script: - ret=0; - ./tools/buildman/buildman -o /tmp -P -E powerpc || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then + ./tools/buildman/buildman -o /tmp -PW powerpc || ret=$?; + if [[ $ret -ne 0 ]]; then ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
@@ -90,8 +86,8 @@ build all other platforms: stage: world build script: - ret=0; - ./tools/buildman/buildman -o /tmp -P -E -x arm,powerpc || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then + ./tools/buildman/buildman -o /tmp -PW -x arm,powerpc || ret=$?; + if [[ $ret -ne 0 ]]; then ./tools/buildman/buildman -o /tmp -seP; exit $ret; fi;
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 15/20 gitlab: Use the buildman -W flag
Quoted 21 lines · Simon Glass
It doesn't seem to make sense to tell buildman to report warning as errors (thus ensuring there are no warnings) and then ignore the warnings.
We can simplify the logic now that we can tell buildman to ignore warnings.
Signed-off-by Simon Glass <[email protected]>---
.gitlab-ci.yml | 24 ++++++++++-------------- 1 file changed, 10 insertions(+), 14 deletions(-).gitlab-ci.yml +0
-1@@ -30,12 +30,8 @@ stages: # From buildman, exit code 129 means warnings only. If we've been asked to # use clang only do one configuration. - export UBOOT_TRAVIS_BUILD_DIR=/tmp/${TEST_PY_BD} - - ret=0;
.gitlab-ci.yml stages: · line 37
Quoted 10 lines
.gitlab-ci.yml +2
-5- tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -E - --board ${TEST_PY_BD} ${OVERRIDE} || ret=$?; - if [[ $ret -ne 0 && $ret -ne 129 ]]; then - exit $ret; - fi + - tools/buildman/buildman -o ${UBOOT_TRAVIS_BUILD_DIR} -w -W + --board ${TEST_PY_BD} ${OVERRIDE} - virtualenv -p /usr/bin/python3 /tmp/venv - . /tmp/venv/bin/activate - pip install -r test/py/requirements.txt
Maybe this whole bit of logic needs some re-thinking. We pass -E to get warnings-as-errors so I think it's better to drop this new flag, fix the test to just check for non-zero and a Fixes tag for 329f5ef51d2e8de79fe2e846f2c2905da9530a71 as that should have done this in the first place I think. Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/a43aa0df/attachment.sig>
PATCH 16/20 gitlab: Enable test_handoff
In reply to Simon GlassEnsure that this SPL test runs on gitlab.
---
.gitlab-ci.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
.gitlab-ci.yml +1 -1
@@ -182,7 +182,7 @@ sandbox_spl test.py: tags: [ 'all' ] variables: TEST_PY_BD: "sandbox_spl" - TEST_PY_TEST_SPEC: "test_ofplatdata" + TEST_PY_TEST_SPEC: "test_ofplatdata or test_handoff" <<: *buildman_and_testpy_dfn evb-ast2500 test.py:
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 17/20 buildman: Be more selective about which directories to remove
In reply to Simon GlassAt present buildman removes any directory it doesn't intend to write output into. This is overly expansive since if the output directory happens to be somewhere with existing files, they may be removed. Using an existing directory for buildman is not a good practice, but since the result might be catastrophic, it is best to guard against it.
A previous commit[1] fixed this by refusing to write to a subdirectory of the current directory, assumed to have U-Boot source code. But we can do better by only removing directories that look like the ones buildman creates.
Update the code to do this and add a test.
[1] 409fc029c40 tools: buildman: Don't use the working dir as build dir
---
tools/buildman/builder.py | 27 ++++++++++++++++++++++----- tools/buildman/test.py | 20 ++++++++++++++++++++ 2 files changed, 42 insertions(+), 5 deletions(-)
tools/buildman/builder.py +22 -5
@@ -485,6 +485,7 @@ class Builder: if self.commits: commit = self.commits[commit_upto] subject = commit.subject.translate(trans_valid_chars) + # See _GetOutputSpaceRemovals() which parses this name commit_dir = ('%02d_of_%02d_g%s_%s' % (commit_upto + 1, self.commit_count, commit.hash, subject[:20])) elif not self.no_subdirs:
@@ -1525,12 +1526,15 @@ class Builder: for thread in range(max_threads): self._PrepareThread(thread, setup_git) - def _PrepareOutputSpace(self): + def _GetOutputSpaceRemovals(self): """Get the output directories ready to receive files. - We delete any output directories which look like ones we need to - create. Having left over directories is confusing when the user wants - to check the output manually. + Figure out what needs to be deleted in the output directory before it + can be used. We only delete old buildman directories which have the + expected name pattern. See _GetOutputDir(). + + Returns: + List of full paths of directories to remove """ if not self.commits: return
@@ -1541,7 +1545,20 @@ class Builder: to_remove = [] for dirname in glob.glob(os.path.join(self.base_dir, '*')): if dirname not in dir_list: - to_remove.append(dirname) + leaf = dirname[len(self.base_dir) + 1:] + m = re.match('[0-9]+_of_[0-9]+_g[0-9a-f]+_.*', leaf) + if m: + to_remove.append(dirname) + return to_remove + + def _PrepareOutputSpace(self): + """Get the output directories ready to receive files. + + We delete any output directories which look like ones we need to + create. Having left over directories is confusing when the user wants + to check the output manually. + """ + to_remove = self._GetOutputSpaceRemovals() if to_remove: Print('Removing %d old build directories' % len(to_remove), newline=False)
tools/buildman/test.py +20 -0
@@ -22,6 +22,7 @@ import commit import terminal import test_util import toolchain +import tools use_network = True
@@ -469,6 +470,25 @@ class TestBuild(unittest.TestCase): self.assertEqual('HOSTCC=clang CC=clang', tc.GetEnvArgs(toolchain.VAR_MAKE_ARGS)) + def testPrepareOutputSpace(self): + def _Touch(fname): + tools.WriteFile(os.path.join(base_dir, fname), b'') + + base_dir = tempfile.mkdtemp() + + # Add various files that we want removed and left alone + to_remove = ['01_of_22_g0982734987_title', '102_of_222_g92bf_title', + '01_of_22_g2938abd8_title'] + to_leave = ['something_else', '01-something.patch', '01_of_22_another'] + for name in to_remove + to_leave: + _Touch(name) + + build = builder.Builder(self.toolchains, base_dir, None, 1, 2) + build.commits = self.commits + build.commit_count = len(commits) + result = set(build._GetOutputSpaceRemovals()) + expected = set([os.path.join(base_dir, f) for f in to_remove]) + self.assertEqual(expected, result) if __name__ == "__main__": unittest.main()
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 18/20 buildman: Allow building within a subdir of the current dir
In reply to Simon GlassThis is useful in some situations, in particular with -w and when building in-tree. Now that we are more careful about what we remove in _PrepareOutputSpace(), it should be save to relax this restriction.
Update the progress information also so it is clear what buildman is doing. Remove files can take a long time.
---
tools/buildman/builder.py | 3 ++- tools/buildman/control.py | 23 ----------------------- tools/buildman/func_test.py | 9 --------- 3 files changed, 2 insertions(+), 33 deletions(-)
tools/buildman/builder.py +2 -1
@@ -1560,10 +1560,11 @@ class Builder: """ to_remove = self._GetOutputSpaceRemovals() if to_remove: - Print('Removing %d old build directories' % len(to_remove), + Print('Removing %d old build directories...' % len(to_remove), newline=False) for dirname in to_remove: shutil.rmtree(dirname) + Print('done') def BuildBoards(self, commits, board_selected, keep_outputs, verbose): """Build all commits for a list of boards
tools/buildman/control.py +0 -23
@@ -85,28 +85,6 @@ def ShowActions(series, why_selected, boards_selected, builder, options, for warning in board_warnings: print(col.Color(col.YELLOW, warning)) -def CheckOutputDir(output_dir): - """Make sure that the output directory is not within the current directory - - If we try to use an output directory which is within the current directory - (which is assumed to hold the U-Boot source) we may end up deleting the - U-Boot source code. Detect this and print an error in this case. - - Args: - output_dir: Output directory path to check - """ - path = os.path.realpath(output_dir) - cwd_path = os.path.realpath('.') - while True: - if os.path.realpath(path) == cwd_path: - Print("Cannot use output directory '%s' since it is within the current directory '%s'" % - (path, cwd_path)) - sys.exit(1) - parent = os.path.dirname(path) - if parent == path: - break - path = parent - def ShowToolchainInfo(boards, toolchains, print_arch, print_prefix): """Show information about a the tool chain used by one or more boards
@@ -331,7 +309,6 @@ def DoBuildman(options, args, toolchains=None, make_func=None, boards=None, output_dir = os.path.join(options.output_dir, dirname) if clean_dir and os.path.exists(output_dir): shutil.rmtree(output_dir) - CheckOutputDir(output_dir) builder = Builder(toolchains, output_dir, options.git_dir, options.threads, options.jobs, gnu_make=gnu_make, checkout=True, show_unknown=options.show_unknown, step=options.step,
tools/buildman/func_test.py +0 -9
@@ -534,15 +534,6 @@ class TestFunctional(unittest.TestCase): self.assertEqual(self._builder.count, self._total_builds) self.assertEqual(self._builder.fail, 0) - def testBadOutputDir(self): - """Test building with an output dir the same as out current dir""" - self._test_branch = '/__dev/__testbranch' - with self.assertRaises(SystemExit): - self._RunControl('-b', self._test_branch, '-o', os.getcwd()) - with self.assertRaises(SystemExit): - self._RunControl('-b', self._test_branch, '-o', - os.path.join(os.getcwd(), 'test')) - def testWorkInOutput(self): """Test the -w option which should write directly to the output dir""" board_list = board.Boards()
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 19/20 test/py: Use buildman to build U-Boot
In reply to Simon GlassIt is a pain to have to set the ARCH and CROSS_COMPILE environment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.
Remove the manual 'make' logic in test/py and use buildman instead.
---
test/py/conftest.py | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-)
test/py/conftest.py +6 -10
@@ -141,17 +141,13 @@ def pytest_configure(config): if config.getoption('build'): if build_dir != source_dir: - o_opt = 'O=%s' % build_dir + dest_args = ['-o', build_dir, '-w'] else: - o_opt = '' - cmds = ( - ['make', o_opt, '-s', board_type + '_defconfig'], - ['make', o_opt, '-s', '-j8'], - ) - with log.section('make'): - runner = log.get_runner('make', sys.stdout) - for cmd in cmds: - runner.run(cmd, cwd=source_dir) + dest_args = ['-i'] + cmd = ['buildman', '--board', board_type] + dest_args + with log.section('buildman'): + runner = log.get_runner('buildman', sys.stdout) + runner.run(cmd, cwd=source_dir) runner.close() log.status_pass('OK')
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
PATCH 19/20 test/py: Use buildman to build U-Boot
Quoted 6 lines · Simon Glass
It is a pain to have to set the ARCH and
CROSS_COMPILEenvironment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.Remove the manual 'make' logic in test/py and use buildman instead.
I far prefer using make here; this requires zero setup of buildman (e.g. the config file and specific toolchains), and so it much *less* of a pain.
PATCH 19/20 test/py: Use buildman to build U-Boot
Quoted 10 lines · Stephen Warren
Quoted 6 lines · Simon Glass
It is a pain to have to set the ARCH and
CROSS_COMPILEenvironment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.Remove the manual 'make' logic in test/py and use buildman instead.
I far prefer using make here; this requires zero setup of buildman (e.g. the config file and specific toolchains), and so it much *less* of a pain.
I have to agree here. Keeping our test suite as dependency-free as possible is important. But... that's also not what's going on in the code. We don't set ARCH from what I can see, and of course don't use it. We don't set the CROSS_COMPILER from the snippet in question, only the output directory. Today, looking at the Travis/GitLab CI scripts we don't even build via test.py but rather buildman prior to calling test.py. And I don't think I saw that changing in this series either.
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/e2ab1525/attachment.sig>
PATCH 19/20 test/py: Use buildman to build U-Boot
Hi Tom,
Quoted 13 lines · Tom Rini
Quoted 10 lines · Stephen Warren
Quoted 6 lines · Simon Glass
It is a pain to have to set the ARCH and
CROSS_COMPILEenvironment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.Remove the manual 'make' logic in test/py and use buildman instead.
I far prefer using make here; this requires zero setup of buildman (e.g. the config file and specific toolchains), and so it much *less* of a pain.
Quoted 7 lines
I have to agree here. Keeping our test suite as dependency-free as possible is important. But... that's also not what's going on in the code. We don't set ARCH from what I can see, and of course don't use it. We don't set the
CROSS_COMPILERfrom the snippet in question, only the output directory. Today, looking at the Travis/GitLab CI scripts we don't even build viatest.pybut rather buildman prior to callingtest.py. And I don't think I saw that changing in this series either.
I mean that to run pytest I have to do:
PATH=$PATH:tools/buildman ARCH=buildman -a zynq_zybo CROSS_COMPILE=buildman -A zynq_zybo \test/py/test.py -B zynq_zybo --id sjg-zynq_zybo --build-dir ../current/zynq_zybo --build
which is a bit of a pain.
With this change I can do:
test/py/test.py -B zynq_zybo --id sjg-zynq_zybo --build-dir ../current/zynq_zybo --build
Regards,
Simon
PATCH 19/20 test/py: Use buildman to build U-Boot
Quoted 31 lines · Simon Glass
Hi Tom,
Quoted 13 lines · Tom Rini
Quoted 10 lines · Stephen Warren
Quoted 6 lines · Simon Glass
It is a pain to have to set the ARCH and
CROSS_COMPILEenvironment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.Remove the manual 'make' logic in test/py and use buildman instead.
I far prefer using make here; this requires zero setup of buildman (e.g. the config file and specific toolchains), and so it much *less* of a pain.
Quoted 7 lines
I have to agree here. Keeping our test suite as dependency-free as possible is important. But... that's also not what's going on in the code. We don't set ARCH from what I can see, and of course don't use it. We don't set the
CROSS_COMPILERfrom the snippet in question, only the output directory. Today, looking at the Travis/GitLab CI scripts we don't even build viatest.pybut rather buildman prior to callingtest.py. And I don't think I saw that changing in this series either.I mean that to run pytest I have to do:
PATH=$PATH:tools/buildman ARCH=
buildman -a zynq_zyboCROSS_COMPILE=buildman -A zynq_zybo\test/py/test.py-Bzynq_zybo--id sjg-zynq_zybo --build-dir ../current/zynq_zybo --build
Quoted 6 lines
which is a bit of a pain.
With this change I can do:
test/py/test.py-Bzynq_zybo--id sjg-zynq_zybo --build-dir ../current/zynq_zybo --build
Right. The commit message isn't clear as the CI loops build the board with buildman first. Second, we don't use ARCH= when building U-Boot, so we could just drop that from buildman I suspect. Third, no, I think it's important to NOT require buildman to be builder here and setting CROSS_COMPILE in the environment is fine and makes integration with other systems easier. Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200312/650cf31a/attachment.sig>
PATCH 19/20 test/py: Use buildman to build U-Boot
In reply to Stephen WarrenHi Stephen,
Quoted 11 lines · Stephen Warren
Quoted 6 lines · Simon Glass
It is a pain to have to set the ARCH and
CROSS_COMPILEenvironment variables when using test.py's --build option. It is possible to get these using the -A and -a options from buildman. But it seems better to just use buildman to do the build.Remove the manual 'make' logic in test/py and use buildman instead.
I far prefer using make here; this requires zero setup of buildman (e.g. the config file and specific toolchains), and so it much *less* of a pain.
For people who don't have an existing setup though, it is tricky. And having to set those environment variables is a hassle.
'buildman --fetch-arch all' will download and set up toolchains.
Perhaps we could make this optional?
Regards,
Simon
PATCH 20/20 gitlab: Simplify the exit code for test.py
In reply to Simon GlassIt seems unnecessary to read the exit code and then check it again. Drop this and just let the test.py provide the exit code directly.
---
.gitlab-ci.yml | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-)
.gitlab-ci.yml +1 -5
@@ -39,11 +39,7 @@ stages: export PYTHONPATH=/tmp/uboot-test-hooks/py/travis-ci; ./test/py/test.py --bd ${TEST_PY_BD} ${TEST_PY_ID} ${TEST_PY_TEST_SPEC:+"-k ${TEST_PY_TEST_SPEC}"} - --build-dir "$UBOOT_TRAVIS_BUILD_DIR"; - ret=$?; - if [[ $ret -ne 0 ]]; then - exit $ret; - fi + --build-dir "$UBOOT_TRAVIS_BUILD_DIR" build all 32bit ARM platforms: tags: [ 'all' ]
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
cover gitlab: Simplify the test script
In reply to Simon GlassQuoted 9 lines · Simon Glass
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directory
Quoted 4 lines
It also
- movestest.pyover to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docs
One general comment is that while it's clear from this series that you're focusing on test.py invocation most of the time, a lot of the commit messages aren't clear that you're changing the buildman_and_testpy_template stanza.
Also, while you're keeping Azure in sync (as it's largely based on the GitLab logic), Travis is being left out. But the GitLab logic is based on the Travis logic and as much as I dislike Travis for being slow and having network induced failures, it's still a valid and supported CI flow that we can't drop, so please keep it in sync with these improvements, thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200309/50e99e93/attachment.sig>
cover gitlab: Simplify the test script
Hi Tom,
Quoted 17 lines · Tom Rini
Quoted 9 lines · Simon Glass
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directoryQuoted 4 lines
It also
- movestest.pyover to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docs
Quoted 4 lines
One general comment is that while it's clear from this series that you're focusing on
test.pyinvocation most of the time, a lot of the commit messages aren't clear that you're changing thebuildman_and_testpy_templatestanza.
I'm not sure what you are looking for there. Do you want the commit message to mention which part of the gitlab file is being changed?
Quoted 7 lines
Also, while you're keeping Azure in sync (as it's largely based on the GitLab logic), Travis is being left out. But the GitLab logic is based on the Travis logic and as much as I dislike Travis for being slow and having network induced failures, it's still a valid and supported CI flow that we can't drop, so please keep it in sync with these improvements, thanks!
OK.
Regards,
Simon
cover gitlab: Simplify the test script
Quoted 20 lines · Simon Glass
Hi Tom,
Quoted 17 lines · Tom Rini
Quoted 13 lines · Simon Glass
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directoryIt also
- movestest.pyover to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docs
Quoted 7 lines
Quoted 4 lines
One general comment is that while it's clear from this series that you're focusing on
test.pyinvocation most of the time, a lot of the commit messages aren't clear that you're changing thebuildman_and_testpy_templatestanza.I'm not sure what you are looking for there. Do you want the commit message to mention which part of the gitlab file is being changed?
I mean a lot of places say something like "Change gitlab ..." but aren't changing any of the world builds, only test.py stuff, so I would prefer "Change gitlab when using test.py ..."
Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200315/0b3dd1b1/attachment.sig>
cover gitlab: Simplify the test script
Hi Tom,
Quoted 27 lines · Tom Rini
Quoted 25 lines · Simon Glass
Hi Tom,
Quoted 21 lines · Tom Rini
Quoted 9 lines · Simon Glass
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directoryQuoted 4 lines
It also
- movestest.pyover to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docsOne general comment is that while it's clear from this series that you're focusing on
test.pyinvocation most of the time, a lot of the commit messages aren't clear that you're changing thebuildman_and_testpy_templatestanza.
Quoted 6 lines
Quoted 2 lines
I'm not sure what you are looking for there. Do you want the commit message to mention which part of the gitlab file is being changed?
I mean a lot of places say something like "Change gitlab ..." but aren't changing any of the world builds, only
test.pystuff, so I would prefer "Change gitlab when usingtest.py..."
OK I will try. I've got the changes to all three environments in a single commit at present, so we don't end up with double the commis. Or would you prefer it split out?
One more thing...I notice that gitlab and azure use 'set -ex' to avoid needing to check errors in the script. Is is possible for travis to do that do, or is there some restriction?
Quoted 4 lines
Thanks!
--
Tom
Regards,
Simon
cover gitlab: Simplify the test script
Quoted 33 lines · Simon Glass
Hi Tom,
Quoted 30 lines · Tom Rini
Quoted 27 lines · Simon Glass
Hi Tom,
Quoted 17 lines · Tom Rini
Quoted 9 lines · Simon Glass
At present there are several things in the gitlab script which work around limitations in buildman. With a few small feature additions these can be removed.
This series adds some new features to buildman and simplifies the script:
- Option to run a single build in a specified output directory
- Allow ignoring warnings
- Removes a restriction on the build output directoryQuoted 4 lines
It also
- movestest.pyover to use buildman for the --build option
- makes one change to azure since the same approach should be possible there
- fixes a few minor problems noticed in main and sandbox docsQuoted 4 lines
One general comment is that while it's clear from this series that you're focusing on
test.pyinvocation most of the time, a lot of the commit messages aren't clear that you're changing thebuildman_and_testpy_templatestanza.I'm not sure what you are looking for there. Do you want the commit message to mention which part of the gitlab file is being changed?
Quoted 7 lines
Quoted 3 lines
I mean a lot of places say something like "Change gitlab ..." but aren't changing any of the world builds, only
test.pystuff, so I would prefer "Change gitlab when usingtest.py..."OK I will try. I've got the changes to all three environments in a single commit at present, so we don't end up with double the commis. Or would you prefer it split out?
Doing all CIs in step makes sense. So something like...:
Azure/GitLab/Travis: Change test.py and buildman ...
Quoted 3 lines
One more thing...I notice that gitlab and azure use 'set -ex' to avoid needing to check errors in the script. Is is possible for travis to do that do, or is there some restriction?
Not sure. When in doubt I always hack the heck out of .travis.yml to just be a testcase of what I want to figure out, push and see. Thanks!
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 659 bytes Desc: not available URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200315/991df2ba/attachment.sig>