Started 3 mo 11 days ago
Took 10 days

Aborted Build clang-d323204-gd021617bf7e-t677-b677.tar.gz (Aug 2, 2019 3:21:09 PM)

Issues

Found 1 issues:
Aborted
Waiting to resume part of lnt-test-suite-x86_64-Os-flto clang-d323204-gd021617bf7e-t677-b677.tar.gz: Jenkins is about to shut down
Waiting to resume part of lnt-test-suite-x86_64-Os-flto clang-d323204-gd021617bf7e-t677-b677.tar.gz: Jenkins is about to shut down
Waiting to resume part of lnt-test-suite-x86_64-Os-flto clang-d323204-gd021617bf7e-t677-b677.tar.gz: Jenkins is about to shut down
Ready to run at Wed Aug 07 09:23:49 PDT 2019
Sending interrupt signal to process
Aborted by Andreas Bergen
Sending interrupt signal to process
Click here to forcibly terminate running steps

Build Log

Revision: 367002
Changes
  1. Use labmaster llvm-project in common.groovy for CTMark/test suite jobs

    It should be faster than checking out the monorepo sources from GitHub. (detail)
    by paquette
  2. Update test suite jobs to use monorepo.

    Changes to relevant test generator files:

    common.groovy:
      - Kill all the SVN stuff and use the monorepo instead.
      - Use GIT_SHA/GIT_DISTANCE

    relay.groovy:
      - Set/use GIT_SHA and GIT_DISTANCE instead of LLVM_REV (detail)
    by paquette
  3. Update URL for last-good-build.properties

    The last_good_build.properties file is no longer hosted at

    http://labmaster2.local/artifacts/clang-stage1-configure-RA

    Update it to point to its new home at

    http://labmaster2.local/artifacts/clang-stage1-RA

    (This might cause some things to break, which were relying on LLVM_REV being
    set. Those things were needlessly spinning on the same compiler though, so
    they really ought to be broken anyway.) (detail)
    by paquette
  4. Removed unused llgo builder. (detail)
    by gkistanova
  5. [monorepo_build] Also write GIT_DISTANCE to the last_good_build.properties file (detail)
    by Azharuddin Mohammed
  6. Use https as vcs_protocol for the systemz builder

    This works around firewall issues we were seeing with http. (detail)
    by uweigand
  7. Slave 'lldb-x86_64-fedora': Switch to normal non-debuginfo builds again

    The racy vscode testcases have been dropped from both main and lldb-mi
    repositories anyway so there is nothing left to debug:
    [lldb-dev] [RFC] Removing lldb-mi
    https://lists.llvm.org/pipermail/lldb-dev/2019-July/015209.html (detail)
    by jankratochvil
  8. Slave 'lldb-x86_64-fedora': Reduce build resource requirements (detail)
    by jankratochvil
  9. monorepo_build.py: Update artifact name

    Incorporate the git sha and a sequentially increasing number (which can be
    generated using "git describe" and an annotated tag). (detail)
    by Azharuddin Mohammed
  10. Reland "[zorg] Add a new build script for use with monorepo"

    Summary:
    monorepo-build.py is similar in functionality to the current build.py, but
    with changes to make it work with the monorepo.

    Differential Revision: https://reviews.llvm.org/D64028 (detail)
    by Azharuddin Mohammed
  11. Revert "[zorg] Add a new build script for use with monorepo"

    This reverts commit c842a5a3fa0d0d7d00e206c47d58fba7288111d1 (r365440). (detail)
    by azharudd
  12. Fix typo

    Buiilds -> Builds (detail)
    by Azharuddin Mohammed
  13. [lldb-cmake-standalone] Pre-populate Xcode build-tree cache with Apple-lldb-Xcode

    This cache does not set install-related options. LLVM cannot install from multi-config generators. (detail)
    by stefan.graenitz
  14. Use https for the source code checkout of buildbot running on ppc64be-clang-lnt-test.

    Patch by Kamau Bridgeman.

    Differential Revision: https://reviews.llvm.org/D64221 (detail)
    by gkistanova
  15. Added MailNotifier for builders "llvm-clang-x86_64-win-fast" and "lld-x86_64-ubuntu-fast". (detail)
    by gkistanova
  16. Removed MailNotifiers for removed builders. (detail)
    by gkistanova
  17. [lldb-cmake-standalone] Pre-populate provided LLVM build-tree cache with Apple-lldb-base (detail)
    by stefan.graenitz
  18. [LLDB] getLLDBCMakeBuildFactory: New parameter testTimeout

    It also uses the new long timeout for slave 'lldb-x86_64-fedora'.

    Differential Revision: https://reviews.llvm.org/D64719 (detail)
    by jankratochvil
  19. Added vcs_protocol property to specify what protocol to use for the source code checkout. (detail)
    by gkistanova
  20. [zorg] Add a new build script for use with monorepo

    Summary:
    monorepo-build.py is similar in functionality to the currently build.py, but
    with changes to make it work with the monorepo.

    Reviewers: cmatthews

    Reviewed By: cmatthews

    Subscribers: llvm-commits

    Tags: #llvm

    Differential Revision: https://reviews.llvm.org/D64028 (detail)
    by azharudd
  21. Remove extraRecipients for sanitizer-x86_64-linux-gn

    Reviewers: eugenis

    Differential Revision: https://reviews.llvm.org/D64363 (detail)
    by Vitaly Buka
  22. Force the LibcxxAndABIBuilder to use libc++abi when building libc++

    (unless otherwise specified) (detail)
    by ericwf
  23. Move GN bot into clang.exp

    Summary:
    Master avoids IRC and emails to bots in category with .exp prefix
    It uses clang.exp to avoid creating new section in console.

    Reviewers: gkistanova, eugenis

    Differential Revision: https://reviews.llvm.org/D63909 (detail)
    by Vitaly Buka
  24. Add sync_source_lists_from_cmake.py step to GN bot (detail)
    by Vitaly Buka
  25. [zorg] Fixes for solaris11-amd64, solaris11-sparcv9  builders

    The first few days of running the Solaris 11 buildbots revealed a couple of
    problems:

    - The builders didn't honor the slaves' configured jobs property, but were
      running with the full parallelism discovered by ninja.  That massivly
      impacts other builders running on these systems, so im explicitly
      configuring jobs in the builder configs, too.

    - Along the same line, the number of paralllel links is the worst problem
      here.  I had experimented with the LLVM_PARALLEL_LINK_JOBS option in
      the past, but only now (in cmake/modules/HandleLLVMOptions.cmake)
      discovered that it's Ninja-only. I'm reducing it to 4 here as an
      experiment.

    - Unlike Solaris/x86, the Solaris/SPARC configure triplet defaults to
      sparc-sun-solaris2.11, which results in clang creating 32-bit
      binaries by default.  It turns out to be an inconsistency in
      config.guess which I'm trying to get resolved.  In the meantime, I'm
      forcing the proper triplet on both sparc and x86.

    Differential Revision: https://reviews.llvm.org/D63757 (detail)
    by ro
  26. Split check-debuginfo and check-lldb into two separate invocations. (detail)
    by Adrian Prantl
  27. Update tests to no longer expect debuginfo-tests in an LLVM build.

    <rdar://problem/51268955> (detail)
    by Adrian Prantl
  28. Add debuginfo-tests to the list of repositories needed by lldb-cmake-matrix. (detail)
    by Adrian Prantl
  29. Tabs to spaces. NFC. (detail)
    by gkistanova
  30. Allow merge build requests for lld-x86_64-win7. (detail)
    by gkistanova
  31. [GN] Report check-clang-tools error as warning (detail)
    by Vitaly Buka
  32. [lldb-cmake-standalone] Streamline labels of build stages (detail)
    by stefan.graenitz
  33. [lldb-cmake-standalone] Build and test CMake-generated Xcode project in Debug mode (detail)
    by stefan.graenitz
  34. [lldb-cmake-standalone] Build provided LLVM tree as RelWithDebInfo (detail)
    by stefan.graenitz
  35. [zorg] Add solaris11-amd64, solaris11-sparcv9 builders

    I'm working to provide two Solaris 11.4 build slaves with a clang builder
    each, one on amd64, the other on sparcv9.  I'm still working out the
    details like parallelism and max_builds, but the attached patch captures
    the basics, intended to be minimal.

    Differential Revision: https://reviews.llvm.org/D63495 (detail)
    by ro
  36. Separate options into separate array elements. (detail)
    by Adrian Prantl
  37. Run the debuginfo-tests as part of the lldb-cmake bot.

    rdar://problem/51799130 (detail)
    by Adrian Prantl
  38. Remove the debuginfo-tests from the default llvm configuration.

    I'm going to make them a separate bot where we have more control over what the host LLDB is.

    rdar://problem/51799130 (detail)
    by Adrian Prantl
  39. Add lit timeout for lldb arm/aarch64 ubuntu builders

    This patch adds a timeout interval of 200 seconds for any hanging tests. (detail)
    by omjavaid
  40. [zorg] Add lldb-arm-ubuntu builder

    This patch adds lldb arm linux builder. It ll run on staging master until tests become stable.

    Differential revision:https://reviews.llvm.org/D63441 (detail)
    by omjavaid
  41. Moved builder lld-x86_64-win7 to another machine.
    Removed slave/worker as-bldslv4. (detail)
    by gkistanova
  42. [lldb-cmake-standalone] CMake-generated Xcode project should build the LLDB.framework (detail)
    by stefan.graenitz
  43. [lldb-cmake-standalone] Invoke llvm-lit manually for CMake-generated Xcode project in order to pass --verbose flag (detail)
    by stefan.graenitz
  44. [lldb-cmake-standalone] Ninja verbose flag is -v (detail)
    by stefan.graenitz
  45. [lldb-cmake-standalone] Remove note about Xcode generator

    This was fixed with https://reviews.llvm.org/rGa0781b6e3a3a (detail)
    by stefan.graenitz
  46. [lldb-cmake-standalone] Enable tests for CMake-generated Xcode project (detail)
    by stefan.graenitz
  47. [lldb-cmake-standalone] Run check-lldb in TERM=vt100 (detail)
    by stefan.graenitz
  48. [lldb-cmake-standalone] Checkout libc++ the same way as other subprojects (detail)
    by stefan.graenitz
  49. [lldb-cmake-standalone] When invoking llmv-lit manually, make sure to build lldb-test-deps first (detail)
    by stefan.graenitz
  50. Add LLVM lab buildslaves for Windows (armv7 target).

    The ARMV7 buildslave configurations for a new hardware in LLVM lab.

    Requested for the production buildbot http://lab.llvm.org:8011

    Patch by Vlad Vereschaka.

    Differential Revision: https://reviews.llvm.org/D63103 (detail)
    by gkistanova
  51. Removed long time off-line builders and slaves.

    Builders:
    llvm-mips-linux
    perf-x86_64-penryn-O3
    clang-openbsd-amd64
    perf-x86_64-penryn-O3-polly-fast
    perf-x86_64-penryn-O3-polly-parallel-fast
    perf-x86_64-penryn-O3-polly-unprofitable
    perf-x86_64-penryn-O3-polly
    perf-x86_64-penryn-O3-polly-detect-only
    lldb-amd64-ninja-freebsd11
    clang-native-aarch64-full
    clang-bpf-build

    Slaves:
    pollyperf1, pollyperf2, pollyperf3, pollyperf4, pollyperf5, pollyperf6, pollyperf7, pollyperf11, pollyperf14, pollyperf15,
    mipsswbrd002, juno-aarch64-01, bpf-build-slave01, lldb-amd64-ninja-freebsd11, openbsd-amd64 (detail)
    by gkistanova
  52. Removed commented out builders and slaves. (detail)
    by gkistanova
  53. [lldb-cmake-standalone] Point lit to build directory not to sources (detail)
    by stefan.graenitz
  54. [lldb-cmake-standalone] Show full test output for failures (detail)
    by stefan.graenitz
  55. [lldb-cmake-standalone] Run tests for lldb-against-buildtree (detail)
    by stefan.graenitz
  56. [lldb-cmake-standalone] Avoid apostrophe in FIXME comment (detail)
    by stefan.graenitz
  57. [lldb-cmake-standalone] Fix unable to find LLDB.framework dir in Xcode build (detail)
    by stefan.graenitz
  58. [lldb-cmake-standalone] Fix LLVMConfig.cmake not found in LLVM install-tree (detail)
    by stefan.graenitz
  59. [lldb-cmake-standalone] Fix loading initial cache script (detail)
    by stefan.graenitz
  60. [lldb-cmnake-standalone] Add stage to build generated Xcode project against buildtree (detail)
    by stefan.graenitz
  61. [lldb-cmnake-standalone] Use CMake cache file, use DESTDIR for ninja install, rename folder to llvm-installtree (detail)
    by stefan.graenitz
  62. Add location for checking out debuginfo-tests (detail)
    by rnk
  63. Try to fix clang windows bot by passing projects variable into run_steps (detail)
    by rnk
  64. Run check-debuginfo on the Windows bot

    cdb tests were recently added, and this will run them continuously. (detail)
    by rnk
  65. [zorg] Add lldb-aarch64-ubuntu builder

    This patch adds lldb aarch64 linux builder. It ll run on staging master
    untill tests become stable and its consistently running green.

    Differential revision: https://reviews.llvm.org/D62473 (detail)
    by omjavaid
  66. Fix build step name (detail)
    by Vitaly Buka
  67. [zorg] Build and test LLVM with GN

    Reviewers: eugenis, morehouse, hctim, thakis, pcc

    Differential Revision: https://reviews.llvm.org/D62020 (detail)
    by Vitaly Buka
  68. builder lldb-x86_64-fedora: more verbosity

    Patch by: Konrad Kleine

    Differential revision: https://reviews.llvm.org/D62182 (detail)
    by jankratochvil
  69. Removed explicit build of lib/LLVMgold.so targer. (detail)
    by gkistanova
  70. Fix cfi-and-supported (detail)
    by Vitaly Buka
  71. [zorg][sanitizers] Update stage1_clang_at_revison to r360832 and build lld (detail)
    by Vitaly Buka
  72. [zorg] Add empty GN buildbot

    Reviewers: eugenis, morehouse, hctim

    Differential Revision: https://reviews.llvm.org/D61965 (detail)
    by Vitaly Buka
  73. [zorg] Switch between from BUILDBOT_MONO_REPO_PATH to BUILDBOT_REVISION (detail)
    by Vitaly Buka
  74. Enabled 3 builds on sanitizer-buildbot1,2
    They looks underutilized. (detail)
    by Vitaly Buka
  75. Revert "Temporary disable modules for LLDB."

    We think we figured out the issue. (detail)
    by Jonas Devlieghere
  76. Temporary disable modules for LLDB.

    We're temporarily disabling modules on the LLDB bots while we
    investigate an issue with the stage-1 host compiler. (detail)
    by Jonas Devlieghere
  77. Updated test build step name to correspond to the actual test targets. (detail)
    by gkistanova
Revision: 367002
Changes
  1. Fix runtest test-suite pgo + multisampling.

    There were are at least 3 issues when combining --pgo with
    --exec-multisample:

    * Make clean was not called between a measurement run and the next
       profiling run.
    * Make all was only called on the first measurement run.
    * Cmake variable TEST_SUITE_PROFILE_USE=OFF was not set on the
       second and later profiling runs.

    These 3 issues are fixed by this patch.

    It is also questionable whether we need to run profile collection
    multiple times. With the instrumentation-based profile collection, the
    profiles should be deterministic between different runs, and hence there
    is no value in collecting profiles multiple times.
    However, if in the future there would also be support for collecting
    profiles e.g. based on performance counter sampling, the profiles might
    be non-deterministic in such a scenario and having multiple profiling
    runs could make a lot more sense. I'm leaving that discussion for
    another time. (detail)
    by kbeyls
  2. in test_suite run_test function, opt and self.opt is same object

    Patch by Lily He

    Differential Revision: https://reviews.llvm.org/D61662 (detail)
    by danilaml
Revision: 367002
Changes
  1. Add geomean column earlier to dataout as Series in compare.py.

    This fixes the following error when using --all.

    Traceback (most recent call last):

        File "test-suite/utils/compare.py", line 374, in <module>
          print_result(data, limit_output, shorten_names, config.show_diff, sortkey)
        File "test-suite/utils/compare.py", line 227, in print_result
          dataout = add_geomean_row(d, dataout)
        File "test-suite/utils/compare.py", line 139, in add_geomean_row
          relative = values1 / values0
        File "pyvenv/lib/python2.7/site-packages/pandas/core/ops.py", line 1583, in wrapper
          result = safe_na_op(lvalues, rvalues)
        File "pyvenv/lib/python2.7/site-packages/pandas/core/ops.py", line 1533, in safe_na_op
          lambda x: op(x, rvalues))
        File "pandas/_libs/algos.pyx", line 690, in pandas._libs.algos.arrmap
        File "pyvenv/lib/python2.7/site-packages/pandas/core/ops.py", line 1533, in <lambda>
          lambda x: op(x, rvalues))
        TypeError: ufunc 'true_divide' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

    Reviewed By: paquette

    Differential Revision: https://reviews.llvm.org/D64066 (detail)
    by fhahn
  2. Skip test earlier if metric is not found in compare.py

    This speeds up processing for inputs with lots of missing metrics, by
    avoiding adding them to names.

    Reviewed By: paquette

    Differential Revision: https://reviews.llvm.org/D64065 (detail)
    by fhahn
  3. Fix more -Werror issues with obsequi. (detail)
    by aemerson
  4. Obsequi: fix implicit declaration of a function, breaks -Werror builds. (detail)
    by aemerson
  5. Enable the obsequi application for Darwin. It seems to run with no issues. (detail)
    by aemerson
  6. Fix support for user mode emulation when using cmake/lit.

    When using user mode emulation, i.e. cross-compiling programs for a
    different target and running them on a host under qemu user mode
    emulation, timeit and fpcmp should have host versions, not target
    versions.

    Running under user mode emulation had been broken for a while,
    presumably since https://reviews.llvm.org/rT341257

    I first tried an alternative approach where fpcmp would be run under qemu user
    mode emulation too.  That in itself worked, but if going for that approach, for
    orthogonality reasons, we probably should also run the other helper programs as
    if they were running on the target, i.e. also under qemu user mode emulation.
    I ran into issues with running timeit under qemu user mode emulation and also
    running RunSafely.sh under user mode emulation doesn't seem trivial.

    In the end, it seemed better to me to explicitly add a cmake option to mark
    that we're running under qemu user mode emulation, and in that mode, only aim
    to run the test/benchmark under qemu user mode emulation, rather than also all
    the helper programs (such as fpcmp, timeit, RunSafely.sh) under it (which is
    what would be needed if we just kept on using only the RUN_UNDER option for
    qemu user mode emulation.

    Differential Revision: https://reviews.llvm.org/D61597 (detail)
    by kbeyls
  7. [test-suite] Fix RISC-V Support in benchmark 1.3.0

    Summary:
    Fixed by backporting the upstream fix from here:
      https://github.com/google/benchmark/pull/833

    Reviewers: lebedev.ri

    Reviewed By: lebedev.ri

    Subscribers: asb, kito-cheng, shiva0217, rogfer01, rkruppe, llvm-commits

    Tags: #llvm

    Differential Revision: https://reviews.llvm.org/D64237 (detail)
    by lenary
  8. Revert "[test-suite] Signal error if llvm-lit was not found"

    As disscussed in post-commit review, the LIT is not specified
    in the zorg/buildbot/builders/ClangBuilder.py, so this broke
    all test-suite bots. That is a bot problem,
    i'm pulling the commit until is is resolved.

    This reverts commit r364388. (detail)
    by lebedevri
  9. [test-suite] Signal error if llvm-lit was not found

    Summary:
    I'm not sure if that was intentional or not?
    I found it quite surprising when trying to integrate a new external project.

    Reviewers: MatzeB, homerdin, hfinkel

    Reviewed By: hfinkel

    Subscribers: mgorny, llvm-commits

    Tags: #llvm

    Differential Revision: https://reviews.llvm.org/D60091 (detail)
    by lebedevri
  10. Made lambda test more robust.

    The test used a lambda that could be optimized out as a constexpr lambda in
    the C++17 mode, leading to a test faiure due to a missing symbol. This patch
    avoids that. (detail)
    by ssrivastava

Started by upstream project relay-lnt-test-suite build number 8614
originally caused by:

This run spent:

  • 57 sec waiting;
  • 10 days build duration;
  • 10 days total from scheduled to completion.
Revision: d021617bf7e1a34ac8f8d25146bea1af97580e3b
  • detached

Aborted by user Andreas Bergen

Identified problems

No identified problem

No problems were identified. If you know why this problem occurred, please add a suitable Cause for it.