[all-commits] [llvm/llvm-project] 52b2ba: [lldb/testsuite] Skip TestEmptyStdModule.py if usi...

[fredriss via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 52b2bae777f2a30d1ed6e87c8812bbffc4f4feeb
      https://github.com/llvm/llvm-project/commit/52b2bae777f2a30d1ed6e87c8812bbffc4f4feeb
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/commands/expression/import-std-module/empty-module/TestEmptyStdModule.py

  Log Message:
  -----------
  [lldb/testsuite] Skip TestEmptyStdModule.py if using a remote platform

The test runs `platform select host`, so it make no sense to run it
when remote debugging.


  Commit: c182be211a4d1a79390703ede8f041dcbaaf7947
      https://github.com/llvm/llvm-project/commit/c182be211a4d1a79390703ede8f041dcbaaf7947
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/functionalities/breakpoint/breakpoint_locations/main.c

  Log Message:
  -----------
  [lldb/testsuite] Tweak TestBreakpointLocations.py to pass for arm64

The test checks that we correctly set the right number of breakpoints
when breaking into an `always_inline` function. The line of this
funstion selected for this test was the return statement, but with
recent compiler, this return statement doesn't necessarily exist after
inlining, even at O0.

Switch the breakpoint to a different line of the inline function.


  Commit: 127b9d9d774dcc593cfd50eccde307dbe96097a2
      https://github.com/llvm/llvm-project/commit/127b9d9d774dcc593cfd50eccde307dbe96097a2
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/functionalities/breakpoint/hardware_breakpoints/hardware_breakpoint_on_multiple_threads/TestHWBreakMultiThread.py

  Log Message:
  -----------
  [lldb/testsuite] Apply @skipIfDarwinEmbedded to part of TestHWBreakMultiThread

The comment in the test wrongfully claimed that we support hardware
breakpoints on darwin for arm64, but we never did.


  Commit: 71db787c4583b5b05b9066509c36eb996924aeda
      https://github.com/llvm/llvm-project/commit/71db787c4583b5b05b9066509c36eb996924aeda
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/lang/cpp/thread_local/TestThreadLocal.py
    M lldb/test/API/lang/cpp/thread_local/main.cpp

  Log Message:
  -----------
  [lldb/testsuite] Rewrite TestThreadLocal.py

It was an inline test before. Clang stopped emitting line information
for the TLS initialization and the inline test didn't have a way to
break before it anymore.

This rewrites the test as a full-fldeged python test and improves the
checking of the error case to verify that the failure we are looking
for is related to the TLS setup not being complete.


  Commit: acd641c19d687c7117b08cdd568a91a381043ebb
      https://github.com/llvm/llvm-project/commit/acd641c19d687c7117b08cdd568a91a381043ebb
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/lang/objc/hidden-ivars/Makefile
    M lldb/test/API/lang/objc/hidden-ivars/TestHiddenIvars.py

  Log Message:
  -----------
  [lldb/testsuite] Slightly rework TestHiddenIvars.py

The test was stripping the binaries from the Python
code. Unfortunately, if running on darwin embedded in a context that
requires code signing, the stripping was invalidating the signature,
thus breaking the test.

This patch moves the stripping to the Makefile and resigns the
stripped binaries if required.


  Commit: 59918d3793a1136e7041b1a76f38a42cf8644474
      https://github.com/llvm/llvm-project/commit/59918d3793a1136e7041b1a76f38a42cf8644474
  Author: Fred Riss <friss at apple.com>
  Date:   2020-03-18 (Wed, 18 Mar 2020)

  Changed paths:
    M lldb/test/API/lang/objc/objc-ivar-stripped/Makefile

  Log Message:
  -----------
  [lldb/testsuite] Make TestObjCIvarStripped.py working with codesigning

This test was stripping a binary generated by Makefile.rules which is
potentially codesigned. Stripping invalidates the code signature, so
we might need to re-sign after stripping.


Compare: https://github.com/llvm/llvm-project/compare/aacf022cd577...59918d3793a1