Daniel Sanders | dc21330 | 2016-08-09 11:50:53 +0000 | [diff] [blame] | 1 | ======================================== |
| 2 | Compiler-rt Testing Infrastructure Guide |
| 3 | ======================================== |
| 4 | |
| 5 | .. contents:: |
| 6 | :local: |
| 7 | |
| 8 | Overview |
| 9 | ======== |
| 10 | |
| 11 | This document is the reference manual for the compiler-rt modifications to the |
| 12 | testing infrastructure. Documentation for the infrastructure itself can be found at |
| 13 | :ref:`llvm_testing_guide`. |
| 14 | |
| 15 | LLVM testing infrastructure organization |
| 16 | ======================================== |
| 17 | |
| 18 | The compiler-rt testing infrastructure contains regression tests which are run |
| 19 | as part of the usual ``make check-all`` and are expected to always pass -- they |
| 20 | should be run before every commit. |
| 21 | |
| 22 | Quick start |
| 23 | =========== |
| 24 | |
| 25 | The regressions tests are in the "compiler-rt" module and are normally checked |
| 26 | out in the directory ``llvm/projects/compiler-rt/test``. Use ``make check-all`` |
| 27 | to run the regression tests after building compiler-rt. |
| 28 | |
| 29 | REQUIRES, XFAIL, etc. |
| 30 | --------------------- |
| 31 | |
| 32 | Sometimes it is necessary to restrict a test to a specific target or mark it as |
| 33 | an "expected fail" or XFAIL. This is normally achieved using ``REQUIRES:`` or |
| 34 | ``XFAIL:`` with a substring of LLVM's default target triple. Unfortunately, the |
| 35 | behaviour of this is somewhat quirky in compiler-rt. There are two main |
| 36 | pitfalls to avoid. |
| 37 | |
| 38 | The first pitfall is that these directives perform a substring match on the |
| 39 | triple and as such ``XFAIL: mips`` affects more triples than expected. For |
| 40 | example, ``mips-linux-gnu``, ``mipsel-linux-gnu``, ``mips64-linux-gnu``, and |
| 41 | ``mips64el-linux-gnu`` will all match a ``XFAIL: mips`` directive. Including a |
| 42 | trailing ``-`` such as in ``XFAIL: mips-`` can help to mitigate this quirk but |
| 43 | even that has issues as described below. |
| 44 | |
| 45 | The second pitfall is that the default target triple is often inappropriate for |
| 46 | compiler-rt tests since compiler-rt tests may be compiled for multiple targets. |
| 47 | For example, a typical build on an ``x86_64-linux-gnu`` host will often run the |
| 48 | tests for both x86_64 and i386. In this situation ``XFAIL: x86_64`` will mark |
| 49 | both the x86_64 and i386 tests as an expected failure while ``XFAIL: i386`` |
| 50 | will have no effect at all. |
| 51 | |
| 52 | To remedy both pitfalls, compiler-rt tests provide a feature string which can |
| 53 | be used to specify a single target. This string is of the form |
| 54 | ``target-is-${arch}`` where ``${arch}}`` is one of the values from the |
| 55 | following lines of the CMake output:: |
| 56 | |
| 57 | -- Compiler-RT supported architectures: x86_64;i386 |
| 58 | -- Builtin supported architectures: i386;x86_64 |
| 59 | |
| 60 | So for example ``XFAIL: target-is-x86_64`` will mark a test as expected to fail |
| 61 | on x86_64 without also affecting the i386 test and ``XFAIL: target-is-i386`` |
| 62 | will mark a test as expected to fail on i386 even if the default target triple |
| 63 | is ``x86_64-linux-gnu``. Directives that use these ``target-is-${arch}`` string |
| 64 | require exact matches so ``XFAIL: target-is-mips``, |
| 65 | ``XFAIL: target-is-mipsel``, ``XFAIL: target-is-mips64``, and |
| 66 | ``XFAIL: target-is-mips64el`` all refer to different MIPS targets. |