GitLab

We get a reference to memcpy@plt rather than memcpy

dynamic ghci can't load .a files

This is more efficient than having to wait for all threads to finish
before each 'alone' test.

With the exception of
    nbody      dph-nbody-copy-opt
    nbody      dph-nbody-vseg-opt
    quickhull  dph-quickhull-copy-opt
    quickhull  dph-quickhull-vseg-opt
    words      dph-words-copy-opt
    words      dph-words-vseg-opt
which are marked as 'high_memory_usage'

They are mutable values, so altering them affects other tests too

This makes it possible to share source files between tests, without
having the .o/.hi files overlap

I'm not sure whether it makes more sense to make this change, or to
change the build system to build -prof -dynamic libraries. But at least
this change suffices to make the two match, so the tests don't all fail
due to missing libraries.

tibbe authored May 31, 2011 and Simon Marlow committed Oct 09, 2012

This test framework feature is inspired by a similar feature in LLVM.
The programmer writes a bit of Cmm

    #include "Cmm.h"

    // Large memcpy's should lower to calls.
    callMemcpy
    {
      W_ dst, src;
      prim %memcpy(dst "ptr", src "ptr", 1024, 4) [];
    }

and asserts what the generated assembly should look like, modulo
register naming.

    callMemcpy:
    movq  ; Move arguments into place
    movq
    movl
    movl
    call memcpy

Patch edited and updated by Simon Marlow, and I also added a test for
unrolling memcpy and a simple constant-propagation test.

(technically, broken if dynamic-by-default is enabled)