testsuite/libffi.bhaible/README

This package contains a test suite for libffi.

This test suite can be compiled with a C compiler. No need for 'expect'
or some other package that is often not installed.

The test suite consists of 81 C functions, each with a different signature.
* test-call verifies that calling each function directly produces the same
  results as calling the function indirectly through 'ffi_call'.
* test-callback verifies that calling each function directly produces the same
  results as calling a function that is a callback (object build by
  'ffi_prep_closure_loc') and simulates the original function.

Each direct or indirect invocation should produce one line of output to
stdout. A correct output consists of paired lines, such as

void f(void):
void f(void):
int f(void):->99
int f(void):->99
int f(int):(1)->2
int f(int):(1)->2
int f(2*int):(1,2)->3
int f(2*int):(1,2)->3
...

The Makefile then creates two files:
* failed-call, which consists of the non-paired lines of output of
  'test-call',
* failed-callback, which consists of the non-paired lines of output of
  'test-callback'.

The test suite passes if both failed-call and failed-callback come out
as empty.


How to use the test suite
-------------------------

1. Modify the Makefile's variables
   prefix = the directory in which libffi was installed
   CC = the C compiler, often with options such as "-m32" or "-m64"
        that enforce a certain ABI,
   CFLAGS = optimization options (need to change them only for non-GCC
            compilers)
2. Run "make". If it fails already in "test-call", run also
   "make check-callback".
3. If this failed, inspect the output files.


How to interpret the results
----------------------------

The failed-call and failed-callback files consist of paired lines:
The first line is the result of the direct invocation.
The second line is the result of invocation through libffi.

For example, this output

uchar f(uchar,ushort,uint,ulong):(97,2,3,4)->255
uchar f(uchar,ushort,uint,ulong):(97,2,3,4)->0

indicates that the arguments were passed correctly, but the return
value came out wrong.

And this output

float f(17*float,3*int,L):(0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.1,1.2,1.3,1.4,1.5,1.6,1.7,1.8,6,7,8,561,1105,1729,2465,2821,6601)->15319.1
float f(17*float,3*int,L):(0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.1,1.2,1.3,1.4,1.5,1.6,1.7,1.8,-140443648,10,268042216,-72537980,-140443648,-140443648,-140443648,-140443648,-140443648)->-6.47158e+08

indicates that integer arguments that come after 17 floating-point arguments
were not passed correctly.


Credits
-------

The test suite is based on the one of GNU libffcall-2.0.
Authors: Bill Triggs, Bruno Haible