Hi,
> 2. how do the test cases get the oracles (i.e. how to decide the
> testing output is Pass, Fail, ...)?
IMHO, it depends. Like in the dg tests ( gcc.dg) there are ways of
specifying in the testcase itself as to what the test aims to check.
For e.g with a /* { dg-final { scan-assembler "add a, b" } } */ we say
that if the instruction add a, b is present in the assembly the test
has passed. For torture tests (gcc.c-torture) usually a testcase that
exits with a non-zero values is said to have failed. See
http://gcc.gnu.org/wiki/HowToPrepareATestcase for more details.
HTH,
Pranav
