Do not run analysis pass again if analysis info is still available.
This fixes PR1441.

llvm-svn: 48476

llvm-svn: 48474

llvm-svn: 48470

llvm-svn: 48469

llvm-svn: 48468

llvm-svn: 48467

llvm-svn: 48466

llvm-svn: 48465

llvm-svn: 48462

llvm-svn: 48461

llvm-svn: 48460

in ppc64 mode.

llvm-svn: 48459

  make test VERBOSE=1
should now produce the same output like
the Makefile normally does.
Without VERBOSE only the names of failed tests
are printed.

I am pretty comfortable with this version, and
I think it is time to switch over to Makefile.parallel
to be the default.

What do you think? Did I forget some piece of feedback?

llvm-svn: 48458

inside of the Output/ directory.
The testclean target could be removed now.

llvm-svn: 48457

This implements second part of
<http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080310/004727.html>.

llvm-svn: 48456

Get rid of "awk" dependence by using same "make"
functionality:

<http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080310/004727.html>

llvm-svn: 48455

filename.
This fixes (e.g.) CodeGen/struct-x86-darwin.c which
raced at writing to "%t1" with other tests when parallel
testing was specified with a high "-j XX" value.

Also gets rid of the file tools/clang/test/%t1
when doing an "svn st" after testing.

llvm-svn: 48454

* print out message when tests start
* testrunner sets status now
* on failed test run print out test name

I tested these changes with the non-parallel makefile
and it did not break.

llvm-svn: 48453

about sret made by the optimizers.

llvm-svn: 48452

llvm-svn: 48451

llvm-svn: 48450

lowering over to SparcCallingConv.td.  We can't make the switch
yet because we can't say to pass f64 registers in 2 x i32 registers
with the td file yet.

llvm-svn: 48449

llvm-svn: 48448

llvm-svn: 48447

the coalescer.  This doesn't really work, but gets us farther than
before.

llvm-svn: 48446

llvm-svn: 48445

other targets.  Use autogenerated calling conv to lower result of
calls.

llvm-svn: 48444

return lowering first.  This fixes a bug where the top and bottom
of i64 values were returned in the wrong registers before.

llvm-svn: 48443

best practices.

llvm-svn: 48442

each lowering, which is 'best practice'.

llvm-svn: 48441

behavior where a callee thinks a param will be
present in memory, even though the ABI doc says
it doesn't have to be.  Handle complex long long
and complex double (4 and 8 return regs).

llvm-svn: 48439

llvm-svn: 48438

llvm-svn: 48437

llvm-svn: 48436

llvm-svn: 48435

like the rest of the classes.

llvm-svn: 48434

the allocation in the class, not in sema.

llvm-svn: 48433

llvm-svn: 48432

llvm-svn: 48431

llvm-svn: 48430