This also makes TableGen able to compute sizes/offsets of synthesized
indices representing tuples.

llvm-svn: 183061

Fixes rdar:14036816, PR16130.

There is an opportunity to compute precise trip counts for 'or'
expressions and multi-exit loops.
rdar:14038809: Optimize trip count computation for multi-exit loops.

To do this we need to record the fact that ExitLimit assumes NSW. When
it does not we can safely assume that the loop trip count is the
minimum ExitLimt across all subexpressions and loop exits.

llvm-svn: 183060

seems to have caused PR16192 and other JIT related failures.

llvm-svn: 183059

llvm-svn: 183058

llvm-svn: 183057

llvm-svn: 183056

llvm-svn: 183055

llvm-svn: 183054

llvm-svn: 183053

llvm-svn: 183052

llvm-svn: 183051

llvm-svn: 183050

All running threads will be detected and stopped on attach and all threads get resumed on detach.

llvm-svn: 183049

'ivar' was used inside a record/union used 
as argument to __typeof. // rdar14037151 pr5984

llvm-svn: 183048

Use ScalarEvolution's getBackedgeTakenCount API instead of getExitCount since
that is really what we want to know. Using the more specific getExitCount was
safe because we made sure that there is only one exiting block.

No functionality change.

llvm-svn: 183047

llvm-svn: 183046

Account for the cost of scaling factor in Loop Strength Reduce when rating the
formulae. This uses a target hook.

The default implementation of the hook is: if the addressing mode is legal, the
scaling factor is free.

<rdar://problem/13806271>

llvm-svn: 183045

- should resolve remaining buildbot issues with debian/clang builder

llvm-svn: 183044

llvm-svn: 183043

llvm-svn: 183042

llvm-svn: 183041

llvm-svn: 183040

- as per review comment from Dimitry Andric!

llvm-svn: 183039

Patch by Ed Maste!

llvm-svn: 183038

Add 'JoinExistingSession' to XPC for root debugging.

llvm-svn: 183037

I renamed the function at one point in time, but forgot to fix the
layout of the arguments.

llvm-svn: 183036

We check that instructions in the loop don't have outside users (except if
they are reduction values). Unfortunately, we skipped this check for
if-convertable PHIs.

Fixes PR16184.

llvm-svn: 183035

llvm-svn: 183034

For integer types of sizes 1, 2, 4 and 8, libcompiler-rt (and libgcc)
provide atomic functions that pass parameters by value and return
results directly.

libgcc and libcompiler-rt only provide optimized libcalls for
__atomic_fetch_*, as generic libcalls on non-integer types would make
little sense. This means that we can finally make __atomic_fetch_* work
on architectures for which we don't provide these operations as builtins
(e.g. ARM).

This should fix the dreaded "cannot compile this atomic library call
yet" error that would pop up once every while.

llvm-svn: 183033

- missing #include <cstdlib> in Mangled.cpp
- missing include dirs in FreeBSD CMakeLists.txt

Patch by Ed Maste!

llvm-svn: 183032

Fixes a leak found by valgrind.

llvm-svn: 183031

Fixing an issue where formats would not propagate from parents to children in all cases
Details follow:
an SBValue has children and those are fetched along with their values
Now, one calls SBValue::SetFormat() on the parent
Technically, the format choices should propagate onto the children (see ValueObject::GetFormat())
But if the children values are already fetched, they won't notice the format change and won't update themselves
This commit fixes that by making ValueObject::GetValueAsCString() check if any format change intervened from the previous call to the current one
A test case is also added

llvm-svn: 183030

Also add the "=" to the matched pattern, to see whether we actually save
the loaded value.

llvm-svn: 183029

llvm-svn: 183028

The latest version of isl includes a new data type isl_val, which properly hides
the multi precision math library used by isl. In Polly we would like to replace
all uses of isl_int with the corresponding isl_val interfaces. This will allow
us to switch the multi precision math library in isl. This is especially
interesting for people who would like to replace libgmp with a non-gpl licensed
library (e.g. imath).

llvm-svn: 183026

The original test case showed a problem with the independet blocks pass and
we decided to XFAIL it for now. Unfortunately the failure is not detected if
we build without asserts and the verification of the independent block pass
is not run. This change tests now for the actual reason of the failure and
should trigger even in a non asserts build. We did not yet solve the underlying
bug, but this should at least make the test suite behavior consistent.

llvm-svn: 183025

llvm-svn: 183024

to be able to compile with ISL master as of today
1df91d8515ec88dc7f7f597168ad0f34f26de5a7

llvm-svn: 183023

<rdar://problem/14005311>

llvm-svn: 183022

Namely, check if the target allows to fold more that one register in the
addressing mode and if yes, adjust the cost accordingly.

Prior to this commit, reg1 + scale * reg2 accesses were artificially preferred
to reg1 + reg2 accesses. Indeed, the cost model wrongly assumed that reg1 + reg2
needs a temporary register for the computation, whereas it was correctly
estimated for reg1 + scale * reg2.

<rdar://problem/13973908>

llvm-svn: 183021