llvm-svn: 133069

llvm-svn: 133068

Note that this actually changes code generation, and someone who
understands this target better should check the changes.

- R12Q is now allocatable. I think it was omitted from the allocation
  order by mistake since it isn't reserved. It as apparently used as a
  GOT pointer sometimes, and it should probably be reserved if that is
  the case.

- The GR64 registers are allocated in a different order now. The
  register allocator will automatically put the CSRs last. There were
  other changes to the order that may have been significant.

The test fix is because r0 and r1 swapped places in the allocation order.

llvm-svn: 133067

llvm-svn: 133066

- llvm.dbg.declare already receives line number information from ParmDecl
- Additional extra stoppoint messes up gdb's understanding of where function body starts.

llvm-svn: 133065

llvm-svn: 133064

Added a test case for handling physreg aliases during pre-RA-sched.

llvm-svn: 133063

llvm-svn: 133062

Add an introduction to the scripting example web page.

llvm-svn: 133061

were just punting on template argument deduction for a number of type
nodes. Most of them, obviously, didn't matter.

As a consequence of this, make extended vector types (via the
ext_vector_type attribute) actually work properly for several
important cases:
  - If the attribute appears in a type-id (i.e, not attached to a
  typedef), actually build a proper vector type
  - Build ExtVectorType whenever the size is constant; previously, we
  were building DependentSizedExtVectorType when the size was constant
  but the type was dependent, which makes no sense at all.
  - Teach template argument deduction to handle
  ExtVectorType/DependentSizedExtVectorType.

llvm-svn: 133060

Patch by Nathan Whitehorn!

llvm-svn: 133059

llvm-svn: 133057

llvm-svn: 133056

before the template parameters have acquired a proper context (e.g.,
because the enclosing context has yet to be built), provide empty
parameter lists for all outer template parameter scopes to inhibit any
substitution for those template parameters. Fixes PR9643 /
<rdar://problem/9251019>.

llvm-svn: 133055

specifying that the target CPU is corei7.

llvm-svn: 133053

Enable the simplification of truncating-store after fixing the usage of
GetDemandBits (which must operate on the vector element type).

Fix the a usage of getZeroExtendInReg which must also be done on scalar types.

llvm-svn: 133052

Replace the statically generated hashtables for checking register relationships with just scanning the (typically tiny) static lists.

At the time I wrote this code (circa 2007), TargetRegisterInfo was using a std::set to perform these queries.  Switching to the static hashtables was an obvious improvement, but in reality there's no reason to do anything other than scan.
With this change, total LLC time on a whole-program 403.gcc is reduced by approximately 1.5%, almost all of which comes from a 15% reduction in LiveVariables time.  It also reduces the binary size of LLC by 86KB, thanks to eliminating a bunch of very large static tables.

llvm-svn: 133051

Unfortunately we can't follow what the rest of the language does (wrapping it
in double-quotes) because that would cause an ambiguity with metadata strings,
so instead we escape any unusual characters with \xx escaping.

llvm-svn: 133050

llvm-svn: 133049

llvm-svn: 133048

llvm-svn: 133047

- (bounded copies) Be more conservative about how much is being copied.
- (str(n)cat) If we can't compute the exact final length of an append operation, we can still lower-bound it.
- (stpcpy) Fix the conjured return value at the end to actually be returned.

This requires these supporting changes:
- C string metadata symbols are still live even when buried in a SymExpr.
- "Hypothetical" C string lengths, to represent a value that /will/ be passed to setCStringLength() if all goes well. (The idea is to allow for temporary constrainable symbols that may end up becoming permanent.)
- The 'checkAdditionOverflow' helper makes sure that the two strings being appended in a strcat don't overflow size_t. This should never *actually* happen; the real effect is to keep the final string length from "wrapping around" in the constraint manager.

This doesn't actually test the "bounded" operations (strncpy and strncat) because they can leave strings unterminated. Next on the list!

llvm-svn: 133046

either imlicitly (for builtins) or explicitly (due to multiple
specification of the same attributes). Fixes <rdar://problem/9612060>.

llvm-svn: 133045

[analyzer] If a C string length is UnknownVal, clear any existing length binding. No tests yet because the only thing that sets string length is strcpy(), and that needs some work anyway.

llvm-svn: 133044

This prepares tablegen to compute register lists from set theoretic dag
expressions. This doesn't really make any difference as long as
Target.td still declares RegisterClass::MemberList as [Register].

llvm-svn: 133043

llvm-svn: 133042

[analyzer] When performing a binary operation on symbolic operands, we convert the symbol values to a common type. But in a relational operation, the result is an 'int' or 'bool', which may not be the appropriate type to convert the operands to. In these cases, use the left-hand operand's type as the conversion type.

There's no associated test for this because fully-constrained symbolic values are evaluated ahead of time in normal expressions. This can only come up in checker-constructed expressions (like the ones in an upcoming patch to CStringChecker).

llvm-svn: 133041

Make the Elements vector private and expose an ArrayRef through
getOrder() instead. getOrder will eventually provide multiple
user-specified allocation orders.

Use the sorted member set for member and subclass tests. Clean up a lot
of ad hoc searches.

llvm-svn: 133040

llvm-svn: 133039

are also greater than the alias.

llvm-svn: 133038

protected in the case where a variable is being initialized by a
trivial default constructor but has a non-trivial destructor.

llvm-svn: 133037

llvm-svn: 133036

Stop using memdep for a check that didn't really make sense with memdep. In terms of specific issues, using memdep here checks irrelevant instructions and won't work properly once we start returning "unknown" more aggressively from memdep.

llvm-svn: 133035

the bits being cleared by the AND are not demanded by the BFI.

The previous BFI dag combine rule was actually incorrect (or used to be
correct until BFI representation changed).

rdar://9609030

llvm-svn: 133034

Change the output for -Wshift-overflow and
-Wshift-sign-overflow to an unsigned hexadecimal. It makes
more sense for looking at bits than a signed decimal does.
Also, change the diagnostic's wording from "overrides"
to "sets".

This uses a new optional argument in APInt::toString()
that adds the '0x' prefix to hexademical numbers.

This fixes PR 9651.

Patch by nobled@dreamwidth.org!

llvm-svn: 133033

toString() now takes an optional bool argument that,
depending on the radix, adds the appropriate prefix
to the integer's string representation that makes it into a
meaningful C literal, e.g.:

hexademical: '-f' becomes '-0xf'
octal: '77' becomes '077'
binary: '110' becomes '0b110'

Patch by nobled@dreamwidth.org!

llvm-svn: 133032

Add "unknown" results for memdep, which mean "I don't know whether a dependence for the given instruction exists in the given block".  This cleans up all the existing hacks in memdep which represent this concept by returning clobber with various unrelated instructions.

llvm-svn: 133031

llvm-svn: 133030

No functional change intended.

llvm-svn: 133029

Add an optimization that looks for a specific pair-wise add pattern and generates a vpaddl instruction instead of scalarizing the add.
Includes a test case.

llvm-svn: 133027