When computing currently-live registers, the register scavenger excludes undef
uses. As a result, undef uses are ignored when computing the restore points of
registers spilled into the emergency slots. While the register scavenger
normally excludes from consideration, when scavenging, registers used by the
current instruction, we need to not exclude undef uses. Otherwise, we might end
up requiring more emergency spill slots than we have (in the case where the
undef use *is* the currently-spilled register).

Another bug found by llvm-stress.

llvm-svn: 186067

llvm-svn: 186065

llvm-svn: 186064

llvm-svn: 186063

llvm-svn: 186062

llvm-svn: 186060

SLPVectorizer: refactor the code that places extracts. Place the code that decides where to put extracts in the build-tree phase. This allows us to take the cost of the extracts into account.

llvm-svn: 186058

Without the changes introduced into this patch, if TRE saw any allocas at all,
TRE would not perform TRE *or* mark callsites with the tail marker.

Because TRE runs after mem2reg, this inadequacy is not a death sentence. But
given a callsite A without escaping alloca argument, A may not be able to have
the tail marker placed on it due to a separate callsite B having a write-back
parameter passed in via an argument with the nocapture attribute.

Assume that B is the only other callsite besides A and B only has nocapture
escaping alloca arguments (*NOTE* B may have other arguments that are not passed
allocas). In this case not marking A with the tail marker is unnecessarily
conservative since:

  1. By assumption A has no escaping alloca arguments itself so it can not
     access the caller's stack via its arguments.

  2. Since all of B's escaping alloca arguments are passed as parameters with
     the nocapture attribute, we know that B does not stash said escaping
     allocas in a manner that outlives B itself and thus could be accessed
     indirectly by A.

With the changes introduced by this patch:

  1. If we see any escaping allocas passed as a capturing argument, we do
     nothing and bail early.

  2. If we do not see any escaping allocas passed as captured arguments but we
     do see escaping allocas passed as nocapture arguments:

       i. We do not perform TRE to avoid PR962 since the code generator produces
          significantly worse code for the dynamic allocas that would be created
          by the TRE algorithm.

       ii. If we do not return twice, mark call sites without escaping allocas
           with the tail marker. *NOTE* This excludes functions with escaping
           nocapture allocas.

  3. If we do not see any escaping allocas at all (whether captured or not):

       i. If we do not have usage of setjmp, mark all callsites with the tail
          marker.

       ii. If there are no dynamic/variable sized allocas in the function,
           attempt to perform TRE on all callsites in the function.

Based off of a patch by Nick Lewycky.

rdar://14324281.

llvm-svn: 186057

I had thought that these tests could be target-neutral, but in practice this is
not the case (on some targets, like Hexagon and Darwin), they trigger an assert
(a different assert than the one that r186044 fixes).

llvm-svn: 186051

Trying to fix the i686-mingw32 build.

llvm-svn: 186046

For some reason, the Hexagon backend does not reject these invalid static
initializer expressions, but instead crashes in AsmPrinter::EmitGlobalConstant.

llvm-svn: 186045

A non-constant-foldable static initializer expression containing insertvalue or
extractvalue had been causing an assert:

  Constants.cpp:1971: Assertion `FC && "ExtractValue constant expr couldn't be
                                 folded!"' failed.

Now we report a more-sensible "Unsupported expression in static initializer"
error instead.

Fixes PR15417.

llvm-svn: 186044

It is not reliable to depend on the output of llvm_unreachable. The original
change will have proper tests when llvm-ar moves to lib/Object (soon).

llvm-svn: 186043

llvm-svn: 186042

llvm-svn: 186041

There is no lib/Archive anymore and some archive tests were in test/Archive and
others in test/Object. Since archive is just one of the formats supported by
lib/Object, test/Object is probably the best location.

llvm-svn: 186038

llvm-svn: 186035

Patch from Игорь Пашев  (I do hope we support utf-8 commit messages; I
also hope he'll forgive me for transliterating it as Igor Pashev in
case things go horribly wrong).

llvm-svn: 186034

llvm-svn: 186032

llvm-svn: 186031

llvm-svn: 186029

llvm-svn: 186027

[objc-arc] Changed 'mode: c++' => 'C++' at Nick Lewycky's suggestion. Also removed unnecessary mode: c++ lines from .cpp files.

llvm-svn: 186026

llvm-svn: 186025

Found by -Wdocumentation.

llvm-svn: 186021

llvm-svn: 186018

llvm-svn: 186017

The status function is already using a syscall that returns the file size.
Remember it and implement file_size as a simple wrapper.

No functionally change, but clients that already use status now can avoid
calling file_size.

llvm-svn: 186016

llvm-svn: 186015

llvm-svn: 186014

llvm-svn: 186013

Enough for the radeonsi driver to use it for calculating derivatives.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 186012

lit test coverage to follow in the next commit.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 186011

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 186010

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 186009

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 186008

llvm-svn: 186006

In discussing this change with Bill Schmidt, it was decided that the original
comment about negative FIs was incorrect. We'll still exclude them for now, but
now with a more-accurate explanation.

llvm-svn: 186005

llvm-svn: 186001

llvm-svn: 186000