[PATCH] Add driver support for Native Client SDK

[Jan Voung]

otherwise LGTM

a little sad that a bunch of tools have to duplicate the "// Silence warning for "clang -g foo.o -o foo"" stuff... =/


================
Comment at: lib/Driver/ToolChains.cpp:2447
@@ +2446,3 @@
+      TheTriple.getEnvironment() == llvm::Triple::UnknownEnvironment)
+    TheTriple.setEnvironment(llvm::Triple::GNUEABIHF);
+  return TheTriple.getTriple();
----------------
dschuff wrote:
> jvoung wrote:
> > Is there anywhere that would warn if plain "gnueabi" was specified?
> This just covers the case if it's not specified. If gnueabi is specified that should override this default and be honored. Since the behavior of the regular arm compiler is just to allow the user to do that, maybe we should too?
If gnueabi is allowed, then perhaps the "StringRef tools::arm::getARMFloatABI()" code should check for it, rather than force hard-float.

================
Comment at: lib/Driver/Tools.cpp:675
@@ +674,3 @@
+    case llvm::Triple::NaCl:
+      FloatABI = "hard";
+      break;
----------------
See reply about allowing gnueabi in ComputeEffectiveClangTriple... how does that interact with this then? Should this check for GNUEABI and set FloatABI = "soft"?

================
Comment at: lib/Driver/Tools.cpp:7947
@@ +7946,3 @@
+void nacltools::Link::ConstructJob(Compilation &C, const JobAction &JA,
+                                  const InputInfo &Output,
+                                  const InputInfoList &Inputs,
----------------
params can be lined up

================
Comment at: lib/Driver/Tools.h:530
@@ +529,3 @@
+    void ConstructJob(Compilation &C, const JobAction &JA,
+                              const InputInfo &Output,
+                              const InputInfoList &Inputs,
----------------
Line up thy parameters.

http://reviews.llvm.org/D8590

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/