Various fixes for 32-bit Linux GN builds that need to be tracked down.

build/config/compiler/BUILD.gn

Index: build/config/compiler/BUILD.gn
diff --git a/build/config/compiler/BUILD.gn b/build/config/compiler/BUILD.gn
index 7cc2c39da9422ac403c70b1ae5fa0ffb5aabef6e..0cf4ce6d911feb1d5a723b4e26c2d2e7dd57cc1a 100644
--- a/build/config/compiler/BUILD.gn
+++ b/build/config/compiler/BUILD.gn
@@ -44,9 +44,10 @@ declare_args() {
   # Requires profiling to be set to true.
   enable_full_stack_frames_for_profiling = false
 
-  # Use 64-bit gold for linking on both 64-bit Linux and 32-bit linux;
-  # 32-bit Gold runs out of address-space on 32-bit bit builds.
-  use_gold = is_linux && (current_cpu == "x64" || current_cpu == "x86")
+  # TODO(GYP): We should be using 64-bit gold for linking on both 64-bit Linux
+  # and 32-bit linux; 32-bit Gold runs out of address-space on 32-bit builds.
+  # However, something isn't quite working right on the 32-bit builds.
+  use_gold = is_linux && current_cpu == "x64"

[Roland McGrath, 2015/09/21 23:59:07]

I investigated the gold crash a bit.  It still happens with upstream trunk of
gold.  I posted on binutils@sourceware.org about the details.
You can work around it by dropping the --icf=all flag.

 
   # When we are going to use gold we need to find it.
   if (use_gold) {