get_concurrent_links.py: give more RAM per job in LTO builds.

build/toolchain/gcc_toolchain.gni

Index: build/toolchain/gcc_toolchain.gni
diff --git a/build/toolchain/gcc_toolchain.gni b/build/toolchain/gcc_toolchain.gni
index 6742f2b9d1b215bec91dea3dc13e2c4bb1f7adbe..e83b94b46b1bde7d1ea5ae0d4d92bc3b2113c842 100644
--- a/build/toolchain/gcc_toolchain.gni
+++ b/build/toolchain/gcc_toolchain.gni
@@ -9,7 +9,12 @@ import("//build/toolchain/goma.gni")
 import("//build/toolchain/toolchain.gni")
 
 # This value will be inherited in the toolchain below.
-concurrent_links = exec_script("get_concurrent_links.py", [], "value")
+if (is_cfi) {
+  concurrent_links =

[krasin, 2015/12/03 23:18:45]

I am not totally sure, if it's an idiomatic GN code.
I would probably prefer something like

concurrent_links = exec_script("get_concurrent_links.py", ["--lto=" + is_cfi],
"value")

if it's possible. Suggestions are welcome.

[krasin, 2015/12/03 23:22:20]

Acknowledged.

+      exec_script("get_concurrent_links.py", [ "--lto" ], "value")
+} else {
+  concurrent_links = exec_script("get_concurrent_links.py", [], "value")
+}
 
 # This template defines a toolchain for something that works like gcc
 # (including clang).