Add helper comment for icecc users.

build/common.gypi

 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 # IMPORTANT:
 # Please don't directly include this file if you are building via gyp_chromium,
 # since gyp_chromium is automatically forcing its inclusion.
 {
   # Variables expected to be overriden on the GYP command line (-D) or by
   # ~/.gyp/include.gypi.
@@ skipping 149 matching lines @@
 
         # Set ARM architecture version.
         'arm_version%': 7,
 
         # Use aurax11 for clipboard implementation. This is true on linux_aura.
         'use_clipboard_aurax11%': 0,
 
         # goma settings.
         # 1 to use goma.
         # If no gomadir is set, it uses the default gomadir.
         'use_goma%': 0,

[dshwang, 2014/04/23 15:05:20]

we already have use_goma, so imo it's reasonable to introduce use_icecc because
most non-googler use icecc.

         'gomadir%': '',
 
+        # When using icecc, we cannot use some build flags. e.g. debug fission, -B
+        'use_icecc%': 0,
+
         # The system root for cross-compiles. Default: none.
         'sysroot%': '',
         'chroot_cmd%': '',
 
         'conditions': [
           # Ash needs Aura.
           ['use_aura==0', {
             'use_ash%': 0,
           }],
 
@@ skipping 71 matching lines @@
       'use_ozone_evdev%': '<(use_ozone_evdev)',
       'use_clipboard_aurax11%': '<(use_clipboard_aurax11)',
       'embedded%': '<(embedded)',
       'use_openssl%': '<(use_openssl)',
       'use_openssl_certs%': '<(use_openssl_certs)',
       'use_system_fontconfig%': '<(use_system_fontconfig)',
       'enable_viewport%': '<(enable_viewport)',
       'enable_hidpi%': '<(enable_hidpi)',
       'android_webview_build%': '<(android_webview_build)',
       'use_goma%': '<(use_goma)',
+      'use_icecc%': '<(use_icecc)',

[Mostyn Bramley-Moore, 2014/04/24 07:36:52]

nitpick: move this down a line, to keep the goma settings together.

[dshwang, 2014/04/24 07:51:37]

yes, I'll fix it.

       'gomadir%': '<(gomadir)',
       'enable_app_list%': '<(enable_app_list)',
       'use_default_render_theme%': '<(use_default_render_theme)',
       'buildtype%': '<(buildtype)',
       'branding%': '<(branding)',
       'arm_version%': '<(arm_version)',
       'sysroot%': '<(sysroot)',
       'chroot_cmd%': '<(chroot_cmd)',
 
       # Set to 1 to enable fast builds. Set to 2 for even faster builds
@@ skipping 446 matching lines @@
           'enable_plugins%': 0,
         }, {
           'enable_plugins%': 1,
         }],
 
         # linux_use_bundled_gold: whether to use the gold linker binary checked
         # into third_party/binutils.  Force this off via GYP_DEFINES when you
         # are using a custom toolchain and need to control -B in ldflags.
         # Gold is not used for 32-bit linux builds as it runs out of address
         # space.
-        ['OS=="linux" and (target_arch=="x64" or target_arch=="arm")', {
+        ['OS=="linux" and (target_arch=="x64" or target_arch=="arm") and use_icecc==0', {

[dshwang, 2014/04/24 04:18:57]

use_icecc already set other build flags here.

If we want to change it to following pattern,
['use_icecc==1', {
  'linux_use_bundled_gold%': 0,
  'linux_use_bundled_binutils%': 0,
}],
these definition should stay more upper level, and linux_use_bundled_gold
variable will be in trouble, because "'linux_use_bundled_gold%': 1" means that
define linux_use_bundled_gold as 1 if linux_use_bundled_gold is not defined.
In this case, linux_use_bundled_gold may or may not be defined according to
use_icecc.

So imho current code is more succinct.

[Mostyn Bramley-Moore, 2014/04/24 07:36:52]

Is this really needed?  linux_use_bundled_gold is only used to modify ldflags,
which is only used in link commands, which icecc doesn't distribute anyway.

[dshwang, 2014/04/24 07:51:37]

That's good point. My machine works well without linux_use_bundled_gold.

However, linux_use_bundled_binutils=0 and linux_use_bundled_gold=1 looks
inconsistent to my eyes, because linux_use_bundled_binutils=0 makes gcc use
system "as" or "objcopy" but linux_use_bundled_gold=1 makes ninja use bundled
"ld".
For example, if system binutils version is 2.22, doesn't it break anything?
because we use 2.22 objcopy with 2.24 ld.

[Lei Zhang, 2014/04/24 19:06:42]

I don't think it'll break anything. Before third_party/binutils, we only had
third_party/gold, so we were using the system toolchain for everything but the
linker, which got version bumps independent of everything else.

           'linux_use_bundled_gold%': 1,
         }, {
           'linux_use_bundled_gold%': 0,
         }],
 
         # linux_use_bundled_binutils: whether to use the binary binutils
         # checked into third_party/binutils.  These are not multi-arch so cannot
         # be used except on x86 and x86-64 (the only two architectures which
         # are currently checke in).  Force this off via GYP_DEFINES when you
         # are using a custom toolchain and need to control -B in cflags.
-        ['OS=="linux" and (target_arch=="x64")', {
+        ['OS=="linux" and (target_arch=="x64") and use_icecc==0', {
           'linux_use_bundled_binutils%': 1,
         }, {
           'linux_use_bundled_binutils%': 0,
         }],
 
         # linux_use_gold_flags: whether to use build flags that rely on gold.
         # On by default for x64 Linux.  Temporarily off for ChromeOS as
         # it failed on a buildbot.
         ['OS=="linux" and target_arch=="x64" and chromeos==0', {
           'linux_use_gold_flags%': 1,
@@ skipping 284 matching lines @@
     'spdy_proxy_auth_origin%': '<(spdy_proxy_auth_origin)',
     'spdy_proxy_auth_property%': '<(spdy_proxy_auth_property)',
     'spdy_proxy_auth_value%': '<(spdy_proxy_auth_value)',
     'data_reduction_proxy_probe_url%': '<(data_reduction_proxy_probe_url)',
     'enable_mdns%' : '<(enable_mdns)',
     'enable_service_discovery%' : '<(enable_service_discovery)',
     'enable_hangout_services_extension%' : '<(enable_hangout_services_extension)',
     'v8_optimized_debug%': '<(v8_optimized_debug)',
     'proprietary_codecs%': '<(proprietary_codecs)',
     'use_goma%': '<(use_goma)',
+    'use_icecc%': '<(use_icecc)',

[Mostyn Bramley-Moore, 2014/04/24 07:36:52]

nitpick: move this down a line, to keep the goma settings together.

     'gomadir%': '<(gomadir)',
 
     # Use system protobuf instead of bundled one.
     'use_system_protobuf%': 0,
 
     # Use system yasm instead of bundled one.
     'use_system_yasm%': 0,
 
     # Use system ICU instead of bundled one.
     'use_system_icu%' : 0,
@@ skipping 2141 matching lines @@
                   '-Wl,--no-as-needed',
                 ],
               }],
               ['debug_unwind_tables==1', {
                 'cflags': ['-funwind-tables'],
               }, {
                 'cflags': ['-fno-unwind-tables', '-fno-asynchronous-unwind-tables'],
               }],
               # http://gcc.gnu.org/wiki/DebugFission
               # Requires gold and gcc >= 4.8 or clang.
-              ['linux_use_gold_flags==1 and (clang==1 or gcc_version>=48) and binutils_version>=223', {
+              ['linux_use_gold_flags==1 and (clang==1 or gcc_version>=48) and binutils_version>=223 and use_icecc==0', {

[dshwang, 2014/04/24 04:18:57]

There is not foo variable for "if (use_icc) {  foo = 1; }"
icecc can use system gold.

[Mostyn Bramley-Moore, 2014/04/24 07:36:52]

IMO (but I think Nico disagreed previously), this is where we should use a new
linux_use_split_dwarf variable (the "foo" in Lei Zhang's suggestion?).

[dshwang, 2014/04/24 07:51:37]

if we introduce linux_use_split_dwarf, that code looks like
['linux_use_gold_flags==1 and (clang==1 or gcc_version>=48) and
binutils_version>=223 and use_icecc==0', {
'linux_use_split_dwarf': 1
}],
...
['linux_use_split_dwarf==1', {
'cflags': ['-gsplit-dwarf'],
'ldflags': ['-Wl,--gdb-index'],
}],

IMO, looks not very advanced.

When Nico disagreed, icecc needs only binutils_version=0 hack, but now icecc
needs this whole CL. So Nico@, could I ask your opinion?

                 'cflags': ['-gsplit-dwarf'],
                 'ldflags': ['-Wl,--gdb-index'],
               }],
             ],
           },
           'Release_Base': {
             'variables': {
               'release_optimize%': '2',
               # Binaries become big and gold is unable to perform GC
               # and remove unused sections for some of test targets
@@ skipping 2002 matching lines @@
     # settings in target dicts.  SYMROOT is a special case, because many other
     # Xcode variables depend on it, including variables such as
     # PROJECT_DERIVED_FILE_DIR.  When a source group corresponding to something
     # like PROJECT_DERIVED_FILE_DIR is added to a project, in order for the
     # files to appear (when present) in the UI as actual files and not red
     # red "missing file" proxies, the correct path to PROJECT_DERIVED_FILE_DIR,
     # and therefore SYMROOT, needs to be set at the project level.
     'SYMROOT': '<(DEPTH)/xcodebuild',
   },
 }