[GN] Use correct toolchain for x64 target on Android

[GN] Use correct toolchain for x64 target on Android

This commit is to fix the linking error:
  ../../v8/src/base/platform/platform-posix.cc:418: error: undefined reference to '__android_log_vprint'

Committed: https://crrev.com/fafcc0d717526d0fc4c2312e8d6fd22ce263030b
Cr-Commit-Position: refs/heads/master@{#27559}

[haltonhuo, 2015-03-27 10:04:55]

halton.huo@intel.com changed reviewers:
+ yangguo@chromium.org

[Yang, 2015-03-27 10:17:11]

yangguo@chromium.org changed reviewers:
+ jochen@chromium.org

[Yang, 2015-03-27 10:17:11]

Jochen, can you give this a look? (CCLA checks out).

[haltonhuo, 2015-03-27 10:40:38]

On 2015/03/27 10:17:11, Yang wrote:
> Jochen, can you give this a look? (CCLA checks out).

This issue only happen on target_cpu with x64, x86 and arm are okay. And the GYP
build for all arch okay.

[jochen (gone - plz use gerrit), 2015-03-27 11:59:56]

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn
File BUILD.gn (right):

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn#newcode1400
BUILD.gn:1400: libs = [ "dl", "log" ]
log is only needed for the target toolchain, no?

I'm surprised you don't need "rt" for host?

[haltonhuo, 2015-03-30 02:15:10]

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn
File BUILD.gn (right):

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn#newcode1400
BUILD.gn:1400: libs = [ "dl", "log" ]
On 2015/03/27 11:59:56, jochen wrote:
> log is only needed for the target toolchain, no?
> 
> I'm surprised you don't need "rt" for host?
Adding rt will cause -lrt not found error. Actually, after try the GYP and GN
build, I found another difference is: GYP use clang++, which GN uses g++. I
think we should change mksnapshot to built with clang.

So I added target_cpu == "x64" to let snapshot_toolchain use clang++. But I met
macro error "error Target architecture x64 is only supported on x64 host".

Any hint?

[haltonhuo, 2015-03-30 07:35:15]

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn
File BUILD.gn (right):

https://codereview.chromium.org/1037193003/diff/1/BUILD.gn#newcode1400
BUILD.gn:1400: libs = [ "dl", "log" ]
On 2015/03/27 11:59:56, jochen wrote:
> log is only needed for the target toolchain, no?
> 
> I'm surprised you don't need "rt" for host?

Uploaded patch set2 is more correct way to go. Please review.

[jochen (gone - plz use gerrit), 2015-03-31 10:23:52]

lgtm

[jochen (gone - plz use gerrit), 2015-03-31 10:24:24]

can you please update the CL description to have a small summary in the first
line

it should also include

BUG=none
LOG=n

[haltonhuo, 2015-03-31 10:50:55]

On 2015/03/31 10:24:24, jochen wrote:
> can you please update the CL description to have a small summary in the first
> line
> 
> it should also include
> 
> BUG=none
> LOG=n

jochen, should I upload a new patch set or just update via "Edit issue"?

[Jakob Kummerow, 2015-03-31 13:40:46]

On 2015/03/31 10:50:55, haltonhuo wrote:
> > BUG=none
> > LOG=n

nit: "LOG=n" is not necessary when "BUG=none". (IMHO "BUG=none" can be dropped
as well, but if Jochen prefers to have it, that's fine with me.)

> jochen, should I upload a new patch set or just update via "Edit issue"?

Just edit it.

[jochen (gone - plz use gerrit), 2015-03-31 13:42:07]

afaik the presubmit check will whine if you just have BUG=

[Jakob Kummerow, 2015-03-31 13:51:21]

On 2015/03/31 13:42:07, jochen wrote:
> afaik the presubmit check will whine if you just have BUG=

We have plenty of counter-examples in recent history :-)
And that's WAI; the LOG= line is only required when a bug is referenced
(rationale: having an associated bug is a necessary but not sufficient condition
for being log-worthy).

[haltonhuo, 2015-03-31 15:07:11]

On 2015/03/31 13:42:07, jochen wrote:
> afaik the presubmit check will whine if you just have BUG=

Done.

Actually I have same problem as
https://code.google.com/p/v8/issues/detail?id=3552 when use "git cl upload". So
I used --bypass-hooks as workaround.

[haltonhuo, 2015-03-31 15:07:24]

The CQ bit was checked by halton.huo@intel.com

[commit-bot: I haz the power, 2015-03-31 15:07:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1037193003/20001

[commit-bot: I haz the power, 2015-03-31 15:11:24]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-03-31 15:11:25]

Try jobs failed on following builders:
  v8_presubmit on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/1658)

[Jakob Kummerow, 2015-03-31 17:05:10]

On 2015/03/31 15:11:25, I haz the power (commit-bot) wrote:
> Try jobs failed on following builders:
>   v8_presubmit on tryserver.v8 (JOB_FAILED,
> http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/1658)

Looks like "BUG=" is fine, but "BUG=none" frightens and confuses the presubmit
check. I've just dropped the entire line.

[haltonhuo, 2015-04-01 01:43:58]

The CQ bit was checked by halton.huo@intel.com

[commit-bot: I haz the power, 2015-04-01 01:44:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1037193003/20001

[commit-bot: I haz the power, 2015-04-01 01:44:29]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-04-01 01:44:40]

Patchset 2 (id:??) landed as
https://crrev.com/fafcc0d717526d0fc4c2312e8d6fd22ce263030b
Cr-Commit-Position: refs/heads/master@{#27559}