Set '-m32' option for host and target platforms

Set '-m32' option for host and target platforms

For different host and target archs, try to check whether
their compilers support '-m32' option individually. If yes,
then add this option into cflags and ldflags. The exception is
that both host and target archs are 'x64'.

Committed: https://code.google.com/p/v8/source/detail?r=11597

[Jakob Kummerow, 2012-05-15 14:05:34]

LGTM if comments are addressed.

Thanks for contributing!

http://codereview.chromium.org/10335014/diff/1/build/common.gypi
File build/common.gypi (right):

http://codereview.chromium.org/10335014/diff/1/build/common.gypi#newcode281
build/common.gypi:281: ['v8_target_arch!="x64"', {
The condition must check for OS as well, as the shell commands are not available
e.g. on Windows.
Also, I have a weak preference to explicitly list the affected architectures,
but I don't feel strongly about that.
The following should work:

['(OS=="linux" or OS=="freebsd" or OS=="openbsd" or \
   OS=="solaris" or OS=="netbsd" or OS=="mac") and \
  (v8_target_arch=="arm" or v8_target_arch=="ia32" or \
   v8_target_arch=="mips")', {

(Mac is tricky since you can use GYP to generate either Xcode project files or
Makefiles there. In order not to break the latter, I think we have to include
OS=="mac" here, but I might be wrong. Do you have a Mac around where you can
test this?)

http://codereview.chromium.org/10335014/diff/1/build/common.gypi#newcode282
build/common.gypi:282: # Check whether the host compiler and target compiler
support '-m32' option and set it if yes
nit: for comments, please keep the 80-col limit per line, and end sentences with
a full stop.

[yongsheng, 2012-05-15 14:20:49]

http://codereview.chromium.org/10335014/diff/1/build/common.gypi
File build/common.gypi (right):

http://codereview.chromium.org/10335014/diff/1/build/common.gypi#newcode281
build/common.gypi:281: ['v8_target_arch!="x64"', {
On 2012/05/15 14:05:34, Jakob wrote:
> The condition must check for OS as well, as the shell commands are not
available
> e.g. on Windows.
> Also, I have a weak preference to explicitly list the affected architectures,
> but I don't feel strongly about that.
> The following should work:
> 
> ['(OS=="linux" or OS=="freebsd" or OS=="openbsd" or \
>    OS=="solaris" or OS=="netbsd" or OS=="mac") and \
>   (v8_target_arch=="arm" or v8_target_arch=="ia32" or \
>    v8_target_arch=="mips")', {
> 
I'll change it. 
> (Mac is tricky since you can use GYP to generate either Xcode project files or
> Makefiles there. In order not to break the latter, I think we have to include
> OS=="mac" here, but I might be wrong. Do you have a Mac around where you can
> test this?)
Sorry, I don't have Mac. Buildbot will test this? right?

http://codereview.chromium.org/10335014/diff/1/build/common.gypi#newcode282
build/common.gypi:282: # Check whether the host compiler and target compiler
support '-m32' option and set it if yes
On 2012/05/15 14:05:34, Jakob wrote:
> nit: for comments, please keep the 80-col limit per line, and end sentences
with
> a full stop.
okay,thanks a lot for your review.

[yongsheng, 2012-05-15 14:36:03]

http://codereview.chromium.org/10335014/diff/1/build/common.gypi
File build/common.gypi (right):

http://codereview.chromium.org/10335014/diff/1/build/common.gypi#newcode281
build/common.gypi:281: ['v8_target_arch!="x64"', {
On 2012/05/15 14:05:34, Jakob wrote:
> The condition must check for OS as well, as the shell commands are not
available
> e.g. on Windows.
> Also, I have a weak preference to explicitly list the affected architectures,
> but I don't feel strongly about that.
> The following should work:
> 
> ['(OS=="linux" or OS=="freebsd" or OS=="openbsd" or \
>    OS=="solaris" or OS=="netbsd" or OS=="mac") and \
>   (v8_target_arch=="arm" or v8_target_arch=="ia32" or \
>    v8_target_arch=="mips")', {
> 
Sorry, I forget "Android". will add it.
> (Mac is tricky since you can use GYP to generate either Xcode project files or
> Makefiles there. In order not to break the latter, I think we have to include
> OS=="mac" here, but I might be wrong. Do you have a Mac around where you can
> test this?)

[yongsheng, 2012-05-15 14:43:42]

I upload the new patch according to your comments. Please have a check.
Anyone else who want to review this? Thanks.

[Jakob Kummerow, 2012-05-15 16:21:54]

Patch set 2 LGTM. I'll be back in the office on Friday, at which point I'll test
it on as many configurations as I can and then land it.

Sure, the buildbots test everything too, but I still prefer to test stuff before
landing it ;-)

[yongsheng, 2012-05-16 00:30:58]

On 2012/05/15 16:21:54, Jakob wrote:
> Patch set 2 LGTM. I'll be back in the office on Friday, at which point I'll
test
> it on as many configurations as I can and then land it.
> 
> Sure, the buildbots test everything too, but I still prefer to test stuff
before
> landing it ;-)
yes, that's a good habit. Great thanks for kindly review and help.

[yongsheng, 2012-05-21 05:56:34]

On 2012/05/15 16:21:54, Jakob wrote:
> Patch set 2 LGTM. I'll be back in the office on Friday, at which point I'll
test
> it on as many configurations as I can and then land it.
If any problem, would you please let me know? Thanks

[Jakob Kummerow, 2012-05-21 09:02:33]

On 2012/05/21 05:56:34, yongsheng.zhu wrote:
> If any problem, would you please let me know? Thanks

Landed as r11597, with a small fix for the "android" target added to the
Makefile.

It also doesn't work for Mac; I have a separate fix for that coming up.

[yongsheng, 2012-05-21 09:03:51]

On 2012/05/21 09:02:33, Jakob wrote:
> On 2012/05/21 05:56:34, yongsheng.zhu wrote:
> > If any problem, would you please let me know? Thanks
> 
> Landed as r11597, with a small fix for the "android" target added to the
> Makefile.
> 
> It also doesn't work for Mac; I have a separate fix for that coming up.
thanks!