Use gold when compiling 32 bit binaries on an x64 machine.

Use gold when compiling 32 bit binaries on an x64 machine.

gold can't normally be used on 32 bit machines because it uses too
much memory, but if the 32-bit binary that is being created is being
created on an x64 machine with sysroot, then it's possible.

BUG=511201
R=dpranke@chromium.org,thakis@chromium.org

Committed: https://crrev.com/f3f4e1b70c954cb30b7c689e4f130ab715fbb49a
Cr-Commit-Position: refs/heads/master@{#339514}

[Daniel Bratell, 2015-07-17 13:53:53]

bratell@opera.com changed reviewers:
+ thakis@chromium.org

[Daniel Bratell, 2015-07-17 13:53:54]

thakis and dpranke, could you take a look at this one liner? It seems this is
already done in gn (build/config/compiler/BUILD.gn only checks current_cpu and
ignores target_arch).

[Dirk Pranke, 2015-07-17 15:59:16]

lgtm. I'm not sure we even support host_arch==ia32 for anything at this point in
chromium (I believe NaCl does support it, though).

[Nico, 2015-07-17 16:02:10]

Please wait with landing until I can take a look (won't be at a laptop for
a few more hours -- I thought Lei did this already and I want to look at
some old bugs about this)
On Jul 17, 2015 8:59 AM, <dpranke@chromium.org> wrote:

> lgtm. I'm not sure we even support host_arch==ia32 for anything at this
> point in
> chromium (I believe NaCl does support it, though).
>
> https://codereview.chromium.org/1234943009/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Daniel Bratell, 2015-07-20 12:37:14]

On 2015/07/17 16:02:10, Nico wrote:
> Please wait with landing until I can take a look (won't be at a laptop for
> a few more hours -- I thought Lei did this already and I want to look at
> some old bugs about this)

Did you find your laptop?

[Nico, 2015-07-20 15:58:39]

Thanks for the ping. https://codereview.chromium.org/897163002 attempted to do
the same thing.
https://code.google.com/p/chromium/codesearch#chromium/src/build/detect_host_...
tries to set host_arch to ia32 for 64-bit kernels with 32-bit userland. Does
that not work for you?

[Daniel Bratell, 2015-07-20 17:07:37]

On 2015/07/20 15:58:39, Nico wrote:
> Thanks for the ping. https://codereview.chromium.org/897163002 attempted to do
> the same thing.
>
https://code.google.com/p/chromium/codesearch#chromium/src/build/detect_host_...
> tries to set host_arch to ia32 for 64-bit kernels with 32-bit userland. Does
> that not work for you?

We're using sysroots with a normal x64 machine base, basically doing a cross
compilation so gyp will not detect anything ia32 when it looks around. I guess
that is why that code doesn't trigger.

[Nico, 2015-07-20 17:09:23]

On 2015/07/20 17:07:37, Daniel Bratell wrote:
> On 2015/07/20 15:58:39, Nico wrote:
> > Thanks for the ping. https://codereview.chromium.org/897163002 attempted to
do
> > the same thing.
> >
>
https://code.google.com/p/chromium/codesearch#chromium/src/build/detect_host_...
> > tries to set host_arch to ia32 for 64-bit kernels with 32-bit userland. Does
> > that not work for you?
> 
> We're using sysroots with a normal x64 machine base, basically doing a cross
> compilation so gyp will not detect anything ia32 when it looks around. I guess
> that is why that code doesn't trigger.

No chroot?

[Daniel Bratell, 2015-07-20 18:19:05]

On 2015/07/20 17:09:23, Nico wrote:
> On 2015/07/20 17:07:37, Daniel Bratell wrote:
 work for you?
> > 
> > We're using sysroots with a normal x64 machine base, basically doing a cross
> > compilation so gyp will not detect anything ia32 when it looks around. I
guess
> > that is why that code doesn't trigger.
> 
> No chroot?

I do not think so. I cannot find any chroots in the build scripts or build logs.
The one I would ask to be certain is on vacation now, but it all seems to be
done with flags to the compilers/linkers.

[Nico, 2015-07-20 18:25:31]

lgtm, but life will probably be easier for you if you use the same setup as our
32-bit bots (i.e. a chroot)

[Mostyn Bramley-Moore, 2015-07-20 18:28:37]

> > No chroot?
> 
> I do not think so. I cannot find any chroots in the build scripts or build
logs.
> The one I would ask to be certain is on vacation now, but it all seems to be
> done with flags to the compilers/linkers.

I can confirm that we're not using a chroot here (even though I don't work
directly on this particular product).

[Daniel Bratell, 2015-07-20 19:31:51]

The CQ bit was checked by bratell@opera.com

[commit-bot: I haz the power, 2015-07-20 19:33:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1234943009/1

[commit-bot: I haz the power, 2015-07-20 21:00:47]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-07-20 21:03:40]

Patchset 1 (id:??) landed as
https://crrev.com/f3f4e1b70c954cb30b7c689e4f130ab715fbb49a
Cr-Commit-Position: refs/heads/master@{#339514}

[odinho, 2015-07-20 21:14:43]

On 2015/07/20 at 18:25:31, thakis wrote:
> lgtm, but life will probably be easier for you if you use the same setup as
our 32-bit bots (i.e. a chroot)

I did try going the chroot route for a while, but I kept hitting road blocks. 
It seemed like that was *not* how you were doing things, based on how not-smooth
everything was.  When I found the sysroot scripts in chrome/installer/linux
everything went much smoother.  Quite happy about our sysroots now, we haven't
had any problems with this, we're even moving 64 bit builds to be build using
sysroots.

I'm surprised you don't use this, considering how well it's been working and how
easy it was to get going compared to chroots.