Linux: Dynamically load libudev.

Linux: Dynamically load libudev.

Funnel all libudev access through a wrapper for the udev library loader.

BUG=415212
Committed: https://crrev.com/ee816c278f12488490cb2e01fda46e4abab45af7
Cr-Commit-Position: refs/heads/master@{#305348}

[Lei Zhang, 2014-11-14 22:47:11]

Patchset #4 (id:60001) has been deleted

[Lei Zhang, 2014-11-18 02:17:43]

thestig@chromium.org changed reviewers:
+ reillyg@chromium.org

[Lei Zhang, 2014-11-18 02:17:44]

reillyg: Care to take a first crack at reviewing this?

Instead of accessing libudev.h, all the udev calls now go through identical
functions in device/udev_linux/udev.h. From there we have:

device::udev_foo() which calls a singleton:
device::UdevLoader::Get()::udev_foo() which selects the right udev loader:
device::Udev[01]Loader::udev_foo() which calls the real library loader:
(dynamically generated) LibUdev[01]Loader::udev_foo()

Most of the changes outside of device/ are just to convert callers to use
device::udev_foo().

[Lei Zhang, 2014-11-18 02:24:05]

Oh, and since we are no longer linking against libudev, any direct calls to udev
functions will cause the linking step to fail. That's how I know I got them all.

[Reilly Grant (use Gerrit), 2014-11-18 22:49:57]

lgtm

[Lei Zhang, 2014-11-18 23:31:23]

thestig@chromium.org changed reviewers:
+ brettw@chromium.org, cpu@chromium.org, spang@chromium.org

[Lei Zhang, 2014-11-18 23:31:24]

Adding OWNERS:

brettw: build/config/linux/BUILD.gn
cpu: third_party/libudev - I added a copy of the libudev headers because no
Linux machine would have headers for libudev0 and libudev1 available at the same
time. They are used for the udev loader generation in build/
spang: ui/events/ozone

[spang, 2014-11-18 23:34:45]

lgtm

[Lei Zhang, 2014-11-20 00:12:23]

thestig@chromium.org changed reviewers:
+ mmoss@chromium.org

[Lei Zhang, 2014-11-20 00:12:24]

+mmoss for additional changes in chrome/installer in patch set 9.

[Michael Moss, 2014-11-20 00:17:56]

lgtm

[brettw, 2014-11-20 18:31:31]

gn build lgtm

[cpu_(ooo_6.6-7.5), 2014-11-20 22:40:40]

if this is a new third party library (which appears to be) then we have some
process to follow.

1- split the new third party lib in a CL with only that

2- make a bug which states intent to add this new third party library, what it
does and why this one.

3- follow http://www.chromium.org/developers/adding-3rd-party-libraries

[Lei Zhang, 2014-11-20 23:12:08]

On 2014/11/20 22:40:40, cpu wrote:
> if this is a new third party library (which appears to be) then we have some
> process to follow.
> 
> 1- split the new third party lib in a CL with only that

https://codereview.chromium.org/742293002

> 2- make a bug which states intent to add this new third party library, what it
> does and why this one.

http://crbug.com/435343
 
> 3- follow http://www.chromium.org/developers/adding-3rd-party-libraries

Done. A lot of it doesn't apply since we are not pulling in anything via DEPS.

[Lei Zhang, 2014-11-22 01:50:05]

The CQ bit was checked by thestig@chromium.org

[commit-bot: I haz the power, 2014-11-22 01:51:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/674703002/200001

[commit-bot: I haz the power, 2014-11-22 02:48:01]

Committed patchset #10 (id:200001)

[commit-bot: I haz the power, 2014-11-22 02:49:32]

Patchset 10 (id:??) landed as
https://crrev.com/ee816c278f12488490cb2e01fda46e4abab45af7
Cr-Commit-Position: refs/heads/master@{#305348}