Linux Sandbox: get everything to compile on Android.

Linux Sandbox: get everything to compile on Android.

We define our own android_arm_ucontext.h file since
signal.h doesn't define ucontext_t on Android.

BUG=166704
NOTRY=true


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=174070

[jln (very slow on Chromium), 2012-12-19 00:33:05]

Markus: this allows us to fully compile on Android.

There's a bit of a hack to compensate for Android's lack of u context.h

[Markus (顧孟勤), 2012-12-19 00:44:54]

lgtm

https://chromiumcodereview.appspot.com/11618010/diff/1/sandbox/linux/seccomp-...
File sandbox/linux/seccomp-bpf/sandbox_bpf.cc (right):

https://chromiumcodereview.appspot.com/11618010/diff/1/sandbox/linux/seccomp-...
sandbox/linux/seccomp-bpf/sandbox_bpf.cc:52: bool GetIsInSigHandler(ucontext_t
*ctx) {
I probably would have left the function signature unchanged, as it looks more
correct with the "const". And I would then have performed a const_cast<>() in
the function body.

But it's really your call. I doesn't much matter either way.

[jln (very slow on Chromium), 2012-12-19 01:45:32]

https://chromiumcodereview.appspot.com/11618010/diff/1/sandbox/linux/seccomp-...
File sandbox/linux/seccomp-bpf/sandbox_bpf.cc (right):

https://chromiumcodereview.appspot.com/11618010/diff/1/sandbox/linux/seccomp-...
sandbox/linux/seccomp-bpf/sandbox_bpf.cc:52: bool GetIsInSigHandler(ucontext_t
*ctx) {
On 2012/12/19 00:44:54, Markus (顧孟勤) wrote:
> I probably would have left the function signature unchanged, as it looks more
> correct with the "const". And I would then have performed a const_cast<>() in
> the function body.
> 
> But it's really your call. I doesn't much matter either way.

I was worried that sigismember() might touch the sigmask, but it is crazy. Added
a const_cast.

[commit-bot: I haz the power, 2012-12-20 01:27:33]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jln@chromium.org/11618010/2011

[commit-bot: I haz the power, 2012-12-20 01:28:54]

Change committed as 174070

[boliu_use_chromium_pls, 2012-12-20 02:07:25]

This broke downstream android webview build and the fix is not obvious to me and
was wondering if you have any pointers on what it might be:

sandbox/linux/services/android_arm_ucontext.h:13:21: error: field 'uc_mcontext'
has incomplete type

So struct sigcontext is not defined, but I don't see any includes in
android_arm_ucontext.h that defines it (or stack_t for that matter).

I'm not suggesting to revert this, but would like to fix it asap. Thanks!

[boliu_use_chromium_pls, 2012-12-20 02:45:18]

Looks like including <asm/sigcontext.h> fixed our downstream build. Waiting on
android trybots finishing: https://codereview.chromium.org/11636039/

[jln (very slow on Chromium), 2012-12-20 03:00:35]

On 2012/12/20 02:45:18, Bo Liu wrote:
> Looks like including <asm/sigcontext.h> fixed our downstream build. Waiting on
> android trybots finishing: https://codereview.chromium.org/11636039/

Sorry about that, and thanks a lot for not reverting.

I'll need to come chat with you guys to know what a "downstream" build is and
why the try bots / waterfall bots didn't fail.

[shashi, 2012-12-20 18:28:55]

This also broke the x86 build downstream.Android also has x86 build.
On 2012/12/20 03:00:35, Julien Tinnes wrote:
> On 2012/12/20 02:45:18, Bo Liu wrote:
> > Looks like including <asm/sigcontext.h> fixed our downstream build. Waiting
on
> > android trybots finishing: https://codereview.chromium.org/11636039/
> 
> Sorry about that, and thanks a lot for not reverting.
> 
> I'll need to come chat with you guys to know what a "downstream" build is and
> why the try bots / waterfall bots didn't fail.