Linux sandbox: add behind-flag USR2 handler for crash debugging.

Linux sandbox: add behind-flag USR2 handler for crash debugging.

When --allow-sandbox-debugging is used, we set-up a new signal
handler in the Zygote (inherited by all renderers) that performs a
chroot().

This allows testing of sandbox violation crash report with official
binaries shipping to users.

BUG=415842
R=mdempsky@chromium.org

Committed: https://chromium.googlesource.com/chromium/src/+/a993af23aca15e7e24643d9892e746a9dfbf6fa6

[jln (very slow on Chromium), 2014-09-19 22:57:26]

jln@chromium.org changed reviewers:
+ mdempsky@chromium.org

[jln (very slow on Chromium), 2014-09-19 22:58:00]

I'm not very proud, but I think this would allow for very useful manual testing.

[mdempsky, 2014-09-19 23:07:46]

lgtm

https://codereview.chromium.org/585123003/diff/1/content/zygote/zygote_main_l...
File content/zygote/zygote_main_linux.cc (right):

https://codereview.chromium.org/585123003/diff/1/content/zygote/zygote_main_l...
content/zygote/zygote_main_linux.cc:71: ignore_result(chroot("/"));
Do we want to do anything special if this returns? E.g., still crash/exit or do
you think it doesn't matter?

[jln (very slow on Chromium), 2014-09-19 23:29:51]

Patchset #3 (id:40001) has been deleted

[jln (very slow on Chromium), 2014-09-19 23:30:18]

> Do we want to do anything special if this returns? E.g., still crash/exit or
do
> you think it doesn't matter?

Good idea. Done (and tested with --disable-seccomp-filter-sandbox). Could you
please take another look?

[mdempsky, 2014-09-19 23:35:40]

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
File content/zygote/zygote_main_linux.cc (right):

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
content/zygote/zygote_main_linux.cc:80: kSecondMessage, "chroot() returned.
Errno is %d.\n", errno);
Maybe go ahead and include chroot()'s return value in the message string too?

Also, no SafeSNPrintf in bafe? :(

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
content/zygote/zygote_main_linux.cc:101: }
(Should have a "// namespace" comment, I think?)

[jln (very slow on Chromium), 2014-09-19 23:42:31]

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
File content/zygote/zygote_main_linux.cc (right):

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
content/zygote/zygote_main_linux.cc:80: kSecondMessage, "chroot() returned.
Errno is %d.\n", errno);
On 2014/09/19 23:35:40, mdempsky wrote:
> Maybe go ahead and include chroot()'s return value in the message string too?

Done

> Also, no SafeSNPrintf in bafe? :(

There is, but SafeSprintf is safe as well (it uses templates and knows the size
of the buffer).

https://codereview.chromium.org/585123003/diff/60001/content/zygote/zygote_ma...
content/zygote/zygote_main_linux.cc:101: }
On 2014/09/19 23:35:40, mdempsky wrote:
> (Should have a "// namespace" comment, I think?)

Done.

[mdempsky, 2014-09-19 23:43:17]

lgtm

[jln (very slow on Chromium), 2014-09-19 23:43:27]

The CQ bit was checked by jln@chromium.org

[commit-bot: I haz the power, 2014-09-19 23:44:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/585123003/80001

[commit-bot: I haz the power, 2014-09-19 23:45:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/585123003/80001

[jln (very slow on Chromium), 2014-09-19 23:54:37]

The CQ bit was unchecked by jln@chromium.org

[jln (very slow on Chromium), 2014-09-19 23:57:32]

The CQ bit was checked by jln@chromium.org

[commit-bot: I haz the power, 2014-09-19 23:58:46]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/585123003/100001

[commit-bot: I haz the power, 2014-09-20 00:15:20]

Patchset 5 (id:??) landed as
https://crrev.com/a993af23aca15e7e24643d9892e746a9dfbf6fa6
Cr-Commit-Position: refs/heads/master@{#295826}

[jln (very slow on Chromium), 2014-09-20 00:15:22]

Committed patchset #5 (id:100001) manually as a993af2 (presubmit successful).