Linux: Catch X errors and force a crash....

Linux: Catch X errors and force a crash.

BUG=45020
TEST=See http://crosbug.com/3487

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=49488

[agl, 2010-06-10 21:52:08]

LGTM

http://codereview.chromium.org/2716007/diff/1/5
File chrome/plugin/plugin_thread.cc (right):

http://codereview.chromium.org/2716007/diff/1/5#newcode15
chrome/plugin/plugin_thread.cc:15: #include <string>
nit: why these two? If it's needed by x11_util.h, should not x11_util.h include
them itself?

[piman, 2010-06-10 21:55:19]

I think you want to do that in the gpu process as well (chrome/gpu/gpu_thread.cc
after gtk_init).

Also, isn't it true that glib catches it and pipes it to our glib log handler ?

[Lei Zhang, 2010-06-10 23:58:51]

On 2010/06/10 21:55:19, piman wrote:
> I think you want to do that in the gpu process as well
(chrome/gpu/gpu_thread.cc
> after gtk_init).

Done in patch set 2.

> Also, isn't it true that glib catches it and pipes it to our glib log handler
?

No, glib doesn't seem to catch X errors.

http://codereview.chromium.org/2716007/diff/1/5
File chrome/plugin/plugin_thread.cc (right):

http://codereview.chromium.org/2716007/diff/1/5#newcode15
chrome/plugin/plugin_thread.cc:15: #include <string>
On 2010/06/10 21:52:10, agl wrote:
> nit: why these two? If it's needed by x11_util.h, should not x11_util.h
include
> them itself?

Eh? plugin_thread.cc uses std::string/vector independent of x11_util.h. Plus,
the linter told me to do it.

[piman, 2010-06-11 00:07:46]

LGTM