Add unit test and supporting code to test process overflow case.

Add unit test and supporting code to test process overflow case.
This test is disabled for now since we don't have a fix yet.  The point 
of checkin in this CL is to make it easier for us to test alternate 
approaches to fixing.

BUG=43448
TEST=RenderProcessHostTest.ProcessOverflow

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=52059

[Erik does not do reviews, 2010-07-09 21:12:02]

Charlie - please review.  Should be straightforward since this code is unchanged
from the other original review and since no real functionality changes in this
change.

Brett and Darin FYI

[brettw, 2010-07-09 22:30:50]

LGTM

http://codereview.chromium.org/2928004/diff/2001/3003
File chrome/browser/renderer_host/site_instance.cc (right):

http://codereview.chromium.org/2928004/diff/2001/3003#newcode227
chrome/browser/renderer_host/site_instance.cc:227: bool
SiteInstance::HasSameRendererType(const GURL& url) {
It doesn't look like this is called anywhere. Is it necessary? This seems like
something that's easy enough for somebody else to do without a special function
anyway.

[Charlie Reis, 2010-07-09 22:41:24]

LGTM with the inline suggestions.

http://codereview.chromium.org/2928004/diff/2001/3003
File chrome/browser/renderer_host/site_instance.cc (right):

http://codereview.chromium.org/2928004/diff/2001/3003#newcode212
chrome/browser/renderer_host/site_instance.cc:212: if
(DOMUIFactory::HasDOMUIScheme(url))
I ran into bugs with HasDOMUIScheme in TabContents::NavigateToPendingEntry,
because it missed certain DOM UI URLs.  You might want to use UseDOMUIForURL
instead.

http://codereview.chromium.org/2928004/diff/2001/3004
File chrome/browser/renderer_host/site_instance.h (right):

http://codereview.chromium.org/2928004/diff/2001/3004#newcode135
chrome/browser/renderer_host/site_instance.h:135: // What is the renderer type
for this URL.
What is -> Returns

http://codereview.chromium.org/2928004/diff/2001/3005
File
chrome/browser/renderer_host/test/render_process_host_manager_browsertest.cc
(right):

http://codereview.chromium.org/2928004/diff/2001/3005#newcode14
chrome/browser/renderer_host/test/render_process_host_manager_browsertest.cc:14:
class RenderProcessHostManagerTest : public InProcessBrowserTest {
Is there a RenderProcessHostManager?  I think there's just RenderProcessHost and
RenderViewHostManager.  Perhaps this should be RenderProcessHostTest.

[Erik does not do reviews, 2010-07-09 22:41:33]

http://codereview.chromium.org/2928004/diff/2001/3003
File chrome/browser/renderer_host/site_instance.cc (right):

http://codereview.chromium.org/2928004/diff/2001/3003#newcode227
chrome/browser/renderer_host/site_instance.cc:227: bool
SiteInstance::HasSameRendererType(const GURL& url) {
On 2010/07/09 22:30:51, brettw wrote:
> It doesn't look like this is called anywhere. Is it necessary? This seems like
> something that's easy enough for somebody else to do without a special
function
> anyway.

removed

[Erik does not do reviews, 2010-07-09 23:32:44]

http://codereview.chromium.org/2928004/diff/2001/3003
File chrome/browser/renderer_host/site_instance.cc (right):

http://codereview.chromium.org/2928004/diff/2001/3003#newcode212
chrome/browser/renderer_host/site_instance.cc:212: if
(DOMUIFactory::HasDOMUIScheme(url))
On 2010/07/09 22:41:24, creis wrote:
> I ran into bugs with HasDOMUIScheme in TabContents::NavigateToPendingEntry,
> because it missed certain DOM UI URLs.  You might want to use UseDOMUIForURL
> instead.

Interesting.  Do you think we should remove HasDOMUIScheme?

Since this change would alter behavior from the old code, I'd like to punt it to
the follow-on work.  I added a TODO.

http://codereview.chromium.org/2928004/diff/2001/3005
File
chrome/browser/renderer_host/test/render_process_host_manager_browsertest.cc
(right):

http://codereview.chromium.org/2928004/diff/2001/3005#newcode14
chrome/browser/renderer_host/test/render_process_host_manager_browsertest.cc:14:
class RenderProcessHostManagerTest : public InProcessBrowserTest {
On 2010/07/09 22:41:24, creis wrote:
> Is there a RenderProcessHostManager?  I think there's just RenderProcessHost
and
> RenderViewHostManager.  Perhaps this should be RenderProcessHostTest.

lol.  I can't believe we all missed this until now.  We're all used to seeing
all of these prefixes and suffixes mashed together.  renamed the file and the
test.