Show gpu process stats in about:tcmalloc

Show gpu process stats in about:tcmalloc


BUG=123939
TEST=launch GPU process by navigating to a page requiring it. open about:tcmalloc. refresh. see GPU process memory information


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=133816

[jamesr, 2012-04-18 22:16:38]

This generalizes the renderer-specific about:tcmalloc IPC hook for all (?)
process types. I don't think I'm iterating through child processes correctly,
because I have to iterate over the RenderProcessHost::AllHostsIterator() on the
UI thread and a BrowserChildProcessHostIterator on the IO thread - seems like
one or the other should be able to capture all children.

I'm also piggy-backing on the about:profiling hooks to get the data back up to
about_ui.cc, specifically the ProfilerMessageFilter and the
content::ProfilerSubscriber interface.  If someone knows a better way to do all
this please let me know.

OWNERS guesses:
jam@ for most of this
sky@ chrome/browser/ui/

[jam, 2012-04-18 23:06:41]

(didn't look through all now)

IPCs don't span content/chrome, which is why you got that checkdeps failure.

since this is sent by chrome, you'll need to keep the ipc in a file in chrome.
for dispatching it, we haven't had that requirement before (i.e. chrome only
hooks in renderer and utility processes, but not worker/plugin/gpu). one option
is that ChildThread can call ContentClient when it starts,  i.e.

void ContentClient::ChildThreadStarted()

and then embedders can add their own message filters on the
ChildThread::current() object

https://chromiumcodereview.appspot.com/10041017/diff/2001/chrome/browser/ui/w...
File chrome/browser/ui/webui/about_ui.cc (right):

https://chromiumcodereview.appspot.com/10041017/diff/2001/chrome/browser/ui/w...
chrome/browser/ui/webui/about_ui.cc:834: // FIXME: Why do I have to iterate over
both this and
1) chrome code uses TODOs not FIXME :)
2) there are no plans to fix this, since too much of the UI code assumes that
RenderProcessHost lives on the UI thread. even if we moved it to the IO thread,
we'd still need another object to represent it on the UI thread..

so no need for this comment

https://chromiumcodereview.appspot.com/10041017/diff/2001/chrome/browser/ui/w...
chrome/browser/ui/webui/about_ui.cc:843:
BrowserThread::PostTask(BrowserThread::IO, FROM_HERE, base::Bind(
nit: i think this is non standard tabbing per the style guide

i.e. the start of each line of parameters have to line up, so i'd do

BrowserThread::PostTask(
    BrowserThread::IO, FROM_HERE,
    base::Bind(&RequestTcmallocStatsFromChildProcesses));

[jamesr, 2012-04-24 00:32:41]

Want to take another look, John?  I've moved all of the stuff down to content

[jam, 2012-04-24 20:14:08]

lgtm

http://codereview.chromium.org/10041017/diff/22001/content/browser/tcmalloc_i...
File content/browser/tcmalloc_internals_request_job.h (right):

http://codereview.chromium.org/10041017/diff/22001/content/browser/tcmalloc_i...
content/browser/tcmalloc_internals_request_job.h:15: #include <map>
nit: this needs to be at the top

http://codereview.chromium.org/10041017/diff/22001/content/browser/tcmalloc_i...
content/browser/tcmalloc_internals_request_job.h:21: typedef
std::map<std::string, std::string> AboutTcmallocOutputsType;
nit: make this private if only this class is using it

http://codereview.chromium.org/10041017/diff/22001/content/common/child_threa...
File content/common/child_thread.cc (right):

http://codereview.chromium.org/10041017/diff/22001/content/common/child_threa...
content/common/child_thread.cc:258: {
nit: on previous line

[commit-bot: I haz the power, 2012-04-24 22:30:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jamesr@chromium.org/10041017/31001

[commit-bot: I haz the power, 2012-04-24 22:30:27]

Presubmit check for 10041017-31001 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for files in these directories:
    chrome/browser/ui/webui

Presubmit checks took 2.3s to calculate.

[sky, 2012-04-24 22:33:25]

LGTM

[commit-bot: I haz the power, 2012-04-24 22:33:56]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jamesr@chromium.org/10041017/31001

[commit-bot: I haz the power, 2012-04-25 00:05:19]

Change committed as 133816