This might also fail on linux/windows, just haven't tried.

If you manually set a HTTP and HTTPS proxy in network settings (to a proxy that can handle both), you can probably repro this, and my quick attempt at that actually got the red failure page once for an http connection.

doh, forgot, this was with Mac 6.0.437.2 dev  (current dev channel build)

(No comment was entered for this change.)

Ok, even easier repro

Setup:
- Open System Preferences -> Network, select the network connection you are using (ethernet), hit Advanced... and switch to the Proxies tab.  Set HTTP and HTTPS proxy to cache.corp.google.com:3128 and hit Ok, then Apply

Repro steps:
1. launch with --use-spdy=npn command line to force spdy on
2. Open corp gmail, and auth so it signs in right as it finishes loading, go to the next step
3. Open a new tab and load build.chromium.org

every time I try, either the whole build page or the iframes in the top banner fail to load with the security issue.  I sometimes see the CSS for the page fail to load so it renders wrong.

thomasvl: thanks for the bug report.  I can't reproduce the bug using a Mac Chromium
I built today.  I wonder if it's necessary to use the VPN.

willchan: do you have any luck reproducing this bug?

I just tried and could not repro on my Mac.  I tried the steps from #4.

Thomas - can you reproduce this and then cut-and-paste the output from about:net-internals into the bug report?  It may not offer clues; do you know what the security complaint is?

We pushed out 6.0.437.3 dev friday, and with that installed, the steps I could show this with are no longer working.  Did something in the networking layers get fixed between .2 and .3? 

I'm at the office today, but I'll try on vpn again later today/tomorrow to confirm that setup also (the setup I'm using right now I remote into last week and did repro it here also).

Mike - which the the iframes from net-internals did you want (if I can get this to repro again).

I'm not aware of anything which would have impacted this during the dev rollout last week.  I know they had some trouble with 6.0.437 in general, I wonder if this crash was not related to spdy?

For the net-internals, there is a tab labeled "data" with a button to "dump to text".  That contains everything good for pasting into a bug.

(No comment was entered for this change.)

mike - there is no crash here, pages get blocked for SSL site checks.

I did find http://src.chromium.org/viewvc/chrome?view=rev&revision=50164 in the list of things that seems to be changes for this time frame of when it suddenly worked.

That has nothing to do with it.  That's a potential crash fix.  As you noted, crashing is not a symptom of this bug.

I can't produce this failure like i did before on any of the machines in question.  I tried pulling 437.2 off servers and reusing that, and i'm still failing.  Which would seem to indicate something else has changed, is there anything server side with respect to SPDY or cache.corp that could have caused some of this?

mbelshe,willchan: here are two screen shots that thomasvl sent to
me by email when he first reported this bug to me.

Here is an excerpt of his email that described these screen shots:

Yesterday my macs got the lastest dev channel push:

    Google Chrome: 6.0.437.2 (Official Build 49871) dev
    WebKit: 534.1
    V8: 2.2.17
    User Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_8; en-US) AppleWebKit/534.1 (KHTML, like Gecko) Chrome/6.0.437.2 Safari/534.1
Command Line: /Applications/Google Chrome.app/Contents/MacOS/Google Chrome -psn_0_16760827

I had a bunch of windows/tabs open (gmail, webkit pages, crbugs - a mix of http & https).  I went to check the waterfall and got the attached Picture 1.  Note the url bar doesn't show "https", so the url is http, but I'm getting the cert checking.  Taking a quick peek in chrome:net-internals, I found the interesting breadcrumb in Picture 2.

	Picture 1.png 288 KB   View   Download

	Picture 2.png 461 KB   View   Download

(No comment was entered for this change.)

This is a design issue with Alternate-Protocol and spdy.  We're discussing in a separate thread the appropriate fix.  The short of it is, we don't have a certificate for build.chromium.org, but we're trying to connect to it via SSL for spdy.  Hence the error.

fyi - i also got the failures connecting to bugs.webkit.org (via https), which should have a cert.  Also some times when bugs.webkit.org did load, the iframes for patch statuses would error, didn't check exactly how those were coming in.

I don't think that bugs.webkit.org could be related, unless they've installed a spdy server there :-)  If it is effected, we've got some really gross bug in chrome.

I filed  issue 47181  for the https://bugs.webkit.org/ certificate error
on the Mac.

(No comment was entered for this change.)

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=50997 

------------------------------------------------------------------------
r50997 | willchan@chromium.org | 2010-06-28 10:30:37 -0700 (Mon, 28 Jun 2010) | 4 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/http/http_network_transaction.cc?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/http/http_network_transaction.h?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_http_stream_unittest.cc?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_network_transaction.cc?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_session.cc?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_session.h?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_session_pool.cc?r1=50997&r2=50996
   M http://src.chromium.org/viewvc/chrome/trunk/src/net/spdy/spdy_session_pool.h?r1=50997&r2=50996

SPDY: Make sure we don't try to send https/wss over an unauthenticated, but encrypted SSL socket.
BUG=46924

Review URL: http://codereview.chromium.org/2805039
------------------------------------------------------------------------

 Issue 36148  has been merged into this issue.

 Issue 47328  has been merged into this issue.

(No comment was entered for this change.)

Couldn't repro using TVL's 2nd set of repro steps.

Verified in 6.0.466.2 (Official Build 52428).

This issue has been closed for some time. No one will pay attention to new comments.
If you are seeing this bug or have new data, please click New Issue to start a new bug.

(No comment was entered for this change.)

(No comment was entered for this change.)