Thanks for filing this issue. Adding some labels to get some more eyes on this.

dimich, could this be an offline page? Do those have UX showing they are offline when this is a restored page?

I'm not  sure what the "offline page" you're talking about is, and I know the question is not directed to me, but just to clarify:

- I am definitely online when this happens
- at least in some of the cases, at step 2, I observed the old page being (re)loaded (the tab would go blank, the progress indicator would show up and take its time to fill up while the page would reload) before I would type the new url/search

AFAIK there was an optimization to used an "offline" or saved view of a page when chrome is being restored, to prevent having to reload the page. I'm not sure in which circumstances this triggers or if it can happen when you're online.

#2 indicates that this probably isn't an offline page which wouldn't need to reload.

When you try to navigate, literally nothing happens on the UI?

> When you try to navigate, literally nothing happens on the UI?

Actually I haven't tried clicking on links and stuff like that (I'll try next time), the only thing I have tried is typing a search or an URL (or even pasting an url copied from some other app). In those cases yes, literally nothing happens below the address bar. As I indicated in the report, in the case of typing a search, the google search url does show up in the address bar, but besides that, nothing (visible) happens.

Should have mentioned: while typing search keywords, suggestions do show up, but after submitting the input, nothing happens

Thanks for the added info. This definitely seems like a navigation issue. Trying links would definitely help pinpoint the problem.

Offline pages are Android-only and not enabled in M53 stable, so we likely can exclude that. I wonder what OS this report is from - it mentions Flash version, but also "tapping Chrome icon on home screen", so it's unclear.

If Chrome stable had this issue in general, we'd see tons more reports, so perhaps a specific device config issue? 

53.0.2785.124 is the version of the previous stable release of Android Chrome according to omahaproxy.appspot.com.

> I wonder what OS this report is from

Android. 

I selected it from the dropdown when filing the report, I wonder why it's not automatically included at the bottom of the report (though I do see it's in the bug info on the left). I didn't bother to delete the content of the "flash version" textfield which had been filled in automatically. This bug tracker is quite pathetic.

It might help to get a network dump of this in action. See:
https://sites.google.com/a/chromium.org/dev/for-testers/providing-network-details

I'm wondering if we're even initiated the navigation request.

> It might help to get a network dump of this in action

It might help a lot, but I've had a look at the instructions to get one, and if you look at the steps to reproduce the issue (note 1-2), it seems pretty hard to do both.

> I'm wondering if we're even initiated the navigation request.

I'm definitely wondering that too.

Could you verify the problem is occurring, then start logging in a separate tab, then reproduce the problem in the original tab?

Do you have to close the problematic tab for other tabs to work?

@12 great idea!
Done that, I have emailed the log to you personally, rather than uploading it here, as I'm afraid it might contain private information.

Err, actually, what I did was follow the steps in  issue 654540  (which I reproduced) and then step 3 in this issue, which I also observed. The two seem to be related, if not the same issue completely.

Note that I had a bunch of other tabs open which I didn't touch during the test.

Thanks for the log. Can you describe which URLs or searches you were trying to do, to help make sense of the log?

Just to clarify:

1 - I clicked a link from facebook, which opened a blank tab in Chrome (not sure if replacing an existing tab or opening a fresh new one, but it appeared blank), with the progress indicator at a small non-zero position, which remained there forever, and the tab remained blank forever (as if it was taking forever to load, but I doubt any request is started at all) as described in  issue 654540 .
2 - I went to another random existing tab, opened chrome://net-export and started logging
3 - I went back to the original tab (the one opened by clicking on the link from Facebook) and tried tapping in the address bar and submitting the same URL again (nothing happened)
4 - I tapped again on the address bar and started typing another url, then selected one from the list of suggestions from visited urls => nothing happened again
5 - then I went back to the tab where I had initiated logging, I stopped logging and emailed the log to myself, which I then forwarded to you.

@14 yes, sorry:

- the link I clicked at step 1 is a page from huffingtonpost.it
- the url I tried to open at step 4 has the substring "sex" in it :$

Fortunately I've just checked that any of the other existing open tab belong to any of those domains.

Hope this helps.

Errata:

> - the url I tried to open at step 4 has the substring "sex" in it :$

I mean in the very domain name

> Fortunately I've just checked that any of the other existing open tab

I meant NONE of the other existing open tabS

The net log shows no navigation requests, as we kind of guessed. However, there are ConnectJobs and associated Sockets to the urls you mention. To me this indicates that we successfully made it to WebContentsObserver::DidStartNavigationToPendingEntry and launched a predictive connection to those hosts. Somehow things must have gone south after this.

+navigation folks for further triage. ping me for access to the net log.

Let's remove the offline component per #7.

FYI, with the latest stable channel update, I started seeing something similar.  Sometimes when I resume Chrome, the tab that's shown starts reloading and becomes stuck.  I see the loading blue bar below the omnibox.  Yet if I wait, nothing happens.  If I click the "..." menu and click the "x", nothing happens (loading bar still there).  I cannot drag down to refresh/reload (I do not get the spinner).  I cannot use the omnibox to go anywhere (I don't recall if I can't type or just using a suggestion doesn't work).  If I press enter to attempt to reload, nothing changes.  In short, I cannot get this tab to do anything anymore.  I can create a new tab.  I can interact fine on other tabs.  I can use a different tab to navigate to the same page, though that doesn't help the frozen tab.

This behavior is very inconsistent.  I don't have concrete repro steps.

The only way I've found to get this tab to be responsive again is to leave Chrome, make sure it quits (I used the Recents menu and press the "X" on the app), then load Chrome again.

53.0.2785.124
Android 5.0.02
LG G2 (LG-D801)

It happened again to me today, so I can provide more context.  This time it happened in an incognito window that I created by opening a URL in a new tab from another incognito window.  Again, the page didn't load, the "X" in the mehu did work to cancel the load, swipe down to refresh didn't work.  The extra detail I can provide here (that I couldn't recall in comment #19) is that yes, I can focus on the omnibox and type.  If I tap a suggestion, the omnibox displays the URL of the suggestion, but no navigation occurs.  The tapped page doesn't load.  When I pressed back, the whole tab closed and returned me to the tab that spawned this one, indicating that no navigation happened in here when I tried to use the suggestion.

Switching to Untriaged because we have two confirmations.

I agree this behavior started recently, possibly with M-53.

Let me know if you need more information.  You already have a netlog from the other reporter so I'm not sure if I have anything useful to add.

I'm stumped. The only other info that I can think of that could shed light on this is to run an about:tracing log with the deviced attached w/ usb debugging using chrome://inspect?tracing.

The point about the omnibox updating w/ the selected URL I think solidifies the claim that we've created a pending entry and started navigating to it, as I believe the omnibox gets data from the underlying virtual url of the entry. It's possible that code is in a bad state, or there is some problem with spawning the renderer process

I'm going to go ahead and remove Loader and Omnibox labels. I think this is definitely in the navigation code. Also -dimich.

Google should URGENTLY release an "update" that rolls everything back to the version before this regression was introduced, while you figure out how to fix it. I wonder how it hasn't been done yet given the gravity of the issue. 

Chrome is almost unusable. Almost every time I open it I can't navigate anywhere. I've just experienced the same issue in a freshly open incognito window. Often, opening a fresh new tab doesn't fix the issue either. Every time I want to open some page I have to cycle through existing tabs until I find one that works, or close and reopen Chrome etc, until a random sequence of steps finally gets it to work.


In the meantime, is there any way we (users) can downgrade to previous versions looking for one without this issue??

I don't know if this helps, but more than once this week I've had this happen with tabs that were not in the foreground.  I.e., when I resume using Chrome, I can do something with the foreground tab, but one of the other tabs, when I switch to it, shows this problem.

I, at least, haven't seen this problem in many months.

Is anyone else still experiencing this bug?

Closing as obsolete.  (No one has mentioned seeing it at all last year.)