dedup of issue 699274. I suspect it may come from the fix https://codereview.chromium.org/2685333005/.

For more details from my experiment: when crash-and-restore, the start-up-browsers for displaying the crash bubble from both profiles show on the same desktop.

I can take a look why it happens.

Issue 699274 has been merged into this issue.

Actually bisect shows the broken CL is: https://codereview.chromium.org/2695803004

+cl owner and reviewers. Assigned to cl owner first. If a chrome os side investigation is needed instead, feel free to assign back to me.

Any update here? This is pretty bad.

Just back from OOO. I will look into this ASAP.

I cannot repro this on ToT. If I force a browser crash with multiple active profiles, their windows are in the correct respective desktops when the browser is restarted.

I'll try to repro on M58.

Hmm. Doesn't repro on 58.0.3029.68 either. 

I've also just tried on 58.0.3029.31 from the original bug report. I cannot get this issue to reproduce. Here's my test procedure:

1. Log into one account
2. Log into secondary account
3. From either browser window, navigate to chrome://inducebrowsercrashforrealz

I don't see the bug. Are there some other conditions which might affect this behavior?

Re #10, that is the same steps I used for bisecting.

I also gave a try on ToT and now could not reproduce it. It may get silently fixed (and merged to M-58) since the bisect date.

Thanks for looking. That was my thought too, but even the release from the
original bug report does not repro for me. I suspect something may still be
broken but the precise conditions are unclear.

FWIW the CL in question should have no impact on the behavior of session
restore, multi-profile or otherwise. I wonder if the behavior is just flaky
and so a bisect is unreliable?

Oh sorry, I overlooked "I've also just tried on 58.0.3029.31 from the original bug report".

hmm, then it could be just flaky. Yes, I could not strongly see the relation of your CL with this bug, that is why I assigned to you.

Things became tricky now. I did multiple tries today but never saw a repro.

If this does not reproduce, do we still think this should be blocking stable?

I don't think this should be a blocker. Multiprofile use is rare, as far as I'm aware, and this is just annoying and not a showstopper. (And hopefully the browser process doesn't crash *that* often, or we have bigger problems...)

 Issue 715015  has been merged into this issue.

Let's still keep trying to repro in 59, but yeah, let's not hold 58 on it.

Passing back to a Chrome OS owner for further investigation given the conclusions from comments #12 and #13. Please let me know if I can help further!

FWIW, I repro'd this on 59.0.3054.0 - I have no repro steps though, it just happened.

I'm still hearing reports of this. It's been P1 for a while now. What are next steps?

Could not repro it on tot on caroline. 

We have a few reports from Googlers about this issue, all on M63 Beta:
- https://listnr.corp.google.com/report/84684122195
- https://listnr.corp.google.com/report/81407794875
- https://listnr.corp.google.com/report/84715451644

#22 reports don't mention chrome crashes. They might be other issues.

Un-cc-ing me from all bugs on my final day.

Can't be reproduced. Close it.