Able to reproduce the issue on reported version 66.0.3359.139, latest stable# 65.0.3325.181 and on latest Beta# 67.0.3396.56, the same is not seen on latest canary 69.0.3443.0 using Windows 10 and Ubuntu 14.04, hence providing reverse bisect info
Note: Issue is not seen on Mac 10.12.6
Reverse Bisect Info:
================
Last Bad build: 68.0.3423.0
First Good build: 68.0.3424.0

You are probably looking for a change made after 556468 (known good), but no later than 556469 (first known bad).
https://chromium.googlesource.com/chromium/src/+log/18502e69bd4080084bdf69c2cbeda8a629ab72a8..87f271283cfb054062dd24bd1e5fc4f768fca91b
Reviewed-on: https://chromium-review.googlesource.com/1027882

@Javier Fernandez: Please confirm the issue and help in re-assigning if it is not related to your change, please help in merging it to M-67 if applicable
Adding ReleaseBlock-Stable for M-67, feel free to remove it if not applicable.

Thanks!

As this issue exists on M66 per comment #0 and #1, we won't block M67 stable for this. And seems like this is already fix in M68/M69. 

I can not confirm the fix is due to my patch in https://chromium-review.googlesource.com/1027882 though.

Umm, I was able to verify the issue in linux at r556468, and it's fixed by my patch in r556469. However, I still don't fully understand which was the root cause of the issue and why the patch identified by the bisect fixed it. 

My theory is that avoiding layouts during intrinsic size computation reduce the chances to be impacted by bugs related to repeated executions of the grid track sizing algorithm. We were aware of these issues long time ago and we indeed fixed many of them, but I think the mentioned CL is a big step forward the definitive solution. 

However, unless this issue's severity is considered high, I'd rather wait to see if the CL doesn't cause any regression before it's merged in the stable channel. 

 jfernandez@

Still seeing the same issue on Windows 10 & Linux. Please take a look into it as it is marked as stable blocker.
Thanks..!

umm, what do you want me to loot at ? per comment #c6 the patch in r556469 does indeed fix the issue, but it's a risky patch, so I'm not sure it's worth to merge in stable. 

I wouldn't oppose, though, if anybody consider the issue critical enough to take the risk.

BTW, the issue still has the Bug-Regression flag, which I think it's not accurate. 

M67 stable has been out since 05/29 and we're not planning any further M67 release unless extremely critical issue arise.

Pls request a merge to M68 if needed. Thank you.

Bulk update: M68 stable cut is scheduled for July 19th. This issue is marked as RB-Stable, so please take a look at it before. Thanks!

Per c#8, removing blocker label. Please feel free to re-add if someone thinks we need this critical merge to M68.