FYI, this also affects many other types of grid items, not just blocks.
Here's an <iframe> example.

Deleted: grid-item-iframe.html 503 bytes

grid-item-iframe.html 503 bytes View Download

I would appreciate it if you fix this before enabling Grid by default,
because otherwise I suspect it will cause web-compat issues.  Thanks.

Affects image grid items as well.

Here's a link to the image grid item test:
https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/reftests/css-grid/grid-item-auto-min-size-clamp-003.html

I'm sorry but I'm not sure that's the expected behavior.
I've been discussing this with fantasai on GitHub:
https://github.com/w3c/csswg-drafts/issues/283#issuecomment-247545477

She says that "stretch" and "start" cases should be different.
I'm attaching a simple example where it's clear the difference between
Chromium and Firefox.

Adding Tab on CC if he wants to clarify the topic.

Deleted: minimum-size.html 981 bytes

minimum-size.html 981 bytes View Download

Deleted: minimum-size-chromium.png 29.1 KB

	minimum-size-chromium.png 29.1 KB View Download

Deleted: minimum-size-firefox.png 45.3 KB

	minimum-size-firefox.png 45.3 KB View Download

Any update on this?

The CSSWG resolution is that, for replaced elements, "nor4mal" equals "start" - that is, no stretching takes places.

This is because, for replaced elements, you really want to pay attention to the intrinsic aspect ratio.  You don't want to fully stretch, as normal elements do, to the dimensions of the grid area, because it distorts the element.

Instead, per WG resolution, we're going to introduce the "contain" keyword to 'width' and 'height', which fills the area while respecting the aspect ratio.  But absent that, replaced elements just default to being start/start aligned.

This bug is not about replaced elements, that one is  bug #666961 .
And yeah we need to review and update our implementation there.
We didn't do anything regarding "normal" and replaced elements yet.
We were waiting for the CSS WG resolution which seems now clear in:
https://github.com/w3c/csswg-drafts/issues/523

This bug is about regular elements and implied minimum size.
I believe current Chromium behavior is correct
given the last comment by fantasai at:
https://github.com/w3c/csswg-drafts/issues/283#issuecomment-268125974

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/71efe49d24b1154125e1f07e772151794a156265

commit 71efe49d24b1154125e1f07e772151794a156265
Author: rego <rego@igalia.com>
Date: Sat Feb 25 15:18:05 2017

[css-grid] Unskip Automatic Implied Minimum Size tests

The tests have been updated upstream on csswg-test repository,
so we don't need to skip them anymore.

There are 2 tests failing out of 18 checking this feature:
* 007 & 009: These are related to transferred size of
  replaced grid items ( crbug.com/555027 ).

BUG= 666940 

Review-Url: https://codereview.chromium.org/2700623002
Cr-Commit-Position: refs/heads/master@{#453093}

[modify] https://crrev.com/71efe49d24b1154125e1f07e772151794a156265/third_party/WebKit/LayoutTests/TestExpectations

I've updated the test suite and we're now passing it on Chromium:
https://github.com/w3c/csswg-drafts/issues/283#issuecomment-279987986

I believe this is the right behavior, but I might be missing something.
Please reopen this if needed.