We fail this WPT:
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-011.html
http://wpt.live/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-011.html

I think this a version of bug 1310551.

Essentially it seems that we size the firstRowFirstColumn element to be ~1em wide, when in fact it should be ~3em wide. If I use devtools to add width:3em to that element, then it looks visually like the test passes (modulo red fringes which I think don't show up in the test harness where antialiasing is disabled for the Ahem font). So that's the size that we're failing to apply automatically (to shrinkwrap our content).

(To analyze/visualize the behavior-difference vs. other browsers here, it's easiest to remove the green background from the block element, and then add a border around firstRowFirstColumn.)

Actually, we fail a handful of tests with this same prefix:
https://wpt.fyi/results/css/css-grid/alignment/self-baseline?q=grid-self-baseline-changes-grid-area-size

Specifically these ones:
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-003.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-006.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-007.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-008.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-009.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-010.html
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-011.html (the one linked in comment 0)
https://wpt.fyi/results/css/css-grid/alignment/self-baseline/grid-self-baseline-changes-grid-area-size-012.html

I haven't looked through all of them in detail, but the tests look similar, and I'm guessing they're largely failing for a shared reason or small set of reasons.