Steps to reproduce:

Create a block with "transform: scaleX(-1)" to flip it horizontally and add clip-path with round corners through clip-path.

Actual results:

Corners are cut as squares.

Expected results:

It should have round corners.

See this codepen:
https://codepen.io/waterplea/pen/ExQrZPv

The Bugbug bot thinks this bug should belong to the 'Core::Graphics: WebRender' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

9:17.61 INFO: No more integration revisions, bisection finished.
9:17.61 INFO: Last good revision: 17ba69318293ff3bcedd0e870a0fab94f47f0d9f
9:17.61 INFO: First bad revision: 13058b0677bb64120fb18ce0a27884db6a781cfc
9:17.61 INFO: Pushlog:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=17ba69318293ff3bcedd0e870a0fab94f47f0d9f&tochange=13058b0677bb64120fb18ce0a27884db6a781cfc

Set release status flags based on info from the regressing bug 1757259

:gw, since you are the author of the regressor, bug 1757259, could you take a look?
For more information, please visit auto_nag documentation.

Set release status flags based on info from the regressing bug 1757259

The severity field is not set for this bug.
:gw, could you have a look please?

For more information, please visit auto_nag documentation.

This is related to how we select surfaces to sometimes have a different raster vs surface spatial node, so that fractional snap offsets work on surfaces in the root coordinate system. I spent a couple hours trying to fix it, but no luck so far. I'm also going to take a look at how much work is involved in a "proper" fix (supporting a unified coordinate system rather than different raster vs. spatial) and see if that might be a better way to fix this case.

gets fixed by the patches in bug 1849661