Use new intrinsic content height when computing min-height: min-content

third_party/WebKit/Source/core/layout/LayoutBox.cpp

Index: third_party/WebKit/Source/core/layout/LayoutBox.cpp
diff --git a/third_party/WebKit/Source/core/layout/LayoutBox.cpp b/third_party/WebKit/Source/core/layout/LayoutBox.cpp
index 988307e39943873e7b46fd9da4ccf07acff1f20d..36b9ed500fc573509c1f8c3350fd0b29c3131938 100644
--- a/third_party/WebKit/Source/core/layout/LayoutBox.cpp
+++ b/third_party/WebKit/Source/core/layout/LayoutBox.cpp
@@ -2673,9 +2673,9 @@ LayoutUnit LayoutBox::computeIntrinsicLogicalContentHeightUsing(const Length& lo
     if (logicalHeightLength.isMinContent() || logicalHeightLength.isMaxContent() || logicalHeightLength.isFitContent()) {
         if (isAtomicInlineLevel())
             return intrinsicSize().height();
-        if (m_intrinsicContentLogicalHeight != -1)
-            return m_intrinsicContentLogicalHeight;
-        return intrinsicContentHeight;
+        if (intrinsicContentHeight != -1)

[eae, 2016/05/12 19:24:56]

Shouldn't we be able to always use intrinsicContentHeight here?
m_intrinsicContentLogicalHeight can be -1 but isn't intrinsicContentHeight
guaranteed to be set here?

[cbiesinger, 2016/05/12 19:49:45]

I believe you're correct, especially given that
computeContentAndScrollbarLogicalHeightUsing (=the main caller of this function)
returns -1 if intrinsic height is -1. Callers should just pass
intrinsicContentHeight() if desired.

availableLogicalHeight() might want to consider that... I added a FIXME comment.
And I fixed another caller to pass something better. Those two are the only ones
where it matters, I think.

+            return intrinsicContentHeight;
+        return m_intrinsicContentLogicalHeight;
     }
     if (logicalHeightLength.isFillAvailable())
         return containingBlock()->availableLogicalHeight(ExcludeMarginBorderPadding) - borderAndPadding;