+++ This bug was initially created as a clone of Bug #325427 +++ This appears to be a regression from the testcase for bug 325427 landing. > 49238 [NNNNN] WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file layout/base/nsCSSFrameConstructor.cpp, line 4010 This warning [1] shows up in the following test suites: > mozilla-central_ubuntu64_vm-debug_test-crashtest-e10s-bm116-tests1-linux64-build33.txt:24426 > mozilla-central_ubuntu64_vm-debug_test-crashtest-bm54-tests1-linux64-build36.txt:24426 > mozilla-central_ubuntu64_vm-debug_test-reftest-e10s-2-bm120-tests1-linux64-build24.txt:153 > mozilla-central_ubuntu64_vm-debug_test-reftest-3-bm117-tests1-linux64-build2.txt:121 > mozilla-central_ubuntu64_vm-debug_test-reftest-4-bm52-tests1-linux64-build1.txt:62 > mozilla-central_ubuntu64_vm-debug_test-reftest-e10s-1-bm121-tests1-linux64-build1.txt:13 > mozilla-central_ubuntu64_vm-debug_test-web-platform-tests-e10s-4-bm114-tests1-linux64-build4.txt:10 > mozilla-central_ubuntu64_vm-debug_test-web-platform-tests-4-bm68-tests1-linux64-build2.txt:10 > mozilla-central_ubuntu64_vm-debug_test-reftest-2-bm68-tests1-linux64-build1.txt:8 > mozilla-central_ubuntu64_vm-debug_test-reftest-1-bm52-tests1-linux64-build3.txt:4 > mozilla-central_ubuntu64_vm-debug_test-mochitest-3-bm115-tests1-linux64-build60.txt:2 > mozilla-central_ubuntu64_vm-debug_test-mochitest-e10s-5-bm113-tests1-linux64-build2.txt:1 > mozilla-central_ubuntu64_vm-debug_test-mochitest-e10s-3-bm68-tests1-linux64-build6.txt:1 > mozilla-central_ubuntu64_vm-debug_test-mochitest-5-bm125-tests1-linux64-build3.txt:1 It shows up in 99 tests. A few of the most prevalent: > 48824 - file:///builds/slave/test/build/tests/reftest/tests/layout/svg/crashtests/325427-1.svg > 42 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/image/image-preserveAspectRatio-04.svg > 42 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/image/image-preserveAspectRatio-03.svg > 40 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/image/image-preserveAspectRatio-02-raster.svg > 40 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/image/image-preserveAspectRatio-01-raster.svg > 30 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/use-01-extref.svg > 8 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/use-02-extref.svg > 6 - file:///builds/slave/test/build/tests/reftest/tests/layout/svg/crashtests/467498-1.svg > 6 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/tspan-rotate-04.svg > 6 - file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/svg/tspan-rotate-02.svg [1] https://hg.mozilla.org/mozilla-central/annotate/f7b746b4e913+/layout/base/nsCSSFrameConstructor.cpp#l4010

The hg.m.o source URL is broken (stray "+") -- corrected URL: https://hg.mozilla.org/mozilla-central/annotate/f7b746b4e913/layout/base/nsCSSFrameConstructor.cpp#l4010 Quoting the source, this is the NS_ENSURE_TRUE here: > 4000 nsresult > 4001 nsCSSFrameConstructor::GetAnonymousContent(nsIContent* aParent, > 4002 nsIFrame* aParentFrame, > 4003 nsTArray<nsIAnonymousContentCreator::ContentInfo>& aContent) > 4004 { > 4005 nsIAnonymousContentCreator* creator = do_QueryFrame(aParentFrame); > 4006 if (!creator) > 4007 return NS_OK; > 4008 > 4009 nsresult rv = creator->CreateAnonymousContent(aContent); > 4010 NS_ENSURE_SUCCESS(rv, rv);

That CreateAnonymousContent line (4009) is returning NS_ERROR_FAILURE, due to failure in its call to SVGUseElement::CreateAnonymousContent() called on its local variable "use". And SVGUseElement::CreateAnonymousContent() is failing due to its "circular loop detection", "check 2". We're detecting a reference loop (correctly) and refusing to create anonymous content. So, this seems like perhaps we shouldn't bother warning about (at least, not with NS_ENSURE_SUCCESS).

Following up on comment 2 -- here's a link to where the failure originates from: https://mxr.mozilla.org/mozilla-central/source/dom/svg/SVGUseElement.cpp?rev=e8c7dfe727cd&mark=241-241,247-247,254-254#241 (It's line 254, which is a "return null" under "check 2", which is testing whether we're cloning something that is already in our clone hierarchy.)

And also for reference, the testcase that's spamming all these warnings is attachment 593788 [details] (recently checked in as "325427-1.svg" as noted in comment 0).

Comment on attachment 8680822 [details] [diff] [review] fix v1: use explicit early-return instead of spammy NS_ENSURE_SUCCESS Can you also add a comment noting this is one of the reasons the call might fail?

Will do. (Just noticed this bug had slipped out of my scope; I'll circle back and land it, with that code-comment, later today.)

Added comment before landing: // CreateAnonymousContent failed, e.g. because the page has a <use> loop.