run-webkit-tests hangs in http/tests/plugins/npapi-response-headers.html. Seen on the build slave many times (see URL) and locally on a G5.
You can reproduce this, Mitz? Have you sampled it (Sharked it) yet during the hang? It apparently started with this commit, although I'm not sure how this change would have caused that test to hang. http://build.webkit.org/post-commit-powerpc-mac-os-x/builds/6415
(In reply to comment #1) > You can reproduce this, Mitz? Yes, 'run-webkit-tests http/tests/plugins'. > Have you sampled it (Sharked it) yet during the > hang? It's just spinning in DRT's main runloop, presumably waiting for the notifyDone().
Created attachment 14061 [details] Sample of DumpRenderTree Sample taken while DumpRenderTree was hanging. There appears to be a lot of activity in CFNetwork.
(In reply to comment #3) > Created an attachment (id=14061) [edit] > Sample of DumpRenderTree > > Sample taken while DumpRenderTree was hanging. There appears to be a lot of > activity in CFNetwork. That's bizarre! There are 1001 invocations of dumpRenderTree() and 995 invocations of runTest()?! Why would dumpRenderTree() be called that many times if the test was really "hung"? What are the arguments to run-webkit-tests on the buildbot (or for this sample)?
I've disabled the test for now.
I am only able to find following reference to this test in Webkit Github mirror: https://github.com/WebKit/WebKit/blob/04657d258e6f723e5a16ee627c29643eedfc2257/LayoutTests/platform/win/TestExpectations#L1032 I think it is now skipped because of lack of NPAPI support. Can we close this bug as "RESOLVED WONTFIX" due to lack of NPAPI support from Safari 14 onward and also not supported in WebkitGTK ports. Thanks!
<rdar://problem/95545541>
The test still exists at LayoutTests/http/tests/plugins/npapi-response-headers.html, we should probably delete the whole directory.