86203 2012-05-11 05:51:45 -0700 REGRESSION (r116659?): Assertion failures in JSC marking 2012-05-11 21:53:08 -0700 1 1 1 Unclassified WebKit JavaScriptCore 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P1 Normal --- 1 koivisto msaboff fpizlo ggaren gustavo mhahnenberg msaboff oldest_to_newest 621534 0 koivisto 2012-05-11 05:51:45 -0700 It looks like we started getting lots of marking related crashes after r116659. http://build.webkit.org/results/Lion%20Debug%20(Tests)/r116675%20(6428)/sputnik/Unicode/Unicode_510/S7.6_A3.1-crash-log.txt http://build.webkit.org/results/Lion%20Debug%20(Tests)/r116675%20(6428)/inspector/debugger/script-formatter-console-crash-log.txt http://build.webkit.org/results/Lion%20Debug%20(Tests)/r116677%20(6430)/inspector/styles/show-user-agent-styles-crash-log.txt 622021 1 ggaren 2012-05-11 15:27:56 -0700 Seems like the best thing here is to roll out the patch and rework it. I'd suggest returning to the model where all strings are hash-const'd, to make bugs less elusive. 622232 2 msaboff 2012-05-11 21:52:37 -0700 Rolled out r116659: <http://trac.webkit.org/changeset/116659> in r116822 <http://trac.webkit.org/changeset/116822>.