before: http://build.webkit.sed.hu/builders/ARMv7%20Linux%20Qt5%20Release%20%28Test%29/builds/7316 after: http://build.webkit.sed.hu/builders/ARMv7%20Linux%20Qt5%20Release%20%28Test%29/builds/7318 Between r133953-r134071 there was build failure, so nobody knows which revision broke tests. FYI: Now 94 JSC tests and 1041 layout tests crash, so ARM traditional assembler is something useless now. Could you fix it as soon as possible? Or we should drop supporing it, and remove it from the trunk if it isn't possible to fix serious regressions in months.
I tested r134071 buildfix applying to r133953 and the bug is still valid. So the culprit is r133953 or r134071.
I see the same trend, that the traditional ARM backends are falling behind. How important is the traditional ARM back-end these days? My gut feeling is that every actually relevant ARM SoC does support thumb-2 instructions.
Removing this bug from the blockers for the release of QtWebKit as part of Qt 5.0.0. The release of 5.0.0 depends on bug #76773 and while sadly the traditional ARM JIT may be falling behind, it is not considered important enough to stop the first major release. That said, if fixes make it into trunk in time, then that's great. But there will be future releases of QtWebKit and Qt :)
Fixed by https://trac.webkit.org/changeset/135717 *** This bug has been marked as a duplicate of bug 98857 ***