153482 2016-01-26 05:54:02 -0800 REGRESSION(r195575): It made all JSC tests crash on ARMv7 Linux 2016-01-27 03:30:25 -0800 1 1 1 Unclassified WebKit JavaScriptCore Other Unspecified Unspecified RESOLVED WORKSFORME P1 Blocker --- 108645 153448 1 ossy webkit-unassigned ggaren kling ossy oldest_to_newest 1158794 0 ossy 2016-01-26 05:54:02 -0800 https://trac.webkit.org/changeset/195575 made all JSC tests crash on ARMv7 Linux. EFL ARMv7 (with ARM instruction set): ====================================== before: https://build.webkit.org/builders/EFL%20Linux%20ARMv7%20Traditional%20Release/builds/16758 after: https://build.webkit.org/builders/EFL%20Linux%20ARMv7%20Traditional%20Release/builds/16752 EFL ARMv7 (with Thumb2 instruction set): ========================================= before: https://build.webkit.org/builders/EFL%20Linux%20ARMv7%20Thumb2%20Release/builds/16899 after: https://build.webkit.org/builders/EFL%20Linux%20ARMv7%20Thumb2%20Release/builds/16906 1158857 1 ossy 2016-01-26 10:23:28 -0800 This asserts hit in debug build: Source/JavaScriptCore/assembler/ARMv7Assembler.h(2206) : static void JSC::ARMv7Assembler::relinkJump(void*, void*) ASSERTION FAILED: !(reinterpret_cast<intptr_t>(to) & 1) 1158867 2 ossy 2016-01-26 10:47:45 -0800 full crash log: ASSERTION FAILED: !(reinterpret_cast<intptr_t>(to) & 1) ../../Source/JavaScriptCore/assembler/ARMv7Assembler.h(2206) : static void JSC::ARMv7Assembler::relinkJump(void*, void*) 1 0xb64c3868 WTFCrash 2 0xb5fe3fe4 JSC::ARMv7Assembler::relinkJump(void*, void*) 3 0xb61f4694 JSC::AbstractMacroAssembler<JSC::ARMv7Assembler, JSC::MacroAssemblerARMv7>::repatchNearCall(JSC::CodeLocationNearCall, JSC::CodeLocationLabel) 4 0xb61f1790 JSC::linkFor(JSC::ExecState*, JSC::CallLinkInfo&, JSC::CodeBlock*, JSC::JSFunction*, JSC::MacroAssemblerCodePtr) 5 0xb61d45ae Segmentation fault Unfortunately it isn't easy to debug this regression, because GDB crashes :( 1159262 3 ossy 2016-01-27 03:30:25 -0800 I don't know what happened, but after a clean build release crashes went away. It seems the debug assert is unrelated to this bug, but still valid. I'm going to file a new bug report for it and start to investigate.