Status: New
Owner: ----
New issue 2512 by [email protected]: JitCodeEvent API emits same
CODE_ADDED event twice
http://code.google.com/p/v8/issues/detail?id=2512
CODE_ADDED code_start=0x1975c71b4e40 code_len=93 name="StoreIC:incoming"
CODE_ADDED code_start=0x1975c71b4e40 code_len=93 name="StoreIC:incoming"
That's non-snipped, i.e. no in-between code events. With backtrace info:
CODE_ADDED code_start=0x1975c71b4e40 code_len=93 name="StoreIC:incoming"
out/Debug/node[0x703242]
out/Debug/node(_ZN2v88internal6Logger19IssueCodeAddedEventEPNS0_4CodeEPKcm+0x62)[0x8ff8bc]
out/Debug/node(_ZN2v88internal6Logger15CodeCreateEventENS1_16LogEventsAndTagsEPNS0_4CodeEPNS0_6StringE+0x102)[0x90145e]
out/Debug/node(_ZN2v88internal9StubCache17ComputeStoreFieldENS0_6HandleINS0_6StringEEENS2_INS0_8JSObjectEEEiNS2_INS0_3MapEEENS0_14StrictModeFlagE+0x1b8)[0xa55b26]
out/Debug/node(_ZN2v88internal7StoreIC12UpdateCachesEPNS0_12LookupResultENS0_16InlineCacheStateENS0_14StrictModeFlagENS0_6HandleINS0_8JSObjectEEENS6_INS0_6StringEEENS6_INS0_6ObjectEEE+0x1c7)[0x8b59e7]
out/Debug/node(_ZN2v88internal7StoreIC5StoreENS0_16InlineCacheStateENS0_14StrictModeFlagENS0_6HandleINS0_6ObjectEEENS4_INS0_6StringEEES6_+0x5df)[0x8b55bb]
out/Debug/node(_ZN2v88internal12StoreIC_MissENS0_9ArgumentsEPNS0_7IsolateE+0x12c)[0x8b87f2]
[0x1975c710654e]
CODE_ADDED code_start=0x1975c71b4e40 code_len=93 name="StoreIC:incoming"
out/Debug/node[0x703242]
out/Debug/node(_ZN2v88internal6Logger19IssueCodeAddedEventEPNS0_4CodeEPKcm+0x62)[0x8ff8bc]
out/Debug/node(_ZN2v88internal6Logger15CodeCreateEventENS1_16LogEventsAndTagsEPNS0_4CodeEPNS0_18SharedFunctionInfoEPNS0_6StringEi+0x193)[0x901b2f]
out/Debug/node(_ZN2v88internal8Compiler25RecordFunctionCompilationENS0_6Logger16LogEventsAndTagsEPNS0_15CompilationInfoENS0_6HandleINS0_18SharedFunctionInfoEEE+0x1b8)[0x7c81a4]
out/Debug/node[0x7c6af9]
out/Debug/node(_ZN2v88internal8Compiler11CompileLazyEPNS0_15CompilationInfoE+0x1db)[0x7c6f7b]
out/Debug/node[0x93e207]
out/Debug/node(_ZN2v88internal10JSFunction11CompileLazyENS0_6HandleIS1_EENS0_18ClearExceptionFlagE+0x119)[0x93ea6d]
out/Debug/node(_ZN2v88internal11CallIC_MissENS0_9ArgumentsEPNS0_7IsolateE+0x195)[0x8b8182]
[0x1975c710654e]
The call graphs are different but the emitted event is the same. Is that an
artefact of the situation described in v8.h or something else?
"Since code removal notifications are not currently issued, the
event_handler may get notifications of code that overlaps earlier
code notifications. This happens when code areas are reused, and the
earlier overlapping code areas should therefore be discarded."
The block of memory that event->code_start points to doesn't change (in the
memcmp() sense) between events.
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
