1487214 - We don't seem to memory-report the memory JitcodeGlobalTable::allocateEntry allocates

Status: NEW

(Core :: JavaScript Engine: JIT, enhancement, P3)

Description

[Boris Zbarsky [:bzbarsky]]

From a DMD report, we have 34 allocations of 16KiB each that are not reported:

Unreported {
  34 blocks in heap block record 36 of 6,988
  557,056 bytes (557,056 requested / 0 slop)
  Individual block sizes: 16,384 x 34
  0.12% of the heap (26.60% cumulative)
  0.31% of unreported (72.28% cumulative)
  Allocated at {
    #01: replace_malloc(unsigned long) (DMD.cpp:1267, in libmozglue.dylib)
    #02: js::LifoAlloc::getOrCreateChunk(unsigned long) (LifoAlloc.cpp:32, in XUL)
    #03: js::jit::JitcodeGlobalTable::allocateEntry() (LifoAlloc.h:609, in XUL)
    #04: js::jit::JitcodeGlobalTable::addEntry(js::jit::JitcodeGlobalEntry const&) (JitcodeMap.cpp:503, in XUL)
    #05: js::jit::BaselineCompiler::compile() (BaselineCompiler.cpp:293, in XUL)
    #06: js::jit::BaselineCompile(JSContext*, JSScript*, bool) (BaselineJIT.cpp:266, in XUL)
    #07: js::jit::CanEnterBaselineMethod(JSContext*, js::RunState&) (BaselineJIT.cpp:310, in XUL)
    #08: js::jit::MaybeEnterJit(JSContext*, js::RunState&) (Jit.cpp:153, in XUL)
  }
}

Comment 1

[Jan de Mooij [:jandem]]

We have plans to remove this code soon, so I think we should hold off doing anything here for now.

Comment 2

[Andrew McCreight [:mccr8]]

From skimming source code, it looks like this still exists.