Hi Deepti, Thanks for your response.
On Wed, Sep 12, 2018 at 7:32 PM, 'Deepti Gandluri' via v8-users < v8-users@googlegroups.com> wrote: > If you are looking to track Wasm allocations, there is a counter on the > Isolate that wasm memory allocations are registered > <https://cs.chromium.org/chromium/src/v8/src/wasm/wasm-memory.cc?q=wasm-memory.cc&sq=package:chromium&dr&l=153> > with - V8.WasmAddressSpaceUsageMiB. The page allocator is the catch all for > allocations, so that should work too. > > For GetHeapStatistics > <https://cs.chromium.org/chromium/src/v8/src/api.cc?dr&q=GetHeapSta&sq=package:chromium&g=0&l=8365>(), > WasmMemory is accounted for under malloced memory, could you elaborate on > the discrepancies that you are seeing? > In real tests, I observe that malloced_memory is always 8192. Even after creating a large WebAssembly.Memory object and populating it, malloced_memory continues to return 8192. I'm not sure what this stat tracks, exactly, but it makes sense that it wouldn't count WebAssembly.Memory backing buffers since they are presumably not allocated with malloc(). It looks like external_memory() was recently added, and this actually measures WASM memory, which solves my problem. (And it looks like while I wait for external_memory() to land in stable, I can call AdjustAmountOfExternalAllocatedMemory(0) as an alternative way to get the same information.) > Ideally, Memory.Grow should honor the same resource constraints that a > JSArrayBuffer allocation does, because it either adjusts the externally > allocated memory, > <https://cs.chromium.org/chromium/src/v8/src/wasm/wasm-objects.cc?dr&q=GrowMem&sq=package:chromium&g=0&l=911> > or allocates a new ArrayBuffer. The WasmMemoryTracker tracks wasm > allocations > <https://cs.chromium.org/chromium/src/v8/src/wasm/wasm-memory.cc?dr&q=wasm-memory.cc&sq=package:chromium&g=0&l=29>, > apart from the isolate counters, there is no public API for the WasmEngine > right now. > For now I've implemented a "patrolling" technique, where I kill an isolate after the fact if it uses too much memory. It would be great if a limit on external memory were added to ResourceConstraints and enforced upfront. If I find time I'll look into adding this and submitting a patch. -Kenton -- -- v8-users mailing list v8-users@googlegroups.com http://groups.google.com/group/v8-users --- You received this message because you are subscribed to the Google Groups "v8-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
