llnode
llnode copied to clipboard
nodejs
An lldb plugin for Node.js and V8, which enables inspection of JavaScript states for insights into Node.js processes and their core dumps.
An error occured while executing `npm install -g llnode`, and I had already installed `lldb`. There is output: ```console [root@63675c114bd6 llnode]# npm install -g llnode > [email protected] preinstall /usr/local/lib/node_modules/llnode >...
Today there's no easy way to get llnode's version, making it harder to report issues with detailed information. We could either have a llnode command `v8 version` or a command...
The V8 7.6 update requires the following adjustments to the postmortem debugging metadata constants: - `v8dbg_class_FixedTypedArrayBase__base_pointer__Object` - Field moved from FixedTypedArrayBase to JSTypedArray. - Postmortem tools should use `v8dbg_class_JSTypedArray__base_pointer__Object`. -...
Our install script tries to fetch branch `release_71` from `llvm-mirror/lldb`, but this branch doesn't exist (there's a `release_70` though).
V8 stopped using JSArrays to store the stack in https://chromium-review.googlesource.com/c/v8/v8/+/1564060. Now the stack is stored in a raw FixedArray. We'll need to change our heuristics to handle this.
- [ ] `v8 findrefs` equivalent - List of retainers as a property of heap instances? - [x] ~`v8 bt` equivalent~ - Accessible through `LLNode.getProcessObject().threads[0]` - [ ] `v8 getactivehandles`...
I added test cases to the command `findrefs -s` and it happens that once in a while an ECONNRESET error is thrown. ``` Error: read ECONNRESET at exports._errnoException (util.js:1020:11) at...
tl;dr: I think we should standarize the output for all types into JSON format. The default output should hide V8 implementation details and should only show properties and elements accessible...
Right now, when you try to inspect an empty object, this is what you see: ``` (llnode) v8 i 0x000038e4d66492e1 0x000038e4d66492e1: ``` This is a bit confusing, upon further inspection...
Current compilation process throws some warnings. Wouldn't it be a good approach to prevent merge of PRs that introduce new warnings? Or maybe add it to the test pipeline? Current...