This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
refactor spawning logic. #474
Comments
|
any progress ? 😢 |
|
Finally made some progress here! Code is not in yet, but the spawning logic rewrite happened yesterday. https://twitter.com/leonjza/status/1452516278451773448 |
|
Muchos grazias @leonjza for update 💯 Let us know when you push to remote, we might be of help |
this commit contains many changes. the most important of which is a completely rewritten Agent class, improving support for early instrumentation. pre this commit, because we we're injecting the frida gadget in target applications, we expected communications to occur over a tcp socket. as far back as frida 12.7+ support injecting the frida-server into applications that are debuggable vastly simplifies this process in jailed environments. as such, we can use normal "usb" comms, and thereby more closely follow the same logic as found in frida-* tools. a small changelog for all the changes: - remove the get_device_info() call that relied on frida scripts to get information about the platform. instead we're now using device.query_system_parameters() - replace the old `explore` command with a new `start` command - support spawning new applications - support attaching to the frontmost application - support enabling the node debug port, accessible via the chrome debugger - support immediately resuming apps - replace os with pathlib in some places - remove the `objection device_type` command - add the `resume` repl command to resume apps - add a prompt token to show the current process state (pause/run) no tests were fixed, so expect most of them to fail :P related issues: #473, #474
|
Just pushed the first WIP for the refactor. Most things work. Theres many breaking changes too, but feel free to experiment :P Most notably, the
|
Well, I can't thank you for this. Who knows how many times I tried typing I will sure be among the first who will (s)t(r)est it out 💟 Will update personal observations here asap edit: attaching by $pid has been ditched as it seems. Notable diff: Initialization of new Agent Object via CLI, Gadget ditch, New Agent logic |
If you provide a pid to |
This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
Right now, the objection spawning logic is not great. We can't reliably do early instrumentation, and with Frida 15, some new API's are around that we can use to improve application selection etc. Some quick thoughts on what needs implementing:
The text was updated successfully, but these errors were encountered: