Repository navigation
Conversation
Have Meson write go-config.json through configure_file(), instead of referring to the file detect.py wrote by name, which Meson warns about and will reject in 2.0.
Otherwise its IOError goes unhandled.
oleavr
force-pushed
the
feature/gum-mapper
branch
6 times, most recently
from
October 28, 2026 10:20
a220526 to
e362df0
Compare
When they're subprojects their vapis are passed by path, so valac never sees the packages they refer to, such as json-glib-1.0 and libsoup-3.0. Use the packages they export through "vala_packages".
The Mach-O parser types were renamed from Gum.Darwin* to Gum.Macho* in frida-gum; follow suit here.
The Mach-O parser types were renamed from Gum.Darwin* to Gum.Macho* in frida-gum; follow suit here.
The Mach-O parser and mapper were renamed from GumDarwin* to GumMacho* in frida-gum, and the mapper now allocates the payload itself, with the image first, followed by the runtime and the extra memory holding our stub code, data and stack. It also no longer drives Mach APIs on the resolver's task, so give it a GumNativeProcessMemory for the target task.
oleavr
force-pushed
the
feature/gum-mapper
branch
3 times, most recently
from
October 28, 2026 16:12
1afb1df to
09f771f
Compare
Map the agent straight into the target with the cross-platform Gum.Mapper instead of writing a temporary DLL and loading it through LoadLibrary. A small emitted stub runs the mapped images' constructors, calls the agent's entrypoint, and runs the destructors when it is not staying resident, in which case the mapper then unmaps it. The agent is copied once into a pagefile-backed section per architecture, whose handle helpers duplicate to map it, so its code pages are shared across every target. The dbghelp and symsrv DLLs are mapped alongside it and handed to the agent, so symbolication doesn't load them either. The agent cloaks these mappings. Gated by the existing mapper option, now enabled by default on Windows, so nothing touches the filesystem; the LoadLibrary path stays for builds with the mapper disabled. On either path the agent is handed the range of the stub, so it stays cloaked for as long as it runs.
The Linux container handed the agent its control socket through a LinuxInjectorState, making it the only non-injected caller that passes an injector state. Encode it in the transport URI instead and pass NULL like on the other OSes.
An agent may make its process exit right away, e.g. through exit() in its entrypoint, and if that happens before we detach, the detach fails with ESRCH even though the injection succeeded.
Map the agent out-of-process instead of having the target dlopen() it. The loader loads any of the agent's dependencies that the target lacks, reserves memory, and reports its address back. Imports are then resolved against the target's libraries, and the loader is sent a plan that it applies: the read-only segments are mapped straight from the agent's file, so their pages are shared across targets, while the rest is streamed over the control socket. The loader also applies protections, runs the constructor, and on unload runs the destructor and unmaps. The target's libraries are opened through /proc/<pid>/map_files when permitted, then through /proc/<pid>/root with an inode check, so targets in containers resolve against their own files. When neither works, e.g. because a library was replaced on disk, the image is copied from the target's memory. Gated by the mapper option, now enabled by default on Linux, and used by the ptrace, procmem and kernel injectors alike. No memfd support is required.
oleavr
force-pushed
the
feature/gum-mapper
branch
from
October 28, 2026 20:32
09f771f to
a8750f0
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.