> Well, the non-using-ptrace feature is only useful in special situations,
> when building local exploits from a secured computer, etc ..

Yes i think this is cool. But I think is more interesting for reverse
engineering, because it allows you to bypass some antidebugging ptrace-
related tricks.

btw, the thing i like from this project is that it brings a new way for
debugging programs from userland, and with a clean code (gdb source code
really sux).

> That's what i think about:
> * The ERSI language's usability is poor, radare perl r00lz.

hehe, well. They have different aims. ERSI have an api for hooking functions,
etc... and radare scripts and perl is just 'the language itself', and
you have to do everything by hand. Would be cool to have a perl module
to ease all these kind of things. But IMHO keeping the raw interface
as a first instance is better, because it gives a better perspective
for a low-level guy.

> * The external call tracing  looks interesting to do reverse ingenieering.

The ELF load is done entirely in userland by ELFsh, so hooking functions
is merely simple (but slower than the system's one).

IMHO radare needs an interface to ease these kind of hooks.

> * libelfsh let's inject a section to redirect the calls for revesing
> purposes, looks interesting.

Just a idea... dumping with radare the 'mapped' symbols of lib* inside
the binary (where they longjmp to the real library). Modifying it 'offline'
and poking it again will be cool, and having a frontend to parse and
support the modification of this is almost simple.

This can be an easy way for hooking functions, but...i think that to provide
a complete environment for radare we will probably have to implement a
frontend for it. Like the 'gtk' one, to ease and automatize complex or
repetitive tasks and so.

--pancake

_______________________________________________
radare mailing list
[email protected]
https://lists.nopcode.org/mailman/listinfo/radare

Reply via email to