|
Message-ID: <CA+fZqCVUYRTOKxfcQ+NAOxqdQdPXOu3+dDffSEXFmrwmRL2djQ@mail.gmail.com> Date: Thu, 7 Dec 2017 14:19:24 +0100 From: ardi <ardillasdelmonte@...il.com> To: musl@...ts.openwall.com Subject: Easiest way of capturing syscalls, all-architectures-wide? Is there some place (some header, I suppose) where I could redefine the syscall interface in a global way, for all CPU architectures, and redirect all syscalls to a custom implementation of the Linux call table? For this scenario I'm thinking in totally bypassing the arch-dependent assembly syscall code, and redirect it to a C-written interface. Can this be done in an smart way that would involve perhaps just editing one file? Maybe you have this feature already implemented for your testing purposes, but it's disabled in release builds? (I'm asking this for not reinventing the wheel in case you already have this feature) I'm looking at src/internal/syscall.h in this very same moment, and it looks like the redirection can be done there, but... would this capture all syscalls, or maybe some parts of musl might still be issuing syscalls that don't go through src/internal/syscall.h ? Also, is there an easy way of disabling the build of the assembly syscall code, just to be 100% confident that I build without the risk of doing direct syscalls? Thanks!! ardi
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.