Hi Kamyar... On Wed, Aug 4, 2010 at 01:44, Kamyar Mohajerani <[email protected]> wrote: > I have studied and tried kprobe too but I guess I doesn't exactly do what I > need here. I could still be wrong.
You mean, kprobe doesn't do what you wanna do? > the printk replacement is an exact copy from printk's definition in > kernel/printk.c. It calls vprintk with it's arguements. I have also removed > all the statements in function's body and tested. The same result. By "with > a one" I mean <a function>, which is my printk replacement. All I know, if something brings havoc in kernel, it could be one of these things: - null reference - sleep inside atomic or interrupt context - stack overflow (won't be obvious...) - deadlock or race condition - possibly many more... So ask yourself, which one it might be? The displayed/ emitted Error message, possibly in kernel panic could give you a clue. But I also have something for you to check: perhaps you also need to flush instruction and/or TLB. The reason: the address of the symbol could be cached somewhere by the current callers. Thus, it clash with the interceptor function. Try to google about it...i completely forgot where I read it...maybe in one of the phrack issues too...which talk about PaX..specifically emulating NX bit. -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to [email protected] Please read the FAQ at http://kernelnewbies.org/FAQ
