Sorry....i don't understand what u meant :-):
On Sat, Jul 26, 2008 at 8:39 PM, Mulyadi Santosa
<mulyadi.santosa@xxxxxxxxx> wrote:
NB: Anyway, I just found out that we know have ftrace in latest kernel...
what is this ftrace to do with the present discussion? I am really
eager to learn this thing...let's discuss:
a. how is ftrace different from utrace? utrace, as i understood
from Roland's email, is that he is getting rid of all the ptrace
stuff, and replacing it with utrace....overcoming all the latency
problems in ptracing.... so theoretically ptracing is supposed to be
much faster. have not understood Roland's patches yet....not sure
how ptrace is being patched by utrace.
Let's learn and share:
http://lwn.net/Articles/224779/
utrace infrastructure has already been used by a Japanese to debug a
userspace program "xeyes"...amazing..
b. ftrace......by Steve from redhat, is for tracing latency
scheduler...if i understood correctly.....but it is least understood
by me....
c. the one amazes me a lot is LTTng....Matthieu mentioned that it
can trace almost anything, even interrupt and memory pagefault
mechanism.......as it avoid the use of spinlocks when in its
infrastructure...
d. now....systemtap is beginning to use the utrace infrastructure as
well....just heard of it only....someone please show a sample
systemtap script that uses utrace infrastructure?
e. btw...what is the difference between tracehook and kprobe? then
jprobe? lots of terminology....but not sure if all boils down to the
same thing....?
Hm....got lots of stuff to checkout..
When do you find time to sleep? ;)