Re: dtrace status

From: Juergen Lock <nox_at_jelal.kn-bremen.de>
Date: Sun, 21 Sep 2008 15:09:39 +0200 (CEST)
In article <20080919114528.5yzyki2ry8044g4s_at_0x20.net> you write:
>Quoting John Birrell <jb_at_what-creek.com>:
>
>> On Wed, Sep 17, 2008 at 12:10:13PM +0200, Roman Divacky wrote:
>>> Dtrace was commited 3 months ago and the only things that prevents
>>> using it "out of the box" is building kernel with WITH_CTF=1.
>>>
>>> When is this going to be enabled on default. What is preventing this
>>> from happening?
>>
>> I wonder whether people generally want it enabled by default.
>
>If it doesn't slow anything down, then why not?
>
At least on 7-stable (I didn't try HEAD) kgdb doesn't seem to like dtrace
bits in the kernel, backtraces look like from a kernel without debug
symbols...  Also there seem to be issues with fbt probes,
	http://lists.freebsd.org/pipermail/freebsd-stable/2008-September/045180.html
and SMP support at least on amd64,
	http://lists.freebsd.org/pipermail/freebsd-stable/2008-September/045093.html

>Are there any FreeBSD specific docs on this? Maybe a short article for  
>/usr/share/doc or a new chapter for the handbook? :-)

 I dunno about FreeBSD specific, but I liked this presentation:
	http://video.google.com/videoplay?docid=-8002801113289007228&ei=XLPNSMv5KpKw2QKysZzBAg&q=dtrace
(video is called Dtrace Review if you need to search it.)

 Thanx,
	Juergen
Received on Sun Sep 21 2008 - 11:38:26 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:35 UTC