From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <45756EDE.9080603@domain.hid> Date: Tue, 05 Dec 2006 14:06:38 +0100 From: Gilles Chanteperdrix MIME-Version: 1.0 Subject: Re: [Adeos-main] [PATCH] I-pipe tracer for ARM References: <4575370B.5070002@domain.hid> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit List-Id: General discussion about Adeos List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Sebastian Smolorz Cc: adeos-main@gna.org Sebastian Smolorz wrote: > Hi Gilles, > > Gilles Chanteperdrix wrote: > >>Sebastian Smolorz wrote: >> >>>Hi, >>> >>>here comes a preliminary version of the missing ARM-parts for the I-pipe >>>tracer. It's not yet complete, some features for the tracing of IRQs-off >>>times are missing, but will soon be added. Besides of that, the tracer >>>seems to work. ;-) >> >>That's nice. >> >> >>>The following traces were recorded on a S3C2440. It would be nice if >>>someone could have a look at them and determine if everything looks sane >>>so far. >> >>The patch looks ok, a few remarks however: >>- you include the full head.S in the patch whereas you most likely just >>added the empty mcount; > > > I diff'ed over a ipipe cvs repository where head.S does not exist. Of course, > if I would create the patch over a ipipe'd linux kernel the empty mcount > would be the only change in head.S > > >>- I see no patch of any Kconfig, did you forget them ? > > > As far as I remember no change in any Kconfig file was needed. > > >>- the arm_return_address function seem to rely on frame pointers, so >>should not the tracer option require the CONFIG_FRAME_POINTER option to >>be enabled ? > > > You mean the existing contents of kernel/ipipe/Kconfig.debug are not enough? > > ... > config IPIPE_TRACE > bool "Latency tracing" > depends on IPIPE_DEBUG > select FRAME_POINTER > ... Sorry for the noise, I was not looking at Adeos cvs. -- Gilles Chanteperdrix