From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Mon, 16 Aug 2010 13:13:43 +0100 (BST) From: edward.robbins@domain.hid MIME-Version: 1.0 Content-Type: text/plain;charset=UTF-8 Content-Transfer-Encoding: quoted-printable In-Reply-To: <4C65583D.3030800@domain.hid> References: <1281709057.650414331@domain.hid> <4C65583D.3030800@domain.hid> Message-ID: <1281960823.030215299@domain.hid> Subject: Re: [Xenomai-help] Very high latencies under stress testing List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gilles Chanteperdrix Cc: xenomai@xenomai.org =0A=0AOn Friday, 13 August, 2010 3:35pm, "Gilles Chanteperdrix" said:=0A=0A> edward.robbins@domain.hid= k wrote:=0A>>=0A>> I wonder if this is to do with graphics? I have set "NoA= ccel" in the=0A>> driver section of Xorg.conf (I am using fbdev). I also no= tice that=0A>> latencies go massive (e.g. 4200uS) the first time I run glxg= ears.=0A>> "glxinfo" reports "Direct rendering: no".=0A> =0A> This is bad n= ews. It could come either from the fact that the Xorg=0A> driver shuts hard= ware interrupts off (there is nothing xenomai can do=0A> about that), or it= could be an SMI issue. What about using the driver=0A> for the real graphi= c card you have ? Probably intel card? Do you observe=0A> any big latencies= if you stress the system with anything else than X?=0A> =0A>> Power manage= ment options (ACPUI, APM): Disable ACPI (Advanced=0A>> Configuration and Po= wer Interface) Support --> Processor Disable CPU=0A>> Frequency Scaling Dis= able CPU idle PM support Disable Power Managment=0A>> Support=0A> =0A> Disa= bling ACPI is a bad idea. As repeated many times, only=0A> ACPI_PROCESSOR s= hould be disabled, not ACPI.=0A> =0A=0AThanks for your helpful responses.= =0A=0AI'm not working on this project again until next week, but on Friday = I was able to=0A=0A o Recompile the kernel with ACPI enabled =0A o Boot wit= h the 'SMI handler' BIOS option set to disabled....=0A according to the h= elp text this option should be disabled if 'an RTOS =0A is installed' - I= presume this means that SMIs are completely disabled =0A and therefore S= MIs cannot be the source of the problem =0A (very convenient BIOS option..= . now I begin to appreciate the expense of this board!) =0A o Changed xorg = to use the intel video driver=0A=0AAfter doing this I still have problems w= ith latencies and X, but did not have a chance to stress the system with no= n-graphics tasks. I am thinking that when I get back to it next week I will= compile the kernel a few times and also try writing a large file a number = of times to the disk as Philippe suggested. Hopefully this will not cause a= ny problems and I can then more confidently attribute the high latency to t= he graphics system. I will also attempt to install the I-pipe tracer as sug= gested. I'll update when (if!?) I get that far.=0A=0AThanks=0AEdward