From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <485B8332.90908@domain.hid> Date: Fri, 20 Jun 2008 12:15:14 +0200 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <5719c6f30805050943p31f3f5eew74d0f7ef3dd2b5a5@domain.hid> <4820B784.3080306@domain.hid> <5719c6f30806190900v5fea670amef6954efaae59882@domain.hid> In-Reply-To: <5719c6f30806190900v5fea670amef6954efaae59882@domain.hid> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Adeos-main] Advice regarding MIPS port List-Id: General discussion about Adeos List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Florent Audebert Cc: adeos-main@gna.org Florent Audebert wrote: > Hi, > > I'm back on my work concerning Adeos port on MIPS. I'm a little bit lost atm. > > My kernel is stuck in calibrate_delay() loop. Timer interrupts seem to > occur well but my i-pipe is stalled. There is no arch-dependent places > where I explicitely unstall root domain. The root domain should be unstalled in the start_kernel function, file init/main.c by the call to local_irq_enable(), which should result in an ipipe_unstall_root(). > I saw fast stall / unstall > calls in other archs entry.S. Are these calls mandatory or are they > here in order to reduce latency ? They are mandatory in a final I-pipe patch, but you can ignore them for a first attempt, without them, you will get big latencies, that is all. > > In case the previous asumption is false, is there any other potential > leads ? (Something wrong in my local_irq_* ?) Well, when debugging, every ugly hack is permitted, so you can put as much printks as you want in the interrupt handler or the calibrate delay loop. > > I didn't post my code so far since nothing is working yet... If you > think it could worth it, I'll be glad to send you a git-diff from my > 2.6.24 working directory. > -- Gilles Chanteperdrix