From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <48C0041A.5020001@domain.hid> Date: Thu, 04 Sep 2008 17:51:54 +0200 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <48BFA4C1.1040800@domain.hid> In-Reply-To: Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 8bit Subject: Re: [Xenomai-help] spurious interrupt with latest release List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: xenomai@xenomai.org stephane ancelot wrote: > Le Thu, 04 Sep 2008 11:05:05 +0200, Gilles Chanteperdrix > a écrit: > >> stephane ancelot wrote: >>> Hi, >>> Finally since I did not understand why this happened and to inhibit any >>> driver writing related problems,I made the following architecture : >>> >>> a single 5ms task triggs a PLX 9050 INTCSR software register to activate >>> INTA (level irq) >>> >>> the rtdm interupt handler checks if IT comes from the PLX, disable PLX >>> IT, >>> resets the software register , reenable PLX it, >>> and ack irq rtdm >>> >>> However at random time , I have always spurious interrupt in this >>> handler. >>> The time it occurs is random . >>> >>> attached my kernel config.gz >> Hi, >> >> Did you check the differences in masking/acking/unmasking interrupts for >> the interrupt controller you use between the I-pipe patch that works and >> the I-pipe patch that does not work ? >> > > I know that it rocks with > kernel 2.6.24.2 and adeos-ipipe-2.6.24-x86-2.0-03.patch > > it does not with : > kernel 2.6.26.2 and adeos-ipipe-2.6.26-x86-2.0-09.patch Well, it would be nice to try 2.6.25. > Having a look at the patch I did not manage to find big differencies only > in entry_32.S sysenter for enabling hw it, but I am not sure if this part > of code may be related to my problem > > The routed IRQ number is 18d > > enabling ipipe and kernel debugging options does not help. > > > It would be useful , if I could reproduce the problem using a standard > component of the PC to trig an IT like the PLX soft register, in order > produce a regression test and some people being able to reproduce > it.Supposing the same problem should appear :-) And if you run cat /proc/interrupts on the two kernels, do you observe any difference, as for instance, irqs changing from edge to level, or from local-apic to IO-apic or to fasteoi, or whatever ? -- Gilles.