From mboxrd@z Thu Jan 1 00:00:00 1970 From: Randolph Chung Subject: Re: [parisc-linux] BUG maxcpus=1 still sees 2 CPUs Date: Sun, 5 Dec 2004 00:46:13 -0800 Message-ID: <20041205084613.GU6359@tausq.org> References: <20041205064112.GA10169@colo.lackof.org> Reply-To: Randolph Chung Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: parisc-linux@lists.parisc-linux.org To: Grant Grundler Return-Path: In-Reply-To: <20041205064112.GA10169@colo.lackof.org> List-Id: parisc-linux developers list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: parisc-linux-bounces@lists.parisc-linux.org > It's late and I won't have a chance to chase this bug I just noticed > when trying to boot 2.6.10-rc3-pa1 (64bit, SMP) kernel on a500. fwiw that kernel booted on ios (a500 with 2xpa8500 @ 440mhz) > I set "maxcpus=1" and still get the following output: > CPU(s): 2 x PA8500 (PCX-W) at 440.000000 MHz > Setting cache flush threshold to 231180 (1 CPUs online) right, since that is an inventory message, and we still inventory all the cpus, it will report all the available CPUs. > I'm thinking that while two CPUs are present, only > one is "online", it doesn't feel right to report both > in the first line above. Is this expected behavior > or should we be indicating something like > CPU(s): 1 (of 2) x PA8500 (PCX-W) at 440.000000 MHz we could do that easily enough if you think it's better... > Other issues: > o the box is hanging after sym2 resets the SCSI bus. > My first guess is something is wrong with interrupt delivery. yes, that would explain this as well: > o omitting "maxcpus=1", the box will hang for about 60 seconds > after printing the inventory. And then prints: > Releasing cpu 1 now, hpa=fffffffffffa2000 > and continues on to the SCSI bus resets. that sounds like we timed out bringing up the 2nd cpu, altho if that happened i would have expected an error message. if the 2nd cpu is not processing interrupts, and we are assigning some of the interrupts to the 2nd cpu, then the scsi bus resets will fail. you can try forcing all the irqs to cpu0 and see if that lets the box boot.... > I have tried both gcc 3.0.4 and 3.3.4 (hppa64 versions) with the > same results (AFAICT). currently i'm using gcc-3.4.2 to build the kernel on ios, but in the recent past i've tried 3.0.4 and 3.3.4 and both worked (but that was on -rc2) randolph -- Randolph Chung Debian GNU/Linux Developer, hppa/ia64 ports http://www.tausq.org/ _______________________________________________ parisc-linux mailing list parisc-linux@lists.parisc-linux.org http://lists.parisc-linux.org/mailman/listinfo/parisc-linux