Hi, > On Thu, Jul 27, 2000 at 06:03:44PM -0500, Stefano Curtarolo wrote: > > > > > Huh ? Any compelling reson, why a fairly recent 2.2 kernel ( I use 2.2.15 > > > ) > > > on a Tsunami arch ? > > > > Tsunamy arch IRQ handling was broken for more than 2 CPUS, until > > 2.4.0-test2 or test3. You get huge troubles with a 4(+)CPUS machine. > > 2.2.x kernels used only CPU 0 for interrupt handling on Alpha; it's > only with 2.4.x that interrupts were able to be delivered to other > CPUs on Alpha.
Ahh. I didn't know that. Well, I begin to see *good* resons to move to 2.4, since I have a bunch of devices running of 4 SCSI channels doing *lots* of stuff - shame on me, I never checked how the interrupts got handled. Well, for the other problem - I don't have it. Only two CPU's here ;-( I once saw a ES40 on a computer show - let me tell you, the d**n thing was too heavy to be carried away by me 8-) Anyway, as for the spinlocks getting stuck sometimes in 2.2.x, is this fixed in 2.4 ( I have checked some of the patches, but by far not all ) ? Reading Alan Cox' ToDo list in the kernel-mailing list somewhat frightens me. However, it is hard to tell, how much of it is Alpha-related, since Alan is - ahem - somewhat x86-centric for that matter ;-) Any serious problems so far in practice ? I would like to try out some of the new things in 2.4... Regards, Thomas Weyergraf ( on late hours again ...) -- Thomas Weyergraf [EMAIL PROTECTED] My Favorite IA64 Opcode-guess ( see arch/ia64/lib/memset.S ) "br.ret.spnt.few" - got back from getting beer, did not spend a lot.
