| To: | Thomas Koeller <thomas@koeller.dyndns.org> |
|---|---|
| Subject: | Re: timer irq setup |
| From: | Ralf Baechle <ralf@linux-mips.org> |
| Date: | Tue, 18 Dec 2007 11:48:34 +0100 |
| Cc: | linux-mips@linux-mips.org |
| In-reply-to: | <200712171135.24569.thomas@koeller.dyndns.org> |
| Original-recipient: | rfc822;linux-mips@linux-mips.org |
| References: | <200712171135.24569.thomas@koeller.dyndns.org> |
| Sender: | linux-mips-bounce@linux-mips.org |
| User-agent: | Mutt/1.5.17 (2007-11-01) |
On Mon, Dec 17, 2007 at 11:35:22AM +0100, Thomas Koeller wrote: > In arch/mips/kernel/traps.c, per_cpu_trap_init() contains > code to set up the global cp0_compare_irq variable. Does > this make sense? I'd say that either the irq setup should > be moved to trap_init(), or cp0_compare_irq should be > changed to a per-cpu variable, depending on what the original > intention was. Technically it would be legal to route the timer interrupt to a different interrupt for each of possibly multiple processors, that's why it's done in per_cpu_trap_init(). Obviously the rest of the code isn't quite there and also to best of my knowledge all actual implementations use the same interrupt across all cores. So something like a global variable and a paranoia check in per_cpu_trap_init() to ensure all CPUs really use the same interrupt would seem reasonable. Ralf |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [UPDATED PATCH] SGISEEQ: use cached memory access to make driver work on IP28, Thomas Bogendoerfer |
|---|---|
| Next by Date: | Re: [PATCH 1/4] Introduced GPI_RM9000 configuration parameter, Ralf Baechle |
| Previous by Thread: | timer irq setup, Thomas Koeller |
| Next by Thread: | [PATCH][MIPS] fix user_cpus_allowed assignment, Pavel Kiryukhin |
| Indexes: | [Date] [Thread] [Top] [All Lists] |