| To: | Ralf Baechle <ralf@linux-mips.org> |
|---|---|
| Subject: | Re: [PATCH] Proposed NMI handling interface... |
| From: | Thomas Horsten <thomas@horsten.com> |
| Date: | Thu, 9 Oct 2003 13:55:12 +0100 (BST) |
| Cc: | "Steven J. Hill" <sjhill@realitydiluted.com>, <linux-mips@linux-mips.org> |
| In-reply-to: | <20031008160600.GA19102@linux-mips.org> |
| Original-recipient: | rfc822;linux-mips@linux-mips.org |
| Sender: | linux-mips-bounce@linux-mips.org |
On Wed, 8 Oct 2003, Ralf Baechle wrote: > That hook you're proposing isn't even necessary in generic code. NMI > on MIPS hardware is a fairly odd type of exception - it goes straight to > 0xbfc00000 which usually is a a firmware address - and lots of firmware > doesn't even offer an NMI hook. So for those cases where it's possible, > you need to do something firmware anyway before jumping to Linux's NMI > handler. An additional problem with the NMI design of MIPS is it's using > ErrorEPC, just like cache errors so you better pray - or simply design > systems only to rely on NMI for debugging and catastrophic failures. On Lasat boards NMI is used to reboot into Service Mode. There's no hook to override this behaviour. // Thomas |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: What toolchain for vr4181, Geert Uytterhoeven |
|---|---|
| Next by Date: | Re: mips 32 bit HIGHMEM support, Ralf Baechle |
| Previous by Thread: | Re: [PATCH] Proposed NMI handling interface..., Ralf Baechle |
| Next by Thread: | Re: [PATCH] Proposed NMI handling interface..., Brian Murphy |
| Indexes: | [Date] [Thread] [Top] [All Lists] |