Re: [PATCH] Longhaul - Fix hole

[Dave Jones <davej@redhat.com>]

On Fri, Dec 29, 2006 at 09:32:29PM +0100, Rafał Bilski wrote:
 > ACPI PM2 register was fallback for "Longhaul ver. 1" CPU's. 
 > My assumption that this register isn't present at 
 > "PowerSaver" motherboards is so far true, but current code 
 > will not work correctly in other case. There are three possible 
 > supports: ACPI C3, PM2 and northbridge. That was my assumption 
 > that ACPI C3 and northbridge is for PS and northbridge and PM2 
 > is for V1. In current code we can only check if it is ACPI 
 > support or not by port22_en. So remove port22_en and add 
 > longhaul_flags. If USE_ACPI_C3 and USE_NORTHBRIDGE are both 
 > clear then it means ACPI PM2 support. Also change order of 
 > support probe from ACPI C3, PM2, northbridge to ACPI C3, 
 > northbridge, ACPI PM2. Paranoid protection against port 0x22 
 > cast as ACPI PM2 register. Bit 1 clear in such case - lockup 
 > on AGP DMA. And obvious (now) fixup for do_powersaver. Use 
 > cx->address only for ACPI C3 ("PowerSaver" processor using 
 > PM2 support).

Much better description, thanks.
I renamed the subject "Fix up powersaver assumptions" which is a bit
more descriptive.
 
 > diff --git a/arch/i386/kernel/cpu/cpufreq/longhaul.c b/arch/i386/kernel/cpu/cpufreq/longhaul.c
 > --- a/arch/i386/kernel/cpu/cpufreq/longhaul.c
 > +++ b/arch/i386/kernel/cpu/cpufreq/longhaul.c
 > @@ -52,6 +52,10 @@
 > #define	CPU_EZRA_T	4
 > #define	CPU_NEHEMIAH	5
 > 
 > +/* Flags */
 > +#define USE_ACPI_C3		(1 << 1)
 > +#define USE_NORTHBRIDGE		(1 << 2)
 > +
 > static int cpu_model;
 > static unsigned int numscales=16;
 > static unsigned int fsb;

The diff got whitespace damaged this time, so I glued the
description above on the old diff, which applied
(after I fixed up the trailing whitespace addition which git hates).

Applied though finally :)

Thanks,

		Dave

-- 
http://www.codemonkey.org.uk