[PATCH] add AMD Geode processor support

[PATCH] add AMD Geode processor support

[Dan Malek]

[-- Attachment #1: Type: text/plain, Size: 223 bytes --]



Hi Marcelo!

Here is a patch for 2.4 that adds the basic AMD Geode GX2/GX3
and GX1/SC1200 support.  This patch updates configuration
scripts, defconfig, and setup files.

Signed-off-by: Dan Malek <dan@embeddedalley.com>


[-- Attachment #2: geode_x86.patch --]
[-- Type: application/octet-stream, Size: 7286 bytes --]

diff -Naru a/arch/i386/config.in b/arch/i386/config.in
--- a/arch/i386/config.in	2004-11-17 03:54:21.000000000 -0800
+++ b/arch/i386/config.in	2005-01-26 17:03:43.000000000 -0800
@@ -42,6 +42,8 @@
 	 Winchip-C6				CONFIG_MWINCHIPC6 \
 	 Winchip-2				CONFIG_MWINCHIP2 \
 	 Winchip-2A/Winchip-3			CONFIG_MWINCHIP3D \
+	 Geode-GX1/SC1200			CONFIG_GEODE_SC1200 \
+	 Geode-GX2/GX3				CONFIG_GEODE_GX2 \
 	 CyrixIII/VIA-C3			CONFIG_MCYRIXIII \
 	 VIA-C3-2				CONFIG_MVIAC3_2" Pentium-Pro
 #
@@ -139,9 +141,31 @@
    define_bool CONFIG_X86_GOOD_APIC y
    define_bool CONFIG_X86_USE_3DNOW y
    define_bool CONFIG_X86_PGE y
+   define_bool CONFIG_X86_USE_PPRO_CHECKSUM y 	
+   define_bool CONFIG_X86_F00F_WORKS_OK y
+fi
+
+if [ "$CONFIG_GEODE_SC1200" = "y" ]; then
+   define_int  CONFIG_X86_L1_CACHE_SHIFT 5
+   define_bool CONFIG_X86_USE_STRING_486 y
+   define_bool CONFIG_X86_ALIGNMENT_16 y
+   define_bool CONFIG_X86_HAS_TSC y
+   define_bool CONFIG_X86_PPRO_FENCE y
+   define_bool CONFIG_X86_F00F_WORKS_OK y
+fi
+
+if [ "$CONFIG_GEODE_GX2" = "y" ]; then
+   define_int  CONFIG_X86_L1_CACHE_SHIFT 5
+   define_bool CONFIG_X86_HAS_TSC y
+   define_bool CONFIG_MTRR n
+   define_bool CONFIG_X86_LOCAL_APIC n
+   define_bool CONFIG_X86_GOOD_APIC n
+   define_bool CONFIG_X86_PGE y
    define_bool CONFIG_X86_USE_PPRO_CHECKSUM y
    define_bool CONFIG_X86_F00F_WORKS_OK y
+   define_bool CONFIG_X86_MCE n
 fi
+
 if [ "$CONFIG_MELAN" = "y" ]; then
    define_int  CONFIG_X86_L1_CACHE_SHIFT 4
    define_bool CONFIG_X86_USE_STRING_486 y
@@ -192,7 +216,9 @@
    define_bool CONFIG_X86_F00F_WORKS_OK y
 fi
 
+if [ "$CONFIG_GEODE_GX2" != "y" ]; then
 bool 'Machine Check Exception' CONFIG_X86_MCE
+fi
 
 tristate 'Toshiba Laptop support' CONFIG_TOSHIBA
 tristate 'Dell laptop support' CONFIG_I8K
@@ -205,10 +231,19 @@
    tristate 'BIOS Enhanced Disk Drive calls determine boot disk (EXPERIMENTAL)' CONFIG_EDD
 fi
 
-choice 'High Memory Support' \
+# GX2/3 does not have PAE support, so don't give em the option of 64G support
+
+if [ "$CONFIG_GEODE_GX2" != "y" ]; then
+	choice 'High Memory Support' \
 	"off    CONFIG_NOHIGHMEM \
 	 4GB    CONFIG_HIGHMEM4G \
 	 64GB   CONFIG_HIGHMEM64G" off
+else
+	choice 'High Memory Support' \
+	"off	CONFIG_NOHIGHMEM \
+	 4GB	CONFIG_HIGHMEM4G" off
+fi
+
 if [ "$CONFIG_HIGHMEM4G" = "y" -o "$CONFIG_HIGHMEM64G" = "y" ]; then
    define_bool CONFIG_HIGHMEM y
 else
@@ -223,16 +258,22 @@
 fi
 
 bool 'Math emulation' CONFIG_MATH_EMULATION
+
+if [ "$CONFIG_GEODE_GX2" != "y" ]; then
 bool 'MTRR (Memory Type Range Register) support' CONFIG_MTRR
+fi
+
 bool 'Symmetric multi-processing support' CONFIG_SMP
 if [ "$CONFIG_SMP" != "y" ]; then
-   bool 'Local APIC support on uniprocessors' CONFIG_X86_UP_APIC
-   dep_bool 'IO-APIC support on uniprocessors' CONFIG_X86_UP_IOAPIC $CONFIG_X86_UP_APIC
-   if [ "$CONFIG_X86_UP_APIC" = "y" ]; then
-      define_bool CONFIG_X86_LOCAL_APIC y
-   fi
-   if [ "$CONFIG_X86_UP_IOAPIC" = "y" ]; then
-      define_bool CONFIG_X86_IO_APIC y
+   if [ "$CONFIG_GEODE_GX2" != "y" ]; then
+   	bool 'Local APIC support on uniprocessors' CONFIG_X86_UP_APIC
+   	dep_bool 'IO-APIC support on uniprocessors' CONFIG_X86_UP_IOAPIC $CONFIG_X86_UP_APIC
+   	if [ "$CONFIG_X86_UP_APIC" = "y" ]; then
+      	define_bool CONFIG_X86_LOCAL_APIC y
+   	fi
+   	if [ "$CONFIG_X86_UP_IOAPIC" = "y" ]; then
+      	define_bool CONFIG_X86_IO_APIC y
+   	fi 
    fi
 else
    int  'Maximum number of CPUs (2-32)' CONFIG_NR_CPUS 32
diff -Naru a/arch/i386/kernel/setup.c b/arch/i386/kernel/setup.c
--- a/arch/i386/kernel/setup.c	2004-08-07 16:26:04.000000000 -0700
+++ b/arch/i386/kernel/setup.c	2005-01-26 17:03:43.000000000 -0800
@@ -74,6 +74,9 @@
  *  Provisions for empty E820 memory regions (reported by certain BIOSes).
  *  Alex Achenbach <xela@slit.de>, December 2002.
  *
+ *  AMD Geode support by Lycoris Solutions Team and NSC Longmont.
+ *  solutions@lycoris.com, john.zulauf@amd.com.  December 2003.
+ *
  */
 
 /*
@@ -1544,6 +1547,52 @@
 	return r;
 }
 
+static int __init init_amd_geode(struct cpuinfo_x86 *c)
+{
+
+	int r;
+
+	/* Bit 31 in normal CPUID used for nonstandard 3DNow ID;
+	   3DNow is IDd by bit 31 in extended CPUID (1*32+31) anyway */
+	clear_bit(0*32+31, &c->x86_capability);
+	
+	r = get_model_name(c);
+
+	switch(c->x86_model)
+	{
+	        case 4: /* GX1/SCxx00 */
+
+			/* GX1 uses bits 16 and 24 differently - 
+			   you could probably just do 
+
+			   clear_bit(0*32+16, &c->x86_capability);
+			   clear_bit(0*32+24, &c->x86_capability);
+
+			   since I don't think the kernel supports
+			   FPU-CMOV or Cyrix MMX.  Unsure tho.
+
+			   Also checking GX1 cache here needs to be done -
+			   display_cacheinfo() won't work according to
+			   AMD specs.
+			*/
+
+			break;
+
+		case 5: /* GX2 */
+			display_cacheinfo(c);
+			break;
+			
+	        case 10: /* GX3 */
+			/* What else needs to happen here? */
+			display_cacheinfo(c);
+			break;
+			
+
+	}
+
+	return r;
+}
+
 /*
  * Read NSC/Cyrix DEVID registers (DIR) to get more detailed info. about the CPU
  */
@@ -2904,11 +2953,24 @@
 		break;
 
 	case X86_VENDOR_NSC:
-	        init_cyrix(c);
+             /* NSC chips with TFMS of 0x54n or 0x55n are Geode 
+		- do a special init */
+		if ( (c->x86 == 5) &&
+		     (c->x86_model >= 4 && c->x86_model <= 5))
+			init_amd_geode(c);
+		else
+		        init_cyrix(c);
 		break;
 
 	case X86_VENDOR_AMD:
-		init_amd(c);
+		/* Unlike its predecessors, the GX3 flavor of the 
+		   Geode family declares itself as an AuthenticAMD,
+		   so check for that here */
+		
+		if (c->x86 == 5 && c->x86_model == 10)
+			init_amd_geode(c);
+		else
+			init_amd(c);
 		break;
 
 	case X86_VENDOR_CENTAUR:
diff -Naru a/include/asm-i386/msr.h b/include/asm-i386/msr.h
--- a/include/asm-i386/msr.h	2003-11-28 10:26:21.000000000 -0800
+++ b/include/asm-i386/msr.h	2005-01-26 17:03:43.000000000 -0800
@@ -1,6 +1,8 @@
 #ifndef __ASM_MSR_H
 #define __ASM_MSR_H
 
+#include <linux/config.h>
+
 /*
  * Access to machine-specific registers (available on 586 and better only)
  * Note: the rd* operations modify the parameters directly (without using
@@ -17,8 +19,14 @@
 			  : /* no outputs */ \
 			  : "c" (msr), "a" (val1), "d" (val2))
 
+#ifdef CONFIG_GEODE_SC1200
+#define rdtsc(low,high) \
+     __asm__ __volatile__("rdtsc" : "=a" (low), "=d" (high)); \
+    if ((unsigned long) low > 0xFFFFFFFC) high--
+#else
 #define rdtsc(low,high) \
      __asm__ __volatile__("rdtsc" : "=a" (low), "=d" (high))
+#endif
 
 #define rdtscl(low) \
      __asm__ __volatile__("rdtsc" : "=a" (low) : : "edx")
diff -Naru a/arch/i386/defconfig b/arch/i386/defconfig
--- a/arch/i386/defconfig	2005-01-19 06:09:25.000000000 -0800
+++ b/arch/i386/defconfig	2005-01-26 17:59:24.000000000 -0800
@@ -36,6 +36,8 @@
 # CONFIG_MWINCHIPC6 is not set
 # CONFIG_MWINCHIP2 is not set
 # CONFIG_MWINCHIP3D is not set
+# CONFIG_GEODE_SC1200 is not set
+# CONFIG_GEODE_GX2 is not set
 # CONFIG_MCYRIXIII is not set
 # CONFIG_MVIAC3_2 is not set
 CONFIG_X86_WP_WORKS_OK=y
@@ -274,6 +276,7 @@
 # CONFIG_BLK_DEV_TRIFLEX is not set
 # CONFIG_BLK_DEV_CY82C693 is not set
 # CONFIG_BLK_DEV_CS5530 is not set
+# CONFIG_BLK_DEV_CS5535 is not set
 # CONFIG_BLK_DEV_HPT34X is not set
 # CONFIG_HPT34X_AUTODMA is not set
 # CONFIG_BLK_DEV_HPT366 is not set

Re: [PATCH] add AMD Geode processor support

[Pavel Machek]

Hi!

> Here is a patch for 2.4 that adds the basic AMD Geode GX2/GX3
> and GX1/SC1200 support.  This patch updates configuration
> scripts, defconfig, and setup files.

Please inline patches...

We do not disable HIGHMEM_64GB for 486, I do not see why we should add
extra code to geode...

If you really want to do that, do it properly for all cpus...
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!

Re: [PATCH] add AMD Geode processor support

[Dan Malek]

On Jan 27, 2005, at 9:15 AM, Pavel Machek wrote:

> We do not disable HIGHMEM_64GB for 486, I do not see why we should add
> extra code to geode...

What about some of the other ones like MTRR and IOAPIC?
I was kinda passing this along from someone I thought knew
better than I, but I didn't like it either.  It seems just setting these
booleans to 'n' should do the trick.

Thanks.	

	-- Dan

Re: [PATCH] add AMD Geode processor support

[Pavel Machek]

Hi!

> >We do not disable HIGHMEM_64GB for 486, I do not see why we should add
> >extra code to geode...
> 
> What about some of the other ones like MTRR and IOAPIC?
> I was kinda passing this along from someone I thought knew
> better than I, but I didn't like it either.  It seems just setting these
> booleans to 'n' should do the trick.

Actually sorry, no, you can't do that.

486 should mean "can run on 486 or better" (at least in 2.4.X
series). That means it should have IOAPIC support compiled in.

Just do not play with HIGHMEM_64GB; someone may want a kernel for
"geode or better".
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!

Re: [PATCH] add AMD Geode processor support

[Alan Cox]

On Iau, 2005-01-27 at 06:54, Dan Malek wrote:
> Hi Marcelo!
> 
> Here is a patch for 2.4 that adds the basic AMD Geode GX2/GX3
> and GX1/SC1200 support.  This patch updates configuration
> scripts, defconfig, and setup files.

GX2 seems no different to the older Cyrix/NS settings. GX3 a little
different.
Also you might not want to magically force settings like highmem because
you want that for multi-target kernels - Geode is a sort of odd case
where it almost makes sense but its different enough to make me dubious.

Re: [PATCH] add AMD Geode processor support

[Dan Malek]

On Jan 31, 2005, at 11:01 AM, Alan Cox wrote:

> Also you might not want to magically force settings like highmem 
> because
> you want that for multi-target kernels - Geode is a sort of odd case
> where it almost makes sense but its different enough to make me 
> dubious.

I've already taken that out.

In fact, I've decided to wait until I get the 2.6 done first, then go 
back
and properly update for 2.4.  This was just a patch that has been around
for a while, I just wanted to get it off of my plate so I could do a 
minor
update later, but I'll just wait.

Apologies for the wasted bandwidth.

Thanks.

	-- Dan