* lots of bugfixes in the assembler code
	* reverted hardware.h back to original
	* enabled hardware DRAM calibration
	* GCC-4 fix: modified GLOBAL_DATA_POINTER macro
This commit is contained in:
Markus Klotzbcher 2006-02-08 18:56:28 +01:00 committed by Markus Klotzbcher
parent af646e865f
commit 9d803d8c0b
3 changed files with 199 additions and 116 deletions

View file

@ -39,6 +39,16 @@ DRAM_SIZE: .long CFG_DRAM_SIZE
.endm
.macro wait time
ldr r2, =OSCR
mov r3, #0
str r3, [r2]
0:
ldr r3, [r2]
cmp r3, \time
bls 0b
.endm
/*
* Memory setup
*/
@ -48,7 +58,7 @@ lowlevel_init:
/* Set up GPIO pins first ----------------------------------------- */
mov r10, lr
/* GPIO41, 42, 43, 44, 45, 46, 47, 48 */
/* Configure GPIO Pins 41 - 48 as UART1 / altern. Fkt. 2 */
ldr r0, =0x40E10438 @ GPIO41 FFRXD
ldr r1, =0x802
str r1, [r0]
@ -82,7 +92,7 @@ lowlevel_init:
str r1, [r0]
/* tebrandt - ASCR, clear the RDH bit */
ldr r0, =ASCR
ldr r0, =ASCR
ldr r1, [r0]
bic r1, r1, #0x80000000
str r1, [r0]
@ -101,16 +111,18 @@ lowlevel_init:
/* FIXME: can be optimized later */
/* ---------------------------------------------------------------- */
ldr r3, =OSCR /* reset the OS Timer Count to zero */
mov r2, #0
str r2, [r3]
ldr r4, =0x300 /* really 0x2E1 is about 200usec, */
/* so 0x300 should be plenty */
1:
ldr r2, [r3]
cmp r4, r2
bgt 1b
/* mk: replaced with wait macro */
/* ldr r3, =OSCR /\* reset the OS Timer Count to zero *\/ */
/* mov r2, #0 */
/* str r2, [r3] */
/* ldr r4, =0x300 /\* really 0x2E1 is about 200usec, *\/ */
/* /\* so 0x300 should be plenty *\/ */
/* 1: */
/* ldr r2, [r3] */
/* cmp r4, r2 */
/* bgt 1b */
wait #300
mem_init:
/* configure the MEMCLKCFG register */
@ -209,8 +221,12 @@ mem_init:
str r2, [r1]
ldr r2, [r1]
/* DDR Read-Strobe Delay Calibration */
/* bl ddr_calibration */
/* Hardware DDR Read-Strobe Delay Calibration */
ldr r0, =DDR_HCAL @ DDR_HCAL
ldr r1, =0x803ffc07 @ the offset is correct? -SC
str r1, [r0]
wait #5
ldr r1, [r0]
/* Here we assume the hardware calibration alwasy be successful. -SC */
/* Set DMCEN bit in MDCNFG Register */
@ -220,10 +236,12 @@ mem_init:
str r1, [r0]
/* scrub/init SDRAM if enabled/present */
ldr r11, =0xa0000000 //RAM_BASE // base address of SDRAM
ldr r12, =0x04000000 // size of memory to scrub
mov r8,r12 // save DRAM size
mov r0, #0 // scrub with 0x0000:0000
/* ldr r11, =0xa0000000 /\* base address of SDRAM (CFG_DRAM_BASE) *\/ */
/* ldr r12, =0x04000000 /\* size of memory to scrub (CFG_DRAM_SIZE) *\/ */
/* mov r8,r12 /\* save DRAM size (mk: why???) *\/ */
ldr r8, =0xa0000000 /* base address of SDRAM (CFG_DRAM_BASE) */
ldr r9, =0x04000000 /* size of memory to scrub (CFG_DRAM_SIZE) */
mov r0, #0 /* scrub with 0x0000:0000 */
mov r1, #0
mov r2, #0
mov r3, #0
@ -232,8 +250,8 @@ mem_init:
mov r6, #0
mov r7, #0
10: /* fastScrubLoop */
subs r12, r12, #32 // 32 bytes/line
stmia r11!, {r0-r7}
subs r9, r9, #32 // 32 bytes/line
stmia r8!, {r0-r7}
beq 15f
b 10b
@ -264,3 +282,94 @@ mem_init:
endlowlevel_init:
mov pc, lr
/*
@********************************************************************************
@ DDR calibration
@
@ This function is used to calibrate DQS delay lines.
@ Monahans supports three ways to do it. One is software
@ calibration. Two is hardware calibration. Three is hybrid
@ calibration.
@
@ TBD
@ -SC
ddr_calibration:
@ Case 1: Write the correct delay value once
@ Configure DDR_SCAL Register
ldr r0, =DDR_SCAL @ DDR_SCAL
q ldr r1, =0xaf2f2f2f
str r1, [r0]
ldr r1, [r0]
*/
/* @ Case 2: Software Calibration
@ Write test pattern to memory
ldr r5, =0x0faf0faf @ Data Pattern
ldr r4, =0xa0000000 @ DDR ram
str r5, [r4]
mov r1, =0x0 @ delay count
mov r6, =0x0
mov r7, =0x0
ddr_loop1:
add r1, r1, =0x1
cmp r1, =0xf
ble end_loop
mov r3, r1
mov r0, r1, lsl #30
orr r3, r3, r0
mov r0, r1, lsl #22
orr r3, r3, r0
mov r0, r1, lsl #14
orr r3, r3, r0
orr r3, r3, =0x80000000
ldr r2, =DDR_SCAL
str r3, [r2]
ldr r2, [r4]
cmp r2, r5
bne ddr_loop1
mov r6, r1
ddr_loop2:
add r1, r1, =0x1
cmp r1, =0xf
ble end_loop
mov r3, r1
mov r0, r1, lsl #30
orr r3, r3, r0
mov r0, r1, lsl #22
orr r3, r3, r0
mov r0, r1, lsl #14
orr r3, r3, r0
orr r3, r3, =0x80000000
ldr r2, =DDR_SCAL
str r3, [r2]
ldr r2, [r4]
cmp r2, r5
be ddr_loop2
mov r7, r2
add r3, r6, r7
lsr r3, r3, =0x1
mov r0, r1, lsl #30
orr r3, r3, r0
mov r0, r1, lsl #22
orr r3, r3, r0
mov r0, r1, lsl #14
orr r3, r3, r0
orr r3, r3, =0x80000000
ldr r2, =DDR_SCAL
end_loop:
@ Case 3: Hardware Calibratoin
ldr r0, =DDR_HCAL @ DDR_HCAL
ldr r1, =0x803ffc07 @ the offset is correct? -SC
str r1, [r0]
wait #5
ldr r1, [r0]
mov pc, lr
*/

View file

@ -8,6 +8,11 @@
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* Note: This file was taken from linux-2.4.19-rmk4-pxa1
*
* - 2003/01/20 implementation specifics activated
* Robert Schwebel <r.schwebel@pengutronix.de>
*/
#ifndef __ASM_ARCH_HARDWARE_H
@ -16,6 +21,16 @@
#include <linux/config.h>
#include <asm/mach-types.h>
/*
* These are statically mapped PCMCIA IO space for designs using it as a
* generic IO bus, typically with ISA parts, hardwired IDE interfaces, etc.
* The actual PCMCIA code is mapping required IO region at run time.
*/
#define PCMCIA_IO_0_BASE 0xf6000000
#define PCMCIA_IO_1_BASE 0xf7000000
/*
* We requires absolute addresses.
*/
@ -29,63 +44,22 @@
#define UNCACHED_ADDR UNCACHED_PHYS_0
/*
* Intel PXA2xx internal register mapping:
* Intel PXA internal I/O mappings:
*
* 0x40000000 - 0x41ffffff <--> 0xf2000000 - 0xf3ffffff
* 0x44000000 - 0x45ffffff <--> 0xf4000000 - 0xf5ffffff
* 0x48000000 - 0x49ffffff <--> 0xf6000000 - 0xf7ffffff
* 0x4c000000 - 0x4dffffff <--> 0xf8000000 - 0xf9ffffff
* 0x50000000 - 0x51ffffff <--> 0xfa000000 - 0xfbffffff
* 0x54000000 - 0x55ffffff <--> 0xfc000000 - 0xfdffffff
* 0x58000000 - 0x59ffffff <--> 0xfe000000 - 0xffffffff
*
* Note that not all PXA2xx chips implement all those addresses, and the
* kernel only maps the minimum needed range of this mapping.
*/
#ifndef CONFIG_CPU_MONAHANS
#define io_p2v(x) (0xf2000000 + ((x) & 0x01ffffff) + (((x) & 0x1c000000) >> 1))
#define io_v2p(x) (0x3c000000 + ((x) & 0x01ffffff) + (((x) & 0x0e000000) << 1))
#else
/* There are too many IO area needed to map, so I divide them into 3 areas
* 0x40000000 - 0x41ffffff <--> 0xf6000000 - 0xf7ffffff Devs
*/
#define io_p2v(x) ((((x) & 0xfc000000)>>4) + 0xf2000000 + ((x)&0x01ffffff))
#define io_v2p(x) (((((x) - 0xf2000000)&0xfc000000)<<4) + ((x)&0x01ffffff))
/*
* 0x42000000 - 0x421fffff <--> 0xf8000000 - 0xf81fffff MMC2 & USIM2
* 0x43000000 - 0x430fffff <--> 0xf8200000 - 0xf82fffff Caddo
* 0x43100000 - 0x431fffff <--> 0xf8300000 - 0xf83fffff NAND
* 0x44000000 - 0x440fffff <--> 0xf8400000 - 0xf84fffff LCD
* 0x46000000 - 0x460fffff <--> 0xf8800000 - 0xf88fffff Mini LCD
* 0x48100000 - 0x481fffff <--> 0xf8d00000 - 0xf8dfffff Dynamic Mem Ctl
* 0x4a000000 - 0x4a0fffff <--> 0xf9000000 - 0xf90fffff Static Mem Ctl
* 0x4c000000 - 0x4c0fffff <--> 0xf9400000 - 0xf94fffff USB Host
* 0x40000000 - 0x41ffffff <--> 0xf8000000 - 0xf9ffffff
* 0x44000000 - 0x45ffffff <--> 0xfa000000 - 0xfbffffff
* 0x48000000 - 0x49ffffff <--> 0xfc000000 - 0xfdffffff
*/
#define io_p2v_2(x) (((((x) - 0x42000000) & 0xff000000) >> 3) + 0xf8000000\
+ ((x) & 0x001fffff))
#define io_v2p_2(x) (((((x) & 0xffe00000) - 0xf8000000) << 3) + 0x42000000\
+ (x & 0x001fffff))
/*
* 0x50000000 - 0x500fffff <--> 0xfa000000 - 0xfa0fffff Camera Interface
* 0x54000000 - 0x540fffff <--> 0xfa400000 - 0xfa4fffff 2D Graphics Ctrl
* 0x54100000 - 0x541fffff <--> 0xfa500000 - 0xfa5fffff USB Device 2.0 Ctrl
* 0x58000000 - 0x580fffff <--> 0xfa800000 - 0xfa8fffff Internal SRAM Ctrl
*/
#define io_p2v_3(x) ((((x) & 0xfc000000) >> 4) + 0xf5000000 + \
((x) & 0x001fffff))
#define io_v2p_3(x) (((((x) - 0xf5000000) & 0x0fc00000) << 4) + \
((x) & 0x001fffff))
#endif /* CONFIG_CPU_MONAHANS */
/* FIXME: Only this does work for u-boot... find out why... [RS] */
#define UBOOT_REG_FIX 1
#ifndef UBOOT_REG_FIX
#ifndef __ASSEMBLY__
#if 0
# define __REG(x) (*((volatile u32 *)io_p2v(x)))
#else
#define io_p2v(x) ( ((x) | 0xbe000000) ^ (~((x) >> 1) & 0x06000000) )
#define io_v2p( x ) ( ((x) & 0x41ffffff) ^ ( ((x) & 0x06000000) << 1) )
/*
* This __REG() version gives the same results as the one above, except
* that we are fooling gcc somehow so it generates far better and smaller
@ -96,56 +70,66 @@
typedef struct { volatile u32 offset[4096]; } __regbase;
# define __REGP(x) ((__regbase *)((x)&~4095))->offset[((x)&4095)>>2]
# define __REG(x) __REGP(io_p2v(x))
#endif
/* __REG_2 is for NAND, LCD etc.
* __REG_3 is for Camera Interface, 2D Graphics, U2D etc.*/
#ifdef CONFIG_CPU_MONAHANS
#define __REG_2(x) __REGP(io_p2v_2(x))
#define __REG_3(x) __REGP(io_p2v_3(x))
#endif /* CONFIG_CPU_MONAHANS */
#endif /* if 0 */
/* With indexed regs we don't want to feed the index through io_p2v()
especially if it is a variable, otherwise horrible code will result. */
# define __REG2(x,y) (*(volatile u32 *)((u32)&__REG(x) + (y)))
/* Let's kick gcc's ass again... */
# define __REG2(x,y) \
( __builtin_constant_p(y) ? (__REG((x) + (y))) \
: (*(volatile u32 *)((u32)&__REG(x) + (y))) )
# define __PREG(x) (io_v2p((u32)&(x)))
#else /* ifndef __ASSEMBLY__ */
#else
# define __REG(x) io_p2v(x)
# define __PREG(x) io_v2p(x)
#ifdef CONFIG_CPU_MONAHANS
# define __REG_2(x) io_p2v(x)
# define __REG_3(x) io_p2v(x)
#endif /* CONFIG_CPU_MONAHANS */
# undef io_p2v
# undef __REG
# ifndef __ASSEMBLY__
# define io_p2v(PhAdd) (PhAdd)
# define __REG(x) (*((volatile u32 *)io_p2v(x)))
# define __REG2(x,y) (*(volatile u32 *)((u32)&__REG(x) + (y)))
# else
# define __REG(x) (x)
# ifdef CONFIG_CPU_MONAHANS /* Hack to make this work with mona's pxa-regs.h */
# define __REG_2(x) (x)
# define __REG_3(x) (x)
# endif
# endif
#endif /* UBOOT_REG_FIX */
#endif /* ifndef __ASSEMBLY__ */
#include "pxa-regs.h"
#ifndef __ASSEMBLY__
#ifdef CONFIG_MACH_ZYLONITE
#include "zylonite.h"
#endif
/*
* GPIO edge detection for IRQs:
* IRQs are generated on Falling-Edge, Rising-Edge, or both.
* This must be called *before* the corresponding IRQ is registered.
* Use this instead of directly setting GRER/GFER.
*/
#define GPIO_FALLING_EDGE 1
#define GPIO_RISING_EDGE 2
#define GPIO_BOTH_EDGES 3
extern void set_GPIO_IRQ_edge( int gpio_nr, int edge_mask );
/*
* Handy routine to set GPIO alternate functions
*/
extern void pxa_gpio_mode( int gpio_mode );
extern void set_GPIO_mode( int gpio_mode );
/*
* Routine to enable or disable CKEN
* return current lclk frequency in units of 10kHz
*/
extern void pxa_set_cken(int clock, int enable);
extern unsigned int get_lclk_frequency_10khz(void);
#endif
/*
* return current memory and LCD clock frequency in units of 10kHz
* Implementation specifics
*/
extern unsigned int get_memclk_frequency_10khz(void);
extern unsigned int get_lcdclk_frequency_10khz(void);
#endif /* __ASSEMBLY__ */
#ifdef CONFIG_ARCH_LUBBOCK
#include "lubbock.h"
@ -159,15 +143,6 @@ extern unsigned int get_lcdclk_frequency_10khz(void);
#include "cerf.h"
#endif
#if CONFIG_CPU_MONAHANS_L2CACHE
#define __cpuc_flush_l2cache_all xscale_flush_l2cache_all
extern void __cpuc_flush_l2cache_all(void);
#define flush_l2cache_all __cpuc_flush_l2cache_all
#else
#define __cpuc_flush_l2cache_all() do {} while (0)
#define flush_l2cache_all() do {} while (0)
#endif
#ifdef CONFIG_ARCH_CSB226
#include "csb226.h"
#endif
@ -180,10 +155,4 @@ extern void __cpuc_flush_l2cache_all(void);
#include "pleb.h"
#endif
#ifdef CONFIG_MACH_MAINSTONE
#include "mainstone.h"
#endif
#include "pxa-regs.h"
#endif /* _ASM_ARCH_HARDWARE_H */
#endif /* _ASM_ARCH_HARDWARE_H */

View file

@ -61,6 +61,11 @@ typedef struct global_data {
#define GD_FLG_DEVINIT 0x00002 /* Devices have been initialized */
#define GD_FLG_SILENT 0x00004 /* Silent mode */
#define DECLARE_GLOBAL_DATA_PTR register volatile gd_t *gd asm ("r8")
#define GCC_4_SCREW_GDP 1
#ifdef GCC_4_SCREW_GDP
# define DECLARE_GLOBAL_DATA_PTR register gd_t* volatile gd asm ("r8");
#else
# define DECLARE_GLOBAL_DATA_PTR register volatile gd_t *gd asm ("r8")
#endif
#endif /* __ASM_GBL_DATA_H */