[ARM] 3377/2: add support for intel xsc3 core

Patch from Lennert Buytenhek

This patch adds support for the new XScale v3 core.  This is an
ARMv5 ISA core with the following additions:

- L2 cache
- I/O coherency support (on select chipsets)
- Low-Locality Reference cache attributes (replaces mini-cache)
- Supersections (v6 compatible)
- 36-bit addressing (v6 compatible)
- Single instruction cache line clean/invalidate
- LRU cache replacement (vs round-robin)

I attempted to merge the XSC3 support into proc-xscale.S, but XSC3
cores have separate errata and have to handle things like L2, so it
is simpler to keep it separate.

L2 cache support is currently a build option because the L2 enable
bit must be set before we enable the MMU and there is no easy way to
capture command line parameters at this point.

There are still optimizations that can be done such as using LLR for
copypage (in theory using the exisiting mini-cache code) but those
can be addressed down the road.

Signed-off-by: Deepak Saxena <dsaxena@plexity.net>
Signed-off-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

1 files changed, 498 insertions, 0 deletions

diff --git a/arch/arm/mm/proc-xsc3.S b/arch/arm/mm/proc-xsc3.S
new file mode 100644
index 0000000..f90513e
--- /dev/null
+++ b/arch/arm/mm/proc-xsc3.S
@@ -0,0 +1,498 @@
+/*
+ * linux/arch/arm/mm/proc-xsc3.S
+ *
+ * Original Author: Matthew Gilbert
+ * Current Maintainer: Deepak Saxena <dsaxena@plexity.net>
+ *
+ * Copyright 2004 (C) Intel Corp.
+ * Copyright 2005 (c) MontaVista Software, Inc.
+ *
+ * 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.
+ *
+ * MMU functions for the Intel XScale3 Core (XSC3).  The XSC3 core is an
+ * extension to Intel's original XScale core that adds the following
+ * features:
+ *
+ * - ARMv6 Supersections
+ * - Low Locality Reference pages (replaces mini-cache)
+ * - 36-bit addressing
+ * - L2 cache
+ * - Cache-coherency if chipset supports it
+ *
+ * Based on orignal XScale code by Nicolas Pitre
+ */
+
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/assembler.h>
+#include <asm/procinfo.h>
+#include <asm/hardware.h>
+#include <asm/pgtable.h>
+#include <asm/page.h>
+#include <asm/ptrace.h>
+#include "proc-macros.S"
+
+/*
+ * This is the maximum size of an area which will be flushed.  If the
+ * area is larger than this, then we flush the whole cache.
+ */
+#define MAX_AREA_SIZE	32768
+
+/*
+ * The cache line size of the I and D cache.
+ */
+#define CACHELINESIZE	32
+
+/*
+ * The size of the data cache.
+ */
+#define CACHESIZE	32768
+
+/*
+ * Run with L2 enabled.
+ */
+#define L2_CACHE_ENABLE	1
+
+/*
+ * Enable the Branch Target Buffer (can cause crashes, see erratum #42.)
+ */
+#define BTB_ENABLE	0
+
+/*
+ * This macro is used to wait for a CP15 write and is needed
+ * when we have to ensure that the last operation to the co-pro
+ * was completed before continuing with operation.
+ */
+	.macro	cpwait_ret, lr, rd
+	mrc	p15, 0, \rd, c2, c0, 0		@ arbitrary read of cp15
+	sub	pc, \lr, \rd, LSR #32		@ wait for completion and
+						@ flush instruction pipeline
+	.endm
+
+/*
+ * This macro cleans & invalidates the entire xsc3 dcache by set & way.
+ */
+
+ 	.macro  clean_d_cache rd, rs
+	mov	\rd, #0x1f00
+	orr	\rd, \rd, #0x00e0
+1:	mcr	p15, 0, \rd, c7, c14, 2		@ clean/inv set/way
+	adds	\rd, \rd, #0x40000000
+	bcc	1b
+	subs	\rd, \rd, #0x20
+	bpl	1b
+	.endm
+
+	.text
+
+/*
+ * cpu_xsc3_proc_init()
+ *
+ * Nothing too exciting at the moment
+ */
+ENTRY(cpu_xsc3_proc_init)
+	mov	pc, lr
+
+/*
+ * cpu_xsc3_proc_fin()
+ */
+ENTRY(cpu_xsc3_proc_fin)
+	str	lr, [sp, #-4]!
+	mov	r0, #PSR_F_BIT|PSR_I_BIT|SVC_MODE
+	msr	cpsr_c, r0
+	bl	xsc3_flush_kern_cache_all	@ clean caches
+	mrc	p15, 0, r0, c1, c0, 0		@ ctrl register
+	bic	r0, r0, #0x1800			@ ...IZ...........
+	bic	r0, r0, #0x0006			@ .............CA.
+	mcr	p15, 0, r0, c1, c0, 0		@ disable caches
+	ldr	pc, [sp], #4
+
+/*
+ * cpu_xsc3_reset(loc)
+ *
+ * Perform a soft reset of the system.  Put the CPU into the
+ * same state as it would be if it had been reset, and branch
+ * to what would be the reset vector.
+ *
+ * loc: location to jump to for soft reset
+ */
+	.align	5
+ENTRY(cpu_xsc3_reset)
+	mov	r1, #PSR_F_BIT|PSR_I_BIT|SVC_MODE
+	msr	cpsr_c, r1			@ reset CPSR
+	mrc	p15, 0, r1, c1, c0, 0		@ ctrl register
+	bic	r1, r1, #0x0086			@ ........B....CA.
+	bic	r1, r1, #0x3900			@ ..VIZ..S........
+	mcr	p15, 0, r1, c1, c0, 0		@ ctrl register
+	mcr	p15, 0, ip, c7, c7, 0		@ invalidate I,D caches & BTB
+	bic	r1, r1, #0x0001			@ ...............M
+	mcr	p15, 0, r1, c1, c0, 0		@ ctrl register
+	@ CAUTION: MMU turned off from this point.  We count on the pipeline
+	@ already containing those two last instructions to survive.
+	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I & D TLBs
+	mov	pc, r0
+
+/*
+ * cpu_xsc3_do_idle()
+ *
+ * Cause the processor to idle
+ *
+ * For now we do nothing but go to idle mode for every case
+ *
+ * XScale supports clock switching, but using idle mode support
+ * allows external hardware to react to system state changes.
+
+ MMG: Come back to this one.
+ */
+	.align	5
+
+ENTRY(cpu_xsc3_do_idle)
+	mov	r0, #1
+	mcr	p14, 0, r0, c7, c0, 0		@ Go to IDLE
+	mov	pc, lr
+
+/* ================================= CACHE ================================ */
+
+/*
+ *	flush_user_cache_all()
+ *
+ *	Invalidate all cache entries in a particular address
+ *	space.
+ */
+ENTRY(xsc3_flush_user_cache_all)
+	/* FALLTHROUGH */
+
+/*
+ *	flush_kern_cache_all()
+ *
+ *	Clean and invalidate the entire cache.
+ */
+ENTRY(xsc3_flush_kern_cache_all)
+	mov	r2, #VM_EXEC
+	mov	ip, #0
+__flush_whole_cache:
+	clean_d_cache r0, r1
+	tst	r2, #VM_EXEC
+	mcrne	p15, 0, ip, c7, c5, 0		@ Invalidate I cache & BTB
+	mcrne	p15, 0, ip, c7, c10, 4		@ Drain Write Buffer
+	mcrne	p15, 0, ip, c7, c5, 4		@ Prefetch Flush
+	mov	pc, lr
+
+/*
+ *	flush_user_cache_range(start, end, vm_flags)
+ *
+ *	Invalidate a range of cache entries in the specified
+ *	address space.
+ *
+ *	- start - start address (may not be aligned)
+ *	- end	- end address (exclusive, may not be aligned)
+ *	- vma	- vma_area_struct describing address space
+ */
+	.align	5
+ENTRY(xsc3_flush_user_cache_range)
+	mov	ip, #0
+	sub	r3, r1, r0			@ calculate total size
+	cmp	r3, #MAX_AREA_SIZE
+	bhs	__flush_whole_cache
+
+1:	tst	r2, #VM_EXEC
+	mcrne	p15, 0, r0, c7, c5, 1		@ Invalidate I cache line
+	mcr	p15, 0, r0, c7, c14, 1		@ Clean/invalidate D cache line
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	tst	r2, #VM_EXEC
+	mcrne	p15, 0, ip, c7, c5, 6		@ Invalidate BTB
+	mcrne	p15, 0, ip, c7, c10, 4		@ Drain Write Buffer
+	mcrne	p15, 0, ip, c7, c5, 4		@ Prefetch Flush
+	mov	pc, lr
+
+/*
+ *	coherent_kern_range(start, end)
+ *
+ *	Ensure coherency between the Icache and the Dcache in the
+ *	region described by start.  If you have non-snooping
+ *	Harvard caches, you need to implement this function.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ *
+ *	Note: single I-cache line invalidation isn't used here since
+ *	it also trashes the mini I-cache used by JTAG debuggers.
+ */
+ENTRY(xsc3_coherent_kern_range)
+/* FALLTHROUGH */
+ENTRY(xsc3_coherent_user_range)
+	bic	r0, r0, #CACHELINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c5, 0		@ Invalidate I cache & BTB
+	mcr	p15, 0, r0, c7, c10, 4		@ Drain Write Buffer
+	mcr	p15, 0, r0, c7, c5, 4		@ Prefetch Flush
+	mov	pc, lr
+
+/*
+ *	flush_kern_dcache_page(void *page)
+ *
+ *	Ensure no D cache aliasing occurs, either with itself or
+ *	the I cache
+ *
+ *	- addr	- page aligned address
+ */
+ENTRY(xsc3_flush_kern_dcache_page)
+	add	r1, r0, #PAGE_SZ
+1:	mcr	p15, 0, r0, c7, c14, 1		@ Clean/Invalidate D Cache line
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c5, 0		@ Invalidate I cache & BTB
+	mcr	p15, 0, r0, c7, c10, 4		@ Drain Write Buffer
+	mcr	p15, 0, r0, c7, c5, 4		@ Prefetch Flush
+	mov	pc, lr
+
+/*
+ *	dma_inv_range(start, end)
+ *
+ *	Invalidate (discard) the specified virtual address range.
+ *	May not write back any entries.  If 'start' or 'end'
+ *	are not cache line aligned, those lines must be written
+ *	back.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(xsc3_dma_inv_range)
+	tst	r0, #CACHELINESIZE - 1
+	bic	r0, r0, #CACHELINESIZE - 1
+	mcrne	p15, 0, r0, c7, c10, 1		@ clean L1 D entry
+	mcrne	p15, 1, r0, c7, c11, 1		@ clean L2 D entry
+	tst	r1, #CACHELINESIZE - 1
+	mcrne	p15, 0, r1, c7, c10, 1		@ clean L1 D entry
+	mcrne	p15, 1, r1, c7, c11, 1		@ clean L2 D entry
+1:	mcr	p15, 0, r0, c7, c6, 1		@ invalidate L1 D entry
+	mcr	p15, 1, r0, c7, c7, 1		@ Invalidate L2 D cache line
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	mcr	p15, 0, r0, c7, c10, 4		@ Drain Write Buffer
+	mov	pc, lr
+
+/*
+ *	dma_clean_range(start, end)
+ *
+ *	Clean the specified virtual address range.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(xsc3_dma_clean_range)
+	bic	r0, r0, #CACHELINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c10, 1		@ clean L1 D entry
+	mcr	p15, 1, r0, c7, c11, 1		@ clean L2 D entry
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	mcr	p15, 0, r0, c7, c10, 4		@ Drain Write Buffer
+	mov	pc, lr
+
+/*
+ *	dma_flush_range(start, end)
+ *
+ *	Clean and invalidate the specified virtual address range.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(xsc3_dma_flush_range)
+	bic	r0, r0, #CACHELINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c14, 1	@ Clean/invalidate L1 D cache line
+	mcr	p15, 1, r0, c7, c11, 1	@ Clean L2 D cache line
+	mcr	p15, 1, r0, c7, c7, 1	@ Invalidate L2 D cache line
+	add	r0, r0, #CACHELINESIZE
+	cmp	r0, r1
+	blo	1b
+	mcr	p15, 0, r0, c7, c10, 4		@ Drain Write Buffer
+	mov	pc, lr
+
+ENTRY(xsc3_cache_fns)
+	.long	xsc3_flush_kern_cache_all
+	.long	xsc3_flush_user_cache_all
+	.long	xsc3_flush_user_cache_range
+	.long	xsc3_coherent_kern_range
+	.long	xsc3_coherent_user_range
+	.long	xsc3_flush_kern_dcache_page
+	.long	xsc3_dma_inv_range
+	.long	xsc3_dma_clean_range
+	.long	xsc3_dma_flush_range
+
+ENTRY(cpu_xsc3_dcache_clean_area)
+1:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
+	add	r0, r0, #CACHELINESIZE
+	subs	r1, r1, #CACHELINESIZE
+	bhi	1b
+	mov	pc, lr
+
+/* =============================== PageTable ============================== */
+
+/*
+ * cpu_xsc3_switch_mm(pgd)
+ *
+ * Set the translation base pointer to be as described by pgd.
+ *
+ * pgd: new page tables
+ */
+	.align	5
+ENTRY(cpu_xsc3_switch_mm)
+	clean_d_cache r1, r2
+	mcr	p15, 0, ip, c7, c5, 0		@ Invalidate I cache & BTB
+	mcr	p15, 0, ip, c7, c10, 4		@ Drain Write Buffer
+	mcr	p15, 0, ip, c7, c5, 4		@ Prefetch Flush
+#ifdef L2_CACHE_ENABLE
+	orr	r0, r0, #0x18			@ cache the page table in L2
+#endif
+	mcr	p15, 0, r0, c2, c0, 0		@ load page table pointer
+	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I & D TLBs
+	cpwait_ret lr, ip
+
+/*
+ * cpu_xsc3_set_pte(ptep, pte)
+ *
+ * Set a PTE and flush it out
+ *
+ */
+	.align	5
+ENTRY(cpu_xsc3_set_pte)
+	str	r1, [r0], #-2048		@ linux version
+
+	bic	r2, r1, #0xff0
+	orr	r2, r2, #PTE_TYPE_EXT		@ extended page
+
+	eor	r3, r1, #L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_WRITE | L_PTE_DIRTY
+
+	tst	r3, #L_PTE_USER			@ User?
+	orrne	r2, r2, #PTE_EXT_AP_URO_SRW	@ yes -> user r/o, system r/w
+
+	tst	r3, #L_PTE_WRITE | L_PTE_DIRTY	@ Write and Dirty?
+	orreq	r2, r2, #PTE_EXT_AP_UNO_SRW	@ yes -> user n/a, system r/w
+						@ combined with user -> user r/w
+
+#if L2_CACHE_ENABLE
+	@ If its cacheable it needs to be in L2 also.
+	eor	ip, r1, #L_PTE_CACHEABLE
+	tst	ip, #L_PTE_CACHEABLE
+	orreq	r2, r2, #PTE_EXT_TEX(0x5)
+#endif
+
+	tst	r3, #L_PTE_PRESENT | L_PTE_YOUNG	@ Present and Young?
+	movne	r2, #0				@ no -> fault
+
+	str	r2, [r0]			@ hardware version
+	mov	ip, #0
+	mcr	p15, 0, r0, c7, c10, 1		@ Clean D cache line mcr
+	mcr	p15, 0, ip, c7, c10, 4		@ Drain Write Buffer
+	mov	pc, lr
+
+	.ltorg
+
+	.align
+
+	__INIT
+
+	.type	__xsc3_setup, #function
+__xsc3_setup:
+	mov	r0, #PSR_F_BIT|PSR_I_BIT|SVC_MODE
+	msr	cpsr_c, r0
+	mcr	p15, 0, ip, c7, c7, 0		@ invalidate I, D caches & BTB
+	mcr	p15, 0, ip, c7, c10, 4		@ Drain Write Buffer
+	mcr	p15, 0, ip, c7, c5, 4		@ Prefetch Flush
+	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I, D TLBs
+#if L2_CACHE_ENABLE
+	orr	r4, r4, #0x18			@ cache the page table in L2
+#endif
+	mcr	p15, 0, r4, c2, c0, 0		@ load page table pointer
+	mov	r0, #1				@ Allow access to CP0 and CP13
+	orr	r0, r0, #1 << 13		@ Its undefined whether this
+	mcr	p15, 0, r0, c15, c1, 0		@ affects USR or SVC modes
+	mrc	p15, 0, r0, c1, c0, 1		@ get auxiliary control reg
+	and	r0, r0, #2			@ preserve bit P bit setting
+#if L2_CACHE_ENABLE
+	orr	r0, r0, #(1 << 10)		@ enable L2 for LLR cache
+#endif
+	mcr	p15, 0, r0, c1, c0, 1		@ set auxiliary control reg
+	mrc	p15, 0, r0, c1, c0, 0		@ get control register
+	bic	r0, r0, #0x0200			@ .... ..R. .... ....
+	bic	r0, r0, #0x0002			@ .... .... .... ..A.
+	orr	r0, r0, #0x0005			@ .... .... .... .C.M
+#if BTB_ENABLE
+	orr	r0, r0, #0x3900			@ ..VI Z..S .... ....
+#else
+	orr	r0, r0, #0x3100			@ ..VI ...S .... ....
+#endif
+#if L2_CACHE_ENABLE
+	orr 	r0, r0, #0x4000000		@ L2 enable
+#endif
+	mov	pc, lr
+
+	.size	__xsc3_setup, . - __xsc3_setup
+
+	__INITDATA
+
+/*
+ * Purpose : Function pointers used to access above functions - all calls
+ *	     come through these
+ */
+
+	.type	xsc3_processor_functions, #object
+ENTRY(xsc3_processor_functions)
+	.word	v5t_early_abort
+	.word	cpu_xsc3_proc_init
+	.word	cpu_xsc3_proc_fin
+	.word	cpu_xsc3_reset
+	.word	cpu_xsc3_do_idle
+	.word	cpu_xsc3_dcache_clean_area
+	.word	cpu_xsc3_switch_mm
+	.word	cpu_xsc3_set_pte
+	.size	xsc3_processor_functions, . - xsc3_processor_functions
+
+	.section ".rodata"
+
+	.type	cpu_arch_name, #object
+cpu_arch_name:
+	.asciz	"armv5te"
+	.size	cpu_arch_name, . - cpu_arch_name
+
+	.type	cpu_elf_name, #object
+cpu_elf_name:
+	.asciz	"v5"
+	.size	cpu_elf_name, . - cpu_elf_name
+
+	.type	cpu_xsc3_name, #object
+cpu_xsc3_name:
+	.asciz	"XScale-Core3"
+	.size	cpu_xsc3_name, . - cpu_xsc3_name
+
+	.align
+
+	.section ".proc.info.init", #alloc, #execinstr
+
+	.type	__xsc3_proc_info,#object
+__xsc3_proc_info:
+	.long	0x69056000
+	.long	0xffffe000
+	.long	0x00000c0e
+	b	__xsc3_setup
+	.long	cpu_arch_name
+	.long	cpu_elf_name
+	.long	HWCAP_SWP|HWCAP_HALF|HWCAP_THUMB|HWCAP_FAST_MULT|HWCAP_EDSP
+	.long	cpu_xsc3_name
+	.long	xsc3_processor_functions
+	.long	v4wbi_tlb_fns
+	.long	xsc3_mc_user_fns
+	.long	xsc3_cache_fns
+	.size	__xsc3_proc_info, . - __xsc3_proc_info