freebsd-nq/sys/mips/include/cache_r4k.h
Pedro F. Giffuni 19d3b47b92 sys/mips: further adoption of SPDX licensing ID tags.
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.

The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
2017-11-27 15:07:26 +00:00

437 lines
16 KiB
C

/* $NetBSD: cache_r4k.h,v 1.10 2003/03/08 04:43:26 rafal Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright 2001 Wasabi Systems, Inc.
* All rights reserved.
*
* Written by Jason R. Thorpe for Wasabi Systems, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed for the NetBSD Project by
* Wasabi Systems, Inc.
* 4. The name of Wasabi Systems, Inc. may not be used to endorse
* or promote products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
* $FreeBSD$
*/
/*
* Cache definitions/operations for R4000-style caches.
*/
#define CACHE_R4K_I 0
#define CACHE_R4K_D 1
#define CACHE_R4K_SI 2
#define CACHE_R4K_SD 3
#define CACHEOP_R4K_INDEX_INV (0 << 2) /* I, SI */
#define CACHEOP_R4K_INDEX_WB_INV (0 << 2) /* D, SD */
#define CACHEOP_R4K_INDEX_LOAD_TAG (1 << 2) /* all */
#define CACHEOP_R4K_INDEX_STORE_TAG (2 << 2) /* all */
#define CACHEOP_R4K_CREATE_DIRTY_EXCL (3 << 2) /* D, SD */
#define CACHEOP_R4K_HIT_INV (4 << 2) /* all */
#define CACHEOP_R4K_HIT_WB_INV (5 << 2) /* D, SD */
#define CACHEOP_R4K_FILL (5 << 2) /* I */
#define CACHEOP_R4K_HIT_WB (6 << 2) /* I, D, SD */
#define CACHEOP_R4K_HIT_SET_VIRTUAL (7 << 2) /* SI, SD */
#if !defined(LOCORE)
/*
* cache_r4k_op_line:
*
* Perform the specified cache operation on a single line.
*/
#define cache_op_r4k_line(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0(%0) \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_8lines_16:
*
* Perform the specified cache operation on 8 16-byte cache lines.
*/
#define cache_r4k_op_8lines_16(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x00(%0); cache %1, 0x10(%0) \n\t" \
"cache %1, 0x20(%0); cache %1, 0x30(%0) \n\t" \
"cache %1, 0x40(%0); cache %1, 0x50(%0) \n\t" \
"cache %1, 0x60(%0); cache %1, 0x70(%0) \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_8lines_32:
*
* Perform the specified cache operation on 8 32-byte cache lines.
*/
#define cache_r4k_op_8lines_32(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x00(%0); cache %1, 0x20(%0) \n\t" \
"cache %1, 0x40(%0); cache %1, 0x60(%0) \n\t" \
"cache %1, 0x80(%0); cache %1, 0xa0(%0) \n\t" \
"cache %1, 0xc0(%0); cache %1, 0xe0(%0) \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_8lines_64:
*
* Perform the specified cache operation on 8 64-byte cache lines.
*/
#define cache_r4k_op_8lines_64(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x000(%0); cache %1, 0x040(%0) \n\t" \
"cache %1, 0x080(%0); cache %1, 0x0c0(%0) \n\t" \
"cache %1, 0x100(%0); cache %1, 0x140(%0) \n\t" \
"cache %1, 0x180(%0); cache %1, 0x1c0(%0) \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_32lines_16:
*
* Perform the specified cache operation on 32 16-byte
* cache lines.
*/
#define cache_r4k_op_32lines_16(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x000(%0); cache %1, 0x010(%0); \n\t" \
"cache %1, 0x020(%0); cache %1, 0x030(%0); \n\t" \
"cache %1, 0x040(%0); cache %1, 0x050(%0); \n\t" \
"cache %1, 0x060(%0); cache %1, 0x070(%0); \n\t" \
"cache %1, 0x080(%0); cache %1, 0x090(%0); \n\t" \
"cache %1, 0x0a0(%0); cache %1, 0x0b0(%0); \n\t" \
"cache %1, 0x0c0(%0); cache %1, 0x0d0(%0); \n\t" \
"cache %1, 0x0e0(%0); cache %1, 0x0f0(%0); \n\t" \
"cache %1, 0x100(%0); cache %1, 0x110(%0); \n\t" \
"cache %1, 0x120(%0); cache %1, 0x130(%0); \n\t" \
"cache %1, 0x140(%0); cache %1, 0x150(%0); \n\t" \
"cache %1, 0x160(%0); cache %1, 0x170(%0); \n\t" \
"cache %1, 0x180(%0); cache %1, 0x190(%0); \n\t" \
"cache %1, 0x1a0(%0); cache %1, 0x1b0(%0); \n\t" \
"cache %1, 0x1c0(%0); cache %1, 0x1d0(%0); \n\t" \
"cache %1, 0x1e0(%0); cache %1, 0x1f0(%0); \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_32lines_32:
*
* Perform the specified cache operation on 32 32-byte
* cache lines.
*/
#define cache_r4k_op_32lines_32(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x000(%0); cache %1, 0x020(%0); \n\t" \
"cache %1, 0x040(%0); cache %1, 0x060(%0); \n\t" \
"cache %1, 0x080(%0); cache %1, 0x0a0(%0); \n\t" \
"cache %1, 0x0c0(%0); cache %1, 0x0e0(%0); \n\t" \
"cache %1, 0x100(%0); cache %1, 0x120(%0); \n\t" \
"cache %1, 0x140(%0); cache %1, 0x160(%0); \n\t" \
"cache %1, 0x180(%0); cache %1, 0x1a0(%0); \n\t" \
"cache %1, 0x1c0(%0); cache %1, 0x1e0(%0); \n\t" \
"cache %1, 0x200(%0); cache %1, 0x220(%0); \n\t" \
"cache %1, 0x240(%0); cache %1, 0x260(%0); \n\t" \
"cache %1, 0x280(%0); cache %1, 0x2a0(%0); \n\t" \
"cache %1, 0x2c0(%0); cache %1, 0x2e0(%0); \n\t" \
"cache %1, 0x300(%0); cache %1, 0x320(%0); \n\t" \
"cache %1, 0x340(%0); cache %1, 0x360(%0); \n\t" \
"cache %1, 0x380(%0); cache %1, 0x3a0(%0); \n\t" \
"cache %1, 0x3c0(%0); cache %1, 0x3e0(%0); \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_32lines_64:
*
* Perform the specified cache operation on 32 64-byte
* cache lines.
*/
#define cache_r4k_op_32lines_64(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x000(%0); cache %1, 0x040(%0); \n\t" \
"cache %1, 0x080(%0); cache %1, 0x0c0(%0); \n\t" \
"cache %1, 0x100(%0); cache %1, 0x140(%0); \n\t" \
"cache %1, 0x180(%0); cache %1, 0x1c0(%0); \n\t" \
"cache %1, 0x200(%0); cache %1, 0x240(%0); \n\t" \
"cache %1, 0x280(%0); cache %1, 0x2c0(%0); \n\t" \
"cache %1, 0x300(%0); cache %1, 0x340(%0); \n\t" \
"cache %1, 0x380(%0); cache %1, 0x3c0(%0); \n\t" \
"cache %1, 0x400(%0); cache %1, 0x440(%0); \n\t" \
"cache %1, 0x480(%0); cache %1, 0x4c0(%0); \n\t" \
"cache %1, 0x500(%0); cache %1, 0x540(%0); \n\t" \
"cache %1, 0x580(%0); cache %1, 0x5c0(%0); \n\t" \
"cache %1, 0x600(%0); cache %1, 0x640(%0); \n\t" \
"cache %1, 0x680(%0); cache %1, 0x6c0(%0); \n\t" \
"cache %1, 0x700(%0); cache %1, 0x740(%0); \n\t" \
"cache %1, 0x780(%0); cache %1, 0x7c0(%0); \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_32lines_128:
*
* Perform the specified cache operation on 32 128-byte
* cache lines.
*/
#define cache_r4k_op_32lines_128(va, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %1, 0x0000(%0); cache %1, 0x0080(%0); \n\t" \
"cache %1, 0x0100(%0); cache %1, 0x0180(%0); \n\t" \
"cache %1, 0x0200(%0); cache %1, 0x0280(%0); \n\t" \
"cache %1, 0x0300(%0); cache %1, 0x0380(%0); \n\t" \
"cache %1, 0x0400(%0); cache %1, 0x0480(%0); \n\t" \
"cache %1, 0x0500(%0); cache %1, 0x0580(%0); \n\t" \
"cache %1, 0x0600(%0); cache %1, 0x0680(%0); \n\t" \
"cache %1, 0x0700(%0); cache %1, 0x0780(%0); \n\t" \
"cache %1, 0x0800(%0); cache %1, 0x0880(%0); \n\t" \
"cache %1, 0x0900(%0); cache %1, 0x0980(%0); \n\t" \
"cache %1, 0x0a00(%0); cache %1, 0x0a80(%0); \n\t" \
"cache %1, 0x0b00(%0); cache %1, 0x0b80(%0); \n\t" \
"cache %1, 0x0c00(%0); cache %1, 0x0c80(%0); \n\t" \
"cache %1, 0x0d00(%0); cache %1, 0x0d80(%0); \n\t" \
"cache %1, 0x0e00(%0); cache %1, 0x0e80(%0); \n\t" \
"cache %1, 0x0f00(%0); cache %1, 0x0f80(%0); \n\t" \
".set reorder" \
: \
: "r" (va), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_16lines_16_2way:
*
* Perform the specified cache operation on 16 16-byte
* cache lines, 2-ways.
*/
#define cache_r4k_op_16lines_16_2way(va1, va2, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %2, 0x000(%0); cache %2, 0x000(%1); \n\t" \
"cache %2, 0x010(%0); cache %2, 0x010(%1); \n\t" \
"cache %2, 0x020(%0); cache %2, 0x020(%1); \n\t" \
"cache %2, 0x030(%0); cache %2, 0x030(%1); \n\t" \
"cache %2, 0x040(%0); cache %2, 0x040(%1); \n\t" \
"cache %2, 0x050(%0); cache %2, 0x050(%1); \n\t" \
"cache %2, 0x060(%0); cache %2, 0x060(%1); \n\t" \
"cache %2, 0x070(%0); cache %2, 0x070(%1); \n\t" \
"cache %2, 0x080(%0); cache %2, 0x080(%1); \n\t" \
"cache %2, 0x090(%0); cache %2, 0x090(%1); \n\t" \
"cache %2, 0x0a0(%0); cache %2, 0x0a0(%1); \n\t" \
"cache %2, 0x0b0(%0); cache %2, 0x0b0(%1); \n\t" \
"cache %2, 0x0c0(%0); cache %2, 0x0c0(%1); \n\t" \
"cache %2, 0x0d0(%0); cache %2, 0x0d0(%1); \n\t" \
"cache %2, 0x0e0(%0); cache %2, 0x0e0(%1); \n\t" \
"cache %2, 0x0f0(%0); cache %2, 0x0f0(%1); \n\t" \
".set reorder" \
: \
: "r" (va1), "r" (va2), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_16lines_32_2way:
*
* Perform the specified cache operation on 16 32-byte
* cache lines, 2-ways.
*/
#define cache_r4k_op_16lines_32_2way(va1, va2, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %2, 0x000(%0); cache %2, 0x000(%1); \n\t" \
"cache %2, 0x020(%0); cache %2, 0x020(%1); \n\t" \
"cache %2, 0x040(%0); cache %2, 0x040(%1); \n\t" \
"cache %2, 0x060(%0); cache %2, 0x060(%1); \n\t" \
"cache %2, 0x080(%0); cache %2, 0x080(%1); \n\t" \
"cache %2, 0x0a0(%0); cache %2, 0x0a0(%1); \n\t" \
"cache %2, 0x0c0(%0); cache %2, 0x0c0(%1); \n\t" \
"cache %2, 0x0e0(%0); cache %2, 0x0e0(%1); \n\t" \
"cache %2, 0x100(%0); cache %2, 0x100(%1); \n\t" \
"cache %2, 0x120(%0); cache %2, 0x120(%1); \n\t" \
"cache %2, 0x140(%0); cache %2, 0x140(%1); \n\t" \
"cache %2, 0x160(%0); cache %2, 0x160(%1); \n\t" \
"cache %2, 0x180(%0); cache %2, 0x180(%1); \n\t" \
"cache %2, 0x1a0(%0); cache %2, 0x1a0(%1); \n\t" \
"cache %2, 0x1c0(%0); cache %2, 0x1c0(%1); \n\t" \
"cache %2, 0x1e0(%0); cache %2, 0x1e0(%1); \n\t" \
".set reorder" \
: \
: "r" (va1), "r" (va2), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_8lines_16_4way:
*
* Perform the specified cache operation on 8 16-byte
* cache lines, 4-ways.
*/
#define cache_r4k_op_8lines_16_4way(va1, va2, va3, va4, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %4, 0x000(%0); cache %4, 0x000(%1); \n\t" \
"cache %4, 0x000(%2); cache %4, 0x000(%3); \n\t" \
"cache %4, 0x010(%0); cache %4, 0x010(%1); \n\t" \
"cache %4, 0x010(%2); cache %4, 0x010(%3); \n\t" \
"cache %4, 0x020(%0); cache %4, 0x020(%1); \n\t" \
"cache %4, 0x020(%2); cache %4, 0x020(%3); \n\t" \
"cache %4, 0x030(%0); cache %4, 0x030(%1); \n\t" \
"cache %4, 0x030(%2); cache %4, 0x030(%3); \n\t" \
"cache %4, 0x040(%0); cache %4, 0x040(%1); \n\t" \
"cache %4, 0x040(%2); cache %4, 0x040(%3); \n\t" \
"cache %4, 0x050(%0); cache %4, 0x050(%1); \n\t" \
"cache %4, 0x050(%2); cache %4, 0x050(%3); \n\t" \
"cache %4, 0x060(%0); cache %4, 0x060(%1); \n\t" \
"cache %4, 0x060(%2); cache %4, 0x060(%3); \n\t" \
"cache %4, 0x070(%0); cache %4, 0x070(%1); \n\t" \
"cache %4, 0x070(%2); cache %4, 0x070(%3); \n\t" \
".set reorder" \
: \
: "r" (va1), "r" (va2), "r" (va3), "r" (va4), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
/*
* cache_r4k_op_8lines_32_4way:
*
* Perform the specified cache operation on 8 32-byte
* cache lines, 4-ways.
*/
#define cache_r4k_op_8lines_32_4way(va1, va2, va3, va4, op) \
do { \
__asm __volatile( \
".set noreorder \n\t" \
"cache %4, 0x000(%0); cache %4, 0x000(%1); \n\t" \
"cache %4, 0x000(%2); cache %4, 0x000(%3); \n\t" \
"cache %4, 0x020(%0); cache %4, 0x020(%1); \n\t" \
"cache %4, 0x020(%2); cache %4, 0x020(%3); \n\t" \
"cache %4, 0x040(%0); cache %4, 0x040(%1); \n\t" \
"cache %4, 0x040(%2); cache %4, 0x040(%3); \n\t" \
"cache %4, 0x060(%0); cache %4, 0x060(%1); \n\t" \
"cache %4, 0x060(%2); cache %4, 0x060(%3); \n\t" \
"cache %4, 0x080(%0); cache %4, 0x080(%1); \n\t" \
"cache %4, 0x080(%2); cache %4, 0x080(%3); \n\t" \
"cache %4, 0x0a0(%0); cache %4, 0x0a0(%1); \n\t" \
"cache %4, 0x0a0(%2); cache %4, 0x0a0(%3); \n\t" \
"cache %4, 0x0c0(%0); cache %4, 0x0c0(%1); \n\t" \
"cache %4, 0x0c0(%2); cache %4, 0x0c0(%3); \n\t" \
"cache %4, 0x0e0(%0); cache %4, 0x0e0(%1); \n\t" \
"cache %4, 0x0e0(%2); cache %4, 0x0e0(%3); \n\t" \
".set reorder" \
: \
: "r" (va1), "r" (va2), "r" (va3), "r" (va4), "i" (op) \
: "memory"); \
} while (/*CONSTCOND*/0)
void r4k_icache_sync_all_16(void);
void r4k_icache_sync_range_16(vm_paddr_t, vm_size_t);
void r4k_icache_sync_range_index_16(vm_paddr_t, vm_size_t);
void r4k_icache_sync_all_32(void);
void r4k_icache_sync_range_32(vm_paddr_t, vm_size_t);
void r4k_icache_sync_range_index_32(vm_paddr_t, vm_size_t);
void r4k_pdcache_wbinv_all_16(void);
void r4k_pdcache_wbinv_range_16(vm_paddr_t, vm_size_t);
void r4k_pdcache_wbinv_range_index_16(vm_paddr_t, vm_size_t);
void r4k_pdcache_inv_range_16(vm_paddr_t, vm_size_t);
void r4k_pdcache_wb_range_16(vm_paddr_t, vm_size_t);
void r4k_pdcache_wbinv_all_32(void);
void r4k_pdcache_wbinv_range_32(vm_paddr_t, vm_size_t);
void r4k_pdcache_wbinv_range_index_32(vm_paddr_t, vm_size_t);
void r4k_pdcache_inv_range_32(vm_paddr_t, vm_size_t);
void r4k_pdcache_wb_range_32(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_all_32(void);
void r4k_sdcache_wbinv_range_32(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_range_index_32(vm_paddr_t, vm_size_t);
void r4k_sdcache_inv_range_32(vm_paddr_t, vm_size_t);
void r4k_sdcache_wb_range_32(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_all_128(void);
void r4k_sdcache_wbinv_range_128(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_range_index_128(vm_paddr_t, vm_size_t);
void r4k_sdcache_inv_range_128(vm_paddr_t, vm_size_t);
void r4k_sdcache_wb_range_128(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_all_generic(void);
void r4k_sdcache_wbinv_range_generic(vm_paddr_t, vm_size_t);
void r4k_sdcache_wbinv_range_index_generic(vm_paddr_t, vm_size_t);
void r4k_sdcache_inv_range_generic(vm_paddr_t, vm_size_t);
void r4k_sdcache_wb_range_generic(vm_paddr_t, vm_size_t);
#endif /* !LOCORE */