/*
|
* Copyright 2003-2013 Broadcom Corporation.
|
* All Rights Reserved.
|
*
|
* This software is available to you under a choice of one of two
|
* licenses. You may choose to be licensed under the terms of the GNU
|
* General Public License (GPL) Version 2, available from the file
|
* COPYING in the main directory of this source tree, or the Broadcom
|
* license below:
|
*
|
* 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.
|
*
|
* THIS SOFTWARE IS PROVIDED BY BROADCOM ``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 BROADCOM OR CONTRIBUTORS 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.
|
*/
|
|
|
#include <asm/asm.h>
|
#include <asm/asm-offsets.h>
|
#include <asm/cpu.h>
|
#include <asm/cacheops.h>
|
#include <asm/regdef.h>
|
#include <asm/mipsregs.h>
|
#include <asm/stackframe.h>
|
#include <asm/asmmacro.h>
|
#include <asm/addrspace.h>
|
|
#include <asm/netlogic/common.h>
|
|
#include <asm/netlogic/xlp-hal/iomap.h>
|
#include <asm/netlogic/xlp-hal/xlp.h>
|
#include <asm/netlogic/xlp-hal/sys.h>
|
#include <asm/netlogic/xlp-hal/cpucontrol.h>
|
|
#define SYS_CPU_COHERENT_BASE CKSEG1ADDR(XLP_DEFAULT_IO_BASE) + \
|
XLP_IO_SYS_OFFSET(0) + XLP_IO_PCI_HDRSZ + \
|
SYS_CPU_NONCOHERENT_MODE * 4
|
|
/* Enable XLP features and workarounds in the LSU */
|
.macro xlp_config_lsu
|
li t0, LSU_DEFEATURE
|
mfcr t1, t0
|
|
lui t2, 0x4080 /* Enable Unaligned Access, L2HPE */
|
or t1, t1, t2
|
mtcr t1, t0
|
|
li t0, ICU_DEFEATURE
|
mfcr t1, t0
|
ori t1, 0x1000 /* Enable Icache partitioning */
|
mtcr t1, t0
|
|
li t0, SCHED_DEFEATURE
|
lui t1, 0x0100 /* Disable BRU accepting ALU ops */
|
mtcr t1, t0
|
.endm
|
|
/*
|
* Allow access to physical mem >64G by enabling ELPA in PAGEGRAIN
|
* register. This is needed before going to C code since the SP can
|
* in this region. Called from all HW threads.
|
*/
|
.macro xlp_early_mmu_init
|
mfc0 t0, CP0_PAGEMASK, 1
|
li t1, (1 << 29) /* ELPA bit */
|
or t0, t1
|
mtc0 t0, CP0_PAGEMASK, 1
|
.endm
|
|
/*
|
* L1D cache has to be flushed before enabling threads in XLP.
|
* On XLP8xx/XLP3xx, we do a low level flush using processor control
|
* registers. On XLPII CPUs, usual cache instructions work.
|
*/
|
.macro xlp_flush_l1_dcache
|
mfc0 t0, CP0_PRID
|
andi t0, t0, PRID_IMP_MASK
|
slt t1, t0, 0x1200
|
beqz t1, 15f
|
nop
|
|
/* XLP8xx low level cache flush */
|
li t0, LSU_DEBUG_DATA0
|
li t1, LSU_DEBUG_ADDR
|
li t2, 0 /* index */
|
li t3, 0x1000 /* loop count */
|
11:
|
sll v0, t2, 5
|
mtcr zero, t0
|
ori v1, v0, 0x3 /* way0 | write_enable | write_active */
|
mtcr v1, t1
|
12:
|
mfcr v1, t1
|
andi v1, 0x1 /* wait for write_active == 0 */
|
bnez v1, 12b
|
nop
|
mtcr zero, t0
|
ori v1, v0, 0x7 /* way1 | write_enable | write_active */
|
mtcr v1, t1
|
13:
|
mfcr v1, t1
|
andi v1, 0x1 /* wait for write_active == 0 */
|
bnez v1, 13b
|
nop
|
addi t2, 1
|
bne t3, t2, 11b
|
nop
|
b 17f
|
nop
|
|
/* XLPII CPUs, Invalidate all 64k of L1 D-cache */
|
15:
|
li t0, 0x80000000
|
li t1, 0x80010000
|
16: cache Index_Writeback_Inv_D, 0(t0)
|
addiu t0, t0, 32
|
bne t0, t1, 16b
|
nop
|
17:
|
.endm
|
|
/*
|
* nlm_reset_entry will be copied to the reset entry point for
|
* XLR and XLP. The XLP cores start here when they are woken up. This
|
* is also the NMI entry point.
|
*
|
* We use scratch reg 6/7 to save k0/k1 and check for NMI first.
|
*
|
* The data corresponding to reset/NMI is stored at RESET_DATA_PHYS
|
* location, this will have the thread mask (used when core is woken up)
|
* and the current NMI handler in case we reached here for an NMI.
|
*
|
* When a core or thread is newly woken up, it marks itself ready and
|
* loops in a 'wait'. When the CPU really needs waking up, we send an NMI
|
* IPI to it, with the NMI handler set to prom_boot_secondary_cpus
|
*/
|
.set noreorder
|
.set noat
|
.set arch=xlr /* for mfcr/mtcr, XLR is sufficient */
|
|
FEXPORT(nlm_reset_entry)
|
dmtc0 k0, $22, 6
|
dmtc0 k1, $22, 7
|
mfc0 k0, CP0_STATUS
|
li k1, 0x80000
|
and k1, k0, k1
|
beqz k1, 1f /* go to real reset entry */
|
nop
|
li k1, CKSEG1ADDR(RESET_DATA_PHYS) /* NMI */
|
ld k0, BOOT_NMI_HANDLER(k1)
|
jr k0
|
nop
|
|
1: /* Entry point on core wakeup */
|
mfc0 t0, CP0_PRID /* processor ID */
|
andi t0, PRID_IMP_MASK
|
li t1, 0x1500 /* XLP 9xx */
|
beq t0, t1, 2f /* does not need to set coherent */
|
nop
|
|
li t1, 0x1300 /* XLP 5xx */
|
beq t0, t1, 2f /* does not need to set coherent */
|
nop
|
|
/* set bit in SYS coherent register for the core */
|
mfc0 t0, CP0_EBASE
|
mfc0 t1, CP0_EBASE
|
srl t1, 5
|
andi t1, 0x3 /* t1 <- node */
|
li t2, 0x40000
|
mul t3, t2, t1 /* t3 = node * 0x40000 */
|
srl t0, t0, 2
|
and t0, t0, 0x7 /* t0 <- core */
|
li t1, 0x1
|
sll t0, t1, t0
|
nor t0, t0, zero /* t0 <- ~(1 << core) */
|
li t2, SYS_CPU_COHERENT_BASE
|
add t2, t2, t3 /* t2 <- SYS offset for node */
|
lw t1, 0(t2)
|
and t1, t1, t0
|
sw t1, 0(t2)
|
|
/* read back to ensure complete */
|
lw t1, 0(t2)
|
sync
|
|
2:
|
/* Configure LSU on Non-0 Cores. */
|
xlp_config_lsu
|
/* FALL THROUGH */
|
|
/*
|
* Wake up sibling threads from the initial thread in a core.
|
*/
|
EXPORT(nlm_boot_siblings)
|
/* core L1D flush before enable threads */
|
xlp_flush_l1_dcache
|
/* save ra and sp, will be used later (only for boot cpu) */
|
dmtc0 ra, $22, 6
|
dmtc0 sp, $22, 7
|
/* Enable hw threads by writing to MAP_THREADMODE of the core */
|
li t0, CKSEG1ADDR(RESET_DATA_PHYS)
|
lw t1, BOOT_THREAD_MODE(t0) /* t1 <- thread mode */
|
li t0, ((CPU_BLOCKID_MAP << 8) | MAP_THREADMODE)
|
mfcr t2, t0
|
or t2, t2, t1
|
mtcr t2, t0
|
|
/*
|
* The new hardware thread starts at the next instruction
|
* For all the cases other than core 0 thread 0, we will
|
* jump to the secondary wait function.
|
|
* NOTE: All GPR contents are lost after the mtcr above!
|
*/
|
mfc0 v0, CP0_EBASE
|
andi v0, 0x3ff /* v0 <- node/core */
|
|
/*
|
* Errata: to avoid potential live lock, setup IFU_BRUB_RESERVE
|
* when running 4 threads per core
|
*/
|
andi v1, v0, 0x3 /* v1 <- thread id */
|
bnez v1, 2f
|
nop
|
|
/* thread 0 of each core. */
|
li t0, CKSEG1ADDR(RESET_DATA_PHYS)
|
lw t1, BOOT_THREAD_MODE(t0) /* t1 <- thread mode */
|
subu t1, 0x3 /* 4-thread per core mode? */
|
bnez t1, 2f
|
nop
|
|
li t0, IFU_BRUB_RESERVE
|
li t1, 0x55
|
mtcr t1, t0
|
_ehb
|
2:
|
beqz v0, 4f /* boot cpu (cpuid == 0)? */
|
nop
|
|
/* setup status reg */
|
move t1, zero
|
#ifdef CONFIG_64BIT
|
ori t1, ST0_KX
|
#endif
|
mtc0 t1, CP0_STATUS
|
|
xlp_early_mmu_init
|
|
/* mark CPU ready */
|
li t3, CKSEG1ADDR(RESET_DATA_PHYS)
|
ADDIU t1, t3, BOOT_CPU_READY
|
sll v1, v0, 2
|
PTR_ADDU t1, v1
|
li t2, 1
|
sw t2, 0(t1)
|
/* Wait until NMI hits */
|
3: wait
|
b 3b
|
nop
|
|
/*
|
* For the boot CPU, we have to restore ra and sp and return, rest
|
* of the registers will be restored by the caller
|
*/
|
4:
|
dmfc0 ra, $22, 6
|
dmfc0 sp, $22, 7
|
jr ra
|
nop
|
EXPORT(nlm_reset_entry_end)
|
|
LEAF(nlm_init_boot_cpu)
|
#ifdef CONFIG_CPU_XLP
|
xlp_config_lsu
|
xlp_early_mmu_init
|
#endif
|
jr ra
|
nop
|
END(nlm_init_boot_cpu)
|
