/*
 * Copyright (c) 2006-2018, RT-Thread Development Team
 *
 * SPDX-License-Identifier: Apache-2.0
 *
 * Change Logs:
 * Date           Author       Notes
 * 2018/10/28     Bernard      The unify RISC-V porting implementation
 * 2018/12/27     Jesven       Add SMP support
 * 2020/11/20     BalanceTWK   Add FPU support
 */

#include "cpuport.h"

#ifdef RT_USING_SMP
#define rt_hw_interrupt_disable rt_hw_local_irq_disable
#define rt_hw_interrupt_enable  rt_hw_local_irq_enable
#endif

/*
 * rt_base_t rt_hw_interrupt_disable(void);
 */
    .globl rt_hw_interrupt_disable
rt_hw_interrupt_disable:
    csrrci a0, mstatus, 8
    ret

/*
 * void rt_hw_interrupt_enable(rt_base_t level);
 */
    .globl rt_hw_interrupt_enable
rt_hw_interrupt_enable:
    csrw mstatus, a0
    ret

/*
 * #ifdef RT_USING_SMP
 * void rt_hw_context_switch_to(rt_ubase_t to, stuct rt_thread *to_thread);
 * #else
 * void rt_hw_context_switch_to(rt_ubase_t to);
 * #endif
 * a0 --> to
 * a1 --> to_thread
 */
    .globl rt_hw_context_switch_to
rt_hw_context_switch_to:
    LOAD sp, (a0)

#ifdef RT_USING_SMP
    mv   a0,   a1
    call  rt_cpus_lock_status_restore
#endif
    LOAD a0,   2 * REGBYTES(sp)
    csrw mstatus, a0
    j    rt_hw_context_switch_exit

/*
 * #ifdef RT_USING_SMP
 * void rt_hw_context_switch(rt_ubase_t from, rt_ubase_t to, struct rt_thread *to_thread);
 * #else
 * void rt_hw_context_switch(rt_ubase_t from, rt_ubase_t to);
 * #endif
 *
 * a0 --> from
 * a1 --> to
 * a2 --> to_thread
 */
    .globl rt_hw_context_switch
rt_hw_context_switch:
    /* saved from thread context
     *     x1/ra       -> sp(0)
     *     x1/ra       -> sp(1)
     *     mstatus.mie -> sp(2)
     *     x(i)        -> sp(i-4)
     */
#ifdef ARCH_RISCV_FPU
    addi    sp, sp, -32 * FREGBYTES

    FSTORE  f0, 0 * FREGBYTES(sp)
    FSTORE  f1, 1 * FREGBYTES(sp)
    FSTORE  f2, 2 * FREGBYTES(sp)
    FSTORE  f3, 3 * FREGBYTES(sp)
    FSTORE  f4, 4 * FREGBYTES(sp)
    FSTORE  f5, 5 * FREGBYTES(sp)
    FSTORE  f6, 6 * FREGBYTES(sp)
    FSTORE  f7, 7 * FREGBYTES(sp)
    FSTORE  f8, 8 * FREGBYTES(sp)
    FSTORE  f9, 9 * FREGBYTES(sp)
    FSTORE  f10, 10 * FREGBYTES(sp)
    FSTORE  f11, 11 * FREGBYTES(sp)
    FSTORE  f12, 12 * FREGBYTES(sp)
    FSTORE  f13, 13 * FREGBYTES(sp)
    FSTORE  f14, 14 * FREGBYTES(sp)
    FSTORE  f15, 15 * FREGBYTES(sp)
    FSTORE  f16, 16 * FREGBYTES(sp)
    FSTORE  f17, 17 * FREGBYTES(sp)
    FSTORE  f18, 18 * FREGBYTES(sp)
    FSTORE  f19, 19 * FREGBYTES(sp)
    FSTORE  f20, 20 * FREGBYTES(sp)
    FSTORE  f21, 21 * FREGBYTES(sp)
    FSTORE  f22, 22 * FREGBYTES(sp)
    FSTORE  f23, 23 * FREGBYTES(sp)
    FSTORE  f24, 24 * FREGBYTES(sp)
    FSTORE  f25, 25 * FREGBYTES(sp)
    FSTORE  f26, 26 * FREGBYTES(sp)
    FSTORE  f27, 27 * FREGBYTES(sp)
    FSTORE  f28, 28 * FREGBYTES(sp)
    FSTORE  f29, 29 * FREGBYTES(sp)
    FSTORE  f30, 30 * FREGBYTES(sp)
    FSTORE  f31, 31 * FREGBYTES(sp)

#endif
    addi  sp,  sp, -32 * REGBYTES
    STORE sp,  (a0)

    STORE x1,   0 * REGBYTES(sp)
    STORE x1,   1 * REGBYTES(sp)

    csrr a0, mstatus
    andi a0, a0, 8
    beqz a0, save_mpie
    li   a0, 0x80
save_mpie:
    STORE a0,   2 * REGBYTES(sp)

    STORE x4,   4 * REGBYTES(sp)
    STORE x5,   5 * REGBYTES(sp)
    STORE x6,   6 * REGBYTES(sp)
    STORE x7,   7 * REGBYTES(sp)
    STORE x8,   8 * REGBYTES(sp)
    STORE x9,   9 * REGBYTES(sp)
    STORE x10, 10 * REGBYTES(sp)
    STORE x11, 11 * REGBYTES(sp)
    STORE x12, 12 * REGBYTES(sp)
    STORE x13, 13 * REGBYTES(sp)
    STORE x14, 14 * REGBYTES(sp)
    STORE x15, 15 * REGBYTES(sp)
    STORE x16, 16 * REGBYTES(sp)
    STORE x17, 17 * REGBYTES(sp)
    STORE x18, 18 * REGBYTES(sp)
    STORE x19, 19 * REGBYTES(sp)
    STORE x20, 20 * REGBYTES(sp)
    STORE x21, 21 * REGBYTES(sp)
    STORE x22, 22 * REGBYTES(sp)
    STORE x23, 23 * REGBYTES(sp)
    STORE x24, 24 * REGBYTES(sp)
    STORE x25, 25 * REGBYTES(sp)
    STORE x26, 26 * REGBYTES(sp)
    STORE x27, 27 * REGBYTES(sp)
    STORE x28, 28 * REGBYTES(sp)
    STORE x29, 29 * REGBYTES(sp)
    STORE x30, 30 * REGBYTES(sp)
    STORE x31, 31 * REGBYTES(sp)

    /* restore to thread context
     * sp(0) -> epc;
     * sp(1) -> ra;
     * sp(i) -> x(i+2)
     */
    LOAD sp,  (a1)

#ifdef RT_USING_SMP
    mv   a0,   a2
    call  rt_cpus_lock_status_restore
#endif /*RT_USING_SMP*/

    j rt_hw_context_switch_exit

#ifdef RT_USING_SMP
/*
 * void rt_hw_context_switch_interrupt(void *context, rt_ubase_t from, rt_ubase_t to, struct rt_thread *to_thread);
 *
 * a0 --> context
 * a1 --> from
 * a2 --> to
 * a3 --> to_thread
 */
    .globl rt_hw_context_switch_interrupt
rt_hw_context_switch_interrupt:

    STORE a0, 0(a1)

    LOAD  sp, 0(a2)
    move  a0, a3
    call rt_cpus_lock_status_restore

    j rt_hw_context_switch_exit

#endif

.global rt_hw_context_switch_exit
rt_hw_context_switch_exit:
#ifdef RT_USING_SMP
#ifdef RT_USING_SIGNALS
    mv a0, sp

    csrr  t0, mhartid
    /* switch interrupt stack of current cpu */
    la    sp, __stack_start__
    addi  t1, t0, 1
    li    t2, __STACKSIZE__
    mul   t1, t1, t2
    add   sp, sp, t1 /* sp = (cpuid + 1) * __STACKSIZE__ + __stack_start__ */

    call rt_signal_check
    mv sp, a0
#endif
#endif
    /* resw ra to mepc */
    LOAD a0,   0 * REGBYTES(sp)
    csrw mepc, a0

    LOAD x1,   1 * REGBYTES(sp)

    li    t0, 0x00007800
    csrw  mstatus, t0
    LOAD a0,   2 * REGBYTES(sp)
    csrs mstatus, a0

    LOAD x4,   4 * REGBYTES(sp)
    LOAD x5,   5 * REGBYTES(sp)
    LOAD x6,   6 * REGBYTES(sp)
    LOAD x7,   7 * REGBYTES(sp)
    LOAD x8,   8 * REGBYTES(sp)
    LOAD x9,   9 * REGBYTES(sp)
    LOAD x10, 10 * REGBYTES(sp)
    LOAD x11, 11 * REGBYTES(sp)
    LOAD x12, 12 * REGBYTES(sp)
    LOAD x13, 13 * REGBYTES(sp)
    LOAD x14, 14 * REGBYTES(sp)
    LOAD x15, 15 * REGBYTES(sp)
    LOAD x16, 16 * REGBYTES(sp)
    LOAD x17, 17 * REGBYTES(sp)
    LOAD x18, 18 * REGBYTES(sp)
    LOAD x19, 19 * REGBYTES(sp)
    LOAD x20, 20 * REGBYTES(sp)
    LOAD x21, 21 * REGBYTES(sp)
    LOAD x22, 22 * REGBYTES(sp)
    LOAD x23, 23 * REGBYTES(sp)
    LOAD x24, 24 * REGBYTES(sp)
    LOAD x25, 25 * REGBYTES(sp)
    LOAD x26, 26 * REGBYTES(sp)
    LOAD x27, 27 * REGBYTES(sp)
    LOAD x28, 28 * REGBYTES(sp)
    LOAD x29, 29 * REGBYTES(sp)
    LOAD x30, 30 * REGBYTES(sp)
    LOAD x31, 31 * REGBYTES(sp)

    addi sp,  sp, 32 * REGBYTES

#ifdef ARCH_RISCV_FPU
    FLOAD   f0, 0 * FREGBYTES(sp)
    FLOAD   f1, 1 * FREGBYTES(sp)
    FLOAD   f2, 2 * FREGBYTES(sp)
    FLOAD   f3, 3 * FREGBYTES(sp)
    FLOAD   f4, 4 * FREGBYTES(sp)
    FLOAD   f5, 5 * FREGBYTES(sp)
    FLOAD   f6, 6 * FREGBYTES(sp)
    FLOAD   f7, 7 * FREGBYTES(sp)
    FLOAD   f8, 8 * FREGBYTES(sp)
    FLOAD   f9, 9 * FREGBYTES(sp)
    FLOAD   f10, 10 * FREGBYTES(sp)
    FLOAD   f11, 11 * FREGBYTES(sp)
    FLOAD   f12, 12 * FREGBYTES(sp)
    FLOAD   f13, 13 * FREGBYTES(sp)
    FLOAD   f14, 14 * FREGBYTES(sp)
    FLOAD   f15, 15 * FREGBYTES(sp)
    FLOAD   f16, 16 * FREGBYTES(sp)
    FLOAD   f17, 17 * FREGBYTES(sp)
    FLOAD   f18, 18 * FREGBYTES(sp)
    FLOAD   f19, 19 * FREGBYTES(sp)
    FLOAD   f20, 20 * FREGBYTES(sp)
    FLOAD   f21, 21 * FREGBYTES(sp)
    FLOAD   f22, 22 * FREGBYTES(sp)
    FLOAD   f23, 23 * FREGBYTES(sp)
    FLOAD   f24, 24 * FREGBYTES(sp)
    FLOAD   f25, 25 * FREGBYTES(sp)
    FLOAD   f26, 26 * FREGBYTES(sp)
    FLOAD   f27, 27 * FREGBYTES(sp)
    FLOAD   f28, 28 * FREGBYTES(sp)
    FLOAD   f29, 29 * FREGBYTES(sp)
    FLOAD   f30, 30 * FREGBYTES(sp)
    FLOAD   f31, 31 * FREGBYTES(sp)

    addi    sp, sp, 32 * FREGBYTES
#endif

    mret