/* $NetBSD: vm_machdep.c,v 1.123 2023/10/06 11:53:26 skrll Exp $ */

/*
 * Copyright (c) 1994, 1995, 1996 Carnegie-Mellon University.
 * All rights reserved.
 *
 * Author: Chris G. Demetriou
 *
 * Permission to use, copy, modify and distribute this software and
 * its documentation is hereby granted, provided that both the copyright
 * notice and this permission notice appear in all copies of the
 * software, derivative works or modified versions, and any portions
 * thereof, and that both notices appear in supporting documentation.
 *
 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
 * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
 *
 * Carnegie Mellon requests users of this software to return to
 *
 *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
 *  School of Computer Science
 *  Carnegie Mellon University
 *  Pittsburgh PA 15213-3890
 *
 * any improvements or extensions that they make and grant Carnegie the
 * rights to redistribute these changes.
 */

#include <sys/cdefs.h>                  /* RCS ID & Copyright macro defns */

__KERNEL_RCSID(0, "$NetBSD: vm_machdep.c,v 1.123 2023/10/06 11:53:26 skrll Exp $");

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/buf.h>
#include <sys/vnode.h>
#include <sys/core.h>
#include <sys/exec.h>

#include <uvm/uvm.h>

#include <machine/cpu.h>
#include <machine/alpha.h>
#include <machine/pmap.h>
#include <machine/reg.h>

void
cpu_lwp_free(struct lwp *l, int proc)
{
        (void) l;
}

void
cpu_lwp_free2(struct lwp *l)
{
        (void) l;
}

/*
 * This is a backstop used to ensure that kernel threads never do
 * something silly like attempt to return to userspace.  We achieve
 * this by putting this at the root of their call graph instead of
 * exception_return().
 */
void
alpha_kthread_backstop(void)
{
        struct lwp * const l = curlwp;

        panic("kthread lwp %p (%s) hit the backstop", l, l->l_name);
}

/*
 * Finish a fork operation, with thread l2 nearly set up.
 * Copy and update the pcb and trap frame, making the child ready to run.
 *
 * Rig the child's kernel stack so that it will start out in
 * lwp_trampoline() and call child_return() with l2 as an
 * argument. This causes the newly-created child thread to go
 * directly to user level with an apparent return value of 0 from
 * fork(), while the parent process returns normally.
 *
 * l1 is the thread being forked; if l1 == &lwp0, we are creating
 * a kernel thread, and the return path and argument are specified with
 * `func' and `arg'.
 *
 * If an alternate user-level stack is requested (with non-zero values
 * in both the stack and stacksize args), set up the user stack pointer
 * accordingly.
 */
void
cpu_lwp_fork(struct lwp *l1, struct lwp *l2, void *stack, size_t stacksize,
    void (*func)(void *), void *arg)
{
        struct pcb *pcb1, *pcb2;
        extern void lwp_trampoline(void);

        pcb1 = lwp_getpcb(l1);
        pcb2 = lwp_getpcb(l2);

        l2->l_md.md_tf = l1->l_md.md_tf;
        l2->l_md.md_flags = l1->l_md.md_flags & MDLWP_FP_C;
        KASSERT(l2->l_md.md_astpending == 0);

        /*
         * Cache the physical address of the pcb, so we can
         * swap to it easily.
         */
        l2->l_md.md_pcbpaddr = (void *)vtophys((vaddr_t)pcb2);

        /*
         * Copy pcb and user stack pointer from proc p1 to p2.
         * If specified, give the child a different stack.
         * Floating point state from the FP chip has already been saved.
         */
        *pcb2 = *pcb1;
        if (stack != NULL) {
                pcb2->pcb_hw.apcb_usp =
                    ((u_long)stack + stacksize) & ~((u_long)STACK_ALIGNBYTES);
        } else {
                pcb2->pcb_hw.apcb_usp = alpha_pal_rdusp();
        }

        /*
         * Put l2 on the kernel's page tables until its first trip
         * through pmap_activate().
         */
        pcb2->pcb_hw.apcb_ptbr =
            ALPHA_K0SEG_TO_PHYS((vaddr_t)kernel_lev1map) >> PGSHIFT;
        pcb2->pcb_hw.apcb_asn = PMAP_ASN_KERNEL;

#ifdef DIAGNOSTIC
        /*
         * If l1 != curlwp && l1 == &lwp0, we are creating a kernel
         * thread.
         */
        if (l1 != curlwp && l1 != &lwp0)
                panic("cpu_lwp_fork: curlwp");
#endif

        /*
         * create the child's kernel stack, from scratch.
         */
        {
                struct trapframe *l2tf;
                uint64_t call_root;

                /*
                 * Pick a stack pointer, leaving room for a trapframe;
                 * copy trapframe from parent so return to user mode
                 * will be to right address, with correct registers.
                 */
                l2tf = l2->l_md.md_tf = (struct trapframe *)
                    (uvm_lwp_getuarea(l2) + USPACE - sizeof(struct trapframe));
                memcpy(l2->l_md.md_tf, l1->l_md.md_tf,
                    sizeof(struct trapframe));

                /*
                 * Set up return-value registers as fork() libc stub expects.
                 */
                l2tf->tf_regs[FRAME_V0] = l1->l_proc->p_pid; /* parent's pid */
                l2tf->tf_regs[FRAME_A3] = 0;            /* no error */
                l2tf->tf_regs[FRAME_A4] = 1;            /* is child */

                /*
                 * Normal LWPs have their return address set to
                 * exception_return() so that they'll pop into
                 * user space.  But kernel threads don't have
                 * a user space, so we put a backtop in place
                 * just in case they try.
                 */
                if (__predict_true(l2->l_proc != &proc0))
                        call_root = (uint64_t)exception_return;
                else
                        call_root = (uint64_t)alpha_kthread_backstop;

                pcb2->pcb_hw.apcb_ksp =
                    (uint64_t)l2->l_md.md_tf;
                pcb2->pcb_context[0] =
                    (uint64_t)func;                     /* s0: pc */
                pcb2->pcb_context[1] =
                    call_root;                          /* s1: ra */
                pcb2->pcb_context[2] =
                    (uint64_t)arg;                      /* s2: arg */
                pcb2->pcb_context[3] =
                    (uint64_t)l2;                       /* s3: lwp */
                pcb2->pcb_context[7] =
                    (uint64_t)lwp_trampoline;           /* ra: assembly magic */
        }
}

/*
 * Map a user I/O request into kernel virtual address space.
 * Note: the pages are already locked by uvm_vslock(), so we
 * do not need to pass an access_type to pmap_enter().
 */
int
vmapbuf(struct buf *bp, vsize_t len)
{
        vaddr_t faddr, taddr, off;
        paddr_t pa;
        struct proc *p;

        if ((bp->b_flags & B_PHYS) == 0)
                panic("vmapbuf");
        p = bp->b_proc;
        bp->b_saveaddr = bp->b_data;
        faddr = trunc_page((vaddr_t)bp->b_data);
        off = (vaddr_t)bp->b_data - faddr;
        len = round_page(off + len);
        taddr = uvm_km_alloc(phys_map, len, 0, UVM_KMF_VAONLY|UVM_KMF_WAITVA);
        bp->b_data = (void *)(taddr + off);
        len = atop(len);
        while (len--) {
                if (pmap_extract(vm_map_pmap(&p->p_vmspace->vm_map), faddr,
                    &pa) == false)
                        panic("vmapbuf: null page frame");
                pmap_enter(vm_map_pmap(phys_map), taddr, trunc_page(pa),
                    VM_PROT_READ|VM_PROT_WRITE, PMAP_WIRED);
                faddr += PAGE_SIZE;
                taddr += PAGE_SIZE;
        }
        pmap_update(vm_map_pmap(phys_map));

        return 0;
}

/*
 * Unmap a previously-mapped user I/O request.
 */
void
vunmapbuf(struct buf *bp, vsize_t len)
{
        vaddr_t addr, off;

        if ((bp->b_flags & B_PHYS) == 0)
                panic("vunmapbuf");
        addr = trunc_page((vaddr_t)bp->b_data);
        off = (vaddr_t)bp->b_data - addr;
        len = round_page(off + len);
        pmap_remove(vm_map_pmap(phys_map), addr, addr + len);
        pmap_update(vm_map_pmap(phys_map));
        uvm_km_free(phys_map, addr, len, UVM_KMF_VAONLY);
        bp->b_data = bp->b_saveaddr;
        bp->b_saveaddr = NULL;
}

#ifdef __HAVE_CPU_UAREA_ROUTINES
static struct evcnt uarea_direct_success =
    EVCNT_INITIALIZER(EVCNT_TYPE_MISC, NULL, "uarea direct", "success");
static struct evcnt uarea_direct_failure =
    EVCNT_INITIALIZER(EVCNT_TYPE_MISC, NULL, "uarea direct", "failure");

EVCNT_ATTACH_STATIC(uarea_direct_success);
EVCNT_ATTACH_STATIC(uarea_direct_failure);

void *
cpu_uarea_alloc(bool system)
{
        struct pglist pglist;
        int error;

        /*
         * Allocate a new physically contiguous uarea which can be
         * direct-mapped.
         */
        error = uvm_pglistalloc(USPACE, 0, ptoa(physmem), 0, 0, &pglist, 1, 1);
        if (error) {
                atomic_inc_ulong(&uarea_direct_failure.ev_count);
                return NULL;
        }
        atomic_inc_ulong(&uarea_direct_success.ev_count);

        /*
         * Get the physical address from the first page.
         */
        const struct vm_page * const pg = TAILQ_FIRST(&pglist);
        KASSERT(pg != NULL);
        const paddr_t pa = VM_PAGE_TO_PHYS(pg);

        /*
         * We need to return a direct-mapped VA for the pa.
         */

        return (void *)PMAP_MAP_POOLPAGE(pa);
}

/*
 * Return true if we freed it, false if we didn't.
 */
bool
cpu_uarea_free(void *vva)
{
        vaddr_t va = (vaddr_t) vva;
        if (va >= VM_MIN_KERNEL_ADDRESS && va < VM_MAX_KERNEL_ADDRESS)
                return false;

        /*
         * Since the pages are physically contiguous, the vm_page structure
         * will be as well.
         */
        struct vm_page *pg = PHYS_TO_VM_PAGE(PMAP_UNMAP_POOLPAGE(va));
        KASSERT(pg != NULL);
        for (size_t i = 0; i < UPAGES; i++, pg++) {
                uvm_pagefree(pg);
        }
        return true;
}
#endif /* __HAVE_CPU_UAREA_ROUTINES */