/*      $NetBSD: mm.c,v 1.25 2024/08/25 11:29:38 uwe Exp $      */

/*-
* Copyright (c) 2002, 2008, 2010 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Christos Zoulas, Joerg Sonnenberger and Mindaugas Rasiukevicius.
*
* 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 THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``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 THE FOUNDATION 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.
*/

/*
* Special /dev/{mem,kmem,zero,null} memory devices.
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: mm.c,v 1.25 2024/08/25 11:29:38 uwe Exp $");

#include "opt_compat_netbsd.h"

#include <sys/param.h>
#include <sys/conf.h>
#include <sys/ioctl.h>
#include <sys/mman.h>
#include <sys/uio.h>
#include <sys/termios.h>

#include <dev/mm.h>

#include <uvm/uvm_extern.h>

static void *           dev_zero_page   __read_mostly;
static kmutex_t         dev_mem_lock    __cacheline_aligned;
static vaddr_t          dev_mem_addr    __read_mostly;

static dev_type_open(mm_open);
static dev_type_read(mm_readwrite);
static dev_type_mmap(mm_mmap);
static dev_type_ioctl(mm_ioctl);

const struct cdevsw mem_cdevsw = {
       .d_open = mm_open,
       .d_close = nullclose,
       .d_read = mm_readwrite,
       .d_write = mm_readwrite,
       .d_ioctl = mm_ioctl,
       .d_stop = nostop,
       .d_tty = notty,
       .d_poll = nopoll,
       .d_mmap = mm_mmap,
       .d_kqfilter = nokqfilter,
       .d_discard = nodiscard,
       .d_flag = D_MPSAFE
};

#ifdef pmax     /* XXX */
const struct cdevsw mem_ultrix_cdevsw = {
       .d_open = nullopen,
       .d_close = nullclose,
       .d_read = mm_readwrite,
       .d_write = mm_readwrite,
       .d_ioctl = mm_ioctl,
       .d_stop = nostop,
       .d_tty = notty,
       .d_poll = nopoll,
       .d_mmap = mm_mmap,
       .d_kqfilter = nokqfilter,
       .d_discard = nodiscard,
       .d_flag = D_MPSAFE
};
#endif

static int
mm_open(dev_t dev, int flag, int mode, struct lwp *l)
{
#ifdef __HAVE_MM_MD_OPEN
       int error;
       if ((error = mm_md_open(dev, flag, mode, l)) != 0)
               return error;
#endif
       l->l_proc->p_flag |= PK_KMEM;
       return 0;
}

/*
* mm_init: initialize memory device driver.
*/
void
mm_init(void)
{
       vaddr_t pg;

       mutex_init(&dev_mem_lock, MUTEX_DEFAULT, IPL_NONE);

       /* Read-only zero-page. */
       pg = uvm_km_alloc(kernel_map, PAGE_SIZE, 0, UVM_KMF_WIRED|UVM_KMF_ZERO);
       KASSERT(pg != 0);
       pmap_protect(pmap_kernel(), pg, pg + PAGE_SIZE, VM_PROT_READ);
       pmap_update(pmap_kernel());
       dev_zero_page = (void *)pg;

#ifndef __HAVE_MM_MD_CACHE_ALIASING
       /* KVA for mappings during I/O. */
       dev_mem_addr = uvm_km_alloc(kernel_map, PAGE_SIZE, 0,
           UVM_KMF_VAONLY|UVM_KMF_WAITVA);
       KASSERT(dev_mem_addr != 0);
#else
       dev_mem_addr = 0;
#endif
}


/*
* dev_mem_getva: get a special virtual address.  If architecture requires,
* allocate VA according to PA, which avoids cache-aliasing issues.  Use a
* constant, general mapping address otherwise.
*/
static inline vaddr_t
dev_mem_getva(paddr_t pa, int color)
{
#ifdef __HAVE_MM_MD_CACHE_ALIASING
       return uvm_km_alloc(kernel_map, PAGE_SIZE,
           color & uvmexp.colormask,
           UVM_KMF_VAONLY | UVM_KMF_WAITVA | UVM_KMF_COLORMATCH);
#else
       return dev_mem_addr;
#endif
}

static inline void
dev_mem_relva(paddr_t pa, vaddr_t va)
{
#ifdef __HAVE_MM_MD_CACHE_ALIASING
       uvm_km_free(kernel_map, va, PAGE_SIZE, UVM_KMF_VAONLY);
#else
       KASSERT(dev_mem_addr == va);
#endif
}

/*
* dev_kmem_readwrite: helper for DEV_MEM (/dev/mem) case of R/W.
*/
static int
dev_mem_readwrite(struct uio *uio, struct iovec *iov)
{
       paddr_t paddr;
       vaddr_t vaddr;
       vm_prot_t prot;
       size_t len, offset;
       bool have_direct;
       int error;
       int color = 0;

       /* Check for wrap around. */
       if ((uintptr_t)uio->uio_offset != uio->uio_offset) {
               return EFAULT;
       }
       paddr = uio->uio_offset & ~PAGE_MASK;
       prot = (uio->uio_rw == UIO_WRITE) ? VM_PROT_WRITE : VM_PROT_READ;
       error = mm_md_physacc(paddr, prot);
       if (error) {
               return error;
       }
       offset = uio->uio_offset & PAGE_MASK;
       len = MIN(uio->uio_resid, PAGE_SIZE - offset);

#ifdef __HAVE_MM_MD_CACHE_ALIASING
       have_direct = mm_md_page_color(paddr, &color);
#else
       have_direct = true;
       color = 0;
#endif

#ifdef __HAVE_MM_MD_DIRECT_MAPPED_PHYS
       /* Is physical address directly mapped?  Return VA. */
       if (have_direct)
               have_direct = mm_md_direct_mapped_phys(paddr, &vaddr);
#else
       vaddr = 0;
       have_direct = false;
#endif
       if (!have_direct) {
               /* Get a special virtual address. */
               const vaddr_t va = dev_mem_getva(paddr, color);

               /* Map selected KVA to physical address. */
               mutex_enter(&dev_mem_lock);
               pmap_kenter_pa(va, paddr, prot, 0);
               pmap_update(pmap_kernel());

               /* Perform I/O. */
               vaddr = va + offset;
               error = uiomove((void *)vaddr, len, uio);

               /* Unmap, flush before unlock. */
               pmap_kremove(va, PAGE_SIZE);
               pmap_update(pmap_kernel());
               mutex_exit(&dev_mem_lock);

               /* "Release" the virtual address. */
               dev_mem_relva(paddr, va);
       } else {
               /* Direct map, just perform I/O. */
               vaddr += offset;
               error = uiomove((void *)vaddr, len, uio);
       }
       return error;
}

/*
* dev_kmem_readwrite: helper for DEV_KMEM (/dev/kmem) case of R/W.
*/
static int
dev_kmem_readwrite(struct uio *uio, struct iovec *iov)
{
       void *addr;
       size_t len, offset;
       vm_prot_t prot;
       int error;
       bool md_kva;

       /* Check for wrap around. */
       addr = (void *)(intptr_t)uio->uio_offset;
       if ((uintptr_t)addr != uio->uio_offset) {
               return EFAULT;
       }
       /*
        * Handle non-page aligned offset.
        * Otherwise, we operate in page-by-page basis.
        */
       offset = uio->uio_offset & PAGE_MASK;
       len = MIN(uio->uio_resid, PAGE_SIZE - offset);
       prot = (uio->uio_rw == UIO_WRITE) ? VM_PROT_WRITE : VM_PROT_READ;

       md_kva = false;

#ifdef __HAVE_MM_MD_DIRECT_MAPPED_IO
       paddr_t paddr;
       /* MD case: is this is a directly mapped address? */
       if (mm_md_direct_mapped_io(addr, &paddr)) {
               /* If so, validate physical address. */
               error = mm_md_physacc(paddr, prot);
               if (error) {
                       return error;
               }
               md_kva = true;
       }
#endif
       if (!md_kva) {
               bool checked = false;

#ifdef __HAVE_MM_MD_KERNACC
               /* MD check for the address. */
               error = mm_md_kernacc(addr, prot, &checked);
               if (error) {
                       return error;
               }
#endif
               /* UVM check for the address (unless MD indicated to not). */
               if (!checked && !uvm_kernacc(addr, len, prot)) {
                       return EFAULT;
               }
       }
       error = uiomove(addr, len, uio);
       return error;
}

/*
* dev_zero_readwrite: helper for DEV_ZERO (/dev/null) case of R/W.
*/
static inline int
dev_zero_readwrite(struct uio *uio, struct iovec *iov)
{
       size_t len;

       /* Nothing to do for the write case. */
       if (uio->uio_rw == UIO_WRITE) {
               uio->uio_resid = 0;
               return 0;
       }
       /*
        * Read in page-by-page basis, caller will continue.
        * Cut appropriately for a single/last-iteration cases.
        */
       len = MIN(iov->iov_len, PAGE_SIZE);
       return uiomove(dev_zero_page, len, uio);
}

/*
* mm_readwrite: general memory R/W function.
*/
static int
mm_readwrite(dev_t dev, struct uio *uio, int flags)
{
       struct iovec *iov;
       int error;

#ifdef __HAVE_MM_MD_READWRITE
       /* If defined - there are extra MD cases. */
       switch (minor(dev)) {
       case DEV_MEM:
       case DEV_KMEM:
       case DEV_NULL:
       case DEV_ZERO:
#if defined(COMPAT_16) && defined(__arm)
       case _DEV_ZERO_oARM:
#endif
               break;
       default:
               return mm_md_readwrite(dev, uio);
       }
#endif
       error = 0;
       while (uio->uio_resid > 0 && error == 0) {
               iov = uio->uio_iov;
               if (iov->iov_len == 0) {
                       /* Processed; next I/O vector. */
                       uio->uio_iov++;
                       uio->uio_iovcnt--;
                       KASSERT(uio->uio_iovcnt >= 0);
                       continue;
               }
               /* Helper functions will process in page-by-page basis. */
               switch (minor(dev)) {
               case DEV_MEM:
                       error = dev_mem_readwrite(uio, iov);
                       break;
               case DEV_KMEM:
                       error = dev_kmem_readwrite(uio, iov);
                       break;
               case DEV_NULL:
                       if (uio->uio_rw == UIO_WRITE) {
                               uio->uio_resid = 0;
                       }
                       /* Break directly out of the loop. */
                       return 0;
               case DEV_FULL:
                       if (uio->uio_rw == UIO_WRITE) {
                               return ENOSPC;
                       }
#if defined(COMPAT_16) && defined(__arm)
                       /* FALLTHROUGH */
               case _DEV_ZERO_oARM:
#endif
               /* FALLTHROUGH */
               case DEV_ZERO:
                       error = dev_zero_readwrite(uio, iov);
                       break;
               default:
                       error = ENXIO;
                       break;
               }
       }
       return error;
}

/*
* mm_mmap: general mmap() handler.
*/
static paddr_t
mm_mmap(dev_t dev, off_t off, int acc)
{
       vm_prot_t prot;

#ifdef __HAVE_MM_MD_MMAP
       /* If defined - there are extra mmap() MD cases. */
       switch (minor(dev)) {
       case DEV_MEM:
       case DEV_KMEM:
       case DEV_NULL:
#if defined(COMPAT_16) && defined(__arm)
       case _DEV_ZERO_oARM:
#endif
       case DEV_ZERO:
               break;
       default:
               return mm_md_mmap(dev, off, acc);
       }
#endif
       /*
        * /dev/null does not make sense, /dev/kmem is volatile and
        * /dev/zero is handled in mmap already.
        */
       if (minor(dev) != DEV_MEM) {
               return -1;
       }

       prot = 0;
       if (acc & PROT_EXEC)
               prot |= VM_PROT_EXECUTE;
       if (acc & PROT_READ)
               prot |= VM_PROT_READ;
       if (acc & PROT_WRITE)
               prot |= VM_PROT_WRITE;

       /* Validate the physical address. */
       if (mm_md_physacc(off, prot) != 0) {
               return -1;
       }
       return off >> PGSHIFT;
}

static int
mm_ioctl(dev_t dev, u_long cmd, void *data, int flag, struct lwp *l)
{

       switch (cmd) {
       case FIONBIO:
               /* We never block anyway. */
               return 0;

       case FIOSETOWN:
       case FIOGETOWN:
       case TIOCGPGRP:
       case TIOCSPGRP:
       case TIOCGETA:
               return ENOTTY;

       case FIOASYNC:
               if ((*(int *)data) == 0) {
                       return 0;
               }
               /* FALLTHROUGH */
       default:
               return EOPNOTSUPP;
       }
}