/*      $NetBSD: uvm_device.c,v 1.80 2022/07/07 13:27:02 riastradh Exp $        */

/*
* Copyright (c) 1997 Charles D. Cranor and Washington University.
* All rights reserved.
*
* 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 AUTHOR ``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 AUTHOR 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.
*
* from: Id: uvm_device.c,v 1.1.2.9 1998/02/06 05:11:47 chs Exp
*/

/*
* uvm_device.c: the device pager.
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: uvm_device.c,v 1.80 2022/07/07 13:27:02 riastradh Exp $");

#include "opt_uvmhist.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/conf.h>
#include <sys/proc.h>
#include <sys/kmem.h>

#include <uvm/uvm.h>
#include <uvm/uvm_device.h>
#include <uvm/uvm_pmap.h>

/*
* private global data structure
*
* we keep a list of active device objects in the system.
*/

LIST_HEAD(udv_list_struct, uvm_device);
static struct udv_list_struct udv_list;
static kmutex_t udv_lock __cacheline_aligned;

/*
* functions
*/

static void     udv_init(void);
static void     udv_reference(struct uvm_object *);
static void     udv_detach(struct uvm_object *);
static int      udv_fault(struct uvm_faultinfo *, vaddr_t,
                         struct vm_page **, int, int, vm_prot_t,
                         int);

/*
* master pager structure
*/

const struct uvm_pagerops uvm_deviceops = {
       .pgo_init = udv_init,
       .pgo_reference = udv_reference,
       .pgo_detach = udv_detach,
       .pgo_fault = udv_fault,
};

/*
* the ops!
*/

/*
* udv_init
*
* init pager private data structures.
*/

static void
udv_init(void)
{
       LIST_INIT(&udv_list);
       mutex_init(&udv_lock, MUTEX_DEFAULT, IPL_NONE);
}

/*
* udv_attach
*
* get a VM object that is associated with a device.   allocate a new
* one if needed.
*
* => caller must _not_ already be holding the lock on the uvm_object.
* => in fact, nothing should be locked so that we can sleep here.
*/

struct uvm_object *
udv_attach(dev_t device, vm_prot_t accessprot,
   voff_t off,         /* used only for access check */
   vsize_t size        /* used only for access check */)
{
       struct uvm_device *udv, *lcv;
       const struct cdevsw *cdev;
       dev_mmap_t *mapfn;

       UVMHIST_FUNC(__func__);
       UVMHIST_CALLARGS(maphist, "(device=%#jx)", device,0,0,0);

       KASSERT(size > 0);

       /*
        * before we do anything, ensure this device supports mmap
        */

       cdev = cdevsw_lookup(device);
       if (cdev == NULL) {
               return NULL;
       }
       mapfn = cdev->d_mmap;
       if (mapfn == NULL || mapfn == nommap) {
               return NULL;
       }

       /*
        * Negative offsets on the object are not allowed, unless the
        * device has affirmatively set D_NEGOFFSAFE.
        */
       if ((cdev->d_flag & D_NEGOFFSAFE) == 0 && off != UVM_UNKNOWN_OFFSET) {
               if (off < 0)
                       return NULL;
#if SIZE_MAX > UINT32_MAX       /* XXX -Wtype-limits */
               if (size > __type_max(voff_t))
                       return NULL;
#endif
               if (off > __type_max(voff_t) - size)
                       return NULL;
       }

       /*
        * Check that the specified range of the device allows the
        * desired protection.
        *
        * XXX assumes VM_PROT_* == PROT_*
        * XXX clobbers off and size, but nothing else here needs them.
        */
       do {
               KASSERTMSG((off % PAGE_SIZE) == 0, "off=%jd", (intmax_t)off);
               KASSERTMSG(size >= PAGE_SIZE, "size=%"PRIuVSIZE, size);
               if (cdev_mmap(device, off, accessprot) == -1)
                       return NULL;
               KASSERT(off <= __type_max(voff_t) - PAGE_SIZE ||
                   (cdev->d_flag & D_NEGOFFSAFE) != 0);
               if (__predict_false(off > __type_max(voff_t) - PAGE_SIZE)) {
                       /*
                        * off += PAGE_SIZE, with two's-complement
                        * wraparound, or
                        *
                        *      off += PAGE_SIZE - 2*(VOFF_MAX + 1).
                        */
                       CTASSERT(MIN_PAGE_SIZE >= 2);
                       off -= __type_max(voff_t);
                       off += PAGE_SIZE - 2;
                       off -= __type_max(voff_t);
               } else {
                       off += PAGE_SIZE;
               }
               size -= PAGE_SIZE;
       } while (size != 0);

       /*
        * keep looping until we get it
        */

       for (;;) {

               /*
                * first, attempt to find it on the main list
                */

               mutex_enter(&udv_lock);
               LIST_FOREACH(lcv, &udv_list, u_list) {
                       if (device == lcv->u_device)
                               break;
               }

               /*
                * got it on main list.  put a hold on it and unlock udv_lock.
                */

               if (lcv) {

                       /*
                        * if someone else has a hold on it, sleep and start
                        * over again.
                        */

                       if (lcv->u_flags & UVM_DEVICE_HOLD) {
                               lcv->u_flags |= UVM_DEVICE_WANTED;
                               UVM_UNLOCK_AND_WAIT(lcv, &udv_lock, false,
                                   "udv_attach",0);
                               continue;
                       }

                       /* we are now holding it */
                       lcv->u_flags |= UVM_DEVICE_HOLD;
                       mutex_exit(&udv_lock);

                       /*
                        * bump reference count, unhold, return.
                        */

                       rw_enter(lcv->u_obj.vmobjlock, RW_WRITER);
                       lcv->u_obj.uo_refs++;
                       rw_exit(lcv->u_obj.vmobjlock);

                       mutex_enter(&udv_lock);
                       if (lcv->u_flags & UVM_DEVICE_WANTED)
                               wakeup(lcv);
                       lcv->u_flags &= ~(UVM_DEVICE_WANTED|UVM_DEVICE_HOLD);
                       mutex_exit(&udv_lock);
                       return &lcv->u_obj;
               }

               /*
                * Did not find it on main list.  Need to allocate a new one.
                */

               mutex_exit(&udv_lock);

               /* Note: both calls may allocate memory and sleep. */
               udv = kmem_alloc(sizeof(*udv), KM_SLEEP);
               uvm_obj_init(&udv->u_obj, &uvm_deviceops, true, 1);

               mutex_enter(&udv_lock);

               /*
                * now we have to double check to make sure no one added it
                * to the list while we were sleeping...
                */

               LIST_FOREACH(lcv, &udv_list, u_list) {
                       if (device == lcv->u_device)
                               break;
               }

               /*
                * did we lose a race to someone else?
                * free our memory and retry.
                */

               if (lcv) {
                       mutex_exit(&udv_lock);
                       uvm_obj_destroy(&udv->u_obj, true);
                       kmem_free(udv, sizeof(*udv));
                       continue;
               }

               /*
                * we have it!   init the data structures, add to list
                * and return.
                */

               udv->u_flags = 0;
               udv->u_device = device;
               LIST_INSERT_HEAD(&udv_list, udv, u_list);
               mutex_exit(&udv_lock);
               return &udv->u_obj;
       }
       /*NOTREACHED*/
}

/*
* udv_reference
*
* add a reference to a VM object.   Note that the reference count must
* already be one (the passed in reference) so there is no chance of the
* udv being released or locked out here.
*
* => caller must call with object unlocked.
*/

static void
udv_reference(struct uvm_object *uobj)
{
       UVMHIST_FUNC(__func__); UVMHIST_CALLED(maphist);

       rw_enter(uobj->vmobjlock, RW_WRITER);
       uobj->uo_refs++;
       UVMHIST_LOG(maphist, "<- done (uobj=%#jx, ref = %jd)",
           (uintptr_t)uobj, uobj->uo_refs,0,0);
       rw_exit(uobj->vmobjlock);
}

/*
* udv_detach
*
* remove a reference to a VM object.
*
* => caller must call with object unlocked and map locked.
*/

static void
udv_detach(struct uvm_object *uobj)
{
       struct uvm_device *udv = (struct uvm_device *)uobj;
       UVMHIST_FUNC(__func__); UVMHIST_CALLED(maphist);

       /*
        * loop until done
        */
again:
       rw_enter(uobj->vmobjlock, RW_WRITER);
       if (uobj->uo_refs > 1) {
               uobj->uo_refs--;
               rw_exit(uobj->vmobjlock);
               UVMHIST_LOG(maphist," <- done, uobj=%#jx, ref=%jd",
                   (uintptr_t)uobj,uobj->uo_refs,0,0);
               return;
       }

       /*
        * is it being held?   if so, wait until others are done.
        */

       mutex_enter(&udv_lock);
       if (udv->u_flags & UVM_DEVICE_HOLD) {
               udv->u_flags |= UVM_DEVICE_WANTED;
               rw_exit(uobj->vmobjlock);
               UVM_UNLOCK_AND_WAIT(udv, &udv_lock, false, "udv_detach",0);
               goto again;
       }

       /*
        * got it!   nuke it now.
        */

       LIST_REMOVE(udv, u_list);
       if (udv->u_flags & UVM_DEVICE_WANTED)
               wakeup(udv);
       mutex_exit(&udv_lock);
       rw_exit(uobj->vmobjlock);

       uvm_obj_destroy(uobj, true);
       kmem_free(udv, sizeof(*udv));
       UVMHIST_LOG(maphist," <- done, freed uobj=%#jx", (uintptr_t)uobj,
           0, 0, 0);
}

/*
* udv_fault: non-standard fault routine for device "pages"
*
* => rather than having a "get" function, we have a fault routine
*      since we don't return vm_pages we need full control over the
*      pmap_enter map in
* => all the usual fault data structured are locked by the caller
*      (i.e. maps(read), amap (if any), uobj)
* => on return, we unlock all fault data structures
* => flags: PGO_ALLPAGES: get all of the pages
*           PGO_LOCKED: fault data structures are locked
*    XXX: currently PGO_LOCKED is always required ... consider removing
*      it as a flag
* => NOTE: vaddr is the VA of pps[0] in ufi->entry, _NOT_ pps[centeridx]
*/

static int
udv_fault(struct uvm_faultinfo *ufi, vaddr_t vaddr, struct vm_page **pps,
   int npages, int centeridx, vm_prot_t access_type,
   int flags)
{
       struct vm_map_entry *entry = ufi->entry;
       struct uvm_object *uobj = entry->object.uvm_obj;
       struct uvm_device *udv = (struct uvm_device *)uobj;
       vaddr_t curr_va;
       off_t curr_offset;
       paddr_t paddr, mdpgno;
       u_int mmapflags;
       int lcv, retval;
       dev_t device;
       vm_prot_t mapprot;
       UVMHIST_FUNC(__func__); UVMHIST_CALLED(maphist);
       UVMHIST_LOG(maphist,"  flags=%#jx", flags,0,0,0);

       /*
        * we do not allow device mappings to be mapped copy-on-write
        * so we kill any attempt to do so here.
        */

       if (UVM_ET_ISCOPYONWRITE(entry)) {
               UVMHIST_LOG(maphist, "<- failed -- COW entry (etype=%#jx)",
                   entry->etype, 0,0,0);
               uvmfault_unlockall(ufi, ufi->entry->aref.ar_amap, uobj);
               return EIO;
       }

       /*
        * get device map function.
        */

       device = udv->u_device;
       if (cdevsw_lookup(device) == NULL) {
               /* XXX This should not happen */
               uvmfault_unlockall(ufi, ufi->entry->aref.ar_amap, uobj);
               return EIO;
       }

       /*
        * now we must determine the offset in udv to use and the VA to
        * use for pmap_enter.  note that we always use orig_map's pmap
        * for pmap_enter (even if we have a submap).   since virtual
        * addresses in a submap must match the main map, this is ok.
        */

       /* udv offset = (offset from start of entry) + entry's offset */
       curr_offset = entry->offset + (vaddr - entry->start);
       /* pmap va = vaddr (virtual address of pps[0]) */
       curr_va = vaddr;

       /*
        * loop over the page range entering in as needed
        */

       retval = 0;
       for (lcv = 0 ; lcv < npages ; lcv++, curr_offset += PAGE_SIZE,
           curr_va += PAGE_SIZE) {
               if ((flags & PGO_ALLPAGES) == 0 && lcv != centeridx)
                       continue;

               if (pps[lcv] == PGO_DONTCARE)
                       continue;

               mdpgno = cdev_mmap(device, curr_offset, access_type);
               if (mdpgno == -1) {
                       retval = EIO;
                       break;
               }
               paddr = pmap_phys_address(mdpgno);
               mmapflags = pmap_mmap_flags(mdpgno);
               mapprot = ufi->entry->protection;
               UVMHIST_LOG(maphist,
                   "  MAPPING: device: pm=%#jx, va=%#jx, pa=%#jx, at=%jd",
                   (uintptr_t)ufi->orig_map->pmap, curr_va, paddr, mapprot);
               if (pmap_enter(ufi->orig_map->pmap, curr_va, paddr, mapprot,
                   PMAP_CANFAIL | mapprot | mmapflags) != 0) {
                       /*
                        * pmap_enter() didn't have the resource to
                        * enter this mapping.  Unlock everything,
                        * wait for the pagedaemon to free up some
                        * pages, and then tell uvm_fault() to start
                        * the fault again.
                        *
                        * XXX Needs some rethinking for the PGO_ALLPAGES
                        * XXX case.
                        */
                       pmap_update(ufi->orig_map->pmap);       /* sync what we have so far */
                       uvmfault_unlockall(ufi, ufi->entry->aref.ar_amap,
                           uobj);
                       return ENOMEM;
               }
       }

       pmap_update(ufi->orig_map->pmap);
       uvmfault_unlockall(ufi, ufi->entry->aref.ar_amap, uobj);
       return retval;
}