/*      $NetBSD: exec_subr.c,v 1.91 2024/12/06 16:48:13 riastradh Exp $ */

/*
* Copyright (c) 1993, 1994, 1996 Christopher G. Demetriou
* 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.
* 3. All advertising materials mentioning features or use of this software
*    must display the following acknowledgement:
*      This product includes software developed by Christopher G. Demetriou.
* 4. The name of the author may not be used to endorse or promote products
*    derived from this software without specific prior written permission
*
* 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.
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: exec_subr.c,v 1.91 2024/12/06 16:48:13 riastradh Exp $");

#include "opt_pax.h"

#include <sys/param.h>
#include <sys/types.h>

#include <sys/device.h>
#include <sys/exec.h>
#include <sys/filedesc.h>
#include <sys/kmem.h>
#include <sys/mman.h>
#include <sys/pax.h>
#include <sys/proc.h>
#include <sys/resourcevar.h>
#include <sys/sdt.h>
#include <sys/systm.h>
#include <sys/vnode.h>

#include <uvm/uvm_extern.h>

#define VMCMD_EVCNT_DECL(name)                                  \
static struct evcnt vmcmd_ev_##name =                           \
   EVCNT_INITIALIZER(EVCNT_TYPE_MISC, NULL, "vmcmd", #name);   \
EVCNT_ATTACH_STATIC(vmcmd_ev_##name)

#define VMCMD_EVCNT_INCR(name)                                  \
   vmcmd_ev_##name.ev_count++

VMCMD_EVCNT_DECL(calls);
VMCMD_EVCNT_DECL(extends);
VMCMD_EVCNT_DECL(kills);

#ifdef DEBUG_STACK
#define DPRINTF(a) uprintf a
#else
#define DPRINTF(a)
#endif

unsigned int user_stack_guard_size = 1024 * 1024;
unsigned int user_thread_stack_guard_size = 64 * 1024;

/*
* new_vmcmd():
*      create a new vmcmd structure and fill in its fields based
*      on function call arguments.  make sure objects ref'd by
*      the vmcmd are 'held'.
*/

void
new_vmcmd(struct exec_vmcmd_set *evsp,
   int (*proc)(struct lwp * l, struct exec_vmcmd *),
   vsize_t len, vaddr_t addr, struct vnode *vp, u_long offset,
   u_int prot, int flags)
{
       struct exec_vmcmd *vcp;

       VMCMD_EVCNT_INCR(calls);
       KASSERT(proc != vmcmd_map_pagedvn || (vp->v_iflag & VI_TEXT));
       KASSERT(vp == NULL || vrefcnt(vp) > 0);

       if (evsp->evs_used >= evsp->evs_cnt)
               vmcmdset_extend(evsp);
       vcp = &evsp->evs_cmds[evsp->evs_used++];
       vcp->ev_proc = proc;
       vcp->ev_len = len;
       vcp->ev_addr = addr;
       if ((vcp->ev_vp = vp) != NULL)
               vref(vp);
       vcp->ev_offset = offset;
       vcp->ev_prot = prot;
       vcp->ev_flags = flags;
}

void
vmcmdset_extend(struct exec_vmcmd_set *evsp)
{
       struct exec_vmcmd *nvcp;
       u_int ocnt;

#ifdef DIAGNOSTIC
       if (evsp->evs_used < evsp->evs_cnt)
               panic("vmcmdset_extend: not necessary");
#endif

       /* figure out number of entries in new set */
       if ((ocnt = evsp->evs_cnt) != 0) {
               evsp->evs_cnt += ocnt;
               VMCMD_EVCNT_INCR(extends);
       } else
               evsp->evs_cnt = EXEC_DEFAULT_VMCMD_SETSIZE;

       /* allocate it */
       nvcp = kmem_alloc(evsp->evs_cnt * sizeof(struct exec_vmcmd), KM_SLEEP);

       /* free the old struct, if there was one, and record the new one */
       if (ocnt) {
               memcpy(nvcp, evsp->evs_cmds,
                   (ocnt * sizeof(struct exec_vmcmd)));
               kmem_free(evsp->evs_cmds, ocnt * sizeof(struct exec_vmcmd));
       }
       evsp->evs_cmds = nvcp;
}

void
kill_vmcmds(struct exec_vmcmd_set *evsp)
{
       struct exec_vmcmd *vcp;
       u_int i;

       VMCMD_EVCNT_INCR(kills);

       if (evsp->evs_cnt == 0)
               return;

       for (i = 0; i < evsp->evs_used; i++) {
               vcp = &evsp->evs_cmds[i];
               if (vcp->ev_vp != NULL)
                       vrele(vcp->ev_vp);
       }
       kmem_free(evsp->evs_cmds, evsp->evs_cnt * sizeof(struct exec_vmcmd));
       evsp->evs_used = evsp->evs_cnt = 0;
}

/*
* vmcmd_map_pagedvn():
*      handle vmcmd which specifies that a vnode should be mmap'd.
*      appropriate for handling demand-paged text and data segments.
*/

static int
vmcmd_get_prot(struct lwp *l, const struct exec_vmcmd *cmd, vm_prot_t *prot,
   vm_prot_t *maxprot)
{
       vm_prot_t extraprot = PROT_MPROTECT_EXTRACT(cmd->ev_prot);

       *prot = cmd->ev_prot & UVM_PROT_ALL;
       *maxprot = PAX_MPROTECT_MAXPROTECT(l, *prot, extraprot, UVM_PROT_ALL);

       if ((*prot & *maxprot) != *prot)
               return SET_ERROR(EACCES);
       return PAX_MPROTECT_VALIDATE(l, *prot);
}

int
vmcmd_map_pagedvn(struct lwp *l, struct exec_vmcmd *cmd)
{
       struct uvm_object *uobj;
       struct vnode *vp = cmd->ev_vp;
       struct proc *p = l->l_proc;
       int error;
       vm_prot_t prot, maxprot;

       KASSERT(vp->v_iflag & VI_TEXT);

       /*
        * map the vnode in using uvm_map.
        */

       if (cmd->ev_len == 0)
               return 0;
       if (cmd->ev_offset & PAGE_MASK)
               return SET_ERROR(EINVAL);
       if (cmd->ev_addr & PAGE_MASK)
               return SET_ERROR(EINVAL);
       if (cmd->ev_len & PAGE_MASK)
               return SET_ERROR(EINVAL);

       if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
               return error;

       /*
        * check the file system's opinion about mmapping the file
        */

       error = VOP_MMAP(vp, prot, l->l_cred);
       if (error)
               return error;

       if ((vp->v_vflag & VV_MAPPED) == 0) {
               vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
               vp->v_vflag |= VV_MAPPED;
               VOP_UNLOCK(vp);
       }

       /*
        * do the map, reference the object for this map entry
        */
       uobj = &vp->v_uobj;
       vref(vp);

       error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr, cmd->ev_len,
               uobj, cmd->ev_offset, 0,
               UVM_MAPFLAG(prot, maxprot, UVM_INH_COPY,
                       UVM_ADV_NORMAL, UVM_FLAG_COPYONW|UVM_FLAG_FIXED));
       if (error) {
               uobj->pgops->pgo_detach(uobj);
       }
       return error;
}

/*
* vmcmd_map_readvn():
*      handle vmcmd which specifies that a vnode should be read from.
*      appropriate for non-demand-paged text/data segments, i.e. impure
*      objects (a la OMAGIC and NMAGIC).
*/
int
vmcmd_map_readvn(struct lwp *l, struct exec_vmcmd *cmd)
{
       struct proc *p = l->l_proc;
       int error;
       long diff;

       if (cmd->ev_len == 0)
               return 0;

       diff = cmd->ev_addr - trunc_page(cmd->ev_addr);
       cmd->ev_addr -= diff;                   /* required by uvm_map */
       cmd->ev_offset -= diff;
       cmd->ev_len += diff;

       error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr,
                       round_page(cmd->ev_len), NULL, UVM_UNKNOWN_OFFSET, 0,
                       UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
                       UVM_ADV_NORMAL,
                       UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW));

       if (error)
               return error;

       return vmcmd_readvn(l, cmd);
}

int
vmcmd_readvn(struct lwp *l, struct exec_vmcmd *cmd)
{
       struct proc *p = l->l_proc;
       int error;
       vm_prot_t prot, maxprot;

       error = vn_rdwr(UIO_READ, cmd->ev_vp, (void *)cmd->ev_addr,
           cmd->ev_len, cmd->ev_offset, UIO_USERSPACE, IO_UNIT,
           l->l_cred, NULL, l);
       if (error)
               return error;

       if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
               return error;

#ifdef PMAP_NEED_PROCWR
       /*
        * we had to write the process, make sure the pages are synched
        * with the instruction cache.
        */
       if (prot & VM_PROT_EXECUTE)
               pmap_procwr(p, cmd->ev_addr, cmd->ev_len);
#endif

       /*
        * we had to map in the area at PROT_ALL so that vn_rdwr()
        * could write to it.   however, the caller seems to want
        * it mapped read-only, so now we are going to have to call
        * uvm_map_protect() to fix up the protection.  ICK.
        */
       if (maxprot != VM_PROT_ALL) {
               error = uvm_map_protect(&p->p_vmspace->vm_map,
                               trunc_page(cmd->ev_addr),
                               round_page(cmd->ev_addr + cmd->ev_len),
                               maxprot, true);
               if (error)
                       return error;
       }

       if (prot != maxprot) {
               error = uvm_map_protect(&p->p_vmspace->vm_map,
                               trunc_page(cmd->ev_addr),
                               round_page(cmd->ev_addr + cmd->ev_len),
                               prot, false);
               if (error)
                       return error;
       }

       return 0;
}

/*
* vmcmd_map_zero():
*      handle vmcmd which specifies a zero-filled address space region.  The
*      address range must be first allocated, then protected appropriately.
*/

int
vmcmd_map_zero(struct lwp *l, struct exec_vmcmd *cmd)
{
       struct proc *p = l->l_proc;
       int error;
       long diff;
       vm_prot_t prot, maxprot;

       diff = cmd->ev_addr - trunc_page(cmd->ev_addr);
       cmd->ev_addr -= diff;                   /* required by uvm_map */
       cmd->ev_len += diff;

       if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
               return error;

       error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr,
                       round_page(cmd->ev_len), NULL, UVM_UNKNOWN_OFFSET, 0,
                       UVM_MAPFLAG(prot, maxprot, UVM_INH_COPY,
                       UVM_ADV_NORMAL,
                       UVM_FLAG_FIXED|UVM_FLAG_COPYONW));
       if (cmd->ev_flags & VMCMD_STACK)
               curproc->p_vmspace->vm_issize += atop(round_page(cmd->ev_len));
       return error;
}

/*
* exec_read():
*
*      Read from vnode into buffer at offset.
*/
int
exec_read(struct lwp *l, struct vnode *vp, u_long off, void *bf, size_t size,
   int ioflg)
{
       int error;
       size_t resid;

       KASSERT((ioflg & IO_NODELOCKED) == 0 || VOP_ISLOCKED(vp) != LK_NONE);

       if ((error = vn_rdwr(UIO_READ, vp, bf, size, off, UIO_SYSSPACE,
           ioflg, l->l_cred, &resid, NULL)) != 0)
               return error;
       /*
        * See if we got all of it
        */
       if (resid != 0)
               return SET_ERROR(ENOEXEC);
       return 0;
}

/*
* exec_setup_stack(): Set up the stack segment for an elf
* executable.
*
* Note that the ep_ssize parameter must be set to be the current stack
* limit; this is adjusted in the body of execve() to yield the
* appropriate stack segment usage once the argument length is
* calculated.
*
* This function returns an int for uniformity with other (future) formats'
* stack setup functions.  They might have errors to return.
*/

int
exec_setup_stack(struct lwp *l, struct exec_package *epp)
{
       vsize_t max_stack_size;
       vaddr_t access_linear_min;
       vsize_t access_size;
       vaddr_t noaccess_linear_min;
       vsize_t noaccess_size;

#ifndef USRSTACK32
#define USRSTACK32      (0x00000000ffffffffL&~PGOFSET)
#endif
#ifndef MAXSSIZ32
#define MAXSSIZ32       (MAXSSIZ >> 2)
#endif

       if (epp->ep_flags & EXEC_32) {
               epp->ep_minsaddr = USRSTACK32;
               max_stack_size = MAXSSIZ32;
       } else {
               epp->ep_minsaddr = USRSTACK;
               max_stack_size = MAXSSIZ;
       }

       DPRINTF(("ep_minsaddr=%#jx max_stack_size=%#jx\n",
           (uintmax_t)epp->ep_minsaddr, (uintmax_t)max_stack_size));

       pax_aslr_stack(epp, &max_stack_size);

       DPRINTF(("[RLIMIT_STACK].lim_cur=%#jx max_stack_size=%#jx\n",
           (uintmax_t)l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur,
           (uintmax_t)max_stack_size));
       epp->ep_ssize = MIN(l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur,
           max_stack_size);

       l->l_proc->p_stackbase = epp->ep_minsaddr;

       epp->ep_maxsaddr = (vaddr_t)STACK_GROW(epp->ep_minsaddr,
           max_stack_size);

       DPRINTF(("ep_ssize=%#jx ep_minsaddr=%#jx ep_maxsaddr=%#jx\n",
           (uintmax_t)epp->ep_ssize, (uintmax_t)epp->ep_minsaddr,
           (uintmax_t)epp->ep_maxsaddr));

       /*
        * set up commands for stack.  note that this takes *two*, one to
        * map the part of the stack which we can access, and one to map
        * the part which we can't.
        *
        * arguably, it could be made into one, but that would require the
        * addition of another mapping proc, which is unnecessary
        */
       access_size = epp->ep_ssize;
       access_linear_min = (vaddr_t)STACK_ALLOC(epp->ep_minsaddr, access_size);
       noaccess_size = max_stack_size - access_size;
       noaccess_linear_min = (vaddr_t)STACK_ALLOC(STACK_GROW(epp->ep_minsaddr,
           access_size), noaccess_size);

       DPRINTF(("access_size=%#jx, access_linear_min=%#jx, "
           "noaccess_size=%#jx, noaccess_linear_min=%#jx\n",
           (uintmax_t)access_size, (uintmax_t)access_linear_min,
           (uintmax_t)noaccess_size, (uintmax_t)noaccess_linear_min));

       if (user_stack_guard_size > 0) {
#ifdef __MACHINE_STACK_GROWS_UP
               vsize_t guard_size = MIN(VM_MAXUSER_ADDRESS - epp->ep_maxsaddr, user_stack_guard_size);
               if (guard_size > 0)
                       NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, guard_size,
                           epp->ep_maxsaddr, NULL, 0, VM_PROT_NONE);
#else
               NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, user_stack_guard_size,
                   epp->ep_maxsaddr - user_stack_guard_size, NULL, 0, VM_PROT_NONE);
#endif
       }
       if (noaccess_size > 0 && noaccess_size <= MAXSSIZ) {
               NEW_VMCMD2(&epp->ep_vmcmds, vmcmd_map_zero, noaccess_size,
                   noaccess_linear_min, NULL, 0,
                   VM_PROT_NONE | PROT_MPROTECT(VM_PROT_READ | VM_PROT_WRITE),
                   VMCMD_STACK);
       }
       KASSERT(access_size > 0);
       KASSERT(access_size <= MAXSSIZ);
       NEW_VMCMD2(&epp->ep_vmcmds, vmcmd_map_zero, access_size,
           access_linear_min, NULL, 0, VM_PROT_READ | VM_PROT_WRITE,
           VMCMD_STACK);

       return 0;
}