731 lines
21 KiB
C
731 lines
21 KiB
C
/*
|
|
* Copyright (c) 1994 Scott Bartram
|
|
* Copyright (c) 1994 Adam Glass
|
|
* Copyright (c) 1993, 1994 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 Scott Bartram.
|
|
* 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.
|
|
*
|
|
* originally from kern/exec_ecoff.c
|
|
*/
|
|
|
|
#include <sys/param.h>
|
|
#include <sys/systm.h>
|
|
#include <sys/proc.h>
|
|
#include <sys/malloc.h>
|
|
#include <sys/vnode.h>
|
|
#include <sys/exec.h>
|
|
#include <sys/resourcevar.h>
|
|
#include <sys/namei.h>
|
|
#include <vm/vm.h>
|
|
|
|
#include <compat/ibcs2/ibcs2_exec.h>
|
|
|
|
#ifdef DEBUG_IBCS2
|
|
#define DPRINTF(s) printf s
|
|
#else
|
|
#define DPRINTF(s)
|
|
#endif
|
|
|
|
int exec_ibcs2_coff_prep_omagic __P((struct proc *, struct exec_package *,
|
|
struct coff_filehdr *,
|
|
struct coff_aouthdr *));
|
|
int exec_ibcs2_coff_prep_nmagic __P((struct proc *, struct exec_package *,
|
|
struct coff_filehdr *,
|
|
struct coff_aouthdr *));
|
|
int exec_ibcs2_coff_prep_zmagic __P((struct proc *, struct exec_package *,
|
|
struct coff_filehdr *,
|
|
struct coff_aouthdr *));
|
|
int exec_ibcs2_coff_setup_stack __P((struct proc *, struct exec_package *));
|
|
void cpu_exec_ibcs2_coff_setup __P((struct proc *, struct exec_package *));
|
|
|
|
int exec_ibcs2_xout_prep_nmagic __P((struct proc *, struct exec_package *,
|
|
struct xexec *, struct xext *));
|
|
int exec_ibcs2_xout_prep_zmagic __P((struct proc *, struct exec_package *,
|
|
struct xexec *, struct xext *));
|
|
int exec_ibcs2_xout_setup_stack __P((struct proc *, struct exec_package *));
|
|
void cpu_exec_ibcs2_xout_setup __P((struct proc *, struct exec_package *));
|
|
|
|
/*
|
|
* exec_ibcs2_coff_makecmds(): Check if it's an coff-format executable.
|
|
*
|
|
* Given a proc pointer and an exec package pointer, see if the referent
|
|
* of the epp is in coff format. Check 'standard' magic numbers for
|
|
* this architecture. If that fails, return failure.
|
|
*
|
|
* This function is responsible for creating a set of vmcmds which can be
|
|
* used to build the process's vm space and inserting them into the exec
|
|
* package.
|
|
*/
|
|
|
|
int
|
|
exec_ibcs2_coff_makecmds(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
u_long midmag, magic;
|
|
u_short mid;
|
|
int error;
|
|
struct coff_filehdr *fp = epp->ep_hdr;
|
|
struct coff_aouthdr *ap;
|
|
|
|
if (epp->ep_hdrvalid < COFF_HDR_SIZE)
|
|
return ENOEXEC;
|
|
|
|
if (COFF_BADMAG(fp))
|
|
return ENOEXEC;
|
|
|
|
ap = epp->ep_hdr + sizeof(struct coff_filehdr);
|
|
switch (ap->a_magic) {
|
|
case COFF_OMAGIC:
|
|
error = exec_ibcs2_coff_prep_omagic(p, epp, fp, ap);
|
|
break;
|
|
case COFF_NMAGIC:
|
|
error = exec_ibcs2_coff_prep_nmagic(p, epp, fp, ap);
|
|
break;
|
|
case COFF_ZMAGIC:
|
|
error = exec_ibcs2_coff_prep_zmagic(p, epp, fp, ap);
|
|
break;
|
|
default:
|
|
return ENOEXEC;
|
|
}
|
|
|
|
if (error == 0) {
|
|
epp->ep_emul = EMUL_IBCS2_COFF;
|
|
epp->ep_setup = cpu_exec_ibcs2_coff_setup;
|
|
}
|
|
|
|
if (error)
|
|
kill_vmcmds(&epp->ep_vmcmds);
|
|
|
|
bad:
|
|
return error;
|
|
}
|
|
|
|
/*
|
|
* exec_ibcs2_coff_setup_stack(): Set up the stack segment for a coff
|
|
* 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_ibcs2_coff_setup_stack(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
/* DPRINTF(("enter exec_ibcs2_coff_setup_stack\n")); */
|
|
|
|
epp->ep_maxsaddr = USRSTACK - MAXSSIZ;
|
|
epp->ep_minsaddr = USRSTACK;
|
|
epp->ep_ssize = p->p_rlimit[RLIMIT_STACK].rlim_cur;
|
|
|
|
/*
|
|
* 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
|
|
*
|
|
* note that in memory, things assumed to be: 0 ....... ep_maxsaddr
|
|
* <stack> ep_minsaddr
|
|
*/
|
|
/* DPRINTF(("VMCMD: addr %x size %d\n", epp->ep_maxsaddr,
|
|
(epp->ep_minsaddr - epp->ep_ssize) - epp->ep_maxsaddr)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero,
|
|
((epp->ep_minsaddr - epp->ep_ssize) - epp->ep_maxsaddr),
|
|
epp->ep_maxsaddr, NULLVP, 0, VM_PROT_NONE);
|
|
/* DPRINTF(("VMCMD: addr %x size %d\n",
|
|
epp->ep_minsaddr - epp->ep_ssize,
|
|
epp->ep_ssize)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, epp->ep_ssize,
|
|
(epp->ep_minsaddr - epp->ep_ssize), NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
* exec_ibcs2_coff_prep_omagic(): Prepare a COFF OMAGIC binary's exec package
|
|
*/
|
|
|
|
int
|
|
exec_ibcs2_coff_prep_omagic(p, epp, fp, ap)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
struct coff_filehdr *fp;
|
|
struct coff_aouthdr *ap;
|
|
{
|
|
epp->ep_taddr = COFF_SEGMENT_ALIGN(ap, ap->a_tstart);
|
|
epp->ep_tsize = ap->a_tsize;
|
|
epp->ep_daddr = COFF_SEGMENT_ALIGN(ap, ap->a_dstart);
|
|
epp->ep_dsize = ap->a_dsize;
|
|
epp->ep_entry = ap->a_entry;
|
|
|
|
/* set up command for text and data segments */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn,
|
|
ap->a_tsize + ap->a_dsize, epp->ep_taddr, epp->ep_vp,
|
|
COFF_TXTOFF(fp, ap),
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
/* set up command for bss segment */
|
|
if (ap->a_bsize > 0)
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, ap->a_bsize,
|
|
COFF_SEGMENT_ALIGN(ap, ap->a_dstart + ap->a_dsize),
|
|
NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
return exec_ibcs2_coff_setup_stack(p, epp);
|
|
}
|
|
|
|
/*
|
|
* exec_ibcs2_coff_prep_nmagic(): Prepare a 'native' NMAGIC COFF binary's exec
|
|
* package.
|
|
*/
|
|
|
|
int
|
|
exec_ibcs2_coff_prep_nmagic(p, epp, fp, ap)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
struct coff_filehdr *fp;
|
|
struct coff_aouthdr *ap;
|
|
{
|
|
epp->ep_taddr = COFF_SEGMENT_ALIGN(ap, ap->a_tstart);
|
|
epp->ep_tsize = ap->a_tsize;
|
|
epp->ep_daddr = COFF_ROUND(ap->a_dstart, COFF_LDPGSZ);
|
|
epp->ep_dsize = ap->a_dsize;
|
|
epp->ep_entry = ap->a_entry;
|
|
|
|
/* set up command for text segment */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn, epp->ep_tsize,
|
|
epp->ep_taddr, epp->ep_vp, COFF_TXTOFF(fp, ap),
|
|
VM_PROT_READ|VM_PROT_EXECUTE);
|
|
|
|
/* set up command for data segment */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn, epp->ep_dsize,
|
|
epp->ep_daddr, epp->ep_vp, COFF_DATOFF(fp, ap),
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
/* set up command for bss segment */
|
|
if (ap->a_bsize > 0)
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, ap->a_bsize,
|
|
COFF_SEGMENT_ALIGN(ap, ap->a_dstart + ap->a_dsize),
|
|
NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
return exec_ibcs2_coff_setup_stack(p, epp);
|
|
}
|
|
|
|
/*
|
|
* coff_find_section - load specified section header
|
|
*
|
|
* TODO - optimize by reading all section headers in at once
|
|
*/
|
|
|
|
static int
|
|
coff_find_section(p, vp, fp, sh, s_type)
|
|
struct proc *p;
|
|
struct vnode *vp;
|
|
struct coff_filehdr *fp;
|
|
struct coff_scnhdr *sh;
|
|
int s_type;
|
|
{
|
|
int i, pos, resid, siz, error;
|
|
|
|
pos = COFF_HDR_SIZE;
|
|
for (i = 0; i < fp->f_nscns; i++, pos += sizeof(struct coff_scnhdr)) {
|
|
siz = sizeof(struct coff_scnhdr);
|
|
if (error = vn_rdwr(UIO_READ, vp, (caddr_t) sh,
|
|
siz, pos,
|
|
UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred,
|
|
&resid, p)) {
|
|
DPRINTF(("section hdr %d read error %d\n", i, error));
|
|
return error;
|
|
}
|
|
siz -= resid;
|
|
if (siz != sizeof(struct coff_scnhdr)) {
|
|
DPRINTF(("incomplete read: hdr %d ask=%d, rem=%d got %d\n",
|
|
s_type, sizeof(struct coff_scnhdr),
|
|
resid, siz));
|
|
return ENOEXEC;
|
|
}
|
|
/* DPRINTF(("found section: %x\n", sh->s_flags)); */
|
|
if (sh->s_flags == s_type)
|
|
return 0;
|
|
}
|
|
return ENOEXEC;
|
|
}
|
|
|
|
/*
|
|
* exec_ibcs2_coff_prep_zmagic(): Prepare a COFF ZMAGIC binary's exec package
|
|
*
|
|
* First, set the various offsets/lengths in the exec package.
|
|
*
|
|
* Then, mark the text image busy (so it can be demand paged) or error
|
|
* out if this is not possible. Finally, set up vmcmds for the
|
|
* text, data, bss, and stack segments.
|
|
*/
|
|
|
|
int
|
|
exec_ibcs2_coff_prep_zmagic(p, epp, fp, ap)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
struct coff_filehdr *fp;
|
|
struct coff_aouthdr *ap;
|
|
{
|
|
int error;
|
|
u_long offset;
|
|
long dsize, baddr, bsize;
|
|
struct coff_scnhdr sh;
|
|
|
|
/* DPRINTF(("enter exec_ibcs2_coff_prep_zmagic\n")); */
|
|
|
|
/* set up command for text segment */
|
|
error = coff_find_section(p, epp->ep_vp, fp, &sh, COFF_STYP_TEXT);
|
|
if (error) {
|
|
DPRINTF(("can't find text section: %d\n", error));
|
|
return error;
|
|
}
|
|
/* DPRINTF(("COFF text addr %x size %d offset %d\n", sh.s_vaddr,
|
|
sh.s_size, sh.s_scnptr)); */
|
|
epp->ep_taddr = COFF_ALIGN(sh.s_vaddr);
|
|
offset = sh.s_scnptr - (sh.s_vaddr - epp->ep_taddr);
|
|
epp->ep_tsize = sh.s_size + (sh.s_vaddr - epp->ep_taddr);
|
|
|
|
#ifdef notyet
|
|
/*
|
|
* check if vnode is in open for writing, because we want to
|
|
* demand-page out of it. if it is, don't do it, for various
|
|
* reasons
|
|
n */
|
|
if ((ap->a_tsize != 0 || ap->a_dsize != 0) &&
|
|
epp->ep_vp->v_writecount != 0) {
|
|
#ifdef DIAGNOSTIC
|
|
if (epp->ep_vp->v_flag & VTEXT)
|
|
panic("exec: a VTEXT vnode has writecount != 0\n");
|
|
#endif
|
|
return ETXTBSY;
|
|
}
|
|
epp->ep_vp->v_flag |= VTEXT;
|
|
#endif
|
|
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n", epp->ep_taddr,
|
|
epp->ep_tsize, offset)); */
|
|
#ifdef notyet
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_pagedvn, epp->ep_tsize,
|
|
epp->ep_taddr, epp->ep_vp, offset,
|
|
VM_PROT_READ|VM_PROT_EXECUTE);
|
|
#else
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn, epp->ep_tsize,
|
|
epp->ep_taddr, epp->ep_vp, offset,
|
|
VM_PROT_READ|VM_PROT_EXECUTE);
|
|
#endif
|
|
|
|
/* set up command for data segment */
|
|
error = coff_find_section(p, epp->ep_vp, fp, &sh, COFF_STYP_DATA);
|
|
if (error) {
|
|
DPRINTF(("can't find data section: %d\n", error));
|
|
return error;
|
|
}
|
|
/* DPRINTF(("COFF data addr %x size %d offset %d\n", sh.s_vaddr,
|
|
sh.s_size, sh.s_scnptr)); */
|
|
epp->ep_daddr = COFF_ALIGN(sh.s_vaddr);
|
|
offset = sh.s_scnptr - (sh.s_vaddr - epp->ep_daddr);
|
|
dsize = sh.s_size + (sh.s_vaddr - epp->ep_daddr);
|
|
epp->ep_dsize = dsize + ap->a_bsize;
|
|
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n", epp->ep_daddr,
|
|
dsize, offset)); */
|
|
#ifdef notyet
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_pagedvn, dsize,
|
|
epp->ep_daddr, epp->ep_vp, offset,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
#else
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn,
|
|
dsize, epp->ep_daddr, epp->ep_vp, offset,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
#endif
|
|
|
|
/* set up command for bss segment */
|
|
baddr = round_page(epp->ep_daddr + dsize);
|
|
bsize = epp->ep_daddr + epp->ep_dsize - baddr;
|
|
if (bsize > 0) {
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n",
|
|
baddr, bsize, 0)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero,
|
|
bsize, baddr, NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
}
|
|
|
|
/* load any shared libraries */
|
|
error = coff_find_section(p, epp->ep_vp, fp, &sh, COFF_STYP_SHLIB);
|
|
if (!error) {
|
|
int resid;
|
|
struct coff_slhdr *slhdr;
|
|
char buf[128], *bufp; /* FIXME */
|
|
int len = sh.s_size, path_index, entry_len;
|
|
|
|
/* DPRINTF(("COFF shlib size %d offset %d\n",
|
|
sh.s_size, sh.s_scnptr)); */
|
|
|
|
error = vn_rdwr(UIO_READ, epp->ep_vp, (caddr_t) buf,
|
|
len, sh.s_scnptr,
|
|
UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred,
|
|
&resid, p);
|
|
if (error) {
|
|
DPRINTF(("shlib section read error %d\n", error));
|
|
return ENOEXEC;
|
|
}
|
|
bufp = buf;
|
|
while (len) {
|
|
slhdr = (struct coff_slhdr *)bufp;
|
|
path_index = slhdr->path_index * sizeof(long);
|
|
entry_len = slhdr->entry_len * sizeof(long);
|
|
|
|
/* DPRINTF(("path_index: %d entry_len: %d name: %s\n",
|
|
path_index, entry_len, slhdr->sl_name)); */
|
|
|
|
error = coff_load_shlib(p, slhdr->sl_name);
|
|
if (error)
|
|
return ENOEXEC;
|
|
bufp += entry_len;
|
|
len -= entry_len;
|
|
}
|
|
}
|
|
|
|
/* set up entry point */
|
|
epp->ep_entry = ap->a_entry;
|
|
|
|
#if 0
|
|
DPRINTF(("text addr: %x size: %d data addr: %x size: %d entry: %x\n",
|
|
epp->ep_taddr, epp->ep_tsize,
|
|
epp->ep_daddr, epp->ep_dsize,
|
|
epp->ep_entry));
|
|
#endif
|
|
|
|
return exec_ibcs2_coff_setup_stack(p, epp);
|
|
}
|
|
|
|
int
|
|
coff_load_shlib(p, path, epp)
|
|
struct proc *p;
|
|
char *path;
|
|
struct exec_package *epp;
|
|
{
|
|
int error, siz, resid;
|
|
int taddr, tsize, daddr, dsize, offset;
|
|
struct nameidata nd;
|
|
struct coff_filehdr fh, *fhp = &fh;
|
|
struct coff_scnhdr sh, *shp = &sh;
|
|
|
|
/*
|
|
* 1. open shlib file
|
|
* 2. read filehdr
|
|
* 3. map text, data, and bss out of it using VM_*
|
|
*/
|
|
NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, path, p);
|
|
/* first get the vnode */
|
|
if (error = namei(&nd)) {
|
|
DPRINTF(("load_coff_shlib: can't find library %s\n", path));
|
|
return error;
|
|
}
|
|
|
|
siz = sizeof(struct coff_filehdr);
|
|
if (error = vn_rdwr(UIO_READ, nd.ni_vp, (caddr_t) fhp, siz, 0,
|
|
UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred,
|
|
&resid, p)) {
|
|
DPRINTF(("filehdr read error %d\n", error));
|
|
vrele(nd.ni_vp);
|
|
return error;
|
|
}
|
|
siz -= resid;
|
|
if (siz != sizeof(struct coff_filehdr)) {
|
|
DPRINTF(("load_coff_shlib: incomplete read: ask=%d, rem=%d got %d\n",
|
|
sizeof(struct coff_filehdr), resid, siz));
|
|
vrele(nd.ni_vp);
|
|
return ENOEXEC;
|
|
}
|
|
|
|
/* load text */
|
|
error = coff_find_section(p, nd.ni_vp, fhp, shp, COFF_STYP_TEXT);
|
|
if (error) {
|
|
DPRINTF(("can't find shlib text section\n"));
|
|
vrele(nd.ni_vp);
|
|
return error;
|
|
}
|
|
/* DPRINTF(("COFF text addr %x size %d offset %d\n", sh.s_vaddr,
|
|
sh.s_size, sh.s_scnptr)); */
|
|
taddr = COFF_ALIGN(shp->s_vaddr);
|
|
offset = shp->s_scnptr - (shp->s_vaddr - taddr);
|
|
tsize = shp->s_size + (shp->s_vaddr - taddr);
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n", taddr, tsize, offset)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn, tsize, taddr,
|
|
nd.ni_vp, offset,
|
|
VM_PROT_READ|VM_PROT_EXECUTE);
|
|
|
|
/* load data */
|
|
error = coff_find_section(p, nd.ni_vp, fhp, shp, COFF_STYP_DATA);
|
|
if (error) {
|
|
DPRINTF(("can't find shlib data section\n"));
|
|
vrele(nd.ni_vp);
|
|
return error;
|
|
}
|
|
/* DPRINTF(("COFF data addr %x size %d offset %d\n", shp->s_vaddr,
|
|
shp->s_size, shp->s_scnptr)); */
|
|
daddr = COFF_ALIGN(shp->s_vaddr);
|
|
offset = shp->s_scnptr - (shp->s_vaddr - daddr);
|
|
dsize = shp->s_size + (shp->s_vaddr - daddr);
|
|
/* epp->ep_dsize = dsize + ap->a_bsize; */
|
|
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n", daddr, dsize, offset)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn,
|
|
dsize, daddr, nd.ni_vp, offset,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
/* load bss */
|
|
error = coff_find_section(p, nd.ni_vp, fhp, shp, COFF_STYP_BSS);
|
|
if (!error) {
|
|
int baddr = round_page(daddr + dsize);
|
|
int bsize = daddr + dsize + shp->s_size - baddr;
|
|
if (bsize > 0) {
|
|
/* DPRINTF(("VMCMD: addr %x size %d offset %d\n",
|
|
baddr, bsize, 0)); */
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero,
|
|
bsize, baddr, NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
}
|
|
}
|
|
vrele(nd.ni_vp);
|
|
|
|
return 0;
|
|
}
|
|
|
|
void
|
|
cpu_exec_ibcs2_coff_setup(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
#if 0
|
|
struct coff_aouthdr *ap;
|
|
|
|
ap = epp->ep_hdr + sizeof(struct coff_filehdr);
|
|
p->p_md.md_regs[GP] = ap->a_gp_value;
|
|
#endif
|
|
}
|
|
|
|
int
|
|
exec_ibcs2_xout_makecmds(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
u_long midmag, magic;
|
|
u_short mid;
|
|
int error;
|
|
struct xexec *xp = epp->ep_hdr;
|
|
struct xext *xep;
|
|
|
|
if (epp->ep_hdrvalid < XOUT_HDR_SIZE)
|
|
return ENOEXEC;
|
|
|
|
if ((xp->x_magic != XOUT_MAGIC) || (xp->x_cpu != XC_386))
|
|
return ENOEXEC;
|
|
if ((xp->x_renv & (XE_ABS | XE_VMOD)) || !(xp->x_renv & XE_EXEC))
|
|
return ENOEXEC;
|
|
|
|
xep = epp->ep_hdr + sizeof(struct xexec);
|
|
#ifdef notyet
|
|
if (xp->x_renv & XE_PURE)
|
|
error = exec_ibcs2_xout_prep_zmagic(p, epp, xp, xep);
|
|
else
|
|
#endif
|
|
error = exec_ibcs2_xout_prep_nmagic(p, epp, xp, xep);
|
|
|
|
if (error == 0) {
|
|
epp->ep_emul = EMUL_IBCS2_XOUT;
|
|
epp->ep_setup = cpu_exec_ibcs2_xout_setup;
|
|
}
|
|
|
|
if (error)
|
|
kill_vmcmds(&epp->ep_vmcmds);
|
|
|
|
return error;
|
|
}
|
|
|
|
void
|
|
cpu_exec_ibcs2_xout_setup(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
|
|
return;
|
|
}
|
|
|
|
/*
|
|
* exec_ibcs2_xout_prep_nmagic(): Prepare a pure x.out binary's exec package
|
|
*
|
|
*/
|
|
|
|
int
|
|
exec_ibcs2_xout_prep_nmagic(p, epp, xp, xep)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
struct xexec *xp;
|
|
struct xext *xep;
|
|
{
|
|
int error, resid, nseg, i;
|
|
long baddr, bsize;
|
|
struct xseg *xs;
|
|
|
|
/* read in segment table */
|
|
xs = (struct xseg *)malloc(xep->xe_segsize, M_TEMP, M_WAITOK);
|
|
error = vn_rdwr(UIO_READ, epp->ep_vp, (caddr_t)xs,
|
|
xep->xe_segsize, xep->xe_segpos,
|
|
UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred,
|
|
&resid, p);
|
|
if (error) {
|
|
DPRINTF(("segment table read error %d\n", error));
|
|
free(xs, M_TEMP);
|
|
return ENOEXEC;
|
|
}
|
|
|
|
for (nseg = xep->xe_segsize / sizeof(*xs), i = 0; i < nseg; i++) {
|
|
switch (xs[i].xs_type) {
|
|
case XS_TTEXT: /* text segment */
|
|
|
|
DPRINTF(("text addr %x psize %d vsize %d off %d\n",
|
|
xs[i].xs_rbase, xs[i].xs_psize,
|
|
xs[i].xs_vsize, xs[i].xs_filpos));
|
|
|
|
epp->ep_taddr = xs[i].xs_rbase; /* XXX - align ??? */
|
|
epp->ep_tsize = xs[i].xs_vsize;
|
|
|
|
DPRINTF(("VMCMD: addr %x size %d offset %d\n",
|
|
epp->ep_taddr, epp->ep_tsize,
|
|
xs[i].xs_filpos));
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn,
|
|
epp->ep_tsize, epp->ep_taddr,
|
|
epp->ep_vp, xs[i].xs_filpos,
|
|
VM_PROT_READ|VM_PROT_EXECUTE);
|
|
break;
|
|
|
|
case XS_TDATA: /* data segment */
|
|
|
|
DPRINTF(("data addr %x psize %d vsize %d off %d\n",
|
|
xs[i].xs_rbase, xs[i].xs_psize,
|
|
xs[i].xs_vsize, xs[i].xs_filpos));
|
|
|
|
epp->ep_daddr = xs[i].xs_rbase; /* XXX - align ??? */
|
|
epp->ep_dsize = xs[i].xs_vsize;
|
|
|
|
DPRINTF(("VMCMD: addr %x size %d offset %d\n",
|
|
epp->ep_daddr, xs[i].xs_psize,
|
|
xs[i].xs_filpos));
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_readvn,
|
|
xs[i].xs_psize, epp->ep_daddr,
|
|
epp->ep_vp, xs[i].xs_filpos,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
/* set up command for bss segment */
|
|
baddr = round_page(epp->ep_daddr + xs[i].xs_psize);
|
|
bsize = epp->ep_daddr + epp->ep_dsize - baddr;
|
|
if (bsize > 0) {
|
|
DPRINTF(("VMCMD: bss addr %x size %d off %d\n",
|
|
baddr, bsize, 0));
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero,
|
|
bsize, baddr, NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|
|
|
VM_PROT_EXECUTE);
|
|
}
|
|
break;
|
|
|
|
default:
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* set up entry point */
|
|
epp->ep_entry = xp->x_entry;
|
|
|
|
DPRINTF(("text addr: %x size: %d data addr: %x size: %d entry: %x\n",
|
|
epp->ep_taddr, epp->ep_tsize,
|
|
epp->ep_daddr, epp->ep_dsize,
|
|
epp->ep_entry));
|
|
|
|
free(xs, M_TEMP);
|
|
return exec_ibcs2_xout_setup_stack(p, epp);
|
|
}
|
|
|
|
/*
|
|
* exec_ibcs2_xout_setup_stack(): Set up the stack segment for a x.out
|
|
* 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_ibcs2_xout_setup_stack(p, epp)
|
|
struct proc *p;
|
|
struct exec_package *epp;
|
|
{
|
|
epp->ep_maxsaddr = USRSTACK - MAXSSIZ;
|
|
epp->ep_minsaddr = USRSTACK;
|
|
epp->ep_ssize = p->p_rlimit[RLIMIT_STACK].rlim_cur;
|
|
|
|
/*
|
|
* 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
|
|
*
|
|
* note that in memory, things assumed to be: 0 ....... ep_maxsaddr
|
|
* <stack> ep_minsaddr
|
|
*/
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero,
|
|
((epp->ep_minsaddr - epp->ep_ssize) - epp->ep_maxsaddr),
|
|
epp->ep_maxsaddr, NULLVP, 0, VM_PROT_NONE);
|
|
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, epp->ep_ssize,
|
|
(epp->ep_minsaddr - epp->ep_ssize), NULLVP, 0,
|
|
VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE);
|
|
|
|
return 0;
|
|
}
|