freebsd-skq/sys/kern/imgact_gzip.c
alc 215850cc69 MFC
File                  Revisions
  kern/imgact_aout.c    1.100
  kern/imgact_elf.c     1.167-1.172, 1.175
  kern/imgact_gzip.c    1.55
  vm/vm_extern.h        1.77
  vm/vm_glue.c          1.214

  Use sf_buf_alloc() instead of vm_map_find() on exec_map to create
  the ephemeral mappings that are used as the source for three copy
  operations from kernel space to user space.  There are two reasons
  for making this change: (1) Under heavy load exec_map can fill up
  causing vm_map_find() to fail.  When it fails, the nascent process
  is aborted (SIGABRT).  Whereas, this reimplementation using
  sf_buf_alloc() sleeps.  (2) Although it is possible to sleep on
  vm_map_find()'s failure until address space becomes available (see
  kmem_alloc_wait()), using sf_buf_alloc() is faster.  Furthermore,
  the reimplementation uses a CPU private mapping, avoiding a TLB
  shootdown on multiprocessors.

  The second argument to vm_map_find() should be NULL instead of 0.

  Correct a long-standing problem in elfN_map_insert(): In order to
  copy a page to user space, the user space mapping must allow write
  access.

  Eliminate an unneeded (vm_prot_t) parameter from two functions.
  Eliminate unnecessary uses of a local variable.

  Maintain the vnode lock throughout elfN_load_file() rather than
  releasing it and reacquiring it in vrele().  Consequently, there is
  no reason to increase the reference count on the vm object caching
  the file's pages.

  Eliminate unused parameters to elfN_load_file().

  Maintain the lock on the vnode for most of exec_elfN_imgact().
  Specifically, it is required for the I/O that may be performed by
  elfN_load_section().

  Avoid an obscure deadlock in the a.out, elf, and gzip image
  activators.  Add a comment describing why the deadlock does not
  occur in the common case and how it might occur in less usual
  circumstances.

  Eliminate an unused variable from exec_aout_imgact().

  Avoid a vm object reference leak in a rarely used code path.

  An executable contains at most one PT_INTERP program header.
  Therefore, the loop that searches for it can terminate after it is
  found rather than iterating over the entire set of program headers.

  Eliminate an unneeded initialization.

Approved by: re (mux)
2006-03-16 00:25:32 +00:00

400 lines
9.9 KiB
C

/*-
* ----------------------------------------------------------------------------
* "THE BEER-WARE LICENSE" (Revision 42):
* <phk@FreeBSD.org> wrote this file. As long as you retain this notice you
* can do whatever you want with this stuff. If we meet some day, and you think
* this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp
* ----------------------------------------------------------------------------
*/
/*
* This module handles execution of a.out files which have been run through
* "gzip". This saves diskspace, but wastes cpu-cycles and VM.
*
* TODO:
* text-segments should be made R/O after being filled
* is the vm-stuff safe ?
* should handle the entire header of gzip'ed stuff.
* inflate isn't quite reentrant yet...
* error-handling is a mess...
* so is the rest...
* tidy up unnecesary includes
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/exec.h>
#include <sys/imgact.h>
#include <sys/imgact_aout.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mman.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/resourcevar.h>
#include <sys/sysent.h>
#include <sys/systm.h>
#include <sys/vnode.h>
#include <sys/inflate.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_kern.h>
#include <vm/vm_extern.h>
struct imgact_gzip {
struct image_params *ip;
struct exec a_out;
int error;
int gotheader;
int where;
u_char *inbuf;
u_long offset;
u_long output;
u_long len;
int idx;
u_long virtual_offset, file_offset, file_end, bss_size;
};
static int exec_gzip_imgact(struct image_params *imgp);
static int NextByte(void *vp);
static int do_aout_hdr(struct imgact_gzip *);
static int Flush(void *vp, u_char *, u_long siz);
static int
exec_gzip_imgact(imgp)
struct image_params *imgp;
{
int error, error2 = 0;
const u_char *p = (const u_char *) imgp->image_header;
struct imgact_gzip igz;
struct inflate infl;
struct vmspace *vmspace;
/* If these four are not OK, it isn't a gzip file */
if (p[0] != 0x1f)
return -1; /* 0 Simply magic */
if (p[1] != 0x8b)
return -1; /* 1 Simply magic */
if (p[2] != 0x08)
return -1; /* 2 Compression method */
if (p[9] != 0x03)
return -1; /* 9 OS compressed on */
/*
* If this one contains anything but a comment or a filename marker,
* we don't want to chew on it
*/
if (p[3] & ~(0x18))
return ENOEXEC; /* 3 Flags */
/* These are of no use to us */
/* 4-7 Timestamp */
/* 8 Extra flags */
bzero(&igz, sizeof igz);
bzero(&infl, sizeof infl);
infl.gz_private = (void *) &igz;
infl.gz_input = NextByte;
infl.gz_output = Flush;
igz.ip = imgp;
igz.idx = 10;
if (p[3] & 0x08) { /* skip a filename */
while (p[igz.idx++])
if (igz.idx >= PAGE_SIZE)
return ENOEXEC;
}
if (p[3] & 0x10) { /* skip a comment */
while (p[igz.idx++])
if (igz.idx >= PAGE_SIZE)
return ENOEXEC;
}
igz.len = imgp->attr->va_size;
error = inflate(&infl);
/*
* The unzipped file may not even have been long enough to contain
* a header giving Flush() a chance to return error. Check for this.
*/
if ( !igz.gotheader )
return ENOEXEC;
if ( !error ) {
vmspace = imgp->proc->p_vmspace;
error = vm_map_protect(&vmspace->vm_map,
(vm_offset_t) vmspace->vm_taddr,
(vm_offset_t) (vmspace->vm_taddr +
(vmspace->vm_tsize << PAGE_SHIFT)) ,
VM_PROT_READ|VM_PROT_EXECUTE,0);
}
if (igz.inbuf) {
error2 =
vm_map_remove(kernel_map, (vm_offset_t) igz.inbuf,
(vm_offset_t) igz.inbuf + PAGE_SIZE);
}
if (igz.error || error || error2) {
printf("Output=%lu ", igz.output);
printf("Inflate_error=%d igz.error=%d error2=%d where=%d\n",
error, igz.error, error2, igz.where);
}
if (igz.error)
return igz.error;
if (error)
return ENOEXEC;
if (error2)
return error2;
return 0;
}
static int
do_aout_hdr(struct imgact_gzip * gz)
{
int error;
struct thread *td = curthread;
struct vmspace *vmspace;
vm_offset_t vmaddr;
/*
* Set file/virtual offset based on a.out variant. We do two cases:
* host byte order and network byte order (for NetBSD compatibility)
*/
switch ((int) (gz->a_out.a_magic & 0xffff)) {
case ZMAGIC:
gz->virtual_offset = 0;
if (gz->a_out.a_text) {
gz->file_offset = PAGE_SIZE;
} else {
/* Bill's "screwball mode" */
gz->file_offset = 0;
}
break;
case QMAGIC:
gz->virtual_offset = PAGE_SIZE;
gz->file_offset = 0;
break;
default:
/* NetBSD compatibility */
switch ((int) (ntohl(gz->a_out.a_magic) & 0xffff)) {
case ZMAGIC:
case QMAGIC:
gz->virtual_offset = PAGE_SIZE;
gz->file_offset = 0;
break;
default:
gz->where = __LINE__;
return (-1);
}
}
gz->bss_size = roundup(gz->a_out.a_bss, PAGE_SIZE);
/*
* Check various fields in header for validity/bounds.
*/
if ( /* entry point must lay with text region */
gz->a_out.a_entry < gz->virtual_offset ||
gz->a_out.a_entry >= gz->virtual_offset + gz->a_out.a_text ||
/* text and data size must each be page rounded */
gz->a_out.a_text & PAGE_MASK || gz->a_out.a_data & PAGE_MASK) {
gz->where = __LINE__;
return (-1);
}
/*
* text/data/bss must not exceed limits
*/
PROC_LOCK(gz->ip->proc);
if ( /* text can't exceed maximum text size */
gz->a_out.a_text > maxtsiz ||
/* data + bss can't exceed rlimit */
gz->a_out.a_data + gz->bss_size >
lim_cur(gz->ip->proc, RLIMIT_DATA)) {
PROC_UNLOCK(gz->ip->proc);
gz->where = __LINE__;
return (ENOMEM);
}
PROC_UNLOCK(gz->ip->proc);
/* Find out how far we should go */
gz->file_end = gz->file_offset + gz->a_out.a_text + gz->a_out.a_data;
/*
* Avoid a possible deadlock if the current address space is destroyed
* and that address space maps the locked vnode. In the common case,
* the locked vnode's v_usecount is decremented but remains greater
* than zero. Consequently, the vnode lock is not needed by vrele().
* However, in cases where the vnode lock is external, such as nullfs,
* v_usecount may become zero.
*/
VOP_UNLOCK(gz->ip->vp, 0, td);
/*
* Destroy old process VM and create a new one (with a new stack)
*/
exec_new_vmspace(gz->ip, &aout_sysvec);
vn_lock(gz->ip->vp, LK_EXCLUSIVE | LK_RETRY, td);
vmspace = gz->ip->proc->p_vmspace;
vmaddr = gz->virtual_offset;
error = vm_mmap(&vmspace->vm_map,
&vmaddr,
gz->a_out.a_text + gz->a_out.a_data,
VM_PROT_ALL, VM_PROT_ALL, MAP_ANON | MAP_FIXED,
OBJT_DEFAULT,
NULL,
0);
if (error) {
gz->where = __LINE__;
return (error);
}
if (gz->bss_size != 0) {
/*
* Allocate demand-zeroed area for uninitialized data.
* "bss" = 'block started by symbol' - named after the
* IBM 7090 instruction of the same name.
*/
vmaddr = gz->virtual_offset + gz->a_out.a_text +
gz->a_out.a_data;
error = vm_map_find(&vmspace->vm_map,
NULL,
0,
&vmaddr,
gz->bss_size,
FALSE, VM_PROT_ALL, VM_PROT_ALL, 0);
if (error) {
gz->where = __LINE__;
return (error);
}
}
/* Fill in process VM information */
vmspace->vm_tsize = gz->a_out.a_text >> PAGE_SHIFT;
vmspace->vm_dsize = (gz->a_out.a_data + gz->bss_size) >> PAGE_SHIFT;
vmspace->vm_taddr = (caddr_t) (uintptr_t) gz->virtual_offset;
vmspace->vm_daddr = (caddr_t) (uintptr_t)
(gz->virtual_offset + gz->a_out.a_text);
/* Fill in image_params */
gz->ip->interpreted = 0;
gz->ip->entry_addr = gz->a_out.a_entry;
gz->ip->proc->p_sysent = &aout_sysvec;
return 0;
}
static int
NextByte(void *vp)
{
int error;
struct imgact_gzip *igz = (struct imgact_gzip *) vp;
if (igz->idx >= igz->len) {
igz->where = __LINE__;
return GZ_EOF;
}
if (igz->inbuf && igz->idx < (igz->offset + PAGE_SIZE)) {
return igz->inbuf[(igz->idx++) - igz->offset];
}
if (igz->inbuf) {
error = vm_map_remove(kernel_map, (vm_offset_t) igz->inbuf,
(vm_offset_t) igz->inbuf + PAGE_SIZE);
if (error) {
igz->where = __LINE__;
igz->error = error;
return GZ_EOF;
}
}
igz->offset = igz->idx & ~PAGE_MASK;
error = vm_mmap(kernel_map, /* map */
(vm_offset_t *) & igz->inbuf, /* address */
PAGE_SIZE, /* size */
VM_PROT_READ, /* protection */
VM_PROT_READ, /* max protection */
0, /* flags */
OBJT_VNODE, /* handle type */
igz->ip->vp, /* vnode */
igz->offset); /* offset */
if (error) {
igz->where = __LINE__;
igz->error = error;
return GZ_EOF;
}
return igz->inbuf[(igz->idx++) - igz->offset];
}
static int
Flush(void *vp, u_char * ptr, u_long siz)
{
struct imgact_gzip *gz = (struct imgact_gzip *) vp;
u_char *p = ptr, *q;
int i;
/* First, find an a.out-header. */
if (gz->output < sizeof gz->a_out) {
q = (u_char *) & gz->a_out;
i = min(siz, sizeof gz->a_out - gz->output);
bcopy(p, q + gz->output, i);
gz->output += i;
p += i;
siz -= i;
if (gz->output == sizeof gz->a_out) {
gz->gotheader = 1;
i = do_aout_hdr(gz);
if (i == -1) {
if (!gz->where)
gz->where = __LINE__;
gz->error = ENOEXEC;
return ENOEXEC;
} else if (i) {
gz->where = __LINE__;
gz->error = i;
return ENOEXEC;
}
if (gz->file_offset == 0) {
q = (u_char *) (uintptr_t) gz->virtual_offset;
copyout(&gz->a_out, q, sizeof gz->a_out);
}
}
}
/* Skip over zero-padded first PAGE if needed */
if (gz->output < gz->file_offset &&
gz->output + siz > gz->file_offset) {
i = min(siz, gz->file_offset - gz->output);
gz->output += i;
p += i;
siz -= i;
}
if (gz->output >= gz->file_offset && gz->output < gz->file_end) {
i = min(siz, gz->file_end - gz->output);
q = (u_char *) (uintptr_t)
(gz->virtual_offset + gz->output - gz->file_offset);
copyout(p, q, i);
gz->output += i;
p += i;
siz -= i;
}
gz->output += siz;
return 0;
}
/*
* Tell kern_execve.c about it, with a little help from the linker.
*/
static struct execsw gzip_execsw = {exec_gzip_imgact, "gzip"};
EXEC_SET(execgzip, gzip_execsw);