提交 d0b5e15f 编写于 作者: R Richard Weinberger

um: Remove SKAS3/4 support

Before we had SKAS0 UML had two modes of operation
TT (tracing thread) and SKAS3/4 (separated kernel address space).
TT was known to be insecure and got removed a long time ago.
SKAS3/4 required a few (3 or 4) patches on the host side which never went
mainline. The last host patch is 10 years old.

With SKAS0 mode (separated kernel address space using 0 host patches),
default since 2005, SKAS3/4 is obsolete and can be removed.
Signed-off-by: NRichard Weinberger <richard@nod.at>
上级 aaeac66b
......@@ -174,7 +174,6 @@ extern unsigned long long os_makedev(unsigned major, unsigned minor);
/* start_up.c */
extern void os_early_checks(void);
extern void can_do_skas(void);
extern void os_check_bugs(void);
extern void check_host_supports_tls(int *supports_tls, int *tls_min);
......@@ -187,7 +186,6 @@ extern int os_process_parent(int pid);
extern void os_stop_process(int pid);
extern void os_kill_process(int pid, int reap_child);
extern void os_kill_ptraced_process(int pid, int reap_child);
extern long os_ptrace_ldt(long pid, long addr, long data);
extern int os_getpid(void);
extern int os_getpgrp(void);
......
/*
* Copyright (C) 2002 Jeff Dike (jdike@karaya.com)
* Licensed under the GPL
*/
#ifndef __SKAS_PROC_MM_H
#define __SKAS_PROC_MM_H
#define MM_MMAP 54
#define MM_MUNMAP 55
#define MM_MPROTECT 56
#define MM_COPY_SEGMENTS 57
struct mm_mmap {
unsigned long addr;
unsigned long len;
unsigned long prot;
unsigned long flags;
unsigned long fd;
unsigned long offset;
};
struct mm_munmap {
unsigned long addr;
unsigned long len;
};
struct mm_mprotect {
unsigned long addr;
unsigned long len;
unsigned int prot;
};
struct proc_mm_op {
int op;
union {
struct mm_mmap mmap;
struct mm_munmap munmap;
struct mm_mprotect mprotect;
int copy_segments;
} u;
};
#endif
......@@ -9,13 +9,10 @@
#include <sysdep/ptrace.h>
extern int userspace_pid[];
extern int proc_mm, ptrace_faultinfo, ptrace_ldt;
extern int skas_needs_stub;
extern int user_thread(unsigned long stack, int flags);
extern void new_thread_handler(void);
extern void handle_syscall(struct uml_pt_regs *regs);
extern int new_mm(unsigned long stack);
extern long execute_syscall_skas(void *r);
extern unsigned long current_stub_stack(void);
......
/*
* Copyright (C) 2000 - 2007 Jeff Dike (jdike@{addtoit,linux.intel}.com)
* Licensed under the GPL
*/
#ifndef __SKAS_PTRACE_H
#define __SKAS_PTRACE_H
#define PTRACE_FAULTINFO 52
#define PTRACE_SWITCH_MM 55
#include <sysdep/skas_ptrace.h>
#endif
......@@ -8,9 +8,6 @@
#include <linux/sched.h>
#include <linux/tracehook.h>
#include <asm/uaccess.h>
#include <skas_ptrace.h>
void user_enable_single_step(struct task_struct *child)
{
......@@ -104,35 +101,6 @@ long arch_ptrace(struct task_struct *child, long request,
ret = ptrace_set_thread_area(child, addr, vp);
break;
case PTRACE_FAULTINFO: {
/*
* Take the info from thread->arch->faultinfo,
* but transfer max. sizeof(struct ptrace_faultinfo).
* On i386, ptrace_faultinfo is smaller!
*/
ret = copy_to_user(p, &child->thread.arch.faultinfo,
sizeof(struct ptrace_faultinfo)) ?
-EIO : 0;
break;
}
#ifdef PTRACE_LDT
case PTRACE_LDT: {
struct ptrace_ldt ldt;
if (copy_from_user(&ldt, p, sizeof(ldt))) {
ret = -EIO;
break;
}
/*
* This one is confusing, so just punt and return -EIO for
* now
*/
ret = -EIO;
break;
}
#endif
default:
ret = ptrace_request(child, request, addr, data);
if (ret == -EIO)
......
......@@ -15,28 +15,21 @@ void (*pm_power_off)(void);
static void kill_off_processes(void)
{
if (proc_mm)
/*
* FIXME: need to loop over userspace_pids
*/
os_kill_ptraced_process(userspace_pid[0], 1);
else {
struct task_struct *p;
int pid;
read_lock(&tasklist_lock);
for_each_process(p) {
struct task_struct *t;
t = find_lock_task_mm(p);
if (!t)
continue;
pid = t->mm->context.id.u.pid;
task_unlock(t);
os_kill_ptraced_process(pid, 1);
}
read_unlock(&tasklist_lock);
struct task_struct *p;
int pid;
read_lock(&tasklist_lock);
for_each_process(p) {
struct task_struct *t;
t = find_lock_task_mm(p);
if (!t)
continue;
pid = t->mm->context.id.u.pid;
task_unlock(t);
os_kill_ptraced_process(pid, 1);
}
read_unlock(&tasklist_lock);
}
void uml_cleanup(void)
......
......@@ -54,35 +54,22 @@ int init_new_context(struct task_struct *task, struct mm_struct *mm)
unsigned long stack = 0;
int ret = -ENOMEM;
if (skas_needs_stub) {
stack = get_zeroed_page(GFP_KERNEL);
if (stack == 0)
goto out;
}
stack = get_zeroed_page(GFP_KERNEL);
if (stack == 0)
goto out;
to_mm->id.stack = stack;
if (current->mm != NULL && current->mm != &init_mm)
from_mm = &current->mm->context;
if (proc_mm) {
ret = new_mm(stack);
if (ret < 0) {
printk(KERN_ERR "init_new_context_skas - "
"new_mm failed, errno = %d\n", ret);
goto out_free;
}
to_mm->id.u.mm_fd = ret;
}
else {
if (from_mm)
to_mm->id.u.pid = copy_context_skas0(stack,
from_mm->id.u.pid);
else to_mm->id.u.pid = start_userspace(stack);
if (to_mm->id.u.pid < 0) {
ret = to_mm->id.u.pid;
goto out_free;
}
if (from_mm)
to_mm->id.u.pid = copy_context_skas0(stack,
from_mm->id.u.pid);
else to_mm->id.u.pid = start_userspace(stack);
if (to_mm->id.u.pid < 0) {
ret = to_mm->id.u.pid;
goto out_free;
}
ret = init_new_ldt(to_mm, from_mm);
......@@ -105,9 +92,6 @@ void uml_setup_stubs(struct mm_struct *mm)
{
int err, ret;
if (!skas_needs_stub)
return;
ret = init_stub_pte(mm, STUB_CODE,
(unsigned long) &__syscall_stub_start);
if (ret)
......@@ -154,25 +138,19 @@ void destroy_context(struct mm_struct *mm)
{
struct mm_context *mmu = &mm->context;
if (proc_mm)
os_close_file(mmu->id.u.mm_fd);
else {
/*
* If init_new_context wasn't called, this will be
* zero, resulting in a kill(0), which will result in the
* whole UML suddenly dying. Also, cover negative and
* 1 cases, since they shouldn't happen either.
*/
if (mmu->id.u.pid < 2) {
printk(KERN_ERR "corrupt mm_context - pid = %d\n",
mmu->id.u.pid);
return;
}
os_kill_ptraced_process(mmu->id.u.pid, 1);
/*
* If init_new_context wasn't called, this will be
* zero, resulting in a kill(0), which will result in the
* whole UML suddenly dying. Also, cover negative and
* 1 cases, since they shouldn't happen either.
*/
if (mmu->id.u.pid < 2) {
printk(KERN_ERR "corrupt mm_context - pid = %d\n",
mmu->id.u.pid);
return;
}
os_kill_ptraced_process(mmu->id.u.pid, 1);
if (skas_needs_stub)
free_page(mmu->id.stack);
free_page(mmu->id.stack);
free_ldt(mmu);
}
......@@ -10,25 +10,6 @@
#include <os.h>
#include <skas.h>
int new_mm(unsigned long stack)
{
int fd, err;
fd = os_open_file("/proc/mm", of_cloexec(of_write(OPENFLAGS())), 0);
if (fd < 0)
return fd;
if (skas_needs_stub) {
err = map_stub_pages(fd, STUB_CODE, STUB_DATA, stack);
if (err) {
os_close_file(fd);
return err;
}
}
return fd;
}
extern void start_kernel(void);
static int __init start_kernel_proc(void *unused)
......@@ -55,14 +36,6 @@ int __init start_uml(void)
{
stack_protections((unsigned long) &cpu0_irqstack);
set_sigstack(cpu0_irqstack, THREAD_SIZE);
if (proc_mm) {
userspace_pid[0] = start_userspace(0);
if (userspace_pid[0] < 0) {
printf("start_uml - start_userspace returned %d\n",
userspace_pid[0]);
exit(1);
}
}
init_new_thread_signals();
......
......@@ -220,7 +220,7 @@ unsigned long segv(struct faultinfo fi, unsigned long ip, int is_user,
panic("Segfault with no mm");
}
if (SEGV_IS_FIXABLE(&fi) || SEGV_MAYBE_FIXABLE(&fi))
if (SEGV_IS_FIXABLE(&fi))
err = handle_page_fault(address, ip, is_write, is_user,
&si.si_code);
else {
......
......@@ -268,7 +268,6 @@ int __init linux_main(int argc, char **argv)
unsigned long stack;
unsigned int i;
int add;
char * mode;
for (i = 1; i < argc; i++) {
if ((i == 1) && (argv[i][0] == ' '))
......@@ -291,15 +290,6 @@ int __init linux_main(int argc, char **argv)
/* OS sanity checks that need to happen before the kernel runs */
os_early_checks();
can_do_skas();
if (proc_mm && ptrace_faultinfo)
mode = "SKAS3";
else
mode = "SKAS0";
printf("UML running in %s mode\n", mode);
brk_start = (unsigned long) sbrk(0);
/*
......
......@@ -16,7 +16,6 @@
#include <init.h>
#include <longjmp.h>
#include <os.h>
#include <skas_ptrace.h>
#define ARBITRARY_ADDR -1
#define FAILURE_PID -1
......@@ -102,21 +101,6 @@ void os_kill_process(int pid, int reap_child)
CATCH_EINTR(waitpid(pid, NULL, __WALL));
}
/* This is here uniquely to have access to the userspace errno, i.e. the one
* used by ptrace in case of error.
*/
long os_ptrace_ldt(long pid, long addr, long data)
{
int ret;
ret = ptrace(PTRACE_LDT, pid, addr, data);
if (ret < 0)
return -errno;
return ret;
}
/* Kill off a ptraced child by all means available. kill it normally first,
* then PTRACE_KILL it, then PTRACE_CONT it in case it's in a run state from
* which it can't exit directly.
......
......@@ -12,7 +12,6 @@
#include <as-layout.h>
#include <mm_id.h>
#include <os.h>
#include <proc_mm.h>
#include <ptrace_user.h>
#include <registers.h>
#include <skas.h>
......@@ -46,8 +45,6 @@ static int __init init_syscall_regs(void)
__initcall(init_syscall_regs);
extern int proc_mm;
static inline long do_syscall_stub(struct mm_id * mm_idp, void **addr)
{
int n, i;
......@@ -56,10 +53,6 @@ static inline long do_syscall_stub(struct mm_id * mm_idp, void **addr)
unsigned long * syscall;
int err, pid = mm_idp->u.pid;
if (proc_mm)
/* FIXME: Need to look up userspace_pid by cpu */
pid = userspace_pid[0];
n = ptrace_setregs(pid, syscall_regs);
if (n < 0) {
printk(UM_KERN_ERR "Registers - \n");
......@@ -178,38 +171,12 @@ int map(struct mm_id * mm_idp, unsigned long virt, unsigned long len, int prot,
int phys_fd, unsigned long long offset, int done, void **data)
{
int ret;
unsigned long args[] = { virt, len, prot,
MAP_SHARED | MAP_FIXED, phys_fd,
MMAP_OFFSET(offset) };
if (proc_mm) {
struct proc_mm_op map;
int fd = mm_idp->u.mm_fd;
map = ((struct proc_mm_op) { .op = MM_MMAP,
.u =
{ .mmap =
{ .addr = virt,
.len = len,
.prot = prot,
.flags = MAP_SHARED |
MAP_FIXED,
.fd = phys_fd,
.offset= offset
} } } );
CATCH_EINTR(ret = write(fd, &map, sizeof(map)));
if (ret != sizeof(map)) {
ret = -errno;
printk(UM_KERN_ERR "map : /proc/mm map failed, "
"err = %d\n", -ret);
}
else ret = 0;
}
else {
unsigned long args[] = { virt, len, prot,
MAP_SHARED | MAP_FIXED, phys_fd,
MMAP_OFFSET(offset) };
ret = run_syscall_stub(mm_idp, STUB_MMAP_NR, args, virt,
data, done);
}
ret = run_syscall_stub(mm_idp, STUB_MMAP_NR, args, virt,
data, done);
return ret;
}
......@@ -218,32 +185,11 @@ int unmap(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
int done, void **data)
{
int ret;
unsigned long args[] = { (unsigned long) addr, len, 0, 0, 0,
0 };
if (proc_mm) {
struct proc_mm_op unmap;
int fd = mm_idp->u.mm_fd;
unmap = ((struct proc_mm_op) { .op = MM_MUNMAP,
.u =
{ .munmap =
{ .addr =
(unsigned long) addr,
.len = len } } } );
CATCH_EINTR(ret = write(fd, &unmap, sizeof(unmap)));
if (ret != sizeof(unmap)) {
ret = -errno;
printk(UM_KERN_ERR "unmap - proc_mm write returned "
"%d\n", ret);
}
else ret = 0;
}
else {
unsigned long args[] = { (unsigned long) addr, len, 0, 0, 0,
0 };
ret = run_syscall_stub(mm_idp, __NR_munmap, args, 0,
data, done);
}
ret = run_syscall_stub(mm_idp, __NR_munmap, args, 0,
data, done);
return ret;
}
......@@ -251,33 +197,11 @@ int unmap(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
int protect(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
unsigned int prot, int done, void **data)
{
struct proc_mm_op protect;
int ret;
unsigned long args[] = { addr, len, prot, 0, 0, 0 };
if (proc_mm) {
int fd = mm_idp->u.mm_fd;
protect = ((struct proc_mm_op) { .op = MM_MPROTECT,
.u =
{ .mprotect =
{ .addr =
(unsigned long) addr,
.len = len,
.prot = prot } } } );
CATCH_EINTR(ret = write(fd, &protect, sizeof(protect)));
if (ret != sizeof(protect)) {
ret = -errno;
printk(UM_KERN_ERR "protect failed, err = %d", -ret);
}
else ret = 0;
}
else {
unsigned long args[] = { addr, len, prot, 0, 0, 0 };
ret = run_syscall_stub(mm_idp, __NR_mprotect, args, 0,
data, done);
}
ret = run_syscall_stub(mm_idp, __NR_mprotect, args, 0,
data, done);
return ret;
}
......@@ -16,11 +16,9 @@
#include <kern_util.h>
#include <mem.h>
#include <os.h>
#include <proc_mm.h>
#include <ptrace_user.h>
#include <registers.h>
#include <skas.h>
#include <skas_ptrace.h>
#include <sysdep/stub.h>
int is_skas_winch(int pid, int fd, void *data)
......@@ -91,50 +89,33 @@ extern unsigned long current_stub_stack(void);
static void get_skas_faultinfo(int pid, struct faultinfo *fi)
{
int err;
unsigned long fpregs[FP_SIZE];
if (ptrace_faultinfo) {
err = ptrace(PTRACE_FAULTINFO, pid, 0, fi);
if (err) {
printk(UM_KERN_ERR "get_skas_faultinfo - "
"PTRACE_FAULTINFO failed, errno = %d\n", errno);
fatal_sigsegv();
}
/* Special handling for i386, which has different structs */
if (sizeof(struct ptrace_faultinfo) < sizeof(struct faultinfo))
memset((char *)fi + sizeof(struct ptrace_faultinfo), 0,
sizeof(struct faultinfo) -
sizeof(struct ptrace_faultinfo));
err = get_fp_registers(pid, fpregs);
if (err < 0) {
printk(UM_KERN_ERR "save_fp_registers returned %d\n",
err);
fatal_sigsegv();
}
else {
unsigned long fpregs[FP_SIZE];
err = get_fp_registers(pid, fpregs);
if (err < 0) {
printk(UM_KERN_ERR "save_fp_registers returned %d\n",
err);
fatal_sigsegv();
}
err = ptrace(PTRACE_CONT, pid, 0, SIGSEGV);
if (err) {
printk(UM_KERN_ERR "Failed to continue stub, pid = %d, "
"errno = %d\n", pid, errno);
fatal_sigsegv();
}
wait_stub_done(pid);
err = ptrace(PTRACE_CONT, pid, 0, SIGSEGV);
if (err) {
printk(UM_KERN_ERR "Failed to continue stub, pid = %d, "
"errno = %d\n", pid, errno);
fatal_sigsegv();
}
wait_stub_done(pid);
/*
* faultinfo is prepared by the stub-segv-handler at start of
* the stub stack page. We just have to copy it.
*/
memcpy(fi, (void *)current_stub_stack(), sizeof(*fi));
/*
* faultinfo is prepared by the stub-segv-handler at start of
* the stub stack page. We just have to copy it.
*/
memcpy(fi, (void *)current_stub_stack(), sizeof(*fi));
err = put_fp_registers(pid, fpregs);
if (err < 0) {
printk(UM_KERN_ERR "put_fp_registers returned %d\n",
err);
fatal_sigsegv();
}
err = put_fp_registers(pid, fpregs);
if (err < 0) {
printk(UM_KERN_ERR "put_fp_registers returned %d\n",
err);
fatal_sigsegv();
}
}
......@@ -198,7 +179,8 @@ extern int __syscall_stub_start;
static int userspace_tramp(void *stack)
{
void *addr;
int err;
int err, fd;
unsigned long long offset;
ptrace(PTRACE_TRACEME, 0, 0, 0);
......@@ -211,36 +193,32 @@ static int userspace_tramp(void *stack)
exit(1);
}
if (!proc_mm) {
/*
* This has a pte, but it can't be mapped in with the usual
* tlb_flush mechanism because this is part of that mechanism
*/
int fd;
unsigned long long offset;
fd = phys_mapping(to_phys(&__syscall_stub_start), &offset);
addr = mmap64((void *) STUB_CODE, UM_KERN_PAGE_SIZE,
PROT_EXEC, MAP_FIXED | MAP_PRIVATE, fd, offset);
/*
* This has a pte, but it can't be mapped in with the usual
* tlb_flush mechanism because this is part of that mechanism
*/
fd = phys_mapping(to_phys(&__syscall_stub_start), &offset);
addr = mmap64((void *) STUB_CODE, UM_KERN_PAGE_SIZE,
PROT_EXEC, MAP_FIXED | MAP_PRIVATE, fd, offset);
if (addr == MAP_FAILED) {
printk(UM_KERN_ERR "mapping mmap stub at 0x%lx failed, "
"errno = %d\n", STUB_CODE, errno);
exit(1);
}
if (stack != NULL) {
fd = phys_mapping(to_phys(stack), &offset);
addr = mmap((void *) STUB_DATA,
UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
MAP_FIXED | MAP_SHARED, fd, offset);
if (addr == MAP_FAILED) {
printk(UM_KERN_ERR "mapping mmap stub at 0x%lx failed, "
"errno = %d\n", STUB_CODE, errno);
printk(UM_KERN_ERR "mapping segfault stack "
"at 0x%lx failed, errno = %d\n",
STUB_DATA, errno);
exit(1);
}
if (stack != NULL) {
fd = phys_mapping(to_phys(stack), &offset);
addr = mmap((void *) STUB_DATA,
UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
MAP_FIXED | MAP_SHARED, fd, offset);
if (addr == MAP_FAILED) {
printk(UM_KERN_ERR "mapping segfault stack "
"at 0x%lx failed, errno = %d\n",
STUB_DATA, errno);
exit(1);
}
}
}
if (!ptrace_faultinfo && (stack != NULL)) {
if (stack != NULL) {
struct sigaction sa;
unsigned long v = STUB_CODE +
......@@ -286,11 +264,7 @@ int start_userspace(unsigned long stub_stack)
sp = (unsigned long) stack + UM_KERN_PAGE_SIZE - sizeof(void *);
flags = CLONE_FILES;
if (proc_mm)
flags |= CLONE_VM;
else
flags |= SIGCHLD;
flags = CLONE_FILES | SIGCHLD;
pid = clone(userspace_tramp, (void *) sp, flags, (void *) stub_stack);
if (pid < 0) {
......@@ -413,8 +387,7 @@ void userspace(struct uml_pt_regs *regs)
switch (sig) {
case SIGSEGV:
if (PTRACE_FULL_FAULTINFO ||
!ptrace_faultinfo) {
if (PTRACE_FULL_FAULTINFO) {
get_skas_faultinfo(pid,
&regs->faultinfo);
(*sig_info[SIGSEGV])(SIGSEGV, (struct siginfo *)&si,
......@@ -571,67 +544,6 @@ int copy_context_skas0(unsigned long new_stack, int pid)
return err;
}
/*
* This is used only, if stub pages are needed, while proc_mm is
* available. Opening /proc/mm creates a new mm_context, which lacks
* the stub-pages. Thus, we map them using /proc/mm-fd
*/
int map_stub_pages(int fd, unsigned long code, unsigned long data,
unsigned long stack)
{
struct proc_mm_op mmop;
int n;
unsigned long long code_offset;
int code_fd = phys_mapping(to_phys((void *) &__syscall_stub_start),
&code_offset);
mmop = ((struct proc_mm_op) { .op = MM_MMAP,
.u =
{ .mmap =
{ .addr = code,
.len = UM_KERN_PAGE_SIZE,
.prot = PROT_EXEC,
.flags = MAP_FIXED | MAP_PRIVATE,
.fd = code_fd,
.offset = code_offset
} } });
CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
if (n != sizeof(mmop)) {
n = errno;
printk(UM_KERN_ERR "mmap args - addr = 0x%lx, fd = %d, "
"offset = %llx\n", code, code_fd,
(unsigned long long) code_offset);
printk(UM_KERN_ERR "map_stub_pages : /proc/mm map for code "
"failed, err = %d\n", n);
return -n;
}
if (stack) {
unsigned long long map_offset;
int map_fd = phys_mapping(to_phys((void *)stack), &map_offset);
mmop = ((struct proc_mm_op)
{ .op = MM_MMAP,
.u =
{ .mmap =
{ .addr = data,
.len = UM_KERN_PAGE_SIZE,
.prot = PROT_READ | PROT_WRITE,
.flags = MAP_FIXED | MAP_SHARED,
.fd = map_fd,
.offset = map_offset
} } });
CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
if (n != sizeof(mmop)) {
n = errno;
printk(UM_KERN_ERR "map_stub_pages : /proc/mm map for "
"data failed, err = %d\n", n);
return -n;
}
}
return 0;
}
void new_thread(void *stack, jmp_buf *buf, void (*handler)(void))
{
(*buf)[0].JB_IP = (unsigned long) handler;
......@@ -728,17 +640,5 @@ void reboot_skas(void)
void __switch_mm(struct mm_id *mm_idp)
{
int err;
/* FIXME: need cpu pid in __switch_mm */
if (proc_mm) {
err = ptrace(PTRACE_SWITCH_MM, userspace_pid[0], 0,
mm_idp->u.mm_fd);
if (err) {
printk(UM_KERN_ERR "__switch_mm - PTRACE_SWITCH_MM "
"failed, errno = %d\n", errno);
fatal_sigsegv();
}
}
else userspace_pid[0] = mm_idp->u.pid;
userspace_pid[0] = mm_idp->u.pid;
}
......@@ -24,7 +24,6 @@
#include <ptrace_user.h>
#include <registers.h>
#include <skas.h>
#include <skas_ptrace.h>
static void ptrace_child(void)
{
......@@ -142,44 +141,6 @@ static int stop_ptraced_child(int pid, int exitcode, int mustexit)
return ret;
}
/* Changed only during early boot */
int ptrace_faultinfo;
static int disable_ptrace_faultinfo;
int ptrace_ldt;
static int disable_ptrace_ldt;
int proc_mm;
static int disable_proc_mm;
int have_switch_mm;
static int disable_switch_mm;
int skas_needs_stub;
static int __init skas0_cmd_param(char *str, int* add)
{
disable_ptrace_faultinfo = 1;
disable_ptrace_ldt = 1;
disable_proc_mm = 1;
disable_switch_mm = 1;
return 0;
}
/* The two __uml_setup would conflict, without this stupid alias. */
static int __init mode_skas0_cmd_param(char *str, int* add)
__attribute__((alias("skas0_cmd_param")));
__uml_setup("skas0", skas0_cmd_param,
"skas0\n"
" Disables SKAS3 and SKAS4 usage, so that SKAS0 is used\n\n");
__uml_setup("mode=skas0", mode_skas0_cmd_param,
"mode=skas0\n"
" Disables SKAS3 and SKAS4 usage, so that SKAS0 is used.\n\n");
/* Changed only during early boot */
static int force_sysemu_disabled = 0;
......@@ -376,121 +337,6 @@ void __init os_early_checks(void)
stop_ptraced_child(pid, 1, 1);
}
static int __init noprocmm_cmd_param(char *str, int* add)
{
disable_proc_mm = 1;
return 0;
}
__uml_setup("noprocmm", noprocmm_cmd_param,
"noprocmm\n"
" Turns off usage of /proc/mm, even if host supports it.\n"
" To support /proc/mm, the host needs to be patched using\n"
" the current skas3 patch.\n\n");
static int __init noptracefaultinfo_cmd_param(char *str, int* add)
{
disable_ptrace_faultinfo = 1;
return 0;
}
__uml_setup("noptracefaultinfo", noptracefaultinfo_cmd_param,
"noptracefaultinfo\n"
" Turns off usage of PTRACE_FAULTINFO, even if host supports\n"
" it. To support PTRACE_FAULTINFO, the host needs to be patched\n"
" using the current skas3 patch.\n\n");
static int __init noptraceldt_cmd_param(char *str, int* add)
{
disable_ptrace_ldt = 1;
return 0;
}
__uml_setup("noptraceldt", noptraceldt_cmd_param,
"noptraceldt\n"
" Turns off usage of PTRACE_LDT, even if host supports it.\n"
" To support PTRACE_LDT, the host needs to be patched using\n"
" the current skas3 patch.\n\n");
static inline void check_skas3_ptrace_faultinfo(void)
{
struct ptrace_faultinfo fi;
int pid, n;
non_fatal(" - PTRACE_FAULTINFO...");
pid = start_ptraced_child();
n = ptrace(PTRACE_FAULTINFO, pid, 0, &fi);
if (n < 0) {
if (errno == EIO)
non_fatal("not found\n");
else
perror("not found");
} else if (disable_ptrace_faultinfo)
non_fatal("found but disabled on command line\n");
else {
ptrace_faultinfo = 1;
non_fatal("found\n");
}
stop_ptraced_child(pid, 1, 1);
}
static inline void check_skas3_ptrace_ldt(void)
{
#ifdef PTRACE_LDT
int pid, n;
unsigned char ldtbuf[40];
struct ptrace_ldt ldt_op = (struct ptrace_ldt) {
.func = 2, /* read default ldt */
.ptr = ldtbuf,
.bytecount = sizeof(ldtbuf)};
non_fatal(" - PTRACE_LDT...");
pid = start_ptraced_child();
n = ptrace(PTRACE_LDT, pid, 0, (unsigned long) &ldt_op);
if (n < 0) {
if (errno == EIO)
non_fatal("not found\n");
else
perror("not found");
} else if (disable_ptrace_ldt)
non_fatal("found, but use is disabled\n");
else {
ptrace_ldt = 1;
non_fatal("found\n");
}
stop_ptraced_child(pid, 1, 1);
#endif
}
static inline void check_skas3_proc_mm(void)
{
non_fatal(" - /proc/mm...");
if (access("/proc/mm", W_OK) < 0)
perror("not found");
else if (disable_proc_mm)
non_fatal("found but disabled on command line\n");
else {
proc_mm = 1;
non_fatal("found\n");
}
}
void can_do_skas(void)
{
non_fatal("Checking for the skas3 patch in the host:\n");
check_skas3_proc_mm();
check_skas3_ptrace_faultinfo();
check_skas3_ptrace_ldt();
if (!proc_mm || !ptrace_faultinfo || !ptrace_ldt)
skas_needs_stub = 1;
}
int __init parse_iomem(char *str, int *add)
{
struct iomem_region *new;
......
......@@ -8,9 +8,7 @@
#include <linux/slab.h>
#include <asm/unistd.h>
#include <os.h>
#include <proc_mm.h>
#include <skas.h>
#include <skas_ptrace.h>
#include <sysdep/tls.h>
extern int modify_ldt(int func, void *ptr, unsigned long bytecount);
......@@ -19,105 +17,20 @@ static long write_ldt_entry(struct mm_id *mm_idp, int func,
struct user_desc *desc, void **addr, int done)
{
long res;
if (proc_mm) {
/*
* This is a special handling for the case, that the mm to
* modify isn't current->active_mm.
* If this is called directly by modify_ldt,
* (current->active_mm->context.skas.u == mm_idp)
* will be true. So no call to __switch_mm(mm_idp) is done.
* If this is called in case of init_new_ldt or PTRACE_LDT,
* mm_idp won't belong to current->active_mm, but child->mm.
* So we need to switch child's mm into our userspace, then
* later switch back.
*
* Note: I'm unsure: should interrupts be disabled here?
*/
if (!current->active_mm || current->active_mm == &init_mm ||
mm_idp != &current->active_mm->context.id)
__switch_mm(mm_idp);
}
if (ptrace_ldt) {
struct ptrace_ldt ldt_op = (struct ptrace_ldt) {
.func = func,
.ptr = desc,
.bytecount = sizeof(*desc)};
u32 cpu;
int pid;
if (!proc_mm)
pid = mm_idp->u.pid;
else {
cpu = get_cpu();
pid = userspace_pid[cpu];
}
res = os_ptrace_ldt(pid, 0, (unsigned long) &ldt_op);
if (proc_mm)
put_cpu();
}
else {
void *stub_addr;
res = syscall_stub_data(mm_idp, (unsigned long *)desc,
(sizeof(*desc) + sizeof(long) - 1) &
~(sizeof(long) - 1),
addr, &stub_addr);
if (!res) {
unsigned long args[] = { func,
(unsigned long)stub_addr,
sizeof(*desc),
0, 0, 0 };
res = run_syscall_stub(mm_idp, __NR_modify_ldt, args,
0, addr, done);
}
void *stub_addr;
res = syscall_stub_data(mm_idp, (unsigned long *)desc,
(sizeof(*desc) + sizeof(long) - 1) &
~(sizeof(long) - 1),
addr, &stub_addr);
if (!res) {
unsigned long args[] = { func,
(unsigned long)stub_addr,
sizeof(*desc),
0, 0, 0 };
res = run_syscall_stub(mm_idp, __NR_modify_ldt, args,
0, addr, done);
}
if (proc_mm) {
/*
* This is the second part of special handling, that makes
* PTRACE_LDT possible to implement.
*/
if (current->active_mm && current->active_mm != &init_mm &&
mm_idp != &current->active_mm->context.id)
__switch_mm(&current->active_mm->context.id);
}
return res;
}
static long read_ldt_from_host(void __user * ptr, unsigned long bytecount)
{
int res, n;
struct ptrace_ldt ptrace_ldt = (struct ptrace_ldt) {
.func = 0,
.bytecount = bytecount,
.ptr = kmalloc(bytecount, GFP_KERNEL)};
u32 cpu;
if (ptrace_ldt.ptr == NULL)
return -ENOMEM;
/*
* This is called from sys_modify_ldt only, so userspace_pid gives
* us the right number
*/
cpu = get_cpu();
res = os_ptrace_ldt(userspace_pid[cpu], 0, (unsigned long) &ptrace_ldt);
put_cpu();
if (res < 0)
goto out;
n = copy_to_user(ptr, ptrace_ldt.ptr, res);
if (n != 0)
res = -EFAULT;
out:
kfree(ptrace_ldt.ptr);
return res;
}
......@@ -145,9 +58,6 @@ static int read_ldt(void __user * ptr, unsigned long bytecount)
bytecount = LDT_ENTRY_SIZE*LDT_ENTRIES;
err = bytecount;
if (ptrace_ldt)
return read_ldt_from_host(ptr, bytecount);
mutex_lock(&ldt->lock);
if (ldt->entry_count <= LDT_DIRECT_ENTRIES) {
size = LDT_ENTRY_SIZE*LDT_DIRECT_ENTRIES;
......@@ -229,17 +139,11 @@ static int write_ldt(void __user * ptr, unsigned long bytecount, int func)
goto out;
}
if (!ptrace_ldt)
mutex_lock(&ldt->lock);
mutex_lock(&ldt->lock);
err = write_ldt_entry(mm_idp, func, &ldt_info, &addr, 1);
if (err)
goto out_unlock;
else if (ptrace_ldt) {
/* With PTRACE_LDT available, this is used as a flag only */
ldt->entry_count = 1;
goto out;
}
if (ldt_info.entry_number >= ldt->entry_count &&
ldt_info.entry_number >= LDT_DIRECT_ENTRIES) {
......@@ -393,91 +297,56 @@ long init_new_ldt(struct mm_context *new_mm, struct mm_context *from_mm)
int i;
long page, err=0;
void *addr = NULL;
struct proc_mm_op copy;
if (!ptrace_ldt)
mutex_init(&new_mm->arch.ldt.lock);
mutex_init(&new_mm->arch.ldt.lock);
if (!from_mm) {
memset(&desc, 0, sizeof(desc));
/*
* We have to initialize a clean ldt.
* Now we try to retrieve info about the ldt, we
* inherited from the host. All ldt-entries found
* will be reset in the following loop
*/
if (proc_mm) {
/*
* If the new mm was created using proc_mm, host's
* default-ldt currently is assigned, which normally
* contains the call-gates for lcall7 and lcall27.
* To remove these gates, we simply write an empty
* entry as number 0 to the host.
*/
err = write_ldt_entry(&new_mm->id, 1, &desc, &addr, 1);
}
else{
/*
* Now we try to retrieve info about the ldt, we
* inherited from the host. All ldt-entries found
* will be reset in the following loop
*/
ldt_get_host_info();
for (num_p=host_ldt_entries; *num_p != -1; num_p++) {
desc.entry_number = *num_p;
err = write_ldt_entry(&new_mm->id, 1, &desc,
&addr, *(num_p + 1) == -1);
if (err)
break;
}
ldt_get_host_info();
for (num_p=host_ldt_entries; *num_p != -1; num_p++) {
desc.entry_number = *num_p;
err = write_ldt_entry(&new_mm->id, 1, &desc,
&addr, *(num_p + 1) == -1);
if (err)
break;
}
new_mm->arch.ldt.entry_count = 0;
goto out;
}
if (proc_mm) {
/*
* We have a valid from_mm, so we now have to copy the LDT of
* from_mm to new_mm, because using proc_mm an new mm with
* an empty/default LDT was created in new_mm()
*/
copy = ((struct proc_mm_op) { .op = MM_COPY_SEGMENTS,
.u =
{ .copy_segments =
from_mm->id.u.mm_fd } } );
i = os_write_file(new_mm->id.u.mm_fd, &copy, sizeof(copy));
if (i != sizeof(copy))
printk(KERN_ERR "new_mm : /proc/mm copy_segments "
"failed, err = %d\n", -i);
}
if (!ptrace_ldt) {
/*
* Our local LDT is used to supply the data for
* modify_ldt(READLDT), if PTRACE_LDT isn't available,
* i.e., we have to use the stub for modify_ldt, which
* can't handle the big read buffer of up to 64kB.
*/
mutex_lock(&from_mm->arch.ldt.lock);
if (from_mm->arch.ldt.entry_count <= LDT_DIRECT_ENTRIES)
memcpy(new_mm->arch.ldt.u.entries, from_mm->arch.ldt.u.entries,
sizeof(new_mm->arch.ldt.u.entries));
else {
i = from_mm->arch.ldt.entry_count / LDT_ENTRIES_PER_PAGE;
while (i-->0) {
page = __get_free_page(GFP_KERNEL|__GFP_ZERO);
if (!page) {
err = -ENOMEM;
break;
}
new_mm->arch.ldt.u.pages[i] =
(struct ldt_entry *) page;
memcpy(new_mm->arch.ldt.u.pages[i],
from_mm->arch.ldt.u.pages[i], PAGE_SIZE);
/*
* Our local LDT is used to supply the data for
* modify_ldt(READLDT), if PTRACE_LDT isn't available,
* i.e., we have to use the stub for modify_ldt, which
* can't handle the big read buffer of up to 64kB.
*/
mutex_lock(&from_mm->arch.ldt.lock);
if (from_mm->arch.ldt.entry_count <= LDT_DIRECT_ENTRIES)
memcpy(new_mm->arch.ldt.u.entries, from_mm->arch.ldt.u.entries,
sizeof(new_mm->arch.ldt.u.entries));
else {
i = from_mm->arch.ldt.entry_count / LDT_ENTRIES_PER_PAGE;
while (i-->0) {
page = __get_free_page(GFP_KERNEL|__GFP_ZERO);
if (!page) {
err = -ENOMEM;
break;
}
new_mm->arch.ldt.u.pages[i] =
(struct ldt_entry *) page;
memcpy(new_mm->arch.ldt.u.pages[i],
from_mm->arch.ldt.u.pages[i], PAGE_SIZE);
}
new_mm->arch.ldt.entry_count = from_mm->arch.ldt.entry_count;
mutex_unlock(&from_mm->arch.ldt.lock);
}
new_mm->arch.ldt.entry_count = from_mm->arch.ldt.entry_count;
mutex_unlock(&from_mm->arch.ldt.lock);
out:
return err;
......@@ -488,7 +357,7 @@ void free_ldt(struct mm_context *mm)
{
int i;
if (!ptrace_ldt && mm->arch.ldt.entry_count > LDT_DIRECT_ENTRIES) {
if (mm->arch.ldt.entry_count > LDT_DIRECT_ENTRIES) {
i = mm->arch.ldt.entry_count / LDT_ENTRIES_PER_PAGE;
while (i-- > 0)
free_page((long) mm->arch.ldt.u.pages[i]);
......
......@@ -27,9 +27,6 @@ struct faultinfo {
/* This is Page Fault */
#define SEGV_IS_FIXABLE(fi) ((fi)->trap_no == 14)
/* SKAS3 has no trap_no on i386, but get_skas_faultinfo() sets it to 0. */
#define SEGV_MAYBE_FIXABLE(fi) ((fi)->trap_no == 0 && ptrace_faultinfo)
#define PTRACE_FULL_FAULTINFO 0
#endif
......@@ -27,9 +27,6 @@ struct faultinfo {
/* This is Page Fault */
#define SEGV_IS_FIXABLE(fi) ((fi)->trap_no == 14)
/* No broken SKAS API, which doesn't pass trap_no, here. */
#define SEGV_MAYBE_FIXABLE(fi) 0
#define PTRACE_FULL_FAULTINFO 1
#endif
/*
* Copyright (C) 2000, 2001, 2002 Jeff Dike (jdike@karaya.com)
* Licensed under the GPL
*/
#ifndef __SYSDEP_X86_SKAS_PTRACE_H
#define __SYSDEP_X86_SKAS_PTRACE_H
struct ptrace_faultinfo {
int is_write;
unsigned long addr;
};
struct ptrace_ldt {
int func;
void *ptr;
unsigned long bytecount;
};
#define PTRACE_LDT 54
#endif
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册