2006-09-26 02:33:01 -04:00
|
|
|
/*
|
2007-10-16 04:27:00 -04:00
|
|
|
* Copyright (C) 2000 - 2007 Jeff Dike (jdike@{addtoit,linux.intel}.com)
|
2005-04-16 18:20:36 -04:00
|
|
|
* Licensed under the GPL
|
|
|
|
*/
|
|
|
|
|
uml: header untangling
Untangle UML headers somewhat and add some includes where they were
needed explicitly, but gotten accidentally via some other header.
arch/um/include/um_uaccess.h loses asm/fixmap.h because it uses no
fixmap stuff and gains elf.h, because it needs FIXADDR_USER_*, and
archsetjmp.h, because it needs jmp_buf.
pmd_alloc_one is uninlined because it needs mm_struct, and that's
inconvenient to provide in asm-um/pgtable-3level.h.
elf_core_copy_fpregs is also uninlined from elf-i386.h and
elf-x86_64.h, which duplicated the code anyway, to
arch/um/kernel/process.c, so that the reference to current_thread
doesn't pull sched.h or anything related into asm/elf.h.
arch/um/sys-i386/ldt.c, arch/um/kernel/tlb.c and
arch/um/kernel/skas/uaccess.c got sched.h because they dereference
task_structs. Its includes of linux and asm headers got turned from
"" to <>.
arch/um/sys-i386/bug.c gets asm/errno.h because it needs errno
constants.
asm/elf-i386 gets asm/user.h because it needs user_regs_struct.
asm/fixmap.h gets page.h because it needs PAGE_SIZE and PAGE_MASK and
system.h for BUG_ON.
asm/pgtable doesn't need sched.h.
asm/processor-generic.h defined mm_segment_t, but didn't use it. So,
that definition is moved to uaccess.h, which defines a bunch of
mm_segment_t-related stuff. thread_info.h uses mm_segment_t, and
includes uaccess.h, which causes a recursion. So, the definition is
placed above the include of thread_info. in uaccess.h. thread_info.h
also gets page.h because it needs PAGE_SIZE.
ObCheckpatchViolationJustification - I'm not adding a typedef; I'm
moving mm_segment_t from one place to another.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 01:30:53 -05:00
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <asm/pgtable.h>
|
|
|
|
#include <asm/tlbflush.h>
|
2007-05-06 17:51:08 -04:00
|
|
|
#include "as-layout.h"
|
2005-04-16 18:20:36 -04:00
|
|
|
#include "mem_user.h"
|
|
|
|
#include "os.h"
|
2007-10-16 04:26:58 -04:00
|
|
|
#include "skas.h"
|
2007-10-16 04:27:00 -04:00
|
|
|
#include "tlb.h"
|
2005-04-16 18:20:36 -04:00
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
struct host_vm_change {
|
|
|
|
struct host_vm_op {
|
|
|
|
enum { NONE, MMAP, MUNMAP, MPROTECT } type;
|
|
|
|
union {
|
|
|
|
struct {
|
|
|
|
unsigned long addr;
|
|
|
|
unsigned long len;
|
|
|
|
unsigned int prot;
|
|
|
|
int fd;
|
|
|
|
__u64 offset;
|
|
|
|
} mmap;
|
|
|
|
struct {
|
|
|
|
unsigned long addr;
|
|
|
|
unsigned long len;
|
|
|
|
} munmap;
|
|
|
|
struct {
|
|
|
|
unsigned long addr;
|
|
|
|
unsigned long len;
|
|
|
|
unsigned int prot;
|
|
|
|
} mprotect;
|
|
|
|
} u;
|
|
|
|
} ops[1];
|
|
|
|
int index;
|
|
|
|
struct mm_id *id;
|
|
|
|
void *data;
|
|
|
|
int force;
|
|
|
|
};
|
|
|
|
|
|
|
|
#define INIT_HVC(mm, force) \
|
|
|
|
((struct host_vm_change) \
|
|
|
|
{ .ops = { { .type = NONE } }, \
|
|
|
|
.id = &mm->context.id, \
|
|
|
|
.data = NULL, \
|
|
|
|
.index = 0, \
|
|
|
|
.force = force })
|
|
|
|
|
|
|
|
static int do_ops(struct host_vm_change *hvc, int end,
|
|
|
|
int finished)
|
|
|
|
{
|
|
|
|
struct host_vm_op *op;
|
|
|
|
int i, ret = 0;
|
|
|
|
|
|
|
|
for (i = 0; i < end && !ret; i++) {
|
|
|
|
op = &hvc->ops[i];
|
2008-02-05 01:31:14 -05:00
|
|
|
switch (op->type) {
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
case MMAP:
|
|
|
|
ret = map(hvc->id, op->u.mmap.addr, op->u.mmap.len,
|
|
|
|
op->u.mmap.prot, op->u.mmap.fd,
|
|
|
|
op->u.mmap.offset, finished, &hvc->data);
|
|
|
|
break;
|
|
|
|
case MUNMAP:
|
|
|
|
ret = unmap(hvc->id, op->u.munmap.addr,
|
|
|
|
op->u.munmap.len, finished, &hvc->data);
|
|
|
|
break;
|
|
|
|
case MPROTECT:
|
|
|
|
ret = protect(hvc->id, op->u.mprotect.addr,
|
|
|
|
op->u.mprotect.len, op->u.mprotect.prot,
|
|
|
|
finished, &hvc->data);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
printk(KERN_ERR "Unknown op type %d in do_ops\n",
|
|
|
|
op->type);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
static int add_mmap(unsigned long virt, unsigned long phys, unsigned long len,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned int prot, struct host_vm_change *hvc)
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
{
|
2006-09-26 02:33:01 -04:00
|
|
|
__u64 offset;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
struct host_vm_op *last;
|
2005-09-03 18:57:50 -04:00
|
|
|
int fd, ret = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
|
|
|
|
fd = phys_mapping(phys, &offset);
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index != 0) {
|
|
|
|
last = &hvc->ops[hvc->index - 1];
|
2007-10-16 04:27:00 -04:00
|
|
|
if ((last->type == MMAP) &&
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
(last->u.mmap.addr + last->u.mmap.len == virt) &&
|
2007-05-06 17:51:48 -04:00
|
|
|
(last->u.mmap.prot == prot) && (last->u.mmap.fd == fd) &&
|
2007-10-16 04:27:00 -04:00
|
|
|
(last->u.mmap.offset + last->u.mmap.len == offset)) {
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
last->u.mmap.len += len;
|
2005-09-03 18:57:50 -04:00
|
|
|
return 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index == ARRAY_SIZE(hvc->ops)) {
|
|
|
|
ret = do_ops(hvc, ARRAY_SIZE(hvc->ops), 0);
|
|
|
|
hvc->index = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
hvc->ops[hvc->index++] = ((struct host_vm_op)
|
|
|
|
{ .type = MMAP,
|
|
|
|
.u = { .mmap = { .addr = virt,
|
|
|
|
.len = len,
|
|
|
|
.prot = prot,
|
|
|
|
.fd = fd,
|
|
|
|
.offset = offset }
|
2005-09-03 18:57:50 -04:00
|
|
|
} });
|
|
|
|
return ret;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static int add_munmap(unsigned long addr, unsigned long len,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
struct host_vm_change *hvc)
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
{
|
|
|
|
struct host_vm_op *last;
|
2005-09-03 18:57:50 -04:00
|
|
|
int ret = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index != 0) {
|
|
|
|
last = &hvc->ops[hvc->index - 1];
|
2007-10-16 04:27:00 -04:00
|
|
|
if ((last->type == MUNMAP) &&
|
|
|
|
(last->u.munmap.addr + last->u.mmap.len == addr)) {
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
last->u.munmap.len += len;
|
2005-09-03 18:57:50 -04:00
|
|
|
return 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index == ARRAY_SIZE(hvc->ops)) {
|
|
|
|
ret = do_ops(hvc, ARRAY_SIZE(hvc->ops), 0);
|
|
|
|
hvc->index = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
hvc->ops[hvc->index++] = ((struct host_vm_op)
|
|
|
|
{ .type = MUNMAP,
|
|
|
|
.u = { .munmap = { .addr = addr,
|
|
|
|
.len = len } } });
|
2005-09-03 18:57:50 -04:00
|
|
|
return ret;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
2007-05-06 17:51:48 -04:00
|
|
|
static int add_mprotect(unsigned long addr, unsigned long len,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned int prot, struct host_vm_change *hvc)
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
{
|
|
|
|
struct host_vm_op *last;
|
2005-09-03 18:57:50 -04:00
|
|
|
int ret = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index != 0) {
|
|
|
|
last = &hvc->ops[hvc->index - 1];
|
2007-10-16 04:27:00 -04:00
|
|
|
if ((last->type == MPROTECT) &&
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
(last->u.mprotect.addr + last->u.mprotect.len == addr) &&
|
2007-10-16 04:27:00 -04:00
|
|
|
(last->u.mprotect.prot == prot)) {
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
last->u.mprotect.len += len;
|
2005-09-03 18:57:50 -04:00
|
|
|
return 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->index == ARRAY_SIZE(hvc->ops)) {
|
|
|
|
ret = do_ops(hvc, ARRAY_SIZE(hvc->ops), 0);
|
|
|
|
hvc->index = 0;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
hvc->ops[hvc->index++] = ((struct host_vm_op)
|
|
|
|
{ .type = MPROTECT,
|
|
|
|
.u = { .mprotect = { .addr = addr,
|
|
|
|
.len = len,
|
|
|
|
.prot = prot } } });
|
2005-09-03 18:57:50 -04:00
|
|
|
return ret;
|
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 18:57:36 -04:00
|
|
|
}
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
#define ADD_ROUND(n, inc) (((n) + (inc)) & ~((inc) - 1))
|
|
|
|
|
2007-05-06 17:51:30 -04:00
|
|
|
static inline int update_pte_range(pmd_t *pmd, unsigned long addr,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned long end,
|
|
|
|
struct host_vm_change *hvc)
|
2007-05-06 17:51:30 -04:00
|
|
|
{
|
|
|
|
pte_t *pte;
|
2007-05-06 17:51:48 -04:00
|
|
|
int r, w, x, prot, ret = 0;
|
2007-05-06 17:51:30 -04:00
|
|
|
|
|
|
|
pte = pte_offset_kernel(pmd, addr);
|
|
|
|
do {
|
uml: cover stubs with a VMA
Give the stubs a VMA. This allows the removal of a truly nasty kludge to make
sure that mm->nr_ptes was correct in exit_mmap. The underlying problem was
always that the stubs, which have ptes, and thus allocated a page table,
weren't covered by a VMA.
This patch fixes that by using install_special_mapping in arch_dup_mmap and
activate_context to create the VMA. The stubs have to be moved, since
shift_arg_pages seems to assume that the stack is the only VMA present at that
point during exec, and uses vma_adjust to fiddle its VMA. However, that
extends the stub VMA by the amount removed from the stack VMA.
To avoid this problem, the stubs were moved to a different fixed location at
the start of the address space.
The init_stub_pte calls were moved from init_new_context to arch_dup_mmap
because I was occasionally seeing arch_dup_mmap not being called, causing
exit_mmap to die. Rather than figure out what was really happening, I decided
it was cleaner to just move the calls so that there's no doubt that both the
pte and VMA creation happen, no matter what. arch_exit_mmap is used to clear
the stub ptes at exit time.
The STUB_* constants in as-layout.h no longer depend on UM_TASK_SIZE, that
that definition is removed, along with the comments complaining about gcc.
Because the stubs are no longer at the top of the address space, some care is
needed while flushing TLBs. update_pte_range checks for addresses in the stub
range and skips them. flush_thread now issues two unmaps, one for the range
before STUB_START and one for the range after STUB_END.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 01:31:01 -05:00
|
|
|
if ((addr >= STUB_START) && (addr < STUB_END))
|
|
|
|
continue;
|
|
|
|
|
2007-05-06 17:51:30 -04:00
|
|
|
r = pte_read(*pte);
|
|
|
|
w = pte_write(*pte);
|
|
|
|
x = pte_exec(*pte);
|
|
|
|
if (!pte_young(*pte)) {
|
|
|
|
r = 0;
|
|
|
|
w = 0;
|
2008-02-05 01:31:07 -05:00
|
|
|
} else if (!pte_dirty(*pte))
|
2007-05-06 17:51:30 -04:00
|
|
|
w = 0;
|
2008-02-05 01:31:07 -05:00
|
|
|
|
2007-05-06 17:51:48 -04:00
|
|
|
prot = ((r ? UM_PROT_READ : 0) | (w ? UM_PROT_WRITE : 0) |
|
|
|
|
(x ? UM_PROT_EXEC : 0));
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->force || pte_newpage(*pte)) {
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pte_present(*pte))
|
2007-05-06 17:51:30 -04:00
|
|
|
ret = add_mmap(addr, pte_val(*pte) & PAGE_MASK,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
PAGE_SIZE, prot, hvc);
|
2008-02-05 01:31:07 -05:00
|
|
|
else
|
|
|
|
ret = add_munmap(addr, PAGE_SIZE, hvc);
|
|
|
|
} else if (pte_newprot(*pte))
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
ret = add_mprotect(addr, PAGE_SIZE, prot, hvc);
|
2007-05-06 17:51:30 -04:00
|
|
|
*pte = pte_mkuptodate(*pte);
|
2008-02-05 01:31:06 -05:00
|
|
|
} while (pte++, addr += PAGE_SIZE, ((addr < end) && !ret));
|
2007-05-06 17:51:30 -04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int update_pmd_range(pud_t *pud, unsigned long addr,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned long end,
|
|
|
|
struct host_vm_change *hvc)
|
2007-05-06 17:51:30 -04:00
|
|
|
{
|
|
|
|
pmd_t *pmd;
|
|
|
|
unsigned long next;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
pmd = pmd_offset(pud, addr);
|
|
|
|
do {
|
|
|
|
next = pmd_addr_end(addr, end);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pmd_present(*pmd)) {
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->force || pmd_newpage(*pmd)) {
|
|
|
|
ret = add_munmap(addr, next - addr, hvc);
|
2007-05-06 17:51:30 -04:00
|
|
|
pmd_mkuptodate(*pmd);
|
|
|
|
}
|
|
|
|
}
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
else ret = update_pte_range(pmd, addr, next, hvc);
|
2008-02-05 01:31:06 -05:00
|
|
|
} while (pmd++, addr = next, ((addr < end) && !ret));
|
2007-05-06 17:51:30 -04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int update_pud_range(pgd_t *pgd, unsigned long addr,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned long end,
|
|
|
|
struct host_vm_change *hvc)
|
2007-05-06 17:51:30 -04:00
|
|
|
{
|
|
|
|
pud_t *pud;
|
|
|
|
unsigned long next;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
pud = pud_offset(pgd, addr);
|
|
|
|
do {
|
|
|
|
next = pud_addr_end(addr, end);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pud_present(*pud)) {
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
if (hvc->force || pud_newpage(*pud)) {
|
|
|
|
ret = add_munmap(addr, next - addr, hvc);
|
2007-05-06 17:51:30 -04:00
|
|
|
pud_mkuptodate(*pud);
|
|
|
|
}
|
|
|
|
}
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
else ret = update_pmd_range(pud, addr, next, hvc);
|
2008-02-05 01:31:06 -05:00
|
|
|
} while (pud++, addr = next, ((addr < end) && !ret));
|
2007-05-06 17:51:30 -04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
void fix_range_common(struct mm_struct *mm, unsigned long start_addr,
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
unsigned long end_addr, int force)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2007-05-06 17:51:30 -04:00
|
|
|
pgd_t *pgd;
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
struct host_vm_change hvc;
|
2007-05-06 17:51:30 -04:00
|
|
|
unsigned long addr = start_addr, next;
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
int ret = 0;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
hvc = INIT_HVC(mm, force);
|
2007-05-06 17:51:30 -04:00
|
|
|
pgd = pgd_offset(mm, addr);
|
|
|
|
do {
|
|
|
|
next = pgd_addr_end(addr, end_addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pgd_present(*pgd)) {
|
|
|
|
if (force || pgd_newpage(*pgd)) {
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
ret = add_munmap(addr, next - addr, &hvc);
|
2007-05-06 17:51:30 -04:00
|
|
|
pgd_mkuptodate(*pgd);
|
2006-09-26 02:33:01 -04:00
|
|
|
}
|
|
|
|
}
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
else ret = update_pud_range(pgd, addr, next, &hvc);
|
2008-02-05 01:31:06 -05:00
|
|
|
} while (pgd++, addr = next, ((addr < end_addr) && !ret));
|
2006-09-26 02:33:01 -04:00
|
|
|
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!ret)
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
ret = do_ops(&hvc, hvc.index, 1);
|
2005-09-03 18:57:50 -04:00
|
|
|
|
2007-05-06 17:51:30 -04:00
|
|
|
/* This is not an else because ret is modified above */
|
2007-10-16 04:27:00 -04:00
|
|
|
if (ret) {
|
|
|
|
printk(KERN_ERR "fix_range_common: failed, killing current "
|
|
|
|
"process\n");
|
2005-09-03 18:57:50 -04:00
|
|
|
force_sig(SIGKILL, current);
|
|
|
|
}
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
int flush_tlb_kernel_range_common(unsigned long start, unsigned long end)
|
|
|
|
{
|
2006-09-26 02:33:01 -04:00
|
|
|
struct mm_struct *mm;
|
|
|
|
pgd_t *pgd;
|
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
pte_t *pte;
|
|
|
|
unsigned long addr, last;
|
|
|
|
int updated = 0, err;
|
|
|
|
|
|
|
|
mm = &init_mm;
|
2007-10-16 04:27:00 -04:00
|
|
|
for (addr = start; addr < end;) {
|
2006-09-26 02:33:01 -04:00
|
|
|
pgd = pgd_offset(mm, addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pgd_present(*pgd)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
last = ADD_ROUND(addr, PGDIR_SIZE);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (last > end)
|
2006-09-26 02:33:01 -04:00
|
|
|
last = end;
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pgd_newpage(*pgd)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
updated = 1;
|
|
|
|
err = os_unmap_memory((void *) addr,
|
|
|
|
last - addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (err < 0)
|
2006-09-26 02:33:01 -04:00
|
|
|
panic("munmap failed, errno = %d\n",
|
|
|
|
-err);
|
|
|
|
}
|
|
|
|
addr = last;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
pud = pud_offset(pgd, addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pud_present(*pud)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
last = ADD_ROUND(addr, PUD_SIZE);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (last > end)
|
2006-09-26 02:33:01 -04:00
|
|
|
last = end;
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pud_newpage(*pud)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
updated = 1;
|
|
|
|
err = os_unmap_memory((void *) addr,
|
|
|
|
last - addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (err < 0)
|
2006-09-26 02:33:01 -04:00
|
|
|
panic("munmap failed, errno = %d\n",
|
|
|
|
-err);
|
|
|
|
}
|
|
|
|
addr = last;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
pmd = pmd_offset(pud, addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pmd_present(*pmd)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
last = ADD_ROUND(addr, PMD_SIZE);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (last > end)
|
2006-09-26 02:33:01 -04:00
|
|
|
last = end;
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pmd_newpage(*pmd)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
updated = 1;
|
|
|
|
err = os_unmap_memory((void *) addr,
|
|
|
|
last - addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (err < 0)
|
2006-09-26 02:33:01 -04:00
|
|
|
panic("munmap failed, errno = %d\n",
|
|
|
|
-err);
|
|
|
|
}
|
|
|
|
addr = last;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
pte = pte_offset_kernel(pmd, addr);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pte_present(*pte) || pte_newpage(*pte)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
updated = 1;
|
|
|
|
err = os_unmap_memory((void *) addr,
|
|
|
|
PAGE_SIZE);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (err < 0)
|
2006-09-26 02:33:01 -04:00
|
|
|
panic("munmap failed, errno = %d\n",
|
|
|
|
-err);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pte_present(*pte))
|
2006-09-26 02:33:01 -04:00
|
|
|
map_memory(addr,
|
|
|
|
pte_val(*pte) & PAGE_MASK,
|
|
|
|
PAGE_SIZE, 1, 1, 1);
|
|
|
|
}
|
2007-10-16 04:27:00 -04:00
|
|
|
else if (pte_newprot(*pte)) {
|
2006-09-26 02:33:01 -04:00
|
|
|
updated = 1;
|
|
|
|
os_protect_memory((void *) addr, PAGE_SIZE, 1, 1, 1);
|
|
|
|
}
|
|
|
|
addr += PAGE_SIZE;
|
|
|
|
}
|
2007-10-16 04:27:00 -04:00
|
|
|
return updated;
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
2007-10-16 04:26:58 -04:00
|
|
|
void flush_tlb_page(struct vm_area_struct *vma, unsigned long address)
|
|
|
|
{
|
|
|
|
pgd_t *pgd;
|
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
pte_t *pte;
|
|
|
|
struct mm_struct *mm = vma->vm_mm;
|
|
|
|
void *flush = NULL;
|
|
|
|
int r, w, x, prot, err = 0;
|
|
|
|
struct mm_id *mm_id;
|
|
|
|
|
|
|
|
address &= PAGE_MASK;
|
|
|
|
pgd = pgd_offset(mm, address);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pgd_present(*pgd))
|
2007-10-16 04:26:58 -04:00
|
|
|
goto kill;
|
|
|
|
|
|
|
|
pud = pud_offset(pgd, address);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pud_present(*pud))
|
2007-10-16 04:26:58 -04:00
|
|
|
goto kill;
|
|
|
|
|
|
|
|
pmd = pmd_offset(pud, address);
|
2007-10-16 04:27:00 -04:00
|
|
|
if (!pmd_present(*pmd))
|
2007-10-16 04:26:58 -04:00
|
|
|
goto kill;
|
|
|
|
|
|
|
|
pte = pte_offset_kernel(pmd, address);
|
|
|
|
|
|
|
|
r = pte_read(*pte);
|
|
|
|
w = pte_write(*pte);
|
|
|
|
x = pte_exec(*pte);
|
|
|
|
if (!pte_young(*pte)) {
|
|
|
|
r = 0;
|
|
|
|
w = 0;
|
|
|
|
} else if (!pte_dirty(*pte)) {
|
|
|
|
w = 0;
|
|
|
|
}
|
|
|
|
|
2007-10-16 04:27:06 -04:00
|
|
|
mm_id = &mm->context.id;
|
2007-10-16 04:26:58 -04:00
|
|
|
prot = ((r ? UM_PROT_READ : 0) | (w ? UM_PROT_WRITE : 0) |
|
|
|
|
(x ? UM_PROT_EXEC : 0));
|
2007-10-16 04:27:00 -04:00
|
|
|
if (pte_newpage(*pte)) {
|
|
|
|
if (pte_present(*pte)) {
|
2007-10-16 04:26:58 -04:00
|
|
|
unsigned long long offset;
|
|
|
|
int fd;
|
|
|
|
|
|
|
|
fd = phys_mapping(pte_val(*pte) & PAGE_MASK, &offset);
|
|
|
|
err = map(mm_id, address, PAGE_SIZE, prot, fd, offset,
|
|
|
|
1, &flush);
|
|
|
|
}
|
|
|
|
else err = unmap(mm_id, address, PAGE_SIZE, 1, &flush);
|
|
|
|
}
|
2007-10-16 04:27:00 -04:00
|
|
|
else if (pte_newprot(*pte))
|
2007-10-16 04:26:58 -04:00
|
|
|
err = protect(mm_id, address, PAGE_SIZE, prot, 1, &flush);
|
|
|
|
|
2007-10-16 04:27:00 -04:00
|
|
|
if (err)
|
2007-10-16 04:26:58 -04:00
|
|
|
goto kill;
|
|
|
|
|
|
|
|
*pte = pte_mkuptodate(*pte);
|
|
|
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
kill:
|
2007-10-16 04:27:00 -04:00
|
|
|
printk(KERN_ERR "Failed to flush page for address 0x%lx\n", address);
|
2007-10-16 04:26:58 -04:00
|
|
|
force_sig(SIGKILL, current);
|
|
|
|
}
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
pgd_t *pgd_offset_proc(struct mm_struct *mm, unsigned long address)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
return pgd_offset(mm, address);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
pud_t *pud_offset_proc(pgd_t *pgd, unsigned long address)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
return pud_offset(pgd, address);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
pmd_t *pmd_offset_proc(pud_t *pud, unsigned long address)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
return pmd_offset(pud, address);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
pte_t *pte_offset_proc(pmd_t *pmd, unsigned long address)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
return pte_offset_kernel(pmd, address);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
pte_t *addr_pte(struct task_struct *task, unsigned long addr)
|
|
|
|
{
|
2006-09-26 02:33:01 -04:00
|
|
|
pgd_t *pgd = pgd_offset(task->mm, addr);
|
|
|
|
pud_t *pud = pud_offset(pgd, addr);
|
|
|
|
pmd_t *pmd = pmd_offset(pud, addr);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2007-10-16 04:27:00 -04:00
|
|
|
return pte_offset_map(pmd, addr);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
void flush_tlb_all(void)
|
|
|
|
{
|
2006-09-26 02:33:01 -04:00
|
|
|
flush_tlb_mm(current->mm);
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
void flush_tlb_kernel_range(unsigned long start, unsigned long end)
|
|
|
|
{
|
2007-10-16 04:26:56 -04:00
|
|
|
flush_tlb_kernel_range_common(start, end);
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
void flush_tlb_kernel_vm(void)
|
|
|
|
{
|
2007-10-16 04:26:56 -04:00
|
|
|
flush_tlb_kernel_range_common(start_vm, end_vm);
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
void __flush_tlb_one(unsigned long addr)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
flush_tlb_kernel_range_common(addr, addr + PAGE_SIZE);
|
2007-10-16 04:26:58 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static void fix_range(struct mm_struct *mm, unsigned long start_addr,
|
|
|
|
unsigned long end_addr, int force)
|
|
|
|
{
|
uml: clean up tlb flush path
Tidy the tlb flushing code.
With tt mode gone, there is no reason to have the capability to have
called directly from do_mmap, do_mprotect, and do_munmap, rather than
calling a function pointer that it is given.
There was a large amount of data that was passed from function to
function, being used at the lowest level, without being changed. This
stuff is now encapsulated in a structure which is initialized at the
top layer and passed down. This simplifies the code, reduces the
amount of code needed to pass the parameters around, and saves on
stack space.
A somewhat more subtle change is the meaning of the current operation
index. It used to start at -1, being pre-incremented when adding an
operation. It now starts at 0, being post-incremented, with
associated adjustments of +/- 1 on comparisons.
In addition, tlb.h contained a couple of declarations which had no
users outside of tlb.c, so they could be moved or deleted.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 04:27:12 -04:00
|
|
|
fix_range_common(mm, start_addr, end_addr, force);
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
void flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
|
|
|
|
unsigned long end)
|
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
if (vma->vm_mm == NULL)
|
|
|
|
flush_tlb_kernel_range_common(start, end);
|
|
|
|
else fix_range(vma->vm_mm, start, end, 0);
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
2008-02-05 01:31:07 -05:00
|
|
|
void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
|
|
|
|
unsigned long end)
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
{
|
2007-10-16 04:27:00 -04:00
|
|
|
/*
|
|
|
|
* Don't bother flushing if this address space is about to be
|
|
|
|
* destroyed.
|
|
|
|
*/
|
|
|
|
if (atomic_read(&mm->mm_users) == 0)
|
|
|
|
return;
|
2007-10-16 04:26:58 -04:00
|
|
|
|
2008-02-05 01:31:07 -05:00
|
|
|
fix_range(mm, start, end, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
void flush_tlb_mm(struct mm_struct *mm)
|
|
|
|
{
|
|
|
|
struct vm_area_struct *vma = mm->mmap;
|
|
|
|
|
|
|
|
while (vma != NULL) {
|
|
|
|
fix_range(mm, vma->vm_start, vma->vm_end, 0);
|
|
|
|
vma = vma->vm_next;
|
|
|
|
}
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
void force_flush_all(void)
|
|
|
|
{
|
2007-10-16 04:26:58 -04:00
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
struct vm_area_struct *vma = mm->mmap;
|
|
|
|
|
2007-10-16 04:27:00 -04:00
|
|
|
while (vma != NULL) {
|
2007-10-16 04:26:58 -04:00
|
|
|
fix_range(mm, vma->vm_start, vma->vm_end, 1);
|
|
|
|
vma = vma->vm_next;
|
|
|
|
}
|
[PATCH] uml: skas0 - separate kernel address space on stock hosts
UML has had two modes of operation - an insecure, slow mode (tt mode) in
which the kernel is mapped into every process address space which requires
no host kernel modifications, and a secure, faster mode (skas mode) in
which the UML kernel is in a separate host address space, which requires a
patch to the host kernel.
This patch implements something very close to skas mode for hosts which
don't support skas - I'm calling this skas0. It provides the security of
the skas host patch, and some of the performance gains.
The two main things that are provided by the skas patch, /proc/mm and
PTRACE_FAULTINFO, are implemented in a way that require no host patch.
For the remote address space changing stuff (mmap, munmap, and mprotect),
we set aside two pages in the process above its stack, one of which
contains a little bit of code which can call mmap et al.
To update the address space, the system call information (system call
number and arguments) are written to the stub page above the code. The
%esp is set to the beginning of the data, the %eip is set the the start of
the stub, and it repeatedly pops the information into its registers and
makes the system call until it sees a system call number of zero. This is
to amortize the cost of the context switch across multiple address space
updates.
When the updates are done, it SIGSTOPs itself, and the kernel process
continues what it was doing.
For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the
child, and let it handle segfaults rather than nullifying them. The
handler is in the same page as the mmap stub. The second page is used as
the stack. The handler reads cr2 and err from the sigcontext, sticks them
at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The
kernel then reads the faultinfo and handles the fault.
A complication on x86_64 is that this involves resetting the registers to
the segfault values when the process is inside the kill system call. This
breaks on x86_64 because %rcx will contain %rip because you tell SYSRET
where to return to by putting the value in %rcx. So, this corrupts $rcx on
return from the segfault. To work around this, I added an
arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces
the child back through the sigreturn. This causes %rcx to be restored by
sigreturn and avoids the corruption. Ultimately, I think I will replace
this with the trick of having it send itself a blocked signal which will be
unblocked by the sigreturn. This will allow it to be stopped just after
the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of
PTRACE_SYSCALLing it through sigreturn.
This runs on a stock host, so theoretically (and hopefully), tt mode isn't
needed any more. We need to make sure that this is better in every way
than tt mode, though. I'm concerned about the speed of address space
updates and page fault handling, since they involve extra round-trips to
the child. We can amortize the round-trip cost for large address space
updates by writing all of the operations to the data page and having the
child execute them all at the same time. This will help fork and exec, but
not page faults, since they involve only one page.
I can't think of any way to help page faults, except to add something like
PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use
siginfo for SIGSEGV (or anything else) because there isn't enough
information in the siginfo struct to handle page faults (the faulting
operation type is missing). Adding that would make PTRACE_SIGINFO a usable
equivalent to PTRACE_FAULTINFO.
As for the code itself:
- The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is
put in its own section of the binary along with stub_segv_handler in
arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub
in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system
call at all, but it's only used for mmap, munmap, and mprotect.
- The x86_64 stub calls sigreturn by hand rather than allowing the normal
sigreturn to happen, because the normal sigreturn is a SA_RESTORER in
UML's address space provided by libc. Needless to say, this is not
available in the child's address space. Also, it does a couple of odd
pops before that which restore the stack to the state it was in at the
time the signal handler was called.
- There is a new field in the arch mmu_context, which is now a union.
This is the pid to be manipulated rather than the /proc/mm file
descriptor. Code which deals with this now checks proc_mm to see whether
it should use the usual skas code or the new code.
- userspace_tramp is now used to create a new host process for every UML
process, rather than one per UML processor. It checks proc_mm and
ptrace_faultinfo to decide whether to map in the pages above its stack.
- start_userspace now makes CLONE_VM conditional on proc_mm since we need
separate address spaces now.
- switch_mm_skas now just sets userspace_pid[0] to the new pid rather
than PTRACE_SWITCH_MM. There is an addition to userspace which updates
its idea of the pid being manipulated each time around the loop. This is
important on exec, when the pid will change underneath userspace().
- The stub page has a pte, but it can't be mapped in using tlb_flush
because it is part of tlb_flush. This is why it's required for it to be
mapped in by userspace_tramp.
Other random things:
- The stub section in uml.lds.S is page aligned. This page is written
out to the backing vm file in setup_physmem because it is mapped from
there into user processes.
- There's some confusion with TASK_SIZE now that there are a couple of
extra pages that the process can't use. TASK_SIZE is considered by the
elf code to be the usable process memory, which is reasonable, so it is
decreased by two pages. This confuses the definition of
USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down
of the uneven division. So we round it to the nearest PGDIR_SIZE rather
than the lower one.
- I added a missing PT_SYSCALL_ARG6_OFFSET macro.
- um_mmu.h was made into a userspace-usable file.
- proc_mm and ptrace_faultinfo are globals which say whether the host
supports these features.
- There is a bad interaction between the mm.nr_ptes check at the end of
exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing
pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't
on the last page table page, the last pte page won't be freed, as it
should be since the stub ptes are there, and exit_mmap will BUG because
there is an unfreed page. To get around this, TASK_SIZE is set to the
next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the
calls to init_stub_pte. This ensures that we know the process stack (and
all other process mappings) will be below the top page table page, and
thus we know that mm->nr_ptes will be one too many, and can be
decremented.
Things that need fixing:
- We may need better assurrences that the stub code is PIC.
- The stub pte is set up in init_new_context_skas.
- alloc_pgdir is probably the right place.
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-07 20:56:49 -04:00
|
|
|
}
|