linux/arch/um/drivers/mconsole_kern.c
Alan Stern e041c68341 [PATCH] Notifier chain update: API changes
The kernel's implementation of notifier chains is unsafe.  There is no
protection against entries being added to or removed from a chain while the
chain is in use.  The issues were discussed in this thread:

    http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2

We noticed that notifier chains in the kernel fall into two basic usage
classes:

	"Blocking" chains are always called from a process context
	and the callout routines are allowed to sleep;

	"Atomic" chains can be called from an atomic context and
	the callout routines are not allowed to sleep.

We decided to codify this distinction and make it part of the API.  Therefore
this set of patches introduces three new, parallel APIs: one for blocking
notifiers, one for atomic notifiers, and one for "raw" notifiers (which is
really just the old API under a new name).  New kinds of data structures are
used for the heads of the chains, and new routines are defined for
registration, unregistration, and calling a chain.  The three APIs are
explained in include/linux/notifier.h and their implementation is in
kernel/sys.c.

With atomic and blocking chains, the implementation guarantees that the chain
links will not be corrupted and that chain callers will not get messed up by
entries being added or removed.  For raw chains the implementation provides no
guarantees at all; users of this API must provide their own protections.  (The
idea was that situations may come up where the assumptions of the atomic and
blocking APIs are not appropriate, so it should be possible for users to
handle these things in their own way.)

There are some limitations, which should not be too hard to live with.  For
atomic/blocking chains, registration and unregistration must always be done in
a process context since the chain is protected by a mutex/rwsem.  Also, a
callout routine for a non-raw chain must not try to register or unregister
entries on its own chain.  (This did happen in a couple of places and the code
had to be changed to avoid it.)

Since atomic chains may be called from within an NMI handler, they cannot use
spinlocks for synchronization.  Instead we use RCU.  The overhead falls almost
entirely in the unregister routine, which is okay since unregistration is much
less frequent that calling a chain.

Here is the list of chains that we adjusted and their classifications.  None
of them use the raw API, so for the moment it is only a placeholder.

  ATOMIC CHAINS
  -------------
arch/i386/kernel/traps.c:		i386die_chain
arch/ia64/kernel/traps.c:		ia64die_chain
arch/powerpc/kernel/traps.c:		powerpc_die_chain
arch/sparc64/kernel/traps.c:		sparc64die_chain
arch/x86_64/kernel/traps.c:		die_chain
drivers/char/ipmi/ipmi_si_intf.c:	xaction_notifier_list
kernel/panic.c:				panic_notifier_list
kernel/profile.c:			task_free_notifier
net/bluetooth/hci_core.c:		hci_notifier
net/ipv4/netfilter/ip_conntrack_core.c:	ip_conntrack_chain
net/ipv4/netfilter/ip_conntrack_core.c:	ip_conntrack_expect_chain
net/ipv6/addrconf.c:			inet6addr_chain
net/netfilter/nf_conntrack_core.c:	nf_conntrack_chain
net/netfilter/nf_conntrack_core.c:	nf_conntrack_expect_chain
net/netlink/af_netlink.c:		netlink_chain

  BLOCKING CHAINS
  ---------------
arch/powerpc/platforms/pseries/reconfig.c:	pSeries_reconfig_chain
arch/s390/kernel/process.c:		idle_chain
arch/x86_64/kernel/process.c		idle_notifier
drivers/base/memory.c:			memory_chain
drivers/cpufreq/cpufreq.c		cpufreq_policy_notifier_list
drivers/cpufreq/cpufreq.c		cpufreq_transition_notifier_list
drivers/macintosh/adb.c:		adb_client_list
drivers/macintosh/via-pmu.c		sleep_notifier_list
drivers/macintosh/via-pmu68k.c		sleep_notifier_list
drivers/macintosh/windfarm_core.c	wf_client_list
drivers/usb/core/notify.c		usb_notifier_list
drivers/video/fbmem.c			fb_notifier_list
kernel/cpu.c				cpu_chain
kernel/module.c				module_notify_list
kernel/profile.c			munmap_notifier
kernel/profile.c			task_exit_notifier
kernel/sys.c				reboot_notifier_list
net/core/dev.c				netdev_chain
net/decnet/dn_dev.c:			dnaddr_chain
net/ipv4/devinet.c:			inetaddr_chain

It's possible that some of these classifications are wrong.  If they are,
please let us know or submit a patch to fix them.  Note that any chain that
gets called very frequently should be atomic, because the rwsem read-locking
used for blocking chains is very likely to incur cache misses on SMP systems.
(However, if the chain's callout routines may sleep then the chain cannot be
atomic.)

The patch set was written by Alan Stern and Chandra Seetharaman, incorporating
material written by Keith Owens and suggestions from Paul McKenney and Andrew
Morton.

[jes@sgi.com: restructure the notifier chain initialization macros]
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 08:44:50 -08:00

778 lines
17 KiB
C

/*
* Copyright (C) 2001 Lennert Buytenhek (buytenh@gnu.org)
* Copyright (C) 2001 - 2003 Jeff Dike (jdike@addtoit.com)
* Licensed under the GPL
*/
#include "linux/kernel.h"
#include "linux/slab.h"
#include "linux/init.h"
#include "linux/notifier.h"
#include "linux/reboot.h"
#include "linux/utsname.h"
#include "linux/ctype.h"
#include "linux/interrupt.h"
#include "linux/sysrq.h"
#include "linux/workqueue.h"
#include "linux/module.h"
#include "linux/file.h"
#include "linux/fs.h"
#include "linux/namei.h"
#include "linux/proc_fs.h"
#include "linux/syscalls.h"
#include "linux/console.h"
#include "asm/irq.h"
#include "asm/uaccess.h"
#include "user_util.h"
#include "kern_util.h"
#include "kern.h"
#include "mconsole.h"
#include "mconsole_kern.h"
#include "irq_user.h"
#include "init.h"
#include "os.h"
#include "umid.h"
#include "irq_kern.h"
#include "choose-mode.h"
static int do_unlink_socket(struct notifier_block *notifier,
unsigned long what, void *data)
{
return(mconsole_unlink_socket());
}
static struct notifier_block reboot_notifier = {
.notifier_call = do_unlink_socket,
.priority = 0,
};
/* Safe without explicit locking for now. Tasklets provide their own
* locking, and the interrupt handler is safe because it can't interrupt
* itself and it can only happen on CPU 0.
*/
static LIST_HEAD(mc_requests);
static void mc_work_proc(void *unused)
{
struct mconsole_entry *req;
unsigned long flags;
while(!list_empty(&mc_requests)){
local_save_flags(flags);
req = list_entry(mc_requests.next, struct mconsole_entry,
list);
list_del(&req->list);
local_irq_restore(flags);
req->request.cmd->handler(&req->request);
kfree(req);
}
}
static DECLARE_WORK(mconsole_work, mc_work_proc, NULL);
static irqreturn_t mconsole_interrupt(int irq, void *dev_id,
struct pt_regs *regs)
{
/* long to avoid size mismatch warnings from gcc */
long fd;
struct mconsole_entry *new;
struct mc_request req;
fd = (long) dev_id;
while (mconsole_get_request(fd, &req)){
if(req.cmd->context == MCONSOLE_INTR)
(*req.cmd->handler)(&req);
else {
new = kmalloc(sizeof(*new), GFP_ATOMIC);
if(new == NULL)
mconsole_reply(&req, "Out of memory", 1, 0);
else {
new->request = req;
list_add(&new->list, &mc_requests);
}
}
}
if(!list_empty(&mc_requests))
schedule_work(&mconsole_work);
reactivate_fd(fd, MCONSOLE_IRQ);
return(IRQ_HANDLED);
}
void mconsole_version(struct mc_request *req)
{
char version[256];
sprintf(version, "%s %s %s %s %s", system_utsname.sysname,
system_utsname.nodename, system_utsname.release,
system_utsname.version, system_utsname.machine);
mconsole_reply(req, version, 0, 0);
}
void mconsole_log(struct mc_request *req)
{
int len;
char *ptr = req->request.data;
ptr += strlen("log ");
len = req->len - (ptr - req->request.data);
printk("%.*s", len, ptr);
mconsole_reply(req, "", 0, 0);
}
/* This is a more convoluted version of mconsole_proc, which has some stability
* problems; however, we need it fixed, because it is expected that UML users
* mount HPPFS instead of procfs on /proc. And we want mconsole_proc to still
* show the real procfs content, not the ones from hppfs.*/
#if 0
void mconsole_proc(struct mc_request *req)
{
struct nameidata nd;
struct file_system_type *proc;
struct super_block *super;
struct file *file;
int n, err;
char *ptr = req->request.data, *buf;
ptr += strlen("proc");
while(isspace(*ptr)) ptr++;
proc = get_fs_type("proc");
if(proc == NULL){
mconsole_reply(req, "procfs not registered", 1, 0);
goto out;
}
super = (*proc->get_sb)(proc, 0, NULL, NULL);
put_filesystem(proc);
if(super == NULL){
mconsole_reply(req, "Failed to get procfs superblock", 1, 0);
goto out;
}
up_write(&super->s_umount);
nd.dentry = super->s_root;
nd.mnt = NULL;
nd.flags = O_RDONLY + 1;
nd.last_type = LAST_ROOT;
/* START: it was experienced that the stability problems are closed
* if commenting out these two calls + the below read cycle. To
* make UML crash again, it was enough to readd either one.*/
err = link_path_walk(ptr, &nd);
if(err){
mconsole_reply(req, "Failed to look up file", 1, 0);
goto out_kill;
}
file = dentry_open(nd.dentry, nd.mnt, O_RDONLY);
if(IS_ERR(file)){
mconsole_reply(req, "Failed to open file", 1, 0);
goto out_kill;
}
/*END*/
buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
if(buf == NULL){
mconsole_reply(req, "Failed to allocate buffer", 1, 0);
goto out_fput;
}
if((file->f_op != NULL) && (file->f_op->read != NULL)){
do {
n = (*file->f_op->read)(file, buf, PAGE_SIZE - 1,
&file->f_pos);
if(n >= 0){
buf[n] = '\0';
mconsole_reply(req, buf, 0, (n > 0));
}
else {
mconsole_reply(req, "Read of file failed",
1, 0);
goto out_free;
}
} while(n > 0);
}
else mconsole_reply(req, "", 0, 0);
out_free:
kfree(buf);
out_fput:
fput(file);
out_kill:
deactivate_super(super);
out: ;
}
#endif
void mconsole_proc(struct mc_request *req)
{
char path[64];
char *buf;
int len;
int fd;
int first_chunk = 1;
char *ptr = req->request.data;
ptr += strlen("proc");
while(isspace(*ptr)) ptr++;
snprintf(path, sizeof(path), "/proc/%s", ptr);
fd = sys_open(path, 0, 0);
if (fd < 0) {
mconsole_reply(req, "Failed to open file", 1, 0);
printk("open %s: %d\n",path,fd);
goto out;
}
buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
if(buf == NULL){
mconsole_reply(req, "Failed to allocate buffer", 1, 0);
goto out_close;
}
for (;;) {
len = sys_read(fd, buf, PAGE_SIZE-1);
if (len < 0) {
mconsole_reply(req, "Read of file failed", 1, 0);
goto out_free;
}
/*Begin the file content on his own line.*/
if (first_chunk) {
mconsole_reply(req, "\n", 0, 1);
first_chunk = 0;
}
if (len == PAGE_SIZE-1) {
buf[len] = '\0';
mconsole_reply(req, buf, 0, 1);
} else {
buf[len] = '\0';
mconsole_reply(req, buf, 0, 0);
break;
}
}
out_free:
kfree(buf);
out_close:
sys_close(fd);
out:
/* nothing */;
}
#define UML_MCONSOLE_HELPTEXT \
"Commands: \n\
version - Get kernel version \n\
help - Print this message \n\
halt - Halt UML \n\
reboot - Reboot UML \n\
config <dev>=<config> - Add a new device to UML; \n\
same syntax as command line \n\
config <dev> - Query the configuration of a device \n\
remove <dev> - Remove a device from UML \n\
sysrq <letter> - Performs the SysRq action controlled by the letter \n\
cad - invoke the Ctrl-Alt-Del handler \n\
stop - pause the UML; it will do nothing until it receives a 'go' \n\
go - continue the UML after a 'stop' \n\
log <string> - make UML enter <string> into the kernel log\n\
proc <file> - returns the contents of the UML's /proc/<file>\n\
stack <pid> - returns the stack of the specified pid\n\
"
void mconsole_help(struct mc_request *req)
{
mconsole_reply(req, UML_MCONSOLE_HELPTEXT, 0, 0);
}
void mconsole_halt(struct mc_request *req)
{
mconsole_reply(req, "", 0, 0);
machine_halt();
}
void mconsole_reboot(struct mc_request *req)
{
mconsole_reply(req, "", 0, 0);
machine_restart(NULL);
}
extern void ctrl_alt_del(void);
void mconsole_cad(struct mc_request *req)
{
mconsole_reply(req, "", 0, 0);
ctrl_alt_del();
}
void mconsole_go(struct mc_request *req)
{
mconsole_reply(req, "Not stopped", 1, 0);
}
void mconsole_stop(struct mc_request *req)
{
deactivate_fd(req->originating_fd, MCONSOLE_IRQ);
os_set_fd_block(req->originating_fd, 1);
mconsole_reply(req, "", 0, 0);
while(mconsole_get_request(req->originating_fd, req)){
if(req->cmd->handler == mconsole_go) break;
(*req->cmd->handler)(req);
}
os_set_fd_block(req->originating_fd, 0);
reactivate_fd(req->originating_fd, MCONSOLE_IRQ);
mconsole_reply(req, "", 0, 0);
}
/* This list is populated by __initcall routines. */
static LIST_HEAD(mconsole_devices);
void mconsole_register_dev(struct mc_device *new)
{
list_add(&new->list, &mconsole_devices);
}
static struct mc_device *mconsole_find_dev(char *name)
{
struct list_head *ele;
struct mc_device *dev;
list_for_each(ele, &mconsole_devices){
dev = list_entry(ele, struct mc_device, list);
if(!strncmp(name, dev->name, strlen(dev->name)))
return(dev);
}
return(NULL);
}
#define CONFIG_BUF_SIZE 64
static void mconsole_get_config(int (*get_config)(char *, char *, int,
char **),
struct mc_request *req, char *name)
{
char default_buf[CONFIG_BUF_SIZE], *error, *buf;
int n, size;
if(get_config == NULL){
mconsole_reply(req, "No get_config routine defined", 1, 0);
return;
}
error = NULL;
size = sizeof(default_buf)/sizeof(default_buf[0]);
buf = default_buf;
while(1){
n = (*get_config)(name, buf, size, &error);
if(error != NULL){
mconsole_reply(req, error, 1, 0);
goto out;
}
if(n <= size){
mconsole_reply(req, buf, 0, 0);
goto out;
}
if(buf != default_buf)
kfree(buf);
size = n;
buf = kmalloc(size, GFP_KERNEL);
if(buf == NULL){
mconsole_reply(req, "Failed to allocate buffer", 1, 0);
return;
}
}
out:
if(buf != default_buf)
kfree(buf);
}
void mconsole_config(struct mc_request *req)
{
struct mc_device *dev;
char *ptr = req->request.data, *name;
int err;
ptr += strlen("config");
while(isspace(*ptr)) ptr++;
dev = mconsole_find_dev(ptr);
if(dev == NULL){
mconsole_reply(req, "Bad configuration option", 1, 0);
return;
}
name = &ptr[strlen(dev->name)];
ptr = name;
while((*ptr != '=') && (*ptr != '\0'))
ptr++;
if(*ptr == '='){
err = (*dev->config)(name);
mconsole_reply(req, "", err, 0);
}
else mconsole_get_config(dev->get_config, req, name);
}
void mconsole_remove(struct mc_request *req)
{
struct mc_device *dev;
char *ptr = req->request.data, *err_msg = "";
char error[256];
int err, start, end, n;
ptr += strlen("remove");
while(isspace(*ptr)) ptr++;
dev = mconsole_find_dev(ptr);
if(dev == NULL){
mconsole_reply(req, "Bad remove option", 1, 0);
return;
}
ptr = &ptr[strlen(dev->name)];
err = 1;
n = (*dev->id)(&ptr, &start, &end);
if(n < 0){
err_msg = "Couldn't parse device number";
goto out;
}
else if((n < start) || (n > end)){
sprintf(error, "Invalid device number - must be between "
"%d and %d", start, end);
err_msg = error;
goto out;
}
err = (*dev->remove)(n);
switch(err){
case -ENODEV:
err_msg = "Device doesn't exist";
break;
case -EBUSY:
err_msg = "Device is currently open";
break;
default:
break;
}
out:
mconsole_reply(req, err_msg, err, 0);
}
static DEFINE_SPINLOCK(console_lock);
static LIST_HEAD(clients);
static char console_buf[MCONSOLE_MAX_DATA];
static int console_index = 0;
static void console_write(struct console *console, const char *string,
unsigned len)
{
struct list_head *ele;
int n;
if(list_empty(&clients))
return;
while(1){
n = min(len, ARRAY_SIZE(console_buf) - console_index);
strncpy(&console_buf[console_index], string, n);
console_index += n;
string += n;
len -= n;
if(len == 0)
return;
list_for_each(ele, &clients){
struct mconsole_entry *entry;
entry = list_entry(ele, struct mconsole_entry, list);
mconsole_reply_len(&entry->request, console_buf,
console_index, 0, 1);
}
console_index = 0;
}
}
static struct console mc_console = { .name = "mc",
.write = console_write,
.flags = CON_ENABLED,
.index = -1 };
static int mc_add_console(void)
{
register_console(&mc_console);
return 0;
}
late_initcall(mc_add_console);
static void with_console(struct mc_request *req, void (*proc)(void *),
void *arg)
{
struct mconsole_entry entry;
unsigned long flags;
INIT_LIST_HEAD(&entry.list);
entry.request = *req;
list_add(&entry.list, &clients);
spin_lock_irqsave(&console_lock, flags);
(*proc)(arg);
mconsole_reply_len(req, console_buf, console_index, 0, 0);
console_index = 0;
spin_unlock_irqrestore(&console_lock, flags);
list_del(&entry.list);
}
#ifdef CONFIG_MAGIC_SYSRQ
static void sysrq_proc(void *arg)
{
char *op = arg;
handle_sysrq(*op, &current->thread.regs, NULL);
}
void mconsole_sysrq(struct mc_request *req)
{
char *ptr = req->request.data;
ptr += strlen("sysrq");
while(isspace(*ptr)) ptr++;
/* With 'b', the system will shut down without a chance to reply,
* so in this case, we reply first.
*/
if(*ptr == 'b')
mconsole_reply(req, "", 0, 0);
with_console(req, sysrq_proc, ptr);
}
#else
void mconsole_sysrq(struct mc_request *req)
{
mconsole_reply(req, "Sysrq not compiled in", 1, 0);
}
#endif
#ifdef CONFIG_MODE_SKAS
static void stack_proc(void *arg)
{
struct task_struct *from = current, *to = arg;
to->thread.saved_task = from;
switch_to(from, to, from);
}
/* Mconsole stack trace
* Added by Allan Graves, Jeff Dike
* Dumps a stacks registers to the linux console.
* Usage stack <pid>.
*/
static void do_stack_trace(struct mc_request *req)
{
char *ptr = req->request.data;
int pid_requested= -1;
struct task_struct *from = NULL;
struct task_struct *to = NULL;
/* Would be nice:
* 1) Send showregs output to mconsole.
* 2) Add a way to stack dump all pids.
*/
ptr += strlen("stack");
while(isspace(*ptr)) ptr++;
/* Should really check for multiple pids or reject bad args here */
/* What do the arguments in mconsole_reply mean? */
if(sscanf(ptr, "%d", &pid_requested) == 0){
mconsole_reply(req, "Please specify a pid", 1, 0);
return;
}
from = current;
to = find_task_by_pid(pid_requested);
if((to == NULL) || (pid_requested == 0)) {
mconsole_reply(req, "Couldn't find that pid", 1, 0);
return;
}
with_console(req, stack_proc, to);
}
#endif /* CONFIG_MODE_SKAS */
void mconsole_stack(struct mc_request *req)
{
/* This command doesn't work in TT mode, so let's check and then
* get out of here
*/
CHOOSE_MODE(mconsole_reply(req, "Sorry, this doesn't work in TT mode",
1, 0),
do_stack_trace(req));
}
/* Changed by mconsole_setup, which is __setup, and called before SMP is
* active.
*/
static char *notify_socket = NULL;
static int mconsole_init(void)
{
/* long to avoid size mismatch warnings from gcc */
long sock;
int err;
char file[256];
if(umid_file_name("mconsole", file, sizeof(file))) return(-1);
snprintf(mconsole_socket_name, sizeof(file), "%s", file);
sock = os_create_unix_socket(file, sizeof(file), 1);
if (sock < 0){
printk("Failed to initialize management console\n");
return(1);
}
register_reboot_notifier(&reboot_notifier);
err = um_request_irq(MCONSOLE_IRQ, sock, IRQ_READ, mconsole_interrupt,
SA_INTERRUPT | SA_SHIRQ | SA_SAMPLE_RANDOM,
"mconsole", (void *)sock);
if (err){
printk("Failed to get IRQ for management console\n");
return(1);
}
if(notify_socket != NULL){
notify_socket = kstrdup(notify_socket, GFP_KERNEL);
if(notify_socket != NULL)
mconsole_notify(notify_socket, MCONSOLE_SOCKET,
mconsole_socket_name,
strlen(mconsole_socket_name) + 1);
else printk(KERN_ERR "mconsole_setup failed to strdup "
"string\n");
}
printk("mconsole (version %d) initialized on %s\n",
MCONSOLE_VERSION, mconsole_socket_name);
return(0);
}
__initcall(mconsole_init);
static int write_proc_mconsole(struct file *file, const char __user *buffer,
unsigned long count, void *data)
{
char *buf;
buf = kmalloc(count + 1, GFP_KERNEL);
if(buf == NULL)
return(-ENOMEM);
if(copy_from_user(buf, buffer, count)){
count = -EFAULT;
goto out;
}
buf[count] = '\0';
mconsole_notify(notify_socket, MCONSOLE_USER_NOTIFY, buf, count);
out:
kfree(buf);
return(count);
}
static int create_proc_mconsole(void)
{
struct proc_dir_entry *ent;
if(notify_socket == NULL) return(0);
ent = create_proc_entry("mconsole", S_IFREG | 0200, NULL);
if(ent == NULL){
printk(KERN_INFO "create_proc_mconsole : create_proc_entry failed\n");
return(0);
}
ent->read_proc = NULL;
ent->write_proc = write_proc_mconsole;
return(0);
}
static DEFINE_SPINLOCK(notify_spinlock);
void lock_notify(void)
{
spin_lock(&notify_spinlock);
}
void unlock_notify(void)
{
spin_unlock(&notify_spinlock);
}
__initcall(create_proc_mconsole);
#define NOTIFY "=notify:"
static int mconsole_setup(char *str)
{
if(!strncmp(str, NOTIFY, strlen(NOTIFY))){
str += strlen(NOTIFY);
notify_socket = str;
}
else printk(KERN_ERR "mconsole_setup : Unknown option - '%s'\n", str);
return(1);
}
__setup("mconsole", mconsole_setup);
__uml_help(mconsole_setup,
"mconsole=notify:<socket>\n"
" Requests that the mconsole driver send a message to the named Unix\n"
" socket containing the name of the mconsole socket. This also serves\n"
" to notify outside processes when UML has booted far enough to respond\n"
" to mconsole requests.\n\n"
);
static int notify_panic(struct notifier_block *self, unsigned long unused1,
void *ptr)
{
char *message = ptr;
if(notify_socket == NULL) return(0);
mconsole_notify(notify_socket, MCONSOLE_PANIC, message,
strlen(message) + 1);
return(0);
}
static struct notifier_block panic_exit_notifier = {
.notifier_call = notify_panic,
.next = NULL,
.priority = 1
};
static int add_notifier(void)
{
atomic_notifier_chain_register(&panic_notifier_list,
&panic_exit_notifier);
return(0);
}
__initcall(add_notifier);
char *mconsole_notify_socket(void)
{
return(notify_socket);
}
EXPORT_SYMBOL(mconsole_notify_socket);