2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* This file contains the procedures for the handling of select and poll
|
|
|
|
*
|
|
|
|
* Created for Linux based loosely upon Mathius Lattner's minix
|
|
|
|
* patches by Peter MacDonald. Heavily edited by Linus.
|
|
|
|
*
|
|
|
|
* 4 February 1994
|
|
|
|
* COFF/ELF binary emulation. If the process has the STICKY_TIMEOUTS
|
|
|
|
* flag set in its personality we do *not* modify the given timeout
|
|
|
|
* parameter to reflect time remaining.
|
|
|
|
*
|
|
|
|
* 24 January 2000
|
|
|
|
* Changed sys_poll()/do_poll() to use PAGE_SIZE chunk-based allocation
|
|
|
|
* of fds to overcome nfds < 16390 descriptors limit (Tigran Aivazian).
|
|
|
|
*/
|
|
|
|
|
2007-05-08 14:29:02 +07:00
|
|
|
#include <linux/kernel.h>
|
2009-10-04 19:11:37 +07:00
|
|
|
#include <linux/sched.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/syscalls.h>
|
2011-11-17 11:57:37 +07:00
|
|
|
#include <linux/export.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/poll.h>
|
|
|
|
#include <linux/personality.h> /* for STICKY_TIMEOUTS */
|
|
|
|
#include <linux/file.h>
|
2008-04-24 18:44:08 +07:00
|
|
|
#include <linux/fdtable.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/fs.h>
|
2005-09-10 03:04:14 +07:00
|
|
|
#include <linux/rcupdate.h>
|
2008-08-31 22:26:40 +07:00
|
|
|
#include <linux/hrtimer.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
#include <asm/uaccess.h>
|
|
|
|
|
2008-09-02 05:55:35 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Estimate expected accuracy in ns from a timeval.
|
|
|
|
*
|
|
|
|
* After quite a bit of churning around, we've settled on
|
|
|
|
* a simple thing of taking 0.1% of the timeout as the
|
|
|
|
* slack, with a cap of 100 msec.
|
|
|
|
* "nice" tasks get a 0.5% slack instead.
|
|
|
|
*
|
|
|
|
* Consider this comment an open invitation to come up with even
|
|
|
|
* better solutions..
|
|
|
|
*/
|
|
|
|
|
2009-09-23 06:43:30 +07:00
|
|
|
#define MAX_SLACK (100 * NSEC_PER_MSEC)
|
|
|
|
|
2008-09-08 06:08:55 +07:00
|
|
|
static long __estimate_accuracy(struct timespec *tv)
|
2008-09-02 05:55:35 +07:00
|
|
|
{
|
2008-09-08 06:08:55 +07:00
|
|
|
long slack;
|
2008-09-02 05:55:35 +07:00
|
|
|
int divfactor = 1000;
|
|
|
|
|
2009-09-23 06:43:30 +07:00
|
|
|
if (tv->tv_sec < 0)
|
|
|
|
return 0;
|
|
|
|
|
2008-09-08 05:31:39 +07:00
|
|
|
if (task_nice(current) > 0)
|
2008-09-02 05:55:35 +07:00
|
|
|
divfactor = divfactor / 5;
|
|
|
|
|
2009-09-23 06:43:30 +07:00
|
|
|
if (tv->tv_sec > MAX_SLACK / (NSEC_PER_SEC/divfactor))
|
|
|
|
return MAX_SLACK;
|
|
|
|
|
2008-09-02 05:55:35 +07:00
|
|
|
slack = tv->tv_nsec / divfactor;
|
|
|
|
slack += tv->tv_sec * (NSEC_PER_SEC/divfactor);
|
|
|
|
|
2009-09-23 06:43:30 +07:00
|
|
|
if (slack > MAX_SLACK)
|
|
|
|
return MAX_SLACK;
|
2008-09-08 06:08:55 +07:00
|
|
|
|
2008-09-02 05:55:35 +07:00
|
|
|
return slack;
|
|
|
|
}
|
|
|
|
|
2010-10-28 05:34:54 +07:00
|
|
|
long select_estimate_accuracy(struct timespec *tv)
|
2008-09-02 05:55:35 +07:00
|
|
|
{
|
|
|
|
unsigned long ret;
|
|
|
|
struct timespec now;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Realtime tasks get a slack of 0 for obvious reasons.
|
|
|
|
*/
|
|
|
|
|
2008-09-08 05:31:39 +07:00
|
|
|
if (rt_task(current))
|
2008-09-02 05:55:35 +07:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
ktime_get_ts(&now);
|
|
|
|
now = timespec_sub(*tv, now);
|
|
|
|
ret = __estimate_accuracy(&now);
|
|
|
|
if (ret < current->timer_slack_ns)
|
|
|
|
return current->timer_slack_ns;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
struct poll_table_page {
|
|
|
|
struct poll_table_page * next;
|
|
|
|
struct poll_table_entry * entry;
|
|
|
|
struct poll_table_entry entries[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
#define POLL_TABLE_FULL(table) \
|
|
|
|
((unsigned long)((table)->entry+1) > PAGE_SIZE + (unsigned long)(table))
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Ok, Peter made a complicated, but straightforward multiple_wait() function.
|
|
|
|
* I have rewritten this, taking some shortcuts: This code may not be easy to
|
|
|
|
* follow, but it should be free of race-conditions, and it's practical. If you
|
|
|
|
* understand what I'm doing here, then you understand how the linux
|
|
|
|
* sleep/wakeup mechanism works.
|
|
|
|
*
|
|
|
|
* Two very simple procedures, poll_wait() and poll_freewait() make all the
|
|
|
|
* work. poll_wait() is an inline-function defined in <linux/poll.h>,
|
|
|
|
* as all select/poll functions have to call it to add an entry to the
|
|
|
|
* poll table.
|
|
|
|
*/
|
2005-05-06 06:16:09 +07:00
|
|
|
static void __pollwait(struct file *filp, wait_queue_head_t *wait_address,
|
|
|
|
poll_table *p);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
void poll_initwait(struct poll_wqueues *pwq)
|
|
|
|
{
|
|
|
|
init_poll_funcptr(&pwq->pt, __pollwait);
|
2009-01-07 05:40:59 +07:00
|
|
|
pwq->polling_task = current;
|
2009-08-16 00:30:24 +07:00
|
|
|
pwq->triggered = 0;
|
2005-04-17 05:20:36 +07:00
|
|
|
pwq->error = 0;
|
|
|
|
pwq->table = NULL;
|
2006-03-28 16:56:33 +07:00
|
|
|
pwq->inline_index = 0;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(poll_initwait);
|
|
|
|
|
2006-03-28 16:56:33 +07:00
|
|
|
static void free_poll_entry(struct poll_table_entry *entry)
|
|
|
|
{
|
2007-05-09 12:10:02 +07:00
|
|
|
remove_wait_queue(entry->wait_address, &entry->wait);
|
2006-03-28 16:56:33 +07:00
|
|
|
fput(entry->filp);
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
void poll_freewait(struct poll_wqueues *pwq)
|
|
|
|
{
|
|
|
|
struct poll_table_page * p = pwq->table;
|
2006-03-28 16:56:33 +07:00
|
|
|
int i;
|
|
|
|
for (i = 0; i < pwq->inline_index; i++)
|
|
|
|
free_poll_entry(pwq->inline_entries + i);
|
2005-04-17 05:20:36 +07:00
|
|
|
while (p) {
|
|
|
|
struct poll_table_entry * entry;
|
|
|
|
struct poll_table_page *old;
|
|
|
|
|
|
|
|
entry = p->entry;
|
|
|
|
do {
|
|
|
|
entry--;
|
2006-03-28 16:56:33 +07:00
|
|
|
free_poll_entry(entry);
|
2005-04-17 05:20:36 +07:00
|
|
|
} while (entry > p->entries);
|
|
|
|
old = p;
|
|
|
|
p = p->next;
|
|
|
|
free_page((unsigned long) old);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(poll_freewait);
|
|
|
|
|
2009-01-07 05:40:59 +07:00
|
|
|
static struct poll_table_entry *poll_get_entry(struct poll_wqueues *p)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
struct poll_table_page *table = p->table;
|
|
|
|
|
2006-03-28 16:56:33 +07:00
|
|
|
if (p->inline_index < N_INLINE_POLL_ENTRIES)
|
|
|
|
return p->inline_entries + p->inline_index++;
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
if (!table || POLL_TABLE_FULL(table)) {
|
|
|
|
struct poll_table_page *new_table;
|
|
|
|
|
|
|
|
new_table = (struct poll_table_page *) __get_free_page(GFP_KERNEL);
|
|
|
|
if (!new_table) {
|
|
|
|
p->error = -ENOMEM;
|
2006-03-28 16:56:33 +07:00
|
|
|
return NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
new_table->entry = new_table->entries;
|
|
|
|
new_table->next = table;
|
|
|
|
p->table = new_table;
|
|
|
|
table = new_table;
|
|
|
|
}
|
|
|
|
|
2006-03-28 16:56:33 +07:00
|
|
|
return table->entry++;
|
|
|
|
}
|
|
|
|
|
2009-06-17 05:33:36 +07:00
|
|
|
static int __pollwake(wait_queue_t *wait, unsigned mode, int sync, void *key)
|
2009-01-07 05:40:59 +07:00
|
|
|
{
|
|
|
|
struct poll_wqueues *pwq = wait->private;
|
|
|
|
DECLARE_WAITQUEUE(dummy_wait, pwq->polling_task);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Although this function is called under waitqueue lock, LOCK
|
|
|
|
* doesn't imply write barrier and the users expect write
|
|
|
|
* barrier semantics on wakeup functions. The following
|
|
|
|
* smp_wmb() is equivalent to smp_wmb() in try_to_wake_up()
|
|
|
|
* and is paired with set_mb() in poll_schedule_timeout.
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
|
|
|
pwq->triggered = 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Perform the default wake up operation using a dummy
|
|
|
|
* waitqueue.
|
|
|
|
*
|
|
|
|
* TODO: This is hacky but there currently is no interface to
|
|
|
|
* pass in @sync. @sync is scheduled to be removed and once
|
|
|
|
* that happens, wake_up_process() can be used directly.
|
|
|
|
*/
|
|
|
|
return default_wake_function(&dummy_wait, mode, sync, key);
|
|
|
|
}
|
|
|
|
|
2009-06-17 05:33:36 +07:00
|
|
|
static int pollwake(wait_queue_t *wait, unsigned mode, int sync, void *key)
|
|
|
|
{
|
|
|
|
struct poll_table_entry *entry;
|
|
|
|
|
|
|
|
entry = container_of(wait, struct poll_table_entry, wait);
|
|
|
|
if (key && !((unsigned long)key & entry->key))
|
|
|
|
return 0;
|
|
|
|
return __pollwake(wait, mode, sync, key);
|
|
|
|
}
|
|
|
|
|
2006-03-28 16:56:33 +07:00
|
|
|
/* Add a new entry */
|
|
|
|
static void __pollwait(struct file *filp, wait_queue_head_t *wait_address,
|
|
|
|
poll_table *p)
|
|
|
|
{
|
2009-01-07 05:40:59 +07:00
|
|
|
struct poll_wqueues *pwq = container_of(p, struct poll_wqueues, pt);
|
|
|
|
struct poll_table_entry *entry = poll_get_entry(pwq);
|
2006-03-28 16:56:33 +07:00
|
|
|
if (!entry)
|
|
|
|
return;
|
|
|
|
get_file(filp);
|
|
|
|
entry->filp = filp;
|
|
|
|
entry->wait_address = wait_address;
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
entry->key = p->_key;
|
2009-01-07 05:40:59 +07:00
|
|
|
init_waitqueue_func_entry(&entry->wait, pollwake);
|
|
|
|
entry->wait.private = pwq;
|
2007-05-09 12:10:02 +07:00
|
|
|
add_wait_queue(wait_address, &entry->wait);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2009-01-07 05:40:59 +07:00
|
|
|
int poll_schedule_timeout(struct poll_wqueues *pwq, int state,
|
|
|
|
ktime_t *expires, unsigned long slack)
|
|
|
|
{
|
|
|
|
int rc = -EINTR;
|
|
|
|
|
|
|
|
set_current_state(state);
|
|
|
|
if (!pwq->triggered)
|
|
|
|
rc = schedule_hrtimeout_range(expires, slack, HRTIMER_MODE_ABS);
|
|
|
|
__set_current_state(TASK_RUNNING);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Prepare for the next iteration.
|
|
|
|
*
|
|
|
|
* The following set_mb() serves two purposes. First, it's
|
|
|
|
* the counterpart rmb of the wmb in pollwake() such that data
|
|
|
|
* written before wake up is always visible after wake up.
|
|
|
|
* Second, the full barrier guarantees that triggered clearing
|
|
|
|
* doesn't pass event check of the next iteration. Note that
|
|
|
|
* this problem doesn't exist for the first iteration as
|
|
|
|
* add_wait_queue() has full barrier semantics.
|
|
|
|
*/
|
|
|
|
set_mb(pwq->triggered, 0);
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(poll_schedule_timeout);
|
|
|
|
|
2008-08-31 22:16:57 +07:00
|
|
|
/**
|
|
|
|
* poll_select_set_timeout - helper function to setup the timeout value
|
|
|
|
* @to: pointer to timespec variable for the final timeout
|
|
|
|
* @sec: seconds (from user space)
|
|
|
|
* @nsec: nanoseconds (from user space)
|
|
|
|
*
|
|
|
|
* Note, we do not use a timespec for the user space value here, That
|
|
|
|
* way we can use the function for timeval and compat interfaces as well.
|
|
|
|
*
|
|
|
|
* Returns -EINVAL if sec/nsec are not normalized. Otherwise 0.
|
|
|
|
*/
|
|
|
|
int poll_select_set_timeout(struct timespec *to, long sec, long nsec)
|
|
|
|
{
|
|
|
|
struct timespec ts = {.tv_sec = sec, .tv_nsec = nsec};
|
|
|
|
|
|
|
|
if (!timespec_valid(&ts))
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* Optimize for the zero timeout value here */
|
|
|
|
if (!sec && !nsec) {
|
|
|
|
to->tv_sec = to->tv_nsec = 0;
|
|
|
|
} else {
|
|
|
|
ktime_get_ts(to);
|
|
|
|
*to = timespec_add_safe(*to, ts);
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int poll_select_copy_remaining(struct timespec *end_time, void __user *p,
|
|
|
|
int timeval, int ret)
|
|
|
|
{
|
|
|
|
struct timespec rts;
|
|
|
|
struct timeval rtv;
|
|
|
|
|
|
|
|
if (!p)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
if (current->personality & STICKY_TIMEOUTS)
|
|
|
|
goto sticky;
|
|
|
|
|
|
|
|
/* No update for zero timeout */
|
|
|
|
if (!end_time->tv_sec && !end_time->tv_nsec)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ktime_get_ts(&rts);
|
|
|
|
rts = timespec_sub(*end_time, rts);
|
|
|
|
if (rts.tv_sec < 0)
|
|
|
|
rts.tv_sec = rts.tv_nsec = 0;
|
|
|
|
|
|
|
|
if (timeval) {
|
2011-01-13 08:00:00 +07:00
|
|
|
if (sizeof(rtv) > sizeof(rtv.tv_sec) + sizeof(rtv.tv_usec))
|
|
|
|
memset(&rtv, 0, sizeof(rtv));
|
2008-08-31 22:16:57 +07:00
|
|
|
rtv.tv_sec = rts.tv_sec;
|
|
|
|
rtv.tv_usec = rts.tv_nsec / NSEC_PER_USEC;
|
|
|
|
|
|
|
|
if (!copy_to_user(p, &rtv, sizeof(rtv)))
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
} else if (!copy_to_user(p, &rts, sizeof(rts)))
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If an application puts its timeval in read-only memory, we
|
|
|
|
* don't want the Linux-specific update to the timeval to
|
|
|
|
* cause a fault after the select has completed
|
|
|
|
* successfully. However, because we're not updating the
|
|
|
|
* timeval, we can't restart the system call.
|
|
|
|
*/
|
|
|
|
|
|
|
|
sticky:
|
|
|
|
if (ret == -ERESTARTNOHAND)
|
|
|
|
ret = -EINTR;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#define FDS_IN(fds, n) (fds->in + n)
|
|
|
|
#define FDS_OUT(fds, n) (fds->out + n)
|
|
|
|
#define FDS_EX(fds, n) (fds->ex + n)
|
|
|
|
|
|
|
|
#define BITS(fds, n) (*FDS_IN(fds, n)|*FDS_OUT(fds, n)|*FDS_EX(fds, n))
|
|
|
|
|
|
|
|
static int max_select_fd(unsigned long n, fd_set_bits *fds)
|
|
|
|
{
|
|
|
|
unsigned long *open_fds;
|
|
|
|
unsigned long set;
|
|
|
|
int max;
|
2005-09-10 03:04:10 +07:00
|
|
|
struct fdtable *fdt;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* handle last in-complete long-word first */
|
2012-07-25 21:40:34 +07:00
|
|
|
set = ~(~0UL << (n & (BITS_PER_LONG-1)));
|
|
|
|
n /= BITS_PER_LONG;
|
2005-09-10 03:04:10 +07:00
|
|
|
fdt = files_fdtable(current->files);
|
2012-02-17 00:49:54 +07:00
|
|
|
open_fds = fdt->open_fds + n;
|
2005-04-17 05:20:36 +07:00
|
|
|
max = 0;
|
|
|
|
if (set) {
|
|
|
|
set &= BITS(fds, n);
|
|
|
|
if (set) {
|
|
|
|
if (!(set & ~*open_fds))
|
|
|
|
goto get_max;
|
|
|
|
return -EBADF;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
while (n) {
|
|
|
|
open_fds--;
|
|
|
|
n--;
|
|
|
|
set = BITS(fds, n);
|
|
|
|
if (!set)
|
|
|
|
continue;
|
|
|
|
if (set & ~*open_fds)
|
|
|
|
return -EBADF;
|
|
|
|
if (max)
|
|
|
|
continue;
|
|
|
|
get_max:
|
|
|
|
do {
|
|
|
|
max++;
|
|
|
|
set >>= 1;
|
|
|
|
} while (set);
|
2012-07-25 21:40:34 +07:00
|
|
|
max += n * BITS_PER_LONG;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
return max;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define POLLIN_SET (POLLRDNORM | POLLRDBAND | POLLIN | POLLHUP | POLLERR)
|
|
|
|
#define POLLOUT_SET (POLLWRBAND | POLLWRNORM | POLLOUT | POLLERR)
|
|
|
|
#define POLLEX_SET (POLLPRI)
|
|
|
|
|
2009-06-17 05:33:36 +07:00
|
|
|
static inline void wait_key_set(poll_table *wait, unsigned long in,
|
|
|
|
unsigned long out, unsigned long bit)
|
|
|
|
{
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_key = POLLEX_SET;
|
|
|
|
if (in & bit)
|
|
|
|
wait->_key |= POLLIN_SET;
|
|
|
|
if (out & bit)
|
|
|
|
wait->_key |= POLLOUT_SET;
|
2009-06-17 05:33:36 +07:00
|
|
|
}
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
int do_select(int n, fd_set_bits *fds, struct timespec *end_time)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2008-08-31 22:26:40 +07:00
|
|
|
ktime_t expire, *to = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct poll_wqueues table;
|
|
|
|
poll_table *wait;
|
2008-08-31 22:26:40 +07:00
|
|
|
int retval, i, timed_out = 0;
|
2008-09-02 05:55:35 +07:00
|
|
|
unsigned long slack = 0;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2005-09-10 03:04:14 +07:00
|
|
|
rcu_read_lock();
|
2005-04-17 05:20:36 +07:00
|
|
|
retval = max_select_fd(n, fds);
|
2005-09-10 03:04:14 +07:00
|
|
|
rcu_read_unlock();
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
if (retval < 0)
|
|
|
|
return retval;
|
|
|
|
n = retval;
|
|
|
|
|
|
|
|
poll_initwait(&table);
|
|
|
|
wait = &table.pt;
|
2008-08-31 22:26:40 +07:00
|
|
|
if (end_time && !end_time->tv_sec && !end_time->tv_nsec) {
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_qproc = NULL;
|
2008-08-31 22:26:40 +07:00
|
|
|
timed_out = 1;
|
|
|
|
}
|
|
|
|
|
2008-09-08 06:08:55 +07:00
|
|
|
if (end_time && !timed_out)
|
2010-10-28 05:34:53 +07:00
|
|
|
slack = select_estimate_accuracy(end_time);
|
2008-09-02 05:55:35 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
retval = 0;
|
|
|
|
for (;;) {
|
|
|
|
unsigned long *rinp, *routp, *rexp, *inp, *outp, *exp;
|
|
|
|
|
|
|
|
inp = fds->in; outp = fds->out; exp = fds->ex;
|
|
|
|
rinp = fds->res_in; routp = fds->res_out; rexp = fds->res_ex;
|
|
|
|
|
|
|
|
for (i = 0; i < n; ++rinp, ++routp, ++rexp) {
|
|
|
|
unsigned long in, out, ex, all_bits, bit = 1, mask, j;
|
|
|
|
unsigned long res_in = 0, res_out = 0, res_ex = 0;
|
2006-03-28 16:56:41 +07:00
|
|
|
const struct file_operations *f_op = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct file *file = NULL;
|
|
|
|
|
|
|
|
in = *inp++; out = *outp++; ex = *exp++;
|
|
|
|
all_bits = in | out | ex;
|
|
|
|
if (all_bits == 0) {
|
2012-07-25 21:40:34 +07:00
|
|
|
i += BITS_PER_LONG;
|
2005-04-17 05:20:36 +07:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2012-07-25 21:40:34 +07:00
|
|
|
for (j = 0; j < BITS_PER_LONG; ++j, ++i, bit <<= 1) {
|
2006-03-28 16:56:34 +07:00
|
|
|
int fput_needed;
|
2005-04-17 05:20:36 +07:00
|
|
|
if (i >= n)
|
|
|
|
break;
|
|
|
|
if (!(bit & all_bits))
|
|
|
|
continue;
|
2006-03-28 16:56:34 +07:00
|
|
|
file = fget_light(i, &fput_needed);
|
2005-04-17 05:20:36 +07:00
|
|
|
if (file) {
|
|
|
|
f_op = file->f_op;
|
|
|
|
mask = DEFAULT_POLLMASK;
|
2009-06-17 05:33:36 +07:00
|
|
|
if (f_op && f_op->poll) {
|
|
|
|
wait_key_set(wait, in, out, bit);
|
|
|
|
mask = (*f_op->poll)(file, wait);
|
|
|
|
}
|
2006-03-28 16:56:34 +07:00
|
|
|
fput_light(file, fput_needed);
|
2005-04-17 05:20:36 +07:00
|
|
|
if ((mask & POLLIN_SET) && (in & bit)) {
|
|
|
|
res_in |= bit;
|
|
|
|
retval++;
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_qproc = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
if ((mask & POLLOUT_SET) && (out & bit)) {
|
|
|
|
res_out |= bit;
|
|
|
|
retval++;
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_qproc = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
if ((mask & POLLEX_SET) && (ex & bit)) {
|
|
|
|
res_ex |= bit;
|
|
|
|
retval++;
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_qproc = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (res_in)
|
|
|
|
*rinp = res_in;
|
|
|
|
if (res_out)
|
|
|
|
*routp = res_out;
|
|
|
|
if (res_ex)
|
|
|
|
*rexp = res_ex;
|
2008-06-23 02:23:15 +07:00
|
|
|
cond_resched();
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
wait->_qproc = NULL;
|
2008-08-31 22:26:40 +07:00
|
|
|
if (retval || timed_out || signal_pending(current))
|
2005-04-17 05:20:36 +07:00
|
|
|
break;
|
2008-04-22 05:15:06 +07:00
|
|
|
if (table.error) {
|
2005-04-17 05:20:36 +07:00
|
|
|
retval = table.error;
|
|
|
|
break;
|
|
|
|
}
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
/*
|
|
|
|
* If this is the first loop and we have a timeout
|
|
|
|
* given, then we convert to ktime_t and set the to
|
|
|
|
* pointer to the expiry value.
|
|
|
|
*/
|
|
|
|
if (end_time && !to) {
|
|
|
|
expire = timespec_to_ktime(*end_time);
|
|
|
|
to = &expire;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
2008-08-31 22:26:40 +07:00
|
|
|
|
2009-01-07 05:40:59 +07:00
|
|
|
if (!poll_schedule_timeout(&table, TASK_INTERRUPTIBLE,
|
|
|
|
to, slack))
|
2008-08-31 22:26:40 +07:00
|
|
|
timed_out = 1;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
poll_freewait(&table);
|
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We can actually return ERESTARTSYS instead of EINTR, but I'd
|
|
|
|
* like to be certain this leads to no problems. So I return
|
|
|
|
* EINTR just for safety.
|
|
|
|
*
|
|
|
|
* Update: ERESTARTSYS breaks at least the xview clock binary, so
|
|
|
|
* I'm trying ERESTARTNOHAND which restart only when you want to.
|
|
|
|
*/
|
2008-04-24 01:05:15 +07:00
|
|
|
int core_sys_select(int n, fd_set __user *inp, fd_set __user *outp,
|
2008-08-31 22:26:40 +07:00
|
|
|
fd_set __user *exp, struct timespec *end_time)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
fd_set_bits fds;
|
2006-04-11 12:52:46 +07:00
|
|
|
void *bits;
|
2006-12-10 17:21:12 +07:00
|
|
|
int ret, max_fds;
|
2006-04-11 12:54:08 +07:00
|
|
|
unsigned int size;
|
2005-09-10 03:04:10 +07:00
|
|
|
struct fdtable *fdt;
|
2006-03-28 16:56:33 +07:00
|
|
|
/* Allocate small arguments on the stack to save memory and be faster */
|
2006-03-31 23:18:57 +07:00
|
|
|
long stack_fds[SELECT_STACK_ALLOC/sizeof(long)];
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
if (n < 0)
|
|
|
|
goto out_nofds;
|
|
|
|
|
2006-12-10 17:21:12 +07:00
|
|
|
/* max_fds can increase, so grab it once to avoid race */
|
2005-09-10 03:04:14 +07:00
|
|
|
rcu_read_lock();
|
2005-09-10 03:04:10 +07:00
|
|
|
fdt = files_fdtable(current->files);
|
2006-12-10 17:21:12 +07:00
|
|
|
max_fds = fdt->max_fds;
|
2005-09-10 03:04:14 +07:00
|
|
|
rcu_read_unlock();
|
2006-12-10 17:21:12 +07:00
|
|
|
if (n > max_fds)
|
|
|
|
n = max_fds;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We need 6 bitmaps (in/out/ex for both incoming and outgoing),
|
|
|
|
* since we used fdset we need to allocate memory in units of
|
|
|
|
* long-words.
|
|
|
|
*/
|
|
|
|
size = FDS_BYTES(n);
|
2006-04-11 12:54:08 +07:00
|
|
|
bits = stack_fds;
|
|
|
|
if (size > sizeof(stack_fds) / 6) {
|
|
|
|
/* Not enough space in on-stack array; must use kmalloc */
|
|
|
|
ret = -ENOMEM;
|
2006-03-28 16:56:33 +07:00
|
|
|
bits = kmalloc(6 * size, GFP_KERNEL);
|
2006-04-11 12:54:08 +07:00
|
|
|
if (!bits)
|
|
|
|
goto out_nofds;
|
|
|
|
}
|
2006-04-11 12:52:46 +07:00
|
|
|
fds.in = bits;
|
|
|
|
fds.out = bits + size;
|
|
|
|
fds.ex = bits + 2*size;
|
|
|
|
fds.res_in = bits + 3*size;
|
|
|
|
fds.res_out = bits + 4*size;
|
|
|
|
fds.res_ex = bits + 5*size;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
if ((ret = get_fd_set(n, inp, fds.in)) ||
|
|
|
|
(ret = get_fd_set(n, outp, fds.out)) ||
|
|
|
|
(ret = get_fd_set(n, exp, fds.ex)))
|
|
|
|
goto out;
|
|
|
|
zero_fd_set(n, fds.res_in);
|
|
|
|
zero_fd_set(n, fds.res_out);
|
|
|
|
zero_fd_set(n, fds.res_ex);
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = do_select(n, &fds, end_time);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
if (ret < 0)
|
|
|
|
goto out;
|
|
|
|
if (!ret) {
|
|
|
|
ret = -ERESTARTNOHAND;
|
|
|
|
if (signal_pending(current))
|
|
|
|
goto out;
|
|
|
|
ret = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (set_fd_set(n, inp, fds.res_in) ||
|
|
|
|
set_fd_set(n, outp, fds.res_out) ||
|
|
|
|
set_fd_set(n, exp, fds.res_ex))
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
out:
|
2006-03-28 16:56:33 +07:00
|
|
|
if (bits != stack_fds)
|
|
|
|
kfree(bits);
|
2005-04-17 05:20:36 +07:00
|
|
|
out_nofds:
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2009-01-14 20:14:25 +07:00
|
|
|
SYSCALL_DEFINE5(select, int, n, fd_set __user *, inp, fd_set __user *, outp,
|
|
|
|
fd_set __user *, exp, struct timeval __user *, tvp)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
{
|
2008-08-31 22:26:40 +07:00
|
|
|
struct timespec end_time, *to = NULL;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
struct timeval tv;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (tvp) {
|
|
|
|
if (copy_from_user(&tv, tvp, sizeof(tv)))
|
|
|
|
return -EFAULT;
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
to = &end_time;
|
2008-10-26 02:41:41 +07:00
|
|
|
if (poll_select_set_timeout(to,
|
|
|
|
tv.tv_sec + (tv.tv_usec / USEC_PER_SEC),
|
|
|
|
(tv.tv_usec % USEC_PER_SEC) * NSEC_PER_USEC))
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = core_sys_select(n, inp, outp, exp, to);
|
|
|
|
ret = poll_select_copy_remaining(&end_time, tvp, 1, ret);
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2009-01-14 20:13:57 +07:00
|
|
|
static long do_pselect(int n, fd_set __user *inp, fd_set __user *outp,
|
|
|
|
fd_set __user *exp, struct timespec __user *tsp,
|
|
|
|
const sigset_t __user *sigmask, size_t sigsetsize)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
{
|
|
|
|
sigset_t ksigmask, sigsaved;
|
2008-08-31 22:26:40 +07:00
|
|
|
struct timespec ts, end_time, *to = NULL;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (tsp) {
|
|
|
|
if (copy_from_user(&ts, tsp, sizeof(ts)))
|
|
|
|
return -EFAULT;
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
to = &end_time;
|
|
|
|
if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec))
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sigmask) {
|
|
|
|
/* XXX: Don't preclude handling different sized sigset_t's. */
|
|
|
|
if (sigsetsize != sizeof(sigset_t))
|
|
|
|
return -EINVAL;
|
|
|
|
if (copy_from_user(&ksigmask, sigmask, sizeof(ksigmask)))
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
sigdelsetmask(&ksigmask, sigmask(SIGKILL)|sigmask(SIGSTOP));
|
|
|
|
sigprocmask(SIG_SETMASK, &ksigmask, &sigsaved);
|
|
|
|
}
|
|
|
|
|
2009-01-14 04:14:48 +07:00
|
|
|
ret = core_sys_select(n, inp, outp, exp, to);
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = poll_select_copy_remaining(&end_time, tsp, 0, ret);
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
|
|
|
if (ret == -ERESTARTNOHAND) {
|
|
|
|
/*
|
|
|
|
* Don't restore the signal mask yet. Let do_signal() deliver
|
|
|
|
* the signal on the way back to userspace, before the signal
|
|
|
|
* mask is restored.
|
|
|
|
*/
|
|
|
|
if (sigmask) {
|
|
|
|
memcpy(¤t->saved_sigmask, &sigsaved,
|
|
|
|
sizeof(sigsaved));
|
2008-04-30 14:53:06 +07:00
|
|
|
set_restore_sigmask();
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
} else if (sigmask)
|
|
|
|
sigprocmask(SIG_SETMASK, &sigsaved, NULL);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Most architectures can't handle 7-argument syscalls. So we provide a
|
|
|
|
* 6-argument version where the sixth argument is a pointer to a structure
|
|
|
|
* which has a pointer to the sigset_t itself followed by a size_t containing
|
|
|
|
* the sigset size.
|
|
|
|
*/
|
2009-01-14 20:14:34 +07:00
|
|
|
SYSCALL_DEFINE6(pselect6, int, n, fd_set __user *, inp, fd_set __user *, outp,
|
|
|
|
fd_set __user *, exp, struct timespec __user *, tsp,
|
|
|
|
void __user *, sig)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
{
|
|
|
|
size_t sigsetsize = 0;
|
|
|
|
sigset_t __user *up = NULL;
|
|
|
|
|
|
|
|
if (sig) {
|
|
|
|
if (!access_ok(VERIFY_READ, sig, sizeof(void *)+sizeof(size_t))
|
2006-02-01 17:26:09 +07:00
|
|
|
|| __get_user(up, (sigset_t __user * __user *)sig)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|| __get_user(sigsetsize,
|
2006-02-01 17:26:09 +07:00
|
|
|
(size_t __user *)(sig+sizeof(void *))))
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
return -EFAULT;
|
|
|
|
}
|
|
|
|
|
2009-01-14 20:13:57 +07:00
|
|
|
return do_pselect(n, inp, outp, exp, tsp, up, sigsetsize);
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
|
2010-03-11 06:21:13 +07:00
|
|
|
#ifdef __ARCH_WANT_SYS_OLD_SELECT
|
|
|
|
struct sel_arg_struct {
|
|
|
|
unsigned long n;
|
|
|
|
fd_set __user *inp, *outp, *exp;
|
|
|
|
struct timeval __user *tvp;
|
|
|
|
};
|
|
|
|
|
|
|
|
SYSCALL_DEFINE1(old_select, struct sel_arg_struct __user *, arg)
|
|
|
|
{
|
|
|
|
struct sel_arg_struct a;
|
|
|
|
|
|
|
|
if (copy_from_user(&a, arg, sizeof(a)))
|
|
|
|
return -EFAULT;
|
|
|
|
return sys_select(a.n, a.inp, a.outp, a.exp, a.tvp);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
struct poll_list {
|
|
|
|
struct poll_list *next;
|
|
|
|
int len;
|
|
|
|
struct pollfd entries[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
#define POLLFD_PER_PAGE ((PAGE_SIZE-sizeof(struct poll_list)) / sizeof(struct pollfd))
|
|
|
|
|
2006-06-23 16:05:16 +07:00
|
|
|
/*
|
|
|
|
* Fish for pollable events on the pollfd->fd file descriptor. We're only
|
|
|
|
* interested in events matching the pollfd->events mask, and the result
|
|
|
|
* matching that mask is both recorded in pollfd->revents and returned. The
|
|
|
|
* pwait poll_table will be used by the fd-provided poll handler for waiting,
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
* if pwait->_qproc is non-NULL.
|
2006-06-23 16:05:16 +07:00
|
|
|
*/
|
|
|
|
static inline unsigned int do_pollfd(struct pollfd *pollfd, poll_table *pwait)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2006-06-23 16:05:16 +07:00
|
|
|
unsigned int mask;
|
|
|
|
int fd;
|
|
|
|
|
|
|
|
mask = 0;
|
|
|
|
fd = pollfd->fd;
|
|
|
|
if (fd >= 0) {
|
|
|
|
int fput_needed;
|
|
|
|
struct file * file;
|
|
|
|
|
|
|
|
file = fget_light(fd, &fput_needed);
|
|
|
|
mask = POLLNVAL;
|
|
|
|
if (file != NULL) {
|
|
|
|
mask = DEFAULT_POLLMASK;
|
2009-06-17 05:33:36 +07:00
|
|
|
if (file->f_op && file->f_op->poll) {
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
pwait->_key = pollfd->events|POLLERR|POLLHUP;
|
2006-06-23 16:05:16 +07:00
|
|
|
mask = file->f_op->poll(file, pwait);
|
2009-06-17 05:33:36 +07:00
|
|
|
}
|
2006-06-23 16:05:16 +07:00
|
|
|
/* Mask out unneeded events. */
|
|
|
|
mask &= pollfd->events | POLLERR | POLLHUP;
|
|
|
|
fput_light(file, fput_needed);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
}
|
2006-06-23 16:05:16 +07:00
|
|
|
pollfd->revents = mask;
|
|
|
|
|
|
|
|
return mask;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static int do_poll(unsigned int nfds, struct poll_list *list,
|
2008-08-31 22:26:40 +07:00
|
|
|
struct poll_wqueues *wait, struct timespec *end_time)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
poll_table* pt = &wait->pt;
|
2008-08-31 22:26:40 +07:00
|
|
|
ktime_t expire, *to = NULL;
|
|
|
|
int timed_out = 0, count = 0;
|
2008-09-02 05:55:35 +07:00
|
|
|
unsigned long slack = 0;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
/* Optimise the no-wait case */
|
2008-08-31 22:26:40 +07:00
|
|
|
if (end_time && !end_time->tv_sec && !end_time->tv_nsec) {
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
pt->_qproc = NULL;
|
2008-08-31 22:26:40 +07:00
|
|
|
timed_out = 1;
|
|
|
|
}
|
2007-10-17 13:26:18 +07:00
|
|
|
|
2008-09-08 06:08:55 +07:00
|
|
|
if (end_time && !timed_out)
|
2010-10-28 05:34:53 +07:00
|
|
|
slack = select_estimate_accuracy(end_time);
|
2008-09-02 05:55:35 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
for (;;) {
|
|
|
|
struct poll_list *walk;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2006-06-23 16:05:16 +07:00
|
|
|
for (walk = list; walk != NULL; walk = walk->next) {
|
|
|
|
struct pollfd * pfd, * pfd_end;
|
|
|
|
|
|
|
|
pfd = walk->entries;
|
|
|
|
pfd_end = pfd + walk->len;
|
|
|
|
for (; pfd != pfd_end; pfd++) {
|
|
|
|
/*
|
|
|
|
* Fish for events. If we found one, record it
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
* and kill poll_table->_qproc, so we don't
|
2006-06-23 16:05:16 +07:00
|
|
|
* needlessly register any other waiters after
|
|
|
|
* this. They'll get immediately deregistered
|
|
|
|
* when we break out and return.
|
|
|
|
*/
|
|
|
|
if (do_pollfd(pfd, pt)) {
|
|
|
|
count++;
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
pt->_qproc = NULL;
|
2006-06-23 16:05:16 +07:00
|
|
|
}
|
|
|
|
}
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
2006-06-23 16:05:16 +07:00
|
|
|
/*
|
|
|
|
* All waiters have already been registered, so don't provide
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
* a poll_table->_qproc to them on the next loop iteration.
|
2006-06-23 16:05:16 +07:00
|
|
|
*/
|
poll: add poll_requested_events() and poll_does_not_wait() functions
In some cases the poll() implementation in a driver has to do different
things depending on the events the caller wants to poll for. An example
is when a driver needs to start a DMA engine if the caller polls for
POLLIN, but doesn't want to do that if POLLIN is not requested but instead
only POLLOUT or POLLPRI is requested. This is something that can happen
in the video4linux subsystem among others.
Unfortunately, the current epoll/poll/select implementation doesn't
provide that information reliably. The poll_table_struct does have it: it
has a key field with the event mask. But once a poll() call matches one
or more bits of that mask any following poll() calls are passed a NULL
poll_table pointer.
Also, the eventpoll implementation always left the key field at ~0 instead
of using the requested events mask.
This was changed in eventpoll.c so the key field now contains the actual
events that should be polled for as set by the caller.
The solution to the NULL poll_table pointer is to set the qproc field to
NULL in poll_table once poll() matches the events, not the poll_table
pointer itself. That way drivers can obtain the mask through a new
poll_requested_events inline.
The poll_table_struct can still be NULL since some kernel code calls it
internally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h). In
that case poll_requested_events() returns ~0 (i.e. all events).
Very rarely drivers might want to know whether poll_wait will actually
wait. If another earlier file descriptor in the set already matched the
events the caller wanted to wait for, then the kernel will return from the
select() call without waiting. This might be useful information in order
to avoid doing expensive work.
A new helper function poll_does_not_wait() is added that drivers can use
to detect this situation. This is now used in sock_poll_wait() in
include/net/sock.h. This was the only place in the kernel that needed
this information.
Drivers should no longer access any of the poll_table internals, but use
the poll_requested_events() and poll_does_not_wait() access functions
instead. In order to enforce that the poll_table fields are now prepended
with an underscore and a comment was added warning against using them
directly.
This required a change in unix_dgram_poll() in unix/af_unix.c which used
the key field to get the requested events. It's been replaced by a call
to poll_requested_events().
For qproc it was especially important to change its name since the
behavior of that field changes with this patch since this function pointer
can now be NULL when that wasn't possible in the past.
Any driver accessing the qproc or key fields directly will now fail to compile.
Some notes regarding the correctness of this patch: the driver's poll()
function is called with a 'struct poll_table_struct *wait' argument. This
pointer may or may not be NULL, drivers can never rely on it being one or
the other as that depends on whether or not an earlier file descriptor in
the select()'s fdset matched the requested events.
There are only three things a driver can do with the wait argument:
1) obtain the key field:
events = wait ? wait->key : ~0;
This will still work although it should be replaced with the new
poll_requested_events() function (which does exactly the same).
This will now even work better, since wait is no longer set to NULL
unnecessarily.
2) use the qproc callback. This could be deadly since qproc can now be
NULL. Renaming qproc should prevent this from happening. There are no
kernel drivers that actually access this callback directly, BTW.
3) test whether wait == NULL to determine whether poll would return without
waiting. This is no longer sufficient as the correct test is now
wait == NULL || wait->_qproc == NULL.
However, the worst that can happen here is a slight performance hit in
the case where wait != NULL and wait->_qproc == NULL. In that case the
driver will assume that poll_wait() will actually add the fd to the set
of waiting file descriptors. Of course, poll_wait() will not do that
since it tests for wait->_qproc. This will not break anything, though.
There is only one place in the whole kernel where this happens
(sock_poll_wait() in include/net/sock.h) and that code will be replaced
by a call to poll_does_not_wait() in the next patch.
Note that even if wait->_qproc != NULL drivers cannot rely on poll_wait()
actually waiting. The next file descriptor from the set might match the
event mask and thus any possible waits will never happen.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Jonathan Corbet <corbet@lwn.net>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-24 05:02:27 +07:00
|
|
|
pt->_qproc = NULL;
|
2007-10-17 13:26:18 +07:00
|
|
|
if (!count) {
|
|
|
|
count = wait->error;
|
|
|
|
if (signal_pending(current))
|
|
|
|
count = -EINTR;
|
|
|
|
}
|
2008-08-31 22:26:40 +07:00
|
|
|
if (count || timed_out)
|
2005-04-17 05:20:36 +07:00
|
|
|
break;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
/*
|
|
|
|
* If this is the first loop and we have a timeout
|
|
|
|
* given, then we convert to ktime_t and set the to
|
|
|
|
* pointer to the expiry value.
|
|
|
|
*/
|
|
|
|
if (end_time && !to) {
|
|
|
|
expire = timespec_to_ktime(*end_time);
|
|
|
|
to = &expire;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
|
2009-01-07 05:40:59 +07:00
|
|
|
if (!poll_schedule_timeout(wait, TASK_INTERRUPTIBLE, to, slack))
|
2008-08-31 22:26:40 +07:00
|
|
|
timed_out = 1;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
2006-03-28 16:56:33 +07:00
|
|
|
#define N_STACK_PPS ((sizeof(stack_pps) - sizeof(struct poll_list)) / \
|
|
|
|
sizeof(struct pollfd))
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
int do_sys_poll(struct pollfd __user *ufds, unsigned int nfds,
|
|
|
|
struct timespec *end_time)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
struct poll_wqueues table;
|
2007-10-17 13:26:17 +07:00
|
|
|
int err = -EFAULT, fdcount, len, size;
|
2006-03-31 23:18:57 +07:00
|
|
|
/* Allocate small arguments on the stack to save memory and be
|
|
|
|
faster - use long to make sure the buffer is aligned properly
|
|
|
|
on 64 bit archs to avoid unaligned access */
|
|
|
|
long stack_pps[POLL_STACK_ALLOC/sizeof(long)];
|
2007-10-17 13:26:17 +07:00
|
|
|
struct poll_list *const head = (struct poll_list *)stack_pps;
|
|
|
|
struct poll_list *walk = head;
|
|
|
|
unsigned long todo = nfds;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2010-03-06 04:42:42 +07:00
|
|
|
if (nfds > rlimit(RLIMIT_NOFILE))
|
2005-04-17 05:20:36 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
len = min_t(unsigned int, nfds, N_STACK_PPS);
|
|
|
|
for (;;) {
|
|
|
|
walk->next = NULL;
|
|
|
|
walk->len = len;
|
|
|
|
if (!len)
|
|
|
|
break;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
if (copy_from_user(walk->entries, ufds + nfds-todo,
|
|
|
|
sizeof(struct pollfd) * walk->len))
|
|
|
|
goto out_fds;
|
|
|
|
|
|
|
|
todo -= walk->len;
|
|
|
|
if (!todo)
|
|
|
|
break;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
len = min(todo, POLLFD_PER_PAGE);
|
|
|
|
size = sizeof(struct poll_list) + sizeof(struct pollfd) * len;
|
|
|
|
walk = walk->next = kmalloc(size, GFP_KERNEL);
|
|
|
|
if (!walk) {
|
|
|
|
err = -ENOMEM;
|
2005-04-17 05:20:36 +07:00
|
|
|
goto out_fds;
|
|
|
|
}
|
|
|
|
}
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
poll_initwait(&table);
|
2008-08-31 22:26:40 +07:00
|
|
|
fdcount = do_poll(nfds, head, &table, end_time);
|
2007-10-17 13:26:17 +07:00
|
|
|
poll_freewait(&table);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
for (walk = head; walk; walk = walk->next) {
|
2005-04-17 05:20:36 +07:00
|
|
|
struct pollfd *fds = walk->entries;
|
|
|
|
int j;
|
|
|
|
|
2007-10-17 13:26:17 +07:00
|
|
|
for (j = 0; j < walk->len; j++, ufds++)
|
|
|
|
if (__put_user(fds[j].revents, &ufds->revents))
|
2005-04-17 05:20:36 +07:00
|
|
|
goto out_fds;
|
|
|
|
}
|
2007-10-17 13:26:17 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
err = fdcount;
|
|
|
|
out_fds:
|
2007-10-17 13:26:17 +07:00
|
|
|
walk = head->next;
|
|
|
|
while (walk) {
|
|
|
|
struct poll_list *pos = walk;
|
|
|
|
walk = walk->next;
|
|
|
|
kfree(pos);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
2007-10-17 13:26:17 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
return err;
|
|
|
|
}
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2007-10-17 13:27:18 +07:00
|
|
|
static long do_restart_poll(struct restart_block *restart_block)
|
|
|
|
{
|
2008-08-31 22:26:40 +07:00
|
|
|
struct pollfd __user *ufds = restart_block->poll.ufds;
|
|
|
|
int nfds = restart_block->poll.nfds;
|
|
|
|
struct timespec *to = NULL, end_time;
|
2007-10-17 13:27:18 +07:00
|
|
|
int ret;
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
if (restart_block->poll.has_timeout) {
|
|
|
|
end_time.tv_sec = restart_block->poll.tv_sec;
|
|
|
|
end_time.tv_nsec = restart_block->poll.tv_nsec;
|
|
|
|
to = &end_time;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = do_sys_poll(ufds, nfds, to);
|
|
|
|
|
2007-10-17 13:27:18 +07:00
|
|
|
if (ret == -EINTR) {
|
|
|
|
restart_block->fn = do_restart_poll;
|
|
|
|
ret = -ERESTART_RESTARTBLOCK;
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2009-01-14 20:14:25 +07:00
|
|
|
SYSCALL_DEFINE3(poll, struct pollfd __user *, ufds, unsigned int, nfds,
|
sys_poll: fix incorrect type for 'timeout' parameter
The 'poll()' system call timeout parameter is supposed to be 'int', not
'long'.
Now, the reason this matters is that right now 32-bit compat mode is
broken on at least x86-64, because the 32-bit code just calls
'sys_poll()' directly on x86-64, and the 32-bit argument will have been
zero-extended, turning a signed 'int' into a large unsigned 'long'
value.
We could just introduce a 'compat_sys_poll()' function for this, and
that may eventually be what we have to do, but since the actual standard
poll() semantics is *supposed* to be 'int', and since at least on x86-64
glibc sign-extends the argument before invocing the system call (so
nobody can actually use a 64-bit timeout value in user space _anyway_,
even in 64-bit binaries), the simpler solution would seem to be to just
fix the definition of the system call to match what it should have been
from the very start.
If it turns out that somebody somehow circumvents the user-level libc
64-bit sign extension and actually uses a large unsigned 64-bit timeout
despite that not being how poll() is supposed to work, we will need to
do the compat_sys_poll() approach.
Reported-by: Thomas Meyer <thomas@m3y3r.de>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-02-22 08:24:20 +07:00
|
|
|
int, timeout_msecs)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
{
|
2008-08-31 22:26:40 +07:00
|
|
|
struct timespec end_time, *to = NULL;
|
2007-10-17 13:27:18 +07:00
|
|
|
int ret;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
if (timeout_msecs >= 0) {
|
|
|
|
to = &end_time;
|
|
|
|
poll_select_set_timeout(to, timeout_msecs / MSEC_PER_SEC,
|
|
|
|
NSEC_PER_MSEC * (timeout_msecs % MSEC_PER_SEC));
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = do_sys_poll(ufds, nfds, to);
|
|
|
|
|
2007-10-17 13:27:18 +07:00
|
|
|
if (ret == -EINTR) {
|
|
|
|
struct restart_block *restart_block;
|
2008-08-31 22:26:40 +07:00
|
|
|
|
2007-10-17 13:27:18 +07:00
|
|
|
restart_block = ¤t_thread_info()->restart_block;
|
|
|
|
restart_block->fn = do_restart_poll;
|
2008-08-31 22:26:40 +07:00
|
|
|
restart_block->poll.ufds = ufds;
|
|
|
|
restart_block->poll.nfds = nfds;
|
|
|
|
|
|
|
|
if (timeout_msecs >= 0) {
|
|
|
|
restart_block->poll.tv_sec = end_time.tv_sec;
|
|
|
|
restart_block->poll.tv_nsec = end_time.tv_nsec;
|
|
|
|
restart_block->poll.has_timeout = 1;
|
|
|
|
} else
|
|
|
|
restart_block->poll.has_timeout = 0;
|
|
|
|
|
2007-10-17 13:27:18 +07:00
|
|
|
ret = -ERESTART_RESTARTBLOCK;
|
|
|
|
}
|
|
|
|
return ret;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
|
2009-01-14 20:14:34 +07:00
|
|
|
SYSCALL_DEFINE5(ppoll, struct pollfd __user *, ufds, unsigned int, nfds,
|
|
|
|
struct timespec __user *, tsp, const sigset_t __user *, sigmask,
|
|
|
|
size_t, sigsetsize)
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
{
|
|
|
|
sigset_t ksigmask, sigsaved;
|
2008-08-31 22:26:40 +07:00
|
|
|
struct timespec ts, end_time, *to = NULL;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (tsp) {
|
|
|
|
if (copy_from_user(&ts, tsp, sizeof(ts)))
|
|
|
|
return -EFAULT;
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
to = &end_time;
|
|
|
|
if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec))
|
|
|
|
return -EINVAL;
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
if (sigmask) {
|
|
|
|
/* XXX: Don't preclude handling different sized sigset_t's. */
|
|
|
|
if (sigsetsize != sizeof(sigset_t))
|
|
|
|
return -EINVAL;
|
|
|
|
if (copy_from_user(&ksigmask, sigmask, sizeof(ksigmask)))
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
sigdelsetmask(&ksigmask, sigmask(SIGKILL)|sigmask(SIGSTOP));
|
|
|
|
sigprocmask(SIG_SETMASK, &ksigmask, &sigsaved);
|
|
|
|
}
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = do_sys_poll(ufds, nfds, to);
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
|
|
|
/* We can restart this syscall, usually */
|
|
|
|
if (ret == -EINTR) {
|
|
|
|
/*
|
|
|
|
* Don't restore the signal mask yet. Let do_signal() deliver
|
|
|
|
* the signal on the way back to userspace, before the signal
|
|
|
|
* mask is restored.
|
|
|
|
*/
|
|
|
|
if (sigmask) {
|
|
|
|
memcpy(¤t->saved_sigmask, &sigsaved,
|
|
|
|
sizeof(sigsaved));
|
2008-04-30 14:53:06 +07:00
|
|
|
set_restore_sigmask();
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
}
|
|
|
|
ret = -ERESTARTNOHAND;
|
|
|
|
} else if (sigmask)
|
|
|
|
sigprocmask(SIG_SETMASK, &sigsaved, NULL);
|
|
|
|
|
2008-08-31 22:26:40 +07:00
|
|
|
ret = poll_select_copy_remaining(&end_time, tsp, 0, ret);
|
[PATCH] Add pselect/ppoll system call implementation
The following implementation of ppoll() and pselect() system calls
depends on the architecture providing a TIF_RESTORE_SIGMASK flag in the
thread_info.
These system calls have to change the signal mask during their
operation, and signal handlers must be invoked using the new, temporary
signal mask. The old signal mask must be restored either upon successful
exit from the system call, or upon returning from the invoked signal
handler if the system call is interrupted. We can't simply restore the
original signal mask and return to userspace, since the restored signal
mask may actually block the signal which interrupted the system call.
The TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit
path to trap into do_signal() just as TIF_SIGPENDING does, and by
causing do_signal() to use the saved signal mask instead of the current
signal mask when setting up the stack frame for the signal handler -- or
by causing do_signal() to simply restore the saved signal mask in the
case where there is no handler to be invoked.
The first patch implements the sys_pselect() and sys_ppoll() system
calls, which are present only if TIF_RESTORE_SIGMASK is defined. That
#ifdef should go away in time when all architectures have implemented
it. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC
kernel (in the -mm tree), and the third patch then removes the
arch-specific implementations of sys_rt_sigsuspend() and replaces them
with generic versions using the same trick.
The fourth and fifth patches, provided by David Howells, implement
TIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch
adds the syscalls to the i386 syscall table.
This patch:
Add the pselect() and ppoll() system calls, providing core routines usable by
the original select() and poll() system calls and also the new calls (with
their semantics w.r.t timeouts).
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-19 08:44:05 +07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|