Revision d80e731ecab420ddcb79ee9d0ac427acbc187b4b authored by Oleg Nesterov on 24 February 2012, 19:07:11 UTC, committed by Linus Torvalds on 24 February 2012, 19:42:50 UTC
This patch is intentionally incomplete to simplify the review. It ignores ep_unregister_pollwait() which plays with the same wqh. See the next change. epoll assumes that the EPOLL_CTL_ADD'ed file controls everything f_op->poll() needs. In particular it assumes that the wait queue can't go away until eventpoll_release(). This is not true in case of signalfd, the task which does EPOLL_CTL_ADD uses its ->sighand which is not connected to the file. This patch adds the special event, POLLFREE, currently only for epoll. It expects that init_poll_funcptr()'ed hook should do the necessary cleanup. Perhaps it should be defined as EPOLLFREE in eventpoll. __cleanup_sighand() is changed to do wake_up_poll(POLLFREE) if ->signalfd_wqh is not empty, we add the new signalfd_cleanup() helper. ep_poll_callback(POLLFREE) simply does list_del_init(task_list). This make this poll entry inconsistent, but we don't care. If you share epoll fd which contains our sigfd with another process you should blame yourself. signalfd is "really special". I simply do not know how we can define the "right" semantics if it used with epoll. The main problem is, epoll calls signalfd_poll() once to establish the connection with the wait queue, after that signalfd_poll(NULL) returns the different/inconsistent results depending on who does EPOLL_CTL_MOD/signalfd_read/etc. IOW: apart from sigmask, signalfd has nothing to do with the file, it works with the current thread. In short: this patch is the hack which tries to fix the symptoms. It also assumes that nobody can take tasklist_lock under epoll locks, this seems to be true. Note: - we do not have wake_up_all_poll() but wake_up_poll() is fine, poll/epoll doesn't use WQ_FLAG_EXCLUSIVE. - signalfd_cleanup() uses POLLHUP along with POLLFREE, we need a couple of simple changes in eventpoll.c to make sure it can't be "lost". Reported-by: Maxime Bizon <mbizon@freebox.fr> Cc: <stable@kernel.org> Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent 855a85f
thrash.c
/*
* mm/thrash.c
*
* Copyright (C) 2004, Red Hat, Inc.
* Copyright (C) 2004, Rik van Riel <riel@redhat.com>
* Released under the GPL, see the file COPYING for details.
*
* Simple token based thrashing protection, using the algorithm
* described in: http://www.cse.ohio-state.edu/hpcs/WWW/HTML/publications/abs05-1.html
*
* Sep 2006, Ashwin Chaugule <ashwin.chaugule@celunite.com>
* Improved algorithm to pass token:
* Each task has a priority which is incremented if it contended
* for the token in an interval less than its previous attempt.
* If the token is acquired, that task's priority is boosted to prevent
* the token from bouncing around too often and to let the task make
* some progress in its execution.
*/
#include <linux/jiffies.h>
#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/swap.h>
#include <linux/memcontrol.h>
#include <trace/events/vmscan.h>
#define TOKEN_AGING_INTERVAL (0xFF)
static DEFINE_SPINLOCK(swap_token_lock);
struct mm_struct *swap_token_mm;
static struct mem_cgroup *swap_token_memcg;
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
static struct mem_cgroup *swap_token_memcg_from_mm(struct mm_struct *mm)
{
struct mem_cgroup *memcg;
memcg = try_get_mem_cgroup_from_mm(mm);
if (memcg)
css_put(mem_cgroup_css(memcg));
return memcg;
}
#else
static struct mem_cgroup *swap_token_memcg_from_mm(struct mm_struct *mm)
{
return NULL;
}
#endif
void grab_swap_token(struct mm_struct *mm)
{
int current_interval;
unsigned int old_prio = mm->token_priority;
static unsigned int global_faults;
static unsigned int last_aging;
global_faults++;
current_interval = global_faults - mm->faultstamp;
if (!spin_trylock(&swap_token_lock))
return;
/* First come first served */
if (!swap_token_mm)
goto replace_token;
/*
* Usually, we don't need priority aging because long interval faults
* makes priority decrease quickly. But there is one exception. If the
* token owner task is sleeping, it never make long interval faults.
* Thus, we need a priority aging mechanism instead. The requirements
* of priority aging are
* 1) An aging interval is reasonable enough long. Too short aging
* interval makes quick swap token lost and decrease performance.
* 2) The swap token owner task have to get priority aging even if
* it's under sleep.
*/
if ((global_faults - last_aging) > TOKEN_AGING_INTERVAL) {
swap_token_mm->token_priority /= 2;
last_aging = global_faults;
}
if (mm == swap_token_mm) {
mm->token_priority += 2;
goto update_priority;
}
if (current_interval < mm->last_interval)
mm->token_priority++;
else {
if (likely(mm->token_priority > 0))
mm->token_priority--;
}
/* Check if we deserve the token */
if (mm->token_priority > swap_token_mm->token_priority)
goto replace_token;
update_priority:
trace_update_swap_token_priority(mm, old_prio, swap_token_mm);
out:
mm->faultstamp = global_faults;
mm->last_interval = current_interval;
spin_unlock(&swap_token_lock);
return;
replace_token:
mm->token_priority += 2;
trace_replace_swap_token(swap_token_mm, mm);
swap_token_mm = mm;
swap_token_memcg = swap_token_memcg_from_mm(mm);
last_aging = global_faults;
goto out;
}
/* Called on process exit. */
void __put_swap_token(struct mm_struct *mm)
{
spin_lock(&swap_token_lock);
if (likely(mm == swap_token_mm)) {
trace_put_swap_token(swap_token_mm);
swap_token_mm = NULL;
swap_token_memcg = NULL;
}
spin_unlock(&swap_token_lock);
}
static bool match_memcg(struct mem_cgroup *a, struct mem_cgroup *b)
{
if (!a)
return true;
if (!b)
return true;
if (a == b)
return true;
return false;
}
void disable_swap_token(struct mem_cgroup *memcg)
{
/* memcg reclaim don't disable unrelated mm token. */
if (match_memcg(memcg, swap_token_memcg)) {
spin_lock(&swap_token_lock);
if (match_memcg(memcg, swap_token_memcg)) {
trace_disable_swap_token(swap_token_mm);
swap_token_mm = NULL;
swap_token_memcg = NULL;
}
spin_unlock(&swap_token_lock);
}
}
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...