epoll: drop max_user_instances and rely only on max_user_watches

commit 9df04e1f25effde823a600e755b51475d438f56b upstream.

Linus suggested to put limits where the money is, and max_user_watches
already does that w/out the need of max_user_instances.  That has the
advantage to mitigate the potential DoS while allowing pretty generous
default behavior.

Allowing top 4% of low memory (per user) to be allocated in epoll watches,
we have:

LOMEM    MAX_WATCHES (per user)
512MB    ~178000
1GB      ~356000
2GB      ~712000

A box with 512MB of lomem, will meet some challenge in hitting 180K
watches, socket buffers math teaches us.  No more max_user_instances
limits then.

Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Michael Kerrisk <mtk.manpages@googlemail.com>
Cc: Bron Gondwana <brong@fastmail.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index cb60f9227c5023189a18bad8644c73c8270f62ed..801de2cada636df18aac277927321e46c5b381a2 100644 (file)
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -234,8 +234,6 @@ struct ep_pqueue {
 /*
  * Configuration options available inside /proc/sys/fs/epoll/
  */
-/* Maximum number of epoll devices, per user */
-static int max_user_instances __read_mostly;
 /* Maximum number of epoll watched descriptors, per user */
 static int max_user_watches __read_mostly;
 
@@ -260,14 +258,6 @@ static struct kmem_cache *pwq_cache __read_mostly;
 static int zero;
 
 ctl_table epoll_table[] = {
-       {
-               .procname       = "max_user_instances",
-               .data           = &max_user_instances,
-               .maxlen         = sizeof(int),
-               .mode           = 0644,
-               .proc_handler   = &proc_dointvec_minmax,
-               .extra1         = &zero,
-       },
        {
                .procname       = "max_user_watches",
                .data           = &max_user_watches,
@@ -491,7 +481,6 @@ static void ep_free(struct eventpoll *ep)
 
        mutex_unlock(&epmutex);
        mutex_destroy(&ep->mtx);
-       atomic_dec(&ep->user->epoll_devs);
        free_uid(ep->user);
        kfree(ep);
 }
@@ -581,10 +570,6 @@ static int ep_alloc(struct eventpoll **pep)
        struct eventpoll *ep;
 
        user = get_current_user();
-       error = -EMFILE;
-       if (unlikely(atomic_read(&user->epoll_devs) >=
-                       max_user_instances))
-               goto free_uid;
        error = -ENOMEM;
        ep = kzalloc(sizeof(*ep), GFP_KERNEL);
        if (unlikely(!ep))
@@ -1137,7 +1122,6 @@ SYSCALL_DEFINE1(epoll_create1, int, flags)
                              flags & O_CLOEXEC);
        if (fd < 0)
                ep_free(ep);
-       atomic_inc(&ep->user->epoll_devs);
 
 error_return:
        DNPRINTK(3, (KERN_INFO "[%p] eventpoll: sys_epoll_create(%d) = %d\n",
@@ -1362,8 +1346,10 @@ static int __init eventpoll_init(void)
        struct sysinfo si;
 
        si_meminfo(&si);
-       max_user_instances = 128;
-       max_user_watches = (((si.totalram - si.totalhigh) / 32) << PAGE_SHIFT) /
+       /*
+        * Allows top 4% of lomem to be allocated for epoll watches (per user).
+        */
+       max_user_watches = (((si.totalram - si.totalhigh) / 25) << PAGE_SHIFT) /
                EP_ITEM_COST;
 
        /* Initialize the structure used to perform safe poll wait head wake ups */

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 086f5e1aee60be07c818eaea0087a238dfd7a46a..03e09027f730767e8bebf3bf8eee6c1d90548e46 100644 (file)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -588,7 +588,6 @@ struct user_struct {
        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 #endif
 #ifdef CONFIG_EPOLL
-       atomic_t epoll_devs;    /* The number of epoll descriptors currently open */
        atomic_t epoll_watches; /* The number of file descriptors currently watched */
 #endif
 #ifdef CONFIG_POSIX_MQUEUE