From: Al Viro <viro@ZenIV.linux.org.uk>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
Marc Zyngier <maz@kernel.org>
Subject: [RFC PATCH 18/27] lift locking/unlocking ep->mtx out of ep_{start,done}_scan()
Date: Sun, 4 Oct 2020 03:39:20 +0100 [thread overview]
Message-ID: <20201004023929.2740074-18-viro@ZenIV.linux.org.uk> (raw)
In-Reply-To: <20201004023929.2740074-1-viro@ZenIV.linux.org.uk>
From: Al Viro <viro@zeniv.linux.org.uk>
get rid of depth/ep_locked arguments there and document
the kludge in ep_item_poll() that has lead to ep_locked existence in
the first place
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
fs/eventpoll.c | 57 ++++++++++++++++++++++++++-------------------------------
1 file changed, 26 insertions(+), 31 deletions(-)
diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index ac996b959e94..f9c567af1f5f 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -554,20 +554,13 @@ static inline void ep_pm_stay_awake_rcu(struct epitem *epi)
rcu_read_unlock();
}
-static void ep_start_scan(struct eventpoll *ep,
- int depth, bool ep_locked,
- struct list_head *txlist)
-{
- lockdep_assert_irqs_enabled();
-
- /*
- * We need to lock this because we could be hit by
- * eventpoll_release_file() and epoll_ctl().
- */
-
- if (!ep_locked)
- mutex_lock_nested(&ep->mtx, depth);
+/*
+ * ep->mutex needs to be held because we could be hit by
+ * eventpoll_release_file() and epoll_ctl().
+ */
+static void ep_start_scan(struct eventpoll *ep, struct list_head *txlist)
+{
/*
* Steal the ready list, and re-init the original one to the
* empty list. Also, set ep->ovflist to NULL so that events
@@ -576,6 +569,7 @@ static void ep_start_scan(struct eventpoll *ep,
* because we want the "sproc" callback to be able to do it
* in a lockless way.
*/
+ lockdep_assert_irqs_enabled();
write_lock_irq(&ep->lock);
list_splice_init(&ep->rdllist, txlist);
WRITE_ONCE(ep->ovflist, NULL);
@@ -583,7 +577,6 @@ static void ep_start_scan(struct eventpoll *ep,
}
static void ep_done_scan(struct eventpoll *ep,
- int depth, bool ep_locked,
struct list_head *txlist)
{
struct epitem *epi, *nepi;
@@ -624,9 +617,6 @@ static void ep_done_scan(struct eventpoll *ep,
list_splice(txlist, &ep->rdllist);
__pm_relax(ep->ws);
write_unlock_irq(&ep->lock);
-
- if (!ep_locked)
- mutex_unlock(&ep->mtx);
}
static void epi_rcu_free(struct rcu_head *head)
@@ -763,11 +753,16 @@ static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt,
ep = epi->ffd.file->private_data;
poll_wait(epi->ffd.file, &ep->poll_wait, pt);
- locked = pt && (pt->_qproc == ep_ptable_queue_proc);
- ep_start_scan(ep, depth, locked, &txlist);
+ // kludge: ep_insert() calls us with ep->mtx already locked
+ locked = pt && (pt->_qproc == ep_ptable_queue_proc);
+ if (!locked)
+ mutex_lock_nested(&ep->mtx, depth);
+ ep_start_scan(ep, &txlist);
res = ep_read_events_proc(ep, &txlist, depth + 1);
- ep_done_scan(ep, depth, locked, &txlist);
+ ep_done_scan(ep, &txlist);
+ if (!locked)
+ mutex_unlock(&ep->mtx);
return res & epi->event.events;
}
@@ -809,9 +804,11 @@ static __poll_t ep_eventpoll_poll(struct file *file, poll_table *wait)
* Proceed to find out if wanted events are really available inside
* the ready list.
*/
- ep_start_scan(ep, 0, false, &txlist);
+ mutex_lock(&ep->mtx);
+ ep_start_scan(ep, &txlist);
res = ep_read_events_proc(ep, &txlist, 1);
- ep_done_scan(ep, 0, false, &txlist);
+ ep_done_scan(ep, &txlist);
+ mutex_unlock(&ep->mtx);
return res;
}
@@ -1573,15 +1570,13 @@ static int ep_send_events(struct eventpoll *ep,
init_poll_funcptr(&pt, NULL);
- ep_start_scan(ep, 0, false, &txlist);
+ mutex_lock(&ep->mtx);
+ ep_start_scan(ep, &txlist);
/*
* We can loop without lock because we are passed a task private list.
- * Items cannot vanish during the loop because ep_scan_ready_list() is
- * holding "mtx" during this call.
+ * Items cannot vanish during the loop we are holding ep->mtx.
*/
- lockdep_assert_held(&ep->mtx);
-
list_for_each_entry_safe(epi, tmp, &txlist, rdllink) {
struct wakeup_source *ws;
__poll_t revents;
@@ -1609,9 +1604,8 @@ static int ep_send_events(struct eventpoll *ep,
/*
* If the event mask intersect the caller-requested one,
- * deliver the event to userspace. Again, ep_scan_ready_list()
- * is holding ep->mtx, so no operations coming from userspace
- * can change the item.
+ * deliver the event to userspace. Again, we are holding ep->mtx,
+ * so no operations coming from userspace can change the item.
*/
revents = ep_item_poll(epi, &pt, 1);
if (!revents)
@@ -1645,7 +1639,8 @@ static int ep_send_events(struct eventpoll *ep,
ep_pm_stay_awake(epi);
}
}
- ep_done_scan(ep, 0, false, &txlist);
+ ep_done_scan(ep, &txlist);
+ mutex_unlock(&ep->mtx);
return res;
}
--
2.11.0
next prev parent reply other threads:[~2020-10-04 2:39 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-04 2:36 [RFC][PATCHSET] epoll cleanups Al Viro
2020-10-04 2:39 ` [RFC PATCH 01/27] epoll: switch epitem->pwqlist to single-linked list Al Viro
2020-10-04 2:39 ` [RFC PATCH 02/27] epoll: get rid of epitem->nwait Al Viro
2020-10-04 2:39 ` [RFC PATCH 03/27] untangling ep_call_nested(): get rid of useless arguments Al Viro
2020-10-04 2:39 ` [RFC PATCH 04/27] untangling ep_call_nested(): it's all serialized on epmutex Al Viro
2020-10-04 2:39 ` [RFC PATCH 05/27] untangling ep_call_nested(): take pushing cookie into a helper Al Viro
2020-10-04 2:39 ` [RFC PATCH 06/27] untangling ep_call_nested(): move push/pop of cookie into the callbacks Al Viro
2020-10-04 2:39 ` [RFC PATCH 07/27] untangling ep_call_nested(): and there was much rejoicing Al Viro
2020-10-04 2:39 ` [RFC PATCH 08/27] reverse_path_check_proc(): sane arguments Al Viro
2020-10-04 2:39 ` [RFC PATCH 09/27] reverse_path_check_proc(): don't bother with cookies Al Viro
2020-10-04 2:39 ` [RFC PATCH 10/27] clean reverse_path_check_proc() a bit Al Viro
2020-10-04 2:39 ` [RFC PATCH 11/27] ep_loop_check_proc(): lift pushing the cookie into callers Al Viro
2020-10-04 2:39 ` [RFC PATCH 12/27] get rid of ep_push_nested() Al Viro
2020-10-04 2:39 ` [RFC PATCH 13/27] ep_loop_check_proc(): saner calling conventions Al Viro
2020-10-04 2:39 ` [RFC PATCH 14/27] ep_scan_ready_list(): prepare to splitup Al Viro
2020-10-04 2:39 ` [RFC PATCH 15/27] lift the calls of ep_read_events_proc() into the callers Al Viro
2020-10-04 2:39 ` [RFC PATCH 16/27] lift the calls of ep_send_events_proc() " Al Viro
2020-10-04 2:39 ` [RFC PATCH 17/27] ep_send_events_proc(): fold into the caller Al Viro
2020-10-04 2:39 ` Al Viro [this message]
2020-10-04 2:39 ` [RFC PATCH 19/27] ep_insert(): don't open-code ep_remove() on failure exits Al Viro
2020-10-04 2:39 ` [RFC PATCH 20/27] ep_insert(): we only need tep->mtx around the insertion itself Al Viro
2020-10-04 12:56 ` [ep_insert()] 9ee1cc5666: WARNING:possible_recursive_locking_detected kernel test robot
2020-10-04 14:17 ` Al Viro
2020-10-04 14:27 ` Al Viro
2020-10-04 2:39 ` [RFC PATCH 21/27] take the common part of ep_eventpoll_poll() and ep_item_poll() into helper Al Viro
2020-10-04 2:39 ` [RFC PATCH 22/27] fold ep_read_events_proc() into the only caller Al Viro
2020-10-04 2:39 ` [RFC PATCH 23/27] ep_insert(): move creation of wakeup source past the fl_ep_links insertion Al Viro
2020-10-04 2:39 ` [RFC PATCH 24/27] convert ->f_ep_links/->fllink to hlist Al Viro
2020-10-04 2:39 ` [RFC PATCH 25/27] lift rcu_read_lock() into reverse_path_check() Al Viro
2020-10-04 2:39 ` [RFC PATCH 26/27] epoll: massage the check list insertion Al Viro
2020-10-04 2:39 ` [RFC PATCH 27/27] epoll: take epitem list out of struct file Al Viro
2020-10-05 20:37 ` Qian Cai
2020-10-05 20:49 ` Al Viro
2020-10-04 2:49 ` [RFC][PATCHSET] epoll cleanups Al Viro
2020-10-04 12:13 ` Matthew Wilcox
2020-10-04 14:15 ` Al Viro
2020-10-04 18:08 ` Linus Torvalds
2020-10-04 20:05 ` Al Viro
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201004023929.2740074-18-viro@ZenIV.linux.org.uk \
--to=viro@zeniv.linux.org.uk \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=maz@kernel.org \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).