Add syscall epoll_pwait2, an epoll_wait variant with nsec resolution that
replaces int timeout with struct timespec. It is equivalent otherwise.
int epoll_pwait2(int fd, struct epoll_event *events,
int maxevents,
const struct timespec *timeout,
const sigset_t *sigset);
The underlying hrtimer is already programmed with nsec resolution.
pselect and ppoll also set nsec resolution timeout with timespec.
The sigset_t in epoll_pwait has a compat variant. epoll_pwait2 needs
the same.
For timespec, only support this new interface on 2038 aware platforms
that define __kernel_timespec_t. So no CONFIG_COMPAT_32BIT_TIME.
Link: https://lkml.kernel.org/r/20201121144401.3727659-3-willemdebruijn.kernel@gmail.com
Signed-off-by: Willem de Bruijn <willemb@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "add epoll_pwait2 syscall", v4.
Enable nanosecond timeouts for epoll.
Analogous to pselect and ppoll, introduce an epoll_wait syscall
variant that takes a struct timespec instead of int timeout.
This patch (of 4):
Make epoll more consistent with select/poll: pass along the timeout as
timespec64 pointer.
In anticipation of additional changes affecting all three polling
mechanisms:
- add epoll_pwait2 syscall with timespec semantics,
and share poll_select_set_timeout implementation.
- compute slack before conversion to absolute time,
to save one ktime_get_ts64 call.
Link: https://lkml.kernel.org/r/20201121144401.3727659-1-willemdebruijn.kernel@gmail.com
Link: https://lkml.kernel.org/r/20201121144401.3727659-2-willemdebruijn.kernel@gmail.com
Signed-off-by: Willem de Bruijn <willemb@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We call ep_events_available() under lock when timeout is 0, and then call
it without locks in the loop for the other cases.
Instead, call ep_events_available() without lock for all cases. For
non-zero timeouts, we will recheck after adding the thread to the wait
queue. For zero timeout cases, by definition, user is opportunistically
polling and will have to call epoll_wait again in the future.
Note that this lock was kept in c5a282e963 because the whole loop was
historically under lock.
This patch results in a 1% CPU/RPC reduction in RPC benchmarks.
Link: https://lkml.kernel.org/r/20201106231635.3528496-9-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The existing loop is pointless, and the labels make it really hard to
follow the structure.
Replace that control structure with a simple loop that returns when there
are new events, there is a signal, or the thread has timed out.
Link: https://lkml.kernel.org/r/20201106231635.3528496-8-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is a no-op change which simplifies the follow up patches.
Link: https://lkml.kernel.org/r/20201106231635.3528496-7-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ep_events_available() is called multiple times around the busy loop logic,
even though the logic is generally not used. ep_reset_busy_poll_napi_id()
is similarly always called, even when busy loop is not used.
Eliminate ep_reset_busy_poll_napi_id() and inline it inside
ep_busy_loop(). Make ep_busy_loop() return whether there are any events
available after the busy loop. This will eliminate unnecessary loads and
branches, and simplifies the loop.
Link: https://lkml.kernel.org/r/20201106231635.3528496-6-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is a no-op change and simply to make the code more coherent.
Link: https://lkml.kernel.org/r/20201106231635.3528496-5-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
To simplify the code, pull in checking the fatal signals into
ep_send_events(). ep_send_events() is called only from ep_poll().
Note that, previously, we were always checking fatal events, but it is
checked only if eavail is true. This should be fine because the goal of
that check is to quickly return from epoll_wait() when there is a pending
fatal signal.
Link: https://lkml.kernel.org/r/20201106231635.3528496-4-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Check signals before locking ep->lock, and immediately return -EINTR if
there is any signal pending.
This saves a few loads, stores, and branches from the hot path and
simplifies the loop structure for follow up patches.
Link: https://lkml.kernel.org/r/20201106231635.3528496-3-soheil.kdev@gmail.com
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Cc: Guantao Liu <guantaol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "simplify ep_poll".
This patch series is a followup based on the suggestions and feedback by
Linus:
https://lkml.kernel.org/r/CAHk-=wizk=OxUyQPbO8MS41w2Pag1kniUV5WdD5qWL-gq1kjDA@mail.gmail.com
The first patch in the series is a fix for the epoll race in presence of
timeouts, so that it can be cleanly backported to all affected stable
kernels.
The rest of the patch series simplify the ep_poll() implementation. Some
of these simplifications result in minor performance enhancements as well.
We have kept these changes under self tests and internal benchmarks for a
few days, and there are minor (1-2%) performance enhancements as a result.
This patch (of 8):
After abc610e01c ("fs/epoll: avoid barrier after an epoll_wait(2)
timeout"), we break out of the ep_poll loop upon timeout, without checking
whether there is any new events available. Prior to that patch-series we
always called ep_events_available() after exiting the loop.
This can cause races and missed wakeups. For example, consider the
following scenario reported by Guantao Liu:
Suppose we have an eventfd added using EPOLLET to an epollfd.
Thread 1: Sleeps for just below 5ms and then writes to an eventfd.
Thread 2: Calls epoll_wait with a timeout of 5 ms. If it sees an
event of the eventfd, it will write back on that fd.
Thread 3: Calls epoll_wait with a negative timeout.
Prior to abc610e01c, it is guaranteed that Thread 3 will wake up either
by Thread 1 or Thread 2. After abc610e01c, Thread 3 can be blocked
indefinitely if Thread 2 sees a timeout right before the write to the
eventfd by Thread 1. Thread 2 will be woken up from
schedule_hrtimeout_range and, with evail 0, it will not call
ep_send_events().
To fix this issue:
1) Simplify the timed_out case as suggested by Linus.
2) while holding the lock, recheck whether the thread was woken up
after its time out has reached.
Note that (2) is different from Linus' original suggestion: It do not set
"eavail = ep_events_available(ep)" to avoid unnecessary contention (when
there are too many timed-out threads and a small number of events), as
well as races mentioned in the discussion thread.
This is the first patch in the series so that the backport to stable
releases is straightforward.
Link: https://lkml.kernel.org/r/20201106231635.3528496-1-soheil.kdev@gmail.com
Link: https://lkml.kernel.org/r/CAHk-=wizk=OxUyQPbO8MS41w2Pag1kniUV5WdD5qWL-gq1kjDA@mail.gmail.com
Link: https://lkml.kernel.org/r/20201106231635.3528496-2-soheil.kdev@gmail.com
Fixes: abc610e01c ("fs/epoll: avoid barrier after an epoll_wait(2) timeout")
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Tested-by: Guantao Liu <guantaol@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Guantao Liu <guantaol@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Khazhismel Kumykov <khazhy@google.com>
Reviewed-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull epoll updates from Al Viro:
"Deal with epoll loop check/removal races sanely (among other things).
The solution merged last cycle (pinning a bunch of struct file
instances) had been forced by the wrong data structures; untangling
that takes a bunch of preparations, but it's worth doing - control
flow in there is ridiculously overcomplicated. Memory footprint has
also gone down, while we are at it.
This is not all I want to do in the area, but since I didn't get
around to posting the followups they'll have to wait for the next
cycle"
* 'work.epoll' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (27 commits)
epoll: take epitem list out of struct file
epoll: massage the check list insertion
lift rcu_read_lock() into reverse_path_check()
convert ->f_ep_links/->fllink to hlist
ep_insert(): move creation of wakeup source past the fl_ep_links insertion
fold ep_read_events_proc() into the only caller
take the common part of ep_eventpoll_poll() and ep_item_poll() into helper
ep_insert(): we only need tep->mtx around the insertion itself
ep_insert(): don't open-code ep_remove() on failure exits
lift locking/unlocking ep->mtx out of ep_{start,done}_scan()
ep_send_events_proc(): fold into the caller
lift the calls of ep_send_events_proc() into the callers
lift the calls of ep_read_events_proc() into the callers
ep_scan_ready_list(): prepare to splitup
ep_loop_check_proc(): saner calling conventions
get rid of ep_push_nested()
ep_loop_check_proc(): lift pushing the cookie into callers
clean reverse_path_check_proc() a bit
reverse_path_check_proc(): don't bother with cookies
reverse_path_check_proc(): sane arguments
...
Currently, the sock_from_file prototype takes an "err" pointer that is
either not set or set to -ENOTSOCK IFF the returned socket is NULL. This
makes the error redundant and it is ignored by a few callers.
This patch simplifies the API by letting callers deduce the error based
on whether the returned socket is NULL or not.
Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Florent Revest <revest@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: KP Singh <kpsingh@google.com>
Link: https://lore.kernel.org/bpf/20201204113609.1850150-1-revest@google.com
This option lets a user set a per socket NAPI budget for
busy-polling. If the options is not set, it will use the default of 8.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/bpf/20201130185205.196029-3-bjorn.topel@gmail.com
The existing busy-polling mode, enabled by the SO_BUSY_POLL socket
option or system-wide using the /proc/sys/net/core/busy_read knob, is
an opportunistic. That means that if the NAPI context is not
scheduled, it will poll it. If, after busy-polling, the budget is
exceeded the busy-polling logic will schedule the NAPI onto the
regular softirq handling.
One implication of the behavior above is that a busy/heavy loaded NAPI
context will never enter/allow for busy-polling. Some applications
prefer that most NAPI processing would be done by busy-polling.
This series adds a new socket option, SO_PREFER_BUSY_POLL, that works
in concert with the napi_defer_hard_irqs and gro_flush_timeout
knobs. The napi_defer_hard_irqs and gro_flush_timeout knobs were
introduced in commit 6f8b12d661 ("net: napi: add hard irqs deferral
feature"), and allows for a user to defer interrupts to be enabled and
instead schedule the NAPI context from a watchdog timer. When a user
enables the SO_PREFER_BUSY_POLL, again with the other knobs enabled,
and the NAPI context is being processed by a softirq, the softirq NAPI
processing will exit early to allow the busy-polling to be performed.
If the application stops performing busy-polling via a system call,
the watchdog timer defined by gro_flush_timeout will timeout, and
regular softirq handling will resume.
In summary; Heavy traffic applications that prefer busy-polling over
softirq processing should use this option.
Example usage:
$ echo 2 | sudo tee /sys/class/net/ens785f1/napi_defer_hard_irqs
$ echo 200000 | sudo tee /sys/class/net/ens785f1/gro_flush_timeout
Note that the timeout should be larger than the userspace processing
window, otherwise the watchdog will timeout and fall back to regular
softirq processing.
Enable the SO_BUSY_POLL/SO_PREFER_BUSY_POLL options on your socket.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/bpf/20201130185205.196029-2-bjorn.topel@gmail.com
Move the head of epitem list out of struct file; for epoll ones it's
moved into struct eventpoll (->refs there), for non-epoll - into
the new object (struct epitem_head). In place of ->f_ep_links we
leave a pointer to the list head (->f_ep).
->f_ep is protected by ->f_lock and it's zeroed as soon as the list
of epitems becomes empty (that can happen only in ep_remove() by
now).
The list of files for reverse path check is *not* going through
struct file now - it's a single-linked list going through epitem_head
instances. It's terminated by ERR_PTR(-1) (== EP_UNACTIVE_POINTER),
so the elements of list can be distinguished by head->next != NULL.
epitem_head instances are allocated at ep_insert() time (by
attach_epitem()) and freed either by ep_remove() (if it empties
the set of epitems *and* epitem_head does not belong to the
reverse path check list) or by clear_tfile_check_list() when
the list is emptied (if the set of epitems is empty by that
point). Allocations are done from a separate slab - minimal kmalloc()
size is too large on some architectures.
As the result, we trim struct file _and_ get rid of the games with
temporary file references.
Locking and barriers are interesting (aren't they always); see unlist_file()
and ep_remove() for details. The non-obvious part is that ep_remove() needs
to decide if it will be the one to free the damn thing *before* actually
storing NULL to head->epitems.first - that's what smp_load_acquire is for
in there. unlist_file() lockless path is safe, since we hit it only if
we observe NULL in head->epitems.first and whoever had done that store is
guaranteed to have observed non-NULL in head->next. IOW, their last access
had been the store of NULL into ->epitems.first and we can safely free
the sucker. OTOH, we are under rcu_read_lock() and both epitem and
epitem->file have their freeing RCU-delayed. So if we see non-NULL
->epitems.first, we can grab ->f_lock (all epitems in there share the
same struct file) and safely recheck the emptiness of ->epitems; again,
->next is still non-NULL, so ep_remove() couldn't have freed head yet.
->f_lock serializes us wrt ep_remove(); the rest is trivial.
Note that once head->epitems becomes NULL, nothing can get inserted into
it - the only remaining reference to head after that point is from the
reverse path check list.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
in the "non-epoll target" cases do it in ep_insert() rather than
in do_epoll_ctl(), so that we do it only with some epitem is already
guaranteed to exist.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
That's the beginning of preparations for taking f_ep_links out of struct file.
If insertion might fail, we will need a new failure exit. Having wakeup
source creation done after that point will simplify life there; ep_remove()
can (and commonly does) live with NULL epi->ws, so it can be used for
cleanup after ep_create_wakeup_source() failure. It can't be used before
the rbtree insertion, though, so if we are to unify all old failure exits,
we need to move that thing down. Then we would be free to do simple
kmem_cache_free() on the failure to insert into f_ep_links - no wakeup source
to leak on that failure exit.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The only reason why ep_item_poll() can't simply call ep_eventpoll_poll()
(or, better yet, call vfs_poll() in all cases) is that we need to tell
lockdep how deep into the hierarchy of ->mtx we are. So let's add
a variant of ep_eventpoll_poll() that would take depth explicitly
and turn ep_eventpoll_poll() into wrapper for that.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
We do need ep->mtx (and we are holding it all along), but that's
the lock on the epoll we are inserting into; locking of the
epoll being inserted is not needed for most of that work -
as the matter of fact, we only need it to provide barriers
for the fastpath check (for now).
Move taking and releasing it into ep_insert(). The caller
(do_epoll_ctl()) doesn't need to bother with that at all.
Moreover, that way we kill the kludge in ep_item_poll() - now
it's always called with tep unlocked.
Signed-off-by: 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>
... and get rid of struct ep_send_events_data - not needed anymore.
The weird way of passing the arguments in (and real return value
out - nominal return value of ep_send_events_proc() is ignored)
was due to the signature forced on ep_scan_ready_list() callbacks.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Expand the calls of ep_scan_ready_list() that get ep_read_events_proc().
As a side benefit we can pass depth to ep_read_events_proc() by value
and not by address - the latter used to be forced by the signature
expected from ep_scan_ready_list() callback.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
1) 'cookie' argument is unused; kill it.
2) 'priv' one is always an epoll struct file, and we only care
about its associated struct eventpoll; pass that instead.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The only remaining user is loop checking. But there we only need
to check that we have not walked into the epoll we are inserting
into - we are adding an edge to acyclic graph, so any loop being
created will have to pass through the source of that edge.
So we don't need that array of cookies - we have only one eventpoll
to watch out for. RIP ep_push_nested(), along with the cookies
array.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
We know there's no loops by the time we call it; the
only thing we care about is too deep reverse paths.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
no need to force its calling conventions to match the callback for
late unlamented ep_call_nested()...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
IOW,
* no locking is needed to protect the list
* the list is actually a stack
* no need to check ->ctx
* it can bloody well be a static 5-element array - nobody is
going to be accessing it in parallel.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
we use it only to indicate allocation failures within queueing
callback back to ep_insert(). Might as well use epq.epi for that
reporting...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
We only traverse it once to destroy all associated eppoll_entry at
epitem destruction time. The order of traversal is irrelevant there.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Checking for the lack of epitems refering to the epoll we want to insert into
is not enough; we might have an insertion of that epoll into another one that
has already collected the set of files to recheck for excessive reverse paths,
but hasn't gotten to creating/inserting the epitem for it.
However, any such insertion in progress can be detected - it will update the
generation count in our epoll when it's done looking through it for files
to check. That gets done under ->mtx of our epoll and that allows us to
detect that safely.
We are *not* holding epmutex here, so the generation count is not stable.
However, since both the update of ep->gen by loop check and (later)
insertion into ->f_ep_link are done with ep->mtx held, we are fine -
the sequence is
grab epmutex
bump loop_check_gen
...
grab tep->mtx // 1
tep->gen = loop_check_gen
...
drop tep->mtx // 2
...
grab tep->mtx // 3
...
insert into ->f_ep_link
...
drop tep->mtx // 4
bump loop_check_gen
drop epmutex
and if the fastpath check in another thread happens for that
eventpoll, it can come
* before (1) - in that case fastpath is just fine
* after (4) - we'll see non-empty ->f_ep_link, slow path
taken
* between (2) and (3) - loop_check_gen is stable,
with ->mtx providing barriers and we end up taking slow path.
Note that ->f_ep_link emptiness check is slightly racy - we are protected
against insertions into that list, but removals can happen right under us.
Not a problem - in the worst case we'll end up taking a slow path for
no good reason.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
epoll_loop_check_proc() can run into a file already committed to destruction;
we can't grab a reference on those and don't need to add them to the set for
reverse path check anyway.
Tested-by: Marc Zyngier <maz@kernel.org>
Fixes: a9ed4a6560 ("epoll: Keep a reference on files added to the check list")
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
When adding a new fd to an epoll, and that this new fd is an
epoll fd itself, we recursively scan the fds attached to it
to detect cycles, and add non-epool files to a "check list"
that gets subsequently parsed.
However, this check list isn't completely safe when deletions
can happen concurrently. To sidestep the issue, make sure that
a struct file placed on the check list sees its f_count increased,
ensuring that a concurrent deletion won't result in the file
disapearing from under our feet.
Cc: stable@vger.kernel.org
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
There is a possible race when ep_scan_ready_list() leaves ->rdllist and
->obflist empty for a short period of time although some events are
pending. It is quite likely that ep_events_available() observes empty
lists and goes to sleep.
Since commit 339ddb53d3 ("fs/epoll: remove unnecessary wakeups of
nested epoll") we are conservative in wakeups (there is only one place
for wakeup and this is ep_poll_callback()), thus ep_events_available()
must always observe correct state of two lists.
The easiest and correct way is to do the final check under the lock.
This does not impact the performance, since lock is taken anyway for
adding a wait entry to the wait queue.
The discussion of the problem can be found here:
https://lore.kernel.org/linux-fsdevel/a2f22c3c-c25a-4bda-8339-a7bdaf17849e@akamai.com/
In this patch barrierless __set_current_state() is used. This is safe
since waitqueue_active() is called under the same lock on wakeup side.
Short-circuit for fatal signals (i.e. fatal_signal_pending() check) is
moved to the line just before actual events harvesting routine. This is
fully compliant to what is said in the comment of the patch where the
actual fatal_signal_pending() check was added: c257a340ed ("fs, epoll:
short circuit fetching events if thread has been killed").
Fixes: 339ddb53d3 ("fs/epoll: remove unnecessary wakeups of nested epoll")
Reported-by: Jason Baron <jbaron@akamai.com>
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jason Baron <jbaron@akamai.com>
Cc: Khazhismel Kumykov <khazhy@google.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20200505145609.1865152-1-rpenyaev@suse.de
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch does two things:
- fixes a lost wakeup introduced by commit 339ddb53d3 ("fs/epoll:
remove unnecessary wakeups of nested epoll")
- improves performance for events delivery.
The description of the problem is the following: if N (>1) threads are
waiting on ep->wq for new events and M (>1) events come, it is quite
likely that >1 wakeups hit the same wait queue entry, because there is
quite a big window between __add_wait_queue_exclusive() and the
following __remove_wait_queue() calls in ep_poll() function.
This can lead to lost wakeups, because thread, which was woken up, can
handle not all the events in ->rdllist. (in better words the problem is
described here: https://lkml.org/lkml/2019/10/7/905)
The idea of the current patch is to use init_wait() instead of
init_waitqueue_entry().
Internally init_wait() sets autoremove_wake_function as a callback,
which removes the wait entry atomically (under the wq locks) from the
list, thus the next coming wakeup hits the next wait entry in the wait
queue, thus preventing lost wakeups.
Problem is very well reproduced by the epoll60 test case [1].
Wait entry removal on wakeup has also performance benefits, because
there is no need to take a ep->lock and remove wait entry from the queue
after the successful wakeup. Here is the timing output of the epoll60
test case:
With explicit wakeup from ep_scan_ready_list() (the state of the
code prior 339ddb53d3):
real 0m6.970s
user 0m49.786s
sys 0m0.113s
After this patch:
real 0m5.220s
user 0m36.879s
sys 0m0.019s
The other testcase is the stress-epoll [2], where one thread consumes
all the events and other threads produce many events:
With explicit wakeup from ep_scan_ready_list() (the state of the
code prior 339ddb53d3):
threads events/ms run-time ms
8 5427 1474
16 6163 2596
32 6824 4689
64 7060 9064
128 6991 18309
After this patch:
threads events/ms run-time ms
8 5598 1429
16 7073 2262
32 7502 4265
64 7640 8376
128 7634 16767
(number of "events/ms" represents event bandwidth, thus higher is
better; number of "run-time ms" represents overall time spent
doing the benchmark, thus lower is better)
[1] tools/testing/selftests/filesystems/epoll/epoll_wakeup_test.c
[2] https://github.com/rouming/test-tools/blob/master/stress-epoll.c
Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jason Baron <jbaron@akamai.com>
Cc: Khazhismel Kumykov <khazhy@google.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Heiher <r@hev.cc>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20200430130326.1368509-2-rpenyaev@suse.de
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Willem de Bruijn
Soheil Hassas Yeganeh
Linus Torvalds
Florent Revest
Björn Töpel
Al Viro
Marc Zyngier
Roman Penyaev