IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
After 12 years we finally got TEVENT_FD_ERROR support :-)
TEVENT_FD_WRITE event handlers never get errors reported
instead the event handler is silently disabled.
There are likely callers relying on that behavior, so
we are not able to chance it.
Now TEVENT_FD_WRITE can be used together with TEVENT_FD_ERROR
in order to get errors reported without waiting for TEVENT_FD_READ.
TEVENT_FD_ERROR can also be used alone in order to detect errors
on sockets in order to cleanup resources.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This allows any number of event handlers per low level fd.
It means the epoll backend behaves like the poll backend now.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Backends may require to map individual tevent_fd instances to
a single low level kernel state (e.g. for epoll).
This generic infrastructure adds helper functions using
a generic (sub)part of struct tevent_fd.
The new code will allow us to support more than 2 tevent_fd
instances per fd, which makes sure all backends can provide
a similar behavior. This will be important when we add
TEVENT_FD_ERROR as a 3rd kind of fd event.
The aim is to use this in order to replace the limited implementation
we already have in tevent_epoll.c.
As these helpers are typically called from within
'void tevent_fd_set_flags(struct tevent_fd *fde, uint16_t flags)'
there's no way to report errors. So in order avoid additional
error handling complexity the helpers try to avoid
any allocations which may fail. It also means the logic in
tevent_epoll.c doesn't have to change much.
These are implemented as static line functions in order to avoid
the function call overhead, which showed up in profiles of the
early implementation.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
It means tevent_trace_fd_callback(TEVENT_EVENT_TRACE_DETACH)
is always called and similar future changes are only
needed in one place.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This is clearer for multiplexed fdes as it means both sides are
already cleared before we call epoll_update_event() again.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
The preparation, function call and cleanup for epoll_check_reopen()
is quite some overhead and not needed most of the time!
So check the pid in the caller avoids most of it.
Review with: git show -w
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Prefer epoll_create1(2) over epoll_create(2) and
always require the former to use epoll(7) interface,
thus saving extra fcntl(2) call to set FD_CLOEXEC.
Signed-off-by: Dmitry Antipov <dantipov@cloudlinux.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
This avoids a getpid() syscall per tevent_loop_once() iteration.
We provide tevent_cached_getpid() also as helper for external consumers
in order to have the logic only once.
Note the change to ABI/tevent-0.12.1.sigs will be reverted
with the bump to 0.13.0.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This allows to specify wrapper tevent_contexts, which adds the ability
to run functions before and after the event handler functions.
This can be used to implement impersonation hooks
or advanced debugging/profiling hooks.
We'll undo the 0.9.36 ABI change on the 0.9.37 release
at the end of this patchset.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We'll undo the 0.9.36 ABI change on the 0.9.37 release
at the end of this patchset.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This is infrastructure to improve our async r/w result handling and latency.
The pthreadpool signalling goes through a pipe. This has downsides: The main
event loop has to go through a read on the pipe before it can ship the result.
Also, it is not guaranteed by poll/epoll that the pthreadpool signal pipe is
handled with top priority. When an async pread/pwrite has finished, we should
immediately ship the result to the client, not waiting for anything else.
This patch enables tevent_immediate structs as job signalling. This means a
busy main tevent loop will handle the threaded job completion before any timed
or file descriptor events. Opposite to Jeremy's tevent_thread_proxy this is
done by a modification of the main event loop by looking at a linked list under
a central mutex.
Regarding performance: In a later commit I've created a test that does nothing
but fire one immediate over and over again. If you add a phread_mutex_lock and
unlock pair in the immediate handler, you lose roughly 25% of rounds per
second, so it is measurable. It is questionable that will be measurable in the
real world, but to counter concerns activation of immediates needs to go
through a new struct tevent_threaded_context. Only if such a
tevent_threaded_context exists for a tevent context, the main loop takes the
hit to look at the mutex'ed list of finished jobs.
This patch by design does not care about talloc hierarchies. The idea is that
the main thread owning the tevent context creates a chunk of memory and
prepares the tevent_immediate indication job completion. The main thread hands
the memory chunk together with the immediate as a job description over to a
helper thread. The helper thread does its job and upon completion calls
tevent_threaded_schedule_immediate with the already-prepared immediate. From
that point on memory ownership is again transferred to the main thread.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Such events were used before we had immediate events.
It's likely that there're a lot of this events
and we need to add new ones in fifo order.
The tricky part is that tevent_common_add_timer()
should not use the optimization as it's used
by broken Samba versions, which don't use
tevent_common_loop_timer_delay() in source3/lib/events.c.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
A Samba autobuild passed without a fallback, so this is
really an error.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We'll use this to handle the EPOLL_ADDITIONAL_FD_FLAG_GOT_ERROR
and EPOLL_ADDITIONAL_FD_FLAG_REPORT_ERROR flags with multiplexed
events in the event loop.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Add a utility function epoll_add_multiplex_fd() and
a new flag EPOLL_ADDITIONAL_FD_FLAG_HAS_MPX.
This will be called by epoll_add_event() to merge two
fde events with the same file descriptor.
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
epoll_{add,mod,del}_event() are only called via epoll_update_event()
and epoll_update_event() should not remove REPORT_ERROR itself.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We'll never leave epoll_check_reopen() with epoll_fd == -1.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This calls TALLOC_FREE(ev->additional_data), which is epoll_ev
within epoll_panic() before calling the fallback handler.
In order to notice that a epoll_panic() happened, a caller can
register a pointer to a bool variable under epoll_ev->panic_state.
As epoll_check_reopen() can fail due to a epoll_panic(),
we need to force the replay flag if we have called any event handler.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This makes sure we only do random panics if a fallback handler
is registered.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We might be called during tevent_re_initialise()
which means we need to free our old additional_data.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Can be set externally, allows us to fallback if epoll
fails at runtime.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
A fallback panic handler will need to know if
there was an error while waiting for events
(replay=true) or if the error happened on modify
(replay=false).
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Currently we can't return from this, but the new fallback
code will change this.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Set/get a single callback function to be invoked at various trace
points. Define "before wait" and "after wait" trace points - more
trace points can be added later if required.
CTDB wants this to log long waits and events.
Pair-programmed-with: Amitay Isaacs <amitay@gmail.com>
Signed-off-by: Martin Schwenke <martin@meltin.net>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
If an application using libtevent starts a new process the epoll file descriptor
is leaked to the new process if the event context is not freed explicitly. By
setting FD_CLOEXEC this is not needed anymore.
Signed-off-by: Simo Sorce <idra@samba.org>
