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.
I did not find a way to do this safely without a mutex per threaded_context.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This prepares tevent run-down with active threads.
It has the advantage to not depend on talloc'ed structs. It is needed to make
talloc_free(tevent_context) safe when tevent_threaded_contexts are still
around.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This makes the reading end of the signalling pipe special: If we have eventfd,
this is the same as the write fd. Without eventfd, it will have to be a
separate fd. This moves the requirement to #ifdef from the writing end to the
reading end. Why? We'll use the writing end somewhere else too soon, and this
patch avoids an #ifdef in that new place.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
According to the manpage, eventfd is cheaper than a pipe. At least, we can save
a file descriptor and space for it in struct tevent_context :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Purely cosmetic change: This moves closing the signal/thread event pipe
to where it's opened. This prepares the eventfd support, making the
"magic" for eventfd more obvious.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@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>
Signalling the main event loop will also happen from threads soon, and
that will use the same mechanism. This also keeps the pipe open after the last
signal handler is removed. Threaded jobs will come and go very frequently, and
always setting up and tearing down the pipe for each job will be expensive.
Also, this is "just" two file descriptors, and with eventfd just one.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
Autobuild-User(master): David Disseldorp <ddiss@samba.org>
Autobuild-Date(master): Sun Feb 15 23:25:07 CET 2015 on sn-devel-104
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>
The define TEVENT_HAS_LOOP_ONCE_TRACE_POINTS can be used to
detect the new feature, without writing configure tests.
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Returns an event ops struct given a string name. Not
yet used, but will be part of the new "standard" fallback
code.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
the tevent destructor is called in tevent_re_initialise(), to reset
the event context back to its original state for creating child
processes. We need the nesting flag to stay the same
Pair-Programmed-With: Andrew Bartlett <abartlet@samba.org>
Autobuild-User: Andrew Tridgell <tridge@samba.org>
Autobuild-Date: Fri Aug 12 05:18:25 CEST 2011 on sn-devel-104
We may be forking from within a loop, so we need to clean-up to avoid
aborts when nesting is not allowed and we are in a new children.
Signed-off-by: Andreas Schneider <asn@samba.org>
This allows us to re-initialise a tevent context without destroying
the pointer. That means that if someone keeps a long term ptr to the
event context across a fork it will still work.
This also brings the memory handling in single and standard process
models much closer together, which means less bugs that we don't find
with make test.
The "hack_fds" were never closed before; now they're inside event_context
they should be closed when that is destroyed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Only tevent_loop_once and tevent_loop_until() should care
about the nesting level.
This fixes the samba3 printing code where we use tevent_loop_wait()
and don't allow nested events.
We still call the nesting hook for all levels, we need to decide
if we really want this...
metze
The event nesting code never triggered as nesting.level was never
greater than 1. The main event loop needs to increase the nesting
level by 1.
I also added a paranoia check to the nesting setup call. The API as
currently written cannot support multiple nesting hooks, so we need to
abort if multiple hooks are tried.
It makes no sense to support aio events because,
the current implementation was based on IOCB_CMD_EPOLL_WAIT
which never made it into the main kernel tree.
The native linux aio can be used with select/epoll
using eventfd(), which means we can implement aio
with fd events and implement aio outside of tevent.
metze
