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.
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Tue Sep 6 23:16:34 CEST 2016 on sn-devel-144
* add tevent_threaded_context_create() and tevent_threaded_schedule_immediate()
They add a way to pass the thread result from a helper thread into
the main event loop.
Signed-off-by: Stefan Metzmacher <metze@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>
No functionality change. This just looks better in objdump --disassemble :-)
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>
This adds 40 bytes, but they are needed for correctness :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Fri Jul 22 23:33:57 CEST 2016 on sn-devel-144
This is one of or hottest code paths, I think every bit counts here.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
This is one of or hottest code paths, I think every bit counts here.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
This reverts commit 2991f7709973fdcc2c0b83bbe15dda3f1ceae9b3.
Breaks compile for older (<= 4.4) gccs.
Needs to be done differently.
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We expect these macros to generate tautological compares
intentionally, so disabling the warning is just fine.
This lets --picky-developer work with gcc6 and newer.
Pair-Programmed-With: Ira Cooper <ira@samba.org>
Signed-off-by: Ira Cooper <ira@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Comment on two similar conditions in tevent_standard.c, which,
otherwise, at a first glance, seem useless, i.e. always true.
The conditions checking glue->epoll_ops for being non-NULL, imply that
it *can* be NULL. A casual reader would not generally expect a "member"
function to modify its container's pointer in a container higher up, and
would assume that glue->epoll_ops could be NULL before the call,
resulting in a near-NULL pointer dereference.
However, in this case epoll_ops is indeed cleared in those "member"
functions, in the case of an epoll interface failure, to signify
fallback to poll interface.
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Uri Simchoni <uri@samba.org>
Autobuild-User(master): Uri Simchoni <uri@samba.org>
Autobuild-Date(master): Tue Jul 12 13:56:41 CEST 2016 on sn-devel-144
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Uri Simchoni <uri@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sun May 29 09:13:30 CEST 2016 on sn-devel-144
some compilers don't tolerate void-casting for warn_unused_result
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Christian Ambach <ambi@samba.org>
some compilers don't tolerate void-casting for warn_unused_result
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Christian Ambach <ambi@samba.org>
some compilers don't tolerate void-casting for warn_unused_result
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Christian Ambach <ambi@samba.org>
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Uri Simchoni <uri@samba.org>
* Fix memory leak when old signal action restored (bug #11742)
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Fri Feb 19 19:12:25 CET 2016 on sn-devel-144
BUG: https://bugzilla.samba.org/show_bug.cgi?id=11742
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Feb 18 01:42:50 CET 2016 on sn-devel-144
* Fix bug in poll backend - poll_event_loop_poll()
exits the for loop on POLLNVAL instead of
continuing to find an event that is ready.
* Fix ETIME handling for Solaris event ports (bug #11728).
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Tue Feb 16 00:00:51 CET 2016 on sn-devel-144
It is possible for port_getn to return -1 with errno set to ETIME and
still return events. If those events aren't processed the association is
lost by samba since the kernel dissacociated them and samba never
processed them so never reassociated them with the event port. The
patch checks the nget return value in the case of ETIME and if it is non
0 it doesn't return and goes through the event processing loop.
Signed-off-by: Nathan Huff <nhuff@acm.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Sun Feb 7 11:26:35 CET 2016 on sn-devel-144
Without this, tests fail ir libtevent is not installed on the system.
Signed-Off-By: Jelmer Vernooij <jelmer@jelmer.uk>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Nov 18 21:40:26 CET 2015 on sn-devel-104
If the (pfd->revents & POLLNVAL) case is triggered,
we do DLIST_REMOVE(ev->fd_events, fde); and then
use fde->next in the loop above.
Save off fde->next for loop interation before
this so we can't use a deleted ->next value.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Both tests run cleanly with valgrind --tool=drd and
valgrind --tool=helgrind
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
* Fix compile error in Solaris ports backend.
* Fix access after free in tevent_common_check_signal(). bug #11308
* Improve pytevent bindings.
* Testsuite fixes.
* Improve the documentation of the tevent_add_fd()
assumtions. It must be talloc_free'ed before closing the fd!
See bug #11141 and bug #11316.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
This gives talloc_set_destructor to verify the type,
which removes a compiler warning.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
