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.
In the includes we define SIGRTMIN to 32 if it's not defined already. This
value could be fairly low and it's better to use NSIG(number of defined
signals) as the lower mark for the available signals.
We have similar defenition in the source3/smbd/aio.c, which can be safely
removed, as it comes from includes.h then.
With regards,
Timur Bakeyev.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
The aio_fork module does not need this, as it does not communicate via signals
but with pipes. Watching a strace log with those become_root() calls in aio.c
is absolutely awful, and it does affect performance.
[MS-SMB] 3.3.5.1 Receiving Any Message says that the seqnum
is incremented by only for ntcancel requests for any other
request it's by incremented by 2, even if it doesn't expect
a response.
metze
This replaces release_level2_oplocks_on_change with
contend_level2_oplock_begin/end in order to contend level2 oplocks
throughout an operation rather than just at the begining. This is
necessary for some kernel oplock implementations, and also lays the
groundwork for better correctness in Samba's standard level2 oplock
handling. The next step for non-kernel oplocks is to add additional
state to the share mode lock struct that prevents any new opens from
granting oplocks while a contending operation is in progress.
All operations that contend level 2 oplocks are now correctly spanned
except for aio and synchronous writes. The two write paths both have
non-trivial error paths that need extra care to get right.
RAW-OPLOCK and the rest of 'make test' are still passing with this
change.
This changelist allows for the addition of custom performance
monitoring modules through smb.conf. Entrypoints in the main message
processing code have been added to capture the command, subop, ioctl,
identity and message size statistics.
We need to become root for AIO read and write to allow the AIO thread
to send a completion signal to the parent process when the IO
completes
(This used to be commit c548e5c69f9d8bc85a654f4d29d64c735a5e780b)
In the negprot reply, we allowed the client to issued "max mux" concurrent
requests. The OS might allow less, for example AFAIK AIX has a configurable
limit of concurrent AIO requests. We will fall back to sync operation for the
requests that are too many when aio_read/aio_write return an error.
Jeremy, please check!
(This used to be commit 8f86f7f25c4eb71bbdfcc6bf2d12eaaae9a8d9ec)
This is for external modules implementing the Posix AIO API that not
necessarily depend on the signal mechanism to indicate completion.
(This used to be commit 9a069d306bb6f8a2cd51b8ba87b433884942d13f)
The logic was wrong: A "SMB_VFS_AIO_FORCE()==False" disabled async I/O, whereas
a "SMB_VFS_AIO_FORCE()==True" should enforce it regardless of other settings.
Alexander, please check!
(This used to be commit 46882ad9927c95caadeb7fb03c1d7491bbe1fb22)
with Volker. Mostly making sure we have data on the incoming
packet type, not stored in the smb header.
Jeremy.
(This used to be commit c4e5a505043965eec77b5bb9bc60957e8f3b97c8)
bugs in various places whilst doing this (places that assumed
BOOL == int). I also need to fix the Samba4 pidl generation
(next checkin).
Jeremy.
(This used to be commit f35a266b3cbb3e5fa6a86be60f34fe340a3ca71f)
Prototype of create_aio_ex_read, set_message and definition of
UNIXERROR have changed in r22389 and r22391? to require an additional
"inbuf" argument. Some places in smbd/aio.c had not been adapted.
This patch fixes the build.
Michael
(This used to be commit 9de1af6dc835fa12a75bae00935cd605f61217a3)
to all callers of smb_setlen (via set_message()
calls). This will allow the server to reflect back
the correct encryption context.
Jeremy.
(This used to be commit 2d80a96120a5fe2fe726f00746d36d85044c4bdb)
server. Allow server to reflect back to calling client
the encryption context that was sent.
Jeremy.
(This used to be commit b49e90335d1e589916b5ab4992e3c4a2d221ca7e)
turns out that this patch actually speeds up the async writes considerably.
I tested writing 100.000 times 65535 bytes with the allowed 10 ops in
parallel. Without this patch it took about 32 seconds on my dual-core 1.6GHz
laptop. With this patch it dropped to about 26 seconds. I can only explain it
by better cache locality, NewInBuffer allocates more than 128k, so we jump
around in memory more.
Jeremy, please check!
Volker
(This used to be commit 452d51bc6fd41771b9c41ba6391664513d7cf2cd)
AIO_PENDING_SIZE signals, not one less.
Jeremy I'm not merging this to 3_0_25, I want you to look at it first!
Volker
(This used to be commit 8fd0ec58360a349826508fc361a943650fdd6694)
