| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
| |
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
|
| |
This return either malloc'ed memory or NULL. Make sure there is no
problem.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
| |
csbuild complains about missing NULL checks here.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
| |
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
| |
Use realpath(3) instead of strncmp(3)
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
|
|
|
|
|
| |
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are applications which do:
for (fd = 0; fd <= getdtablesize(); fd++) {
close(fd)
};
This produced millions of error messsages.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
| |
Found by csbuild.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
| |
Pair-Programmed-With: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Andreas Schneider <asn@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
|
|
|
| |
This will allow that we can deal with duplicates if the default value is
used.
Pair-Programmed-With: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Andreas Schneider <asn@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
|
|
| |
We also need a mutex to protect access to it.
Pair-Programmed-With: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Andreas Schneider <asn@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
| |
Fixes the build on OmniOS
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
|
|
| |
CID 184220
CID 184221
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
|
| |
Fixes a helgrind error.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
| |
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
|
|
|
|
|
|
| |
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
|
|
|
|
| |
CID 175897
CID 175898
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes the following bug:
As we are using a doubly linked list, we need a mutex which needs to be
locked when we are reading it that we do not end up with invalid
pointers.
The following can happen:
We are in swrap_close() which calls find_socket_info_fd() this locks the
mutex for the linked list. Now we get a singal SIGCHILD and the signal
handler is called. The signal handler calls swrap_write() and we try to
find out if the socket is managed by socket_wrapper calling
find_socket_info_fd() again -> DEADLOCK!
By moving to an array to handle the socket fds and using the fd as the
array access, we do not need a mutex for reading anymore. All we need is
a memory barrier.
This change also improves the performance as we move from the a linked
list to a hash table!
Pair-Programmed-With: Andreas Schneider <asn@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
|
| |
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
| |
Separating out the metadata related information to another
sub-structure to make it more clean and structured.
Pair-Programmed-With: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added a new mutex variable to socket_info structure along with
new macros for locking and unlocking mutex corresponding to
each socket_info entry. Apart from individual mutex defined in
socket_info structure, 4 new mutexes are added to protect the
concurrent access of globally used swrap parameters from different
threads.
All other individual wrappers and helper routines are also made
capable of acquiring relevant mutex locks before operating on such
global parameters.
Pair-Programmed-With: Michael Adam <obnox@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
In preparation to implement thread safety, re-ordering lines
of code to properly align to lockign calls
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
In preparation to implement thread safety, re-ordering lines
of code to properly align to locking calls.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
swrap_first_free_index is no longer used as the whole logic now
implemented within swrap_create_socket.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
Replace the current logic of socket creation within swrap_accept
with more cleaner version using swrap_create_socket.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
Replace the current logic of socket creation within swrap_socket
with more cleaner version using swrap_create_socket.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
| |
A new function named swrap_create_socket is introduced which
cleanly performs all stuff related to creation of new socket
file descriptors and updation of relevant metadata including
the free-list and reference counter.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The change basically splits socket_info structure into
two structures, namely,
- socket_info: to store the core information
corresponding to a socket entry.
- socket_info_container: wrapping structure to hold
the socket_info data as well as metadata(currently
refcount and next_free).
Pair-Programmed-With: Anoop C S <anoopcs@redhat.com>
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
| |
- swrap_get_next_free
- swrap_set_next_free
to avoid accessing socket_info.next_free directly
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
| |
- swrap_get_refcount
- swrap_alter_refcount
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
| |
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
| |
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
| |
Compilers that don't support fallthrough will end up with an empty
FALL_THROUGH define and just see a semicolon. The probably will warn
that there are double semicolons in the code.
Signed-off-by: Andreas Schneider <asn@samba.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If there is a signal handler defined, e.g. for SIGCHILD and this signal
handler is called during pthread_atfork() has just locked all mutexes
and we want to check if a symbol is bound, we end up in a deadlock.
(gdb) bt
0 __lll_lock_wait () at ../sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:135
1 0x00007fe48837aef5 in __GI___pthread_mutex_lock (mutex=0x7fe4889a71c0 <libc_symbol_binding_mutex>) at ../nptl/pthread_mutex_lock.c:80
2 0x00007fe48879b2f1 in libc_write (fd=14, buf=0x7ffdad6436e8, count=8) at /home/asn/workspace/projects/socket_wrapper/src/socket_wrapper.c:1070
3 0x00007fe4887a21cb in swrap_write (s=14, buf=0x7ffdad6436e8, len=8) at /home/asn/workspace/projects/socket_wrapper/src/socket_wrapper.c:5137
4 0x00007fe4887a2359 in write (s=14, buf=0x7ffdad6436e8, len=8) at /home/asn/workspace/projects/socket_wrapper/src/socket_wrapper.c:5171
5 0x00007fe48713e5f2 in tevent_common_wakeup_fd (fd=14) at ../lib/tevent/tevent.c:959
6 0x00007fe48713e61f in tevent_common_wakeup (ev=ev@entry=0x55d04922e280) at ../lib/tevent/tevent.c:975
7 0x00007fe487141d38 in tevent_common_signal_handler (signum=<optimized out>) at ../lib/tevent/tevent_signal.c:109
8 <signal handler called>
9 0x00007fe4844de081 in __libc_fork () at ../sysdeps/nptl/fork.c:135
10 0x000055d047e8ddb3 in smbd_accept_connection (ev=0x55d04922e280, fde=<optimized out>, flags=<optimized out>, private_data=<optimized out>)
at ../source3/smbd/server.c:992
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
|
|
|
| |
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|