Richard Fuchs
d8776907f8
need to lock the callstream a bit earlier
14 years ago
Richard Fuchs
ab3b8c60f0
fix a silly, threading-unrelated, statistics-skewing bug
14 years ago
Richard Fuchs
45f34b9784
turns out you actually have to free/destroy mutexes. who woulda thunk?
14 years ago
Richard Fuchs
54353022ff
om nom nom memory leak... well, reference count leak
14 years ago
Richard Fuchs
b7a68659b1
analoguous to the last race, call_destroy() must not assume that it's getting the last reference to the call
14 years ago
Richard Fuchs
683b716885
here we go with the subtle race conditions
14 years ago
Richard Fuchs
806920cc52
dont abort polling on EINTR
14 years ago
Richard Fuchs
0d6349f14c
and for the interesting part: multiple poller threads
14 years ago
Richard Fuchs
942c96c96d
move poller timers and polling into a separate thread
14 years ago
Richard Fuchs
d25d1c2009
making main() pretty by moving stuff out of the way
14 years ago
Richard Fuchs
ddffe5eeec
global xmlrpc init must be done prior to any threading
14 years ago
Richard Fuchs
d1fbf9bef9
alarms arent usable in a threaded environment, so use curl internal timeout instead
14 years ago
Richard Fuchs
feeeee2fab
must extract needed tags for xmlrpc call teardown prior to destroying the calls
14 years ago
Richard Fuchs
973d92d9c7
moving call teardown due to timeout to a separate thread
14 years ago
Richard Fuchs
47a9a00a11
it makes no sense to return a value from an effectively detached thread
14 years ago
Richard Fuchs
31a5ff9c29
detached threads considered harmful, so simulate them by keeping a threads-to-join list
14 years ago
Richard Fuchs
bc647862e5
use sigtimedwait for proper sighandler thread shutdown
14 years ago
Richard Fuchs
badae5a874
lets get some threads going then, starting with a signal handler
14 years ago
Richard Fuchs
2dd6cfe68c
keeping the lock hold time to a minimum in lengthy calls_status operation
14 years ago
Richard Fuchs
8c2c69f5e6
make the stats collection thread-safe
14 years ago
Richard Fuchs
d9e09f33bc
lets try with a lot of large, nested, naive locks
14 years ago
Richard Fuchs
ae5a36f5ce
getting rid of call->log_info and introducing call->lock
14 years ago
Richard Fuchs
900f76a17e
make the callmaster struct private - also requires update to redis module
14 years ago
Richard Fuchs
aeb5398b0d
granular lock for "ports used" bit field
14 years ago
Richard Fuchs
a8680cc2d9
TOS for ipv6 needs an int to work
14 years ago
Richard Fuchs
0b0b41e67f
gonna need more than one lock in callmaster, so rename the one we have
14 years ago
Richard Fuchs
4cb63b5edf
need to handle the no-reply case to avoid deadlocks with bad cookies
14 years ago
Richard Fuchs
4588e13a76
getting rid of more global variables
14 years ago
Richard Fuchs
e033d1da1c
EINTR requires special handling
14 years ago
Richard Fuchs
1ec2c96655
living on the edge with epoll edge-triggered behavior
14 years ago
Richard Fuchs
db15692f21
using thread-local-storage for eliminating constant calls to time()
14 years ago
Richard Fuchs
45945f6c7f
it helps to actually initialize the condition var
14 years ago
Richard Fuchs
26824993e2
preliminary attempt to make the udp controller thread-safe
14 years ago
Richard Fuchs
5f5e158d76
private structs make for safer threading
14 years ago
Richard Fuchs
ebd2893e96
missing a streambuf lock here
14 years ago
Richard Fuchs
419f6bb230
and the tcp controller becomes thread-safe
14 years ago
Richard Fuchs
6e474ccfc0
add appropriate threading compile flags
14 years ago
Richard Fuchs
c75aecc282
make use of "make debug" target
14 years ago
Richard Fuchs
7dc00bd258
making timers thread-safe
14 years ago
Richard Fuchs
4b74e6f476
use obj_* to catch double-free errors
14 years ago
Richard Fuchs
b3ef1995fc
add mutex_trylock primitives
14 years ago
Richard Fuchs
de33b8da40
add basic locking to callmaster struct
14 years ago
Richard Fuchs
9dab795a17
disable object debugging, seems to work fine
14 years ago
Richard Fuchs
8dcd2c89c4
add rwlock type and primitives
14 years ago
Richard Fuchs
26f29aefeb
object size isnt really a useful piece of information
14 years ago
Richard Fuchs
8e5ad59558
make the obj_ primitives more liberal to make for easier calling
14 years ago
Richard Fuchs
f05156a58e
catch the inevitable disaster early and define our own mutex types. sigh. and there i was, thinking that using the glib types would make things easier.
14 years ago
Richard Fuchs
806c97246c
glib < 2.32 requires use of distinct static mutex type
14 years ago
Richard Fuchs
145bbd1f7c
Prepare for multi-threaded operation
Introduce generic "objects" with reference counting primitives and automatic
resource deallocation when no refs are left ("garbage collection"). Overhaul
poller framework to make extensive use of these objects to allow for future
thread-concurrent operation.
No locking added anywhere yet except in poller. Poller is still not 100%
thread safe, but close.
Valgrind reports no errors or memleaks.
14 years ago
Richard Fuchs
5d1410c61a
remove obsolete targets from makefile
14 years ago