Age | Commit message (Collapse) | Author |
|
- Avoid deadlock when activating signals. Fixes bug 3048812.
Based on patch by Nicholas Marriott.
The deadlock was ultimately fixed in a different way (by disabling
reinit - see event.c r1.25). Add it now for consistency but without
the Windows compatibility code. Convert the fnctl() calls to
SOCK_CLOEXEC | SOCK_NONBLOCK to simplify the code.
OK nicm@
|
|
OK nicm@
|
|
useless. Let libevent call the libc functions directly.
OK nicm@
|
|
OK nicm@
|
|
Do not include sys/param.h.
OK nicm@
|
|
OK nicm@
|
|
OK nicm@
|
|
pattern.
This commit does not fix the non-obvious bloody horror of select.c.
|
|
evsignal_info does not change, so no library crank.
OK nicm@ deraadt@
|
|
ok otto
|
|
From their change log:
o Fix memory-leak of signal handler array with kqueue. [backport]
o Make evutil_make_socket_nonblocking() leave any other flags alone.
o Adjusted fcntl() retval comparison on evutil_make_socket_nonblocking().
o Re-add event_siglcb; some old code _was_ still using it. :(
o Fix a free(NULL) in min_heap.h
o Clean up properly when adding a signal handler fails.
Also a local change to use an int rather than a long for fcntl().
ok guenther deraadt
|
|
This is the core library only, the DNS parts are removed and it does not
include the other extra bits (HTTP, DNS, and RPC), a separate port for
these will appear in due course.
Thanks to jsg, sthen, alek, gilles, jacekm, bernd and any others I've
forgotten for testing/comments.
Note that /usr/include/evdns.h should be removed after updating.
ok deraadt
|
|
and pass a pointer to that instead of passing SIG_DFL directly.
(Probably the result of a bad signal()-->sigaction() translation...)
ok deraadt@, nicm@, miod@
|
|
"No objection" millert@
"the diff looks and works fine" reyk@
|
|
|
|
every signal received writes 1 byte to the socketpair. if you are outside
the libevent main loop too long, the socketpair might fill up. that write
would then block, in a signal handler. related to this, yesterdy we changed
the signal handler to not trash errno.
as for the read end, there will be multiple libevent reads off the socketpair
(100 bytes at a time) until the socketpair is empty again
ok provos
|
|
|
|
only access sig_atomic_t variables, to make it further save
ok aaron
|
|
ok brad@ beck@
|
|
no shared lib so no ABI/API check is necessary
thanks to Alexander von Gernler for submitting
another diff in an attempt to update libevent
and for a use-after-free fix.
ok henning@ deraadt@
|
|
|
|
|
|
ok markus@
|
|
|
|
|