There was a bug here where AF_INET6 was used instead of AF_UNIX in the
HAVE_SYS_UN_H section. This masked another bug, which was that the
HAVE_SYS_UN_H section inadvertently extended into what was supposed to
be the HAVE_IPV6 section.
I'm not sure why I wasn't seeing this error before:
engine_poll.c: In function ‘poll_loop’:
engine_poll.c:352:46: error: ‘EINTR’ undeclared (first use in this function)
engine_poll.c:352:46: note: each undeclared identifier is reported only once for each function it appears in
All the information passed as arguments is present in an iod, and we
only call this with members of one iod. Change it to accept just an iod
as an argument.
Implementation of UNIX-domain sockets functionality
for Nsock. Also some minor necessary changes to
existing Nsock functions, to work properly with
UNIX-domain sockets.
- Do not set the X flags as events to be watched.
- Do not use POLLPRI as a R flag.
Not doing so causes WSAPoll() to fail with an "invalid argument" error.
This happens on Windows when two sockets bind to the same local port and
connect to the same remote port on a remote host. I tried this on Linux;
Linux prevents the second bind from succeeding.
http://seclists.org/nmap-dev/2012/q2/513
Otherwise WinSock error numbers are not understood; you get "Unknown
error" instead of "Only one usage of each socket address
(protocol/network address/port) is normally permitted." for example.
It's used two times, in two separate blocks of the function. Therefore it was
declared twice (once per block), then got moved toplevel but the second
declaration was forgotten somehow.
This doesn't actually change anything (identical objdump -d diff) but makes code
nicer.
http://seclists.org/nmap-dev/2012/q3/56. r29134 already addressed the issue but
was incomplete.
This replaces r29134 with an engine-agnostic approach, and additionally enforces
the reset of IOD flags before use or re-use.
that were internally closed and replaced by other ones. This happened during
reconnect attempts.
--
When reconnecting with SSL_OP_NO_SSLv2 (nsock_core.c:472), the libary closes the
fd of the current IOD, and replaces it by a new one.
The man page for epoll_ctl states that a close() on a fd makes it removed from
any epoll set it was in. Therefore, if epoll_ctl(EPOLL_CTL_MOD, ...) returns
ENOENT, we retry with EPOLL_CTL_ADD.
- Always format function name without parens at the beginning of a message.
- Added a nsi_new notification message.
- Only trace the first call to nsi_delete() of a given IOD (i.e. don't log the
calls issued from the resulting callbacks).
Solaris 11 uses BPF packet capture rather than DLPI, which requires
different handling in many cases. The new preprocessor symbol tells when
this is the case; it is additional granularity on top of SOLARIS.
The Nsock select engine stores readiness information in a mask instead
of doing an FD_ISSET against the descriptor every time. This is a good
idea, but it removed a special case in the pre-engines code: in case
PCAP_CAN_DO_SELECT is not defined, we can't do FD_ISSET on the
descriptor because it is -1. Here we check for that occurrence and set
EV_READ in the mask so that a non-blocking pcap read can happen every
iteration.
This was being done manually for various types of events, and not doing
it in every case was causing hard-to-find bugs. See the log messages for
r19970 in /nsock and r28292 in /nmap.
nsock_tod is Nsock's idea of the current time. It is updated when an
nsock_pool is initialized, on each iteration of nsock_loop, and in a few
other places. What could go wrong, with respect to timers, is a sequence
like this:
nsp_new
[... some long delay ...]
nsock_create_timer(timeout)
nsock_loop
The time elapsed after the creatino of the timer until it fires would
not be timeout, but rather timeout - delay. If the delay was long
enough, the timer would fire as loop as nsock_loop was entered.
This showed itself in IPv6 OS detection. We schedule 6 timers
immediately, 100 ms apart. If the pcap_open or anything else took too
long, then the timers would fire all at once. This messed up the
calculation of the TCP_ISR feature.
Perhaps we should do this when any new event is created? It is already
done manually at the beginning of each of the connect functions.
