who wants to shut down a connection calls connection_mark_for_close instead
of setting marked_for_close to 1. This automatically removes the connection
from the DNS cache if needed, sends a RELAY END cell if appropriate, and can
be changed to do whatever else is needed.
Still to do:
- The same for circuits, maybe.
- Add some kind of hold_connection_open_until_flushed flag, maybe.
- Change stuff that closes connections with return -1 to use mark_for_close,
maybe.
svn:r1145
add some more data to be flushed but never turn POLLOUT on. not sure
how commonly this bug was hit, but it would be a doozy.
Also add some asserts to see if it happens elsewhere.
svn:r1142
We were telling a child to die by closing the parent's file descriptor
to him. But newer children were inheriting the open file descriptor from
the parent, and since they weren't closing them, the socket never closed,
so the child never read eof, so he never knew to exit.
As a side effect to this bug, we were probably failing to properly close
connections to remote hosts, ORs, and OPs, after a dns child was born.
I'm surprised Tor worked at all.
svn:r974
split 7-byte stream_id string into 2-byte recognized and 2-byte stream_id
fix two seg faults in fetch_from_buf_http
fix several lurking seg faults in handling unexpected relay cells
still need to
* clean up relay_crypt
* use relay dummies if there's going to be a conflict with rh.recognized
* check for a conflict when generating stream_ids
svn:r953
* now we know for sure if an acceptable node is available; we
don't have to keep guessing and checking
* we try options.EntryNodes first for picking the first node
svn:r904
directory is the string that dirserv.c and directory.c deal with
routerlist is routerinfo's that are bundled together in routers.c
rename some of the get_routerlist functions to set_routerlist
preparing to break into router.c for stuff the router does,
and routerlist.c for handling routerlist.
svn:r886
bugfix in connection_ap_can_use_exit: it was using the wrong port
bugfix: the OP now handles a port of '*' correctly when the IP is not
yet known and it's trying to guess whether a router's exit policy
might accept it.
we now don't ever pick exit routers which will reject *:*
attach_circuit now fails a new stream outright if it will never work.
when you get an 'end' cell that resolves an IP, now it will fail the circuit outright if no safe exit nodes exist for that IP.
don't try building a new circuit after an 'end' if a suitable one is
already on the way.
svn:r874
Increment failure counts only when circuits close without having been built.
Reset failure counts only on the second, and when circuits are done building.
svn:r847
bugfix: keep going when a circ fails in circuit_n_conn_open
(make circuit_enumerate_by_naddr_nport obsolete)
bugfix: make circuit_n_conn_open only look at circ's that start at us
bugfix: only try circuit_n_conn_open if we're an OP. Otherwise we
expect connections to always already be up.
bugfix: when choosing path length, pay attention to whether the directory
says a router is down.
bugfix: when picking good exit, skip routers which are known to be down
(more work needs to be done on this one)
svn:r838
- Exit policies now support bitmasks (18.0.0.0/255.0.0.0) and bitcounts
18.0.0.0/8. Policies are parsed on startup, not when comparing to them.
- desired_path_len is now part of an opaque cpath_build_state_t structure.
- END_REASON_EXITPOLICY cells no longer include a port.
- RELAY_COMMAND_CONNECTED cells now include the IP address we've connected
to.
- connection_edge now has a client_dns cache to remember resolved addresses.
It gets populated by RELAY_COMMAND_CONNECTED cells and END_REASON_EXITPOLICY
cells. It gets used by connection_ap_handshake_send_begin. We don't
compare it to exit policies yet.
svn:r812
