Make current loglevel and current log callback global
to the library. Attaching them to the nsock pool doesn't
bring any benefit and prevents from logging activity in
code sections that don't have access to a pool (such as
proxy chain specification parsing).
Updated external calls and nsock tests accordingly.
Some errors were not properly propagated. Ensure proxy_resolve() returns
a negative error code and make the caller fatal() when unable to complete
proxy_chain initialization for whatever reason.
This prevents nsock from iterating over the whole list of events at
each runloop, thus improving performance.
It made it necessary to have pointers from the msevents to the event
lists they belong to. The patch therefore also changes gh_list from
autonomous containers to embedded structures.
Added unit tests accordingly and cosmetic changes to make things look
more consistent.
This allows to generically handle all kinds of connexions instead of manually
providing a handler for tcp connect, ssl connect...
The drawback is that would it makes it harder to implement support of SSL
proxies. Not sure whether there's a need though, looks like regular clients
don't handle them at least.
When establishing the tunnel through proxy chain, we need to track status of
each proxy (storing R/W buffers, stats, retries...).
This patch lets proxies store and manage whatever structure they want to have
for this in a Nsock IOD. Since types can differ between proxy types, the
proxy_info are stored as a list of void *, ordered like the proxy nodes.
Proxy backends are selected at runtime. Each proxy exports a list of operations
to the rest of the library. This is similar to the way IO engines are
implemented within nsock.
Externally:
The calling application can build a proxychain object and assign it to one (or
more) NSP. Once a NSP get assigned a proxychain it's not possible to remove
it so that consistency is (should be...) ensured.
Internally:
An IOD comes with a proxychain context structure storing the whole tunnel
state. Also each proxy type now has a table of associated functions to use
as hooks for TCP connects(), read() and write() requests. As a result, adding
support of new proxy type should be easier. Code also gains readability in
comparison to large switch/cases that redirect the execution flow according to
the given proxy type.
