lws has been able to generate client multipart mime as shown
in minimal-http-client-post, but it requires a lot of user
boilerplate to handle the boundary, related transaction header,
and multipart headers.
This patch adds a client creation flag to indicate it will
carry multipart mime, which autocreates the boundary string
and applies the transaction header with it, and an api to
form the boundary headers between the different mime parts
and the terminating boundary.
This affects max header size since we use the latter half
of the pt_serv_buf to prepare the (possibly huge) auth token.
Adapt the pt_serv_buf_size in the hugeurl example.
Some servers set the tx credit to the absolute max and then add to it... this is illegal
(and checked for in h2spec). Add a quirk flag that works around it by reducing the
initial tx credit size by a factor of 16.
This shouldn't be necessary; just END_HEADERS flag should be enough.
But nghttp2 will not talk to us unless we end the stream from our side.
Unfortunately ending the stream at the time we sent the headers means
we cannot support the long poll half-close scheme. So add a quirk
flag to optionally support this behaviour of nghttp2 when the client
is creating the connection.
Refactor everything around ping / pong handling in ws and h2, so there
is instead a protocol-independent validity lws_sul tracking how long it
has been since the last exchange that confirms the operation of the
network connection in both directions.
Clean out periodic role callback and replace the last two role users
with discrete lws_sul for each pt.
It was already correct but add helpers to isolate and deduplicate
processing adding and closing a generically immortal stream.
Change the default 31s h2 network connection timeout to be settable
by .keepalive_timeout if nonzero.
Add a public api allowing a client h2 stream to transition to
half-closed LOCAL (by sending a 0-byte DATA with END_STREAM) and
mark itself as immortal to create a read-only long-poll stream
if the server allows it.
Add a vhost server option flag LWS_SERVER_OPTION_VH_H2_HALF_CLOSED_LONG_POLL
which allows the vhost to treat half-closed remotes as immortal long
poll streams.
Remove LWS_LATENCY.
Add the option LWS_WITH_DETAILED_LATENCY, allowing lws to collect very detailed
information on every read and write, and allow the user code to provide
a callback to process events.
When creating the stream from the nwsi, the stream was created with
its own user_space that gets overwritten with the nwsi one as it is
demoted to be the stream.
Stop that leaking.
lws has been able to proxy h2 or h1 inbound connections to an
h1 onward connection for a while now. It's simple to use just
build with LWS_WITH_HTTP_PROXY and make a mount where the origin
is the onward connection details. Unix sockets can also be
used as the onward connection.
This patch extends the support to be able to also do the same for
inbound h2 or h1 ws upgrades to an h1 ws onward connection as well.
This allows you to offer completely different services in a
common URL space, including ones that connect back by ws / wss.
Since the NUL was always taken care of at the end of buf, the worst it could
do was log some junk. This makes it explicitly log it as oversize separately.
This lets you build using the runtime Address Sanitizer in gcc.
LWS is heavily tested with valgrind routinely during development. But ASAN
did find some theoretical-only issues with shifting, strictly ~(1 << 31) is
a signed int, it should be ~(1u << 31). Gcc does the same for both, but it's
good to have the ability to find these.
This allows the client stuff to understand that addresses beginning with '+'
represent unix sockets.
If the first character after the '+' is '@', it understands that the '@'
should be read as '\0', in order to use Linux "abstract namespace"
sockets.
Further the lws_parse_uri() helper is extended to understand the convention
that an address starting with + is a unix socket, and treats the socket
path as delimited by ':', eg
http://+/var/run/mysocket:/my/path
HTTP Proxy is updated to allow mounts to these unix socket paths.
Proxy connections go out on h1, but are dynamically translated to h1 or h2
on the incoming side.
Proxy usage of libhubbub is separated out... LWS_WITH_HTTP_PROXY is on by
default, and LWS_WITH_HUBBUB is off by default.
This adds a plugin that interfaces to libjsongit2
https://warmcat.com/git/libjsongit2
to provide a per-vhost service for presenting bare git repos in a
web interface.
This changes the vhost destroy flow to only hand off the listen
socket if another vhost sharing it, and mark the vhost as
being_destroyed.
Each tsi calls lws_check_deferred_free() once a second, if it sees
any vhost being_destroyed there, it closes all wsi on its tsi on
the same vhost, one time.
As the wsi on the vhost complete close (ie, after libuv async close
if on libuv event loop), they decrement a reference count for all
wsi open on the vhost. The tsi who closes the last one then
completes the destroy flow for the vhost itself... it's random
which tsi completes the vhost destroy but since there are no
wsi left on the vhost, and it holds the context lock, nothing
can conflict.
The advantage of this is that owning tsi do the close for wsi
that are bound to the vhost under destruction, at a time when
they are guaranteed to be idle for service, and they do it with
both vhost and context locks owned, so no other service thread
can conflict for stuff protected by those either.
For the situation the user code may have allocations attached to
the vhost, this adds args to lws_vhost_destroy() to allow destroying
the user allocations just before the vhost is freed.
