This makes it so the clients don't have to wait for RS to become
responsive, potentially allowing them to do other things while RS
handles the connections.
Fixes#23306.
This really only implements a heuristic, assuming that HTTP/1.0 servers
cannot handle having multiple active connections; this assumption has
lots of false positives, but ultimately HTTP/1.0 is an out-of-date HTTP
version and people using it should just switch to a newer standard
anyway.
Specifically, python's "SimpleHTTPRequestHandler" utilises a
single-threaded HTTP/1.0 server, which means no keepalive and more
importantly, hangs and races with more than a single connection present.
This commit makes it so we serialise all requests to servers that are
known to serve only a single request per connection (aka HTTP/1.0 with
our setup, as we unconditionally request keepalive)
Fixes#22582
Previously, the job and the (cache of them) would lead to a UAF, as
after `.start()` was called on the job it'd be immediately destroyed.
Example of previous bug:
```
// Note due to the cache &jobA == &jobB
auto& jobA = Job::ensure("https://r.bing.com/");
auto& jobB = Job::ensure("https://r.bing.com/");
// Previously, the first .start() free'd the job
jobA.start();
// So the second .start() was a UAF
jobB.start();
```
This commit un-deprecates DeprecatedString, and repurposes it as a byte
string.
As the null state has already been removed, there are no other
particularly hairy blockers in repurposing this type as a byte string
(what it _really_ is).
This commit is auto-generated:
$ xs=$(ack -l \bDeprecatedString\b\|deprecated_string AK Userland \
Meta Ports Ladybird Tests Kernel)
$ perl -pie 's/\bDeprecatedString\b/ByteString/g;
s/deprecated_string/byte_string/g' $xs
$ clang-format --style=file -i \
$(git diff --name-only | grep \.cpp\|\.h)
$ gn format $(git ls-files '*.gn' '*.gni')
Prior to this commit, it was possible to get a socket stuck in a state
with enqueued entries, waiting for a nonexistent request to finish.
This state could be entered by issuing a request immediately after the
completion of another one, and before deferred_invoke execution in the
event loop.
This commit closes this hole by making sure the socket is never in a
state where it can queue requests without an active job.
Since 2023-09-08, Clang trunk has had a bug which causes a segfault when
evaluating certain `requires` expressions inside templated lambdas.
There isn't an imminent fix on the horizon, so let's work around the
issue by specifying the type of the offending lambda arguments
explicitly.
See https://github.com/llvm/llvm-project/issues/67260
In order to follow spec text to achieve this, we need to change the
underlying representation of a host in AK::URL to deserialized format.
Before this, we were parsing the host and then immediately serializing
it again.
Making that change resulted in a whole bunch of fallout.
After this change, callers can access the serialized data through
this concept-host-serializer. The functional end result of this
change is that IPv6 hosts are now correctly serialized to be
surrounded with '[' and ']'.
We have a new, improved string type coming up in AK (OOM aware, no null
state), and while it's going to use UTF-8, the name UTF8String is a
mouthful - so let's free up the String name by renaming the existing
class.
Making the old one have an annoying name will hopefully also help with
quick adoption :^)
`it.is_end()` could be updated to return false for a previously-invalid
iterator after we append a new socket, copy its value out to a local
variable to not hit this behaviour.
This commit converts TLS::TLSv12 to a Core::Stream object, and in the
process allows TLS to now wrap other Core::Stream::Socket objects.
As a large part of LibHTTP and LibGemini depend on LibTLS's interface,
this also converts those to support Core::Stream, which leads to a
simplification of LibHTTP (as there's no need to care about the
underlying socket type anymore).
Note that RequestServer now controls the TLS socket options, which is a
better place anyway, as RS is the first receiver of the user-requested
options (though this is currently not particularly useful).
We previously only replaced disconnected sockets on the queued-request
path, leading to attempts to send requests on a disconnected socket if
the disconnection happened in the deletion grace period.
vdbgln() was responsible for ~10% of samples on pv's flamegraph for
RequestServer (under request_did_finish) when loading github.com in
Browser and recording a whole-system profile. This makes that almost
completely disappear.
Until we're confident that RequestServer doesn't need this runtime debug
dump helper, it's much nicer if everyone has it built in, so they can
simply send a SIGINFO if they see it acting up.
Just as removing individual connections can cause the vector entries to
change positions, adding or removing connections to the cache can also
move the connections around, which would make it possible for a
connection to avoid being deleted (and make the RS spin on the Notifier
for that connection).
This commit makes it so that no connection cache is left when it's
supposed to be deleted.
Fixes a few more RS spins.
Otherwise we'd end up trying to delete the wrong connection if a
connection made before us is deleted.
Fixes _some_ RequestServer spins (though not all...).
This commit also adds a small debug mechanism to RequestServer (which
can be enabled by turning REQUEST_SERVER_DEBUG on), that can dump all
the current active connections in the cache, what they're doing, and how
long they've been doing that by sending it a SIGINFO.
This makes connections (particularly TLS-based ones) do the handshaking
stuff only once.
Currently the cache is configured to keep at most two connections evenly
balanced in queue size, and with a grace period of 10s after the last
queued job has finished (after which the connection will be dropped).
