Merge branch 'develop' of github.com:matrix-org/synapse into erikj/admin_exfiltrate_data

84 files changed, 1237 insertions, 1027 deletions

diff --git a/synapse/__init__.py b/synapse/__init__.py
index 119359be68..cf22fabd61 100644
--- a/synapse/__init__.py
+++ b/synapse/__init__.py
@@ -35,4 +35,4 @@ try:
 except ImportError:
     pass
 
-__version__ = "1.0.0"
+__version__ = "1.1.0"
diff --git a/synapse/app/_base.py b/synapse/app/_base.py
index d50a9840d4..1ebb7ae539 100644
--- a/synapse/app/_base.py
+++ b/synapse/app/_base.py
@@ -27,7 +27,7 @@ from twisted.protocols.tls import TLSMemoryBIOFactory
 import synapse
 from synapse.app import check_bind_error
 from synapse.crypto import context_factory
-from synapse.util import PreserveLoggingContext
+from synapse.logging.context import PreserveLoggingContext
 from synapse.util.async_helpers import Linearizer
 from synapse.util.rlimit import change_resource_limit
 from synapse.util.versionstring import get_version_string
@@ -93,33 +93,36 @@ def start_reactor(
     install_dns_limiter(reactor)
 
     def run():
-        # make sure that we run the reactor with the sentinel log context,
-        # otherwise other PreserveLoggingContext instances will get confused
-        # and complain when they see the logcontext arbitrarily swapping
-        # between the sentinel and `run` logcontexts.
-        with PreserveLoggingContext():
-            logger.info("Running")
-
-            change_resource_limit(soft_file_limit)
-            if gc_thresholds:
-                gc.set_threshold(*gc_thresholds)
-            reactor.run()
-
-    if daemonize:
-        if print_pidfile:
-            print(pid_file)
-
-        daemon = Daemonize(
-            app=appname,
-            pid=pid_file,
-            action=run,
-            auto_close_fds=False,
-            verbose=True,
-            logger=logger,
-        )
-        daemon.start()
-    else:
-        run()
+        logger.info("Running")
+        change_resource_limit(soft_file_limit)
+        if gc_thresholds:
+            gc.set_threshold(*gc_thresholds)
+        reactor.run()
+
+    # make sure that we run the reactor with the sentinel log context,
+    # otherwise other PreserveLoggingContext instances will get confused
+    # and complain when they see the logcontext arbitrarily swapping
+    # between the sentinel and `run` logcontexts.
+    #
+    # We also need to drop the logcontext before forking if we're daemonizing,
+    # otherwise the cputime metrics get confused about the per-thread resource usage
+    # appearing to go backwards.
+    with PreserveLoggingContext():
+        if daemonize:
+            if print_pidfile:
+                print(pid_file)
+
+            daemon = Daemonize(
+                app=appname,
+                pid=pid_file,
+                action=run,
+                auto_close_fds=False,
+                verbose=True,
+                logger=logger,
+            )
+            daemon.start()
+        else:
+            run()
 
 
 def quit_with_error(error_string):
diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py
index 9120bdb143..be44249ed6 100644
--- a/synapse/app/appservice.py
+++ b/synapse/app/appservice.py
@@ -26,6 +26,7 @@ from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext, run_in_background
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
@@ -36,7 +37,6 @@ from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/client_reader.py b/synapse/app/client_reader.py
index 90bc79cdda..ff11beca82 100644
--- a/synapse/app/client_reader.py
+++ b/synapse/app/client_reader.py
@@ -27,6 +27,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -64,7 +65,6 @@ from synapse.rest.client.versions import VersionsRestServlet
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py
index ff522e4499..cacad25eac 100644
--- a/synapse/app/event_creator.py
+++ b/synapse/app/event_creator.py
@@ -27,6 +27,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -59,7 +60,6 @@ from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.storage.user_directory import UserDirectoryStore
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/federation_reader.py b/synapse/app/federation_reader.py
index 9421420930..11e80dbae0 100644
--- a/synapse/app/federation_reader.py
+++ b/synapse/app/federation_reader.py
@@ -28,6 +28,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.federation.transport.server import TransportLayerServer
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -48,7 +49,6 @@ from synapse.rest.key.v2 import KeyApiV2Resource
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py
index 969be58d0b..97da7bdcbf 100644
--- a/synapse/app/federation_sender.py
+++ b/synapse/app/federation_sender.py
@@ -27,6 +27,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.federation import send_queue
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext, run_in_background
 from synapse.metrics import RegistryProxy
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
@@ -44,7 +45,6 @@ from synapse.storage.engines import create_engine
 from synapse.types import ReadReceipt
 from synapse.util.async_helpers import Linearizer
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py
index 2fd7d57ebf..417a10bbd2 100644
--- a/synapse/app/frontend_proxy.py
+++ b/synapse/app/frontend_proxy.py
@@ -29,6 +29,7 @@ from synapse.config.logger import setup_logging
 from synapse.http.server import JsonResource
 from synapse.http.servlet import RestServlet, parse_json_object_from_request
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -41,7 +42,6 @@ from synapse.rest.client.v2_alpha._base import client_patterns
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py
index 49da105cf6..639b1429c0 100755
--- a/synapse/app/homeserver.py
+++ b/synapse/app/homeserver.py
@@ -54,6 +54,7 @@ from synapse.federation.transport.server import TransportLayerServer
 from synapse.http.additional_resource import AdditionalResource
 from synapse.http.server import RootRedirect
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
@@ -72,7 +73,6 @@ from synapse.storage.engines import IncorrectDatabaseSetup, create_engine
 from synapse.storage.prepare_database import UpgradeDatabaseException, prepare_database
 from synapse.util.caches import CACHE_SIZE_FACTOR
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.module_loader import load_module
 from synapse.util.rlimit import change_resource_limit
diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py
index cf0e2036c3..f23b9b6eda 100644
--- a/synapse/app/media_repository.py
+++ b/synapse/app/media_repository.py
@@ -27,6 +27,7 @@ from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -40,7 +41,6 @@ from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.storage.media_repository import MediaRepositoryStore
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py
index df29ea5ecb..4f929edf86 100644
--- a/synapse/app/pusher.py
+++ b/synapse/app/pusher.py
@@ -26,6 +26,7 @@ from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext, run_in_background
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import __func__
@@ -38,7 +39,6 @@ from synapse.server import HomeServer
 from synapse.storage import DataStore
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py
index 858949910d..de4797fddc 100644
--- a/synapse/app/synchrotron.py
+++ b/synapse/app/synchrotron.py
@@ -31,6 +31,7 @@ from synapse.config.logger import setup_logging
 from synapse.handlers.presence import PresenceHandler, get_interested_parties
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext, run_in_background
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore, __func__
@@ -57,7 +58,6 @@ from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.storage.presence import UserPresenceState
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.stringutils import random_string
 from synapse.util.versionstring import get_version_string
diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py
index 2d9d2e1bbc..1177ddd72e 100644
--- a/synapse/app/user_dir.py
+++ b/synapse/app/user_dir.py
@@ -28,6 +28,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.logging.context import LoggingContext, run_in_background
 from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
@@ -46,7 +47,6 @@ from synapse.storage.engines import create_engine
 from synapse.storage.user_directory import UserDirectoryStore
 from synapse.util.caches.stream_change_cache import StreamChangeCache
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/appservice/scheduler.py b/synapse/appservice/scheduler.py
index b54bf5411f..e5b36494f5 100644
--- a/synapse/appservice/scheduler.py
+++ b/synapse/appservice/scheduler.py
@@ -53,8 +53,8 @@ import logging
 from twisted.internet import defer
 
 from synapse.appservice import ApplicationServiceState
+from synapse.logging.context import run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util.logcontext import run_in_background
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/config/emailconfig.py b/synapse/config/emailconfig.py
index fcd55d3e3d..8381b8eb29 100644
--- a/synapse/config/emailconfig.py
+++ b/synapse/config/emailconfig.py
@@ -112,13 +112,17 @@ class EmailConfig(Config):
             missing = []
             for k in required:
                 if k not in email_config:
-                    missing.append(k)
+                    missing.append("email." + k)
+
+            if config.get("public_baseurl") is None:
+                missing.append("public_base_url")
 
             if len(missing) > 0:
                 raise RuntimeError(
-                    "email.password_reset_behaviour is set to 'local' "
-                    "but required keys are missing: %s"
-                    % (", ".join(["email." + k for k in missing]),)
+                    "Password resets emails are configured to be sent from "
+                    "this homeserver due to a partial 'email' block. "
+                    "However, the following required keys are missing: %s"
+                    % (", ".join(missing),)
                 )
 
             # Templates for password reset emails
@@ -156,13 +160,6 @@ class EmailConfig(Config):
                 filepath, "email.password_reset_template_success_html"
             )
 
-            if config.get("public_baseurl") is None:
-                raise RuntimeError(
-                    "email.password_reset_behaviour is set to 'local' but no "
-                    "public_baseurl is set. This is necessary to generate password "
-                    "reset links"
-                )
-
         if self.email_enable_notifs:
             required = [
                 "smtp_host",
diff --git a/synapse/config/logger.py b/synapse/config/logger.py
index 931aec41c0..0f5554211c 100644
--- a/synapse/config/logger.py
+++ b/synapse/config/logger.py
@@ -24,7 +24,7 @@ from twisted.logger import STDLibLogObserver, globalLogBeginner
 
 import synapse
 from synapse.app import _base as appbase
-from synapse.util.logcontext import LoggingContextFilter
+from synapse.logging.context import LoggingContextFilter
 from synapse.util.versionstring import get_version_string
 
 from ._base import Config
@@ -40,7 +40,7 @@ formatters:
 
 filters:
     context:
-        (): synapse.util.logcontext.LoggingContextFilter
+        (): synapse.logging.context.LoggingContextFilter
         request: ""
 
 handlers:
diff --git a/synapse/config/ratelimiting.py b/synapse/config/ratelimiting.py
index 8c587f3fd2..33f31cf213 100644
--- a/synapse/config/ratelimiting.py
+++ b/synapse/config/ratelimiting.py
@@ -23,7 +23,7 @@ class RateLimitConfig(object):
 
 class FederationRateLimitConfig(object):
     _items_and_default = {
-        "window_size": 10000,
+        "window_size": 1000,
         "sleep_limit": 10,
         "sleep_delay": 500,
         "reject_limit": 50,
@@ -54,7 +54,7 @@ class RatelimitConfig(Config):
 
         # Load the new-style federation config, if it exists. Otherwise, fall
         # back to the old method.
-        if "federation_rc" in config:
+        if "rc_federation" in config:
             self.rc_federation = FederationRateLimitConfig(**config["rc_federation"])
         else:
             self.rc_federation = FederationRateLimitConfig(
diff --git a/synapse/config/saml2_config.py b/synapse/config/saml2_config.py
index 872a1ba934..6a8161547a 100644
--- a/synapse/config/saml2_config.py
+++ b/synapse/config/saml2_config.py
@@ -12,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+from synapse.python_dependencies import DependencyException, check_requirements
 
 from ._base import Config, ConfigError
 
@@ -25,6 +26,11 @@ class SAML2Config(Config):
         if not saml2_config or not saml2_config.get("enabled", True):
             return
 
+        try:
+            check_requirements("saml2")
+        except DependencyException as e:
+            raise ConfigError(e.message)
+
         self.saml2_enabled = True
 
         import saml2.config
@@ -37,6 +43,11 @@ class SAML2Config(Config):
         if config_path is not None:
             self.saml2_sp_config.load_file(config_path)
 
+        # session lifetime: in milliseconds
+        self.saml2_session_lifetime = self.parse_duration(
+            saml2_config.get("saml_session_lifetime", "5m")
+        )
+
     def _default_saml_config_dict(self):
         import saml2
 
@@ -72,6 +83,12 @@ class SAML2Config(Config):
         # so it is not normally necessary to specify them unless you need to
         # override them.
         #
+        # Once SAML support is enabled, a metadata file will be exposed at
+        # https://<server>:<port>/_matrix/saml2/metadata.xml, which you may be able to
+        # use to configure your SAML IdP with. Alternatively, you can manually configure
+        # the IdP to use an ACS location of
+        # https://<server>:<port>/_matrix/saml2/authn_response.
+        #
         #saml2_config:
         #  sp_config:
         #    # point this to the IdP's metadata. You can use either a local file or
@@ -81,7 +98,15 @@ class SAML2Config(Config):
         #      remote:
         #        - url: https://our_idp/metadata.xml
         #
-        #    # The rest of sp_config is just used to generate our metadata xml, and you
+        #    # By default, the user has to go to our login page first. If you'd like to
+        #    # allow IdP-initiated login, set 'allow_unsolicited: True' in a
+        #    # 'service.sp' section:
+        #    #
+        #    #service:
+        #    #  sp:
+        #    #    allow_unsolicited: True
+        #
+        #    # The examples below are just used to generate our metadata xml, and you
         #    # may well not need it, depending on your setup. Alternatively you
         #    # may need a whole lot more detail - see the pysaml2 docs!
         #
@@ -104,6 +129,12 @@ class SAML2Config(Config):
         #  # separate pysaml2 configuration file:
         #  #
         #  config_path: "%(config_dir_path)s/sp_conf.py"
+        #
+        #  # the lifetime of a SAML session. This defines how long a user has to
+        #  # complete the authentication process, if allow_unsolicited is unset.
+        #  # The default is 5 minutes.
+        #  #
+        #  # saml_session_lifetime: 5m
         """ % {
             "config_dir_path": config_dir_path
         }
diff --git a/synapse/crypto/keyring.py b/synapse/crypto/keyring.py
index 10c2eb7f0f..341c863152 100644
--- a/synapse/crypto/keyring.py
+++ b/synapse/crypto/keyring.py
@@ -44,15 +44,16 @@ from synapse.api.errors import (
     RequestSendFailed,
     SynapseError,
 )
-from synapse.storage.keys import FetchKeyResult
-from synapse.util import logcontext, unwrapFirstError
-from synapse.util.async_helpers import yieldable_gather_results
-from synapse.util.logcontext import (
+from synapse.logging.context import (
     LoggingContext,
     PreserveLoggingContext,
+    make_deferred_yieldable,
     preserve_fn,
     run_in_background,
 )
+from synapse.storage.keys import FetchKeyResult
+from synapse.util import unwrapFirstError
+from synapse.util.async_helpers import yieldable_gather_results
 from synapse.util.metrics import Measure
 from synapse.util.retryutils import NotRetryingDestination
 
@@ -140,7 +141,7 @@ class Keyring(object):
         """
         req = VerifyJsonRequest(server_name, json_object, validity_time, request_name)
         requests = (req,)
-        return logcontext.make_deferred_yieldable(self._verify_objects(requests)[0])
+        return make_deferred_yieldable(self._verify_objects(requests)[0])
 
     def verify_json_objects_for_server(self, server_and_json):
         """Bulk verifies signatures of json objects, bulk fetching keys as
@@ -557,7 +558,7 @@ class BaseV2KeyFetcher(object):
 
         signed_key_json_bytes = encode_canonical_json(signed_key_json)
 
-        yield logcontext.make_deferred_yieldable(
+        yield make_deferred_yieldable(
             defer.gatherResults(
                 [
                     run_in_background(
@@ -612,7 +613,7 @@ class PerspectivesKeyFetcher(BaseV2KeyFetcher):
 
             defer.returnValue({})
 
-        results = yield logcontext.make_deferred_yieldable(
+        results = yield make_deferred_yieldable(
             defer.gatherResults(
                 [run_in_background(get_key, server) for server in self.key_servers],
                 consumeErrors=True,
diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py
index a96cdada3d..a9545e6c1b 100644
--- a/synapse/events/snapshot.py
+++ b/synapse/events/snapshot.py
@@ -19,7 +19,7 @@ from frozendict import frozendict
 
 from twisted.internet import defer
 
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 
 
 class EventContext(object):
diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py
index 1e925b19e7..f7bb806ae7 100644
--- a/synapse/federation/federation_base.py
+++ b/synapse/federation/federation_base.py
@@ -27,8 +27,14 @@ from synapse.crypto.event_signing import check_event_content_hash
 from synapse.events import event_type_from_format_version
 from synapse.events.utils import prune_event
 from synapse.http.servlet import assert_params_in_dict
+from synapse.logging.context import (
+    LoggingContext,
+    PreserveLoggingContext,
+    make_deferred_yieldable,
+    preserve_fn,
+)
 from synapse.types import get_domain_from_id
-from synapse.util import logcontext, unwrapFirstError
+from synapse.util import unwrapFirstError
 
 logger = logging.getLogger(__name__)
 
@@ -73,7 +79,7 @@ class FederationBase(object):
         @defer.inlineCallbacks
         def handle_check_result(pdu, deferred):
             try:
-                res = yield logcontext.make_deferred_yieldable(deferred)
+                res = yield make_deferred_yieldable(deferred)
             except SynapseError:
                 res = None
 
@@ -102,10 +108,10 @@ class FederationBase(object):
 
             defer.returnValue(res)
 
-        handle = logcontext.preserve_fn(handle_check_result)
+        handle = preserve_fn(handle_check_result)
         deferreds2 = [handle(pdu, deferred) for pdu, deferred in zip(pdus, deferreds)]
 
-        valid_pdus = yield logcontext.make_deferred_yieldable(
+        valid_pdus = yield make_deferred_yieldable(
             defer.gatherResults(deferreds2, consumeErrors=True)
         ).addErrback(unwrapFirstError)
 
@@ -115,7 +121,7 @@ class FederationBase(object):
             defer.returnValue([p for p in valid_pdus if p])
 
     def _check_sigs_and_hash(self, room_version, pdu):
-        return logcontext.make_deferred_yieldable(
+        return make_deferred_yieldable(
             self._check_sigs_and_hashes(room_version, [pdu])[0]
         )
 
@@ -133,14 +139,14 @@ class FederationBase(object):
               * returns a redacted version of the event (if the signature
                 matched but the hash did not)
               * throws a SynapseError if the signature check failed.
-            The deferreds run their callbacks in the sentinel logcontext.
+            The deferreds run their callbacks in the sentinel
         """
         deferreds = _check_sigs_on_pdus(self.keyring, room_version, pdus)
 
-        ctx = logcontext.LoggingContext.current_context()
+        ctx = LoggingContext.current_context()
 
         def callback(_, pdu):
-            with logcontext.PreserveLoggingContext(ctx):
+            with PreserveLoggingContext(ctx):
                 if not check_event_content_hash(pdu):
                     # let's try to distinguish between failures because the event was
                     # redacted (which are somewhat expected) vs actual ball-tampering
@@ -178,7 +184,7 @@ class FederationBase(object):
 
         def errback(failure, pdu):
             failure.trap(SynapseError)
-            with logcontext.PreserveLoggingContext(ctx):
+            with PreserveLoggingContext(ctx):
                 logger.warn(
                     "Signature check failed for %s: %s",
                     pdu.event_id,
diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py
index 3883eb525e..3cb4b94420 100644
--- a/synapse/federation/federation_client.py
+++ b/synapse/federation/federation_client.py
@@ -39,10 +39,10 @@ from synapse.api.room_versions import (
 )
 from synapse.events import builder, room_version_to_event_format
 from synapse.federation.federation_base import FederationBase, event_from_pdu_json
-from synapse.util import logcontext, unwrapFirstError
+from synapse.logging.context import make_deferred_yieldable, run_in_background
+from synapse.logging.utils import log_function
+from synapse.util import unwrapFirstError
 from synapse.util.caches.expiringcache import ExpiringCache
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
-from synapse.util.logutils import log_function
 from synapse.util.retryutils import NotRetryingDestination
 
 logger = logging.getLogger(__name__)
@@ -207,7 +207,7 @@ class FederationClient(FederationBase):
         ]
 
         # FIXME: We should handle signature failures more gracefully.
-        pdus[:] = yield logcontext.make_deferred_yieldable(
+        pdus[:] = yield make_deferred_yieldable(
             defer.gatherResults(
                 self._check_sigs_and_hashes(room_version, pdus), consumeErrors=True
             ).addErrback(unwrapFirstError)
diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py
index 2e0cebb638..8c0a18b120 100644
--- a/synapse/federation/federation_server.py
+++ b/synapse/federation/federation_server.py
@@ -42,6 +42,8 @@ from synapse.federation.federation_base import FederationBase, event_from_pdu_js
 from synapse.federation.persistence import TransactionActions
 from synapse.federation.units import Edu, Transaction
 from synapse.http.endpoint import parse_server_name
+from synapse.logging.context import nested_logging_context
+from synapse.logging.utils import log_function
 from synapse.replication.http.federation import (
     ReplicationFederationSendEduRestServlet,
     ReplicationGetQueryRestServlet,
@@ -50,8 +52,6 @@ from synapse.types import get_domain_from_id
 from synapse.util import glob_to_regex
 from synapse.util.async_helpers import Linearizer, concurrently_execute
 from synapse.util.caches.response_cache import ResponseCache
-from synapse.util.logcontext import nested_logging_context
-from synapse.util.logutils import log_function
 
 # when processing incoming transactions, we try to handle multiple rooms in
 # parallel, up to this limit.
diff --git a/synapse/federation/persistence.py b/synapse/federation/persistence.py
index 7535f79203..44edcabed4 100644
--- a/synapse/federation/persistence.py
+++ b/synapse/federation/persistence.py
@@ -21,9 +21,7 @@ These actions are mostly only used by the :py:mod:`.replication` module.
 
 import logging
 
-from twisted.internet import defer
-
-from synapse.util.logutils import log_function
+from synapse.logging.utils import log_function
 
 logger = logging.getLogger(__name__)
 
@@ -63,33 +61,3 @@ class TransactionActions(object):
         return self.store.set_received_txn_response(
             transaction.transaction_id, origin, code, response
         )
-
-    @defer.inlineCallbacks
-    @log_function
-    def prepare_to_send(self, transaction):
-        """ Persists the `Transaction` we are about to send and works out the
-        correct value for the `prev_ids` key.
-
-        Returns:
-            Deferred
-        """
-        transaction.prev_ids = yield self.store.prep_send_transaction(
-            transaction.transaction_id,
-            transaction.destination,
-            transaction.origin_server_ts,
-        )
-
-    @log_function
-    def delivered(self, transaction, response_code, response_dict):
-        """ Marks the given `Transaction` as having been successfully
-        delivered to the remote homeserver, and what the response was.
-
-        Returns:
-            Deferred
-        """
-        return self.store.delivered_txn(
-            transaction.transaction_id,
-            transaction.destination,
-            response_code,
-            response_dict,
-        )
diff --git a/synapse/federation/sender/__init__.py b/synapse/federation/sender/__init__.py
index 766c5a37cd..d46f4aaeb1 100644
--- a/synapse/federation/sender/__init__.py
+++ b/synapse/federation/sender/__init__.py
@@ -26,6 +26,11 @@ from synapse.federation.sender.per_destination_queue import PerDestinationQueue
 from synapse.federation.sender.transaction_manager import TransactionManager
 from synapse.federation.units import Edu
 from synapse.handlers.presence import get_interested_remotes
+from synapse.logging.context import (
+    make_deferred_yieldable,
+    preserve_fn,
+    run_in_background,
+)
 from synapse.metrics import (
     LaterGauge,
     event_processing_loop_counter,
@@ -33,7 +38,6 @@ from synapse.metrics import (
     events_processed_counter,
 )
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util import logcontext
 from synapse.util.metrics import measure_func
 
 logger = logging.getLogger(__name__)
@@ -210,10 +214,10 @@ class FederationSender(object):
                 for event in events:
                     events_by_room.setdefault(event.room_id, []).append(event)
 
-                yield logcontext.make_deferred_yieldable(
+                yield make_deferred_yieldable(
                     defer.gatherResults(
                         [
-                            logcontext.run_in_background(handle_room_events, evs)
+                            run_in_background(handle_room_events, evs)
                             for evs in itervalues(events_by_room)
                         ],
                         consumeErrors=True,
@@ -360,7 +364,7 @@ class FederationSender(object):
         for queue in queues:
             queue.flush_read_receipts_for_room(room_id)
 
-    @logcontext.preserve_fn  # the caller should not yield on this
+    @preserve_fn  # the caller should not yield on this
     @defer.inlineCallbacks
     def send_presence(self, states):
         """Send the new presence states to the appropriate destinations.
diff --git a/synapse/federation/sender/transaction_manager.py b/synapse/federation/sender/transaction_manager.py
index c987bb9a0d..0460a8c4ac 100644
--- a/synapse/federation/sender/transaction_manager.py
+++ b/synapse/federation/sender/transaction_manager.py
@@ -63,8 +63,6 @@ class TransactionManager(object):
             len(edus),
         )
 
-        logger.debug("TX [%s] Persisting transaction...", destination)
-
         transaction = Transaction.create_new(
             origin_server_ts=int(self.clock.time_msec()),
             transaction_id=txn_id,
@@ -76,9 +74,6 @@ class TransactionManager(object):
 
         self._next_txn_id += 1
 
-        yield self._transaction_actions.prepare_to_send(transaction)
-
-        logger.debug("TX [%s] Persisted transaction", destination)
         logger.info(
             "TX [%s] {%s} Sending transaction [%s]," " (PDUs: %d, EDUs: %d)",
             destination,
@@ -118,10 +113,6 @@ class TransactionManager(object):
 
         logger.info("TX [%s] {%s} got %d response", destination, txn_id, code)
 
-        yield self._transaction_actions.delivered(transaction, code, response)
-
-        logger.debug("TX [%s] {%s} Marked as delivered", destination, txn_id)
-
         if code == 200:
             for e_id, r in response.get("pdus", {}).items():
                 if "error" in r:
diff --git a/synapse/federation/transport/client.py b/synapse/federation/transport/client.py
index aecd142309..1aae9ec9e7 100644
--- a/synapse/federation/transport/client.py
+++ b/synapse/federation/transport/client.py
@@ -22,7 +22,7 @@ from twisted.internet import defer
 
 from synapse.api.constants import Membership
 from synapse.api.urls import FEDERATION_V1_PREFIX, FEDERATION_V2_PREFIX
-from synapse.util.logutils import log_function
+from synapse.logging.utils import log_function
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/federation/transport/server.py b/synapse/federation/transport/server.py
index 955f0f4308..2efdcff7ef 100644
--- a/synapse/federation/transport/server.py
+++ b/synapse/federation/transport/server.py
@@ -36,8 +36,8 @@ from synapse.http.servlet import (
     parse_json_object_from_request,
     parse_string_from_args,
 )
+from synapse.logging.context import run_in_background
 from synapse.types import ThirdPartyInstanceID, get_domain_from_id
-from synapse.util.logcontext import run_in_background
 from synapse.util.ratelimitutils import FederationRateLimiter
 from synapse.util.versionstring import get_version_string
 
diff --git a/synapse/groups/attestations.py b/synapse/groups/attestations.py
index e73757570c..f497711133 100644
--- a/synapse/groups/attestations.py
+++ b/synapse/groups/attestations.py
@@ -43,9 +43,9 @@ from signedjson.sign import sign_json
 from twisted.internet import defer
 
 from synapse.api.errors import HttpResponseException, RequestSendFailed, SynapseError
+from synapse.logging.context import run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import get_domain_from_id
-from synapse.util.logcontext import run_in_background
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/handlers/account_validity.py b/synapse/handlers/account_validity.py
index 0719da3ab7..1f1708ba7d 100644
--- a/synapse/handlers/account_validity.py
+++ b/synapse/handlers/account_validity.py
@@ -22,9 +22,10 @@ from email.mime.text import MIMEText
 from twisted.internet import defer
 
 from synapse.api.errors import StoreError
+from synapse.logging.context import make_deferred_yieldable
+from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import UserID
 from synapse.util import stringutils
-from synapse.util.logcontext import make_deferred_yieldable
 
 try:
     from synapse.push.mailer import load_jinja2_templates
@@ -67,7 +68,14 @@ class AccountValidityHandler(object):
             )
 
             # Check the renewal emails to send and send them every 30min.
-            self.clock.looping_call(self.send_renewal_emails, 30 * 60 * 1000)
+            def send_emails():
+                # run as a background process to make sure that the database transactions
+                # have a logcontext to report to
+                return run_as_background_process(
+                    "send_renewals", self.send_renewal_emails
+                )
+
+            self.clock.looping_call(send_emails, 30 * 60 * 1000)
 
     @defer.inlineCallbacks
     def send_renewal_emails(self):
diff --git a/synapse/handlers/appservice.py b/synapse/handlers/appservice.py
index 5cc89d43f6..8f089f0e33 100644
--- a/synapse/handlers/appservice.py
+++ b/synapse/handlers/appservice.py
@@ -23,13 +23,13 @@ from twisted.internet import defer
 
 import synapse
 from synapse.api.constants import EventTypes
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.metrics import (
     event_processing_loop_counter,
     event_processing_loop_room_count,
 )
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.util import log_failure
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.metrics import Measure
 
 logger = logging.getLogger(__name__)
diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py
index c8c1ed3246..ef5585aa99 100644
--- a/synapse/handlers/auth.py
+++ b/synapse/handlers/auth.py
@@ -36,9 +36,9 @@ from synapse.api.errors import (
     SynapseError,
 )
 from synapse.api.ratelimiting import Ratelimiter
+from synapse.logging.context import defer_to_thread
 from synapse.module_api import ModuleApi
 from synapse.types import UserID
-from synapse.util import logcontext
 from synapse.util.caches.expiringcache import ExpiringCache
 
 from ._base import BaseHandler
@@ -987,7 +987,7 @@ class AuthHandler(BaseHandler):
                 bcrypt.gensalt(self.bcrypt_rounds),
             ).decode("ascii")
 
-        return logcontext.defer_to_thread(self.hs.get_reactor(), _do_hash)
+        return defer_to_thread(self.hs.get_reactor(), _do_hash)
 
     def validate_hash(self, password, stored_hash):
         """Validates that self.hash(password) == stored_hash.
@@ -1013,7 +1013,7 @@ class AuthHandler(BaseHandler):
             if not isinstance(stored_hash, bytes):
                 stored_hash = stored_hash.encode("ascii")
 
-            return logcontext.defer_to_thread(self.hs.get_reactor(), _do_validate_hash)
+            return defer_to_thread(self.hs.get_reactor(), _do_validate_hash)
         else:
             return defer.succeed(False)
 
diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py
index 807900fe52..55b4ab3a1a 100644
--- a/synapse/handlers/e2e_keys.py
+++ b/synapse/handlers/e2e_keys.py
@@ -23,8 +23,8 @@ from canonicaljson import encode_canonical_json, json
 from twisted.internet import defer
 
 from synapse.api.errors import CodeMessageException, FederationDeniedError, SynapseError
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.types import UserID, get_domain_from_id
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.retryutils import NotRetryingDestination
 
 logger = logging.getLogger(__name__)
diff --git a/synapse/handlers/events.py b/synapse/handlers/events.py
index 5836d3c639..6a38328af3 100644
--- a/synapse/handlers/events.py
+++ b/synapse/handlers/events.py
@@ -21,8 +21,8 @@ from twisted.internet import defer
 from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import AuthError, SynapseError
 from synapse.events import EventBase
+from synapse.logging.utils import log_function
 from synapse.types import UserID
-from synapse.util.logutils import log_function
 from synapse.visibility import filter_events_for_client
 
 from ._base import BaseHandler
diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py
index 02d397c498..57be968c67 100644
--- a/synapse/handlers/federation.py
+++ b/synapse/handlers/federation.py
@@ -45,6 +45,13 @@ from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersions
 from synapse.crypto.event_signing import compute_event_signature
 from synapse.event_auth import auth_types_for_event
 from synapse.events.validator import EventValidator
+from synapse.logging.context import (
+    make_deferred_yieldable,
+    nested_logging_context,
+    preserve_fn,
+    run_in_background,
+)
+from synapse.logging.utils import log_function
 from synapse.replication.http.federation import (
     ReplicationCleanRoomRestServlet,
     ReplicationFederationSendEventsRestServlet,
@@ -52,10 +59,9 @@ from synapse.replication.http.federation import (
 from synapse.replication.http.membership import ReplicationUserJoinedLeftRoomRestServlet
 from synapse.state import StateResolutionStore, resolve_events_with_store
 from synapse.types import UserID, get_domain_from_id
-from synapse.util import logcontext, unwrapFirstError
+from synapse.util import unwrapFirstError
 from synapse.util.async_helpers import Linearizer
 from synapse.util.distributor import user_joined_room
-from synapse.util.logutils import log_function
 from synapse.util.retryutils import NotRetryingDestination
 from synapse.visibility import filter_events_for_server
 
@@ -338,7 +344,7 @@ class FederationHandler(BaseHandler):
 
                         room_version = yield self.store.get_room_version(room_id)
 
-                        with logcontext.nested_logging_context(p):
+                        with nested_logging_context(p):
                             # note that if any of the missing prevs share missing state or
                             # auth events, the requests to fetch those events are deduped
                             # by the get_pdu_cache in federation_client.
@@ -532,7 +538,7 @@ class FederationHandler(BaseHandler):
                 event_id,
                 ev.event_id,
             )
-            with logcontext.nested_logging_context(ev.event_id):
+            with nested_logging_context(ev.event_id):
                 try:
                     yield self.on_receive_pdu(origin, ev, sent_to_us_directly=False)
                 except FederationError as e:
@@ -725,10 +731,10 @@ class FederationHandler(BaseHandler):
                     missing_auth - failed_to_fetch,
                 )
 
-                results = yield logcontext.make_deferred_yieldable(
+                results = yield make_deferred_yieldable(
                     defer.gatherResults(
                         [
-                            logcontext.run_in_background(
+                            run_in_background(
                                 self.federation_client.get_pdu,
                                 [dest],
                                 event_id,
@@ -994,10 +1000,8 @@ class FederationHandler(BaseHandler):
         event_ids = list(extremities.keys())
 
         logger.debug("calling resolve_state_groups in _maybe_backfill")
-        resolve = logcontext.preserve_fn(
-            self.state_handler.resolve_state_groups_for_events
-        )
-        states = yield logcontext.make_deferred_yieldable(
+        resolve = preserve_fn(self.state_handler.resolve_state_groups_for_events)
+        states = yield make_deferred_yieldable(
             defer.gatherResults(
                 [resolve(room_id, [e]) for e in event_ids], consumeErrors=True
             )
@@ -1171,7 +1175,7 @@ class FederationHandler(BaseHandler):
             # lots of requests for missing prev_events which we do actually
             # have. Hence we fire off the deferred, but don't wait for it.
 
-            logcontext.run_in_background(self._handle_queued_pdus, room_queue)
+            run_in_background(self._handle_queued_pdus, room_queue)
 
         defer.returnValue(True)
 
@@ -1191,7 +1195,7 @@ class FederationHandler(BaseHandler):
                     p.event_id,
                     p.room_id,
                 )
-                with logcontext.nested_logging_context(p.event_id):
+                with nested_logging_context(p.event_id):
                     yield self.on_receive_pdu(origin, p, sent_to_us_directly=True)
             except Exception as e:
                 logger.warn(
@@ -1610,7 +1614,7 @@ class FederationHandler(BaseHandler):
             success = True
         finally:
             if not success:
-                logcontext.run_in_background(
+                run_in_background(
                     self.store.remove_push_actions_from_staging, event.event_id
                 )
 
@@ -1629,7 +1633,7 @@ class FederationHandler(BaseHandler):
         @defer.inlineCallbacks
         def prep(ev_info):
             event = ev_info["event"]
-            with logcontext.nested_logging_context(suffix=event.event_id):
+            with nested_logging_context(suffix=event.event_id):
                 res = yield self._prep_event(
                     origin,
                     event,
@@ -1639,12 +1643,9 @@ class FederationHandler(BaseHandler):
                 )
             defer.returnValue(res)
 
-        contexts = yield logcontext.make_deferred_yieldable(
+        contexts = yield make_deferred_yieldable(
             defer.gatherResults(
-                [
-                    logcontext.run_in_background(prep, ev_info)
-                    for ev_info in event_infos
-                ],
+                [run_in_background(prep, ev_info) for ev_info in event_infos],
                 consumeErrors=True,
             )
         )
@@ -2106,10 +2107,10 @@ class FederationHandler(BaseHandler):
 
         room_version = yield self.store.get_room_version(event.room_id)
 
-        different_events = yield logcontext.make_deferred_yieldable(
+        different_events = yield make_deferred_yieldable(
             defer.gatherResults(
                 [
-                    logcontext.run_in_background(
+                    run_in_background(
                         self.store.get_event, d, allow_none=True, allow_rejected=False
                     )
                     for d in different_auth
diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py
index a1fe9d116f..54c966c8a6 100644
--- a/synapse/handlers/initial_sync.py
+++ b/synapse/handlers/initial_sync.py
@@ -21,12 +21,12 @@ from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import AuthError, Codes, SynapseError
 from synapse.events.validator import EventValidator
 from synapse.handlers.presence import format_user_presence_state
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.streams.config import PaginationConfig
 from synapse.types import StreamToken, UserID
 from synapse.util import unwrapFirstError
 from synapse.util.async_helpers import concurrently_execute
 from synapse.util.caches.snapshot_cache import SnapshotCache
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.visibility import filter_events_for_client
 
 from ._base import BaseHandler
diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py
index 683da6bf32..eaeda7a5cb 100644
--- a/synapse/handlers/message.py
+++ b/synapse/handlers/message.py
@@ -34,13 +34,13 @@ from synapse.api.errors import (
 from synapse.api.room_versions import RoomVersions
 from synapse.api.urls import ConsentURIBuilder
 from synapse.events.validator import EventValidator
+from synapse.logging.context import run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.replication.http.send_event import ReplicationSendEventRestServlet
 from synapse.storage.state import StateFilter
 from synapse.types import RoomAlias, UserID, create_requester
 from synapse.util.async_helpers import Linearizer
 from synapse.util.frozenutils import frozendict_json_encoder
-from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import measure_func
 from synapse.visibility import filter_events_for_client
 
diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py
index 76ee97ddd3..20bcfed334 100644
--- a/synapse/handlers/pagination.py
+++ b/synapse/handlers/pagination.py
@@ -20,10 +20,10 @@ from twisted.python.failure import Failure
 
 from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import SynapseError
+from synapse.logging.context import run_in_background
 from synapse.storage.state import StateFilter
 from synapse.types import RoomStreamToken
 from synapse.util.async_helpers import ReadWriteLock
-from synapse.util.logcontext import run_in_background
 from synapse.util.stringutils import random_string
 from synapse.visibility import filter_events_for_client
 
diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py
index c80dc2eba0..6f3537e435 100644
--- a/synapse/handlers/presence.py
+++ b/synapse/handlers/presence.py
@@ -34,14 +34,14 @@ from twisted.internet import defer
 import synapse.metrics
 from synapse.api.constants import EventTypes, Membership, PresenceState
 from synapse.api.errors import SynapseError
+from synapse.logging.context import run_in_background
+from synapse.logging.utils import log_function
 from synapse.metrics import LaterGauge
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.storage.presence import UserPresenceState
 from synapse.types import UserID, get_domain_from_id
 from synapse.util.async_helpers import Linearizer
 from synapse.util.caches.descriptors import cachedInlineCallbacks
-from synapse.util.logcontext import run_in_background
-from synapse.util.logutils import log_function
 from synapse.util.metrics import Measure
 from synapse.util.wheel_timer import WheelTimer
 
diff --git a/synapse/handlers/saml_handler.py b/synapse/handlers/saml_handler.py
new file mode 100644
index 0000000000..a1ce6929cf
--- /dev/null
+++ b/synapse/handlers/saml_handler.py
@@ -0,0 +1,123 @@
+# -*- coding: utf-8 -*-
+# Copyright 2019 The Matrix.org Foundation C.I.C.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import logging
+
+import attr
+import saml2
+from saml2.client import Saml2Client
+
+from synapse.api.errors import SynapseError
+from synapse.http.servlet import parse_string
+from synapse.rest.client.v1.login import SSOAuthHandler
+
+logger = logging.getLogger(__name__)
+
+
+class SamlHandler:
+    def __init__(self, hs):
+        self._saml_client = Saml2Client(hs.config.saml2_sp_config)
+        self._sso_auth_handler = SSOAuthHandler(hs)
+
+        # a map from saml session id to Saml2SessionData object
+        self._outstanding_requests_dict = {}
+
+        self._clock = hs.get_clock()
+        self._saml2_session_lifetime = hs.config.saml2_session_lifetime
+
+    def handle_redirect_request(self, client_redirect_url):
+        """Handle an incoming request to /login/sso/redirect
+
+        Args:
+            client_redirect_url (bytes): the URL that we should redirect the
+                client to when everything is done
+
+        Returns:
+            bytes: URL to redirect to
+        """
+        reqid, info = self._saml_client.prepare_for_authenticate(
+            relay_state=client_redirect_url
+        )
+
+        now = self._clock.time_msec()
+        self._outstanding_requests_dict[reqid] = Saml2SessionData(creation_time=now)
+
+        for key, value in info["headers"]:
+            if key == "Location":
+                return value
+
+        # this shouldn't happen!
+        raise Exception("prepare_for_authenticate didn't return a Location header")
+
+    def handle_saml_response(self, request):
+        """Handle an incoming request to /_matrix/saml2/authn_response
+
+        Args:
+            request (SynapseRequest): the incoming request from the browser. We'll
+                respond to it with a redirect.
+
+        Returns:
+            Deferred[none]: Completes once we have handled the request.
+        """
+        resp_bytes = parse_string(request, "SAMLResponse", required=True)
+        relay_state = parse_string(request, "RelayState", required=True)
+
+        # expire outstanding sessions before parse_authn_request_response checks
+        # the dict.
+        self.expire_sessions()
+
+        try:
+            saml2_auth = self._saml_client.parse_authn_request_response(
+                resp_bytes,
+                saml2.BINDING_HTTP_POST,
+                outstanding=self._outstanding_requests_dict,
+            )
+        except Exception as e:
+            logger.warning("Exception parsing SAML2 response: %s", e)
+            raise SynapseError(400, "Unable to parse SAML2 response: %s" % (e,))
+
+        if saml2_auth.not_signed:
+            logger.warning("SAML2 response was not signed")
+            raise SynapseError(400, "SAML2 response was not signed")
+
+        if "uid" not in saml2_auth.ava:
+            logger.warning("SAML2 response lacks a 'uid' attestation")
+            raise SynapseError(400, "uid not in SAML2 response")
+
+        self._outstanding_requests_dict.pop(saml2_auth.in_response_to, None)
+
+        username = saml2_auth.ava["uid"][0]
+        displayName = saml2_auth.ava.get("displayName", [None])[0]
+
+        return self._sso_auth_handler.on_successful_auth(
+            username, request, relay_state, user_display_name=displayName
+        )
+
+    def expire_sessions(self):
+        expire_before = self._clock.time_msec() - self._saml2_session_lifetime
+        to_expire = set()
+        for reqid, data in self._outstanding_requests_dict.items():
+            if data.creation_time < expire_before:
+                to_expire.add(reqid)
+        for reqid in to_expire:
+            logger.debug("Expiring session id %s", reqid)
+            del self._outstanding_requests_dict[reqid]
+
+
+@attr.s
+class Saml2SessionData:
+    """Data we track about SAML2 sessions"""
+
+    # time the session was created, in milliseconds
+    creation_time = attr.ib()
diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py
index a3f550554f..cd1ac0a27a 100644
--- a/synapse/handlers/sync.py
+++ b/synapse/handlers/sync.py
@@ -25,6 +25,7 @@ from prometheus_client import Counter
 from twisted.internet import defer
 
 from synapse.api.constants import EventTypes, Membership
+from synapse.logging.context import LoggingContext
 from synapse.push.clientformat import format_push_rules_for_user
 from synapse.storage.roommember import MemberSummary
 from synapse.storage.state import StateFilter
@@ -33,7 +34,6 @@ from synapse.util.async_helpers import concurrently_execute
 from synapse.util.caches.expiringcache import ExpiringCache
 from synapse.util.caches.lrucache import LruCache
 from synapse.util.caches.response_cache import ResponseCache
-from synapse.util.logcontext import LoggingContext
 from synapse.util.metrics import Measure, measure_func
 from synapse.visibility import filter_events_for_client
 
diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py
index f8062c8671..c3e0c8fc7e 100644
--- a/synapse/handlers/typing.py
+++ b/synapse/handlers/typing.py
@@ -19,9 +19,9 @@ from collections import namedtuple
 from twisted.internet import defer
 
 from synapse.api.errors import AuthError, SynapseError
+from synapse.logging.context import run_in_background
 from synapse.types import UserID, get_domain_from_id
 from synapse.util.caches.stream_change_cache import StreamChangeCache
-from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import Measure
 from synapse.util.wheel_timer import WheelTimer
 
diff --git a/synapse/http/client.py b/synapse/http/client.py
index 9bc7035c8d..45d5010952 100644
--- a/synapse/http/client.py
+++ b/synapse/http/client.py
@@ -45,9 +45,9 @@ from synapse.http import (
     cancelled_to_request_timed_out_error,
     redact_uri,
 )
+from synapse.logging.context import make_deferred_yieldable
 from synapse.util.async_helpers import timeout_deferred
 from synapse.util.caches import CACHE_SIZE_FACTOR
-from synapse.util.logcontext import make_deferred_yieldable
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/http/federation/matrix_federation_agent.py b/synapse/http/federation/matrix_federation_agent.py
index 414cde0777..054c321a20 100644
--- a/synapse/http/federation/matrix_federation_agent.py
+++ b/synapse/http/federation/matrix_federation_agent.py
@@ -30,9 +30,9 @@ from twisted.web.http_headers import Headers
 from twisted.web.iweb import IAgent
 
 from synapse.http.federation.srv_resolver import SrvResolver, pick_server_from_list
+from synapse.logging.context import make_deferred_yieldable
 from synapse.util import Clock
 from synapse.util.caches.ttlcache import TTLCache
-from synapse.util.logcontext import make_deferred_yieldable
 from synapse.util.metrics import Measure
 
 # period to cache .well-known results for by default
diff --git a/synapse/http/federation/srv_resolver.py b/synapse/http/federation/srv_resolver.py
index 1f22f78a75..ecc88f9b96 100644
--- a/synapse/http/federation/srv_resolver.py
+++ b/synapse/http/federation/srv_resolver.py
@@ -25,7 +25,7 @@ from twisted.internet.error import ConnectError
 from twisted.names import client, dns
 from twisted.names.error import DNSNameError, DomainError
 
-from synapse.util.logcontext import make_deferred_yieldable
+from synapse.logging.context import make_deferred_yieldable
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py
index 5ef8bb60a3..dee3710f68 100644
--- a/synapse/http/matrixfederationclient.py
+++ b/synapse/http/matrixfederationclient.py
@@ -48,8 +48,8 @@ from synapse.api.errors import (
 from synapse.http import QuieterFileBodyProducer
 from synapse.http.client import BlacklistingAgentWrapper, IPBlacklistingResolver
 from synapse.http.federation.matrix_federation_agent import MatrixFederationAgent
+from synapse.logging.context import make_deferred_yieldable
 from synapse.util.async_helpers import timeout_deferred
-from synapse.util.logcontext import make_deferred_yieldable
 from synapse.util.metrics import Measure
 
 logger = logging.getLogger(__name__)
diff --git a/synapse/http/request_metrics.py b/synapse/http/request_metrics.py
index 62045a918b..46af27c8f6 100644
--- a/synapse/http/request_metrics.py
+++ b/synapse/http/request_metrics.py
@@ -19,8 +19,8 @@ import threading
 
 from prometheus_client.core import Counter, Histogram
 
+from synapse.logging.context import LoggingContext
 from synapse.metrics import LaterGauge
-from synapse.util.logcontext import LoggingContext
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/http/server.py b/synapse/http/server.py
index f067c163c1..72a3d67eb6 100644
--- a/synapse/http/server.py
+++ b/synapse/http/server.py
@@ -39,8 +39,8 @@ from synapse.api.errors import (
     SynapseError,
     UnrecognizedRequestError,
 )
+from synapse.logging.context import preserve_fn
 from synapse.util.caches import intern_dict
-from synapse.util.logcontext import preserve_fn
 
 logger = logging.getLogger(__name__)
 
@@ -65,8 +65,8 @@ def wrap_json_request_handler(h):
     The handler method must have a signature of "handle_foo(self, request)",
     where "request" must be a SynapseRequest.
 
-    The handler must return a deferred. If the deferred succeeds we assume that
-    a response has been sent. If the deferred fails with a SynapseError we use
+    The handler must return a deferred or a coroutine. If the deferred succeeds
+    we assume that a response has been sent. If the deferred fails with a SynapseError we use
     it to send a JSON response with the appropriate HTTP reponse code. If the
     deferred fails with any other type of error we send a 500 reponse.
     """
@@ -353,16 +353,22 @@ class DirectServeResource(resource.Resource):
         """
         Render the request, using an asynchronous render handler if it exists.
         """
-        render_callback_name = "_async_render_" + request.method.decode("ascii")
+        async_render_callback_name = "_async_render_" + request.method.decode("ascii")
 
-        if hasattr(self, render_callback_name):
-            # Call the handler
-            callback = getattr(self, render_callback_name)
-            defer.ensureDeferred(callback(request))
+        # Try and get the async renderer
+        callback = getattr(self, async_render_callback_name, None)
 
-            return NOT_DONE_YET
-        else:
-            super().render(request)
+        # No async renderer for this request method.
+        if not callback:
+            return super().render(request)
+
+        resp = callback(request)
+
+        # If it's a coroutine, turn it into a Deferred
+        if isinstance(resp, types.CoroutineType):
+            defer.ensureDeferred(resp)
+
+        return NOT_DONE_YET
 
 
 def _options_handler(request):
diff --git a/synapse/http/site.py b/synapse/http/site.py
index 93f679ea48..df5274c177 100644
--- a/synapse/http/site.py
+++ b/synapse/http/site.py
@@ -19,7 +19,7 @@ from twisted.web.server import Request, Site
 
 from synapse.http import redact_uri
 from synapse.http.request_metrics import RequestMetrics, requests_counter
-from synapse.util.logcontext import LoggingContext, PreserveLoggingContext
+from synapse.logging.context import LoggingContext, PreserveLoggingContext
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/logging/__init__.py b/synapse/logging/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
--- /dev/null
+++ b/synapse/logging/__init__.py
diff --git a/synapse/logging/context.py b/synapse/logging/context.py
new file mode 100644
index 0000000000..30dfa1d6b2
--- /dev/null
+++ b/synapse/logging/context.py
@@ -0,0 +1,693 @@
+# Copyright 2014-2016 OpenMarket Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+""" Thread-local-alike tracking of log contexts within synapse
+
+This module provides objects and utilities for tracking contexts through
+synapse code, so that log lines can include a request identifier, and so that
+CPU and database activity can be accounted for against the request that caused
+them.
+
+See doc/log_contexts.rst for details on how this works.
+"""
+
+import logging
+import threading
+import types
+
+from twisted.internet import defer, threads
+
+logger = logging.getLogger(__name__)
+
+try:
+    import resource
+
+    # Python doesn't ship with a definition of RUSAGE_THREAD but it's defined
+    # to be 1 on linux so we hard code it.
+    RUSAGE_THREAD = 1
+
+    # If the system doesn't support RUSAGE_THREAD then this should throw an
+    # exception.
+    resource.getrusage(RUSAGE_THREAD)
+
+    def get_thread_resource_usage():
+        return resource.getrusage(RUSAGE_THREAD)
+
+
+except Exception:
+    # If the system doesn't support resource.getrusage(RUSAGE_THREAD) then we
+    # won't track resource usage by returning None.
+    def get_thread_resource_usage():
+        return None
+
+
+# get an id for the current thread.
+#
+# threading.get_ident doesn't actually return an OS-level tid, and annoyingly,
+# on Linux it actually returns the same value either side of a fork() call. However
+# we only fork in one place, so it's not worth the hoop-jumping to get a real tid.
+#
+get_thread_id = threading.get_ident
+
+
+class ContextResourceUsage(object):
+    """Object for tracking the resources used by a log context
+
+    Attributes:
+        ru_utime (float): user CPU time (in seconds)
+        ru_stime (float): system CPU time (in seconds)
+        db_txn_count (int): number of database transactions done
+        db_sched_duration_sec (float): amount of time spent waiting for a
+            database connection
+        db_txn_duration_sec (float): amount of time spent doing database
+            transactions (excluding scheduling time)
+        evt_db_fetch_count (int): number of events requested from the database
+    """
+
+    __slots__ = [
+        "ru_stime",
+        "ru_utime",
+        "db_txn_count",
+        "db_txn_duration_sec",
+        "db_sched_duration_sec",
+        "evt_db_fetch_count",
+    ]
+
+    def __init__(self, copy_from=None):
+        """Create a new ContextResourceUsage
+
+        Args:
+            copy_from (ContextResourceUsage|None): if not None, an object to
+                copy stats from
+        """
+        if copy_from is None:
+            self.reset()
+        else:
+            self.ru_utime = copy_from.ru_utime
+            self.ru_stime = copy_from.ru_stime
+            self.db_txn_count = copy_from.db_txn_count
+
+            self.db_txn_duration_sec = copy_from.db_txn_duration_sec
+            self.db_sched_duration_sec = copy_from.db_sched_duration_sec
+            self.evt_db_fetch_count = copy_from.evt_db_fetch_count
+
+    def copy(self):
+        return ContextResourceUsage(copy_from=self)
+
+    def reset(self):
+        self.ru_stime = 0.0
+        self.ru_utime = 0.0
+        self.db_txn_count = 0
+
+        self.db_txn_duration_sec = 0
+        self.db_sched_duration_sec = 0
+        self.evt_db_fetch_count = 0
+
+    def __repr__(self):
+        return (
+            "<ContextResourceUsage ru_stime='%r', ru_utime='%r', "
+            "db_txn_count='%r', db_txn_duration_sec='%r', "
+            "db_sched_duration_sec='%r', evt_db_fetch_count='%r'>"
+        ) % (
+            self.ru_stime,
+            self.ru_utime,
+            self.db_txn_count,
+            self.db_txn_duration_sec,
+            self.db_sched_duration_sec,
+            self.evt_db_fetch_count,
+        )
+
+    def __iadd__(self, other):
+        """Add another ContextResourceUsage's stats to this one's.
+
+        Args:
+            other (ContextResourceUsage): the other resource usage object
+        """
+        self.ru_utime += other.ru_utime
+        self.ru_stime += other.ru_stime
+        self.db_txn_count += other.db_txn_count
+        self.db_txn_duration_sec += other.db_txn_duration_sec
+        self.db_sched_duration_sec += other.db_sched_duration_sec
+        self.evt_db_fetch_count += other.evt_db_fetch_count
+        return self
+
+    def __isub__(self, other):
+        self.ru_utime -= other.ru_utime
+        self.ru_stime -= other.ru_stime
+        self.db_txn_count -= other.db_txn_count
+        self.db_txn_duration_sec -= other.db_txn_duration_sec
+        self.db_sched_duration_sec -= other.db_sched_duration_sec
+        self.evt_db_fetch_count -= other.evt_db_fetch_count
+        return self
+
+    def __add__(self, other):
+        res = ContextResourceUsage(copy_from=self)
+        res += other
+        return res
+
+    def __sub__(self, other):
+        res = ContextResourceUsage(copy_from=self)
+        res -= other
+        return res
+
+
+class LoggingContext(object):
+    """Additional context for log formatting. Contexts are scoped within a
+    "with" block.
+
+    If a parent is given when creating a new context, then:
+        - logging fields are copied from the parent to the new context on entry
+        - when the new context exits, the cpu usage stats are copied from the
+          child to the parent
+
+    Args:
+        name (str): Name for the context for debugging.
+        parent_context (LoggingContext|None): The parent of the new context
+    """
+
+    __slots__ = [
+        "previous_context",
+        "name",
+        "parent_context",
+        "_resource_usage",
+        "usage_start",
+        "main_thread",
+        "alive",
+        "request",
+        "tag",
+    ]
+
+    thread_local = threading.local()
+
+    class Sentinel(object):
+        """Sentinel to represent the root context"""
+
+        __slots__ = []
+
+        def __str__(self):
+            return "sentinel"
+
+        def copy_to(self, record):
+            pass
+
+        def start(self):
+            pass
+
+        def stop(self):
+            pass
+
+        def add_database_transaction(self, duration_sec):
+            pass
+
+        def add_database_scheduled(self, sched_sec):
+            pass
+
+        def record_event_fetch(self, event_count):
+            pass
+
+        def __nonzero__(self):
+            return False
+
+        __bool__ = __nonzero__  # python3
+
+    sentinel = Sentinel()
+
+    def __init__(self, name=None, parent_context=None, request=None):
+        self.previous_context = LoggingContext.current_context()
+        self.name = name
+
+        # track the resources used by this context so far
+        self._resource_usage = ContextResourceUsage()
+
+        # If alive has the thread resource usage when the logcontext last
+        # became active.
+        self.usage_start = None
+
+        self.main_thread = get_thread_id()
+        self.request = None
+        self.tag = ""
+        self.alive = True
+
+        self.parent_context = parent_context
+
+        if self.parent_context is not None:
+            self.parent_context.copy_to(self)
+
+        if request is not None:
+            # the request param overrides the request from the parent context
+            self.request = request
+
+    def __str__(self):
+        if self.request:
+            return str(self.request)
+        return "%s@%x" % (self.name, id(self))
+
+    @classmethod
+    def current_context(cls):
+        """Get the current logging context from thread local storage
+
+        Returns:
+            LoggingContext: the current logging context
+        """
+        return getattr(cls.thread_local, "current_context", cls.sentinel)
+
+    @classmethod
+    def set_current_context(cls, context):
+        """Set the current logging context in thread local storage
+        Args:
+            context(LoggingContext): The context to activate.
+        Returns:
+            The context that was previously active
+        """
+        current = cls.current_context()
+
+        if current is not context:
+            current.stop()
+            cls.thread_local.current_context = context
+            context.start()
+        return current
+
+    def __enter__(self):
+        """Enters this logging context into thread local storage"""
+        old_context = self.set_current_context(self)
+        if self.previous_context != old_context:
+            logger.warn(
+                "Expected previous context %r, found %r",
+                self.previous_context,
+                old_context,
+            )
+        self.alive = True
+
+        return self
+
+    def __exit__(self, type, value, traceback):
+        """Restore the logging context in thread local storage to the state it
+        was before this context was entered.
+        Returns:
+            None to avoid suppressing any exceptions that were thrown.
+        """
+        current = self.set_current_context(self.previous_context)
+        if current is not self:
+            if current is self.sentinel:
+                logger.warning("Expected logging context %s was lost", self)
+            else:
+                logger.warning(
+                    "Expected logging context %s but found %s", self, current
+                )
+        self.previous_context = None
+        self.alive = False
+
+        # if we have a parent, pass our CPU usage stats on
+        if self.parent_context is not None and hasattr(
+            self.parent_context, "_resource_usage"
+        ):
+            self.parent_context._resource_usage += self._resource_usage
+
+            # reset them in case we get entered again
+            self._resource_usage.reset()
+
+    def copy_to(self, record):
+        """Copy logging fields from this context to a log record or
+        another LoggingContext
+        """
+
+        # 'request' is the only field we currently use in the logger, so that's
+        # all we need to copy
+        record.request = self.request
+
+    def start(self):
+        if get_thread_id() != self.main_thread:
+            logger.warning("Started logcontext %s on different thread", self)
+            return
+
+        # If we haven't already started record the thread resource usage so
+        # far
+        if not self.usage_start:
+            self.usage_start = get_thread_resource_usage()
+
+    def stop(self):
+        if get_thread_id() != self.main_thread:
+            logger.warning("Stopped logcontext %s on different thread", self)
+            return
+
+        # When we stop, let's record the cpu used since we started
+        if not self.usage_start:
+            logger.warning("Called stop on logcontext %s without calling start", self)
+            return
+
+        utime_delta, stime_delta = self._get_cputime()
+        self._resource_usage.ru_utime += utime_delta
+        self._resource_usage.ru_stime += stime_delta
+
+        self.usage_start = None
+
+    def get_resource_usage(self):
+        """Get resources used by this logcontext so far.
+
+        Returns:
+            ContextResourceUsage: a *copy* of the object tracking resource
+                usage so far
+        """
+        # we always return a copy, for consistency
+        res = self._resource_usage.copy()
+
+        # If we are on the correct thread and we're currently running then we
+        # can include resource usage so far.
+        is_main_thread = get_thread_id() == self.main_thread
+        if self.alive and self.usage_start and is_main_thread:
+            utime_delta, stime_delta = self._get_cputime()
+            res.ru_utime += utime_delta
+            res.ru_stime += stime_delta
+
+        return res
+
+    def _get_cputime(self):
+        """Get the cpu usage time so far
+
+        Returns: Tuple[float, float]: seconds in user mode, seconds in system mode
+        """
+        current = get_thread_resource_usage()
+
+        utime_delta = current.ru_utime - self.usage_start.ru_utime
+        stime_delta = current.ru_stime - self.usage_start.ru_stime
+
+        # sanity check
+        if utime_delta < 0:
+            logger.error(
+                "utime went backwards! %f < %f",
+                current.ru_utime,
+                self.usage_start.ru_utime,
+            )
+            utime_delta = 0
+
+        if stime_delta < 0:
+            logger.error(
+                "stime went backwards! %f < %f",
+                current.ru_stime,
+                self.usage_start.ru_stime,
+            )
+            stime_delta = 0
+
+        return utime_delta, stime_delta
+
+    def add_database_transaction(self, duration_sec):
+        if duration_sec < 0:
+            raise ValueError("DB txn time can only be non-negative")
+        self._resource_usage.db_txn_count += 1
+        self._resource_usage.db_txn_duration_sec += duration_sec
+
+    def add_database_scheduled(self, sched_sec):
+        """Record a use of the database pool
+
+        Args:
+            sched_sec (float): number of seconds it took us to get a
+                connection
+        """
+        if sched_sec < 0:
+            raise ValueError("DB scheduling time can only be non-negative")
+        self._resource_usage.db_sched_duration_sec += sched_sec
+
+    def record_event_fetch(self, event_count):
+        """Record a number of events being fetched from the db
+
+        Args:
+            event_count (int): number of events being fetched
+        """
+        self._resource_usage.evt_db_fetch_count += event_count
+
+
+class LoggingContextFilter(logging.Filter):
+    """Logging filter that adds values from the current logging context to each
+    record.
+    Args:
+        **defaults: Default values to avoid formatters complaining about
+            missing fields
+    """
+
+    def __init__(self, **defaults):
+        self.defaults = defaults
+
+    def filter(self, record):
+        """Add each fields from the logging contexts to the record.
+        Returns:
+            True to include the record in the log output.
+        """
+        context = LoggingContext.current_context()
+        for key, value in self.defaults.items():
+            setattr(record, key, value)
+
+        # context should never be None, but if it somehow ends up being, then
+        # we end up in a death spiral of infinite loops, so let's check, for
+        # robustness' sake.
+        if context is not None:
+            context.copy_to(record)
+
+        return True
+
+
+class PreserveLoggingContext(object):
+    """Captures the current logging context and restores it when the scope is
+    exited. Used to restore the context after a function using
+    @defer.inlineCallbacks is resumed by a callback from the reactor."""
+
+    __slots__ = ["current_context", "new_context", "has_parent"]
+
+    def __init__(self, new_context=None):
+        if new_context is None:
+            new_context = LoggingContext.sentinel
+        self.new_context = new_context
+
+    def __enter__(self):
+        """Captures the current logging context"""
+        self.current_context = LoggingContext.set_current_context(self.new_context)
+
+        if self.current_context:
+            self.has_parent = self.current_context.previous_context is not None
+            if not self.current_context.alive:
+                logger.debug("Entering dead context: %s", self.current_context)
+
+    def __exit__(self, type, value, traceback):
+        """Restores the current logging context"""
+        context = LoggingContext.set_current_context(self.current_context)
+
+        if context != self.new_context:
+            if context is LoggingContext.sentinel:
+                logger.warning("Expected logging context %s was lost", self.new_context)
+            else:
+                logger.warning(
+                    "Expected logging context %s but found %s",
+                    self.new_context,
+                    context,
+                )
+
+        if self.current_context is not LoggingContext.sentinel:
+            if not self.current_context.alive:
+                logger.debug("Restoring dead context: %s", self.current_context)
+
+
+def nested_logging_context(suffix, parent_context=None):
+    """Creates a new logging context as a child of another.
+
+    The nested logging context will have a 'request' made up of the parent context's
+    request, plus the given suffix.
+
+    CPU/db usage stats will be added to the parent context's on exit.
+
+    Normal usage looks like:
+
+        with nested_logging_context(suffix):
+            # ... do stuff
+
+    Args:
+        suffix (str): suffix to add to the parent context's 'request'.
+        parent_context (LoggingContext|None): parent context. Will use the current context
+            if None.
+
+    Returns:
+        LoggingContext: new logging context.
+    """
+    if parent_context is None:
+        parent_context = LoggingContext.current_context()
+    return LoggingContext(
+        parent_context=parent_context, request=parent_context.request + "-" + suffix
+    )
+
+
+def preserve_fn(f):
+    """Function decorator which wraps the function with run_in_background"""
+
+    def g(*args, **kwargs):
+        return run_in_background(f, *args, **kwargs)
+
+    return g
+
+
+def run_in_background(f, *args, **kwargs):
+    """Calls a function, ensuring that the current context is restored after
+    return from the function, and that the sentinel context is set once the
+    deferred returned by the function completes.
+
+    Useful for wrapping functions that return a deferred or coroutine, which you don't
+    yield or await on (for instance because you want to pass it to
+    deferred.gatherResults()).
+
+    Note that if you completely discard the result, you should make sure that
+    `f` doesn't raise any deferred exceptions, otherwise a scary-looking
+    CRITICAL error about an unhandled error will be logged without much
+    indication about where it came from.
+    """
+    current = LoggingContext.current_context()
+    try:
+        res = f(*args, **kwargs)
+    except:  # noqa: E722
+        # the assumption here is that the caller doesn't want to be disturbed
+        # by synchronous exceptions, so let's turn them into Failures.
+        return defer.fail()
+
+    if isinstance(res, types.CoroutineType):
+        res = defer.ensureDeferred(res)
+
+    if not isinstance(res, defer.Deferred):
+        return res
+
+    if res.called and not res.paused:
+        # The function should have maintained the logcontext, so we can
+        # optimise out the messing about
+        return res
+
+    # The function may have reset the context before returning, so
+    # we need to restore it now.
+    ctx = LoggingContext.set_current_context(current)
+
+    # The original context will be restored when the deferred
+    # completes, but there is nothing waiting for it, so it will
+    # get leaked into the reactor or some other function which
+    # wasn't expecting it. We therefore need to reset the context
+    # here.
+    #
+    # (If this feels asymmetric, consider it this way: we are
+    # effectively forking a new thread of execution. We are
+    # probably currently within a ``with LoggingContext()`` block,
+    # which is supposed to have a single entry and exit point. But
+    # by spawning off another deferred, we are effectively
+    # adding a new exit point.)
+    res.addBoth(_set_context_cb, ctx)
+    return res
+
+
+def make_deferred_yieldable(deferred):
+    """Given a deferred, make it follow the Synapse logcontext rules:
+
+    If the deferred has completed (or is not actually a Deferred), essentially
+    does nothing (just returns another completed deferred with the
+    result/failure).
+
+    If the deferred has not yet completed, resets the logcontext before
+    returning a deferred. Then, when the deferred completes, restores the
+    current logcontext before running callbacks/errbacks.
+
+    (This is more-or-less the opposite operation to run_in_background.)
+    """
+    if not isinstance(deferred, defer.Deferred):
+        return deferred
+
+    if deferred.called and not deferred.paused:
+        # it looks like this deferred is ready to run any callbacks we give it
+        # immediately. We may as well optimise out the logcontext faffery.
+        return deferred
+
+    # ok, we can't be sure that a yield won't block, so let's reset the
+    # logcontext, and add a callback to the deferred to restore it.
+    prev_context = LoggingContext.set_current_context(LoggingContext.sentinel)
+    deferred.addBoth(_set_context_cb, prev_context)
+    return deferred
+
+
+def _set_context_cb(result, context):
+    """A callback function which just sets the logging context"""
+    LoggingContext.set_current_context(context)
+    return result
+
+
+def defer_to_thread(reactor, f, *args, **kwargs):
+    """
+    Calls the function `f` using a thread from the reactor's default threadpool and
+    returns the result as a Deferred.
+
+    Creates a new logcontext for `f`, which is created as a child of the current
+    logcontext (so its CPU usage metrics will get attributed to the current
+    logcontext). `f` should preserve the logcontext it is given.
+
+    The result deferred follows the Synapse logcontext rules: you should `yield`
+    on it.
+
+    Args:
+        reactor (twisted.internet.base.ReactorBase): The reactor in whose main thread
+            the Deferred will be invoked, and whose threadpool we should use for the
+            function.
+
+            Normally this will be hs.get_reactor().
+
+        f (callable): The function to call.
+
+        args: positional arguments to pass to f.
+
+        kwargs: keyword arguments to pass to f.
+
+    Returns:
+        Deferred: A Deferred which fires a callback with the result of `f`, or an
+            errback if `f` throws an exception.
+    """
+    return defer_to_threadpool(reactor, reactor.getThreadPool(), f, *args, **kwargs)
+
+
+def defer_to_threadpool(reactor, threadpool, f, *args, **kwargs):
+    """
+    A wrapper for twisted.internet.threads.deferToThreadpool, which handles
+    logcontexts correctly.
+
+    Calls the function `f` using a thread from the given threadpool and returns
+    the result as a Deferred.
+
+    Creates a new logcontext for `f`, which is created as a child of the current
+    logcontext (so its CPU usage metrics will get attributed to the current
+    logcontext). `f` should preserve the logcontext it is given.
+
+    The result deferred follows the Synapse logcontext rules: you should `yield`
+    on it.
+
+    Args:
+        reactor (twisted.internet.base.ReactorBase): The reactor in whose main thread
+            the Deferred will be invoked. Normally this will be hs.get_reactor().
+
+        threadpool (twisted.python.threadpool.ThreadPool): The threadpool to use for
+            running `f`. Normally this will be hs.get_reactor().getThreadPool().
+
+        f (callable): The function to call.
+
+        args: positional arguments to pass to f.
+
+        kwargs: keyword arguments to pass to f.
+
+    Returns:
+        Deferred: A Deferred which fires a callback with the result of `f`, or an
+            errback if `f` throws an exception.
+    """
+    logcontext = LoggingContext.current_context()
+
+    def g():
+        with LoggingContext(parent_context=logcontext):
+            return f(*args, **kwargs)
+
+    return make_deferred_yieldable(threads.deferToThreadPool(reactor, threadpool, g))
diff --git a/synapse/logging/formatter.py b/synapse/logging/formatter.py
new file mode 100644
index 0000000000..fbf570c756
--- /dev/null
+++ b/synapse/logging/formatter.py
@@ -0,0 +1,53 @@
+# -*- coding: utf-8 -*-
+# Copyright 2017 New Vector Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import logging
+import traceback
+
+from six import StringIO
+
+
+class LogFormatter(logging.Formatter):
+    """Log formatter which gives more detail for exceptions
+
+    This is the same as the standard log formatter, except that when logging
+    exceptions [typically via log.foo("msg", exc_info=1)], it prints the
+    sequence that led up to the point at which the exception was caught.
+    (Normally only stack frames between the point the exception was raised and
+    where it was caught are logged).
+    """
+
+    def __init__(self, *args, **kwargs):
+        super(LogFormatter, self).__init__(*args, **kwargs)
+
+    def formatException(self, ei):
+        sio = StringIO()
+        (typ, val, tb) = ei
+
+        # log the stack above the exception capture point if possible, but
+        # check that we actually have an f_back attribute to work around
+        # https://twistedmatrix.com/trac/ticket/9305
+
+        if tb and hasattr(tb.tb_frame, "f_back"):
+            sio.write("Capture point (most recent call last):\n")
+            traceback.print_stack(tb.tb_frame.f_back, None, sio)
+
+        traceback.print_exception(typ, val, tb, None, sio)
+        s = sio.getvalue()
+        sio.close()
+        if s[-1:] == "\n":
+            s = s[:-1]
+        return s
diff --git a/synapse/util/logutils.py b/synapse/logging/utils.py
index 7df0fa6087..7df0fa6087 100644
--- a/synapse/util/logutils.py
+++ b/synapse/logging/utils.py
diff --git a/synapse/metrics/background_process_metrics.py b/synapse/metrics/background_process_metrics.py
index 167e2c068a..edd6b42db3 100644
--- a/synapse/metrics/background_process_metrics.py
+++ b/synapse/metrics/background_process_metrics.py
@@ -22,7 +22,7 @@ from prometheus_client.core import REGISTRY, Counter, GaugeMetricFamily
 
 from twisted.internet import defer
 
-from synapse.util.logcontext import LoggingContext, PreserveLoggingContext
+from synapse.logging.context import LoggingContext, PreserveLoggingContext
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/notifier.py b/synapse/notifier.py
index d398078eed..918ef64897 100644
--- a/synapse/notifier.py
+++ b/synapse/notifier.py
@@ -23,12 +23,12 @@ from twisted.internet import defer
 from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import AuthError
 from synapse.handlers.presence import format_user_presence_state
+from synapse.logging.context import PreserveLoggingContext
+from synapse.logging.utils import log_function
 from synapse.metrics import LaterGauge
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import StreamToken
 from synapse.util.async_helpers import ObservableDeferred, timeout_deferred
-from synapse.util.logcontext import PreserveLoggingContext
-from synapse.util.logutils import log_function
 from synapse.util.metrics import Measure
 from synapse.visibility import filter_events_for_client
 
diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py
index 809199fe88..521c6e2cd7 100644
--- a/synapse/push/mailer.py
+++ b/synapse/push/mailer.py
@@ -29,6 +29,7 @@ from twisted.internet import defer
 
 from synapse.api.constants import EventTypes
 from synapse.api.errors import StoreError
+from synapse.logging.context import make_deferred_yieldable
 from synapse.push.presentable_names import (
     calculate_room_name,
     descriptor_from_member_events,
@@ -36,7 +37,6 @@ from synapse.push.presentable_names import (
 )
 from synapse.types import UserID
 from synapse.util.async_helpers import concurrently_execute
-from synapse.util.logcontext import make_deferred_yieldable
 from synapse.visibility import filter_events_for_client
 
 logger = logging.getLogger(__name__)
diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py
index 97efb835ad..5ffdf2675d 100644
--- a/synapse/replication/tcp/protocol.py
+++ b/synapse/replication/tcp/protocol.py
@@ -62,9 +62,9 @@ from twisted.internet import defer
 from twisted.protocols.basic import LineOnlyReceiver
 from twisted.python.failure import Failure
 
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.metrics import LaterGauge
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.stringutils import random_string
 
 from .commands import (
diff --git a/synapse/rest/client/transactions.py b/synapse/rest/client/transactions.py
index 36404b797d..6da71dc46f 100644
--- a/synapse/rest/client/transactions.py
+++ b/synapse/rest/client/transactions.py
@@ -17,8 +17,8 @@
 to ensure idempotency when performing PUTs using the REST API."""
 import logging
 
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.util.async_helpers import ObservableDeferred
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py
index ede6bc8b1e..f961178235 100644
--- a/synapse/rest/client/v1/login.py
+++ b/synapse/rest/client/v1/login.py
@@ -86,6 +86,7 @@ class LoginRestServlet(RestServlet):
         self.jwt_enabled = hs.config.jwt_enabled
         self.jwt_secret = hs.config.jwt_secret
         self.jwt_algorithm = hs.config.jwt_algorithm
+        self.saml2_enabled = hs.config.saml2_enabled
         self.cas_enabled = hs.config.cas_enabled
         self.auth_handler = self.hs.get_auth_handler()
         self.registration_handler = hs.get_registration_handler()
@@ -97,6 +98,9 @@ class LoginRestServlet(RestServlet):
         flows = []
         if self.jwt_enabled:
             flows.append({"type": LoginRestServlet.JWT_TYPE})
+        if self.saml2_enabled:
+            flows.append({"type": LoginRestServlet.SSO_TYPE})
+            flows.append({"type": LoginRestServlet.TOKEN_TYPE})
         if self.cas_enabled:
             flows.append({"type": LoginRestServlet.SSO_TYPE})
 
@@ -319,12 +323,12 @@ class LoginRestServlet(RestServlet):
             raise LoginError(401, "Invalid JWT", errcode=Codes.UNAUTHORIZED)
 
         user_id = UserID(user, self.hs.hostname).to_string()
+        device_id = login_submission.get("device_id")
+        initial_display_name = login_submission.get("initial_device_display_name")
 
         auth_handler = self.auth_handler
         registered_user_id = yield auth_handler.check_user_exists(user_id)
         if registered_user_id:
-            device_id = login_submission.get("device_id")
-            initial_display_name = login_submission.get("initial_device_display_name")
             device_id, access_token = yield self.registration_handler.register_device(
                 registered_user_id, device_id, initial_display_name
             )
@@ -338,11 +342,8 @@ class LoginRestServlet(RestServlet):
             user_id, access_token = (
                 yield self.registration_handler.register(localpart=user)
             )
-
-            device_id = login_submission.get("device_id")
-            initial_display_name = login_submission.get("initial_device_display_name")
             device_id, access_token = yield self.registration_handler.register_device(
-                registered_user_id, device_id, initial_display_name
+                user_id, device_id, initial_display_name
             )
 
             result = {
@@ -354,27 +355,49 @@ class LoginRestServlet(RestServlet):
         defer.returnValue(result)
 
 
-class CasRedirectServlet(RestServlet):
+class BaseSSORedirectServlet(RestServlet):
+    """Common base class for /login/sso/redirect impls"""
+
     PATTERNS = client_patterns("/login/(cas|sso)/redirect", v1=True)
 
+    def on_GET(self, request):
+        args = request.args
+        if b"redirectUrl" not in args:
+            return 400, "Redirect URL not specified for SSO auth"
+        client_redirect_url = args[b"redirectUrl"][0]
+        sso_url = self.get_sso_url(client_redirect_url)
+        request.redirect(sso_url)
+        finish_request(request)
+
+    def get_sso_url(self, client_redirect_url):
+        """Get the URL to redirect to, to perform SSO auth
+
+        Args:
+            client_redirect_url (bytes): the URL that we should redirect the
+                client to when everything is done
+
+        Returns:
+            bytes: URL to redirect to
+        """
+        # to be implemented by subclasses
+        raise NotImplementedError()
+
+
+class CasRedirectServlet(BaseSSORedirectServlet):
     def __init__(self, hs):
         super(CasRedirectServlet, self).__init__()
         self.cas_server_url = hs.config.cas_server_url.encode("ascii")
         self.cas_service_url = hs.config.cas_service_url.encode("ascii")
 
-    def on_GET(self, request):
-        args = request.args
-        if b"redirectUrl" not in args:
-            return (400, "Redirect URL not specified for CAS auth")
+    def get_sso_url(self, client_redirect_url):
         client_redirect_url_param = urllib.parse.urlencode(
-            {b"redirectUrl": args[b"redirectUrl"][0]}
+            {b"redirectUrl": client_redirect_url}
         ).encode("ascii")
         hs_redirect_url = self.cas_service_url + b"/_matrix/client/r0/login/cas/ticket"
         service_param = urllib.parse.urlencode(
             {b"service": b"%s?%s" % (hs_redirect_url, client_redirect_url_param)}
         ).encode("ascii")
-        request.redirect(b"%s/login?%s" % (self.cas_server_url, service_param))
-        finish_request(request)
+        return b"%s/login?%s" % (self.cas_server_url, service_param)
 
 
 class CasTicketServlet(RestServlet):
@@ -457,6 +480,16 @@ class CasTicketServlet(RestServlet):
         return user, attributes
 
 
+class SAMLRedirectServlet(BaseSSORedirectServlet):
+    PATTERNS = client_patterns("/login/sso/redirect", v1=True)
+
+    def __init__(self, hs):
+        self._saml_handler = hs.get_saml_handler()
+
+    def get_sso_url(self, client_redirect_url):
+        return self._saml_handler.handle_redirect_request(client_redirect_url)
+
+
 class SSOAuthHandler(object):
     """
     Utility class for Resources and Servlets which handle the response from a SSO
@@ -532,3 +565,5 @@ def register_servlets(hs, http_server):
     if hs.config.cas_enabled:
         CasRedirectServlet(hs).register(http_server)
         CasTicketServlet(hs).register(http_server)
+    elif hs.config.saml2_enabled:
+        SAMLRedirectServlet(hs).register(http_server)
diff --git a/synapse/rest/media/v1/_base.py b/synapse/rest/media/v1/_base.py
index 3318638d3e..5fefee4dde 100644
--- a/synapse/rest/media/v1/_base.py
+++ b/synapse/rest/media/v1/_base.py
@@ -25,7 +25,7 @@ from twisted.protocols.basic import FileSender
 
 from synapse.api.errors import Codes, SynapseError, cs_error
 from synapse.http.server import finish_request, respond_with_json
-from synapse.util import logcontext
+from synapse.logging.context import make_deferred_yieldable
 from synapse.util.stringutils import is_ascii
 
 logger = logging.getLogger(__name__)
@@ -75,9 +75,7 @@ def respond_with_file(request, media_type, file_path, file_size=None, upload_nam
         add_file_headers(request, media_type, file_size, upload_name)
 
         with open(file_path, "rb") as f:
-            yield logcontext.make_deferred_yieldable(
-                FileSender().beginFileTransfer(f, request)
-            )
+            yield make_deferred_yieldable(FileSender().beginFileTransfer(f, request))
 
         finish_request(request)
     else:
diff --git a/synapse/rest/media/v1/media_repository.py b/synapse/rest/media/v1/media_repository.py
index df3d985a38..65afffbb42 100644
--- a/synapse/rest/media/v1/media_repository.py
+++ b/synapse/rest/media/v1/media_repository.py
@@ -33,8 +33,8 @@ from synapse.api.errors import (
     RequestSendFailed,
     SynapseError,
 )
+from synapse.logging.context import defer_to_thread
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util import logcontext
 from synapse.util.async_helpers import Linearizer
 from synapse.util.retryutils import NotRetryingDestination
 from synapse.util.stringutils import random_string
@@ -463,7 +463,7 @@ class MediaRepository(object):
         )
 
         thumbnailer = Thumbnailer(input_path)
-        t_byte_source = yield logcontext.defer_to_thread(
+        t_byte_source = yield defer_to_thread(
             self.hs.get_reactor(),
             self._generate_thumbnail,
             thumbnailer,
@@ -511,7 +511,7 @@ class MediaRepository(object):
         )
 
         thumbnailer = Thumbnailer(input_path)
-        t_byte_source = yield logcontext.defer_to_thread(
+        t_byte_source = yield defer_to_thread(
             self.hs.get_reactor(),
             self._generate_thumbnail,
             thumbnailer,
@@ -596,7 +596,7 @@ class MediaRepository(object):
             return
 
         if thumbnailer.transpose_method is not None:
-            m_width, m_height = yield logcontext.defer_to_thread(
+            m_width, m_height = yield defer_to_thread(
                 self.hs.get_reactor(), thumbnailer.transpose
             )
 
@@ -616,11 +616,11 @@ class MediaRepository(object):
         for (t_width, t_height, t_type), t_method in iteritems(thumbnails):
             # Generate the thumbnail
             if t_method == "crop":
-                t_byte_source = yield logcontext.defer_to_thread(
+                t_byte_source = yield defer_to_thread(
                     self.hs.get_reactor(), thumbnailer.crop, t_width, t_height, t_type
                 )
             elif t_method == "scale":
-                t_byte_source = yield logcontext.defer_to_thread(
+                t_byte_source = yield defer_to_thread(
                     self.hs.get_reactor(), thumbnailer.scale, t_width, t_height, t_type
                 )
             else:
diff --git a/synapse/rest/media/v1/media_storage.py b/synapse/rest/media/v1/media_storage.py
index eff86836fb..25e5ac2848 100644
--- a/synapse/rest/media/v1/media_storage.py
+++ b/synapse/rest/media/v1/media_storage.py
@@ -24,9 +24,8 @@ import six
 from twisted.internet import defer
 from twisted.protocols.basic import FileSender
 
-from synapse.util import logcontext
+from synapse.logging.context import defer_to_thread, make_deferred_yieldable
 from synapse.util.file_consumer import BackgroundFileConsumer
-from synapse.util.logcontext import make_deferred_yieldable
 
 from ._base import Responder
 
@@ -65,7 +64,7 @@ class MediaStorage(object):
 
         with self.store_into_file(file_info) as (f, fname, finish_cb):
             # Write to the main repository
-            yield logcontext.defer_to_thread(
+            yield defer_to_thread(
                 self.hs.get_reactor(), _write_file_synchronously, source, f
             )
             yield finish_cb()
diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py
index 0337b64dc2..5871737bfd 100644
--- a/synapse/rest/media/v1/preview_url_resource.py
+++ b/synapse/rest/media/v1/preview_url_resource.py
@@ -42,11 +42,11 @@ from synapse.http.server import (
     wrap_json_request_handler,
 )
 from synapse.http.servlet import parse_integer, parse_string
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.rest.media.v1._base import get_filename_from_headers
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches.expiringcache import ExpiringCache
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.stringutils import random_string
 
 from ._base import FileInfo
@@ -95,6 +95,7 @@ class PreviewUrlResource(DirectServeResource):
         )
 
     def render_OPTIONS(self, request):
+        request.setHeader(b"Allow", b"OPTIONS, GET")
         return respond_with_json(request, 200, {}, send_cors=True)
 
     @wrap_json_request_handler
diff --git a/synapse/rest/media/v1/storage_provider.py b/synapse/rest/media/v1/storage_provider.py
index 359b45ebfc..e8f559acc1 100644
--- a/synapse/rest/media/v1/storage_provider.py
+++ b/synapse/rest/media/v1/storage_provider.py
@@ -20,8 +20,7 @@ import shutil
 from twisted.internet import defer
 
 from synapse.config._base import Config
-from synapse.util import logcontext
-from synapse.util.logcontext import run_in_background
+from synapse.logging.context import defer_to_thread, run_in_background
 
 from .media_storage import FileResponder
 
@@ -125,7 +124,7 @@ class FileStorageProviderBackend(StorageProvider):
         if not os.path.exists(dirname):
             os.makedirs(dirname)
 
-        return logcontext.defer_to_thread(
+        return defer_to_thread(
             self.hs.get_reactor(), shutil.copyfile, primary_fname, backup_fname
         )
 
diff --git a/synapse/rest/saml2/response_resource.py b/synapse/rest/saml2/response_resource.py
index 939c87306c..69ecc5e4b4 100644
--- a/synapse/rest/saml2/response_resource.py
+++ b/synapse/rest/saml2/response_resource.py
@@ -13,17 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-import logging
 
-import saml2
-from saml2.client import Saml2Client
-
-from synapse.api.errors import CodeMessageException
 from synapse.http.server import DirectServeResource, wrap_html_request_handler
-from synapse.http.servlet import parse_string
-from synapse.rest.client.v1.login import SSOAuthHandler
-
-logger = logging.getLogger(__name__)
 
 
 class SAML2ResponseResource(DirectServeResource):
@@ -33,32 +24,8 @@ class SAML2ResponseResource(DirectServeResource):
 
     def __init__(self, hs):
         super().__init__()
-
-        self._saml_client = Saml2Client(hs.config.saml2_sp_config)
-        self._sso_auth_handler = SSOAuthHandler(hs)
+        self._saml_handler = hs.get_saml_handler()
 
     @wrap_html_request_handler
     async def _async_render_POST(self, request):
-        resp_bytes = parse_string(request, "SAMLResponse", required=True)
-        relay_state = parse_string(request, "RelayState", required=True)
-
-        try:
-            saml2_auth = self._saml_client.parse_authn_request_response(
-                resp_bytes, saml2.BINDING_HTTP_POST
-            )
-        except Exception as e:
-            logger.warning("Exception parsing SAML2 response", exc_info=1)
-            raise CodeMessageException(400, "Unable to parse SAML2 response: %s" % (e,))
-
-        if saml2_auth.not_signed:
-            raise CodeMessageException(400, "SAML2 response was not signed")
-
-        if "uid" not in saml2_auth.ava:
-            raise CodeMessageException(400, "uid not in SAML2 response")
-
-        username = saml2_auth.ava["uid"][0]
-
-        displayName = saml2_auth.ava.get("displayName", [None])[0]
-        return self._sso_auth_handler.on_successful_auth(
-            username, request, relay_state, user_display_name=displayName
-        )
+        return await self._saml_handler.handle_saml_response(request)
diff --git a/synapse/server.py b/synapse/server.py
index a9592c396c..9e28dba2b1 100644
--- a/synapse/server.py
+++ b/synapse/server.py
@@ -194,6 +194,7 @@ class HomeServer(object):
         "sendmail",
         "registration_handler",
         "account_validity_handler",
+        "saml_handler",
         "event_client_serializer",
     ]
 
@@ -524,6 +525,11 @@ class HomeServer(object):
     def build_account_validity_handler(self):
         return AccountValidityHandler(self)
 
+    def build_saml_handler(self):
+        from synapse.handlers.saml_handler import SamlHandler
+
+        return SamlHandler(self)
+
     def build_event_client_serializer(self):
         return EventClientSerializer(self)
 
diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py
index 1b454a56a1..9f708fa205 100644
--- a/synapse/state/__init__.py
+++ b/synapse/state/__init__.py
@@ -28,11 +28,11 @@ from twisted.internet import defer
 from synapse.api.constants import EventTypes
 from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, StateResolutionVersions
 from synapse.events.snapshot import EventContext
+from synapse.logging.utils import log_function
 from synapse.state import v1, v2
 from synapse.util.async_helpers import Linearizer
 from synapse.util.caches import get_cache_factor_for
 from synapse.util.caches.expiringcache import ExpiringCache
-from synapse.util.logutils import log_function
 from synapse.util.metrics import Measure
 
 logger = logging.getLogger(__name__)
diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py
index 29589853c6..2f940dbae6 100644
--- a/synapse/storage/_base.py
+++ b/synapse/storage/_base.py
@@ -30,12 +30,12 @@ from prometheus_client import Histogram
 from twisted.internet import defer
 
 from synapse.api.errors import StoreError
+from synapse.logging.context import LoggingContext, PreserveLoggingContext
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.storage.engines import PostgresEngine, Sqlite3Engine
 from synapse.types import get_domain_from_id
 from synapse.util import batch_iter
 from synapse.util.caches.descriptors import Cache
-from synapse.util.logcontext import LoggingContext, PreserveLoggingContext
 from synapse.util.stringutils import exception_to_unicode
 
 # import a function which will return a monotonic time, in seconds
diff --git a/synapse/storage/events.py b/synapse/storage/events.py
index fefba39ea1..b486ca50eb 100644
--- a/synapse/storage/events.py
+++ b/synapse/storage/events.py
@@ -33,6 +33,8 @@ from synapse.api.constants import EventTypes
 from synapse.api.errors import SynapseError
 from synapse.events import EventBase  # noqa: F401
 from synapse.events.snapshot import EventContext  # noqa: F401
+from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable
+from synapse.logging.utils import log_function
 from synapse.metrics import BucketCollector
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.state import StateResolutionStore
@@ -45,8 +47,6 @@ from synapse.util import batch_iter
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches.descriptors import cached, cachedInlineCallbacks
 from synapse.util.frozenutils import frozendict_json_encoder
-from synapse.util.logcontext import PreserveLoggingContext, make_deferred_yieldable
-from synapse.util.logutils import log_function
 from synapse.util.metrics import Measure
 
 logger = logging.getLogger(__name__)
@@ -253,7 +253,14 @@ class EventsStore(
         )
 
         # Read the extrems every 60 minutes
-        hs.get_clock().looping_call(self._read_forward_extremities, 60 * 60 * 1000)
+        def read_forward_extremities():
+            # run as a background process to make sure that the database transactions
+            # have a logcontext to report to
+            return run_as_background_process(
+                "read_forward_extremities", self._read_forward_extremities
+            )
+
+        hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000)
 
     @defer.inlineCallbacks
     def _read_forward_extremities(self):
diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py
index 6d680d405a..09db872511 100644
--- a/synapse/storage/events_worker.py
+++ b/synapse/storage/events_worker.py
@@ -29,14 +29,14 @@ from synapse.api.room_versions import EventFormatVersions
 from synapse.events import FrozenEvent, event_type_from_format_version  # noqa: F401
 from synapse.events.snapshot import EventContext  # noqa: F401
 from synapse.events.utils import prune_event
-from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.types import get_domain_from_id
-from synapse.util.logcontext import (
+from synapse.logging.context import (
     LoggingContext,
     PreserveLoggingContext,
     make_deferred_yieldable,
     run_in_background,
 )
+from synapse.metrics.background_process_metrics import run_as_background_process
+from synapse.types import get_domain_from_id
 from synapse.util.metrics import Measure
 
 from ._base import SQLBaseStore
diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py
index 983ce13291..13a3d5208b 100644
--- a/synapse/storage/registration.py
+++ b/synapse/storage/registration.py
@@ -25,6 +25,7 @@ from twisted.internet import defer
 
 from synapse.api.constants import UserTypes
 from synapse.api.errors import Codes, StoreError, ThreepidValidationError
+from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.storage import background_updates
 from synapse.storage._base import SQLBaseStore
 from synapse.types import UserID
@@ -619,9 +620,15 @@ class RegistrationStore(
         )
 
         # Create a background job for culling expired 3PID validity tokens
-        hs.get_clock().looping_call(
-            self.cull_expired_threepid_validation_tokens, THIRTY_MINUTES_IN_MS
-        )
+        def start_cull():
+            # run as a background process to make sure that the database transactions
+            # have a logcontext to report to
+            return run_as_background_process(
+                "cull_expired_threepid_validation_tokens",
+                self.cull_expired_threepid_validation_tokens,
+            )
+
+        hs.get_clock().looping_call(start_cull, THIRTY_MINUTES_IN_MS)
 
     @defer.inlineCallbacks
     def _backgroud_update_set_deactivated_flag(self, progress, batch_size):
diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py
index 7b5b3b8c8d..f8e3007d67 100644
--- a/synapse/storage/stream.py
+++ b/synapse/storage/stream.py
@@ -41,12 +41,12 @@ from six.moves import range
 
 from twisted.internet import defer
 
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.storage._base import SQLBaseStore
 from synapse.storage.engines import PostgresEngine
 from synapse.storage.events_worker import EventsWorkerStore
 from synapse.types import RoomStreamToken
 from synapse.util.caches.stream_change_cache import StreamChangeCache
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py
index b1188f6bcb..fd18619178 100644
--- a/synapse/storage/transactions.py
+++ b/synapse/storage/transactions.py
@@ -133,34 +133,6 @@ class TransactionStore(SQLBaseStore):
             desc="set_received_txn_response",
         )
 
-    def prep_send_transaction(self, transaction_id, destination, origin_server_ts):
-        """Persists an outgoing transaction and calculates the values for the
-        previous transaction id list.
-
-        This should be called before sending the transaction so that it has the
-        correct value for the `prev_ids` key.
-
-        Args:
-            transaction_id (str)
-            destination (str)
-            origin_server_ts (int)
-
-        Returns:
-            list: A list of previous transaction ids.
-        """
-        return defer.succeed([])
-
-    def delivered_txn(self, transaction_id, destination, code, response_dict):
-        """Persists the response for an outgoing transaction.
-
-        Args:
-            transaction_id (str)
-            destination (str)
-            code (int)
-            response_json (str)
-        """
-        pass
-
     @defer.inlineCallbacks
     def get_destination_retry_timings(self, destination):
         """Gets the current retry timings (if any) for a given destination.
diff --git a/synapse/util/__init__.py b/synapse/util/__init__.py
index dcc747cac1..f506b2a695 100644
--- a/synapse/util/__init__.py
+++ b/synapse/util/__init__.py
@@ -21,7 +21,7 @@ import attr
 
 from twisted.internet import defer, task
 
-from synapse.util.logcontext import PreserveLoggingContext
+from synapse.logging import context
 
 logger = logging.getLogger(__name__)
 
@@ -46,7 +46,7 @@ class Clock(object):
     @defer.inlineCallbacks
     def sleep(self, seconds):
         d = defer.Deferred()
-        with PreserveLoggingContext():
+        with context.PreserveLoggingContext():
             self._reactor.callLater(seconds, d.callback, seconds)
             res = yield d
         defer.returnValue(res)
@@ -62,7 +62,10 @@ class Clock(object):
     def looping_call(self, f, msec):
         """Call a function repeatedly.
 
-         Waits `msec` initially before calling `f` for the first time.
+        Waits `msec` initially before calling `f` for the first time.
+
+        Note that the function will be called with no logcontext, so if it is anything
+        other than trivial, you probably want to wrap it in run_as_background_process.
 
         Args:
             f(function): The function to call repeatedly.
@@ -77,6 +80,9 @@ class Clock(object):
     def call_later(self, delay, callback, *args, **kwargs):
         """Call something later
 
+        Note that the function will be called with no logcontext, so if it is anything
+        other than trivial, you probably want to wrap it in run_as_background_process.
+
         Args:
             delay(float): How long to wait in seconds.
             callback(function): Function to call
@@ -85,10 +91,10 @@ class Clock(object):
         """
 
         def wrapped_callback(*args, **kwargs):
-            with PreserveLoggingContext():
+            with context.PreserveLoggingContext():
                 callback(*args, **kwargs)
 
-        with PreserveLoggingContext():
+        with context.PreserveLoggingContext():
             return self._reactor.callLater(delay, wrapped_callback, *args, **kwargs)
 
     def cancel_call_later(self, timer, ignore_errs=False):
diff --git a/synapse/util/async_helpers.py b/synapse/util/async_helpers.py
index 7757b8708a..58a6b8764f 100644
--- a/synapse/util/async_helpers.py
+++ b/synapse/util/async_helpers.py
@@ -23,13 +23,12 @@ from twisted.internet import defer
 from twisted.internet.defer import CancelledError
 from twisted.python import failure
 
-from synapse.util import Clock, logcontext, unwrapFirstError
-
-from .logcontext import (
+from synapse.logging.context import (
     PreserveLoggingContext,
     make_deferred_yieldable,
     run_in_background,
 )
+from synapse.util import Clock, unwrapFirstError
 
 logger = logging.getLogger(__name__)
 
@@ -153,7 +152,7 @@ def concurrently_execute(func, args, limit):
         except StopIteration:
             pass
 
-    return logcontext.make_deferred_yieldable(
+    return make_deferred_yieldable(
         defer.gatherResults(
             [run_in_background(_concurrently_execute_inner) for _ in range(limit)],
             consumeErrors=True,
@@ -174,7 +173,7 @@ def yieldable_gather_results(func, iter, *args, **kwargs):
         Deferred[list]: Resolved when all functions have been invoked, or errors if
         one of the function calls fails.
     """
-    return logcontext.make_deferred_yieldable(
+    return make_deferred_yieldable(
         defer.gatherResults(
             [run_in_background(func, item, *args, **kwargs) for item in iter],
             consumeErrors=True,
diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py
index d2f25063aa..675db2f448 100644
--- a/synapse/util/caches/descriptors.py
+++ b/synapse/util/caches/descriptors.py
@@ -24,7 +24,8 @@ from six import itervalues, string_types
 
 from twisted.internet import defer
 
-from synapse.util import logcontext, unwrapFirstError
+from synapse.logging.context import make_deferred_yieldable, preserve_fn
+from synapse.util import unwrapFirstError
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches import get_cache_factor_for
 from synapse.util.caches.lrucache import LruCache
@@ -388,7 +389,7 @@ class CacheDescriptor(_CacheDescriptorBase):
 
             except KeyError:
                 ret = defer.maybeDeferred(
-                    logcontext.preserve_fn(self.function_to_call), obj, *args, **kwargs
+                    preserve_fn(self.function_to_call), obj, *args, **kwargs
                 )
 
                 def onErr(f):
@@ -408,7 +409,7 @@ class CacheDescriptor(_CacheDescriptorBase):
                 observer = result_d.observe()
 
             if isinstance(observer, defer.Deferred):
-                return logcontext.make_deferred_yieldable(observer)
+                return make_deferred_yieldable(observer)
             else:
                 return observer
 
@@ -563,7 +564,7 @@ class CacheListDescriptor(_CacheDescriptorBase):
 
                 cached_defers.append(
                     defer.maybeDeferred(
-                        logcontext.preserve_fn(self.function_to_call), **args_to_call
+                        preserve_fn(self.function_to_call), **args_to_call
                     ).addCallbacks(complete_all, errback)
                 )
 
@@ -571,7 +572,7 @@ class CacheListDescriptor(_CacheDescriptorBase):
                 d = defer.gatherResults(cached_defers, consumeErrors=True).addCallbacks(
                     lambda _: results, unwrapFirstError
                 )
-                return logcontext.make_deferred_yieldable(d)
+                return make_deferred_yieldable(d)
             else:
                 return results
 
diff --git a/synapse/util/caches/response_cache.py b/synapse/util/caches/response_cache.py
index cbe54d45dd..d6908e169d 100644
--- a/synapse/util/caches/response_cache.py
+++ b/synapse/util/caches/response_cache.py
@@ -16,9 +16,9 @@ import logging
 
 from twisted.internet import defer
 
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches import register_cache
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 
 logger = logging.getLogger(__name__)
 
@@ -78,7 +78,7 @@ class ResponseCache(object):
 
         *deferred* should run its callbacks in the sentinel logcontext (ie,
         you should wrap normal synapse deferreds with
-        logcontext.run_in_background).
+        synapse.logging.context.run_in_background).
 
         Can return either a new Deferred (which also doesn't follow the synapse
         logcontext rules), or, if *deferred* was already complete, the actual
diff --git a/synapse/util/distributor.py b/synapse/util/distributor.py
index 5a79db821c..45af8d3eeb 100644
--- a/synapse/util/distributor.py
+++ b/synapse/util/distributor.py
@@ -17,8 +17,8 @@ import logging
 
 from twisted.internet import defer
 
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/util/file_consumer.py b/synapse/util/file_consumer.py
index 629ed44149..8b17d1c8b8 100644
--- a/synapse/util/file_consumer.py
+++ b/synapse/util/file_consumer.py
@@ -17,7 +17,7 @@ from six.moves import queue
 
 from twisted.internet import threads
 
-from synapse.util.logcontext import make_deferred_yieldable, run_in_background
+from synapse.logging.context import make_deferred_yieldable, run_in_background
 
 
 class BackgroundFileConsumer(object):
diff --git a/synapse/util/logcontext.py b/synapse/util/logcontext.py
index 6b0d2deea0..40e5c10a49 100644
--- a/synapse/util/logcontext.py
+++ b/synapse/util/logcontext.py
@@ -1,4 +1,4 @@
-# Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2019 The Matrix.org Foundation C.I.C.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,668 +12,28 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-""" Thread-local-alike tracking of log contexts within synapse
-
-This module provides objects and utilities for tracking contexts through
-synapse code, so that log lines can include a request identifier, and so that
-CPU and database activity can be accounted for against the request that caused
-them.
-
-See doc/log_contexts.rst for details on how this works.
+"""
+Backwards compatibility re-exports of ``synapse.logging.context`` functionality.
 """
 
-import logging
-import threading
-
-from twisted.internet import defer, threads
-
-logger = logging.getLogger(__name__)
-
-try:
-    import resource
-
-    # Python doesn't ship with a definition of RUSAGE_THREAD but it's defined
-    # to be 1 on linux so we hard code it.
-    RUSAGE_THREAD = 1
-
-    # If the system doesn't support RUSAGE_THREAD then this should throw an
-    # exception.
-    resource.getrusage(RUSAGE_THREAD)
-
-    def get_thread_resource_usage():
-        return resource.getrusage(RUSAGE_THREAD)
-
-
-except Exception:
-    # If the system doesn't support resource.getrusage(RUSAGE_THREAD) then we
-    # won't track resource usage by returning None.
-    def get_thread_resource_usage():
-        return None
-
-
-class ContextResourceUsage(object):
-    """Object for tracking the resources used by a log context
-
-    Attributes:
-        ru_utime (float): user CPU time (in seconds)
-        ru_stime (float): system CPU time (in seconds)
-        db_txn_count (int): number of database transactions done
-        db_sched_duration_sec (float): amount of time spent waiting for a
-            database connection
-        db_txn_duration_sec (float): amount of time spent doing database
-            transactions (excluding scheduling time)
-        evt_db_fetch_count (int): number of events requested from the database
-    """
-
-    __slots__ = [
-        "ru_stime",
-        "ru_utime",
-        "db_txn_count",
-        "db_txn_duration_sec",
-        "db_sched_duration_sec",
-        "evt_db_fetch_count",
-    ]
-
-    def __init__(self, copy_from=None):
-        """Create a new ContextResourceUsage
-
-        Args:
-            copy_from (ContextResourceUsage|None): if not None, an object to
-                copy stats from
-        """
-        if copy_from is None:
-            self.reset()
-        else:
-            self.ru_utime = copy_from.ru_utime
-            self.ru_stime = copy_from.ru_stime
-            self.db_txn_count = copy_from.db_txn_count
-
-            self.db_txn_duration_sec = copy_from.db_txn_duration_sec
-            self.db_sched_duration_sec = copy_from.db_sched_duration_sec
-            self.evt_db_fetch_count = copy_from.evt_db_fetch_count
-
-    def copy(self):
-        return ContextResourceUsage(copy_from=self)
-
-    def reset(self):
-        self.ru_stime = 0.0
-        self.ru_utime = 0.0
-        self.db_txn_count = 0
-
-        self.db_txn_duration_sec = 0
-        self.db_sched_duration_sec = 0
-        self.evt_db_fetch_count = 0
-
-    def __repr__(self):
-        return (
-            "<ContextResourceUsage ru_stime='%r', ru_utime='%r', "
-            "db_txn_count='%r', db_txn_duration_sec='%r', "
-            "db_sched_duration_sec='%r', evt_db_fetch_count='%r'>"
-        ) % (
-            self.ru_stime,
-            self.ru_utime,
-            self.db_txn_count,
-            self.db_txn_duration_sec,
-            self.db_sched_duration_sec,
-            self.evt_db_fetch_count,
-        )
-
-    def __iadd__(self, other):
-        """Add another ContextResourceUsage's stats to this one's.
-
-        Args:
-            other (ContextResourceUsage): the other resource usage object
-        """
-        self.ru_utime += other.ru_utime
-        self.ru_stime += other.ru_stime
-        self.db_txn_count += other.db_txn_count
-        self.db_txn_duration_sec += other.db_txn_duration_sec
-        self.db_sched_duration_sec += other.db_sched_duration_sec
-        self.evt_db_fetch_count += other.evt_db_fetch_count
-        return self
-
-    def __isub__(self, other):
-        self.ru_utime -= other.ru_utime
-        self.ru_stime -= other.ru_stime
-        self.db_txn_count -= other.db_txn_count
-        self.db_txn_duration_sec -= other.db_txn_duration_sec
-        self.db_sched_duration_sec -= other.db_sched_duration_sec
-        self.evt_db_fetch_count -= other.evt_db_fetch_count
-        return self
-
-    def __add__(self, other):
-        res = ContextResourceUsage(copy_from=self)
-        res += other
-        return res
-
-    def __sub__(self, other):
-        res = ContextResourceUsage(copy_from=self)
-        res -= other
-        return res
-
-
-class LoggingContext(object):
-    """Additional context for log formatting. Contexts are scoped within a
-    "with" block.
-
-    If a parent is given when creating a new context, then:
-        - logging fields are copied from the parent to the new context on entry
-        - when the new context exits, the cpu usage stats are copied from the
-          child to the parent
-
-    Args:
-        name (str): Name for the context for debugging.
-        parent_context (LoggingContext|None): The parent of the new context
-    """
-
-    __slots__ = [
-        "previous_context",
-        "name",
-        "parent_context",
-        "_resource_usage",
-        "usage_start",
-        "main_thread",
-        "alive",
-        "request",
-        "tag",
-    ]
-
-    thread_local = threading.local()
-
-    class Sentinel(object):
-        """Sentinel to represent the root context"""
-
-        __slots__ = []
-
-        def __str__(self):
-            return "sentinel"
-
-        def copy_to(self, record):
-            pass
-
-        def start(self):
-            pass
-
-        def stop(self):
-            pass
-
-        def add_database_transaction(self, duration_sec):
-            pass
-
-        def add_database_scheduled(self, sched_sec):
-            pass
-
-        def record_event_fetch(self, event_count):
-            pass
-
-        def __nonzero__(self):
-            return False
-
-        __bool__ = __nonzero__  # python3
-
-    sentinel = Sentinel()
-
-    def __init__(self, name=None, parent_context=None, request=None):
-        self.previous_context = LoggingContext.current_context()
-        self.name = name
-
-        # track the resources used by this context so far
-        self._resource_usage = ContextResourceUsage()
-
-        # If alive has the thread resource usage when the logcontext last
-        # became active.
-        self.usage_start = None
-
-        self.main_thread = threading.current_thread()
-        self.request = None
-        self.tag = ""
-        self.alive = True
-
-        self.parent_context = parent_context
-
-        if self.parent_context is not None:
-            self.parent_context.copy_to(self)
-
-        if request is not None:
-            # the request param overrides the request from the parent context
-            self.request = request
-
-    def __str__(self):
-        if self.request:
-            return str(self.request)
-        return "%s@%x" % (self.name, id(self))
-
-    @classmethod
-    def current_context(cls):
-        """Get the current logging context from thread local storage
-
-        Returns:
-            LoggingContext: the current logging context
-        """
-        return getattr(cls.thread_local, "current_context", cls.sentinel)
-
-    @classmethod
-    def set_current_context(cls, context):
-        """Set the current logging context in thread local storage
-        Args:
-            context(LoggingContext): The context to activate.
-        Returns:
-            The context that was previously active
-        """
-        current = cls.current_context()
-
-        if current is not context:
-            current.stop()
-            cls.thread_local.current_context = context
-            context.start()
-        return current
-
-    def __enter__(self):
-        """Enters this logging context into thread local storage"""
-        old_context = self.set_current_context(self)
-        if self.previous_context != old_context:
-            logger.warn(
-                "Expected previous context %r, found %r",
-                self.previous_context,
-                old_context,
-            )
-        self.alive = True
-
-        return self
-
-    def __exit__(self, type, value, traceback):
-        """Restore the logging context in thread local storage to the state it
-        was before this context was entered.
-        Returns:
-            None to avoid suppressing any exceptions that were thrown.
-        """
-        current = self.set_current_context(self.previous_context)
-        if current is not self:
-            if current is self.sentinel:
-                logger.warning("Expected logging context %s was lost", self)
-            else:
-                logger.warning(
-                    "Expected logging context %s but found %s", self, current
-                )
-        self.previous_context = None
-        self.alive = False
-
-        # if we have a parent, pass our CPU usage stats on
-        if self.parent_context is not None and hasattr(
-            self.parent_context, "_resource_usage"
-        ):
-            self.parent_context._resource_usage += self._resource_usage
-
-            # reset them in case we get entered again
-            self._resource_usage.reset()
-
-    def copy_to(self, record):
-        """Copy logging fields from this context to a log record or
-        another LoggingContext
-        """
-
-        # 'request' is the only field we currently use in the logger, so that's
-        # all we need to copy
-        record.request = self.request
-
-    def start(self):
-        if threading.current_thread() is not self.main_thread:
-            logger.warning("Started logcontext %s on different thread", self)
-            return
-
-        # If we haven't already started record the thread resource usage so
-        # far
-        if not self.usage_start:
-            self.usage_start = get_thread_resource_usage()
-
-    def stop(self):
-        if threading.current_thread() is not self.main_thread:
-            logger.warning("Stopped logcontext %s on different thread", self)
-            return
-
-        # When we stop, let's record the cpu used since we started
-        if not self.usage_start:
-            logger.warning("Called stop on logcontext %s without calling start", self)
-            return
-
-        utime_delta, stime_delta = self._get_cputime()
-        self._resource_usage.ru_utime += utime_delta
-        self._resource_usage.ru_stime += stime_delta
-
-        self.usage_start = None
-
-    def get_resource_usage(self):
-        """Get resources used by this logcontext so far.
-
-        Returns:
-            ContextResourceUsage: a *copy* of the object tracking resource
-                usage so far
-        """
-        # we always return a copy, for consistency
-        res = self._resource_usage.copy()
-
-        # If we are on the correct thread and we're currently running then we
-        # can include resource usage so far.
-        is_main_thread = threading.current_thread() is self.main_thread
-        if self.alive and self.usage_start and is_main_thread:
-            utime_delta, stime_delta = self._get_cputime()
-            res.ru_utime += utime_delta
-            res.ru_stime += stime_delta
-
-        return res
-
-    def _get_cputime(self):
-        """Get the cpu usage time so far
-
-        Returns: Tuple[float, float]: seconds in user mode, seconds in system mode
-        """
-        current = get_thread_resource_usage()
-
-        utime_delta = current.ru_utime - self.usage_start.ru_utime
-        stime_delta = current.ru_stime - self.usage_start.ru_stime
-
-        # sanity check
-        if utime_delta < 0:
-            logger.error(
-                "utime went backwards! %f < %f",
-                current.ru_utime,
-                self.usage_start.ru_utime,
-            )
-            utime_delta = 0
-
-        if stime_delta < 0:
-            logger.error(
-                "stime went backwards! %f < %f",
-                current.ru_stime,
-                self.usage_start.ru_stime,
-            )
-            stime_delta = 0
-
-        return utime_delta, stime_delta
-
-    def add_database_transaction(self, duration_sec):
-        if duration_sec < 0:
-            raise ValueError("DB txn time can only be non-negative")
-        self._resource_usage.db_txn_count += 1
-        self._resource_usage.db_txn_duration_sec += duration_sec
-
-    def add_database_scheduled(self, sched_sec):
-        """Record a use of the database pool
-
-        Args:
-            sched_sec (float): number of seconds it took us to get a
-                connection
-        """
-        if sched_sec < 0:
-            raise ValueError("DB scheduling time can only be non-negative")
-        self._resource_usage.db_sched_duration_sec += sched_sec
-
-    def record_event_fetch(self, event_count):
-        """Record a number of events being fetched from the db
-
-        Args:
-            event_count (int): number of events being fetched
-        """
-        self._resource_usage.evt_db_fetch_count += event_count
-
-
-class LoggingContextFilter(logging.Filter):
-    """Logging filter that adds values from the current logging context to each
-    record.
-    Args:
-        **defaults: Default values to avoid formatters complaining about
-            missing fields
-    """
-
-    def __init__(self, **defaults):
-        self.defaults = defaults
-
-    def filter(self, record):
-        """Add each fields from the logging contexts to the record.
-        Returns:
-            True to include the record in the log output.
-        """
-        context = LoggingContext.current_context()
-        for key, value in self.defaults.items():
-            setattr(record, key, value)
-
-        # context should never be None, but if it somehow ends up being, then
-        # we end up in a death spiral of infinite loops, so let's check, for
-        # robustness' sake.
-        if context is not None:
-            context.copy_to(record)
-
-        return True
-
-
-class PreserveLoggingContext(object):
-    """Captures the current logging context and restores it when the scope is
-    exited. Used to restore the context after a function using
-    @defer.inlineCallbacks is resumed by a callback from the reactor."""
-
-    __slots__ = ["current_context", "new_context", "has_parent"]
-
-    def __init__(self, new_context=None):
-        if new_context is None:
-            new_context = LoggingContext.sentinel
-        self.new_context = new_context
-
-    def __enter__(self):
-        """Captures the current logging context"""
-        self.current_context = LoggingContext.set_current_context(self.new_context)
-
-        if self.current_context:
-            self.has_parent = self.current_context.previous_context is not None
-            if not self.current_context.alive:
-                logger.debug("Entering dead context: %s", self.current_context)
-
-    def __exit__(self, type, value, traceback):
-        """Restores the current logging context"""
-        context = LoggingContext.set_current_context(self.current_context)
-
-        if context != self.new_context:
-            if context is LoggingContext.sentinel:
-                logger.warning("Expected logging context %s was lost", self.new_context)
-            else:
-                logger.warning(
-                    "Expected logging context %s but found %s",
-                    self.new_context,
-                    context,
-                )
-
-        if self.current_context is not LoggingContext.sentinel:
-            if not self.current_context.alive:
-                logger.debug("Restoring dead context: %s", self.current_context)
-
-
-def nested_logging_context(suffix, parent_context=None):
-    """Creates a new logging context as a child of another.
-
-    The nested logging context will have a 'request' made up of the parent context's
-    request, plus the given suffix.
-
-    CPU/db usage stats will be added to the parent context's on exit.
-
-    Normal usage looks like:
-
-        with nested_logging_context(suffix):
-            # ... do stuff
-
-    Args:
-        suffix (str): suffix to add to the parent context's 'request'.
-        parent_context (LoggingContext|None): parent context. Will use the current context
-            if None.
-
-    Returns:
-        LoggingContext: new logging context.
-    """
-    if parent_context is None:
-        parent_context = LoggingContext.current_context()
-    return LoggingContext(
-        parent_context=parent_context, request=parent_context.request + "-" + suffix
-    )
-
-
-def preserve_fn(f):
-    """Function decorator which wraps the function with run_in_background"""
-
-    def g(*args, **kwargs):
-        return run_in_background(f, *args, **kwargs)
-
-    return g
-
-
-def run_in_background(f, *args, **kwargs):
-    """Calls a function, ensuring that the current context is restored after
-    return from the function, and that the sentinel context is set once the
-    deferred returned by the function completes.
-
-    Useful for wrapping functions that return a deferred which you don't yield
-    on (for instance because you want to pass it to deferred.gatherResults()).
-
-    Note that if you completely discard the result, you should make sure that
-    `f` doesn't raise any deferred exceptions, otherwise a scary-looking
-    CRITICAL error about an unhandled error will be logged without much
-    indication about where it came from.
-    """
-    current = LoggingContext.current_context()
-    try:
-        res = f(*args, **kwargs)
-    except:  # noqa: E722
-        # the assumption here is that the caller doesn't want to be disturbed
-        # by synchronous exceptions, so let's turn them into Failures.
-        return defer.fail()
-
-    if not isinstance(res, defer.Deferred):
-        return res
-
-    if res.called and not res.paused:
-        # The function should have maintained the logcontext, so we can
-        # optimise out the messing about
-        return res
-
-    # The function may have reset the context before returning, so
-    # we need to restore it now.
-    ctx = LoggingContext.set_current_context(current)
-
-    # The original context will be restored when the deferred
-    # completes, but there is nothing waiting for it, so it will
-    # get leaked into the reactor or some other function which
-    # wasn't expecting it. We therefore need to reset the context
-    # here.
-    #
-    # (If this feels asymmetric, consider it this way: we are
-    # effectively forking a new thread of execution. We are
-    # probably currently within a ``with LoggingContext()`` block,
-    # which is supposed to have a single entry and exit point. But
-    # by spawning off another deferred, we are effectively
-    # adding a new exit point.)
-    res.addBoth(_set_context_cb, ctx)
-    return res
-
-
-def make_deferred_yieldable(deferred):
-    """Given a deferred, make it follow the Synapse logcontext rules:
-
-    If the deferred has completed (or is not actually a Deferred), essentially
-    does nothing (just returns another completed deferred with the
-    result/failure).
-
-    If the deferred has not yet completed, resets the logcontext before
-    returning a deferred. Then, when the deferred completes, restores the
-    current logcontext before running callbacks/errbacks.
-
-    (This is more-or-less the opposite operation to run_in_background.)
-    """
-    if not isinstance(deferred, defer.Deferred):
-        return deferred
-
-    if deferred.called and not deferred.paused:
-        # it looks like this deferred is ready to run any callbacks we give it
-        # immediately. We may as well optimise out the logcontext faffery.
-        return deferred
-
-    # ok, we can't be sure that a yield won't block, so let's reset the
-    # logcontext, and add a callback to the deferred to restore it.
-    prev_context = LoggingContext.set_current_context(LoggingContext.sentinel)
-    deferred.addBoth(_set_context_cb, prev_context)
-    return deferred
-
-
-def _set_context_cb(result, context):
-    """A callback function which just sets the logging context"""
-    LoggingContext.set_current_context(context)
-    return result
-
-
-def defer_to_thread(reactor, f, *args, **kwargs):
-    """
-    Calls the function `f` using a thread from the reactor's default threadpool and
-    returns the result as a Deferred.
-
-    Creates a new logcontext for `f`, which is created as a child of the current
-    logcontext (so its CPU usage metrics will get attributed to the current
-    logcontext). `f` should preserve the logcontext it is given.
-
-    The result deferred follows the Synapse logcontext rules: you should `yield`
-    on it.
-
-    Args:
-        reactor (twisted.internet.base.ReactorBase): The reactor in whose main thread
-            the Deferred will be invoked, and whose threadpool we should use for the
-            function.
-
-            Normally this will be hs.get_reactor().
-
-        f (callable): The function to call.
-
-        args: positional arguments to pass to f.
-
-        kwargs: keyword arguments to pass to f.
-
-    Returns:
-        Deferred: A Deferred which fires a callback with the result of `f`, or an
-            errback if `f` throws an exception.
-    """
-    return defer_to_threadpool(reactor, reactor.getThreadPool(), f, *args, **kwargs)
-
-
-def defer_to_threadpool(reactor, threadpool, f, *args, **kwargs):
-    """
-    A wrapper for twisted.internet.threads.deferToThreadpool, which handles
-    logcontexts correctly.
-
-    Calls the function `f` using a thread from the given threadpool and returns
-    the result as a Deferred.
-
-    Creates a new logcontext for `f`, which is created as a child of the current
-    logcontext (so its CPU usage metrics will get attributed to the current
-    logcontext). `f` should preserve the logcontext it is given.
-
-    The result deferred follows the Synapse logcontext rules: you should `yield`
-    on it.
-
-    Args:
-        reactor (twisted.internet.base.ReactorBase): The reactor in whose main thread
-            the Deferred will be invoked. Normally this will be hs.get_reactor().
-
-        threadpool (twisted.python.threadpool.ThreadPool): The threadpool to use for
-            running `f`. Normally this will be hs.get_reactor().getThreadPool().
-
-        f (callable): The function to call.
-
-        args: positional arguments to pass to f.
-
-        kwargs: keyword arguments to pass to f.
-
-    Returns:
-        Deferred: A Deferred which fires a callback with the result of `f`, or an
-            errback if `f` throws an exception.
-    """
-    logcontext = LoggingContext.current_context()
-
-    def g():
-        with LoggingContext(parent_context=logcontext):
-            return f(*args, **kwargs)
-
-    return make_deferred_yieldable(threads.deferToThreadPool(reactor, threadpool, g))
+from synapse.logging.context import (
+    LoggingContext,
+    LoggingContextFilter,
+    PreserveLoggingContext,
+    defer_to_thread,
+    make_deferred_yieldable,
+    nested_logging_context,
+    preserve_fn,
+    run_in_background,
+)
+
+__all__ = [
+    "defer_to_thread",
+    "LoggingContext",
+    "LoggingContextFilter",
+    "make_deferred_yieldable",
+    "nested_logging_context",
+    "preserve_fn",
+    "PreserveLoggingContext",
+    "run_in_background",
+]
diff --git a/synapse/util/logformatter.py b/synapse/util/logformatter.py
index fbf570c756..320e8f8174 100644
--- a/synapse/util/logformatter.py
+++ b/synapse/util/logformatter.py
@@ -1,5 +1,4 @@
-# -*- coding: utf-8 -*-
-# Copyright 2017 New Vector Ltd
+# Copyright 2019 The Matrix.org Foundation C.I.C.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,41 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+"""
+Backwards compatibility re-exports of ``synapse.logging.formatter`` functionality.
+"""
 
-import logging
-import traceback
+from synapse.logging.formatter import LogFormatter
 
-from six import StringIO
-
-
-class LogFormatter(logging.Formatter):
-    """Log formatter which gives more detail for exceptions
-
-    This is the same as the standard log formatter, except that when logging
-    exceptions [typically via log.foo("msg", exc_info=1)], it prints the
-    sequence that led up to the point at which the exception was caught.
-    (Normally only stack frames between the point the exception was raised and
-    where it was caught are logged).
-    """
-
-    def __init__(self, *args, **kwargs):
-        super(LogFormatter, self).__init__(*args, **kwargs)
-
-    def formatException(self, ei):
-        sio = StringIO()
-        (typ, val, tb) = ei
-
-        # log the stack above the exception capture point if possible, but
-        # check that we actually have an f_back attribute to work around
-        # https://twistedmatrix.com/trac/ticket/9305
-
-        if tb and hasattr(tb.tb_frame, "f_back"):
-            sio.write("Capture point (most recent call last):\n")
-            traceback.print_stack(tb.tb_frame.f_back, None, sio)
-
-        traceback.print_exception(typ, val, tb, None, sio)
-        s = sio.getvalue()
-        sio.close()
-        if s[-1:] == "\n":
-            s = s[:-1]
-        return s
+__all__ = ["LogFormatter"]
diff --git a/synapse/util/metrics.py b/synapse/util/metrics.py
index 01284d3cf8..c30b6de19c 100644
--- a/synapse/util/metrics.py
+++ b/synapse/util/metrics.py
@@ -20,8 +20,8 @@ from prometheus_client import Counter
 
 from twisted.internet import defer
 
+from synapse.logging.context import LoggingContext
 from synapse.metrics import InFlightGauge
-from synapse.util.logcontext import LoggingContext
 
 logger = logging.getLogger(__name__)
 
diff --git a/synapse/util/ratelimitutils.py b/synapse/util/ratelimitutils.py
index 06defa8199..5ca4521ce3 100644
--- a/synapse/util/ratelimitutils.py
+++ b/synapse/util/ratelimitutils.py
@@ -20,7 +20,7 @@ import logging
 from twisted.internet import defer
 
 from synapse.api.errors import LimitExceededError
-from synapse.util.logcontext import (
+from synapse.logging.context import (
     PreserveLoggingContext,
     make_deferred_yieldable,
     run_in_background,
@@ -36,9 +36,11 @@ class FederationRateLimiter(object):
             clock (Clock)
             config (FederationRateLimitConfig)
         """
-        self.clock = clock
-        self._config = config
-        self.ratelimiters = {}
+
+        def new_limiter():
+            return _PerHostRatelimiter(clock=clock, config=config)
+
+        self.ratelimiters = collections.defaultdict(new_limiter)
 
     def ratelimit(self, host):
         """Used to ratelimit an incoming request from given host
@@ -53,11 +55,9 @@ class FederationRateLimiter(object):
             host (str): Origin of incoming request.
 
         Returns:
-            _PerHostRatelimiter
+            context manager which returns a deferred.
         """
-        return self.ratelimiters.setdefault(
-            host, _PerHostRatelimiter(clock=self.clock, config=self._config)
-        ).ratelimit()
+        return self.ratelimiters[host].ratelimit()
 
 
 class _PerHostRatelimiter(object):
@@ -122,7 +122,7 @@ class _PerHostRatelimiter(object):
         self.request_times.append(time_now)
 
         def queue_request():
-            if len(self.current_processing) > self.concurrent_requests:
+            if len(self.current_processing) >= self.concurrent_requests:
                 queue_defer = defer.Deferred()
                 self.ready_request_queue[request_id] = queue_defer
                 logger.info(
diff --git a/synapse/util/retryutils.py b/synapse/util/retryutils.py
index 1a77456498..d8d0ceae51 100644
--- a/synapse/util/retryutils.py
+++ b/synapse/util/retryutils.py
@@ -17,7 +17,7 @@ import random
 
 from twisted.internet import defer
 
-import synapse.util.logcontext
+import synapse.logging.context
 from synapse.api.errors import CodeMessageException
 
 logger = logging.getLogger(__name__)
@@ -225,4 +225,4 @@ class RetryDestinationLimiter(object):
                 logger.exception("Failed to store destination_retry_timings")
 
         # we deliberately do this in the background.
-        synapse.util.logcontext.run_in_background(store_retry_timings)
+        synapse.logging.context.run_in_background(store_retry_timings)