commit	d206b437e3fb79a89205a9fbe748b017d312ddfa	[log] [tgz]
author	Ed Tanous <ed@tanous.net>	Thu Aug 27 11:45:42 2020 -0700
committer	Ed Tanous <ed@tanous.net>	Thu Sep 03 18:55:17 2020 +0000
tree	87286ecb08d1306c1caca9130c628c29199cc932
parent	f97ddba7015a553f13639233f7e5d3fb0152d0b2 [diff]

Revert http::Request::socket() callback

Details on why this revert is needed are here.

https://lists.ozlabs.org/pipermail/openbmc/2020-August/022478.html

Appu and Ravi still have not commented.

It should be noted, this also causes a memory leak in http connection,
where connections refuse to be freed, because of a bad usage of
shared_from_this.

This code wasn't very well thought through, and needs rearchitected to
not break the unit testability of bmcweb, nor cause memory leaks.

https://github.com/openbmc/bmcweb/blob/218bd4746130aac22366968c8c9a34a929e45a3d/http/http_connection.h#L351

Is the memory leak in question.

Specifically, this reverts:
The /attachment download in LogServices.  This needs reimplemented
properly, but is an OEM property, so it shouldn't be a big deal to
revert, and shouldn't break our redfish compliance.

The IpAddress property in SessionService.  I have no idea why this was
injected, and it's functionally incorrect.  IpAddresses are not related
to a session, and IP addresses can change over the course of a session,
so this property is already broken as written.  I suspect the author
really wanted RedfishEvent type logging, but that was too complex, so
they half implemented this.

Redfish SSE properties.  This needs to be reimplemented similar to the
patchset here:
https://gerrit.openbmc-project.xyz/c/openbmc/bmcweb/+/13948
Where the ownership of the HTTP connection does not leave the http
framework.  As written, the SSE implementation causes ownership issues,
as there's no clear delineation of the ownership between HttpConnection
and the SSE framework.

Tested:
On current master, running this command:
wget -O- --no-http-keep-alive --no-check-certificate https://{bmc
hostname}:18080/redfish/v1
Which should download the service root, then immediately close and
destroy the connection, prints:

(2020-08-28 16:55:24) [DEBUG "routing.h":1258] Matched rule
'/redfish/v1/' 2 / 4
(2020-08-28 16:55:24) [DEBUG "http_response.h":130] calling completion
handler
(2020-08-28 16:55:24) [DEBUG "http_response.h":133] completion handler
was valid
(2020-08-28 16:55:24) [INFO "http_connection.h":429] Response: 0x1e1ee28
/redfish/v1 200 keepalive=0
(2020-08-28 16:55:24) [DEBUG "timer_queue.h":48] timer add inside:
0x1d3d1a8 7
(2020-08-28 16:55:24) [DEBUG "http_connection.h":751] 0x1e1ee28 timer
added: 0x1d3d1a8 7
(2020-08-28 16:55:24) [DEBUG "http_connection.h":655] 0x1e1ee28 doWrite
(2020-08-28 16:55:24) [DEBUG "http_connection.h":663] 0x1e1ee28
async_write 1555 bytes
(2020-08-28 16:55:24) [DEBUG "http_connection.h":697] 0x1e1ee28 timer
cancelled: 0x1d3d1a8 7
(2020-08-28 16:55:24) [DEBUG "http_connection.h":676] 0x1e1ee28 from
write(1)

Then stops.  Note, that the connection was not destroyed, and has
leaked.  Once this patchset is added, the connection closes and destroys
properly, and doesn't leak, so it prints the above, but also prints.

(2020-08-28 16:27:10) [DEBUG "http_connection.h":305] 0x1d15c90
Connection closed, total 1

Ran Redfish service validator.  Saw one unrelated failure due to UUID,
all other things pass.

Signed-off-by: Ed Tanous <ed@tanous.net>
Change-Id: I18686037bf58f20389d31facc0d77020274d38a1

6 files changed

tree: 87286ecb08d1306c1caca9130c628c29199cc932

README.md

OpenBMC webserver

This component attempts to be a "do everything" embedded webserver for openbmc.

Capabilities

At this time, the webserver implements a few interfaces:

Authentication middleware that supports cookie and token based authentication, as well as CSRF prevention backed by linux PAM authentication credentials.
An (incomplete) attempt at replicating phosphor-dbus-rest interfaces in C++. Right now, a few of the endpoint definitions work as expected, but there is still a lot of work to be done. The portions of the interface that are functional are designed to work correctly for phosphor-webui, but may not yet be complete.
Replication of the rest-dbus backend interfaces to allow bmc debug to logged in users.
An initial attempt at a read-only redfish interface. Currently the redfish interface targets ServiceRoot, SessionService, AccountService, Roles, and ManagersService. Some functionality here has been shimmed to make development possible. For example, there exists only a single user role.
SSL key generation at runtime. See the configuration section for details.
Static file hosting. Currently, static files are hosted from the fixed location at /usr/share/www. This is intended to allow loose coupling with yocto projects, and allow overriding static files at build time.
Dbus-monitor over websocket. A generic endpoint that allows UIs to open a websocket and register for notification of events to avoid polling in single page applications. (this interface may be modified in the future due to security concerns.

Configuration

BMCWeb is configured by setting -D flags that correspond to options in bmcweb/CMakeLists.txt and then compiling. For example, cmake -DBMCWEB_ENABLE_KVM=NO ... followed by make. The option names become C++ preprocessor symbols that control which code is compiled into the program.

When BMCWeb starts running, it reads persistent configuration data (such as UUID and session data) from a local file. If this is not usable, it generates a new configuration.

When BMCWeb SSL support is enabled and a usable certificate is not found, it will generate a self-sign a certificate before launching the server. The keys are generated by the secp384r1 algorithm. The certificate

is issued by C=US, O=OpenBMC, CN=testhost,
is valid for 10 years,
has a random serial number, and
is signed using the SHA-256 algorithm.