this patch makes ubsan (-fsanitize-undefined-trap-on-error) happy.
let's look at the following scenario:
1. netlink sends RTM_NEWADDR/RTM_DELADDR message
2. rops_handle_POLLIN_netlink() handles message attributes as RTA_*
3. rops_handle_POLLIN_netlink() finds RTA_PRIORITY attribute and
handles it as u32 value, but it's IFA_CACHEINFO attribute in reality
and payload holds another type
4. ubsan throws SIGTRAP
so, don't handle RTM_NEWADDR/RTM_DELADDR attributes as RTA_*
The attribute indicating that a wsi belongs to an SS object also must be
migrated when we split out the original transaction wsi into a new nwsi, if we're
not going to lose track of its affiliation.
Likewise if the affiliated SS object points to the original wsi, we have to
migrate his pointer when we migrate the wsi.
Taking care of this book-keeping is necessary to get correct behaviours at
close-time.
We basically hear about HUP out-of-band with respect to
pending rx... now we mark the wsi as "unusable" once we
see the HUP - the connection has ended.
This means we have to differentiate between the wsi
being unusable for new things now, like replying, and
what it has already sent still being servicible and
pending.
Improve how we report what happened with http_proxy... if
we can't recognize the response as http/1.0 or http/1.1
do a CCE with "http_proxy fail", or if we did get valid h1
response but no 200 response code, do a CCE with
"http_proxy -> xxx" where xxx is the decimal representation
of the response code, eg, "http_proxy -> 407"
This provides a build option LWS_WITH_CONMON that lets user code recover
detailed connection stats on client connections with the LCCSCF_CONMON
flag.
In addition to latencies for dns, socket connection, tls and first protocol
response where possible, it also provides the user code an unfiltered list
of DNS responses that the client received, and the peer it actually
succeded to connect to.
There are a few build options that are trying to keep and report
various statistics
- DETAILED_LATENCY
- SERVER_STATUS
- WITH_STATS
remove all those and establish a generic rplacement, lws_metrics.
lws_metrics makes its stats available via an lws_system ops function
pointer that the user code can set.
Openmetrics export is supported, for, eg, prometheus scraping.
For SMP case, it was desirable to have a netlink listener per pt so they
could deal with pt-level changes in the pt's local service thread. But
Linux restricts the process to just one netlink listener.
We worked around it by only listening on pt[0], this aligns us a bit more
with the reality and moves to a single routing table in the context.
There's still more to do for SMP case locking.
On OSX, POLLHUP is always |POLLIN, even if you did not wait
on POLLIN. This causes a loop because we don't want to
ack the POLLHUP until we cleared any pollin (but there is
no pending POLLIN...)
