epicsScanDouble accepts "inf" and "nan", so envGetDoubleConfigParam
returns success for them and the parsed value flows into getMaxPeriod's
only range gate, maxPeriod < maxSearchPeriodLowerLimit. That comparison
is false for both inf and NaN, so a non-finite period reaches
getNTimers(), whose static_cast<unsigned> of 1 + log(maxPeriod/...)/log2
is an undefined float->unsigned conversion: inf aborts the client in
malloc via a garbage nTimers allocation, and NaN drives the search-timer
wheel off a NaN period where every deadline comparison is false.
Gate the parsed value on finite() (from epicsMath.h) and positivity
alongside the parse status, and fall back to the default period. A
non-positive period is not a "low" period to be clamped up to the 60 s
lower limit; it takes the same invalid-input path as inf/NaN.
getMaxPeriod() is getNTimers()'s only caller, so this makes the
conversion finite and positive by construction.
A successful parse of EPICS_CA_CONN_TMO stored the value into connTMO
unchecked, so EPICS_CA_CONN_TMO=-5 or =0 left the connection-verify
watchdog deadline permanently in the past. The watchdog then fires
continuously, flooding stderr with "Virtual circuit unresponsive" and
pinning a CPU while updates still arrive. NaN and inf (which
epicsScanDouble also accepts) slip past any ordering comparison: a NaN
period leaves the watchdog deadline arithmetic undefined, and an
infinite period silently disables circuit verification.
Fold the validity check into the existing parse-failure branch, so an
unparsable, non-finite, or non-positive value takes the same path:
warn once and fall back to the 30 s CA_CONN_VERIFY_PERIOD default.
This mirrors the sibling EPICS_CA_MAX_ARRAY_BYTES handling a few lines
below, which already gates a parse failure and an out-of-range value
in one condition, and matches the finite() gate on
EPICS_CA_MAX_SEARCH_PERIOD in udpiiu.cpp.
The ca_repeater function has some failure cases which can cause it to
return. In such a situation, a reference to the current thread's
epicsThreadId is kept in the taskwd, causing an eventual use-after-free
when the taskwd tries to determine the task's status.
We fix this problem by removing the thread from the taskwd if
ca_repeater ever returns.
Added debug messages for client verification.
With -d the repeater shows client events, e.g.
$ caRepeater -d
CA Repeater: Attached and initialized
New client on port 60307
Verified 1 active clients
New client on port 52541
Verified 2 active clients
Client on port 60307 refused message
Deleted client on port 60307
Client on port 52541 refused message
Deleted client on port 52541
^C
Running "caRepeater -dd" also shows beacon forwarding and
individual client verification:
$ caRepeater -dd
CA Repeater: Attached and initialized
New client on port 49968
Client on port 49968 is alive
Verified 1 active clients
Sent to port 49968
Sent to port 49968
Sent to port 49968
New client on port 65377
Sent to port 49968
Client on port 49968 is alive
Client on port 65377 is alive
Verified 2 active clients
Sent to port 49968
Sent to port 65377
Sent to port 49968
Sent to port 65377
Client on port 49968 refused message
Deleted client on port 49968
Sent to port 65377
Sent to port 65377
Client on port 65377 refused message
Deleted client on port 65377
^C
Fix EditorConfig failure in modules/libcom/RTEMS/epicsNtp.c
Add dbDefs to Sphinx API index, move devLib to libcom
Rename internal CA enum tag to fix Doxygen complaint
Set Clang's minimum MacOS version for aarch64 CPUs
Most errors found with:
codespell -L cach,thst,odly,aslo,parm,parms,inpu,ges,prset,pevent,ptd,pring,valu,noo,noe,ned,inout,ro,siz,froms,nd,fo,singl,sart,multy,tthe,allong,ment,inate,nodel,tring,alse,ture,thi,wille,numer
Some more manually found (its -> it's)
c++20: Do not use apostrophe (e.g. can't) in unquoted #error message
* Allow CA clients to determine the server protocol version
Adds a call to the CA client API that allows a client to determine the server's
protocol minor version number. This is needed to allow the ca-nameserver to
report a server's protocol version correctly to a client.
* ca_host_minor_protocol return for disconnected channels
ca_host_minor_protocol now explicitly returns CA_UKN_MINOR_VERSION for a disconnected channel.
Import all function documentations, return code reference
into the header files, as doc comments,
in the hopes of moving the CAref.html manual to docs.epics-controls.org,
which will be done in a later PR.
Also added doc comments to some other low-hanging fruits,
for example when plain comments where near, or when it was obvious to do so.
Markdown can be generated from .pod and .dbd.pod files.
Markdown will be generated and installed by "make inc".
Co-authored-by: Timo Korhonen <timo.korhonen@ess.eu>
Co-authored-by: Andrew Johnson <anj@anl.gov>
Found here at ESS:
(all in one line)
EPICS_CA_ADDR_LIST=averylonghostname.mylabnetwork.technicalnetwork.example.com
EPICS_CA_AUTO_ADDR_LIST=NO
caget somePVnam
leads to something like this:
CA.Client.Exception...............................................
Warning: "Empty PV search address list"
Source File: ../udpiiu.cpp line 403
Current Time: Thu Jun 09 2022 10:10:47.804161447
Problem desription:
addAddrToChannelAccessAddressList() will collect what ever we specify in
EPICS_CA_ADDR_LIST for channel access.
That function will add IP-addresses to the search list.
hostnames are possible, but are ignored if longer than 32 bytes,
because buf is too short.
If a hostname can be resolved into an IP, that is fine,
if not that is "fine as well" (better say: silently ignored):
If, and only if, EPICS_CA_AUTO_ADDR_LIST=NO is given then the one
and only too long hostname will be ignored and lead to an
"Empty PV search address list".
If EPICS_CA_AUTO_ADDR_LIST=YES (or nothing) is specified, the search list
is not empty, and EPICS will search all broadcast addresses.
This will eventually lead into a timout.
Solution:
Increase the buf size in addAddrToChannelAccessAddressList() from 32 to 256
The maximum length for a hostname is 255:
https://www.ietf.org/rfc/rfc1034.txt
If we add one byte for the string terminating '\0', we need 256 bytes.
And yes, this patch neglets the fact that a user can specify HOSTNAME:PORT,
which may need another 6 bytes. To put it the other way around:
In this case the hostname length is limited to 250 bytes,
which is still long enough in practice.
libCa shouldn't cause an application to abort with an uncaught
exception on a machine that has timezone problems just because
we can't convert the current time into a string.
The other Perl scripts that use FindBin for their lib path could
have the same changes applied, but these two scripts are the ones
normally run by hand so might get soft-linked by an installer.
Without this change these scripts don't actually work if run from
a soft-link instead of directly.
Sang Woo Kim
zimoch