optimizations and some fault tolerance #54
Conversation
Ends up causing problems with some of our macros. We null check device in the LOG_XXX macros, in one case we use `this` as the device pointer The compiler then complains about comparing this with nullptr, which is assumed to always be true (this assumption is fine, we can ignore the warning)
JJL772
force-pushed
the
enh_optimizations_and_fault_tolerance
branch
from
c9da01e to
c499d9f
Compare
Module should be nearly 1:1 with what we have internally now.
JJL772
force-pushed
the
enh_optimizations_and_fault_tolerance
branch
from
c499d9f to
0491dd5
Compare
configure/RELEASE
Outdated
| # RELEASE.$(EPICS_HOST_ARCH).$(T_A) | ||
| # | ||
| # This file is parsed by both GNUmake and an EPICS Perl script, | ||
| # so it may ONLY contain definititions of paths to other support |
There was a problem hiding this comment.
| # so it may ONLY contain definititions of paths to other support | |
| # so it may ONLY contain definitions of paths to other support |
It may be in our other RELEASE files but at least ek9k can spell it right 😁
configure/RELEASE
Outdated
| # or | ||
| # FOO = /Full/Path/To/Development/Version | ||
| # ========================================================== | ||
| ASYN=/reg/g/pcds/epics/R7.0.2-2.0/modules/asyn/$(ASYN_MODULE_VERSION) |
There was a problem hiding this comment.
| ASYN=/reg/g/pcds/epics/R7.0.2-2.0/modules/asyn/$(ASYN_MODULE_VERSION) | |
| ASYN=$(EPICS_MODULES)/asyn/$(ASYN_MODULE_VERSION) |
ek9000App/src/devEK9000.h
Outdated
| @@ -218,6 +220,10 @@ class devEK9000 : public drvModbusAsyn { | |||
| /* Buffer for status info */ | |||
| uint16_t m_status_buf[EK9000_STATUS_END - EK9000_STATUS_START + 1]; | |||
|
|
|||
| /* Cache of terminal layout */ | |||
| uint16_t m_terminals[0xFF]; | |||
There was a problem hiding this comment.
Magic number 0xff - yes, a well known one to be fair - should probably be a const/define somewhere?
|
|
||
| memset(m_terminals, 0, sizeof(m_terminals)); | ||
| for (int off = 0; off < 255; off += 125) { | ||
| if (off > 0 && m_terminals[off] == 0) |
There was a problem hiding this comment.
Trying to understand the logic of this loop...
So in iterations 2 and 3, the final element was written by the previous iteration (1 and 2) of the loop, and the final element should have something in it? What is that final element representing?
There was a problem hiding this comment.
Input register address 0x6000 to 0x60FF contain a numerical ID for each terminal. 0x6000 will always contain 9000 for ek9000. If the next terminal is EL4004, 0x6001 contains 4004 (in decimal), and so on. The rest of the registers are set to 0, so if the last element read is 0, there are no more terminals on the rail and we can stop reading.
There was a problem hiding this comment.
Gotcha. Seems worthwhile to toss in a comment to that effect 👍
iocBoot/iocEK9KTest/st.tpl
Outdated
| @@ -11,6 +11,7 @@ ek9000Test_registerRecordDeviceDriver pdbbase | |||
|
|
|||
| # Configure the device (EK9K_NAME, IP, PORT, TERMINAL_COUNT) | |||
| ek9000Configure("${EK9K}", "${IP}", ${PORT}, ${NUM_TERMS}) | |||
| asynSetTraceMask("${EK9K}", 0, 0x21) | |||
There was a problem hiding this comment.
What's 0x21 again? Probably good to mark it in a comment as ASYN_TRACE_WARNING (0x20) |ASYN_TRACE_ERROR (0x01)
JJL772
force-pushed
the
enh_optimizations_and_fault_tolerance
branch
from
dce6720 to
aeeba6f
Compare
|
Post-review tweaks LGTM - merge at will |
Closes #53
Tested on
bhc-tst-ek9000