snmp mib2nut wrapper macros #2285
Conversation
…ULL fields for snmp_info_t and info_lkp_t [DMF] Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
… propose snmp_info_default(...) instead of fixed C structure items Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
… propose info_lkp_default(...) instead of fixed C structure items Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…e short-hand snmp_info_sentinel and info_lkp_sentinel Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…s2l() and info_lkp_fun_s2l() wrappers Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…t al wrapping Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
jimklimov
added
SNMP
refactor/fightwarn
…acros Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
| /* no-op without DMF extensions */ | ||
| # define info_lkp_fun_s2l(_1, _2, _3) {_1, _2, NULL, NULL} | ||
| # define info_lkp_nuf_vp2s(_1, _2, _3) {_1, _2, NULL, NULL} | ||
| # endif /* WITH_DMFMIB */ |
There was a problem hiding this comment.
There is a mapping precedent in original eaton-marlin MIB, which uses the third pointer (so info_lkp_fun_s2l) in DMF branch, but the second pointer as the only one available in master. Not sure OTOH which one is correct; it could have been the case why the extra couple appeared in the first place.
Ideally, someone in the context should revise it more attentively, maybe test build of this branch against a device. @aquette - would you be up to this?
At the moment info_lkp_fun_s2l for non-DMF wrapper would ignore that function. Maybe they should re-use the first and second pointers, if the old (current master) codebase was so inclined?
There was a problem hiding this comment.
NOTE: Updated the macros to conflate the two s2l (string-to-long) and vp2s (void-pointer-to-string) macro names and so mapping table fields of the vanilla NUT, which became role-separated (forward and reverse fun/nuf conversions) in the DMF branch.
| static info_lkp_t marlin_device_count_info[] = { | ||
| { 1, "dummy", NULL, marlin_device_count_fun }, | ||
| { 0, NULL, NULL, NULL } | ||
| /* TOTHINK: DMF branch used info_lkp_fun_s2l()? */ | ||
| info_lkp_nuf_s2l(1, "dummy", marlin_device_count_fun), | ||
| info_lkp_sentinel | ||
| }; |
There was a problem hiding this comment.
Here is the part about function wrappers (second or third pointer) I mentioned in another comment. The DMF/FTY branch has it different than this code, e.g.
nut/drivers/eaton-pdu-marlin-mib.c
Lines 819 to 848 in c4970c9
There was a problem hiding this comment.
Note: the PR was later amended to at least use the pointer to function in non-DMF builds (conflate with the other ..._s2l() method, though).
…ib2nut-wrappers Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…t al wrapping Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…to snmp_info_default() macros Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…_info[] => eaton_ats16_nm2_ambient_drycontacts_info[] to match other table naming patterns Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
… NUT_UNUSED_VARIABLE() Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…ith info_lkp_fun_s2l() and info_lkp_nuf_vp2s() Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
This comment was marked as outdated.
This comment was marked as outdated.
…kp_sentinel Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
| info_lkp_default(11, "BYPASS"), /* maintenanceBypass */ | ||
| info_lkp_default(11, "OL"), /* essMode */ | ||
| /* FIXME: is "11" correct here or in the line above? */ |
There was a problem hiding this comment.
@aquette : the two 11 values here, also in original code before the change - which one is right and which is a typo?
Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
|
@arnaudquette-eaton : I have finally gone over the differences visually, most things fit well; I would however welcome a review for the few content-related questions raised above. |
Re-apply commit d3b6ea4 to modernized codebase of PR networkupstools#2285 and newly cover drivers/hpe-pdu-mib.c as well Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
jimklimov
force-pushed
the
issue-2275-snmp-mib2nut-wrappers
branch
from
270ea61 to
91f107b
Compare
|
CI fault is not relevant here (busy port issue during self-tests). |
|
Offloaded remaining questions to separate issues so they can be addressed later if needed: they are not strictly related to changes of this PR, just were found during the big look at lots of code for its review. |
Closes: networkupstools#2302 Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…s_info[] Found per https://www.circitor.fr/Mibs/Html/P/PDU2-MIB.php#SensorStateEnumeration Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…nticipated by this vendor MIB Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
…2_outlet_status_info[] sentinel) Closes: networkupstools#2304 Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
jimklimov
force-pushed
the
issue-2275-snmp-mib2nut-wrappers
branch
from
87e3394 to
5621018
Compare
…le [#2285] Signed-off-by: Jim Klimov <jimklimov+nut@gmail.com>
Idea raised in PR #2275 allowing to juggle several codebases (main and DMF/FTY until merged) with different C structure property counts and minimal difference in comparable sources. For this, macros
snmp_info_defaultandinfo_lkp_default(and a few for function pointers within) are provided.This PR also clearly ends the tables with an entry literally named
..._sentinelso it is role is well understood (and spelling of the entry fields is never botched), and it is better visible where someone forgot to add one. So a bonus for maintainability.NOTE: Not updating sub-driver versions (except for a few files with mapping content changes) since the tables after C pre-processor expanding the macros should be equivalent to codebase before this PR.