Attention is currently required from: neels, laforge.
iedemam has posted comments on this change. (
https://gerrit.osmocom.org/c/osmo-bsc/+/27081 )
Change subject: stats: new trackers for lchan life duration
......................................................................
Patch Set 14:
(8 comments)
Patchset:
PS12:
Thanks for the confidence boost. […]
Done
Patchset:
PS14:
Hi Neels,
The feedback was really helpful. I think the patch is looking more like what you have in
mind now.
I am fighting a new handover test failure. Apologies for the silly question but how can I
even see what is going on there? I see the expout and testsuite.log files but they
don't offer any detail other than pointing out which test failed. Currently, I only
know that test_max_ta.ho_vty is the failing sub-test. Obviously I'm missing something
basic in the Osmo dev env / work cycle.
Many thanks again,
-Michael
File include/osmocom/bsc/gsm_data.h:
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/5d9f6d10_ab744b10
PS13, Line 771: struct osmo_time_cc timekeeper;
here is your problem: the lchan is completely cleared
out every time an lchan is activated / unused. […]
Thanks! I'm understanding the
lifecycle better now.
File src/osmo-bsc/gsm_data.c:
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/259bec1c_e6b62b68
PS13, Line 354: !&lchan->activate.timekeeper
a) the address of a struct member is *always*
non-NULL. […]
Done
File src/osmo-bsc/lchan_fsm.c:
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/d202e612_9f1d93c4
PS13, Line 226: lchan->activate.timekeeper = (struct osmo_time_cc){
also you probably want to do this only once during
lchan allocation, i.e. […]
Done
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/65022e02_97ddbd57
PS13, Line 228: .gran_usec = 1*1000000,
if you want milliseconds, 1000 is your factor. […]
Changed to milliseconds and changed forget_sum_usec to 1 to discard remainders.
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/2dc3c41c_a7ab9d1b
PS13, Line 237: /* TMP HACK: this fixes "make check" handover tests... */
yeah because you re-init the lost state after
lchan_reset().
Done
https://gerrit.osmocom.org/c/osmo-bsc/+/27081/comment/4d0227f9_b8723593
PS13, Line 558: if (&lchan->activate.timekeeper) {
this condition is never false, as above. simply set
the the flag without condition.
The check has been changed. I think it's still
necessary as there isn't any sanity checking in osmo_time_cc_set_flag itself and not
all lchans will have a timer attached.
--
To view, visit
https://gerrit.osmocom.org/c/osmo-bsc/+/27081
To unsubscribe, or for help writing mail filters, visit
https://gerrit.osmocom.org/settings
Gerrit-Project: osmo-bsc
Gerrit-Branch: master
Gerrit-Change-Id: I1b0670c47cb5e0b7776eda89d1e71545ba0e3347
Gerrit-Change-Number: 27081
Gerrit-PatchSet: 14
Gerrit-Owner: iedemam <michael(a)kapsulate.com>
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: laforge <laforge(a)osmocom.org>
Gerrit-Reviewer: neels <nhofmeyr(a)sysmocom.de>
Gerrit-CC: pespin <pespin(a)sysmocom.de>
Gerrit-Attention: neels <nhofmeyr(a)sysmocom.de>
Gerrit-Attention: laforge <laforge(a)osmocom.org>
Gerrit-Comment-Date: Fri, 04 Mar 2022 18:31:32 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Comment-In-Reply-To: iedemam <michael(a)kapsulate.com>
Comment-In-Reply-To: neels <nhofmeyr(a)sysmocom.de>
Comment-In-Reply-To: laforge <laforge(a)osmocom.org>
Gerrit-MessageType: comment