Project

General

Profile

Bug #1759

Wrong calculation of DL window size for DL assignment

Added by arvind.sirsikar almost 2 years ago. Updated about 2 months ago.

Status:
Stalled
Priority:
High
Assignee:
sysmocom
Target version:
-
Start date:
06/28/2016
Due date:
% Done:

30%

Spec Reference:

Description

Hi,

When we create DOWNLINK TBF without existing UPLINK TBF i.e DL assignment on PCH case, The calculation of window size is found to be incorrect.

Description:
4 time slots is configured for DL and osmo-pcu.cfg is configured as window-size 64 104.

When we try to do IPERF in DL direction, PCU allocates window-size as 160 but configures 4 time slots
as seen by PCU VTY. Below is the result of VTY output.
DL TBFs
TBF: TFI=0 TLLI=0xf73d2ece (valid) DIR=DL IMSI=901555000001280
created=1095 state=0000000a 1st_TS=4 1st_cTS=6 ctrl_TS=6 MS_CLASS=0/1
TS_alloc=4 5 6! 7 CS=MCS-9 WS=160 V(A)=12 V(S)=12 nBSN=138
But we see proper calculation of window-size when DL assignement is done on PACCH. as seen by VTY interface below
DL TBFs
TBF: TFI=0 TLLI=0xf73d2ece (valid) DIR=DL IMSI=901555000001280
created=1095 state=0000000a 1st_TS=4 1st_cTS=6 ctrl_TS=6 MS_CLASS=0/1
TS_alloc=4 5 6! 7 CS=MCS-9 WS=480 V(A)=138 V(S)=139 nBSN=138

Thanks,
Aravind Sirsikar


Related issues

Related to OsmoPCU - Feature #1533: Separate the window handling from the TBF more clearly Feedback 02/22/2016
Related to OsmoPCU - Bug #1524: PACCH on the wrong timeslot Stalled 02/22/2016

History

#1 Updated by laforge 8 months ago

#2 Updated by laforge 8 months ago

  • Assignee changed from arvind.sirsikar to sysmocom

#3 Updated by laforge 6 months ago

  • Assignee changed from sysmocom to msuraev
  • Priority changed from High to Normal

#4 Updated by laforge 6 months ago

  • Priority changed from Normal to High

#5 Updated by msuraev 6 months ago

I don't see explicit tests for it - only as part of larger TBF tests. So the first step would to add such a test.

#6 Updated by msuraev 5 months ago

  • Project changed from Cellular Network Infrastructure to OsmoPCU

#7 Updated by msuraev 4 months ago

  • Related to Feature #1533: Separate the window handling from the TBF more clearly added

#8 Updated by msuraev 4 months ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 20

Related gerrit 5355, 5336-5341 are under review. Remaining parts: clearly differenciate between CCCH and PACCH assignment, figure out whyit makes a difference to window size calculations and test with different window sizes settings and available channels.

#9 Updated by msuraev 3 months ago

  • Status changed from In Progress to Stalled

#10 Updated by msuraev 3 months ago

  • Status changed from Stalled to In Progress
  • % Done changed from 20 to 30

Patches were merged. Additional gerrit 5786 was sent for review.

#11 Updated by msuraev 3 months ago

5786 is merged, needs additional testing.

#12 Updated by msuraev 3 months ago

The window size parameter in vty defined as follows:

window-size <0-1024> [<0-256>]
Window size configuration (b + N_PDCH * f)
Base value (b)
Factor for number of PDCH (f)

So the configuration window-size 64 104 means that WS(1TS) = 64 + 1 * 104 = 168 and WS(4TS) = 64 + 4 * 104 = 480. By default f = 0.

#13 Updated by msuraev 3 months ago

  • Related to Bug #1524: PACCH on the wrong timeslot added

#14 Updated by msuraev 3 months ago

In addition to difference between PACCH and CCCH, the TBF can be assigned via CCCH but configrmed via PACCH (see rcv_control_ack() function). This and the related #1524 makes it harder to reproduce this reliably.

Related gerrit 6239 is under review.

#15 Updated by laforge 3 months ago

Hi Max,
On Mon, Feb 05, 2018 at 03:45:50PM +0000, msuraev [REDMINE] wrote:

In addition to difference between PACCH and CCCH, the TBF can be assigned via CCCH but configrmed via PACCH (see rcv_control_ack() function). This and the related #1524 makes it harder to reproduce this reliably.

would it be possible to disable certain variants as a temporary hack?

#16 Updated by msuraev about 2 months ago

  • Status changed from In Progress to Stalled

Needs a way to reproduce this reliably.

#17 Updated by laforge about 2 months ago

  • Assignee changed from msuraev to sysmocom

Also available in: Atom PDF