calc-cbs-params.py uses kbit/s and bytes inconsistently

[nmuh]

Hello,
I'm currently researching on CBS and my computations on the CBS parameters differed from the script provided in this project.
I had a look into it and found a potential bug in lines 28-29:

tsn-doc/misc/calc-cbs-params.py

Lines 28 to 29 in 722928b

	hicredit = math.ceil(idleslope_a * frame_non_sr / link_speed)
	locredit = math.ceil(sendslope_a * max_frame_size_a / link_speed)

The variables idleslope_a, sendslope_a and linkspeed are measured in kbit/s

tsn-doc/misc/calc-cbs-params.py

Line 85 in 722928b

idleslope += int(stream['rate']) * frame_size * 8 / 1000

tsn-doc/misc/calc-cbs-params.py

Lines 137 to 138 in 722928b

	parser.add_argument('-S', dest='link_speed', default=1000000, type=int,
	help='Link speed in kbps')

tsn-doc/misc/calc-cbs-params.py

Line 150 in 722928b

sendslope_a = idleslope_a - args.link_speed

while max_frame_size_a and frame_non_sr are in Byte (if I did not miss anything).
The usage of kbit/s and byte in one equation sounds wrong to me. Also Annex L section L.2 a) shows an example, where all computations are done in bit and bit/s.

If I missed a point somewhere in the code or the specification, I'd be thankful for pointing it out.

Thanks!

[edersondisouza]

Hi @nmuh,

You didn't missing anything on the code or specification. Note that calc-cbs-params.py returns results to be used to set the CBS qdisc (man 8 tc-cbs), which expects hicredit and locredit to be in bytes.

So it's more of a convenience to just copy its output as input for setting up CBS qdisc - without having to convert to bytes.