PRIO(8)

💖 PRIO(8) — Priority qdisc 💖

✨ NAME ✨

okay so the manual name is PRIO - Priority qdisc 🧚 that's the whole vibe

🎀 SYNOPSIS 🎀

tc qdisc ... dev dev ( parent classid | root) [ handle major: ] prio [
bands bands ] [ priomap band band band...  ] [ estimator interval
timeconstant ]

🌸 DESCRIPTION 🌸

okay so the PRIO qdisc is a simple classful queueing discipline 💅 it holds an arbitrary number of classes, each with a different priority. the classes get dequeued in numerical descending order of priority, so the highest priority goes first. PRIO is a scheduler and it never delays packets 🫶 that makes it a work-conserving qdisc, though the qdiscs sitting inside the classes might not be.

this is so useful for lowering latency when you don't need to slow traffic down 🦋

🦋 ALGORITHM 🦋

when you make one with tc qdisc add, a fixed number of bands gets created. each band is a class 👑 you can't add classes with tc qdisc add though. instead you specify the number of bands on the command line when you attach PRIO to its root.

when it's dequeueing, band 0 gets tried first. only if it didn't deliver a packet does PRIO try band 1, and then onwards. so the most reliability-hungry packets should go to band 0, minimum delay to band 1, and the rest to band 2 🍓

the PRIO qdisc itself has minor number 0, so band 0 is actually major:1, band 1 is major:2, and so on. for major, swap in the major number that got assigned to the qdisc on tc qdisc add with the handle parameter ✨

🧚 CLASSIFICATION 🧚

there are three ways PRIO can decide which band a packet lands in 💜

From userspace

• a process with sufficient privileges can encode the destination class directly with SO_PRIORITY, see socket(7).

With a tc filter

• a tc filter attached to the root qdisc can point traffic directly to a class.

With the priomap

• this one is based on the packet priority, which is derived from the Type of Service assigned to the packet.

only the priomap is specific to this qdisc 🙈

💅 QDISC PARAMETERS 💅

• bands — the number of bands 🌈 if you change it from the default of 3, you have to update the priomap as well, no skipping that step bestie.
• priomap — this maps the priority of a packet to a class ✨ the priority can be set straight from userspace, or derived from the Type of Service of the packet. it's decided by the TOS octet of the packet, which looks like this:

0   1   2   3   4   5   6   7
+---+---+---+---+---+---+---+---+
|           |               |   |
|PRECEDENCE |      TOS      |MBZ|
|           |               |   |
+---+---+---+---+---+---+---+---+

the four TOS bits (the 'TOS field') are defined as:

Binary Decimal  Meaning
-----------------------------------------
1000   8         Minimize delay (md)
0100   4         Maximize throughput (mt)
0010   2         Maximize reliability (mr)
0001   1         Minimize monetary cost (mmc)
0000   0         Normal Service

real talk about the math 🤓 there's 1 bit sitting to the right of those four bits, so the actual value of the TOS field is double the value of the TOS bits. tcpdump -v -v shows you the value of the entire TOS field, not just the four bits. it's the value you see in the first column of this table:

TOS     Bits  Means                    Linux Priority    Band
------------------------------------------------------------
0x0     0     Normal Service           0 Best Effort     1
0x2     1     Minimize Monetary Cost   0 Best Effort     1
0x4     2     Maximize Reliability     0 Best Effort     1
0x6     3     mmc+mr                   0 Best Effort     1
0x8     4     Maximize Throughput      2 Bulk            2
0xa     5     mmc+mt                   2 Bulk            2
0xc     6     mr+mt                    2 Bulk            2
0xe     7     mmc+mr+mt                2 Bulk            2
0x10    8     Minimize Delay           6 Interactive     0
0x12    9     mmc+md                   6 Interactive     0
0x14    10    mr+md                    6 Interactive     0
0x16    11    mmc+mr+md                6 Interactive     0
0x18    12    mt+md                    4 Int. Bulk       1
0x1a    13    mmc+mt+md                4 Int. Bulk       1
0x1c    14    mr+mt+md                 4 Int. Bulk       1
0x1e    15    mmc+mr+mt+md             4 Int. Bulk       1

the second column is the value of the relevant four TOS bits, followed by what they translate to. for example, 15 means a packet wanting Minimal Monetary Cost, Maximum Reliability, Maximum Throughput AND Minimum Delay. chaotic, I love it 💀

the fourth column shows how the Linux kernel interprets the TOS bits, meaning which Priority they get mapped to 👑

the last column is the result of the default priomap. on the command line, the default priomap looks like this:

    1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1

so priority 4, for example, gets mapped to band number 1. the priomap also lets you list higher priorities (> 7) that don't match any TOS mapping, but get set by other means 🎀

this table from RFC 1349 (go read it for more details 🙈) explains how applications might very well set their TOS bits:

TELNET                   1000           (minimize delay)
FTP
        Control          1000           (minimize delay)
        Data             0100           (maximize throughput)

TFTP                     1000           (minimize delay)

SMTP
        Command phase    1000           (minimize delay)
        DATA phase       0100           (maximize throughput)

Domain Name Service
        UDP Query        1000           (minimize delay)
        TCP Query        0000
        Zone Transfer    0100           (maximize throughput)

NNTP                     0001           (minimize monetary cost)

ICMP
        Errors           0000
        Requests         0000 (mostly)
        Responses        <same as request> (mostly)

🫧 CLASSES 🫧

PRIO classes can't be configured any further 🙅‍♀️ they get automatically created the moment the PRIO qdisc is attached. each class can hold yet another qdisc though, so you can layer it up 🍰

🚨 BUGS 🚨

⚠️ big amounts of traffic in the lower bands can starve the higher bands. you can prevent that by attaching a shaper, for example tc-tbf(8), to these bands so they can't dominate the link. don't let the lower bands hog everything, bestie 😭

👑 AUTHORS 👑

Alexey N. Kuznetsov, <kuznet@ms2.inr.ac.ru>, and J Hadi Salim <hadi@cyberus.ca>. this manpage is maintained by bert hubert <ahu@ds9a.nl> 🫶

🌈 okay that's the whole page, periodt 💅 source is iproute2, last updated 16 December 2001, and this was a Linux manual page. love you, mwah 💖🦋