bpfctl Guide
bpfctl
is the command line tool for interacting with bpfd
.
bpfctl
allows the user to load
, unload
, get
and list
eBPF programs.
Notes For This Guide
As described in other sections, bpfd
can be run as either a privileged process or
a systemd service.
If run as a privileged process, bpfctl
will most likely be run from your local
development branch and will require sudo
.
Example:
If run as a systemd service, bpfctl
will most likely be installed in your $PATH,
and will also require sudo
.
Example:
The examples here use sudo bpfctl
in place of sudo ./target/debug/bpfctl
for readability,
use as your system is deployed.
eBPF object files used in the examples are taken from the
examples and
integration-test directories
from the bpfd
repository.
Basic Syntax
Below are the commands supported by bpfctl
.
sudo bpfctl --help
A client for working with bpfd
Usage: bpfctl <COMMAND>
------
Commands:
---------
load-from-file Load an eBPF program from a local .o file
load-from-image Load an eBPF program packaged in a OCI container image from a given registry
unload Unload an eBPF program using the program id
list List all eBPF programs loaded via bpfd
get Get a program's metadata by program id
pull-bytecode Pull a bytecode image for future use by a load command
help Print this message or the help of the given subcommand(s)
Options:
--------
-h, --help Print help
-V, --version Print version
bpfctl load
The bpfctl load-from-file
and bpfctl load-from-image
commands are used to load eBPF programs.
The bpfctl load-from-file
command is used to load a locally built eBPF program.
The bpfctl load-from-image
command is used to load an eBPF program packaged in a OCI container
image from a given registry.
Each program type (i.e. <COMMAND>
) has it's own set of attributes specific to the program type,
and those attributes MUST come after the program type is entered.
There are a common set of attributes, and those MUST come before the program type is entered.
sudo bpfctl load-from-file --help
Load an eBPF program from a local .o file
Usage: bpfctl load-from-file [OPTIONS] --path <PATH> --name <NAME> <COMMAND>
------
Commands:
---------
xdp Install an eBPF program on the XDP hook point for a given interface
tc Install an eBPF program on the TC hook point for a given interface
tracepoint Install an eBPF program on a Tracepoint
kprobe Install an eBPF kprobe or kretprobe
uprobe Install an eBPF uprobe or uretprobe
help Print this message or the help of the given subcommand(s)
Options:
--------
-p, --path <PATH>
Required: Location of local bytecode file as fully qualified file path.
Example: --path $HOME/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o
-n, --name <NAME>
Required: The name of the function that is the entry point for the BPF program
-g, --global <GLOBAL>...
Optional: Global variables to be set when program is loaded.
Format: <NAME>=<Hex Value>
This is a very low level primitive. The caller is responsible for formatting
the byte string appropriately considering such things as size, endianness,
alignment and packing of data structures.
-m, --metadata <METADATA>
Optional: Specify Key/Value metadata to be attached to a program when it
is loaded by bpfd.
Format: <KEY>=<VALUE>
This can later be used to list a certain subset of programs which contain
the specified metadata.
--map-owner-id <MAP_OWNER_ID>
Optional: Program id of loaded eBPF program this eBPF program will share a map with.
Only used when multiple eBPF programs need to share a map.
Example: --map-owner-id 63178
-h, --help
Print help (see a summary with '-h')
and
sudo bpfctl load-from-image --help
Load an eBPF program packaged in a OCI container image from a given registry
Usage: bpfctl load-from-image [OPTIONS] --image-url <IMAGE_URL> <COMMAND>
Commands:
xdp Install an eBPF program on the XDP hook point for a given interface
tc Install an eBPF program on the TC hook point for a given interface
tracepoint Install an eBPF program on a Tracepoint
kprobe Install an eBPF kprobe or kretprobe
uprobe Install an eBPF uprobe or uretprobe
help Print this message or the help of the given subcommand(s)
Options:
-i, --image-url <IMAGE_URL>
Required: Container Image URL.
Example: --image-url quay.io/bpfd-bytecode/xdp_pass:latest
-r, --registry-auth <REGISTRY_AUTH>
Optional: Registry auth for authenticating with the specified image registry.
This should be base64 encoded from the '<username>:<password>' string just like
it's stored in the docker/podman host config.
Example: --registry_auth "YnjrcKw63PhDcQodiU9hYxQ2"
-p, --pull-policy <PULL_POLICY>
Optional: Pull policy for remote images.
[possible values: Always, IfNotPresent, Never]
[default: IfNotPresent]
-n, --name <NAME>
Optional: The name of the function that is the entry point for the BPF program.
If not provided, the program name defined as part of the bytecode image will be used.
[default: ]
-g, --global <GLOBAL>...
Optional: Global variables to be set when program is loaded.
Format: <NAME>=<Hex Value>
This is a very low level primitive. The caller is responsible for formatting
the byte string appropriately considering such things as size, endianness,
alignment and packing of data structures.
-m, --metadata <METADATA>
Optional: Specify Key/Value metadata to be attached to a program when it
is loaded by bpfd.
Format: <KEY>=<VALUE>
This can later be used to list a certain subset of programs which contain
the specified metadata.
Example: --metadata owner=acme
--map-owner-id <MAP_OWNER_ID>
Optional: Program id of loaded eBPF program this eBPF program will share a map with.
Only used when multiple eBPF programs need to share a map.
Example: --map-owner-id 63178
-h, --help
Print help (see a summary with '-h')
When using either load command, --path
, --image-url
, --registry-auth
, --pull-policy
, --name
,
--global
, --metadata
and --map-owner-id
must be entered before the <COMMAND>
(xdp
, tc
,
tracepoint
, etc) is entered.
Then each <COMMAND>
has its own custom parameters (same for both bpfctl load-from-file
and
bpfctl load-from-image
):
sudo bpfctl load-from-file xdp --help
Install an eBPF program on the XDP hook point for a given interface
Usage: bpfctl load-from-file --path <PATH> --name <NAME> xdp [OPTIONS] --iface <IFACE> --priority <PRIORITY>
------
Options:
--------
-i, --iface <IFACE>
Required: Interface to load program on
-p, --priority <PRIORITY>
Required: Priority to run program in chain. Lower value runs first
--proceed-on <PROCEED_ON>...
Optional: Proceed to call other programs in chain on this exit code.
Multiple values supported by repeating the parameter.
Example: --proceed-on "pass" --proceed-on "drop"
[possible values: aborted, drop, pass, tx, redirect, dispatcher_return]
[default: pass, dispatcher_return]
-h, --help
Print help (see a summary with '-h')
Example loading from local file (--path
is the fully qualified path):
sudo bpfctl load-from-file --path $HOME/src/bpfd/tests/integration-test/bpf/.output/xdp_pass.bpf.o --name "pass" xdp --iface vethb2795c7 --priority 100
Example from image in remote repository (Note: --name
is built into the image and is not required):
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/xdp_pass:latest xdp --iface vethb2795c7 --priority 100
The tc
command is similar to xdp
, but it also requires the direction
option
and the proceed-on
values are different.
sudo bpfctl load-from-file tc -h
Install an eBPF program on the TC hook point for a given interface
Usage: bpfctl load-from-file --path <PATH> --name <NAME> tc [OPTIONS] --direction <DIRECTION> --iface <IFACE> --priority <PRIORITY>
------
Options:
--------
-d, --direction <DIRECTION>
Required: Direction to apply program.
[possible values: ingress, egress]
-i, --iface <IFACE>
Required: Interface to load program on
-p, --priority <PRIORITY>
Required: Priority to run program in chain. Lower value runs first
--proceed-on <PROCEED_ON>...
Optional: Proceed to call other programs in chain on this exit code.
Multiple values supported by repeating the parameter.
Example: --proceed-on "ok" --proceed-on "pipe"
[possible values: unspec, ok, reclassify, shot, pipe, stolen, queued,
repeat, redirect, trap, dispatcher_return]
[default: ok, pipe, dispatcher_return]
-h, --help
Print help (see a summary with '-h')
The following is an example of the tc
command using short option names:
sudo bpfctl load-from-file -p $HOME/src/bpfd/tests/integration-test/bpf/.output/tc_pass.bpf.o -n "pass" tc -d ingress -i mynet1 -p 40
For the tc_pass.bpf.o
program loaded with the command above, the name
would be set as shown in the following snippet:
Additional bpfctl Load Examples
Below are some additional examples of bpfctl load
commands:
XDP
sudo bpfctl load-from-file --path $HOME/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o --name "xdp_stats" xdp --iface vethb2795c7 --priority 35
TC
sudo bpfctl load-from-file --path $HOME/src/bpfd/examples/go-tc-counter/bpf_bpfel.o --name "stats"" tc --direction ingress --iface vethb2795c7 --priority 110
Kprobe
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/kprobe:latest kprobe -f try_to_wake_up
Kretprobe
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/kretprobe:latest kprobe -f try_to_wake_up -r
Uprobe
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/uprobe:latest uprobe -f "malloc" -t "libc"
Uretprobe
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/uretprobe:latest uprobe -f "malloc" -t "libc" -r
Setting Global Variables in eBPF Programs
Global variables can be set for any eBPF program type when loading as follows:
sudo bpfctl load-from-file -p $HOME/src/bpfd/tests/integration-test/bpf/.output/tc_pass.bpf.o -g GLOBAL_u8=01020304 GLOBAL_u32=0A0B0C0D -n "pass" tc -d ingress -i mynet1 -p 40
Note, that when setting global variables, the eBPF program being loaded must have global variables named with the strings given, and the size of the value provided must match the size of the given variable. For example, the above command can be used to update the following global variables in an eBPF program.
Modifying the Proceed-On Behavior
The proceed-on
setting applies to xdp
and tc
programs. For both of these
program types, an ordered list of eBPF programs is maintained per attach point.
The proceed-on
setting determines whether processing will "proceed" to the
next eBPF program in the list, or terminate processing and return, based on the
program's return value. For example, the default proceed-on
configuration for
an xdp
program can be modified as follows:
sudo bpfctl load-from-file -p $HOME/src/bpfd/tests/integration-test/bpf/.output/xdp_pass.bpf.o -n "pass" xdp -i mynet1 -p 30 --proceed-on drop pass dispatcher_return
Sharing Maps Between eBPF Programs
WARNING Currently for the map sharing feature to work the LIBBPF_PIN_BY_NAME flag MUST be set in the shared bpf map definitions. Please see this aya issue for future work that will change this requirement.
To share maps between eBPF programs, first load the eBPF program that owns the maps. One eBPF program must own the maps.
sudo bpfctl load-from-file --path $HOME/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o -n "xdp_stats" xdp --iface vethb2795c7 --priority 100
6371
Next, load additional eBPF programs that will share the existing maps by passing
the program id of the eBPF program that owns the maps using the --map-owner-id
parameter:
sudo bpfctl load-from-file --path $HOME/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o -n "xdp_stats" --map-owner-id 6371 xdp --iface vethff657c7 --priority 100
6373
Use the bpfctl get <ID>
command to display the configuration:
sudo bpfctl list
Program ID Name Type Load Time
6371 xdp_stats xdp 2023-07-18T16:50:46-0400
6373 xdp_stats xdp 2023-07-18T16:51:06-0400
sudo bpfctl get 6371
Bpfd State
---------------
Name: xdp_stats
Path: /home/<$USER>/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o
Global: None
Metadata: None
Map Pin Path: /run/bpfd/fs/maps/6371
Map Owner ID: None
Map Used By: 6371
6373
Priority: 50
Iface: vethff657c7
Position: 1
Proceed On: pass, dispatcher_return
:
sudo bpfctl get 6373
Bpfd State
---------------
Name: xdp_stats
Path: /home/<$USER>/src/bpfd/examples/go-xdp-counter/bpf_bpfel.o
Global: None
Metadata: None
Map Pin Path: /run/bpfd/fs/maps/6371
Map Owner ID: 6371
Map Used By: 6371
6373
Priority: 50
Iface: vethff657c7
Position: 0
Proceed On: pass, dispatcher_return
:
As the output shows, the first program (6371
) owns the map, with Map Owner ID
of None
and the Map Pin Path
(/run/bpfd/fs/maps/6371
) that includes its own ID.
The second program (6373
) references the first program via the Map Owner ID
set
to 6371
and the Map Pin Path
(/run/bpfd/fs/maps/6371
) set to same directory as
the first program, which includes the first program's ID.
The output for both commands shows the map is being used by both programs via
the Map Used By
with values of 6371
and 6373
.
The eBPF programs can be unloaded any order, the Map Pin Path
will not be deleted
until all the programs referencing the maps are unloaded:
bpfctl list
The bpfctl list
command lists all the bpfd loaded eBPF programs:
sudo bpfctl list
Program ID Name Type Load Time
6201 pass xdp 2023-07-17T17:17:53-0400
6202 sys_enter_openat tracepoint 2023-07-17T17:19:09-0400
6204 stats tc 2023-07-17T17:20:14-0400
To see all eBPF programs loaded on the system, include the --all
option.
sudo bpfctl list --all
Program ID Name Type Load Time
52 restrict_filesy lsm 2023-05-03T12:53:34-0400
166 dump_bpf_map tracing 2023-05-03T12:53:52-0400
167 dump_bpf_prog tracing 2023-05-03T12:53:52-0400
455 cgroup_device 2023-05-03T12:58:26-0400
:
6190 cgroup_skb 2023-07-17T17:15:23-0400
6191 cgroup_device 2023-07-17T17:15:23-0400
6192 cgroup_skb 2023-07-17T17:15:23-0400
6193 cgroup_skb 2023-07-17T17:15:23-0400
6194 cgroup_device 2023-07-17T17:15:23-0400
6201 pass xdp 2023-07-17T17:17:53-0400
6202 sys_enter_openat tracepoint 2023-07-17T17:19:09-0400
6203 dispatcher tc 2023-07-17T17:20:14-0400
6204 stats tc 2023-07-17T17:20:14-0400
6207 xdp xdp 2023-07-17T17:27:13-0400
To filter on a given program type, include the --program-type
parameter:
sudo bpfctl list --all --program-type tc
Program ID Name Type Load Time
6203 dispatcher tc 2023-07-17T17:20:14-0400
6204 stats tc 2023-07-17T17:20:14-0400
bpfctl get
To retrieve detailed information for a loaded eBPF program, use the
bpfctl get <ID>
command.
If the eBPF program was loaded via bpfd, then there will be a Bpfd State
section with bpfd related attributes and a Kernel State
section with
kernel information.
If the eBPF program was loaded outside of bpfd, then the Bpfd State
section will be empty and Kernel State
section will be populated.
sudo bpfctl get 6204
Bpfd State
---------------
Name: stats
Image URL: quay.io/bpfd-bytecode/go-tc-counter:latest
Pull Policy: IfNotPresent
Global: None
Metadata: None
Map Pin Path: /run/bpfd/fs/maps/6204
Map Owner ID: None
Map Used By: 6204
Priority: 100
Iface: vethff657c7
Position: 0
Direction: eg
Proceed On: pipe, dispatcher_return
Kernel State
----------------------------------
ID: 6204
Name: stats
Type: tc
Loaded At: 2023-07-17T17:20:14-0400
Tag: ead94553702a3742
GPL Compatible: true
Map IDs: [2705]
BTF ID: 2821
Size Translated (bytes): 176
JITed: true
Size JITed (bytes): 116
Kernel Allocated Memory (bytes): 4096
Verified Instruction Count: 24
sudo bpfctl get 6190
Bpfd State
---------------
NONE
Kernel State
----------------------------------
ID: 6190
Name: None
Type: cgroup_skb
Loaded At: 2023-07-17T17:15:23-0400
Tag: 6deef7357e7b4530
GPL Compatible: true
Map IDs: []
BTF ID: 0
Size Translated (bytes): 64
JITed: true
Size JITed (bytes): 55
Kernel Allocated Memory (bytes): 4096
Verified Instruction Count: 8
bpfctl unload
The bpfctl unload
command takes the program id from the load or list command as a parameter,
and unloads the requested eBPF program:
sudo bpfctl list
Program ID Name Type Load Time
6201 pass xdp 2023-07-17T17:17:53-0400
6202 sys_enter_openat tracepoint 2023-07-17T17:19:09-0400
bpfctl pull-bytecode
The bpfctl pull-bytecode
command pulls a given bytecode image for future use
by a load command.
sudo bpfctl pull-bytecode --help
Pull a bytecode image for future use by a load command
Usage: bpfctl pull-bytecode [OPTIONS] --image-url <IMAGE_URL>
Options:
-i, --image-url <IMAGE_URL>
Required: Container Image URL.
Example: --image-url quay.io/bpfd-bytecode/xdp_pass:latest
-r, --registry-auth <REGISTRY_AUTH>
Optional: Registry auth for authenticating with the specified image registry.
This should be base64 encoded from the '<username>:<password>' string just like
it's stored in the docker/podman host config.
Example: --registry_auth "YnjrcKw63PhDcQodiU9hYxQ2"
-p, --pull-policy <PULL_POLICY>
Optional: Pull policy for remote images.
[possible values: Always, IfNotPresent, Never]
[default: IfNotPresent]
-h, --help
Print help (see a summary with '-h')
Example usage:
sudo bpfctl pull-bytecode --image-url quay.io/bpfd-bytecode/xdp_pass:latest
Successfully downloaded bytecode
Then when loaded, the local image will be used:
sudo bpfctl load-from-image --image-url quay.io/bpfd-bytecode/xdp_pass:latest --pull-policy IfNotPresent xdp --iface vethff657c7 --priority 100
Bpfd State
---------------
Name: pass
Image URL: quay.io/bpfd-bytecode/xdp_pass:latest
Pull Policy: IfNotPresent
Global: None
Metadata: None
Map Pin Path: /run/bpfd/fs/maps/406681
Map Owner ID: None
Maps Used By: None
Priority: 100
Iface: vethff657c7
Position: 2
Proceed On: pass, dispatcher_return
Kernel State
----------------------------------
ID: 406681
Name: pass
Type: xdp
Loaded At: 1917-01-27T01:37:06-0500
Tag: 4b9d1b2c140e87ce
GPL Compatible: true
Map IDs: [736646]
BTF ID: 555560
Size Translated (bytes): 96
JITted: true
Size JITted: 67
Kernel Allocated Memory (bytes): 4096
Verified Instruction Count: 9