Loading...
1# SPDX-License-Identifier: GPL-2.0-only
2
3# BPF interpreter that, for example, classic socket filters depend on.
4config BPF
5 bool
6
7# Used by archs to tell that they support BPF JIT compiler plus which
8# flavour. Only one of the two can be selected for a specific arch since
9# eBPF JIT supersedes the cBPF JIT.
10
11# Classic BPF JIT (cBPF)
12config HAVE_CBPF_JIT
13 bool
14
15# Extended BPF JIT (eBPF)
16config HAVE_EBPF_JIT
17 bool
18
19# Used by archs to tell that they want the BPF JIT compiler enabled by
20# default for kernels that were compiled with BPF JIT support.
21config ARCH_WANT_DEFAULT_BPF_JIT
22 bool
23
24menu "BPF subsystem"
25
26config BPF_SYSCALL
27 bool "Enable bpf() system call"
28 select BPF
29 select IRQ_WORK
30 select TASKS_RCU if PREEMPTION
31 select TASKS_TRACE_RCU
32 select BINARY_PRINTF
33 select NET_SOCK_MSG if NET
34 select NET_XGRESS if NET
35 select PAGE_POOL if NET
36 default n
37 help
38 Enable the bpf() system call that allows to manipulate BPF programs
39 and maps via file descriptors.
40
41config BPF_JIT
42 bool "Enable BPF Just In Time compiler"
43 depends on BPF
44 depends on HAVE_CBPF_JIT || HAVE_EBPF_JIT
45 depends on MODULES
46 help
47 BPF programs are normally handled by a BPF interpreter. This option
48 allows the kernel to generate native code when a program is loaded
49 into the kernel. This will significantly speed-up processing of BPF
50 programs.
51
52 Note, an admin should enable this feature changing:
53 /proc/sys/net/core/bpf_jit_enable
54 /proc/sys/net/core/bpf_jit_harden (optional)
55 /proc/sys/net/core/bpf_jit_kallsyms (optional)
56
57config BPF_JIT_ALWAYS_ON
58 bool "Permanently enable BPF JIT and remove BPF interpreter"
59 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
60 help
61 Enables BPF JIT and removes BPF interpreter to avoid speculative
62 execution of BPF instructions by the interpreter.
63
64 When CONFIG_BPF_JIT_ALWAYS_ON is enabled, /proc/sys/net/core/bpf_jit_enable
65 is permanently set to 1 and setting any other value than that will
66 return failure.
67
68config BPF_JIT_DEFAULT_ON
69 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
70 depends on HAVE_EBPF_JIT && BPF_JIT
71
72config BPF_UNPRIV_DEFAULT_OFF
73 bool "Disable unprivileged BPF by default"
74 default y
75 depends on BPF_SYSCALL
76 help
77 Disables unprivileged BPF by default by setting the corresponding
78 /proc/sys/kernel/unprivileged_bpf_disabled knob to 2. An admin can
79 still reenable it by setting it to 0 later on, or permanently
80 disable it by setting it to 1 (from which no other transition to
81 0 is possible anymore).
82
83 Unprivileged BPF could be used to exploit certain potential
84 speculative execution side-channel vulnerabilities on unmitigated
85 affected hardware.
86
87 If you are unsure how to answer this question, answer Y.
88
89source "kernel/bpf/preload/Kconfig"
90
91config BPF_LSM
92 bool "Enable BPF LSM Instrumentation"
93 depends on BPF_EVENTS
94 depends on BPF_SYSCALL
95 depends on SECURITY
96 depends on BPF_JIT
97 help
98 Enables instrumentation of the security hooks with BPF programs for
99 implementing dynamic MAC and Audit Policies.
100
101 If you are unsure how to answer this question, answer N.
102
103endmenu # "BPF subsystem"
1# SPDX-License-Identifier: GPL-2.0-only
2
3# BPF interpreter that, for example, classic socket filters depend on.
4config BPF
5 bool
6
7# Used by archs to tell that they support BPF JIT compiler plus which
8# flavour. Only one of the two can be selected for a specific arch since
9# eBPF JIT supersedes the cBPF JIT.
10
11# Classic BPF JIT (cBPF)
12config HAVE_CBPF_JIT
13 bool
14
15# Extended BPF JIT (eBPF)
16config HAVE_EBPF_JIT
17 bool
18
19# Used by archs to tell that they want the BPF JIT compiler enabled by
20# default for kernels that were compiled with BPF JIT support.
21config ARCH_WANT_DEFAULT_BPF_JIT
22 bool
23
24menu "BPF subsystem"
25
26config BPF_SYSCALL
27 bool "Enable bpf() system call"
28 select BPF
29 select IRQ_WORK
30 select TASKS_TRACE_RCU
31 select BINARY_PRINTF
32 select NET_SOCK_MSG if INET
33 default n
34 help
35 Enable the bpf() system call that allows to manipulate BPF programs
36 and maps via file descriptors.
37
38config BPF_JIT
39 bool "Enable BPF Just In Time compiler"
40 depends on BPF
41 depends on HAVE_CBPF_JIT || HAVE_EBPF_JIT
42 depends on MODULES
43 help
44 BPF programs are normally handled by a BPF interpreter. This option
45 allows the kernel to generate native code when a program is loaded
46 into the kernel. This will significantly speed-up processing of BPF
47 programs.
48
49 Note, an admin should enable this feature changing:
50 /proc/sys/net/core/bpf_jit_enable
51 /proc/sys/net/core/bpf_jit_harden (optional)
52 /proc/sys/net/core/bpf_jit_kallsyms (optional)
53
54config BPF_JIT_ALWAYS_ON
55 bool "Permanently enable BPF JIT and remove BPF interpreter"
56 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
57 help
58 Enables BPF JIT and removes BPF interpreter to avoid speculative
59 execution of BPF instructions by the interpreter.
60
61config BPF_JIT_DEFAULT_ON
62 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
63 depends on HAVE_EBPF_JIT && BPF_JIT
64
65config BPF_UNPRIV_DEFAULT_OFF
66 bool "Disable unprivileged BPF by default"
67 depends on BPF_SYSCALL
68 help
69 Disables unprivileged BPF by default by setting the corresponding
70 /proc/sys/kernel/unprivileged_bpf_disabled knob to 2. An admin can
71 still reenable it by setting it to 0 later on, or permanently
72 disable it by setting it to 1 (from which no other transition to
73 0 is possible anymore).
74
75source "kernel/bpf/preload/Kconfig"
76
77config BPF_LSM
78 bool "Enable BPF LSM Instrumentation"
79 depends on BPF_EVENTS
80 depends on BPF_SYSCALL
81 depends on SECURITY
82 depends on BPF_JIT
83 help
84 Enables instrumentation of the security hooks with BPF programs for
85 implementing dynamic MAC and Audit Policies.
86
87 If you are unsure how to answer this question, answer N.
88
89endmenu # "BPF subsystem"