linux-toradex.git/arch/loongarch/net, branch master

bpf: Add struct bpf_tramp_node object

2026-06-07T17:03:01+00:00

Adding struct bpf_tramp_node to decouple the link out of the trampoline
attachment info.

At the moment the object for attaching bpf program to the trampoline is
'struct bpf_tramp_link':

  struct bpf_tramp_link {
       struct bpf_link link;
       struct hlist_node tramp_hlist;
       u64 cookie;
  }

The link holds the bpf_prog pointer and forces one link - one program
binding logic. In following changes we want to attach program to multiple
trampolines but we want to keep just one bpf_link object.

Splitting struct bpf_tramp_link into:

  struct bpf_tramp_link {
       struct bpf_link link;
       struct bpf_tramp_node node;
  };

  struct bpf_tramp_node {
       struct bpf_link *link;
       struct hlist_node tramp_hlist;
       u64 cookie;
  };

The 'struct bpf_tramp_link' defines standard single trampoline link
and 'struct bpf_tramp_node' is the attachment trampoline object with
pointer to the bpf_link object.

This will allow us to define link for multiple trampolines, like:

  struct bpf_tracing_multi_link {
       struct bpf_link link;
       ...
       int nodes_cnt;
       struct bpf_tracing_multi_node nodes[] __counted_by(nodes_cnt);
  };

Cc: Hengqi Chen 
Signed-off-by: Jiri Olsa 
Link: https://lore.kernel.org/r/20260606123955.345967-9-jolsa@kernel.org
Signed-off-by: Alexei Starovoitov

Merge tag 'loongarch-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson

2026-04-24T16:54:45+00:00

Pull LoongArch updates from Huacai Chen:

 - Adjust build infrastructure for 32BIT/64BIT

 - Add HIGHMEM (PKMAP and FIX_KMAP) support

 - Show and handle CPU vulnerabilites correctly

 - Batch the icache maintenance for jump_label

 - Add more atomic instructions support for BPF JIT

 - Add more features (e.g. fsession) support for BPF trampoline

 - Some bug fixes and other small changes

* tag 'loongarch-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson: (21 commits)
  selftests/bpf: Enable CAN_USE_LOAD_ACQ_STORE_REL for LoongArch
  LoongArch: BPF: Add fsession support for trampolines
  LoongArch: BPF: Introduce emit_store_stack_imm64() helper
  LoongArch: BPF: Support up to 12 function arguments for trampoline
  LoongArch: BPF: Support small struct arguments for trampoline
  LoongArch: BPF: Open code and remove invoke_bpf_mod_ret()
  LoongArch: BPF: Support load-acquire and store-release instructions
  LoongArch: BPF: Support 8 and 16 bit read-modify-write instructions
  LoongArch: BPF: Add the default case in emit_atomic() and rename it
  LoongArch: Define instruction formats for AM{SWAP/ADD}.{B/H} and DBAR
  LoongArch: Batch the icache maintenance for jump_label
  LoongArch: Add flush_icache_all()/local_flush_icache_all()
  LoongArch: Add spectre boundry for syscall dispatch table
  LoongArch: Show CPU vulnerabilites correctly
  LoongArch: Make arch_irq_work_has_interrupt() true only if IPI HW exist
  LoongArch: Use get_random_canary() for stack canary init
  LoongArch: Improve the logging of disabling KASLR
  LoongArch: Align FPU register state to 32 bytes
  LoongArch: Handle CONFIG_32BIT in syscall_get_arch()
  LoongArch: Add HIGHMEM (PKMAP and FIX_KMAP) support
  ...

LoongArch: BPF: Add fsession support for trampolines

2026-04-23T04:49:36+00:00

Implement BPF_TRACE_FSESSION support in LoongArch BPF JIT. The logic
here is almost identical to what has been done in RISC-V JIT.

The key changes are:
  - Allocate stack space for function meta and session cookies
  - Introduce invoke_bpf() as a wrapper around invoke_bpf_prog() that
    populates session cookies before each invocation
  - Implement bpf_jit_supports_fsession() callback

Tested-by: Vincent Li 
Reviewed-by: Menglong Dong 
Signed-off-by: Hengqi Chen 
Signed-off-by: Huacai Chen

LoongArch: BPF: Introduce emit_store_stack_imm64() helper

2026-04-23T04:49:36+00:00

Introduce a helper to store 64-bit immediate on the trampoline stack.
The helper will be used in the next patch. Also refactor the existing
code to use this helper.

Tested-by: Vincent Li 
Reviewed-by: Menglong Dong 
Signed-off-by: Hengqi Chen 
Signed-off-by: Huacai Chen

LoongArch: BPF: Support up to 12 function arguments for trampoline

2026-04-22T07:45:34+00:00

Currently, LoongArch bpf trampoline supports up to 8 function arguments.
According to the statistics from commit 473e3150e30a ("bpf, x86: allow
function arguments up to 12 for TRACING"), there are over 200 functions
accept 9 to 12 arguments, so add 12 arguments support for trampoline.

With this patch, the following related testcases passed:

  sudo ./test_progs -a tracing_struct/struct_many_args
  sudo ./test_progs -a fentry_test/fentry_many_args
  sudo ./test_progs -a fexit_test/fexit_many_args

Acked-by: Hengqi Chen 
Tested-by: Hengqi Chen 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen

LoongArch: BPF: Support small struct arguments for trampoline

2026-04-22T07:45:34+00:00

In the current BPF code, the struct argument size is at most 16 bytes,
enforced by the verifier. According to the Procedure Call Standard for
LoongArch, the struct argument size below 16 bytes are provided as part
of the 8 argument registers, that is to say, the struct argument may be
passed in a pair of registers if its size is more than 8 bytes and no
more than 16 bytes.

Extend the BPF trampoline JIT to support attachment to functions that
take small structures (up to 16 bytes) as argument, save and restore a
number of "argument registers" rather than a number of arguments.

With this patch, the following related testcases passed:

  sudo ./test_progs -a tracing_struct/struct_args
  sudo ./test_progs -a tracing_struct/union_args

Link: https://github.com/loongson/la-abi-specs/blob/release/lapcs.adoc#structures
Acked-by: Hengqi Chen 
Tested-by: Hengqi Chen 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen

LoongArch: BPF: Open code and remove invoke_bpf_mod_ret()

2026-04-22T07:45:34+00:00

invoke_bpf_mod_ret() is a small wrapper over invoke_bpf_prog(), it
should check the return value of invoke_bpf_prog() and then return
immediately if invoke_bpf_prog() failed, just open code and remove
it due to it is called only once.

Acked-by: Hengqi Chen 
Tested-by: Hengqi Chen 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen

LoongArch: BPF: Support load-acquire and store-release instructions

2026-04-22T07:45:34+00:00

Use the LoongArch common memory access instructions with the barrier
'dbar' to support the BPF load-acquire and store-release instructions.

With this patch, the following testcases passed on LoongArch if the
macro CAN_USE_LOAD_ACQ_STORE_REL is usable in bpf selftests:

  sudo ./test_progs -t verifier_load_acquire
  sudo ./test_progs -t verifier_store_release
  sudo ./test_progs -t verifier_precision/bpf_load_acquire
  sudo ./test_progs -t verifier_precision/bpf_store_release
  sudo ./test_progs -t compute_live_registers/atomic_load_acq_store_rel

Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen

LoongArch: BPF: Support 8 and 16 bit read-modify-write instructions

2026-04-22T07:45:34+00:00

The 8 and 16 bit read-modify-write instructions {amadd/amswap}.{b/h}
were newly added in the latest LoongArch Reference Manual, use them to
avoid the error of unknown opcode if possible.

Acked-by: Hengqi Chen 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen

LoongArch: BPF: Add the default case in emit_atomic() and rename it

2026-04-22T07:45:34+00:00

Like the other archs such as x86 and riscv, add the default case
in emit_atomic() to print an error message for the invalid opcode
and return -EINVAL, then make its return type as int.

While at it, given that all of the instructions in emit_atomic()
are only read-modify-write instructions, rename emit_atomic() to
emit_atomic_rmw() to make it clear, because there will be a new
function emit_atomic_ld_st() for load-acquire and store-release
instructions in the later patch.

Acked-by: Hengqi Chen 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Huacai Chen