Implement stack argument passing for BPF-to-BPF and kfunc calls with
more than 5 parameters on arm64, following the AAPCS64 calling
convention.
BPF R1-R5 already map to x0-x4. With BPF_REG_0 moved to x8 by the
previous commit, x5-x7 are free for arguments 6-8. Arguments 9-12
spill onto the stack at [SP+0], [SP+8], ... and the callee reads
them from [FP+16], [FP+24], ... (above the saved FP/LR pair).
BPF convention uses fixed offsets from BPF_REG_PARAMS (r11): off=-8 is
always arg 6, off=-16 arg 7, etc. The verifier invalidates all outgoing
stack arg slots after each call, so the compiler must re-store before
every call. This means x5-x7 don't need to be saved on stack.
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045158.2402494-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Move the BPF return value register from x7 to x8, freeing x7 for use
as an argument register. AAPCS64 designates x8 as the indirect result
location register; it is caller-saved and not used for argument
passing, making it a suitable home for BPF_REG_0.
This is a prerequisite for stack argument support, which needs x5-x7
to pass arguments 6-8 to native kfuncs following the AAPCS64 calling
convention.
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045153.2402197-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a test that verifies precision backtracking works correctly
across BPF-to-BPF calls when stack arguments are involved.
The test passes a size value as incoming stack arg (arg6) to a
subprog, which forwards it as the mem__sz parameter (outgoing arg7)
to bpf_kfunc_call_stack_arg_mem. The expected __msg annotations
verify that precision propagates from the kfunc's mem__sz argument
back through the subprog frame to the caller's outgoing stack arg
store.
A companion BTF file (btf__stack_arg_precision.c) provides named
parameter BTF for the __naked subprog via __btf_func_path.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045148.2400087-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add inline-asm based verifier tests that exercise stack argument
validation logic directly.
Positive tests:
- subprog call with 6 arg's
- Two sequential calls to different subprogs (6-arg and 7-arg)
- Share a r11 store for both branches
Negative tests — verifier rejection:
- Read from uninitialized incoming stack arg slot
- Gap in outgoing slots: only r11-16 written, r11-8 missing
- Write at r11-80, exceeding max 7 stack args
- Missing store on one branch with a shared store
- First call has proper stack arguments and the second
call intends to inherit stack arguments but not working
- r11 load ordering issue
Negative tests — pointer/ref tracking:
- Pruning type mismatch: one branch stores PTR_TO_STACK, the
other stores a scalar, callee dereferences — must not prune
- Release invalidation: bpf_sk_release invalidates a socket
pointer stored in a stack arg slot
- Packet pointer invalidation: bpf_skb_pull_data invalidates
a packet pointer stored in a stack arg slot
- Null propagation: PTR_TO_MAP_VALUE_OR_NULL stored in stack
arg slot, null branch attempts dereference via callee
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045143.2399278-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
When __naked subprogs are used in verifier tests, clang drops
parameter names from their BTF FUNC_PROTO entries. This prevents
the verifier from resolving stack argument slots by name.
Add a __btf_func_path(path) annotation that points to a separate
BTF file containing properly-named FUNC entries. The test_loader
matches FUNC entries by name, detects anonymous parameters, and
replaces the FUNC_PROTO with a new one that carries parameter
names from the custom file while preserving the original type IDs.
The custom BTF file also serves as btf_custom_path for kfunc
resolution when no separate btf_custom_path is specified.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045138.2398886-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add x86_64 JIT support for BPF functions and kfuncs with more than
5 arguments. The extra arguments are passed through a stack area
addressed by register r11 (BPF_REG_PARAMS) in BPF bytecode,
which the JIT translates to native code.
The JIT follows the x86-64 calling convention for both BPF-to-BPF
and kfunc calls:
- Arg 6 is passed in the R9 register
- Args 7+ are passed on the stack
Incoming arg 6 (BPF r11+8) is translated to a MOV from R9 rather
than a memory load. Incoming args 7+ (BPF r11+16, r11+24, ...) map
directly to [rbp + 16], [rbp + 24], ..., matching the x86-64 stack
layout after CALL + PUSH RBP, so no offset adjustment is needed.
tail_call_reachable is rejected by the verifier and priv_stack is
disabled by the JIT when stack args exist, so R9 is always
available. When BPF bytecode writes to the arg-6 stack slot
(offset -8), the JIT emits a MOV into R9 instead of a memory store.
Outgoing args 7+ are placed at [rsp] in a pre-allocated area below
callee-saved registers, using:
native_off = outgoing_arg_base - outgoing_rsp - bpf_off - 16
The native x86_64 stack layout with stack arguments:
high address
+-------------------------+
| incoming stack arg N | [rbp + 16 + (N-7)*8] (from caller)
| ... |
| incoming stack arg 7 | [rbp + 16]
+-------------------------+
| return address | [rbp + 8]
| saved rbp | [rbp]
+-------------------------+
| BPF program stack | (round_up(stack_depth, 8) bytes)
+-------------------------+
| callee-saved regs | (r12, rbx, r13, r14, r15 as needed)
+-------------------------+
| outgoing arg M | [rsp + (M-7)*8]
| ... |
| outgoing arg 7 | [rsp]
+-------------------------+ rsp
low address
Acked-by: Puranjay Mohan <puranjay@kernel.org>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045122.2393118-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Extend the stack argument mechanism to kfunc calls, allowing kfuncs
with more than 5 parameters to receive additional arguments via the
r11-based stack arg area.
For kfuncs, the caller is a BPF program and the callee is a kernel
function. The BPF program writes outgoing args at negative r11
offsets, following the same convention as BPF-to-BPF calls:
Outgoing: r11 - 8 (arg6), ..., r11 - N*8 (last arg)
The following is an example:
int foo(int a1, int a2, int a3, int a4, int a5, int a6, int a7) {
...
kfunc1(a1, a2, a3, a4, a5, a6, a7, a8);
...
kfunc2(a1, a2, a3, a4, a5, a6, a7, a8, a9);
...
}
Caller (foo), generated by llvm
===============================
Incoming (positive offsets):
r11+8: [incoming arg 6]
r11+16: [incoming arg 7]
Outgoing for kfunc1 (negative offsets):
r11-8: [outgoing arg 6]
r11-16: [outgoing arg 7]
r11-24: [outgoing arg 8]
Outgoing for kfunc2 (negative offsets):
r11-8: [outgoing arg 6]
r11-16: [outgoing arg 7]
r11-24: [outgoing arg 8]
r11-32: [outgoing arg 9]
Later JIT will marshal outgoing arguments to the native calling
convention for kfunc1() and kfunc2().
For kfunc calls where stack args are used as constant or size
parameters, a mark_stack_arg_precision() helper is used to propagate
precision and do proper backtracking.
There are two places where meta->release_regno needs to keep
regno for later releasing the reference. Also, 'cur_aux(env)->arg_prog = regno'
is also keeping regno for later fixup. Since stack arguments don't have a valid
register number (regno is negative), these three cases are rejected for now
if the argument is on the stack.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045104.2391543-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
BPF_REG_PARAMS (r11) is used for stack argument accesses and
the following are only insns with r11 presence:
- load incoming stack arg
- store register to outgoing stack arg
- store immediate to outgoing stack arg
The detailed insn format can be found in is_stack_arg_ldx/st/stx()
helpers. After this patch, stack arg ldx/st/stx insns become valid
for kernel and these insns can be properly checked by verifier.
The LLVM compiler [1] implemented the above BPF_REG_PARAMS insns.
[1] https://github.com/llvm/llvm-project/pull/189060
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045059.2391192-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add bpf_jit_supports_stack_args() as a weak function defaulting to
false. Architectures that implement JIT support for stack arguments
override it to return true.
Reject BPF functions with more than 5 parameters at verification
time if the architecture does not support stack arguments.
Acked-by: Puranjay Mohan <puranjay@kernel.org>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045054.2390945-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
BPF_REG_PARAMS (R11) is at index MAX_BPF_REG, which is beyond the
register tracking arrays in const_fold.c and liveness.c. Handle it
explicitly to avoid out-of-bounds accesses.
Extend the arg tracking dataflow to cover stack arg slots. Otherwise,
pointers passed through stack args are invisible to liveness, causing
the pointed-to stack slots to be incorrectly poisoned.
Extend the at_out tracking array to MAX_AT_TRACK_REGS (registers
plus stack arg slots) so that outgoing stack arg stores are tracked
alongside registers. Add a separate at_stack_arg_entry array in
compute_subprog_args(), passed to arg_track_xfer(), to restore
FP-derived values on incoming stack arg reads.
Extend record_call_access() to check stack arg slots for FP-derived
pointers at kfunc call sites, reusing the record_arg_access() helper
extracted in the previous patch. Pass stack arg state from caller to
callee in analyze_subprog() so that callees can track pointers received
through stack args, hence avoid poisoning.
Skip stack arg instructions in record_load_store_access(). Stack arg
STX uses dst_reg=BPF_REG_PARAMS (index 11), but at[11] is repurposed
to track the value stored in stack arg slot 0. Without the skip, if a
prior stack arg STX stored an FP-derived pointer (e.g., fp-64) into
slot 0, a subsequent stack arg STX would read that FP-derived value as
the base pointer and spuriously mark a regular stack slot (e.g., fp-72
from -64 + -8) as accessed in the liveness bitmap.
Extend arg_track_log() to log state transitions for outgoing stack arg
slots at indices MAX_BPF_REG through MAX_AT_TRACK_REGS-1. Without this,
changes to at_out[11..17] caused by stack arg store instructions are
silently omitted from BPF_LOG_LEVEL2 output. For example, when a
caller passes fp-64 through a stack argument:
subprog#0:
10: (bf) r6 = r10
11: (07) r6 += -64
12: (7b) *(u64 *)(r11 -8) = r6
sa0: none -> fp0-64
13: (85) call pc+5
Without the fix, the "sa0: none -> fp0-64" transition at insn 12
would not appear.
Extend print_subprog_arg_access() to include stack arg slots in the
per-instruction FP-derived state dump. For example:
subprog#0:
12: (7b) *(u64 *)(r11 - 8) = r6 // r6=fp0-64
13: (85) call pc+5 // r6=fp0-64 sa0=fp0-64
Without the fix, the "sa0=fp0-64" annotation at insn 13 would not
appear, making it harder to debug liveness analysis for programs
that pass FP-derived pointers through stack arguments.
Extend has_fp_args() to also check stack arg slots for FP-derived
pointers, so that callees receiving pointers only through stack args
are still recursively analyzed.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045043.2389049-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Replace "frame != ARG_NONE" with arg_is_fp() in has_fp_args().
The function's purpose is to check whether any argument is derived
from a frame pointer, which is exactly what arg_is_fp() tests
(frame >= 0 || frame == ARG_IMPRECISE). Using the dedicated
predicate is clearer and more consistent with the rest of the file.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045035.2388671-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Extract the per-argument FP-derived pointer handling from
record_call_access() into a new record_arg_access() helper.
The existing loop body — checking arg_is_fp, querying stack access
bytes, and calling record_stack_access/record_imprecise — will be
reused for stack argument slots in the next patch. Factoring it out
now avoids duplicating the logic.
No functional change.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045030.2388067-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Extend the precision marking and backtracking infrastructure to
support stack argument slots (r11-based accesses). Without this,
precision demands for scalar values passed through stack arguments
are silently dropped, which could allow the verifier to incorrectly
prune states with different constant values in stack arg slots.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045025.2387526-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Move stack slot index (spi) and frame number out of the flags field
in bpf_jmp_history_entry into dedicated bitfields. This simplifies
the encoding and makes room for new flags.
Previously, spi and frame were packed into the lower 9 bits of the
12-bit flags field (3 bits frame + 6 bits spi), with INSN_F_STACK_ACCESS
at BIT(9) and INSN_F_DST/SRC_REG_STACK at BIT(10)/BIT(11).
But this has no room for an INSN_F_* flag for stack arguments.
To resolve this issue, bpf_jmp_history_entry field idx is narrowed to
20 bits (sufficient for insn indices up to 1M), and the freed bits hold
spi (6 bits) and frame (3 bits) as dedicated struct fields. The flags
enum is simplified accordingly:
INSN_F_STACK_ACCESS -> BIT(0)
INSN_F_DST_REG_STACK -> BIT(1)
INSN_F_SRC_REG_STACK -> BIT(2)
which allows more room for additional INSN_F_* flags.
bpf_push_jmp_history() now takes explicit spi and frame parameters
instead of encoding them into flags. The insn_stack_access_flags(),
insn_stack_access_spi(), and insn_stack_access_frameno() helpers are
removed.
No functional change.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045020.2385962-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently BPF functions (subprogs) are limited to 5 register arguments.
With [1], the compiler can emit code that passes additional arguments
via a dedicated stack area through bpf register BPF_REG_PARAMS (r11),
introduced in an earlier patch ([2]).
The compiler uses positive r11 offsets for incoming (callee-side) args
and negative r11 offsets for outgoing (caller-side) args, following the
x86_64/arm64 calling convention direction. There is an 8-byte gap at
offset 0 separating two regions:
Incoming (callee reads): r11+8 (arg6), r11+16 (arg7), ...
Outgoing (caller writes): r11-8 (arg6), r11-16 (arg7), ...
The following is an example to show how stack arguments are saved
and transferred between caller and callee:
int foo(int a1, int a2, int a3, int a4, int a5, int a6, int a7) {
...
bar(a1, a2, a3, a4, a5, a6, a7, a8);
...
}
Caller (foo) Callee (bar)
============ ============
Incoming (positive offsets): Incoming (positive offsets):
r11+8: [incoming arg 6] r11+8: [incoming arg 6] <-+
r11+16: [incoming arg 7] r11+16: [incoming arg 7] <-|+
r11+24: [incoming arg 8] <-||+
Outgoing (negative offsets): |||
r11-8: [outgoing arg 6 to bar] -------->-------------------------+||
r11-16: [outgoing arg 7 to bar] -------->--------------------------+|
r11-24: [outgoing arg 8 to bar] -------->---------------------------+
If the bpf function has more than one call:
int foo(int a1, int a2, int a3, int a4, int a5, int a6, int a7) {
...
bar1(a1, a2, a3, a4, a5, a6, a7, a8);
...
bar2(a1, a2, a3, a4, a5, a6, a7, a8, a9);
...
}
Caller (foo) Callee (bar2)
============ ==============
Incoming (positive offsets): Incoming (positive offsets):
r11+8: [incoming arg 6] r11+8: [incoming arg 6] <+
r11+16: [incoming arg 7] r11+16: [incoming arg 7] <|+
r11+24: [incoming arg 8] <||+
Outgoing for bar2 (negative offsets): r11+32: [incoming arg 9] <|||+
r11-8: [outgoing arg 6] ---->----------->-------------------------+|||
r11-16: [outgoing arg 7] ---->----------->--------------------------+||
r11-24: [outgoing arg 8] ---->----------->---------------------------+|
r11-32: [outgoing arg 9] ---->----------->----------------------------+
The verifier tracks outgoing stack arguments in stack_arg_regs[] and
out_stack_arg_cnt in bpf_func_state, separately from the regular
r10 stack. The callee does not copy incoming args — it reads them
directly from the caller's outgoing slots at positive r11 offsets.
Similar to stacksafe(), introduce stack_arg_safe() to do pruning
check.
Outgoing stack arg slots are invalidated when the callee returns
(e.g. in prepare_func_exit), not at call time. This allows the callee to
read incoming args from the caller's outgoing slots during
verification. The following are a few examples.
Example 1:
*(u64 *)(r11 - 8) = r6;
*(u64 *)(r11 - 16) = r7;
call bar1; // arg6 = r6, arg7 = r7
call bar2; // expected with 2 stack arguments, failed
Example 2:
To fix the Example 1:
*(u64 *)(r11 - 8) = r6;
*(u64 *)(r11 - 16) = r7;
call bar1; // arg6 = r6, arg7 = r7
*(u64 *)(r11 - 8) = r8;
*(u64 *)(r11 - 16) = r9;
call bar2; // arg6 = r8, arg7 = r9
Example 3:
The compiler can hoist the shared stack arg stores above the branch:
*(u64 *)(r11 - 16) = r7;
if cond goto else;
*(u64 *)(r11 - 8) = r8;
call bar1; // arg6 = r8, arg7 = r7
goto end;
else:
*(u64 *)(r11 - 8) = r9;
call bar2; // arg6 = r9, arg7 = r7
end:
Example 4:
Within a loop:
loop:
*(u64 *)(r11 - 8) = r6; // arg6, before loop
call bar; // reuses arg6 each iteration
if ... goto loop;
A separate max_out_stack_arg_cnt field in bpf_subprog_info tracks
the deepest outgoing slot actually written. This intends to
reject programs that write to slots beyond what any callee expects.
It is necessary for JIT.
Similar to typical compiler generated code, enforce the following
orderings:
- all stack arg reads must be ahead of any stack arg write
- all stack arg reads must be before any bpf func, kfunc and helpers
This is needed as JIT may emit 'mov' insns for read/write with
the same register and bpf function, kfunc and helper will invalidate
all arguments immediately after the call.
Callback functions with stack arguments need kernel setup parameter
types (including stack parameters) properly and then callback function
can retrieve such information for verification purpose.
Global subprogs and freplace with >5 args are not yet supported.
[1] https://github.com/llvm/llvm-project/pull/189060
[2] https://lore.kernel.org/bpf/20260423033506.2542005-1-yonghong.song@linux.dev/
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045015.2385013-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Move the "sub->arg_cnt = nargs" assignment to immediately after
nargs is computed from btf_type_vlen(), instead of at the end of
btf_prepare_func_args().
btf_prepare_func_args() can return -EINVAL early in several cases,
e.g. when a static function has some non-int/enum arguments.
Since -EINVAL from btf_prepare_func_args() does not immediately
reject verification, arg_cnt remains zero after the early return.
This causes later stack argument based load/store insns to
incorrectly assume the function has no arguments.
Setting arg_cnt right after nargs ensures it is available regardless
of which path btf_prepare_func_args() takes.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045010.2384635-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add three static inline helper functions — is_stack_arg_ldx(),
is_stack_arg_st(), and is_stack_arg_stx() — that identify r11-based
(BPF_REG_PARAMS) instructions used for stack argument passing. These
helpers encapsulate the detailed encoding requirements (operand size,
register, offset alignment and sign) and hide raw BPF_REG_PARAMS usage
from the verifier, making call sites more readable and explicit.
A later patch ("bpf: Enable r11 based insns") will wire these helpers
into the verifier. Until then, check_and_resolve_insns() rejects any
r11-based registers.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513045005.2383881-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Convert the bpf_get_spilled_reg() macro to a static inline function
for better type safety and readability. This also simplifies the macro
definition in preparation for upcoming stack argument support which
will introduce additional macros.
No functional change.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260513044954.2382693-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Leon Hwang says:
====================
bpf: Extend BPF syscall with common attributes support
This patch series builds upon the discussion in
"[PATCH bpf-next v4 0/4] bpf: Improve error reporting for freplace attachment failure" [1].
This patch series introduces support for *common attributes* in the BPF
syscall, providing a unified mechanism for passing shared metadata across
all BPF commands, initially used by BPF_PROG_LOAD, BPF_BTF_LOAD, and
BPF_MAP_CREATE.
The initial set of common attributes includes:
1. 'log_buf': User-provided buffer for storing log output.
2. 'log_size': Size of the provided log buffer.
3. 'log_level': Verbosity level for logging.
4. 'log_true_size': Actual log size reported by kernel.
With this extension, the BPF syscall will be able to return meaningful
error messages (e.g., map creation failures), improving debuggability
and user experience.
Links:
[1] https://lore.kernel.org/bpf/20250224153352.64689-1-leon.hwang@linux.dev/
Changes:
v13 -> v14:
* Replace __u64 with __aligned_u64 in struct bpf_common_attr in patch #1
(per bot+bpf-ci).
* Add a single line comment for preserving original error in patch #6
(per bot+bpf-ci and Alexei).
* Drop unused label in patch #6 (per bot+bpf-ci).
* v13: https://lore.kernel.org/bpf/20260511152817.89191-1-leon.hwang@linux.dev/
v12 -> v13:
* Rebase on bpf-next tree to resolve code conflict.
* Report log_true_size on success path in patch #6.
* Check size instead of common->log_buf in patch #6 (per bot+bpf-ci).
* v12: https://lore.kernel.org/bpf/20260420141804.27179-1-leon.hwang@linux.dev/
v11 -> v12:
* Drop "log_" prefix in struct bpf_log_attr in patch #3.
* Drop "log_" prefix in struct bpf_log_opts in patch #7.
* Copy log_true_size using copy_to_bpfptr_offset() in patch #3 (per Alexei).
* v11: https://lore.kernel.org/bpf/20260216150445.68278-1-leon.hwang@linux.dev/
v10 -> v11:
* Collect Acked-by from Andrii, thanks.
* Validate whether log_buf, log_size, and log_level are valid by reusing
bpf_verifier_log_attr_valid() in patch #4 (per Andrii).
* v10: https://lore.kernel.org/bpf/20260211151115.78013-1-leon.hwang@linux.dev/
v9 -> v10:
* Collect Acked-by from Andrii, thanks.
* Address comments from Andrii:
* Drop log NULL check in bpf_log_attr_finalize().
* Return -EFAULT early in bpf_log_attr_finalize().
* Validate whether log_buf, log_size, and log_level are set.
* Keep log_buf, log_size, log_level, and user-pointer log_true_size in struct
bpf_log_attr.
* Make prog_load and btf_load work with the new struct bpf_log_attr.
* Add comment to log_true_size of struct bpf_log_opts in libbpf.
* Address comment from Alexei:
* Avoid using BPF_LOG_FIXED as log_level in tests.
* v9: https://lore.kernel.org/bpf/20260202144046.30651-1-leon.hwang@linux.dev/
v8 -> v9:
* Rework reporting 'log_true_size' for prog_load, btf_load, and map_create to
simplify struct bpf_log_attr (per Alexei).
* v8: https://lore.kernel.org/bpf/20260126151409.52072-1-leon.hwang@linux.dev/
v7 -> v8:
* Return 0 when fd < 0 and errno != EFAULT in probe_sys_bpf_ext(), then simplify
probe_bpf_syscall_common_attrs() (per Alexei and Andrii).
* v7: https://lore.kernel.org/bpf/20260123032445.125259-1-leon.hwang@linux.dev/
v6 -> v7:
* Return -errno when fd < 0 and errno != EFAULT in probe_sys_bpf_ext().
* Convert return value of probe_sys_bpf_ext() to bool in
probe_bpf_syscall_common_attrs().
* Address comments from Andrii:
* Drop the comment, and handle fd >= 0 case explicitly in
probe_sys_bpf_ext().
* Return an error when fd >= 0 in probe_sys_bpf_ext().
* v6: https://lore.kernel.org/bpf/20260120152424.40766-1-leon.hwang@linux.dev/
v5 -> v6:
* Address comments from Andrii:
* Update some variables' name.
* Drop unnecessary 'close(fd)' in libbpf.
* Rename FEAT_EXTENDED_SYSCALL to FEAT_BPF_SYSCALL_COMMON_ATTRS with
updated description in libbpf.
* Use EINVAL instead of EUSERS, as EUSERS is not used in bpf yet.
* Rename struct bpf_syscall_common_attr_opts to bpf_log_opts in libbpf.
* Add 'OPTS_SET(log_opts, log_true_size, 0);' in libbpf's 'bpf_map_create()'.
* v5: https://lore.kernel.org/bpf/20260112145616.44195-1-leon.hwang@linux.dev/
v4 -> v5:
* Rework reporting 'log_true_size' for prog_load, btf_load, and map_create
(per Alexei).
* v4: https://lore.kernel.org/bpf/20260106172018.57757-1-leon.hwang@linux.dev/
RFC v3 -> v4:
* Drop RFC.
* Address comments from Andrii:
* Add parentheses in 'sys_bpf_ext()'.
* Avoid creating new fd in 'probe_sys_bpf_ext()'.
* Add a new struct to wrap log fields in libbpf.
* Address comments from Alexei:
* Do not skip writing to user space when log_true_size is zero.
* Do not use 'bool' arguments.
* Drop the adding WARN_ON_ONCE()'s.
* v3: https://lore.kernel.org/bpf/20251002154841.99348-1-leon.hwang@linux.dev/
RFC v2 -> RFC v3:
* Rename probe_sys_bpf_extended to probe_sys_bpf_ext.
* Refactor reporting 'log_true_size' for prog_load.
* Refactor reporting 'btf_log_true_size' for btf_load.
* Add warnings for internal bugs in map_create.
* Check log_true_size in test cases.
* Address comment from Alexei:
* Change kvzalloc/kvfree to kzalloc/kfree.
* Address comments from Andrii:
* Move BPF_COMMON_ATTRS to 'enum bpf_cmd' alongside brief comment.
* Add bpf_check_uarg_tail_zero() for extra checks.
* Rename sys_bpf_extended to sys_bpf_ext.
* Rename sys_bpf_fd_extended to sys_bpf_ext_fd.
* Probe the new feature using NULL and -EFAULT.
* Move probe_sys_bpf_ext to libbpf_internal.h and drop LIBBPF_API.
* Return -EUSERS when log attrs are conflict between bpf_attr and
bpf_common_attr.
* Avoid touching bpf_vlog_init().
* Update the reason messages in map_create.
* Finalize the log using __cleanup().
* Report log size to users.
* Change type of log_buf from '__u64' to 'const char *' and cast type
using ptr_to_u64() in bpf_map_create().
* Do not return -EOPNOTSUPP when kernel doesn't support this feature
in bpf_map_create().
* Add log_level support for map creation for consistency.
* Address comment from Eduard:
* Use common_attrs->log_level instead of BPF_LOG_FIXED.
* v2: https://lore.kernel.org/bpf/20250911163328.93490-1-leon.hwang@linux.dev/
RFC v1 -> RFC v2:
* Fix build error reported by test bot.
* Address comments from Alexei:
* Drop new uapi for freplace.
* Add common attributes support for prog_load and btf_load.
* Add common attributes support for map_create.
* v1: https://lore.kernel.org/bpf/20250728142346.95681-1-leon.hwang@linux.dev/
====================
Link: https://patch.msgid.link/20260512153157.28382-1-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
With the previous commit adding common attribute support for
BPF_MAP_CREATE, users can now retrieve detailed error messages when map
creation fails via the log_buf field.
Introduce struct bpf_log_opts with the following fields:
log_buf, log_size, log_level, and log_true_size.
Extend bpf_map_create_opts with a new field log_opts, allowing users to
capture and inspect log messages on map creation failures.
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-8-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Many BPF_MAP_CREATE validation failures currently return -EINVAL without
any explanation to userspace.
Plumb common syscall log attributes into map_create(), create a verifier
log from bpf_common_attr::log_buf/log_size/log_level, and report
map-creation failure reasons through that buffer.
This improves debuggability by allowing userspace to inspect why map
creation failed and read back log_true_size from common attributes.
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-7-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
BPF_BTF_LOAD can now take log parameters from both union bpf_attr and
struct bpf_common_attr, with the same merge rules as BPF_PROG_LOAD:
- if both sides provide a complete log tuple (buf/size/level) and they
match, use it;
- if only one side provides log parameters, use that one;
- if both sides provide complete tuples but they differ, return -EINVAL.
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-6-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
BPF_PROG_LOAD can now take log parameters from both union bpf_attr and
struct bpf_common_attr. The merge rules are:
- if both sides provide a complete log tuple (buf/size/level) and they
match, use it;
- if only one side provides log parameters, use that one;
- if both sides provide complete tuples but they differ, return -EINVAL.
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-5-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The next commit will add support for reporting logs via extended common
attributes, including 'log_true_size'.
To prepare for that, refactor the 'log_true_size' reporting logic by
introducing a new struct bpf_log_attr to encapsulate log-related behavior:
* bpf_log_attr_init(): initialize log fields, which will support
extended common attributes in the next commit.
* bpf_log_attr_finalize(): handle log finalization and write back
'log_true_size' to userspace.
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-4-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
To support the extended BPF syscall introduced in the previous commit,
introduce the following internal APIs:
* 'sys_bpf_ext()'
* 'sys_bpf_ext_fd()'
They wrap the raw 'syscall()' interface to support passing extended
attributes.
* 'probe_sys_bpf_ext()'
Check whether current kernel supports the BPF syscall common attributes.
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-3-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add generic BPF syscall support for passing common attributes.
The initial set of common attributes includes:
1. 'log_buf': User-provided buffer for storing logs.
2. 'log_size': Size of the log buffer.
3. 'log_level': Log verbosity level.
4. 'log_true_size': Actual log size reported by kernel.
The common-attribute pointer and its size are passed as the 4th and 5th
syscall arguments. A new command bit, 'BPF_COMMON_ATTRS' ('1 << 16'),
indicates that common attributes are supplied.
This commit adds syscall and uapi plumbing. Command-specific handling is
added in follow-up patches.
Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
Link: https://lore.kernel.org/r/20260512153157.28382-2-leon.hwang@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Puranjay Mohan says:
====================
selftests/bpf: Add XDP load-balancer benchmark
Changelog:
RFC: https://lore.kernel.org/all/20260420111726.2118636-1-puranjay@kernel.org/
Changes in v1:
- Replace bpf_get_cpu_time_counter() with bpf_ktime_get_ns()
- Replace bpf_repeat() with plain for loop and may_goto
- Refactor collect_measurements() to reuse bench_force_done()
- Remove histogram, verbose calibration output, and per-scenario status prints
- Trim run script table to p50/stddev/p99
- Set env.quiet when --machine-readable is passed
- Add || true to run script benchmark invocation for set -e safety
- Add bpf-nop benchmark as timing overhead baseline (patch 3)
- Use named struct for LRU inner map to fix build on older toolchains
This series adds an XDP load-balancer benchmark (based on Katran) to the BPF
selftest bench framework.
Motivation
----------
Existing BPF bench tests measure individual operations (map lookups,
kprobes, ring buffers) in isolation. Production BPF programs combine
parsing, map lookups, branching, and packet rewriting in a single call
chain. The performance characteristics of such programs depend on the
interaction of these operations -- register pressure, spills, inlining
decisions, branch layout -- which isolated micro-benchmarks do not
capture.
This benchmark implements a simplified L4 load-balancer modeled after
katran [1]. The BPF program reproduces katran's core datapath:
L3/L4 parsing -> VIP hash lookup -> per-CPU LRU connection table
with consistent-hash fallback -> real server selection -> per-VIP
and per-real stats -> IPIP/IP6IP6 encapsulation
The BPF code exercises hash maps, array-of-maps (per-CPU LRU),
percpu arrays, jhash, bpf_xdp_adjust_head(), bpf_ktime_get_ns(),
and bpf_get_smp_processor_id() in a single pipeline.
This is intended as the first in a series of BPF workload benchmarks
covering other use cases (sched_ext, etc.).
Design
------
A userspace loop calling bpf_prog_test_run_opts(repeat=1) would
measure syscall overhead, not BPF program cost -- the ~4 ns early-exit
paths would be buried under kernel entry/exit. Using repeat=N is
also unsuitable: the kernel re-runs the same packet without resetting
state between iterations, so the second iteration of an encap scenario
would process an already-encapsulated packet.
Instead, timing is measured inside the BPF program using
bpf_ktime_get_ns(). BENCH_BPF_LOOP() brackets N iterations with
timestamp reads using a plain for loop with may_goto, runs a
caller-supplied reset block between iterations to undo side effects
(e.g. strip encapsulation), and records the elapsed time per batch.
One extra untimed iteration runs afterward for output validation.
Auto-calibration picks a batch size targeting ~10 ms per invocation.
A proportionality sanity check verifies that 2N iterations take ~2x
as long as N.
24 scenarios cover the code-path matrix:
- Protocol: TCP, UDP
- Address family: IPv4, IPv6, cross-AF (IPv4-in-IPv6)
- LRU state: hit, miss (16M flow space), diverse (4K flows), cold
- Consistent-hash: direct (LRU bypass)
- TCP flags: SYN (skip LRU, force CH), RST (skip LRU insert)
- Early exits: unknown VIP, non-IP, ICMP, fragments, IP options
Each scenario validates correctness before benchmarking by comparing
the output packet byte-for-byte against a pre-built expected packet
and checking BPF map counters.
Sample single-scenario output:
$ sudo ./bench xdp-lb --scenario tcp-v4-lru-hit
Setting up benchmark 'xdp-lb'...
Benchmark 'xdp-lb' started.
tcp-v4-lru-hit: median 74.51 ns/op, stddev 0.11, p99 74.81 (202 samples)
Sample run script output:
$ ./benchs/run_bench_xdp_lb.sh
XDP load-balancer benchmark
===========================
+----------------------------------+----------+---------+----------+
| Single-flow baseline | p50 | stddev | p99 |
+----------------------------------+----------+---------+----------+
| tcp-v4-lru-hit | 74.30 | 0.08 | 74.48 |
| tcp-v4-ch | 101.73 | 0.11 | 102.01 |
| tcp-v6-lru-hit | 76.77 | 0.14 | 77.04 |
| tcp-v6-ch | 121.40 | 0.10 | 121.65 |
| udp-v4-lru-hit | 107.42 | 0.22 | 107.90 |
| udp-v6-lru-hit | 110.21 | 0.12 | 110.45 |
| tcp-v4v6-lru-hit | 74.82 | 0.35 | 75.43 |
+----------------------------------+----------+---------+----------+
| Diverse flows (4K src addrs) | p50 | stddev | p99 |
+----------------------------------+----------+---------+----------+
| tcp-v4-lru-diverse | 86.63 | 0.37 | 89.04 |
| tcp-v4-ch-diverse | 104.09 | 0.19 | 105.67 |
| tcp-v6-lru-diverse | 89.34 | 0.42 | 90.70 |
| tcp-v6-ch-diverse | 122.20 | 0.21 | 123.78 |
| udp-v4-lru-diverse | 119.37 | 0.58 | 123.10 |
+----------------------------------+----------+---------+----------+
| TCP flags | p50 | stddev | p99 |
+----------------------------------+----------+---------+----------+
| tcp-v4-syn | 165.52 | 15.68 | 198.34 |
| tcp-v4-rst-miss | 161.34 | 2.69 | 172.64 |
+----------------------------------+----------+---------+----------+
| LRU stress | p50 | stddev | p99 |
+----------------------------------+----------+---------+----------+
| tcp-v4-lru-miss | 440.39 | 35.75 | 550.62 |
| udp-v4-lru-miss | 571.88 | 57.38 | 680.61 |
| tcp-v4-lru-warmup | 317.75 | 9.55 | 356.20 |
+----------------------------------+----------+---------+----------+
| Early exits | p50 | stddev | p99 |
+----------------------------------+----------+---------+----------+
| pass-v4-no-vip | 18.26 | 0.13 | 18.66 |
| pass-v6-no-vip | 19.08 | 0.01 | 19.10 |
| pass-v4-icmp | 6.81 | 0.02 | 6.86 |
| pass-non-ip | 5.71 | 0.03 | 5.76 |
| drop-v4-frag | 6.09 | 0.01 | 6.10 |
| drop-v4-options | 5.88 | 0.00 | 5.89 |
| drop-v6-frag | 6.00 | 0.03 | 6.04 |
+----------------------------------+----------+---------+----------+
Patches
-------
Patch 1 adds bench_force_done() to the bench framework so benchmarks
can signal early completion when enough samples have been collected.
Patch 2 adds the shared BPF batch-timing library (BPF-side timing
arrays, BENCH_BPF_LOOP macro, userspace statistics and calibration).
Patch 3 adds a bpf-nop benchmark as a timing overhead baseline and
usage example for the timing library.
Patch 4 adds the common header shared between the BPF program and
userspace (flow_key, vip_definition, real_definition, encap helpers).
Patch 5 adds the XDP load-balancer BPF program.
Patch 6 adds the userspace benchmark driver with 24 scenarios,
packet construction, validation, and bench framework integration.
Patch 7 adds the run script for running all scenarios.
[1] https://github.com/facebookincubator/katran
====================
Link: https://patch.msgid.link/20260427232313.1582588-1-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Wire up the userspace side of the XDP load-balancer benchmark.
24 scenarios cover the full code-path matrix: TCP/UDP, IPv4/IPv6,
cross-AF encap, LRU hit/miss/diverse/cold, consistent-hash bypass,
SYN/RST flag handling, and early exits (unknown VIP, non-IP, ICMP,
fragments, IP options).
Before benchmarking each scenario validates correctness: the output
packet is compared byte-for-byte against a pre-built expected packet
and BPF map counters are checked against the expected values.
Usage:
sudo ./bench -a -w3 -p1 xdp-lb --scenario tcp-v4-lru-hit
sudo ./bench xdp-lb --list-scenarios
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-7-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add the BPF datapath for the XDP load-balancer benchmark, a
simplified L4 load-balancer inspired by katran.
The pipeline: L3/L4 parse -> VIP lookup -> per-CPU LRU connection
table or consistent-hash fallback -> real server lookup -> per-VIP
and per-real stats -> IPIP/IP6IP6 encapsulation. TCP SYN forces
the consistent-hash path (skipping LRU); TCP RST skips LRU insert
to avoid polluting the table.
process_packet() is marked __noinline so that the BENCH_BPF_LOOP
reset block (which strips encapsulation) operates on valid packet
pointers after bpf_xdp_adjust_head().
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-6-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add the shared header for the XDP load-balancer benchmark. This
defines the data structures used by both the BPF program and
userspace: flow_key, vip_definition, real_definition, and the
stats/control structures.
Also provides the encapsulation source-address helpers shared
between the BPF datapath (for encap) and userspace (for building
expected output packets used in validation).
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-5-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a minimal benchmark that measures the overhead of the batch-timing
infrastructure itself. The BPF program runs an empty BENCH_BPF_LOOP body
(~1.5-2 ns/op), establishing the floor cost that all timing-library
benchmarks include.
[root@virtme-ng tools/testing/selftests/bpf]# sudo ./bench -a -p8 bpf-nop
Setting up benchmark 'bpf-nop'...
Benchmark 'bpf-nop' started.
bpf-nop: median 1.82 ns/op, stddev 0.01, p99 1.86 (1754 samples)
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-4-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a reusable timing library for BPF benchmarks that need to measure
BPF program execution time.
The BPF side (progs/bench_bpf_timing.bpf.h) provides per-CPU sample
arrays and BENCH_BPF_LOOP(), a macro that brackets batch_iters
iterations with bpf_ktime_get_ns() reads and records the elapsed time.
One extra untimed iteration runs afterward for output validation.
The userspace side (benchs/bench_bpf_timing.c) collects samples from
the skeleton BSS, computes percentile statistics, and auto-calibrates
batch_iters to target ~10 ms per batch.
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-3-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The bench framework waits for duration_sec to elapse before collecting
results. Benchmarks that know exactly how many samples they need can
call bench_force_done() to signal completion early, avoiding wasted
wall-clock time.
Also refactor collect_measurements() to reuse bench_force_done()
instead of open-coding the same mutex/cond_signal sequence.
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Link: https://lore.kernel.org/r/20260427232313.1582588-2-puranjay@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Pull EDAC fix from Borislav Petkov:
- Fix a string leak in the versalnet driver
* tag 'edac_urgent_for_v7.1_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/versalnet: Fix device name memory leak
The DATA-packet handler in rxrpc_input_call_event() and the RESPONSE
handler in rxrpc_verify_response() copy the skb to a linear one before
calling into the security ops only when skb_cloned() is true. An skb
that is not cloned but still carries externally-owned paged fragments
(e.g. SKBFL_SHARED_FRAG set by splice() into a UDP socket via
__ip_append_data, or a chained skb_has_frag_list()) falls through to
the in-place decryption path, which binds the frag pages directly into
the AEAD/skcipher SGL via skb_to_sgvec().
Extend the gate to also unshare when skb_has_frag_list() or
skb_has_shared_frag() is true. This catches the splice-loopback vector
and other externally-shared frag sources while preserving the
zero-copy fast path for skbs whose frags are kernel-private (e.g. NIC
page_pool RX, GRO). The OOM/trace handling already in place is reused.
Fixes: d0d5c0cd1e ("rxrpc: Use skb_unshare() rather than skb_cow_data()")
Cc: stable@vger.kernel.org
Signed-off-by: Hyunwoo Kim <imv4bel@gmail.com>
Reviewed-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull clk driver fixes from Stephen Boyd:
- Mark the DDR bus clk critical in the SpaceMiT driver so that
boot doesn't fail
- Fix boot on Mobile EyeQ by creating the auxiliary device for
the ethernet PHY
- Plug an OF node leak in Rockchip rk808 clk driver
* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
clk: rk808: fix OF node reference imbalance
MAINTAINERS: add myself as a reviewer for the clk subsystem
reset: eyeq: drop device_set_of_node_from_dev() done by parent
clk: eyeq: add EyeQ5 children auxiliary device for generic PHYs
clk: eyeq: use the auxiliary device creation helper
clk: spacemit: k3: mark top_dclk as CLK_IS_CRITICAL
BPF_MAP_TYPE_ARENA accepts BPF_PSEUDO_MAP_VALUE offsets at exactly
the end of the arena mapping (off == arena_size). The boundary check
in arena_map_direct_value_addr() uses `>` instead of `>=`, which
incorrectly allows a one-past-end pointer to be accepted.
Change the condition to `>=` to correctly reject offsets that fall
outside the valid arena user_vm range.
Fixes: 317460317a ("bpf: Introduce bpf_arena.")
Signed-off-by: Junyoung Jang <graypanda.inzag@gmail.com>
Reviewed-by: Emil Tsalapatis <emil@etsalapatis.com>
Link: https://lore.kernel.org/r/20260426172505.1947915-1-graypanda.inzag@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
XSKMAP entries are used as redirect targets for incoming XDP frames.
A TX-only AF_XDP socket lacks an Rx ring and cannot handle redirected
traffic, but xsk_map_update_elem() currently allows such sockets to
be inserted into the map.
Redirecting packets to such a socket on the veth generic-XDP path
causes a kernel crash in xsk_generic_rcv().
This became possible after xsk_is_setup_for_bpf_map() was removed from
the XSKMAP update path, which allowed bound TX-only sockets to be
inserted into the map.
Reject TX-only sockets during XSKMAP updates to avoid the crash.
They remain fully operational for pure Tx purposes outside XSKMAP.
Fixes: 968be23cea ("xsk: Fix possible segfault at xskmap entry insertion")
Reported-by: Juefei Pu <tomapufckgml@gmail.com>
Reported-by: Yuan Tan <yuantan098@gmail.com>
Reported-by: Xin Liu <bird@lzu.edu.cn>
Signed-off-by: Yifan Wu <yifanwucs@gmail.com>
Signed-off-by: Linpu Yu <linpu5433@gmail.com>
Reviewed-by: Jason Xing <kerneljasonxing@gmail.com>
Link: https://lore.kernel.org/r/20260508144344.694-1-linpu5433@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
