[PATCH 1/2] bpf: Fix register_bpf_struct_ops() dummy

[PATCH 1/2] bpf: Fix register_bpf_struct_ops() dummy

[Geert Uytterhoeven]

If CONFIG_BPF_SYSCALL=y, but CONFIG_BPF_JIT=n:

    net/smc/smc_hs_bpf.c: In function ‘bpf_smc_hs_ctrl_init’:
    include/linux/bpf.h:2068:50: error: statement with no effect [-Werror=unused-value]
     2068 | #define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
	  |                                                  ^~~~~~~~~~~~~~~~
    net/smc/smc_hs_bpf.c:139:16: note: in expansion of macro ‘register_bpf_struct_ops’
      139 |         return register_bpf_struct_ops(&bpf_smc_hs_ctrl_ops, smc_hs_ctrl);
	  |                ^~~~~~~~~~~~~~~~~~~~~~~

As type is not a variable, but a variable type, this cannot be fixed by
just converting register_bpf_struct_ops() into a static inline function.
Hence fix this by introducing a static inline intermediate dummy.

Fixes: f6be98d19985411c ("bpf, net: switch to dynamic registration")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 include/linux/bpf.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 6498be4c44f8c275..bb69905c28a761e7 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -2065,7 +2065,11 @@ int bpf_struct_ops_desc_init(struct bpf_struct_ops_desc *st_ops_desc,
 void bpf_map_struct_ops_info_fill(struct bpf_map_info *info, struct bpf_map *map);
 void bpf_struct_ops_desc_release(struct bpf_struct_ops_desc *st_ops_desc);
 #else
-#define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
+static inline int __register_bpf_struct_ops(struct bpf_struct_ops *st_ops)
+{
+	return 0;
+}
+#define register_bpf_struct_ops(st_ops, type) __register_bpf_struct_ops(st_ops)
 static inline bool bpf_try_module_get(const void *data, struct module *owner)
 {
 	return try_module_get(owner);
-- 
2.43.0

Re: [PATCH 1/2] bpf: Fix register_bpf_struct_ops() dummy

[Martin KaFai Lau]

On 12/4/25 2:29 AM, Geert Uytterhoeven wrote:
> If CONFIG_BPF_SYSCALL=y, but CONFIG_BPF_JIT=n:
> 
>      net/smc/smc_hs_bpf.c: In function ‘bpf_smc_hs_ctrl_init’:
>      include/linux/bpf.h:2068:50: error: statement with no effect [-Werror=unused-value]
>       2068 | #define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
> 	  |                                                  ^~~~~~~~~~~~~~~~
>      net/smc/smc_hs_bpf.c:139:16: note: in expansion of macro ‘register_bpf_struct_ops’
>        139 |         return register_bpf_struct_ops(&bpf_smc_hs_ctrl_ops, smc_hs_ctrl);
> 	  |                ^~~~~~~~~~~~~~~~~~~~~~~
> 
> As type is not a variable, but a variable type, this cannot be fixed by
> just converting register_bpf_struct_ops() into a static inline function.
> Hence fix this by introducing a static inline intermediate dummy.
> 
> Fixes: f6be98d19985411c ("bpf, net: switch to dynamic registration")
> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> ---
>   include/linux/bpf.h | 6 +++++-
>   1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index 6498be4c44f8c275..bb69905c28a761e7 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -2065,7 +2065,11 @@ int bpf_struct_ops_desc_init(struct bpf_struct_ops_desc *st_ops_desc,
>   void bpf_map_struct_ops_info_fill(struct bpf_map_info *info, struct bpf_map *map);
>   void bpf_struct_ops_desc_release(struct bpf_struct_ops_desc *st_ops_desc);
>   #else
> -#define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
> +static inline int __register_bpf_struct_ops(struct bpf_struct_ops *st_ops)
> +{
> +	return 0;
> +}
> +#define register_bpf_struct_ops(st_ops, type) __register_bpf_struct_ops(st_ops)

Only patch 2 is needed. This empty register_bpf_struct_ops should be 
removed in the bpf-next tree as a cleanup.

[PATCH 2/2] net: smc: SMC_HS_CTRL_BPF should depend on BPF_JIT

[Geert Uytterhoeven]

If CONFIG_BPF_SYSCALL=y, but CONFIG_BPF_JIT=n:

    net/smc/smc_hs_bpf.c: In function ‘bpf_smc_hs_ctrl_init’:
    include/linux/bpf.h:2068:50: error: statement with no effect [-Werror=unused-value]
     2068 | #define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
	  |                                                  ^~~~~~~~~~~~~~~~
    net/smc/smc_hs_bpf.c:139:16: note: in expansion of macro ‘register_bpf_struct_ops’
      139 |         return register_bpf_struct_ops(&bpf_smc_hs_ctrl_ops, smc_hs_ctrl);
	  |                ^~~~~~~~~~~~~~~~~~~~~~~

While this compile error is caused by a bug in <linux/bpf.h>, none of
the code in net/smc/smc_hs_bpf.c becomes effective if CONFIG_BPF_JIT is
not enabled.  Hence add a dependency on BPF_JIT.

While at it, add the missing newline at the end of the file.

Fixes: 15f295f55656658e ("net/smc: bpf: Introduce generic hook for handshake flow")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 net/smc/Kconfig | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/smc/Kconfig b/net/smc/Kconfig
index 325addf83cc69f6c..277ef504bc26ef89 100644
--- a/net/smc/Kconfig
+++ b/net/smc/Kconfig
@@ -22,10 +22,10 @@ config SMC_DIAG
 
 config SMC_HS_CTRL_BPF
 	bool "Generic eBPF hook for SMC handshake flow"
-	depends on SMC && BPF_SYSCALL
+	depends on SMC && BPF_JIT && BPF_SYSCALL
 	default y
 	help
 	  SMC_HS_CTRL_BPF enables support to register generic eBPF hook for SMC
 	  handshake flow, which offer much greater flexibility in modifying the behavior
 	  of the SMC protocol stack compared to a complete kernel-based approach. Select
-	  this option if you want filtring the handshake process via eBPF programs.
\ No newline at end of file
+	  this option if you want filtring the handshake process via eBPF programs.
-- 
2.43.0

Re: [PATCH 2/2] net: smc: SMC_HS_CTRL_BPF should depend on BPF_JIT

[Martin KaFai Lau]

On 12/4/25 2:29 AM, Geert Uytterhoeven wrote:
> If CONFIG_BPF_SYSCALL=y, but CONFIG_BPF_JIT=n:
> 
>      net/smc/smc_hs_bpf.c: In function ‘bpf_smc_hs_ctrl_init’:
>      include/linux/bpf.h:2068:50: error: statement with no effect [-Werror=unused-value]
>       2068 | #define register_bpf_struct_ops(st_ops, type) ({ (void *)(st_ops); 0; })
> 	  |                                                  ^~~~~~~~~~~~~~~~
>      net/smc/smc_hs_bpf.c:139:16: note: in expansion of macro ‘register_bpf_struct_ops’
>        139 |         return register_bpf_struct_ops(&bpf_smc_hs_ctrl_ops, smc_hs_ctrl);
> 	  |                ^~~~~~~~~~~~~~~~~~~~~~~
> 
> While this compile error is caused by a bug in <linux/bpf.h>, none of
> the code in net/smc/smc_hs_bpf.c becomes effective if CONFIG_BPF_JIT is
> not enabled.  Hence add a dependency on BPF_JIT.
> 
> While at it, add the missing newline at the end of the file.
> 
> Fixes: 15f295f55656658e ("net/smc: bpf: Introduce generic hook for handshake flow")
> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> ---
>   net/smc/Kconfig | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/net/smc/Kconfig b/net/smc/Kconfig
> index 325addf83cc69f6c..277ef504bc26ef89 100644
> --- a/net/smc/Kconfig
> +++ b/net/smc/Kconfig
> @@ -22,10 +22,10 @@ config SMC_DIAG
>   
>   config SMC_HS_CTRL_BPF
>   	bool "Generic eBPF hook for SMC handshake flow"
> -	depends on SMC && BPF_SYSCALL
> +	depends on SMC && BPF_JIT && BPF_SYSCALL
>   	default y
>   	help
>   	  SMC_HS_CTRL_BPF enables support to register generic eBPF hook for SMC
>   	  handshake flow, which offer much greater flexibility in modifying the behavior
>   	  of the SMC protocol stack compared to a complete kernel-based approach. Select
> -	  this option if you want filtring the handshake process via eBPF programs.
> \ No newline at end of file
> +	  this option if you want filtring the handshake process via eBPF programs.

I have applied patch 2 to the bpf tree. Thanks.