[PATCH rdma-next] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration

[PATCH rdma-next] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration

[Leon Romanovsky]

From: Leon Romanovsky <leonro@nvidia.com>

The call to rdma_uattrs_has_raw_cap() is placed in mlx5 fs.c file,
which is compiled without relation to CONFIG_INFINIBAND_USER_ACCESS.

Despite the check is used only in flows with CONFIG_INFINIBAND_USER_ACCESS=y|m,
the compilers generate the following error for CONFIG_INFINIBAND_USER_ACCESS=n
builds.

>> ERROR: modpost: "rdma_uattrs_has_raw_cap" [drivers/infiniband/hw/mlx5/mlx5_ib.ko] undefined!

Fixes: f458ccd2aa2c ("RDMA/uverbs: Check CAP_NET_RAW in user namespace for flow create")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202507080725.bh7xrhpg-lkp@intel.com/
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 include/rdma/ib_verbs.h | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index c263327a0205..ce4180c4db14 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -4841,15 +4841,19 @@ struct ib_ucontext *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile);
 
 #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)
 int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs);
+bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
 #else
 static inline int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs)
 {
 	return 0;
 }
+static inline bool
+rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs)
+{
+	return false;
+}
 #endif
 
-bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
-
 struct net_device *rdma_alloc_netdev(struct ib_device *device, u32 port_num,
 				     enum rdma_netdev_t type, const char *name,
 				     unsigned char name_assign_type,
-- 
2.50.0

RE: [PATCH rdma-next] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration

[Parav Pandit]

> From: Leon Romanovsky <leon@kernel.org>
> Sent: 08 July 2025 03:02 PM
> From: Leon Romanovsky <leonro@nvidia.com>
> 
> The call to rdma_uattrs_has_raw_cap() is placed in mlx5 fs.c file, which is
> compiled without relation to CONFIG_INFINIBAND_USER_ACCESS.
> 
> Despite the check is used only in flows with
> CONFIG_INFINIBAND_USER_ACCESS=y|m,
> the compilers generate the following error for
> CONFIG_INFINIBAND_USER_ACCESS=n builds.
>
 
> >> ERROR: modpost: "rdma_uattrs_has_raw_cap"
> [drivers/infiniband/hw/mlx5/mlx5_ib.ko] undefined!
> 
> Fixes: f458ccd2aa2c ("RDMA/uverbs: Check CAP_NET_RAW in user namespace
> for flow create")
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202507080725.bh7xrhpg-
> lkp@intel.com/
> Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
> ---
>  include/rdma/ib_verbs.h | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index
> c263327a0205..ce4180c4db14 100644
> --- a/include/rdma/ib_verbs.h
> +++ b/include/rdma/ib_verbs.h
> @@ -4841,15 +4841,19 @@ struct ib_ucontext
> *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile);
> 
>  #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)
>  int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs);
> +bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
>  #else
>  static inline int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs)
> {
>  	return 0;
>  }
> +static inline bool
> +rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs) {
> +	return false;
> +}
>  #endif
> 
> -bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
> -
>  struct net_device *rdma_alloc_netdev(struct ib_device *device, u32
> port_num,
>  				     enum rdma_netdev_t type, const char
> *name,
>  				     unsigned char name_assign_type,
> --
> 2.50.0

Reviewed-by: Parav Pandit <parav@nvidia.com>

Re: [PATCH rdma-next] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration

[yanjun.zhu]

On 7/8/25 2:31 AM, Leon Romanovsky wrote:
> From: Leon Romanovsky <leonro@nvidia.com>
> 
> The call to rdma_uattrs_has_raw_cap() is placed in mlx5 fs.c file,
> which is compiled without relation to CONFIG_INFINIBAND_USER_ACCESS.
> 
> Despite the check is used only in flows with CONFIG_INFINIBAND_USER_ACCESS=y|m,
> the compilers generate the following error for CONFIG_INFINIBAND_USER_ACCESS=n
> builds.
> 
>>> ERROR: modpost: "rdma_uattrs_has_raw_cap" [drivers/infiniband/hw/mlx5/mlx5_ib.ko] undefined!
> 
> Fixes: f458ccd2aa2c ("RDMA/uverbs: Check CAP_NET_RAW in user namespace for flow create")
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202507080725.bh7xrhpg-lkp@intel.com/
> Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
> ---
>   include/rdma/ib_verbs.h | 8 ++++++--
>   1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
> index c263327a0205..ce4180c4db14 100644
> --- a/include/rdma/ib_verbs.h
> +++ b/include/rdma/ib_verbs.h
> @@ -4841,15 +4841,19 @@ struct ib_ucontext *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile);
>   
>   #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)
>   int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs);
> +bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
>   #else
>   static inline int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs)
>   {
>   	return 0;
>   }
> +static inline bool
> +rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs)
> +{
> +	return false;
> +}
>   #endif

The function rdma_uattrs_has_raw_cap is implemented in the file 
rdma_core.c. This file is included in the build via the Makefile:

"
...
ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \
                rdma_core.o uverbs_std_types.o uverbs_ioctl.o \
                ...
"
As a result, rdma_core.o is compiled only when 
CONFIG_INFINIBAND_USER_ACCESS is enabled.

In other words, the implementation of the rdma_uattrs_has_raw_cap 
function is guarded by CONFIG_INFINIBAND_USER_ACCESS.

Given this, it is reasonable to move the declaration of 
rdma_uattrs_has_raw_cap into the scope of CONFIG_INFINIBAND_USER_ACCESS 
in ib_verbs.h.

Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>

Zhu Yanjun

>   
> -bool rdma_uattrs_has_raw_cap(const struct uverbs_attr_bundle *attrs);
> -
>   struct net_device *rdma_alloc_netdev(struct ib_device *device, u32 port_num,
>   				     enum rdma_netdev_t type, const char *name,
>   				     unsigned char name_assign_type,

Re: [PATCH rdma-next] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration

[Leon Romanovsky]

On Tue, 08 Jul 2025 12:31:39 +0300, Leon Romanovsky wrote:
> The call to rdma_uattrs_has_raw_cap() is placed in mlx5 fs.c file,
> which is compiled without relation to CONFIG_INFINIBAND_USER_ACCESS.
> 
> Despite the check is used only in flows with CONFIG_INFINIBAND_USER_ACCESS=y|m,
> the compilers generate the following error for CONFIG_INFINIBAND_USER_ACCESS=n
> builds.
> 
> [...]

Applied, thanks!

[1/1] RDMA/uverbs: Add empty rdma_uattrs_has_raw_cap() declaration
      https://git.kernel.org/rdma/rdma/c/98269398c02ab2

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>