Re: [PATCH] inifiniband: ucaps: avoid format-security warning

[Jason Gunthorpe <jgg@nvidia.com>]

On Mon, Mar 17, 2025 at 01:50:00PM +0200, Leon Romanovsky wrote:
> On Fri, Mar 14, 2025 at 04:57:15PM +0100, Arnd Bergmann wrote:
> > From: Arnd Bergmann <arnd@arndb.de>
> > 
> > Passing a non-constant format string to dev_set_name causes a warning:
> > 
> > drivers/infiniband/core/ucaps.c:173:33: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
> >   173 |         ret = dev_set_name(&ucap->dev, ucap_names[type]);
> >       |                                        ^~~~~~~~~~~~~~~~
> > drivers/infiniband/core/ucaps.c:173:33: note: treat the string as an argument to avoid this
> >   173 |         ret = dev_set_name(&ucap->dev, ucap_names[type]);
> >       |                                        ^
> >       |                                        "%s",
> > 
> > Turn the name into thet %s argument as suggested by gcc.
> > 
> > Fixes: 61e51682816d ("RDMA/uverbs: Introduce UCAP (User CAPabilities) API")
> > Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> > ---
> >  drivers/infiniband/core/ucaps.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Why don't you change ucap_names[] declaration instead?
> 
> diff --git a/drivers/infiniband/core/ucaps.c b/drivers/infiniband/core/ucaps.c
> index 6853c6d078f9..90ac41bc0f07 100644
> --- a/drivers/infiniband/core/ucaps.c
> +++ b/drivers/infiniband/core/ucaps.c
> @@ -23,7 +23,7 @@ struct ib_ucap {
>         struct kref ref;
>  };
> 
> -static const char *ucap_names[RDMA_UCAP_MAX] = {
> +static const char *const ucap_names[RDMA_UCAP_MAX] = {
>         [RDMA_UCAP_MLX5_CTRL_LOCAL] = "mlx5_perm_ctrl_local",
>         [RDMA_UCAP_MLX5_CTRL_OTHER_VHCA] = "mlx5_perm_ctrl_other_vhca"
>  };

This should be done anyone, but it won't fix the warning. gcc is
worried that the format string could have an embedded "%XXX" or
something that would cause stack corruption. It is a security error to
push user controlled data into the format string. This isn't happening
here, but silencing gcc will need the %s.

Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>

For both

Jason