linux-kernel - Re: [RFC PATCH 1/4] sysctl: add SYSCTL_ENTRY/SYSCTL_RANGE_ENTRY helpers for ctl

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <psot4oeauxi3yyj2w4ajm3tfgtcsvao4rhv5sgd5s6ymmjgojk@p3vrj3qluban>
Date: Thu, 22 Jan 2026 10:56:42 +0100
From: Joel Granados <joel.granados@...nel.org>
To: wen.yang@...ux.dev
Cc: linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 1/4] sysctl: add SYSCTL_ENTRY/SYSCTL_RANGE_ENTRY
 helpers for ctl_table init

On Wed, Jan 14, 2026 at 01:40:30AM +0800, wen.yang@...ux.dev wrote:
> From: Wen Yang <wen.yang@...ux.dev>
> 
> Introduce SYSCTL_ENTRY() and SYSCTL_RANGE_ENTRY() (built on __SYSCTL_ENTRY())
> to consolidate common struct ctl_table initializers in one place.
> This is a preparatory refactor in support of future tree-wide sysctl
> cleanups by making the how to build a ctl_table entry logic uniform.
> 
> Also add SYSCTL_NULL for explicitly passing a NULL .data pointer through
> the new macros, and convert a few in-tree users (kernel/sysctl.c) to
> validate the new helpers.
> 
> No functional change intended.
> 
> Selftest were also made:
> 
> [19:31:40] Starting KUnit Kernel (1/1)...
> [19:31:40] ============================================================
> Running tests with:
> $ .kunit/linux kunit.filter_glob=sysctl_test kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
> [19:31:40] ================ sysctl_test (10 subtests) =================
> [19:31:40] [PASSED] sysctl_test_api_dointvec_null_tbl_data
> [19:31:40] [PASSED] sysctl_test_api_dointvec_table_maxlen_unset
> [19:31:40] [PASSED] sysctl_test_api_dointvec_table_len_is_zero
> [19:31:40] [PASSED] sysctl_test_api_dointvec_table_read_but_position_set
> [19:31:40] [PASSED] sysctl_test_dointvec_read_happy_single_positive
> [19:31:40] [PASSED] sysctl_test_dointvec_read_happy_single_negative
> [19:31:40] [PASSED] sysctl_test_dointvec_write_happy_single_positive
> [19:31:40] [PASSED] sysctl_test_dointvec_write_happy_single_negative
> [19:31:40] [PASSED] sysctl_test_api_dointvec_write_single_less_int_min
> [19:31:40] [PASSED] sysctl_test_api_dointvec_write_single_greater_int_max
> [19:31:40] =================== [PASSED] sysctl_test ===================
> [19:31:40] ============================================================
> [19:31:40] Testing complete. Ran 10 tests: passed: 10
> [19:31:40] Elapsed time: 26.755s total, 0.002s configuring, 26.632s building, 0.102s running
> 
> Suggested-by: Joel Granados <joel.granados@...nel.org>
> Link: https://lore.kernel.org/all/7mevcgca7xsd5gckrap22byjinhrgqnelu4y7hzo2r5t2cq7w4@nyaa42khwqez/#t
> Signed-off-by: Wen Yang <wen.yang@...ux.dev>
> ---
>  include/linux/sysctl.h |  18 ++++++
>  kernel/sysctl-test.c   | 131 ++++++++++-------------------------------
>  kernel/sysctl.c        |  43 ++------------
>  3 files changed, 56 insertions(+), 136 deletions(-)
> 
> diff --git a/include/linux/sysctl.h b/include/linux/sysctl.h
> index 2886fbceb5d6..683422bc3e7f 100644
> --- a/include/linux/sysctl.h
> +++ b/include/linux/sysctl.h
> @@ -53,6 +53,7 @@ struct ctl_dir;
>  #define SYSCTL_MAXOLDUID		((void *)&sysctl_vals[10])
>  #define SYSCTL_NEG_ONE			((void *)&sysctl_vals[11])
>  
> +#define SYSCTL_NULL			((void *)NULL)
>  extern const int sysctl_vals[];
>  
>  #define SYSCTL_LONG_ZERO	((void *)&sysctl_long_vals[0])
> @@ -175,6 +176,23 @@ struct ctl_table {
>  	void *extra2;
>  } __randomize_layout;
>  
> +#define __SYSCTL_ENTRY(NAME, DATA, TYPE, MODE, HANDLER, SMIN, SMAX)\
> +	{							\
> +		.procname	= NAME,				\
> +		.data		= DATA,				\
I would like to see at least one lvalue check for the macro parameters.
Something like this

.data = (*(typeof(DATA)*)&(DATA))

In such a way that the compiler fails if an expression is passed.
We can handle these cases (if they exist) in some other way

> +		.maxlen		= sizeof(TYPE),			\
> +		.mode		= MODE,				\
> +		.proc_handler	= HANDLER,			\
> +		.extra1		= SMIN,				\
> +		.extra2		= SMAX,				\
> +	}
> +
> +#define SYSCTL_ENTRY(NAME, DATA, TYPE, MODE)			\
> +	__SYSCTL_ENTRY(NAME, DATA, TYPE, MODE, proc_do##TYPE##vec, NULL, NULL)
> +
> +#define SYSCTL_RANGE_ENTRY(NAME, DATA, TYPE, MODE, SMIN, SMAX)	\
I'm leaning more towards SYSCTL_ENTRY_RANGE or SYSCTL_ENTRY_R.

I am also keeping notes here if you are interested
https://sysctl-dev-rtd.readthedocs.io/en/latest/notes/ctltable_entry_macro.html#table-entry-macro

Best


> +	__SYSCTL_ENTRY(NAME, DATA, TYPE, MODE, proc_do##TYPE##vec_minmax, SMIN, SMAX)
> +
>  struct ctl_node {
>  	struct rb_node node;
>  	struct ctl_table_header *header;
> diff --git a/kernel/sysctl-test.c b/kernel/sysctl-test.c
> index 92f94ea28957..7fb0e7f1e62f 100644
> --- a/kernel/sysctl-test.c
> +++ b/kernel/sysctl-test.c
> @@ -15,20 +15,14 @@
....
> -	{
> -		.procname	= "ignore-unaligned-usertrap",
> -		.data		= &no_unaligned_warning,
> -		.maxlen		= sizeof (int),
> -		.mode		= 0644,
> -		.proc_handler	= proc_dointvec,
> -	},
> +	SYSCTL_ENTRY("ignore-unaligned-usertrap", &no_unaligned_warning, int, 0644),
>  #endif
>  };
>  
> -- 
> 2.25.1
> 

-- 

Joel Granados

Download attachment "signature.asc" of type "application/pgp-signature" (660 bytes)