linux-kernel - Re: [PATCH] sysctl: unregister sysctl table after testing

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: <jedmwyiggspxnr76ugyax73zwotbnrwpccy7gafdeq6vyweb6z@4c3ivqegpgkd>
Date: Mon, 6 Jan 2025 15:15:13 +0100
From: Joel Granados <joel.granados@...nel.org>
To: John Sperbeck <jsperbeck@...gle.com>
Cc: Kees Cook <kees@...nel.org>, Wen Yang <wen.yang@...ux.dev>, 
	linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH] sysctl: unregister sysctl table after testing


On Tue, Dec 24, 2024 at 09:11:24AM -0800, John Sperbeck wrote:
> In commit b5ffbd139688 ("sysctl: move the extra1/2 boundary check
> of u8 to sysctl_check_table_array"), a kunit test was added that
> registers a sysctl table.  If the test is run as a module, then a
> lingering reference to the module is left behind, and a 'sysctl -a'
> leads to a panic.

Very good catch indeed!!!.
> 
> This can be reproduced with these kernel config settings:
> 
>     CONFIG_KUNIT=y
>     CONFIG_SYSCTL_KUNIT_TEST=m
> 
> Then run these commands:
> 
>     modprobe sysctl-test
>     rmmod sysctl-test
>     sysctl -a
> 
> The panic varies but generally looks something like this:
> 
>     BUG: unable to handle page fault for address: ffffa4571c0c7db4
>     #PF: supervisor read access in kernel mode
>     #PF: error_code(0x0000) - not-present page
>     PGD 100000067 P4D 100000067 PUD 100351067 PMD 114f5e067 PTE 0
>     Oops: Oops: 0000 [#1] SMP NOPTI
>     ... ... ...
>     RIP: 0010:proc_sys_readdir+0x166/0x2c0
>     ... ... ...
>     Call Trace:
>      <TASK>
>      iterate_dir+0x6e/0x140
>      __se_sys_getdents+0x6e/0x100
>      do_syscall_64+0x70/0x150
>      entry_SYSCALL_64_after_hwframe+0x76/0x7e
> 
> If we unregister the test sysctl table, then the failure is gone.
> 
> Fixes: b5ffbd139688 ("sysctl: move the extra1/2 boundary check of u8 to sysctl_check_table_array")
> Signed-off-by: John Sperbeck <jsperbeck@...gle.com>
> ---
>  kernel/sysctl-test.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/sysctl-test.c b/kernel/sysctl-test.c
> index 3ac98bb7fb82..2184c1813b1d 100644
> --- a/kernel/sysctl-test.c
> +++ b/kernel/sysctl-test.c
> @@ -373,6 +373,7 @@ static void sysctl_test_api_dointvec_write_single_greater_int_max(
>  static void sysctl_test_register_sysctl_sz_invalid_extra_value(
>  		struct kunit *test)
>  {
> +	struct ctl_table_header *hdr;
>  	unsigned char data = 0;
>  	struct ctl_table table_foo[] = {
>  		{
> @@ -412,7 +413,9 @@ static void sysctl_test_register_sysctl_sz_invalid_extra_value(
>  
>  	KUNIT_EXPECT_NULL(test, register_sysctl("foo", table_foo));
>  	KUNIT_EXPECT_NULL(test, register_sysctl("foo", table_bar));
> -	KUNIT_EXPECT_NOT_NULL(test, register_sysctl("foo", table_qux));
> +	hdr = register_sysctl("foo", table_qux);
> +	KUNIT_EXPECT_NOT_NULL(test, hdr);
> +	unregister_sysctl_table(hdr);
This indeed fixes the behaviour, but it is not what should be done
and this is why:
1. sysctl-test.c is part of the unit tests for sysctl and actually
   trying to execute a register here does not really make sense.
2. The file that actually does the regression testing is
   lib/test_sysctl.c

If you are up for it this is what needs to be done:
1. change what is in sysctl-test.c to call sysctl_check_table_array
   directly and not worry about keeping track of the registration.
2. Add a similar regression test in lib/test_sysctl.c where we actually
   check for the error.

Please tell me if you are up for it (if not I can add it to my todos)

Best

>  }
>  
>  static struct kunit_case sysctl_test_cases[] = {
> -- 
> 2.47.1.613.gc27f4b7a9f-goog
> 

-- 

Joel Granados