WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[kernel test robot]

Hi Pawel,

FYI, the error/warning still remains.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
head:   b311c1b497e51a628aa89e7cb954481e5f9dced2
commit: 982b0192db455d288fc1deb06632f529c35daa15 ice: Refactor finding advertised link speed
date:   10 months ago
config: xtensa-randconfig-r123-20240821 (https://download.01.org/0day-ci/archive/20240822/202408220755.LlaA10C6-lkp@intel.com/config)
compiler: xtensa-linux-gcc (GCC) 14.1.0
reproduce: (https://download.01.org/0day-ci/archive/20240822/202408220755.LlaA10C6-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202408220755.LlaA10C6-lkp@intel.com/

All warnings (new ones prefixed by >>, old ones prefixed by <<):

WARNING: modpost: missing MODULE_DESCRIPTION() in vmlinux.o
WARNING: modpost: vmlinux: section mismatch in reference: put_page+0x78 (section: .text.unlikely) -> initcall_level_names (section: .init.data)
>> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x30 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x4c (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x68 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x84 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xa0 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xbc (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xd8 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xf4 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: WARNING: [xtensa] modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[Przemek Kitszel]

On 8/22/24 01:24, kernel test robot wrote:
> Hi Pawel,
> 
> FYI, the error/warning still remains.

Apologies that we did not replied to the original report.

> 
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> head:   b311c1b497e51a628aa89e7cb954481e5f9dced2
> commit: 982b0192db455d288fc1deb06632f529c35daa15 ice: Refactor finding advertised link speed
> date:   10 months ago
> config: xtensa-randconfig-r123-20240821 (https://download.01.org/0day-ci/archive/20240822/202408220755.LlaA10C6-lkp@intel.com/config)

I wonder why there is 6.6 kernel used, perhaps not a big deal.

> compiler: xtensa-linux-gcc (GCC) 14.1.0
> reproduce: (https://download.01.org/0day-ci/archive/20240822/202408220755.LlaA10C6-lkp@intel.com/reproduce)
> 
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202408220755.LlaA10C6-lkp@intel.com/
> 
> All warnings (new ones prefixed by >>, old ones prefixed by <<):
> 
> WARNING: modpost: missing MODULE_DESCRIPTION() in vmlinux.o
> WARNING: modpost: vmlinux: section mismatch in reference: put_page+0x78 (section: .text.unlikely) -> initcall_level_names (section: .init.data)
>>> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

I have spent just half of hour on that and I'm clueless.

For reference, the driver code is:
	static const u32 arr_name[] __initconst = {
		SOME_CONST,
	};
and core kernel has:
	#define __initconst __section(.init.rodata)


@Max Filippov, you have authored much of xtensa arch for kernel,
especially XIP support, and touched .init.rodata back then;
perhaps you have any idea what is going here?

> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x30 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x4c (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x68 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x84 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xa0 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xbc (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xd8 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0xf4 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
> 

Re: WARNING: [xtensa] modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[Max Filippov]

Hi Przemek,

On Thu, Aug 22, 2024 at 1:23 AM Przemek Kitszel
<przemyslaw.kitszel@intel.com> wrote:
> > WARNING: modpost: missing MODULE_DESCRIPTION() in vmlinux.o
> > WARNING: modpost: vmlinux: section mismatch in reference: put_page+0x78 (section: .text.unlikely) -> initcall_level_names (section: .init.data)
> >>> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
>
> I have spent just half of hour on that and I'm clueless.
>
> For reference, the driver code is:
>         static const u32 arr_name[] __initconst = {
>                 SOME_CONST,
>         };
> and core kernel has:
>         #define __initconst __section(.init.rodata)
>
>
> @Max Filippov, you have authored much of xtensa arch for kernel,
> especially XIP support, and touched .init.rodata back then;
> perhaps you have any idea what is going here?

I see the following:

static struct ethtool_forced_speed_map ice_adv_lnk_speed_maps[]
__ro_after_init = {
       ETHTOOL_FORCED_SPEED_MAP(ice_adv_lnk_speed, 100),

that array goes into the .data, but ETHTOOL_FORCED_SPEED_MAP
expands to the following:

#define ETHTOOL_FORCED_SPEED_MAP(prefix, value)                         \
{                                                                       \
       .speed          = SPEED_##value,                                \
       .cap_arr        = prefix##_##value,                             \
       .arr_size       = ARRAY_SIZE(prefix##_##value),                 \
}

so the first entry of that array quoted above above gets the following
initializer:

  .cap_arr = ice_adv_lnk_speed_100,

and ice_adv_lnk_speed_100 is defined as

  static const u32 ice_adv_lnk_speed_100[] __initconst = {

so this array goes into .init.rodata.
That's a reference from .data to .init.rodata that upsets the modpost checker.
I see that modpost incorrectly deduces where this link points (it's
ice_adv_lnk_speed_*, not __setup_str_initcall_blacklist).
I also see that this link is not used after the init phase, so it's harmless.

-- 
Thanks.
-- Max

Re: WARNING: [xtensa] modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[Przemek Kitszel]

On 8/24/24 08:41, Max Filippov wrote:
> Hi Przemek,
> 
> On Thu, Aug 22, 2024 at 1:23 AM Przemek Kitszel
> <przemyslaw.kitszel@intel.com> wrote:
>>> WARNING: modpost: missing MODULE_DESCRIPTION() in vmlinux.o
>>> WARNING: modpost: vmlinux: section mismatch in reference: put_page+0x78 (section: .text.unlikely) -> initcall_level_names (section: .init.data)
>>>>> WARNING: modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)
>>
>> I have spent just half of hour on that and I'm clueless.
>>
>> For reference, the driver code is:
>>          static const u32 arr_name[] __initconst = {
>>                  SOME_CONST,
>>          };
>> and core kernel has:
>>          #define __initconst __section(.init.rodata)
>>
>>
>> @Max Filippov, you have authored much of xtensa arch for kernel,
>> especially XIP support, and touched .init.rodata back then;
>> perhaps you have any idea what is going here?
> 
> I see the following:
> 
> static struct ethtool_forced_speed_map ice_adv_lnk_speed_maps[]
> __ro_after_init = {
>         ETHTOOL_FORCED_SPEED_MAP(ice_adv_lnk_speed, 100),
> 
> that array goes into the .data, but ETHTOOL_FORCED_SPEED_MAP
> expands to the following:
> 
> #define ETHTOOL_FORCED_SPEED_MAP(prefix, value)                         \
> {                                                                       \
>         .speed          = SPEED_##value,                                \
>         .cap_arr        = prefix##_##value,                             \
>         .arr_size       = ARRAY_SIZE(prefix##_##value),                 \
> }
> 
> so the first entry of that array quoted above above gets the following
> initializer:
> 
>    .cap_arr = ice_adv_lnk_speed_100,
> 
> and ice_adv_lnk_speed_100 is defined as
> 
>    static const u32 ice_adv_lnk_speed_100[] __initconst = {
> 
> so this array goes into .init.rodata.
> That's a reference from .data to .init.rodata that upsets the modpost checker.
> I see that modpost incorrectly deduces where this link points (it's
> ice_adv_lnk_speed_*, not __setup_str_initcall_blacklist).
> I also see that this link is not used after the init phase, so it's harmless.
> 

Thank you very much for the analysis :)

Re: WARNING: [xtensa] modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[Al Viro]

On Fri, Aug 23, 2024 at 11:41:17PM -0700, Max Filippov wrote:

> static struct ethtool_forced_speed_map ice_adv_lnk_speed_maps[]
> __ro_after_init = {
>        ETHTOOL_FORCED_SPEED_MAP(ice_adv_lnk_speed, 100),
> 
> that array goes into the .data,

... due to

#define __ro_after_init __read_mostly

in your asm/cache.h, instead of the usual .data..ro_after_init,
which would be enough for modpost to assume that driver knows
what it's doing and won't access the dangling pointers to
.init.rodata in there after the initmem had been freed.

The same goes for qed and the same thing happens on openrisc,
for exact same reason.

While we are at it, that might as well had been
#define __ro_after_init
since __read_mostly is not defined on xtensa and default is empty.

If you don't want that stuff to go into RODATA, why not
simply define an empty RO_AFTER_INIT_DATA in your vmlinux.lds
and put those sections explicitly there, along with the data ones,
as e.g. s390 does?  Or arch/arm/kernel/vmlinux-xip.lds.S,
for that matter...

Re: WARNING: [xtensa] modpost: vmlinux: section mismatch in reference: ice_adv_lnk_speed_maps+0x14 (section: .data) -> __setup_str_initcall_blacklist (section: .init.rodata)

[Max Filippov]

On Wed, Dec 4, 2024 at 3:05 PM Al Viro <viro@zeniv.linux.org.uk> wrote:
>
> On Fri, Aug 23, 2024 at 11:41:17PM -0700, Max Filippov wrote:
>
> > static struct ethtool_forced_speed_map ice_adv_lnk_speed_maps[]
> > __ro_after_init = {
> >        ETHTOOL_FORCED_SPEED_MAP(ice_adv_lnk_speed, 100),
> >
> > that array goes into the .data,
>
> ... due to
>
> #define __ro_after_init __read_mostly
>
> in your asm/cache.h,

Oh, the trail leads back to me (:

> instead of the usual .data..ro_after_init,
> which would be enough for modpost to assume that driver knows
> what it's doing and won't access the dangling pointers to
> .init.rodata in there after the initmem had been freed.
>
> The same goes for qed and the same thing happens on openrisc,
> for exact same reason.
>
> While we are at it, that might as well had been
> #define __ro_after_init
> since __read_mostly is not defined on xtensa and default is empty.
>
> If you don't want that stuff to go into RODATA, why not
> simply define an empty RO_AFTER_INIT_DATA in your vmlinux.lds
> and put those sections explicitly there, along with the data ones,
> as e.g. s390 does?  Or arch/arm/kernel/vmlinux-xip.lds.S,
> for that matter...

I'll do that, thanks for digging into it.

-- 
Thanks.
-- Max