On 10/21/19 11:37 PM, Vinod Koul wrote:
> On 17-10-19, 08:45, Pierre-Louis Bossart wrote:
>>
>>
>> On 10/17/19 2:44 AM, kbuild test robot wrote:
>>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
>>> head:   3ef845da3c3b180ddd386e228ac3228d84a522d3
>>> commit: df72b71921195015e3a8ce772a9ad79442bf0de3 [4732/5057] soundwire: intel: add missing headers for cross-compilation
>>> config: i386-randconfig-f004-201941 (attached as .config)
>>> compiler: gcc-7 (Debian 7.4.0-13) 7.4.0
>>> reproduce:
>>>           git checkout df72b71921195015e3a8ce772a9ad79442bf0de3
>>>           # save the attached .config to linux build tree
>>>           make ARCH=i386
>>>
>>> If you fix the issue, kindly add following tag
>>> Reported-by: kbuild test robot <lkp@intel.com>
>>>
>>> All errors (new ones prefixed by >>):
>>>
>>>      In file included from drivers//soundwire/intel_init.c:12:0:
>>>      include/linux/iomap.h: In function 'iomap_sector':
>>>>> include/linux/iomap.h:76:48: error: 'SECTOR_SHIFT' undeclared (first use in this function); did you mean 'S_DT_SHIFT'?
>>>        return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
>>>                                                      ^~~~~~~~~~~~
>>>                                                      S_DT_SHIFT
>>
>> SECTOR_SHIFT is defined if CONFIG_BLOCK is selected. It isn't selected in
>> this config, so that's probably a dependency that needs to be captured.
>>
>> Alternatively, maybe we should use readl instead, not sure why ioread32 was
>> used in the first place?
> 
> Coz we are using 32 bit writes which are guaranteed to be 32bits
> irrespective of the arch used.

what's wrong with readl/writel? and why do we care anyways, this is used 
in intel_init() so all we care about is that it works on intel hardware?

> 
> Relooking at the change, something does not seem right. If you care
> about ioread32 etc why are you adding iomap.h in intel_init.c and io.h
> intel.c?
> 
> That does not look correct to me. We should include io.h in both the
> places which is the right header which defines the ioread and friends
> and includes asm/io.h
> 
> I did a quick cross compile on couple of arch and this seemed fine.

ok. I can't recall what I did, this was a month ago and the kbuild bot 
complained without iomap.h
If you want to apply fine with me, that's small potatoes compared to the 
rest I am working on.

> 
> -->8---
> 
> From: Vinod Koul <vkoul@kernel.org>
> Date: Tue, 22 Oct 2019 10:03:08 +0530
> Subject: [PATCH] soundwire: intel: use correct header for io calls
> 
> Commit df72b7192119 ("soundwire: intel: add missing headers for
> cross-compilation") tried to fix cross compilation but erroneously used
> wrong header in one of the file. Fix it by using correct io.h header.
> 
> Reported-by: kbuild test robot <lkp@intel.com>
> Fixes: df72b7192119 ("soundwire: intel: add missing headers for cross-compilation")
> Signed-off-by: Vinod Koul <vkoul@kernel.org>
> ---
>   drivers/soundwire/intel_init.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/soundwire/intel_init.c b/drivers/soundwire/intel_init.c
> index d488c44fcbae..2a2b4d8df462 100644
> --- a/drivers/soundwire/intel_init.c
> +++ b/drivers/soundwire/intel_init.c
> @@ -9,7 +9,7 @@
>   
>   #include <linux/acpi.h>
>   #include <linux/export.h>
> -#include <linux/iomap.h>
> +#include <linux/io.h>
>   #include <linux/module.h>
>   #include <linux/platform_device.h>
>   #include <linux/soundwire/sdw_intel.h>
>