Re: [PATCH] video: ep93xx-fb: add missing include of linux/module.h

[Ryan Mallon <rmallon@gmail.com>]

On 23/08/11 02:40, H Hartley Sweeten wrote:
> On Sunday, August 21, 2011 5:31 PM, Ryan Mallon wrote:
>> On 22/08/11 10:06, Axel Lin wrote:
>>> 2011/8/22 Ryan Mallon<rmallon@gmail.com>:
>>>> On 22/08/11 09:41, Ryan Mallon wrote:
>>>>> On 22/08/11 00:39, Axel Lin wrote:
>>>>>> ep93xx-fb.c uses interfaces from linux/module.h,
>>>>>> so it should include that file.  This patch fixes below build errors.
>>>>> What actually changed to make these files broken? Did some other header
>>>>> previously include module.h for us? How many other drivers are broken?
>>>>>
>>>>> Anyway, the change is okay.
>>>>>
>>>>> Acked-by: Ryan Mallon<rmallon@gmail.com>
>>>> Actually, having a second look at this it does not look right.
>>>>
>>>> drivers/video/ep93xx-fb.c includes linux/platform.h (as its first include),
>>>> which includes linux/driver.h, which includes linux/module.h.
>>>>
>>>> Just tested on Linus' latest tree and both this driver and the ep93xx
>>>> backlight driver build fine. What kernel version are you using?
>>>>
>>>> ~Ryan
>>> hi Ryan,
>>>
>>> The patch is against linux-next tree.
>>> I got build error for ep93xx-fb.c and ep93xx_bl.c on linux-next tree.
>>> ( next-20110819 )
>> Ok, I see now. The change which caused the breakage is fdb697c:
>> "include: replace linux/module.h with "struct module" wherever
>> possible". How many other drivers got broken now that device.h does not
>> include module.h?
> Probably a lot...  Which is one of the reasons linux-next exists.. . ;-)

Does anybody know how we can quickly determine which drivers are broken 
short of doing an allyesconfig? I tried to do some quick tricks by 
passing files which contain THIS_MODULE/MODULE_* through cpp, but I get 
loads of errors in headers files because I'm missing some config 
includes. Is there an easy way to get the kbuild arguments for the 
current .config so I can pass them to cpp?

> Actually, Paul Gortmaker caused this breakage with the commit.  He should
> take a deeper look and see what it broke.  From his commit:
>
>      Most of the implicit dependencies on module.h being present by
>      these headers pulling it in have been now weeded out, so we can
>      finally make this change with hopefully minimal breakage.

Quick glance:

ryanm@kiwi:linux-2.6$ grep -lR "^MODULE_" drivers/  | xargs grep -L 
"linux/module.h\|linux/moduleloader.h\|linux/miscdevice.h\|linux/regmap.h\|linux/irq.h"  
| wc -l
579

ryanm@kiwi:linux-2.6$ grep -lR "THIS_MODULE" drivers/ | xargs grep -L 
"linux/module.h\|linux/moduleloader.h\|linux/miscdevice.h\|linux/regmap.h\|linux/irq.h\|linux/export.h\|acpi/platform/aclinux.h\|xen/xenbus.h"   
| wc -l
399

Not sure how many of those are really broken though since there may be a 
few other ways to include module.h/export.h. I would also think there 
would be a lot more build failure reports if all of those were broken :-).

~Ryan