Re: [akpm-mm:mm-unstable 75/80] drivers/video/fbdev/core/fb_defio.c:272:17: error: implicit declaration of function 'mapping_wrprotect_page'

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Sun, Feb 02, 2025 at 12:16:55AM +0800, kernel test robot wrote:
> Hi Lorenzo,
>
> FYI, the error/warning was bisected to this commit, please ignore it if it's irrelevant.
>
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-unstable
> head:   7de6fd8ab65003f050aa58e705592745717ed318
> commit: 6a9318f28d7e371692838677c364145fe04b1093 [75/80] fb_defio: do not use deprecated page->mapping, index fields
> config: sh-randconfig-002-20250201 (https://download.01.org/0day-ci/archive/20250202/202502020030.MnEJ847Z-lkp@intel.com/config)
> compiler: sh4-linux-gcc (GCC) 14.2.0
> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250202/202502020030.MnEJ847Z-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/202502020030.MnEJ847Z-lkp@intel.com/
>
> All errors (new ones prefixed by >>):
>
>    drivers/video/fbdev/core/fb_defio.c: In function 'fb_deferred_io_work':
> >> drivers/video/fbdev/core/fb_defio.c:272:17: error: implicit declaration of function 'mapping_wrprotect_page' [-Wimplicit-function-declaration]
>      272 |                 mapping_wrprotect_page(fbdefio->mapping, pgoff, 1, page);
>          |                 ^~~~~~~~~~~~~~~~~~~~~~

This seems entirely bogus to me, you're building defio on a non-MMU arch,
whereas defio literally relies on an MMU to function at all.

I mean what does it actually mean to wrprotect otherwise? :)

I think this is because CONFIG_FB_DEFERRED_IO should be made to depend on
CONFIG_MMU but isn't.

>
>
> vim +/mapping_wrprotect_page +272 drivers/video/fbdev/core/fb_defio.c
>
>    258
>    259	/* workqueue callback */
>    260	static void fb_deferred_io_work(struct work_struct *work)
>    261	{
>    262		struct fb_info *info = container_of(work, struct fb_info, deferred_work.work);
>    263		struct fb_deferred_io_pageref *pageref, *next;
>    264		struct fb_deferred_io *fbdefio = info->fbdefio;
>    265
>    266		/* here we wrprotect the page's mappings, then do all deferred IO. */
>    267		mutex_lock(&fbdefio->lock);
>    268		list_for_each_entry(pageref, &fbdefio->pagereflist, list) {
>    269			struct page *page = pageref->page;
>    270			pgoff_t pgoff = pageref->offset >> PAGE_SHIFT;
>    271
>  > 272			mapping_wrprotect_page(fbdefio->mapping, pgoff, 1, page);
>    273		}
>    274
>    275		/* driver's callback with pagereflist */
>    276		fbdefio->deferred_io(info, &fbdefio->pagereflist);
>    277
>    278		/* clear the list */
>    279		list_for_each_entry_safe(pageref, next, &fbdefio->pagereflist, list)
>    280			fb_deferred_io_pageref_put(pageref, info);
>    281
>    282		mutex_unlock(&fbdefio->lock);
>    283	}
>    284
>
> --
> 0-DAY CI Kernel Test Service
> https://github.com/intel/lkp-tests/wiki

I mean I'm not sure it's worth doing anything about this since this is just
an invalid config combo, which was masked before because for instance
folio_mkclean() has a !CONFIG_MMU stub, however it really does not make
sense for there to be such a stub for a function which explicitly refers to
wrprotect afaic.

But if we want to stop randconfig build bots tripping this, probably best
to solve by making CONFIG_FB_DEFERRED_IO depend on CONFIG_MMU, so I will
send a fix-patch to that effect I guess.

I mean quite literally deferred I/O _relies_ on an MMU doing page faulting
to function at all. So this dependency was always an oversight (I guess
they assumed nobody would be crazy enough to figure out a config that
allows the combination of the two things as clearly no real hardware would
:)