Re: [PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)

[Andrew Morton <akpm@osdl.org>]

OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> wrote:
>
> This patch add EXPORT_SYMBOL(filemap_write_and_wait) and use it.
> 
>  See mm/filemap.c:
> 
>  And changes the filemap_write_and_wait() and filemap_write_and_wait_range().
> 
>  Current filemap_write_and_wait() doesn't wait if filemap_fdatawrite()
>  returns error. However, even if filemap_fdatawrite() returns error, it
>  may be submiting data pages . (e.g. in the case of -ENOSPC)
> 
>  However, even if filemap_fdatawrite() returned an error, it may have
>  submitted the partially data pages to the device.
> 
>  I think we should wait those submitted data pages.

This behaviour is deliberate, although I wouldn't claim that a lot of
thought went into it.

If filemap_fdatawrite() returns an error, this might be due to some I/O
problem: dead disk, unplugged cable, etc.  Given the generally crappy
quality of the kernel's handling of such exceptions, there's a good chance
that the filemap_fdatawait() will get stuck in D state forever.

I don't know how useful that really is - probably not very.  Plus, yes, we
should wait on writeout after a -ENOSPC.

So hum, not sure.  My gut feeling is that anything which we can do to help
the kernel limp along after an I/O error is a good thing, hence the
don't-wait-after-EIO feature should remain.