On Thu, Apr 06 2017, Jeff Layton wrote:

>
> I tried to avoid updating things unnecesssarily. I could use some
> guidance on how to specify the constants in terms of MAX_ERRNO as well.

ilog2() defined in include/linux/log2.h

And you have MAX_ERROR in one comment, instead of MAX_ERRNO :-)

>  
> -  flush: called by the close(2) system call to flush a file
> +  flush: called by the close(2) system call to flush a file. Writeback
> +	errors not previously reported via fsync should be reported
> +	here as you would for fsync.

"could", not "should".  I think it is agreed that this is a good
idea, is it?

> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 7251f7bb45e8..f33857113ff4 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -394,6 +394,7 @@ struct address_space {
>  	gfp_t			gfp_mask;	/* implicit gfp mask for allocations */
>  	struct list_head	private_list;	/* ditto */
>  	void			*private_data;	/* ditto */
> +	u32			wb_err;

I would rather this was a wb_err_t or similar, and that the functions
which implement it take a pointer to a wb_err_t.
Then the 'error' in 'struct nfs_open_context' could become a 'wb_err_t',
and nfs could use these functions to do error tracking the way it wants
to.

Thanks - looking good.

NeilBrown