Web lists-archives.com

Re: [PATCH 14/14] mm: turn on vm_fault_t type checking




On Wed, May 16, 2018 at 07:43:48AM +0200, Christoph Hellwig wrote:
> Switch vm_fault_t to point to an unsigned int with __bіtwise annotations.
> This both catches any old ->fault or ->page_mkwrite instance with plain
> compiler type checking, as well as finding more intricate problems with
> sparse.
> 
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---

<ULTRASNIP>

For the iomap and xfs parts,
Reviewed-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>

That said...

> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index 54f1e05ecf3e..da2b77a19911 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -22,7 +22,8 @@
>  #endif
>  #define AT_VECTOR_SIZE (2*(AT_VECTOR_SIZE_ARCH + AT_VECTOR_SIZE_BASE + 1))
>  
> -typedef int vm_fault_t;
> +typedef unsigned __bitwise vm_fault_t;
> +
>  
>  struct address_space;
>  struct mem_cgroup;
> @@ -619,7 +620,7 @@ struct vm_special_mapping {
>  	 * If non-NULL, then this is called to resolve page faults
>  	 * on the special mapping.  If used, .pages is not checked.
>  	 */
> -	int (*fault)(const struct vm_special_mapping *sm,
> +	vm_fault_t (*fault)(const struct vm_special_mapping *sm,

Uh, we're changing function signatures /and/ redefinining vm_fault_t?
All in the same 90K patch?

I /was/ expecting a series of "convert XXXXX and all callers/users"
patches followed by a trivial one to switch the definition, not a giant
pile of change.  FWIW I don't mind so much if you make a patch
containing a change for some super-common primitive and a hojillion
little diff hunks tree-wide, but only one logical change at a time for a
big patch, please...

I quite prefer seeing the whole series from start to finish all packaged
up in one series, but wow this was overwhelming. :/

--D

<ULTRASNIP>