Re: Slab: Remove kmem_cache_t

[Linus Torvalds <torvalds@osdl.org>]

On Tue, 28 Nov 2006, Andrew Morton wrote:

> On Wed, 29 Nov 2006 15:42:44 +1100
> Nick Piggin <nickpiggin@yahoo.com.au> wrote:
> 
> > So what exactly is wrong with
> > a kmem_cache_t declaration in include files, then?
> 
> a) it's a typedef and
> 
> b) it's a typedef, and you cannot forward-declare typedefs.  We've hit this
>    a couple of times.  Header files need to include slab.h just to be able to do
> 
> 	extern kmem_cache_t *wozzle;

Yeah. There really is a good reason to never really use typedef's at all.

The _only_ valid reason to use typedefs ever is because it's a type that 
depends on some configuration option (where "architecture" is obviously 
the biggest config option of all).

If the type has the same name regardless of config options, it should 
never be a typedef. And "struct kmem_cache" obviously doesn't change names 
just because of config options (it may change some of the _members_ it 
contains, but that's a totally different thing, and has no bearing 
what-so-ever on whether you should use a typedef).

So typedefs are good for

 - "u8"/"u16"/"u32"/"u64" kind of things, where the underlying types 
   really are potentially different on different architectures.

 - "sector_t"-like things which may be 32-bit or 64-bit depending on some 
   CONFIG_LBD option or other.

 - as a special case, "sparse" actually makes bitwise typedefs have real 
   meaning as types, so if you are using sparse to distinguish between a 
   little-endian 16-bit entity or a big-endian 16-bit entity, the typedef 
   there is actually important and has real meaning to sparse (without the 
   typedef, each bitwise type declaration would be strictly a _different_ 
   type from another bitwise type declaration that otherwise looks the 
   same).

But typedefs are NOT good for:

 - trying to avoid typing a few characters:

	"kmem_cache_t" is strictly _worse_ than "struct kmem_cache", not 
	just because it causes declaration issues. It also hides the fact 
	that the thing really is a structure (and hiding the fact that 
	it's a pointer is a shooting offense: things like "voidptr_t" 
	should not be allowed at all)

 - incorrect "portability". 

	the POSIX "socklen_t" was not only a really bad way to write
	"int", it actually caused a lot of NON-portability, and made some 
	people think it should be "size_t" or something equally broken.

The one excuse for typedefs in the "typing" sense can be complicated 
function pointer types. Some function pointers are just too easy to screw 
up, and using a

	typedef (*myfn_t)(int, ...);

can be preferable over forcing people to write that really complex kind of 
type out every time. But that shouldn't be overused either (but we use it 
for things like "readdir_t", for example, for exactly this reason).

		Linus

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>