move fat_cache_init/fat_cache_destroy to a common
header file in fs/fat so that inode.c and cache.c
see the same definition, and to stop warnings
from sparse about undeclared functions
Signed-off-by: Ben Dooks <[email protected]>
diff -urpN -X ../dontdiff linux-2.6.14-git3/fs/fat/cache.c linux-2.6.14-git3-bjd1/fs/fat/cache.c
--- linux-2.6.14-git3/fs/fat/cache.c 2005-09-01 21:02:38.000000000 +0100
+++ linux-2.6.14-git3-bjd1/fs/fat/cache.c 2005-10-31 11:31:32.000000000 +0000
@@ -12,6 +12,8 @@
#include <linux/msdos_fs.h>
#include <linux/buffer_head.h>
+#include "cache.h"
+
/* this must be > 0. */
#define FAT_MAX_CACHE 8
diff -urpN -X ../dontdiff linux-2.6.14-git3/fs/fat/cache.h linux-2.6.14-git3-bjd1/fs/fat/cache.h
--- linux-2.6.14-git3/fs/fat/cache.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.14-git3-bjd1/fs/fat/cache.h 2005-10-31 11:32:37.000000000 +0000
@@ -0,0 +1,8 @@
+/* linux/fs/fat/cache.h
+ *
+ * Written 1992,1993 by Werner Almesberger
+ * header, 2005, Ben Dooks
+*/
+
+extern int __init fat_cache_init(void);
+extern void fat_cache_destroy(void);
diff -urpN -X ../dontdiff linux-2.6.14-git3/fs/fat/inode.c linux-2.6.14-git3-bjd1/fs/fat/inode.c
--- linux-2.6.14-git3/fs/fat/inode.c 2005-10-28 11:28:31.000000000 +0100
+++ linux-2.6.14-git3-bjd1/fs/fat/inode.c 2005-10-31 11:32:29.000000000 +0000
@@ -24,6 +24,8 @@
#include <linux/parser.h>
#include <asm/unaligned.h>
+#include "cache.h"
+
#ifndef CONFIG_FAT_DEFAULT_IOCHARSET
/* if user don't select VFAT, this is undefined. */
#define CONFIG_FAT_DEFAULT_IOCHARSET ""
@@ -1346,9 +1348,6 @@ out_fail:
EXPORT_SYMBOL(fat_fill_super);
-int __init fat_cache_init(void);
-void fat_cache_destroy(void);
-
static int __init init_fat_fs(void)
{
int err;
On Mon, Oct 31, 2005 at 11:36:39AM +0000, Ben Dooks wrote:
> move fat_cache_init/fat_cache_destroy to a common
> header file in fs/fat so that inode.c and cache.c
> see the same definition, and to stop warnings
> from sparse about undeclared functions
> --- linux-2.6.14-git3/fs/fat/cache.h
> +++ linux-2.6.14-git3-bjd1/fs/fat/cache.h
> @@ -0,0 +1,8 @@
> +/* linux/fs/fat/cache.h
Every editor can show the name of the current file.
> + *
> + * Written 1992,1993 by Werner Almesberger
> + * header, 2005, Ben Dooks
> +*/
Too much for two-line file.
> +
> +extern int __init fat_cache_init(void);
> +extern void fat_cache_destroy(void);
Where did you pick up "extern"? Prototypes in *.c don't have it.
Ben Dooks <[email protected]> writes:
> move fat_cache_init/fat_cache_destroy to a common
> header file in fs/fat so that inode.c and cache.c
> see the same definition, and to stop warnings
> from sparse about undeclared functions
The fs/fat/* has many internal functions, it is in
include/linux/msdos_fs.h. Please move those internal functions to one
internal header (probably fs/fat/fat.h?).
This seems be just for sparse, please do real cleanup instead.
--
OGAWA Hirofumi <[email protected]>
On Mon, Oct 31, 2005 at 11:15:34PM +0900, OGAWA Hirofumi wrote:
> Ben Dooks <[email protected]> writes:
>
> > move fat_cache_init/fat_cache_destroy to a common
> > header file in fs/fat so that inode.c and cache.c
> > see the same definition, and to stop warnings
> > from sparse about undeclared functions
>
> The fs/fat/* has many internal functions, it is in
> include/linux/msdos_fs.h. Please move those internal functions to one
> internal header (probably fs/fat/fat.h?).
>
> This seems be just for sparse, please do real cleanup instead.
It's not only for sparse.
The -Wmissing-prototypes flag to gcc gives similar warnings, and I'm
also cleaning up code for adding this flag to the CFLAGS.
Why?
It sometimes happens that the signature of a function changes and it is
forgotten to update all prototypes.
If the prototype is in a header file, gcc tells about the mistake.
If the prototype is in the C file gcc can't help us and it might take
some time until someone tracks the source of the nasty runtime problems
this might cause.
It's your choice as subsystem maintainer which header file the
prototypes should go into - it is only important that both the file with
the actual function and all users of this function #include this header.
> OGAWA Hirofumi <[email protected]>
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
Adrian Bunk <[email protected]> writes:
> It sometimes happens that the signature of a function changes and it is
> forgotten to update all prototypes.
>
> If the prototype is in a header file, gcc tells about the mistake.
>
> If the prototype is in the C file gcc can't help us and it might take
> some time until someone tracks the source of the nasty runtime problems
> this might cause.
>
> It's your choice as subsystem maintainer which header file the
> prototypes should go into - it is only important that both the file with
> the actual function and all users of this function #include this header.
Sounds reasonable, although that's unlikely, because those are
init/exit functions. OK, I'll do this in header cleanup as soon as possible.
Thanks.
--
OGAWA Hirofumi <[email protected]>