Re: [PATCH v3] err.h: allow IS_ERR_VALUE to handle properly more types

[Arnd Bergmann]

On Wednesday 03 February 2016 14:15:28 Andrzej Hajda wrote:
> diff --git a/include/linux/err.h b/include/linux/err.h
> index 56762ab..b7d4a9f 100644
> --- a/include/linux/err.h
> +++ b/include/linux/err.h
> @@ -18,7 +18,9 @@
>  
>  #ifndef __ASSEMBLY__
>  
> -#define IS_ERR_VALUE(x) unlikely((x) >= (unsigned long)-MAX_ERRNO)
> +#define IS_ERR_VALUE(x) ((typeof(x))(-1) <= 0 \
> +                               ? unlikely((x) <= -1) \
> +                               : unlikely((x) >= (typeof(x))-MAX_ERRNO))
>  
>  static inline void * __must_check ERR_PTR(long error)
>  {
> 

This has caused a warning to reappear that I had fixed before:

fs/gfs2/dir.c: In function 'get_first_leaf':
fs/gfs2/dir.c:802:9: warning: 'leaf_no' may be used uninitialized in this function [-Wmaybe-uninitialized]
   error = get_leaf(dip, leaf_no, bh_out);
         ^
fs/gfs2/dir.c: In function 'dir_split_leaf':
fs/gfs2/dir.c:1021:8: warning: 'leaf_no' may be used uninitialized in this function [-Wmaybe-uninitialized]
  error = get_leaf(dip, leaf_no, &obh);

See my original patch that was applied at
http://www.gossamer-threads.com/lists/linux/kernel/2353964

Apparently the new version is complex enough to prevent gcc from doing
some optimizations it should do.

I have tried to come up with a new variant that does not bring
the warning back and that should work in all cases:

diff --git a/include/linux/err.h b/include/linux/err.h
index b7d4a9ff6342..bd4936a2c352 100644
--- a/include/linux/err.h
+++ b/include/linux/err.h
@@ -18,9 +18,7 @@
 
 #ifndef __ASSEMBLY__
 
-#define IS_ERR_VALUE(x) ((typeof(x))(-1) <= 0 \
-				? unlikely((x) <= -1) \
-				: unlikely((x) >= (typeof(x))-MAX_ERRNO))
+#define IS_ERR_VALUE(x)	(unlikely((unsigned long long)(x) >= (unsigned long long)(typeof(x))-MAX_ERRNO))
 
 static inline void * __must_check ERR_PTR(long error)
 {


I'm not sure if the cast to 'unsigned long long' might cause less
efficient code to be generated by gcc. I would hope that it is smart
enough to not actually extend shorter variables to 64 bit before
doing the comparison but I have not checked yet.

	Arnd