Re: Fix UDF-FS potentially dereferencing null

[Linus Torvalds]

Again, "dir" cannot be NULL here, that would be a much more fundamental 
bug and just impossible (the way we get to this thing is to follow the 
directory operations - which we find by looking at "dir").

Maybe we could tell the compiler that "dir" is safe to dereference some 
way? Or add a 'sparse' annotation about safe pointers?

I'd rather just remove the bogus check for a NULL dir pointer..

		Linus

On Fri, 16 Apr 2004, Dave Jones wrote:
>
> Move size instantiation after null check for 'dir', nearer
> to where its first used.
> 
> 		Dave
> 
> --- linux-2.6.5/fs/udf/namei.c~	2004-04-16 22:38:28.000000000 +0100
> +++ linux-2.6.5/fs/udf/namei.c	2004-04-16 22:39:25.000000000 +0100
> @@ -159,7 +159,7 @@
>  	char *nameptr;
>  	uint8_t lfi;
>  	uint16_t liu;
> -	loff_t size = (udf_ext0_offset(dir) + dir->i_size) >> 2;
> +	loff_t size;
>  	lb_addr bloc, eloc;
>  	uint32_t extoffset, elen, offset;
>  	struct buffer_head *bh = NULL;
> @@ -202,6 +202,8 @@
>  		return NULL;
>  	}
>  
> +	size = (udf_ext0_offset(dir) + dir->i_size) >> 2;
> +
>  	while ( (f_pos < size) )
>  	{
>  		fi = udf_fileident_read(dir, &f_pos, fibh, cfi, &bloc, &extoffset, &eloc, &elen, &offset, &bh);
> 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/