On Thu, 18 Oct 2007 13:35:51 +0400
Dmitry Antipov <antipov@dev.rtsoft.ru> wrote:

quoted text
> This patch proposes an additional error checking performed within setfl().
> 
> As a result, fcntl (fd, F_SETFL, O_ASYNC) will return -1 and set errno
> to -EINVAL if filp->f_op->fasync is NULL for file specified by fd. This
> is possible, for example, if fd is a descriptor returned by inotify_init().
> 
> 
> Dmitry
> 
> 
> 
> [2.6.23-fcntl-fasync-check.patch  text/x-patch (353B)]
> Signed-off-by: Dmitry Antipov <antipov@dev.rtsoft.ru>
> 
> --- .orig-2.6.23/fs/fcntl.c	2007-10-17 15:26:06.000000000 +0400
> +++ 2.6.23/fs/fcntl.c	2007-10-17 15:25:27.000000000 +0400
> @@ -240,6 +240,9 @@
>  			error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
>  			if (error < 0)
>  				goto out;
> +		} else {
> +			error = -EINVAL;
> +			goto out;
>  		}
>  	}

This would have made sense whent he code was originally written
but it now has a (small) potential to break existing applications.

I guess if the _only_ fd's which don't implement fasync are inotify,
signalfd and other such new-and-obscure things then the risk is
probably acceptably low.

But is the proposed change actually very useful?
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/