Re: [PATCH] mm: msync: require either MS_ASYNC or MS_SYNC

[Michael Kerrisk (man-pages)]

Richard,

On 04/01/2014 08:25 PM, Richard Hansen wrote:
> For the flags parameter, POSIX says "Either MS_ASYNC or MS_SYNC shall
> be specified, but not both." [1]  There was already a test for the
> "both" condition.  Add a test to ensure that the caller specified one
> of the flags; fail with EINVAL if neither are specified.
> 
> Without this change, specifying neither is the same as specifying
> flags=MS_ASYNC because nothing in msync() is conditioned on the
> MS_ASYNC flag.  This has not always been true, 

I am curious (since such things should be documented)--when was
it not true?

> and there's no good
> reason to believe that this behavior would have persisted
> indefinitely.
> 
> The msync(2) man page (as currently written in man-pages.git) is
> silent on the behavior if both flags are unset, so this change should
> not break an application written by somone who carefully reads the
> Linux man pages or the POSIX spec.

Sadly, people do not always carefully read man pages, so there
remains the chance that a change like this will break applications.
Aside from standards conformance, what do you see as the benefit
of the change?

Thanks,

Michael


> [1] http://pubs.opengroup.org/onlinepubs/9699919799/functions/msync.html
> 
> Signed-off-by: Richard Hansen <rhansen@xxxxxxx>
> Reported-by: Greg Troxel <gdt@xxxxxxxxxx>
> Reviewed-by: Greg Troxel <gdt@xxxxxxxxxx>
> ---
> 
> This is a resend of:
> http://article.gmane.org/gmane.linux.kernel/1554416
> I didn't get any feedback from that submission, so I'm resending it
> without changes.
> 
>  mm/msync.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/mm/msync.c b/mm/msync.c
> index 632df45..472ad3e 100644
> --- a/mm/msync.c
> +++ b/mm/msync.c
> @@ -42,6 +42,8 @@ SYSCALL_DEFINE3(msync, unsigned long, start, size_t,
> len, int, flags)
>  		goto out;
>  	if ((flags & MS_ASYNC) && (flags & MS_SYNC))
>  		goto out;
> +	if (!(flags & (MS_ASYNC | MS_SYNC)))
> +		goto out;
>  	error = -ENOMEM;
>  	len = (len + ~PAGE_MASK) & PAGE_MASK;
>  	end = start + len;
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
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/