[RESEND] SUNRPC: fix include for cmpxchg_relaxed()

Currently net/sunrpc/xprtmultipath.c is the only file outside of arch/
headers and asm-generic/ headers to include <asm/cmpxhcg.h>, apparently
for the use of cmpxchg_relaxed().

However, many architectures do not provide cmpxchg_relaxed() in their
<asm/cmpxhcg.h>, and it is necessary to include <linux/atomic.h> to get
this definition, as noted in Documentation/core-api/atomic_ops.rst:

  If someone wants to use xchg(), cmpxchg() and their variants,
  linux/atomic.h should be included rather than asm/cmpxchg.h, unless
  the code is in arch/* and can take care of itself.

Evidently we're getting the right header this via some transitive
include today, but this isn't something we can/should rely upon,
especially with ongoing rework of the atomic headers for KASAN
instrumentation.

Let's fix the code to include <linux/atomic.h>, avoiding fragility.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Trond Myklebust <trond.myklebust@primarydata.com>
Cc: Anna Schumaker <anna.schumaker@netapp.com>
Cc: J. Bruce Fields <bfields@fieldses.org>
Cc: Jeff Layton <jlayton@poochiereds.net>
Cc: David S. Miller <davem@davemloft.net>
Cc: linux-nfs@vger.kernel.org
Cc: netdev@vger.kernel.org
---
 net/sunrpc/xprtmultipath.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

I sent this about a year ago [1], but got no response. This still applies atop
of v4.17-rc3.

I'm currently trying to implement instrumented atomics for arm64, and it would
be great to have this fixed.

Mark.

[1] https://lkml.kernel.org/r/1489574142-20856-1-git-send-email-mark.rutland@arm.com

On Thu, 2018-05-03 at 14:50 +0100, Mark Rutland wrote:
> Currently net/sunrpc/xprtmultipath.c is the only file outside of arch/
> headers and asm-generic/ headers to include <asm/cmpxhcg.h>, apparently
> for the use of cmpxchg_relaxed().
> 
> However, many architectures do not provide cmpxchg_relaxed() in their
> <asm/cmpxhcg.h>, and it is necessary to include <linux/atomic.h> to get
> this definition, as noted in Documentation/core-api/atomic_ops.rst:
> 
>   If someone wants to use xchg(), cmpxchg() and their variants,
>   linux/atomic.h should be included rather than asm/cmpxchg.h, unless
>   the code is in arch/* and can take care of itself.
> 
> Evidently we're getting the right header this via some transitive
> include today, but this isn't something we can/should rely upon,
> especially with ongoing rework of the atomic headers for KASAN
> instrumentation.
> 
> Let's fix the code to include <linux/atomic.h>, avoiding fragility.
> 
> Signed-off-by: Mark Rutland <mark.rutland@arm.com>
> Cc: Trond Myklebust <trond.myklebust@primarydata.com>
> Cc: Anna Schumaker <anna.schumaker@netapp.com>
> Cc: J. Bruce Fields <bfields@fieldses.org>
> Cc: Jeff Layton <jlayton@poochiereds.net>
> Cc: David S. Miller <davem@davemloft.net>
> Cc: linux-nfs@vger.kernel.org
> Cc: netdev@vger.kernel.org
> ---
>  net/sunrpc/xprtmultipath.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> I sent this about a year ago [1], but got no response. This still applies atop
> of v4.17-rc3.
> 
> I'm currently trying to implement instrumented atomics for arm64, and it would
> be great to have this fixed.
> 
> Mark.
> 
> [1] https://lkml.kernel.org/r/1489574142-20856-1-git-send-email-mark.rutland@arm.com
> 
> diff --git a/net/sunrpc/xprtmultipath.c b/net/sunrpc/xprtmultipath.c
> index e2d64c7138c3..d897f41be244 100644
> --- a/net/sunrpc/xprtmultipath.c
> +++ b/net/sunrpc/xprtmultipath.c
> @@ -13,7 +13,7 @@
>  #include <linux/rcupdate.h>
>  #include <linux/rculist.h>
>  #include <linux/slab.h>
> -#include <asm/cmpxchg.h>
> +#include <linux/atomic.h>
>  #include <linux/spinlock.h>
>  #include <linux/sunrpc/xprt.h>
>  #include <linux/sunrpc/addr.h>

Looks fine.

Reviewed-by: Jeff Layton <jlayton@kernel.org>