Re: [PATCH v2 17/17] arch, x86: pmem api for ensuring durability of persistent memory updates

From: Dan Carpenter <dan.carpenter@oracle.com>
To: Dan Williams <dan.j.williams@intel.com>,
	Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: linux-nvdimm@ml01.01.org, axboe@kernel.dk, boaz@plexistor.com,
	toshi.kani@hp.com, mingo@kernel.org,
	linux-kernel@vger.kernel.org, hch@lst.de,
	linux-acpi@vger.kernel.org, Ingo Molnar <mingo@redhat.com>,
	"H. Peter Anvin" <hpa@zytor.com>,
	linux-fsdevel@vger.kernel.org,
	Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH v2 17/17] arch, x86: pmem api for ensuring durability of persistent memory updates
Date: Tue, 30 Jun 2015 13:21:37 +0300	[thread overview]
Message-ID: <20150630102137.GC12986@mwanda> (raw)
In-Reply-To: <20150625093749.40066.25475.stgit@dwillia2-desk3.jf.intel.com>

On Thu, Jun 25, 2015 at 05:37:49AM -0400, Dan Williams wrote:
> diff --git a/include/linux/compiler.h b/include/linux/compiler.h
> index 867722591be2..9a528d945498 100644
> --- a/include/linux/compiler.h
> +++ b/include/linux/compiler.h
> @@ -21,6 +21,7 @@
>  # define __rcu		__attribute__((noderef, address_space(4)))

On this side of the #ifdef CONFIG_SPARSE_RCU_POINTER statement then
__pmem isn't defined so it leads to a build error running a CHECKER on
today's linux-next.

I would define __pmem away, but I don't understand why __pmem and
CONFIG_SPARSE_RCU_POINTER are related at all.  Maybe it should be
outside the if statement?

>  #else
>  # define __rcu
> +# define __pmem		__attribute__((noderef, address_space(5)))
>  #endif
>  extern void __chk_user_ptr(const volatile void __user *);
>  extern void __chk_io_ptr(const volatile void __iomem *);
> @@ -42,6 +43,7 @@ extern void __chk_io_ptr(const volatile void __iomem *);
>  # define __cond_lock(x,c) (c)
>  # define __percpu
>  # define __rcu
> +# define __pmem
>  #endif
>  
>  /* Indirect macros required for expanded argument pasting, eg. __LINE__. */

regards,
dan carpenter