Subject: Re: [rcu:mmap_sem.2017.08.09a 6/16] mm/internal.h:59:26: error:
 'struct vm_fault' has no member named 'sequence'
To: kbuild test robot <fengguang.wu@intel.com>,
        Peter Zijlstra <peterz@infradead.org>
Cc: kbuild-all@01.org, linux-kernel@vger.kernel.org,
        "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
References: <201708100538.jnTepUPA%fengguang.wu@intel.com>
From: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Date: Thu, 10 Aug 2017 10:41:26 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.1
MIME-Version: 1.0
In-Reply-To: <201708100538.jnTepUPA%fengguang.wu@intel.com>
Content-Type: text/plain; charset=windows-1252
Content-Language: fr
Content-Transfer-Encoding: 7bit
Message-Id: <4e820304-2b59-eb97-4c48-6e0ca94d26a8@linux.vnet.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2360
Lines: 61

On 09/08/2017 23:47, kbuild test robot wrote:
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git mmap_sem.2017.08.09a
> head:   39782fbd9252a3ace9b49a55f4dd2a41a6ced31f
> commit: 1b801585f6b03276db9722dc725a50ca11439467 [6/16] mm: RCU free VMAs
> config: i386-tinyconfig (attached as .config)
> compiler: gcc-6 (Debian 6.2.0-3) 6.2.0 20160901
> reproduce:
>         git checkout 1b801585f6b03276db9722dc725a50ca11439467
>         # save the attached .config to linux build tree
>         make ARCH=i386 
> 
> Note: the rcu/mmap_sem.2017.08.09a HEAD 39782fbd9252a3ace9b49a55f4dd2a41a6ced31f builds fine.
>       It only hurts bisectibility.
> 
> All errors (new ones prefixed by >>):
> 
>    In file included from mm/filemap.c:39:0:
>    mm/internal.h: In function 'vma_has_changed':
>>> mm/internal.h:59:26: error: 'struct vm_fault' has no member named 'sequence'
>      return ret || seq != vmf->sequence;
>                              ^~
> --
>    In file included from lib/vsprintf.c:40:0:
>    lib/../mm/internal.h: In function 'vma_has_changed':
>>> lib/../mm/internal.h:59:26: error: 'struct vm_fault' has no member named 'sequence'
>      return ret || seq != vmf->sequence;>                              ^~

This is because I remove the sequence parameter to this service to rely on
vmf->sequence. But the sequence field is added to the vm fault structure in
the next patch. I'll move the define of vma_has_changed to the next patch
introducing the speculative page fault handler since this service is not
used in this patch.

Cheers,
Laurent.

> vim +59 mm/internal.h
> 
>     44	
>     45	extern struct vm_area_struct *find_vma_srcu(struct mm_struct *mm,
>     46						    unsigned long addr);
>     47	
>     48	static inline bool vma_has_changed(struct vm_fault *vmf)
>     49	{
>     50		int ret = RB_EMPTY_NODE(&vmf->vma->vm_rb);
>     51		unsigned seq = ACCESS_ONCE(vmf->vma->vm_sequence.sequence);
>     52	
>     53		/*
>     54		 * Matches both the wmb in write_seqlock_{begin,end}() and
>     55		 * the wmb in vma_rb_erase().
>     56		 */
>     57		smp_rmb();
>     58	
>   > 59		return ret || seq != vmf->sequence;
>     60	}
>     61	
> 
> ---
> 0-DAY kernel test infrastructure                Open Source Technology Center
> https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
>