Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH v8 18/24] mm: Provide speculative fault infrastructure
To:     Daniel Jordan <daniel.m.jordan@oracle.com>,
        paulmck@linux.vnet.ibm.com, peterz@infradead.org,
        akpm@linux-foundation.org, kirill@shutemov.name,
        ak@linux.intel.com, mhocko@kernel.org, dave@stgolabs.net,
        jack@suse.cz, Matthew Wilcox <willy@infradead.org>,
        benh@kernel.crashing.org, mpe@ellerman.id.au, paulus@samba.org,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, hpa@zytor.com,
        Will Deacon <will.deacon@arm.com>,
        Sergey Senozhatsky <sergey.senozhatsky@gmail.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Alexei Starovoitov <alexei.starovoitov@gmail.com>,
        kemi.wang@intel.com, sergey.senozhatsky.work@gmail.com
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        haren@linux.vnet.ibm.com, khandual@linux.vnet.ibm.com,
        npiggin@gmail.com, bsingharora@gmail.com,
        Tim Chen <tim.c.chen@linux.intel.com>,
        linuxppc-dev@lists.ozlabs.org, x86@kernel.org
References: <1518794738-4186-1-git-send-email-ldufour@linux.vnet.ibm.com>
 <1518794738-4186-19-git-send-email-ldufour@linux.vnet.ibm.com>
 <5b16d6ce-6b62-e4ca-2d78-c25bb008e27e@oracle.com>
From:   Laurent Dufour <ldufour@linux.vnet.ibm.com>
Date:   Mon, 5 Mar 2018 11:50:21 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.6.0
MIME-Version: 1.0
In-Reply-To: <5b16d6ce-6b62-e4ca-2d78-c25bb008e27e@oracle.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Message-Id: <1d6c1ea9-6488-5e25-68b7-6f4aa07d1945@linux.vnet.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi Jordan,

Thanks for reporting this.

On 26/02/2018 18:16, Daniel Jordan wrote:
> Hi Laurent,
> 
> This series doesn't build for me[*] when CONFIG_TRANSPARENT_HUGEPAGE is unset.
> 
> The problem seems to be that the BUILD_BUG() version of pmd_same is called in
> pte_map_lock:
> 
> On 02/16/2018 10:25 AM, Laurent Dufour wrote:
>> +static bool pte_map_lock(struct vm_fault *vmf)
>> +{
> ...snip...
>> +    if (!pmd_same(pmdval, vmf->orig_pmd))
>> +        goto out;
> 
> Since SPF can now call pmd_same without THP, maybe the way to fix it is just
> 
> diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
> index 2cfa3075d148..e130692db24a 100644
> --- a/include/asm-generic/pgtable.h
> +++ b/include/asm-generic/pgtable.h
> @@ -375,7 +375,8 @@ static inline int pte_unused(pte_t pte)
>  #endif
>  
>  #ifndef __HAVE_ARCH_PMD_SAME
> -#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> +#if defined(CONFIG_TRANSPARENT_HUGEPAGE) || \
> +    defined(CONFIG_SPECULATIVE_PAGE_FAULT)
>  static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
>  {
>         return pmd_val(pmd_a) == pmd_val(pmd_b);

We can't fix that this way because some architectures define their own
pmd_same() function (like ppc64), thus forcing the define here will be useless
since __HAVE_ARCH_PMD_SAME is set in that case.

The right way to fix that is to _not check_ for the PMD value in pte_spinlock()
when CONFIG_TRANSPARENT_HUGEPAGE is not set since there is no risk of
collapsing operation in our back if THP are disabled.

I'll fix that in the next version.

Laurent.

> ?
> 
> Daniel
> 
> 
> [*]  The errors are:
> 
> In file included from /home/dmjordan/src/linux/include/linux/kernel.h:10:0,
>                  from /home/dmjordan/src/linux/include/linux/list.h:9,
>                  from /home/dmjordan/src/linux/include/linux/smp.h:12,
>                  from /home/dmjordan/src/linux/include/linux/kernel_stat.h:5,
>                  from /home/dmjordan/src/linux/mm/memory.c:41:
> In function ‘pmd_same.isra.104’,
>     inlined from ‘pte_map_lock’ at /home/dmjordan/src/linux/mm/memory.c:2380:7:
> /home/dmjordan/src/linux/include/linux/compiler.h:324:38: error: call to
> ‘__compiletime_assert_391’ declared with attribute error: BUILD_BUG failed
>   _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>                                       ^
> /home/dmjordan/src/linux/include/linux/compiler.h:304:4: note: in definition of
> macro ‘__compiletime_assert’
>     prefix ## suffix();    \
>     ^~~~~~
> /home/dmjordan/src/linux/include/linux/compiler.h:324:2: note: in expansion of
> macro ‘_compiletime_assert’
>   _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>   ^~~~~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/linux/build_bug.h:45:37: note: in expansion of
> macro ‘compiletime_assert’
>  #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
>                                      ^~~~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/linux/build_bug.h:79:21: note: in expansion of
> macro ‘BUILD_BUG_ON_MSG’
>  #define BUILD_BUG() BUILD_BUG_ON_MSG(1, "BUILD_BUG failed")
>                      ^~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/asm-generic/pgtable.h:391:2: note: in
> expansion of macro ‘BUILD_BUG’
>   BUILD_BUG();
>   ^~~~~~~~~
>   CC      block/elevator.o
>   CC      crypto/crypto_wq.o
> In function ‘pmd_same.isra.104’,
>     inlined from ‘pte_spinlock’ at /home/dmjordan/src/linux/mm/memory.c:2326:7,
>     inlined from ‘handle_pte_fault’ at
> /home/dmjordan/src/linux/mm/memory.c:4181:7:
> /home/dmjordan/src/linux/include/linux/compiler.h:324:38: error: call to
> ‘__compiletime_assert_391’ declared with attribute error: BUILD_BUG failed
>   _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>                                       ^
> /home/dmjordan/src/linux/include/linux/compiler.h:304:4: note: in definition of
> macro ‘__compiletime_assert’
>     prefix ## suffix();    \
>     ^~~~~~
> /home/dmjordan/src/linux/include/linux/compiler.h:324:2: note: in expansion of
> macro ‘_compiletime_assert’
>   _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>   ^~~~~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/linux/build_bug.h:45:37: note: in expansion of
> macro ‘compiletime_assert’
>  #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
>                                      ^~~~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/linux/build_bug.h:79:21: note: in expansion of
> macro ‘BUILD_BUG_ON_MSG’
>  #define BUILD_BUG() BUILD_BUG_ON_MSG(1, "BUILD_BUG failed")
>                      ^~~~~~~~~~~~~~~~
> /home/dmjordan/src/linux/include/asm-generic/pgtable.h:391:2: note: in
> expansion of macro ‘BUILD_BUG’
>   BUILD_BUG();
>   ^~~~~~~~~
> ...
> make[2]: *** [/home/dmjordan/src/linux/scripts/Makefile.build:316: mm/memory.o]
> Error 1
> make[1]: *** [/home/dmjordan/src/linux/Makefile:1047: mm] Error 2
>