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 v2 4/8] powerpc/vdso32: inline __get_datapage()
To:     Santosh Sivaraj <santosh@fossix.org>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Michael Ellerman <mpe@ellerman.id.au>
Cc:     linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
References: <cover.1566491310.git.christophe.leroy@c-s.fr>
 <194fb7bc973ef2ce43016c97dd32f2b2dcbae4e7.1566491310.git.christophe.leroy@c-s.fr>
 <87h864iiq9.fsf@santosiv.in.ibm.com>
 <a95c4a58-bc59-3e75-46db-414f6d0f1412@c-s.fr>
 <87blvonwzz.fsf@santosiv.in.ibm.com>
From:   Christophe Leroy <christophe.leroy@c-s.fr>
Message-ID: <559296a2-37f6-dc83-7f60-07567637a9a8@c-s.fr>
Date:   Fri, 13 Sep 2019 15:50:07 +0200
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.9.0
MIME-Version: 1.0
In-Reply-To: <87blvonwzz.fsf@santosiv.in.ibm.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: fr
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


Le 13/09/2019 à 15:31, Santosh Sivaraj a écrit :
> Christophe Leroy <christophe.leroy@c-s.fr> writes:
> 
>> Hi Santosh,
>>
>> Le 26/08/2019 à 07:44, Santosh Sivaraj a écrit :
>>> Hi Christophe,
>>>
>>> Christophe Leroy <christophe.leroy@c-s.fr> writes:
>>>
>>>> __get_datapage() is only a few instructions to retrieve the
>>>> address of the page where the kernel stores data to the VDSO.
>>>>
>>>> By inlining this function into its users, a bl/blr pair and
>>>> a mflr/mtlr pair is avoided, plus a few reg moves.
>>>>
>>>> The improvement is noticeable (about 55 nsec/call on an 8xx)
>>>>
>>>> vdsotest before the patch:
>>>> gettimeofday:    vdso: 731 nsec/call
>>>> clock-gettime-realtime-coarse:    vdso: 668 nsec/call
>>>> clock-gettime-monotonic-coarse:    vdso: 745 nsec/call
>>>>
>>>> vdsotest after the patch:
>>>> gettimeofday:    vdso: 677 nsec/call
>>>> clock-gettime-realtime-coarse:    vdso: 613 nsec/call
>>>> clock-gettime-monotonic-coarse:    vdso: 690 nsec/call
>>>>
>>>> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
>>>> ---
>>>>    arch/powerpc/kernel/vdso32/cacheflush.S   | 10 +++++-----
>>>>    arch/powerpc/kernel/vdso32/datapage.S     | 29 ++++-------------------------
>>>>    arch/powerpc/kernel/vdso32/datapage.h     | 11 +++++++++++
>>>>    arch/powerpc/kernel/vdso32/gettimeofday.S | 13 ++++++-------
>>>>    4 files changed, 26 insertions(+), 37 deletions(-)
>>>>    create mode 100644 arch/powerpc/kernel/vdso32/datapage.h
>>>
>>> The datapage.h file should ideally be moved under include/asm, then we can use
>>> the same for powerpc64 too.
>>
>> I have a more ambitious project indeed.
>>
>> Most of the VDSO code is duplicated between vdso32 and vdso64. I'm
>> aiming at merging everything into a single source code.
>>
>> This means we would have to generate vdso32.so and vdso64.so out of the
>> same source files. Any idea on how to do that ? I'm not too good at
>> creating Makefiles. I guess we would have everything in
>> arch/powerpc/kernel/vdso/ and would have to build the objects twice,
>> once in arch/powerpc/kernel/vdso32/ and once in arch/powerpc/kernel/vdso64/
> 
> Should we need to build the objects twice? For 64 bit config it is going to be
> a 64 bit build else a 32 bit build. It should suffice to get the single source
> code compile for both, maybe with macros or (!)CONFIG_PPC64 conditional
> compilation. Am I missing something when you say build twice?
> 

IIUC, on PPC64 we build vdso64 for 64bits user apps and vdso32 for 
32bits user apps.

In arch/powerpc/kernel/Makefile, you have:

obj-$(CONFIG_VDSO32)		+= vdso32/
obj-$(CONFIG_PPC64)		+= vdso64/

And in arch/powerpc/platforms/Kconfig.cputype, you have:

config VDSO32
	def_bool y
	depends on PPC32 || CPU_BIG_ENDIAN
	help
	  This symbol controls whether we build the 32-bit VDSO. We obviously
	  want to do that if we're building a 32-bit kernel. If we're building
	  a 64-bit kernel then we only want a 32-bit VDSO if we're building for
	  big endian. That is because the only little endian configuration we
	  support is ppc64le which is 64-bit only.


Christophe