From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Cercueil Date: Sat, 05 Apr 2014 17:53:55 +0200 Subject: [Buildroot] [PATCH 02/11] arch: MIPS: Add config option BR2_GCC_TARGET_TUNE In-Reply-To: <20140403221215.GA25471@free.fr> References: <1396558881-29631-1-git-send-email-paul@crapouillou.net> <1396558881-29631-2-git-send-email-paul@crapouillou.net> <20140403211925.GC17328@free.fr> <533DD7B6.8080009@crapouillou.net> <20140403221215.GA25471@free.fr> Message-ID: <53402713.9070801@crapouillou.net> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hi, On 04/04/2014 00:12, Yann E. MORIN wrote: > Paul, All, > > On 2014-04-03 23:50 +0200, Paul Cercueil spake thusly: >> This patch would pass the same value to --with-arch and --with-tune, unless >> you define a different value for BR2_GCC_TARGET_TUNE in your defconfig. > I see what you expect, but since BR2_GCC_TARGET_TUNE is a prompt-less > option, setting it in a defconfig will be overriden when you use that > defconfig. > > For example: > > $ cat defconfig > BR2_MIPS=y > BR2_GCC_TARGET_ARCH="FOO-YEM" > $ make BR2_DEFCONFIG=$(pwd)/defconfig defconfig > [...] > $ grep FOO .config > [empty, nada, zilch] > > So, even if you set it in your defconfig, BR2_GCC_TARGET_ARCH is lost. > > Ditto for BR2_GCC_TARGET_TUNE. > >> We >> use that for the Ingenic jz4740 processor, which is a mips32 processor but >> running better the code tuned for mips32r2. > Oh, I see. But that's not gonna happen with this code, I'm afraid. > > Can you double-check that it is indeed working for you? Because if it > does, we have a really big bug in Kconfig (Oh, no, not one more...) > > Regards, > Yann E. MORIN. I confirm it works. Why is it a bug? Paul > >> On 03/04/2014 23:19, Yann E. MORIN wrote: >>> Paul, All, >>> >>> On 2014-04-03 23:01 +0200, Paul Cercueil spake thusly: >>>> This option is actually already used in GCC's package. >>>> >>>> This allows to optimize the toolchain for a specific MIPS processor >>>> while supporting more than one family of processors. >>> Is that really needed? man gcc says: >>> >>> When this option is not used, GCC optimizes for the processor >>> specified by -march. >>> >>> Since this patch would pass the same value to --with-arch and >>> --with-tune, and since this is the default of gcc, is it really >>> needed? >>> >>> Neither ACKing nor NAKing this patch. Can you explain a bit more why we >>> would want that, given the above explanations? >>> >>> Regards, >>> Yann E. MORIN. >>> >>>> Signed-Off-By: Paul Cercueil >>>> --- >>>> arch/Config.in.mips | 10 ++++++++++ >>>> 1 file changed, 10 insertions(+) >>>> >>>> diff --git a/arch/Config.in.mips b/arch/Config.in.mips >>>> index 20951e0..e4160a2 100644 >>>> --- a/arch/Config.in.mips >>>> +++ b/arch/Config.in.mips >>>> @@ -83,6 +83,16 @@ config BR2_GCC_TARGET_ARCH >>>> default "mips64" if BR2_mips_64 >>>> default "mips64r2" if BR2_mips_64r2 >>>> +config BR2_GCC_TARGET_TUNE >>>> + default "mips1" if BR2_mips_1 >>>> + default "mips2" if BR2_mips_2 >>>> + default "mips3" if BR2_mips_3 >>>> + default "mips4" if BR2_mips_4 >>>> + default "mips32" if BR2_mips_32 >>>> + default "mips32r2" if BR2_mips_32r2 >>>> + default "mips64" if BR2_mips_64 >>>> + default "mips64r2" if BR2_mips_64r2 >>>> + >>>> config BR2_MIPS_OABI32 >>>> bool >>>> default y if BR2_mips || BR2_mipsel >>>> -- >>>> 1.9.0 >>>> >>>> _______________________________________________ >>>> buildroot mailing list >>>> buildroot at busybox.net >>>> http://lists.busybox.net/mailman/listinfo/buildroot