Date: Sat, 23 Feb 2013 23:19:45 +0100 From: Dimitry Andric <dim@FreeBSD.org> To: Konstantin Belousov <kostikbel@gmail.com> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r247166 - head/contrib/llvm/tools/clang/lib/Driver Message-ID: <51294081.1060505@FreeBSD.org> In-Reply-To: <20130223201857.GD2454@kib.kiev.ua> References: <201302222240.r1MMeAsu042087@svn.freebsd.org> <20130222235332.GB2454@kib.kiev.ua> <5128AED6.2020906@FreeBSD.org> <20130223201857.GD2454@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2013-02-23 21:18, Konstantin Belousov wrote: ... > Let me rephrase my question. > > What is the processor targeted by the cc on i386, when no -march flag is > specified, for cc coming from clang, for both .c and .s files. Lets ignore > known bugs, like long nops or cmovs. For i386 arch on FreeBSD, the default has always been i486. The cc1 stage, which compiles .c and .cpp to .o files, uses either this default target CPU, or otherwise the CPU specified on the command line. The cc1as stage, which assembles .s to .o files, only uses the target CPU setting to determine whether to emit long nops or not, at least on x86 arch. After the last fix, the default target is i486, so it will not emit long nops either, unless a higher CPU is specified.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?51294081.1060505>