Re: [PATCH 5/5] MIPS: Fallback CPU -march CFLAG to ISA level if unsupported

From: Jiaxun Yang <jiaxun.yang@flygoat.com>
To: Nick Desaulniers <ndesaulniers@google.com>
Cc: "linux-mips@vger.kernel.org" <linux-mips@vger.kernel.org>,
	llvm@lists.linux.dev,
	Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
	Nathan Chancellor <nathan@kernel.org>
Subject: Re: [PATCH 5/5] MIPS: Fallback CPU -march CFLAG to ISA level if unsupported
Date: Sat, 8 Apr 2023 10:47:13 +0100	[thread overview]
Message-ID: <7A857648-974B-42DA-AF8B-ED982096E847@flygoat.com> (raw)
In-Reply-To: <CAKwvOdmNR=c4toqkzR_x7ub+L5j8V5Y1PZFOi8s37YRGj9nfMw@mail.gmail.com>

> 2023年4月7日 23:23，Nick Desaulniers <ndesaulniers@google.com> 写道：
> 
> On Fri, Apr 7, 2023 at 3:27 AM Jiaxun Yang <jiaxun.yang@flygoat.com> wrote:
>> 
>> LLVM does not implement some of -march option. However those march
>> does not provide any special functionality in most cases, they just
>> serves as compiler's tuning target.
> 
> Then should we change these then to be in the form:
> 
> -cflags-$(CONFIG_CPU_R3000)     += -march=r3000
> +cflags-$(CONFIG_CPU_R3000)     += -march=mips1 $(call cc-option,-mtune=r3000)

Just tested with GCC and it seems like there are still some difference in generated code
for -march=r4600 and -march=mips3 -mtune=r4600.
As I’m unable to test on a real system I’d prefer just leave them as is.

I’ll try to reword commit message.

Thanks
Jiaxun

> 
> -- 
> Thanks,
> ~Nick Desaulniers