Re: [RFC PATCH] tools/x86: add kcpuid tool to show raw CPU features

From: Borislav Petkov <bp@alien8.de>
To: Feng Tang <feng.tang@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>, "H . Peter Anvin" <hpa@zytor.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Dave Hansen <dave.hansen@intel.com>,
	x86@kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC PATCH] tools/x86: add kcpuid tool to show raw CPU features
Date: Wed, 2 Sep 2020 17:40:18 +0200	[thread overview]
Message-ID: <20200902154018.GA21537@zn.tnic> (raw)
In-Reply-To: <1598514543-90152-1-git-send-email-feng.tang@intel.com>

On Thu, Aug 27, 2020 at 03:49:03PM +0800, Feng Tang wrote:
> output of the tool
> ------------------
> 
> 	CPUID leafs total: 28
> 
> 	cpu family	: 6
> 	model		: 13
> 	stepping	: 7

Yeah, this should dump model etc and those numbers should be in hex and
additionally in dec if people prefer them.

> 	CPU features
> 	------------
> 	 sse3
> 	 pclmlqdq
> 	 dtes64
> 	 mwait
> 	 ds_cpl
> 	 vmx
> 	 smx
> 	 eist
> 	 tm2
> 	 ssse3
> 	 cx16
> 	 pctxid
> 	 dca
> 	 sse4_1
> 	 sse4_2
> 	 x2apic
> 	 tsc_deadline_timer
> 	 aes
> 	 xsave
> 	 osxsave
> 	 avx
> 	 fpu
> 	 ...

I guess that's good for grepping. With a lot of leafs, leaf output
should probably be controlled by cmdline opts.

> [1] http://sr71.net/~dave/intel/stupid-cpuid.c
> 
> Originally-by: Borislav Petkov <bp@alien8.de>
> Suggested-by: Dave Hansen <dave.hansen@intel.com>
> Signed-off-by: Feng Tang <feng.tang@intel.com>
> ---
>  tools/arch/x86/tools/kcpuid/Makefile |  21 ++
>  tools/arch/x86/tools/kcpuid/kcpuid.c | 422 +++++++++++++++++++++++++++++++++++
>  2 files changed, 443 insertions(+)
>  create mode 100644 tools/arch/x86/tools/kcpuid/Makefile
>  create mode 100644 tools/arch/x86/tools/kcpuid/kcpuid.c

Let's drop the second "tools" from the path:

tools/arch/x86/kcpuid/Makefile
tools/arch/x86/kcpuid/kcpuid.c

> +struct reg_01_a {
> +	u32 stepping: 4;	/* bit 0 */
> +	u32 model: 4;
> +	u32 family: 4;
> +	u32 type: 2;
> +	u32 pad1: 2;
> +	u32 model_ext: 4;	/* bit 16 */
> +	u32 family_ext: 8;
> +};

Yeah, instead of defining a separate struct for each leaf I think it
would be smarter/better to have a text file in a machine parseable
format which defines your leafs.

When you need to add a new leaf, you simply extend the text file and the
tool parses it anew and has its all CPUID info uptodate. This way you
won't even have to recompile it. Adding new CPUID leafs would be adding new
lines to the file.

For example:

LEAF<num>,SUBLEAF<num>,[EAX,EBX,ECX,EDX]{[width]<Mnemonic>|<Long text>,...}

LEAF07,SUBLEAF00,EAX{[31:0]max_value|Max input value for supported subleafs}
LEAF07,SUBLEAF00,EBX{[0]FSGSBASE|RDFSBASE/RDGSBASE/WRFSBASE/WRGSBASE if 1.,
		     [1]TSC_ADJUST|IA32_TSC_ADJUST MSR is supported if 1.,
		     [2]SGX|Supports Intel® Software Guard Extensions (Intel® SGX Extensions) if 1.,
		     ...
}
LEAF07,SUBLEAF00,ECX{[0]PREFETCHWT1|(Intel® Xeon PhiTM only.),
		     [1]AVX512_VBMI|,
		     ...

This is just a dumb attempt but I hope it comes across where I'm getting
with this.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette