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;
Date:   Thu, 5 Dec 2019 10:13:11 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     Wei Yang <richardw.yang@linux.intel.com>
Cc:     x86@kernel.org, linux-kernel@vger.kernel.org,
        richard.weiyang@gmail.com, dave.hansen@linux.intel.com,
        luto@kernel.org, peterz@infradead.or, tglx@linutronix.de
Subject: Re: [Patch v2 4/6] x86/mm: Refine debug print string retrieval
 function
Message-ID: <20191205091311.GD2810@hirez.programming.kicks-ass.net>
References: <20191205021403.25606-1-richardw.yang@linux.intel.com>
 <20191205021403.25606-5-richardw.yang@linux.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20191205021403.25606-5-richardw.yang@linux.intel.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Dec 05, 2019 at 10:14:01AM +0800, Wei Yang wrote:
> Generally, the mapping page size are:
> 
>    4K, 2M, 1G
> 
> except in case 32-bit without PAE, the mapping page size are:
> 
>    4K, 4M
> 
> Based on PG_LEVEL_X definition and mr->page_size_mask, we can calculate
> the mapping page size from a predefined string array.
> 
> Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
> ---
>  arch/x86/mm/init.c | 39 +++++++++++++--------------------------
>  1 file changed, 13 insertions(+), 26 deletions(-)
> 
> diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
> index 0eb5edb63fa2..ded58a31c679 100644
> --- a/arch/x86/mm/init.c
> +++ b/arch/x86/mm/init.c
> @@ -308,29 +308,20 @@ static void __ref adjust_range_page_size_mask(struct map_range *mr,
>  	}
>  }
>  
> +static void __meminit mr_print(struct map_range *mr, unsigned int maxidx)
>  {
> +#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE)
> +	static const char *sz[2] = { "4K", "4M" };
> +#else
> +	static const char *sz[4] = { "4K", "2M", "1G", "" };
> +#endif
> +	unsigned int idx, s;
>  
> +	for (idx = 0; idx < maxidx; idx++, mr++) {
> +		s = (mr->page_size_mask >> PG_LEVEL_2M) & (ARRAY_SIZE(sz) - 1);

Is it at all possible for !PAE to have 1G here, if you use the sz[4]
definition unconditionally?

> +		pr_debug(" [mem %#010lx-%#010lx] page size %s\n",
> +			 mr->start, mr->end - 1, sz[s]);
> +	}
>  }