Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <2eff7546-5ab5-d4cd-43f6-d66b6490d725@intel.com>
Date:   Tue, 28 Jun 2022 08:52:00 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.9.1
Subject: Re: [RFC v2] memmap: introduce cmdline parameter "memmap=nn[KMG]$"
 without start addr
Content-Language: en-US
To:     lizhe.67@bytedance.com, tglx@linutronix.de, mingo@redhat.com,
        bp@alien8.de, dave.hansen@linux.intel.com, x86@kernel.org,
        hpa@zytor.com
Cc:     lizefan.x@bytedance.com, linux-doc@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <20220628124529.15431-1-lizhe.67@bytedance.com>
From:   Dave Hansen <dave.hansen@intel.com>
In-Reply-To: <20220628124529.15431-1-lizhe.67@bytedance.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 6/28/22 05:45, lizhe.67@bytedance.com wrote:
> From: Li Zhe <lizhe.67@bytedance.com>
> 
> In current kernel we can use memmap=nn[KMG]$ss[KMG] to reserve an
> area of memory for userspace usage through /dev/mem. We have to
> determine the start addr and length. In our scenario, we need
> reserve or alloc large continuous physical memory at least 256M in
> 512G's machine, and need reserve more memory in larger machine, at
> just boot phase for a userspace program. And these memories will not
> be freed to system before system reboot. The userspace program can
> use the memory through /dev/mem to store some data. Besides, a
> hardware will need the data stored by the userspace program to do
> it's job. Why we need continuous memory is that the hardware can
> only access memory without mmu.

That's still a rather anemic description.

I don't think we want this code in the kernel with that description, sorry.

If you'd like to come back with an actual description, naming the exact
hardware that you have *AND* explaining sane reasons for why the
hardware doesn't have a normal kernel driver (or interface) like every
other piece of hardware, then we can talk.  But, otherwise, this isn't
something that can go upstream.