Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20201231053156.24276-1-zhuling8@huawei.com>
In-Reply-To: <20201231053156.24276-1-zhuling8@huawei.com>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Mon, 4 Jan 2021 21:42:59 -0800
Message-ID: <CAPcyv4i9mn7SJ6CSY4uH_74f6pRny5aLQhg6Ubf3cDw8kJoWzg@mail.gmail.com>
Subject: Re: [PATCH] arm64: add pmem module for kernel update
To:     Zhuling <zhuling8@huawei.com>
Cc:     Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        linux-nvdimm <linux-nvdimm@lists.01.org>,
        Linux MM <linux-mm@kvack.org>, Will Deacon <will@kernel.org>,
        Vishal L Verma <vishal.l.verma@intel.com>,
        Dave Jiang <dave.jiang@intel.com>, luanjianhai@huawei.com,
        luchunhua@huawei.com, sangyan@huawei.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Zhuling,

On Wed, Dec 30, 2020 at 10:18 PM Zhuling <zhuling8@huawei.com> wrote:
>
> Category: feature
> Bugzilla: NA
> CVE: NA

These tags can be dropped.

>
> Use reserved memory to create a pmem device to store the
> processes information that dumped before kernel update.
> When you want to use this feature you need to declare by
> "pmemmem=pmem_size:pmem_phystart" in cmdline.
> (exp: pmemmem=100M:0x202000000000)
>

Interesting. I like the feature, but it's not clear to me that a new
command line based configuration scheme is needed. There is the
existing memmap= parameter that on x86 describes a
IORES_DESC_PERSISTENT_MEMORY_LEGACY range. The driver/nvdimm/e820.c
driver could be reworked to attach to the same thing on ARM64.

Then as far as assigning memory to different kernel usages there is
the existing capability in libnvdimm to attach a "personality" to an
nvdimm namespace. I imagine you could write a special signature to the
namespace that libnvdimm would recognize as a KUP reservation
namespace and work generically across any arch.