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;
From:   "Huang, Ying" <ying.huang@intel.com>
To:     Zi Yan <ziy@nvidia.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>, <linux-mm@kvack.org>,
        <linux-kernel@vger.kernel.org>, Yang Shi <shy828301@gmail.com>,
        Baolin Wang <baolin.wang@linux.alibaba.com>,
        Oscar Salvador <osalvador@suse.de>,
        "Matthew Wilcox" <willy@infradead.org>,
        Bharata B Rao <bharata@amd.com>,
        "Alistair Popple" <apopple@nvidia.com>,
        haoxin <xhao@linux.alibaba.com>
Subject: Re: [PATCH 5/8] migrate_pages: batch _unmap and _move
References: <20221227002859.27740-1-ying.huang@intel.com>
        <20221227002859.27740-6-ying.huang@intel.com>
        <05A09BF6-A30C-45A5-952D-0CCFC43FE2D7@nvidia.com>
Date:   Wed, 04 Jan 2023 08:34:46 +0800
In-Reply-To: <05A09BF6-A30C-45A5-952D-0CCFC43FE2D7@nvidia.com> (Zi Yan's
        message of "Tue, 03 Jan 2023 14:01:37 -0500")
Message-ID: <87lemjnnmh.fsf@yhuang6-desk2.ccr.corp.intel.com>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=ascii
Precedence: bulk

Zi Yan <ziy@nvidia.com> writes:

> On 26 Dec 2022, at 19:28, Huang Ying wrote:
>
>> In this patch the _unmap and _move stage of the folio migration is
>> batched.  That for, previously, it is,
>>
>>   for each folio
>>     _unmap()
>>     _move()
>>
>> Now, it is,
>>
>>   for each folio
>>     _unmap()
>>   for each folio
>>     _move()
>
> Also worth adding some notes here, we need extra code to undo the _unmap()
> if _move() fails. Andrew has asked for comments on *_undo_src/dst(),
> but I think it would be better to provide a high level new workflow,
> in the form of pseudo code, in git log and the comment for migrate_pages().
> The extra cleanup code for a failed _move() with a previously successful
> _unmap() might not be obvious to everyone.

Here, I removed the detailed error processing to make it easier to
understand the basic flow changing.  So, I would rather to keep the
pseudo code here as simple as possible and keep the detailed flow in the
diff below.

Best Regards,
Huang, Ying

>>
>> Based on this, we can batch the TLB flushing and use some hardware
>> accelerator to copy folios between batched _unmap and batched _move
>> stages.
>>

[snip]