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
From:   Hao Lee <haolee.swjtu@gmail.com>
Date:   Mon, 28 Sep 2020 22:37:17 +0800
Message-ID: <CA+PpKPkAmnVhkwgZU84udJ0BEi+Pd4UOBaJJDVmvd+P=Depbng@mail.gmail.com>
Subject: Can we remove the checking of CONFIG_ARCH_HAS_PTE_SPECIAL in __vm_insert_mixed?
To:     npiggin@gmail.com
Cc:     dan.j.wiilliams@intel.com, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Sorry, the previous email is corrupted. This is a resending.

In the current code, if the architecture supports PTE_SPECIAL,
__vm_insert_mixed() becomes a simple encapsulation of insert_pfn()
because it always calls insert_pfn() and never calls insert_page(). This
behavior is inconsistent with the function's name. As described in commit
423bad600443 ("mm: add vm_insert_mixed"), this function is designed to
supports _both_ raw pfns and pfns with struct page, but in architecture
with PTE_SPECIAL, it treats the latter as the former. So, I think we'd
better remove the checking of CONFIG_ARCH_HAS_PTE_SPECIAL to make the
behavior consistent for all architectures. If my understanding is right,
I can submit a patch to fix this. Thanks.

Regards,
Hao Lee