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;
MIME-Version: 1.0
References: <20190504094549.10021-1-carmeli.tamir@gmail.com>
 <20190504094549.10021-2-carmeli.tamir@gmail.com> <20190504134503.GA16963@bombadil.infradead.org>
In-Reply-To: <20190504134503.GA16963@bombadil.infradead.org>
From:   Tamir Carmeli <carmeli.tamir@gmail.com>
Date:   Sun, 5 May 2019 21:25:21 +0300
Message-ID: <CAKxm1-H9cgym_RQ-oLcZWEPpyUf5NrZPt_Zu3U=mpU=E38SbvQ@mail.gmail.com>
Subject: Re: [PATCH 1/2] Use list.h instead of file_system_type next
To:     Matthew Wilcox <willy@infradead.org>
Cc:     viro@zeniv.linux.org.uk, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

I just found it weird that there is a proprietary implementation of a
linked list while surely the kernel already offers well established
data structures.
IMO, the current code is a bit hard to understand, especially the
addition of a new struct to the list in the line "*p = fs" after
find_filesystem returned the last member.
Correct, I'm not familiar with all the use cases of the code.

I'm not sure that XArray is a good choice since there is no notion of
an index attached to the pointer, it's really just a linked list of
pointers.
Tell me if you think there is any way to improve my suggestion.
Thanks for you attention.


On Sat, May 4, 2019 at 4:45 PM Matthew Wilcox <willy@infradead.org> wrote:
>
> On Sat, May 04, 2019 at 05:45:48AM -0400, Carmeli Tamir wrote:
> > Changed file_system_type next field to list_head and refactored
> > the code to use list.h functions.
>
> What might be interesting is getting rid of this list and using an XArray
> instead.  This would be a more in-depth change; getting rid of the rwlock
> in favour of using RCU accesses for the read-side and the xa_lock for
> write accesses to the filesystem list.