Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:7 as permitted sender) client-ip=2620:137:e000::3:7;
Message-ID: <acfd4fdd-1e43-2118-1b44-a926b6f09b3b@landley.net>
Date:   Wed, 1 Nov 2023 09:48:55 -0500
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.0
Subject: Re: [RFC PATCH] rootfs: Use tmpfs for rootfs even if root= is given
Content-Language: en-US
To:     Stefan Berger <stefanb@linux.ibm.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc:     linux-kernel@vger.kernel.org,
        "Milton D. Miller II" <mdmii@outlook.com>,
        Jeff Layton <jlayton@kernel.org>, Jens Axboe <axboe@kernel.dk>,
        Jim Cromie <jim.cromie@gmail.com>,
        Sam Ravnborg <sam@ravnborg.org>,
        "Eric W. Biederman" <ebiederm@xmission.com>,
        Alexander Viro <viro@zeniv.linux.org.uk>,
        "H. Peter Anvin" <hpa@zytor.com>, Mimi Zohar <zohar@linux.ibm.com>
References: <20231031154417.621742-1-stefanb@linux.ibm.com>
 <2023103159-punctuate-amount-f09d@gregkh>
 <6dae6aa6-e6c6-89d6-f9d7-7563708f7662@landley.net>
 <e30c9f81-959f-418d-a4b1-d34d47391e09@linux.ibm.com>
From:   Rob Landley <rob@landley.net>
In-Reply-To: <e30c9f81-959f-418d-a4b1-d34d47391e09@linux.ibm.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 11/1/23 07:11, Stefan Berger wrote:
> On 11/1/23 07:35, Rob Landley wrote:
>> On 10/31/23 11:56, Greg Kroah-Hartman wrote:
>>> On Tue, Oct 31, 2023 at 11:44:17AM -0400, Stefan Berger wrote:
>>>> rootfs currently does not use tmpfs if the root= boot option is passed
>>>> even though the documentation about rootfs (added in 6e19eded3684) in
>>>> Documentation/filesystems/ramfs-rootfs-initramfs.rst states:
>>>>
>>>>    If CONFIG_TMPFS is enabled, rootfs will use tmpfs instead of ramfs by
>>>>    default.  To force ramfs, add "rootfstype=ramfs" to the kernel command
>>>>    line.
>>>
>>> At this point in time, is there even any difference between ramfs and
>>> tmpfs anymore?  Why would you want to choose one over the other here?
>> 
>> I submitted a patch to fix this to the list multiple times, which got ignored as
>> always. Most recently here:
>> 
>> https://lore.kernel.org/lkml/8244c75f-445e-b15b-9dbf-266e7ca666e2@landley.net/
> 
> I just tried it with your patch and the machine I am using this with 
> (OpenBMC) uses the boot command line 'console=ttyS4,115200n8 
> root=/dev/ram rw'. When I append rootfstype=tmpfs to this boot command 
> line then it starts actually using tmpfs. So I think this would work for me.
> I can add my Tested-by tag to the patch if this helps to get it merged. 
> Ideally it would also propagate back with a Fixes tag...

Regarding the "why ramfs" question, one bug report I periodically get is people
using initramfs.cpio.gz that works on ramfs but fails to extract when they try
tmpfs, and the reason is tmpfs defaults to size=50% of memory and their archive
extracts to more than that. Since ramfs hasn't got a limit, it extracted and ran
fine (generally a dedicated init app doing its IoT thing) and so far they've
always gone back to ramfs as their fix.

I vaguely recall I had some todo item to let them supply arguments so they could
specify their own size= for initmpfs (ramfs doesn't take any so it hadn't been
wired up last time I looked), but somewhere between
https://lkml.org/lkml/2016/6/22/686 and
https://lkml.iu.edu/hypermail/linux/kernel/2302.2/05597.html still being out of
tree 7 years later I kind of lost interest...

Rob