From: stefan@agner.ch (Stefan Agner)
To: linux-arm-kernel@lists.infradead.org
Subject: specifying order of /dev/mmcblk devices via device-tree?
Date: Fri, 18 Nov 2016 17:23:22 -0800 [thread overview]
Message-ID: <f07a7474de83ed8185f898deb70ffab6@agner.ch> (raw)
In-Reply-To: <CAPDyKFrpqmvVAKsfQBP1m7vdOjPRx7puVkmmCe3yxe1gVFwgzw@mail.gmail.com>
On 2016-11-16 06:45, Ulf Hansson wrote:
> On 28 October 2016 at 17:37, Mark Rutland <mark.rutland@arm.com> wrote:
>> On Fri, Oct 28, 2016 at 08:23:04AM -0700, Tim Harvey wrote:
>>> Greetings,
>>>
>>> I have an IMX6 board that has the following:
>>> sdhc1: mmc0: sdio radio
>>> sdhc2: mmc1: /dev/mmcblk1: microSD connector
>>> sdhc3: mmc2: /dev/mmcblk2: on-board eMMC
>>>
>>> I would like to have sdhc3 registered as /dev/mmcblk0 and sdhc2
>>> registered as /dev/mmcblk1 so that permanent storage is the first
>>> mmcblk device as I think this is more intuitive however currently
>>> these get instanced in the order they appear in the imx6qdl.dtsi
>>> device-tree configuration and are not able to be mapped the way I want
>>> them in my dts file.
>>>
>>> Is there a way, or if not is there a desire for a way, to specify the
>>> order of /dev/mmcblk devices via device-tree?
>>
>> As with many other devices, there is no standard way of controlling the
>> Linux enumeration (and given the ID space is shared with other dynamic
>> devices it's not something that could generally work).
>>
>> These should be refererd to by UUID if possible.
>>
>> If not, we could cosider adding a by-dt-path or something like that.
>
> So does that mean you think using "DT aliases" would be okay? As
> Javier pointed out, there have been some attempts [1] for that, but
> they didn't make it.
> Perhaps we need to re-consider, and if so please re-review the DT
> bindings patch from that series.
I really would like to see some sort of stable MMC block device
ordering.
by-dt-path solutions requires initramfs, which adds complexity and boot
time. Both not very welcome in simple embedded devices.
PARTUUID is relying on a particular instance of a partition, which is
not the same as to say boot from the second partition of device X.
The main problem I see in my patchset is that it works around the fact
MMC is a bus. Adding the mmc controller number to the block device name
would avoid that issue (mmc0blk0...), not sure if a such a drastic
change would be acceptable.
--
Stefan
prev parent reply other threads:[~2016-11-19 1:23 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-28 15:23 specifying order of /dev/mmcblk devices via device-tree? Tim Harvey
2016-10-28 15:33 ` Fabio Estevam
2016-10-28 15:37 ` Javier Martinez Canillas
2016-10-28 15:37 ` Mark Rutland
2016-10-28 16:45 ` Tim Harvey
2016-11-14 19:08 ` Russell King - ARM Linux
2016-11-15 20:27 ` Tim Harvey
2016-11-15 21:35 ` Russell King - ARM Linux
2016-11-15 21:39 ` Tim Harvey
2016-11-15 22:10 ` Russell King - ARM Linux
2016-11-15 23:55 ` Russell King - ARM Linux
2016-11-16 0:33 ` Fabio Estevam
2016-11-19 1:18 ` Stefan Agner
2016-11-16 14:45 ` Ulf Hansson
2016-11-19 1:23 ` Stefan Agner [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=f07a7474de83ed8185f898deb70ffab6@agner.ch \
--to=stefan@agner.ch \
--cc=linux-arm-kernel@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).