[PATCH] ARM: dts: orion5x: add mtd flash support to linkstation lswtgl

Roger Shimizu rogershimizu at gmail.com
Fri Jan 29 05:48:04 PST 2016


Dear Andrew,

On Thu, Jan 28, 2016 at 10:45 PM, Roger Shimizu <rogershimizu at gmail.com> wrote:
> Dear Andrew,
>
> On Thu, Jan 28, 2016 at 3:22 AM, Roger Shimizu <rogershimizu at gmail.com> wrote:
>> On Thu, Jan 28, 2016 at 2:32 AM, Andrew Lunn <andrew at lunn.ch> wrote:
>>> Same comment i just sent to Thomas:
>>>
>>> The partition table should be a subnode of the mtd node and should be named
>>> 'partitions'. This node should have the following property:
>>> - compatible : (required) must be "fixed-partitions"
>>> Partitions are then defined in subnodes of the partitions node.
>>>
>>> For backwards compatibility partitions as direct subnodes of the mtd device are
>>> supported. This use is discouraged.
>>>
>>>            Andrew
>>
>> Thanks for your comments!
>>
>> Using the patch I submitted result in:
>>
>> [    1.667440] Found: SST 39LF020
>> [    1.670613] f4000000.flash: Found 1 x8 devices at 0x0 in 8-bit bank
>> [    1.676948] number of JEDEC chips: 1
>> [    1.698943] 3 ofpart partitions found on MTD device f4000000.flash
>> [    1.705222] Creating 3 MTD partitions on "f4000000.flash":
>> [    1.710803] 0x000000000000-0x000000030000 : "header"
>> [    1.719051] 0x000000030000-0x00000003f000 : "uboot"
>> [    1.727182] 0x00000003f000-0x000000040000 : "uboot_env"
>>
>> Applying the partition DT proposed in
>> Documentation/devicetree/bindings/mtd/partition.txt, result in:
>>
>> [    1.667879] Found: SST 39LF020
>> [    1.671053] f4000000.flash: Found 1 x8 devices at 0x0 in 8-bit bank
>> [    1.677389] number of JEDEC chips: 1
>>
>> So the partitions are gone.
>
> Now I'm thinking for my dts's case, which is orion5x, refer
> - Documentation/devicetree/bindings/mtd/mtd-physmap.txt
> - Documentation/devicetree/bindings/mtd/orion-nand.txt
> would be more proper.
> And in the sample of mtd-physmap.txt, the structure described in
> partition.txt is not mentioned.
>
> Could we go as the dts is, like the original patch I submitted?

I find "Documentation/devicetree/bindings/mtd/partition.txt" is quite
new and was just updated in kernel 4.4.
So I tried "linux-image-4.4.0-trunk-orion5x" in Debian experimental
repo, YES, with my updated dts, the MTD goes back to work as
partitioned as before.

In this way, if I write dts for partition like this, user cannot
simply take dtb from kernel 4.4/4.5 to their stable kernel (such as
Debian Jessie's 3.16-ckt series).
For some advanced functionality, this kind of non-backport capable
change is deserved; but for partitions IMHO it doesn't deserve,
especially for a device/platform existing for more than 5 years.

What do you think?

-- 
Roger Shimizu, GMT +9 Tokyo
PGP/GPG: 17B3ACB1



More information about the linux-arm-kernel mailing list