[PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gordon Bos]

Excito B3 boards use SPI flash for booting and several userland
functions require access to this flash memory, including power down
control. However /proc/mtd shows an empty list and the userland
tools respond with "device not found".

Change driver reference for spi@10600 to M25P80

Verified on my own Excito B3

Signed-off-by: Gordon Bos <gordon-PMRpws9CsXpfxmBXdGbrbQ@public.gmane.org>
---
 arch/arm/boot/dts/kirkwood-b3.dts | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm/boot/dts/kirkwood-b3.dts b/arch/arm/boot/dts/kirkwood-b3.dts
index c9247f8..4b1e414 100644
--- a/arch/arm/boot/dts/kirkwood-b3.dts
+++ b/arch/arm/boot/dts/kirkwood-b3.dts
@@ -71,10 +71,10 @@
 		spi@10600 {
 			status = "okay";
 
-			m25p16@0 {
+			m25p80@0 {
 				#address-cells = <1>;
 				#size-cells = <1>;
-				compatible = "st,m25p16";
+				compatible = "st,m25p80";
 				reg = <0>;
 				spi-max-frequency = <40000000>;
 				mode = <0>;
-- 
2.2.2

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gregory CLEMENT]

Hi Gordon,


On 06/02/2015 09:45, Gordon Bos wrote:
> Excito B3 boards use SPI flash for booting and several userland
> functions require access to this flash memory, including power down
> control. However /proc/mtd shows an empty list and the userland
> tools respond with "device not found".

The change in arm/boot/dts don't go through the device tree subsystem
but through the arm-soc one. Especially it it is only a fix and if
you don't introduce or modifying a binding. For this kind of patch
you should cc the mvebu maintainers (I added the others one)

However it seems that the get_maintainer.pl script is misleading on this
point we should maybe be updating the MAINTAINERS file.

> 
> Change driver reference for spi@10600 to M25P80
> 
> Verified on my own Excito B3

This change looks OK for me however it is too late to merge it
and it will be part of the next merge window.

Andrew, as you initially wrote this dts would you have a look on it?

As it is a fix we will also apply to the stable branches. It would
help if you could tell us since when this bug is present.


Thanks,

Gregory


> 
> Signed-off-by: Gordon Bos <gordon-PMRpws9CsXpfxmBXdGbrbQ@public.gmane.org>
> ---
>  arch/arm/boot/dts/kirkwood-b3.dts | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm/boot/dts/kirkwood-b3.dts b/arch/arm/boot/dts/kirkwood-b3.dts
> index c9247f8..4b1e414 100644
> --- a/arch/arm/boot/dts/kirkwood-b3.dts
> +++ b/arch/arm/boot/dts/kirkwood-b3.dts
> @@ -71,10 +71,10 @@
>  		spi@10600 {
>  			status = "okay";
>  
> -			m25p16@0 {
> +			m25p80@0 {
>  				#address-cells = <1>;
>  				#size-cells = <1>;
> -				compatible = "st,m25p16";
> +				compatible = "st,m25p80";
>  				reg = <0>;
>  				spi-max-frequency = <40000000>;
>  				mode = <0>;
> 


-- 
Gregory Clement, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gordon Bos]

[-- Attachment #1: Type: text/plain, Size: 2831 bytes --]

Hi,

I've had some further discussion on this with another B3 owner.
According to an old patch that Excito corporation built for kernel
version 2.6 the flash memory is in fact a ||Numonyx MP25P16
The driver however must be m25p80.

It would therefore seem cleaner to just correct the driver reference
and leave the label as is. The patch thus becomes:

Signed-off-by: Gordon Bos <gordon@bosvangennip.nl>
---
  arch/arm/boot/dts/kirkwood-b3.dts | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/boot/dts/kirkwood-b3.dts b/arch/arm/boot/dts/kirkwood-b3.dts
index c9247f8..5c6e8d3 100644
--- a/arch/arm/boot/dts/kirkwood-b3.dts
+++ b/arch/arm/boot/dts/kirkwood-b3.dts
@@ -74,7 +74,7 @@
  			m25p16@0 {
  				#address-cells = <1>;
  				#size-cells = <1>;
-				compatible = "st,m25p16";
+				compatible = "st,m25p80";
  				reg = <0>;
  				spi-max-frequency = <40000000>;
  				mode = <0>;
---

Regards,
Gordon Bos


On 02/06/2015 10:09 AM, Gregory CLEMENT wrote:
> Hi Gordon,
>
>
> On 06/02/2015 09:45, Gordon Bos wrote:
>> Excito B3 boards use SPI flash for booting and several userland
>> functions require access to this flash memory, including power down
>> control. However /proc/mtd shows an empty list and the userland
>> tools respond with "device not found".
> The change in arm/boot/dts don't go through the device tree subsystem
> but through the arm-soc one. Especially it it is only a fix and if
> you don't introduce or modifying a binding. For this kind of patch
> you should cc the mvebu maintainers (I added the others one)
>
> However it seems that the get_maintainer.pl script is misleading on this
> point we should maybe be updating the MAINTAINERS file.
>
>> Change driver reference for spi@10600 to M25P80
>>
>> Verified on my own Excito B3
> This change looks OK for me however it is too late to merge it
> and it will be part of the next merge window.
>
> Andrew, as you initially wrote this dts would you have a look on it?
>
> As it is a fix we will also apply to the stable branches. It would
> help if you could tell us since when this bug is present.
>
>
> Thanks,
>
> Gregory
>
>
>> Signed-off-by: Gordon Bos <gordon@bosvangennip.nl>
>> ---
>>   arch/arm/boot/dts/kirkwood-b3.dts | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/arm/boot/dts/kirkwood-b3.dts b/arch/arm/boot/dts/kirkwood-b3.dts
>> index c9247f8..4b1e414 100644
>> --- a/arch/arm/boot/dts/kirkwood-b3.dts
>> +++ b/arch/arm/boot/dts/kirkwood-b3.dts
>> @@ -71,10 +71,10 @@
>>   		spi@10600 {
>>   			status = "okay";
>>   
>> -			m25p16@0 {
>> +			m25p80@0 {
>>   				#address-cells = <1>;
>>   				#size-cells = <1>;
>> -				compatible = "st,m25p16";
>> +				compatible = "st,m25p80";
>>   				reg = <0>;
>>   				spi-max-frequency = <40000000>;
>>   				mode = <0>;
>>
>


[-- Attachment #2: Type: text/html, Size: 3752 bytes --]

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Andrew Lunn]

On Sat, Feb 07, 2015 at 04:04:59PM +0100, Gordon Bos wrote:
> Hi,
> 
> I've had some further discussion on this with another B3 owner.
> According to an old patch that Excito corporation built for kernel
> version 2.6 the flash memory is in fact a ||Numonyx MP25P16
> The driver however must be m25p80.

Hi Gordon

I need to reboot my B3 in order to see what it says. However, with the
current DT i do have access to the MTD devices.

What problems are you actually having?

Thanks

	Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Andrew Lunn]

On Sat, Feb 07, 2015 at 04:04:59PM +0100, Gordon Bos wrote:
> Hi,
> 
> I've had some further discussion on this with another B3 owner.
> According to an old patch that Excito corporation built for kernel
> version 2.6 the flash memory is in fact a ||Numonyx MP25P16
> The driver however must be m25p80.

Hi Gordon

Here is what my B3 says during boot:

m25p80 spi0.0: m25p16 (2048 Kbytes)
3 ofpart partitions found on MTD device spi0.0
Creating 3 MTD partitions on "spi0.0":
0x000000000000-0x0000000c0000 : "u-boot"
0x0000000c0000-0x0000000e0000 : "u-boot env"
0x0000000e0000-0x000000200000 : "data"

m25p80 is the driver being used. The device is an m25p16.

>  			m25p16@0 {
>  				#address-cells = <1>;
>  				#size-cells = <1>;
> -				compatible = "st,m25p16";
> +				compatible = "st,m25p80";

Here, compatibility indicates what the device is. The kernel will then
find a driver for that device. If you look in
drivers/mtd/devices/m25p80.c you see a long list of devices this
driver supports, of which m25p16.

Also, if you specify the wrong device here, the kernel will complain
when it asks the device to identify itself and find it is different:

http://lxr.free-electrons.com/source/drivers/mtd/spi-nor/spi-nor.c#L915

it will print a message: "found %s, expected %s\n"

Are you seeing such a message? Maybe your B3 has a different device
than mine?

Thanks
	Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gordon Bos]

[-- Attachment #1: Type: text/plain, Size: 2959 bytes --]

Hi Andrew,

Unsure how it can work for you, but this is the corresponding entry from 
the Excito release:

---
|/*****************************************************************************
  * 2048KB SPI Flash on Boot Device (Numonyx MP25P16)
  ****************************************************************************/

static struct mtd_partition bubba3_flash_parts[] = {
  {
  .name = "u-boot",
  .size = SZ_512K+SZ_256K,
  .offset = 0,
  },
  {
  .name = "env",
  .size = SZ_128K,
  .offset = MTDPART_OFS_NXTBLK,
  },
  {
  .name = "data",
  .size = MTDPART_SIZ_FULL,
  .offset = MTDPART_OFS_NXTBLK,
  },
};

static const struct flash_platform_data bubba3_flash = {
  .type = "m25p16",
  .name = "spi_flash",
  .parts = bubba3_flash_parts,
  .nr_parts = ARRAY_SIZE(bubba3_flash_parts),
};

static struct spi_board_info __initdata bubba3_spi_slave_info[] = {
  {
  .modalias = "m25p80",
  .platform_data = &bubba3_flash,
  .irq = -1,
  .max_speed_hz = 40000000,
  .bus_num = 0,
  .chip_select = 0,
  },
};
---

I originally started out with a kernel build by user 'sakaki-' on git
(project gentoo-on-b3) and found that it had no access to MTD. The
readme on this project also states that the front led on the B3 will
go off for some 40 seconds or so and then go back on, which is achieved
by an openrc script after sysfs is mounted.

The behaviour we're seeing therefore seems to contradict what you must
be seeing, since you explicitly turn the green led off which would
indicate that the leds default to 'on' position in your case while on
multiple B3's we've now confirmed that the default is in fact off and
blue and red should be explicitly turned on.

As for MTD access I found that sakaki's original kernel did not have
CONFIG_MTD_M25P80 enabled but this did not fix access to MTD (see
discussion at http://forum.mybubba.org/viewtopic.php?f=7&t=5680&start=16
). Only after I changed the driver 'compatible' reference to match
m25p80 I got access to MTD (with that very same kernel).

As you may know, being a B3 owner, performing a shutdown requires
writing a 'magic' DWORD value to a specific location in mtd2 and mtd1
holds a key that is required for a dyndns service that is exclusive for
Excito B3 owners.

The submitted MTD patch has been confirmed to fix the issue on four
different B3's now, running gentoo patched kernels version 3.17.1, 3.18.3
and 3.18.5.


Regards,
Gordon Bos


|On 07/02/2015 17:09, Andrew Lunn wrote:
> On Sat, Feb 07, 2015 at 04:04:59PM +0100, Gordon Bos wrote:
>> Hi,
>>
>> I've had some further discussion on this with another B3 owner.
>> According to an old patch that Excito corporation built for kernel
>> version 2.6 the flash memory is in fact a ||Numonyx MP25P16
>> The driver however must be m25p80.
> Hi Gordon
>
> I need to reboot my B3 in order to see what it says. However, with the
> current DT i do have access to the MTD devices.
>
> What problems are you actually having?
>
> Thanks
>
> 	Andrew


[-- Attachment #2: Type: text/html, Size: 4677 bytes --]

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Andrew Lunn]

On Sat, Feb 07, 2015 at 06:15:12PM +0100, Gordon Bos wrote:
> Hi Andrew,
> 
> Unsure how it can work for you, but this is the corresponding entry
> from the Excito release:
> 
> ---
> |/*****************************************************************************
>  * 2048KB SPI Flash on Boot Device (Numonyx MP25P16)
>  ****************************************************************************/
> 
> static struct mtd_partition bubba3_flash_parts[] = {
>  {
>  .name = "u-boot",
>  .size = SZ_512K+SZ_256K,
>  .offset = 0,
>  },
>  {
>  .name = "env",
>  .size = SZ_128K,
>  .offset = MTDPART_OFS_NXTBLK,
>  },
>  {
>  .name = "data",
>  .size = MTDPART_SIZ_FULL,
>  .offset = MTDPART_OFS_NXTBLK,
>  },
> };
> 
> static const struct flash_platform_data bubba3_flash = {
>  .type = "m25p16",

So the device is a m25p16. 

>  .name = "spi_flash",
>  .parts = bubba3_flash_parts,
>  .nr_parts = ARRAY_SIZE(bubba3_flash_parts),
> };
> 
> static struct spi_board_info __initdata bubba3_spi_slave_info[] = {
>  {
>  .modalias = "m25p80",

And the driver is m25p80.

This is consistent with the current device tree description. 

So what actually happens when you use the mainline DT on your device?

I'm not going to accept any changes until we understand what problems
you have and why it works for me.

Thanks
    Andrew



--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gordon Bos]

[-- Attachment #1: Type: text/plain, Size: 1936 bytes --]

Hi Andrew,

I may be mistaken here, but this here references the driver that is to 
be used, or not?

|            compatible = "st,m25p16";

|For me that does not work with CONFIG_MTD_M25P80. As far as I can see 
there is also no
reference whatsoever to the kernel selectable m25p80 in the current DT 
for kirkwood-b3,
so I'm not sure why you'd find this consistent with it.

In the kernels I tried, m25p80 driver is simply not happy with m25p16 
being defined in DT.

Regards,
Gordon Bos

On 07/02/2015 18:36, Andrew Lunn wrote:
> On Sat, Feb 07, 2015 at 06:15:12PM +0100, Gordon Bos wrote:
>> Hi Andrew,
>>
>> Unsure how it can work for you, but this is the corresponding entry
>> from the Excito release:
>>
>> ---
>> |/*****************************************************************************
>>   * 2048KB SPI Flash on Boot Device (Numonyx MP25P16)
>>   ****************************************************************************/
>>
>> static struct mtd_partition bubba3_flash_parts[] = {
>>   {
>>   .name = "u-boot",
>>   .size = SZ_512K+SZ_256K,
>>   .offset = 0,
>>   },
>>   {
>>   .name = "env",
>>   .size = SZ_128K,
>>   .offset = MTDPART_OFS_NXTBLK,
>>   },
>>   {
>>   .name = "data",
>>   .size = MTDPART_SIZ_FULL,
>>   .offset = MTDPART_OFS_NXTBLK,
>>   },
>> };
>>
>> static const struct flash_platform_data bubba3_flash = {
>>   .type = "m25p16",
> So the device is a m25p16.
>
>>   .name = "spi_flash",
>>   .parts = bubba3_flash_parts,
>>   .nr_parts = ARRAY_SIZE(bubba3_flash_parts),
>> };
>>
>> static struct spi_board_info __initdata bubba3_spi_slave_info[] = {
>>   {
>>   .modalias = "m25p80",
> And the driver is m25p80.
>
> This is consistent with the current device tree description.
>
> So what actually happens when you use the mainline DT on your device?
>
> I'm not going to accept any changes until we understand what problems
> you have and why it works for me.
>
> Thanks
>      Andrew
>
>
>


[-- Attachment #2: Type: text/html, Size: 2614 bytes --]

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Andrew Lunn]

On Sat, Feb 07, 2015 at 07:22:13PM +0100, Gordon Bos wrote:
> Hi Andrew,
> 
> I may be mistaken here, but this here references the driver that is
> to be used, or not?
> 
> |            compatible = "st,m25p16";


Nope, that is the device, not the driver.

The driver has a list of devices it is capable of driving. The kernel
uses that list to find the right driver.

> In the kernels I tried, m25p80 driver is simply not happy with
> m25p16 being defined in DT.

Please define "not happy". Please show me the kernel log of it being
not happy.

We need to compare and contrast my kernel log, where it is happy with
"st,m25p16" against your kernel log where it is unhappy. Then we might
have some clues to understand what is going on.

Thanks
	Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Gordon Bos]

[-- Attachment #1: Type: text/plain, Size: 1569 bytes --]

Hi Andrew,

There is a difference, and apparently it somewhere got solved:

 From running kernel 3.17.1

      ~ # dmesg | grep spi

Running an upgraded kernel 3.18.5 with the same DT blob

      ~ # dmesg | grep spi
    [    0.928832] m25p80 spi0.0: found m25p16, expected m25p80
    [    0.934124] m25p80 spi0.0: m25p16 (2048 Kbytes)
    [    0.938666] 3 ofpart partitions found on MTD device spi0.0
    [    0.944124] Creating 3 MTD partitions on "spi0.0":


In 3.17.1 the MTD devices were missing without the patch, in 3.18.5 they 
also show
up without the patch.

You may therefore consider the MTD issue as solved.

Which leaves just the (aesthetic) issue of the boot led being turned off 
while loading
the kernel.


Regards,
Gordon Bos


On 07/02/2015 19:40, Andrew Lunn wrote:
> On Sat, Feb 07, 2015 at 07:22:13PM +0100, Gordon Bos wrote:
>> Hi Andrew,
>>
>> I may be mistaken here, but this here references the driver that is
>> to be used, or not?
>>
>> |            compatible = "st,m25p16";
>
> Nope, that is the device, not the driver.
>
> The driver has a list of devices it is capable of driving. The kernel
> uses that list to find the right driver.
>
>> In the kernels I tried, m25p80 driver is simply not happy with
>> m25p16 being defined in DT.
> Please define "not happy". Please show me the kernel log of it being
> not happy.
>
> We need to compare and contrast my kernel log, where it is happy with
> "st,m25p16" against your kernel log where it is unhappy. Then we might
> have some clues to understand what is going on.
>
> Thanks
> 	Andrew


[-- Attachment #2: Type: text/html, Size: 2463 bytes --]

Re: [PATCH 1/1] arm: Fix unavailable MTD userland devices on Excito B3 boards

[Andrew Lunn]

On Sat, Feb 07, 2015 at 08:59:46PM +0100, Gordon Bos wrote:
> Hi Andrew,
> 
> There is a difference, and apparently it somewhere got solved:
> 
> From running kernel 3.17.1
> 
>      ~ # dmesg | grep spi
> 
> Running an upgraded kernel 3.18.5 with the same DT blob
> 
>      ~ # dmesg | grep spi
>    [    0.928832] m25p80 spi0.0: found m25p16, expected m25p80
>    [    0.934124] m25p80 spi0.0: m25p16 (2048 Kbytes)

So this is with your patch applied. You can see the driver is
complaining the DT says there should be a m25p80, but it actually
found a m25p16. If you remove your change, this warning will go away.

> You may therefore consider the MTD issue as solved.

Great.
 
 Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html