Call for testing: device nodes as kobjects

Call for testing: device nodes as kobjects

[Grant Likely]

Hi folks,

Could I get some testing support please? I've applied the patch that
makes device nodes into kobjects into the following branch and build
tested it on a bunch of platforms. What I need now is boot testing on
different architectures.

If you can build this tree and make sure it doesn't break your
platforms with OF support I would greatly appreciate it.

Also, if you turn off CONFIG_PROC_DEVICETREE, a symlink will be
created to the /sys/firmware  representation which should be
functionally identical. For those of you using tools that read
/proc/device-tree (like kexec), I would also appreciate if you can
test that those tools still work with CONFIG_PROC_DEVICETREE turned
off.

git://git.secretlab.ca/git/linux devicetree/experimental

Thanks,
g.

Re: Call for testing: device nodes as kobjects

[Sascha Hauer]

Hi Grant,

On Wed, Apr 17, 2013 at 04:48:22PM +0100, Grant Likely wrote:
> Hi folks,
> 
> Could I get some testing support please? I've applied the patch that
> makes device nodes into kobjects into the following branch and build
> tested it on a bunch of platforms. What I need now is boot testing on
> different architectures.
> 
> If you can build this tree and make sure it doesn't break your
> platforms with OF support I would greatly appreciate it.
> 
> Also, if you turn off CONFIG_PROC_DEVICETREE, a symlink will be
> created to the /sys/firmware  representation which should be
> functionally identical. For those of you using tools that read
> /proc/device-tree (like kexec), I would also appreciate if you can
> test that those tools still work with CONFIG_PROC_DEVICETREE turned
> off.
> 
> git://git.secretlab.ca/git/linux devicetree/experimental

I gave it a quick test on i.MX53 (qsb) and i.MX27 (phycore i.MX27).
Both seem to work just normally. I also made a

diff -urN /proc/device-tree /sys/firmware/ofw/device-tree

with CONFIG_PROC_DEVICETREE enabled (so no symlink). Indeed both
trees are identical.

So:

Tested-by: Sascha Hauer <s.hauer-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org>

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: Call for testing: device nodes as kobjects

[Grant Likely]

On Wed, Apr 17, 2013 at 6:16 PM, Sascha Hauer <s.hauer-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org> wrote:
> Hi Grant,
>
> On Wed, Apr 17, 2013 at 04:48:22PM +0100, Grant Likely wrote:
>> Hi folks,
>>
>> Could I get some testing support please? I've applied the patch that
>> makes device nodes into kobjects into the following branch and build
>> tested it on a bunch of platforms. What I need now is boot testing on
>> different architectures.
>>
>> If you can build this tree and make sure it doesn't break your
>> platforms with OF support I would greatly appreciate it.
>>
>> Also, if you turn off CONFIG_PROC_DEVICETREE, a symlink will be
>> created to the /sys/firmware  representation which should be
>> functionally identical. For those of you using tools that read
>> /proc/device-tree (like kexec), I would also appreciate if you can
>> test that those tools still work with CONFIG_PROC_DEVICETREE turned
>> off.
>>
>> git://git.secretlab.ca/git/linux devicetree/experimental
>
> I gave it a quick test on i.MX53 (qsb) and i.MX27 (phycore i.MX27).
> Both seem to work just normally. I also made a
>
> diff -urN /proc/device-tree /sys/firmware/ofw/device-tree
>
> with CONFIG_PROC_DEVICETREE enabled (so no symlink). Indeed both
> trees are identical.
>
> So:
>
> Tested-by: Sascha Hauer <s.hauer-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org>

Great, thanks!

g.

Re: Call for testing: device nodes as kobjects

[Grant Likely]

On Wed, Apr 17, 2013 at 4:48 PM, Grant Likely <grant.likely-s3s/WqlpOiPyB63q8FvJNQ@public.gmane.org> wrote:
> Hi folks,
>
> Could I get some testing support please? I've applied the patch that
> makes device nodes into kobjects into the following branch and build
> tested it on a bunch of platforms. What I need now is boot testing on
> different architectures.
>
> If you can build this tree and make sure it doesn't break your
> platforms with OF support I would greatly appreciate it.
>
> Also, if you turn off CONFIG_PROC_DEVICETREE, a symlink will be
> created to the /sys/firmware  representation which should be
> functionally identical. For those of you using tools that read
> /proc/device-tree (like kexec), I would also appreciate if you can
> test that those tools still work with CONFIG_PROC_DEVICETREE turned
> off.
>
> git://git.secretlab.ca/git/linux devicetree/experimental

Note; I just updated the branch with a build fix for PowerPC

g.

Re: Call for testing: device nodes as kobjects

[Grant Likely]

On Thu, Apr 25, 2013 at 6:11 PM, Nathan Fontenot
<nfont-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org> wrote:
> Hi Grant,
>
> I finally got some time to play around with this on powerpc/pseries.
> Everything looks the same in the device tree, though I am seeing an
> issue with of_property_update(). The routine appears to work and update
> the property but when you read the file from userspace you still get
> the same content that was present before the update.
>
> The root cause appears to be that of_update_property adds the new
> property to the property list on the device node but does not update
> the binary attribute for the file created for that property.
>
> I tried to poke around a bit but couldn't get anything to work that
> would resolve the issue. I'll try to look around some more and send
> something your way if I can get it to work.

Great, thanks for the testing Nathan. I'll take a look also.

g.