[Xenomai] Pre/Post patches

[Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

Hi all,

I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
armv5, but it should already work for armv6 and armv7 processors, it is
available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
by mainline, please simply send patches to the for-ipipe-3.14.0 branch.

Regards.

-- 
Gilles.

Re: [Xenomai] Pre/Post patches

[Matthew Fornero]

Looks like Xilinx hasn't merged 3.14 into their tree yet. There is
some support for Zynq in mainline, but I'm not sure how functional it
is (I've always run from the Xilinx tree). I'm still aiming for a 3.10
build (-pre/-post patches) sometime soon, but I've been completely out
of bandwidth the last month or so.

-Matt

On Sat, Apr 19, 2014 at 2:02 PM, Gilles Chanteperdrix
<gilles.chanteperdrix@xenomai.org> wrote:
>
> Hi all,
>
> I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
> armv5, but it should already work for armv6 and armv7 processors, it is
> available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
> a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
> by mainline, please simply send patches to the for-ipipe-3.14.0 branch.
>
> Regards.
>
> --
> Gilles.

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

Le 20/04/2014 20:22, Matthew Fornero a écrit :
> Looks like Xilinx hasn't merged 3.14 into their tree yet. There is
> some support for Zynq in mainline, but I'm not sure how functional it
> is (I've always run from the Xilinx tree). I'm still aiming for a 3.10
> build (-pre/-post patches) sometime soon, but I've been completely out
> of bandwidth the last month or so.

The advantage of being integrated in the mainline I-pipe tree is that
from that point I handle the kernel version upgrades, which leaves you
time to do less boring and repetitive things, such as getting the Xilinx
tree patches integrated into the mainline kernel.


-- 
Gilles.

Re: [Xenomai] Pre/Post patches

[Matthew Fornero]

No question on the advantages of getting it into the mainline I-pipe
tree. The issue is that the current state of Linux mainline
(kernel.org) is not terribly usable on Zynq AFAIK, and Xilinx doesn't
have 3.14 in their tree yet for me to do the pre/post patching.

That being said, it's still probably worthwhile for me to try a quick
build of your tree on Zynq, just to see how close it is, though from a
quick glance it looks like some important bits will be missing (e.g.
Ethernet).

-Matt

On Sun, Apr 20, 2014 at 7:00 PM, Gilles Chanteperdrix
<gilles.chanteperdrix@xenomai.org> wrote:
> Le 20/04/2014 20:22, Matthew Fornero a écrit :
>> Looks like Xilinx hasn't merged 3.14 into their tree yet. There is
>> some support for Zynq in mainline, but I'm not sure how functional it
>> is (I've always run from the Xilinx tree). I'm still aiming for a 3.10
>> build (-pre/-post patches) sometime soon, but I've been completely out
>> of bandwidth the last month or so.
>
> The advantage of being integrated in the mainline I-pipe tree is that
> from that point I handle the kernel version upgrades, which leaves you
> time to do less boring and repetitive things, such as getting the Xilinx
> tree patches integrated into the mainline kernel.
>
>
> --
> Gilles.

Re: [Xenomai] Pre/Post patches

[Matthew Fornero]

On Mon, Apr 21, 2014 at 8:35 AM, Matthew Fornero
<mfornero.lists@gmail.com> wrote:
> No question on the advantages of getting it into the mainline I-pipe
> tree. The issue is that the current state of Linux mainline
> (kernel.org) is not terribly usable on Zynq AFAIK, and Xilinx doesn't
> have 3.14 in their tree yet for me to do the pre/post patching.
>
> That being said, it's still probably worthwhile for me to try a quick
> build of your tree on Zynq, just to see how close it is, though from a
> quick glance it looks like some important bits will be missing (e.g.
> Ethernet).

Apologies for the top posting, it's the gmail default and I've been
forgetting to re-format it.

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

On 04/21/2014 02:35 PM, Matthew Fornero wrote:
> No question on the advantages of getting it into the mainline I-pipe
> tree. The issue is that the current state of Linux mainline
> (kernel.org) is not terribly usable on Zynq AFAIK, and Xilinx doesn't
> have 3.14 in their tree yet for me to do the pre/post patching.

My point being: it may be a more useful effort to add what is missing to
the mainline kernel (the ethernet driver for instance) from the Xilinx
tree and get Xenomai working on that kernel embryo which is bound to
evolve, than to continue generating pre and post patches for the Xilinx
tree. Vendor forks tend to become obsolete very quickly, and lack the
review of their patches by the kernel community.


-- 
                                                                Gilles.

Re: [Xenomai] Pre/Post patches

[Huub Van Niekerk]

>On Saturday, April 19, 2014 8:07 PM, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
 

>Hi all,

>I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
>armv5, but it should already work for armv6 and armv7 processors, it is
>available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
>a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
>by mainline, please simply send patches to the for-ipipe-3.14.0 branch.

Hi,

When looking at http://git.xenomai.org/ipipe-gch.git/?h=for-ipipe-3.14.0 , I think I see several patches but I'm not sure how to act to get a patch for the armv7 (armhf / beagle-xm).

Thank you,

hvn

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

Le 21/04/2014 08:28, Huub Van Niekerk a écrit :
>>On Saturday, April 19, 2014 8:07 PM, Gilles Chanteperdrix
> <gilles.chanteperdrix@xenomai.org> wrote:
> 
>>Hi all,
> 
>>I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
>>armv5, but it should already work for armv6 and armv7 processors, it is
>>available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
>>a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
>>by mainline, please simply send patches to the for-ipipe-3.14.0 branch.
> 
> Hi,
> 
> When looking at http://git.xenomai.org/ipipe-gch.git/?h=for-ipipe-3.14.0
> , I think I see several patches but I'm not sure how to act to get a
> patch for the armv7 (armhf / beagle-xm).

Note that the mail was intended for BeagleBone, Raspberry and Zynq. For
beagleboard, the kernel should already work, and you can wait for the
final I-pipe patch to be released. But if you want to validate the
kernel early, please go ahead, it will not hurt.

Simply checkout this tree, and use it with the "prepare-kernel.sh"
script, it will detect that the tree already contains the I-pipe
changes, and will not try and apply an I-pipe patch.

-- 
Gilles.

Re: [Xenomai] Pre/Post patches

[Huub Van Niekerk]

On Monday, April 21, 2014 10:15 AM, Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org> wrote:
 
Le 21/04/2014 08:28, Huub Van Niekerk a écrit :
>>On Saturday, April 19, 2014 8:07 PM, Gilles Chanteperdrix
> <gilles.chanteperdrix@xenomai.org> wrote:
> 
>>Hi all,
> 
>>I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
>>armv5, but it should already work for armv6 and armv7 processors, it is
>>available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
>>a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
>>by mainline, please simply send patches to the for-ipipe-3.14.0 branch.
> 
> Hi,
> 
> When looking at http://git.xenomai.org/ipipe-gch.git/?h=for-ipipe-3.14.0
> , I think I see several patches but I'm not sure how to act to get a
> patch for the armv7 (armhf / beagle-xm).

>Note that the mail was intended for BeagleBone, Raspberry and Zynq. For
>beagleboard, the kernel should already work, and you can wait for the
>final I-pipe patch to be released. But if you want to validate the
>kernel early, please go ahead, it will not hurt.

>Simply checkout this tree, and use it with the "prepare-kernel.sh"
>script, it will detect that the tree already contains the I-pipe
>changes, and will not try and apply an I-pipe patch.

Ok, I think I will wait for the final I-pipe patch then (assuming it won't be a long wait). BTW, with this be for version 2.6.3 or a new version ?

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

On 04/21/2014 02:38 PM, Huub Van Niekerk wrote:
> On Monday, April 21, 2014 10:15 AM, Gilles Chanteperdrix
> <gilles.chanteperdrix@xenomai.org> wrote:
> 
> Le 21/04/2014 08:28, Huub Van Niekerk a écrit :
>>> On Saturday, April 19, 2014 8:07 PM, Gilles Chanteperdrix
>> <gilles.chanteperdrix@xenomai.org> wrote:
>> 
>>> Hi all,
>> 
>>> I am currently validating the I-pipe patch for Linux 3.14 on
>>> armv4 and armv5, but it should already work for armv6 and armv7
>>> processors, it is available as a branch for-ipipe-3.14.0 in
>>> ipipe-gch git. Could you have a look at it on BeagleBone,
>>> Raspberry and Zynq? For the SOCs supported by mainline, please
>>> simply send patches to the for-ipipe-3.14.0 branch.
>> 
>> Hi,
>> 
>> When looking at
>> http://git.xenomai.org/ipipe-gch.git/?h=for-ipipe-3.14.0 , I think
>> I see several patches but I'm not sure how to act to get a patch
>> for the armv7 (armhf / beagle-xm).
> 
>> Note that the mail was intended for BeagleBone, Raspberry and Zynq.
>> For beagleboard, the kernel should already work, and you can wait
>> for the final I-pipe patch to be released. But if you want to
>> validate the kernel early, please go ahead, it will not hurt.
> 
>> Simply checkout this tree, and use it with the "prepare-kernel.sh" 
>> script, it will detect that the tree already contains the I-pipe 
>> changes, and will not try and apply an I-pipe patch.
> 
> Ok, I think I will wait for the final I-pipe patch then (assuming it
> won't be a long wait). BTW, with this be for version 2.6.3 or a new
> version ?

We will have to release 2.6.4.

-- 
                                                                Gilles.

Re: [Xenomai] Pre/Post patches

[Charles Steinkuehler]

On 4/19/2014 1:02 PM, Gilles Chanteperdrix wrote:
> 
> I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
> armv5, but it should already work for armv6 and armv7 processors, it is
> available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
> a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
> by mainline, please simply send patches to the for-ipipe-3.14.0 branch.

I'm still working on building a kernel for the BeagleBone Black.

I tried the standard method of generating the ipipe patch and applying
it to the BeagleBone patched kernel source.  The first problem I
encountered was an unknown architecture error for the
net/ethernet/cadence/at91_ether.c driver.  I edited the genpatches.sh
script and forced this to x86 (since I'm pretty sure I don't need this
driver for the BeagleBone).  Note the patch set for the 3.14 BeagleBone
kernel is mostly device-tree changes and about four fairly simple
back-ported fixes.

The next problem comes when building the kernel, I get an undefined
warning from the raid456.ko module (see below).  I'll keep plugging
away, but any advice would be appreciated.

Kernel build error:
  <snip>
  Building modules, stage 2.
  OBJCOPY arch/arm/boot/Image
  Kernel: arch/arm/boot/Image is ready
  AS      arch/arm/boot/compressed/head.o
  XZKERN  arch/arm/boot/compressed/piggy.xzkern
  CC      arch/arm/boot/compressed/misc.o
  CC      arch/arm/boot/compressed/decompress.o
  CC      arch/arm/boot/compressed/string.o
  AS      arch/arm/boot/compressed/hyp-stub.o
  AS      arch/arm/boot/compressed/lib1funcs.o
  MODPOST 1950 modules
ERROR: "__bad_lock_type" [drivers/md/raid456.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2
make: *** Waiting for unfinished jobs....
  AS      arch/arm/boot/compressed/piggy.xzkern.o
  LD      arch/arm/boot/compressed/vmlinux
  OBJCOPY arch/arm/boot/zImage
  Kernel: arch/arm/boot/zImage is ready
charles@cubox:~/linux-dev-3.14$


-- 
Charles Steinkuehler
charles@steinkuehler.net

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
URL: <http://www.xenomai.org/pipermail/xenomai/attachments/20140421/49f8aee0/attachment.sig>

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

On 04/21/2014 09:24 PM, Charles Steinkuehler wrote:
> On 4/19/2014 1:02 PM, Gilles Chanteperdrix wrote:
>>
>> I am currently validating the I-pipe patch for Linux 3.14 on armv4 and
>> armv5, but it should already work for armv6 and armv7 processors, it is
>> available as a branch for-ipipe-3.14.0 in ipipe-gch git. Could you have
>> a look at it on BeagleBone, Raspberry and Zynq? For the SOCs supported
>> by mainline, please simply send patches to the for-ipipe-3.14.0 branch.
> 
> I'm still working on building a kernel for the BeagleBone Black.
> 
> I tried the standard method of generating the ipipe patch and applying
> it to the BeagleBone patched kernel source.  The first problem I
> encountered was an unknown architecture error for the
> net/ethernet/cadence/at91_ether.c driver.  I edited the genpatches.sh
> script and forced this to x86 (since I'm pretty sure I don't need this
> driver for the BeagleBone).

Well, at91 are arms, will fix.

> Note the patch set for the 3.14 BeagleBone
> kernel is mostly device-tree changes and about four fairly simple
> back-ported fixes.
> 
> The next problem comes when building the kernel, I get an undefined
> warning from the raid456.ko module (see below).  I'll keep plugging
> away, but any advice would be appreciated.

probably something wrong with the configuration. Why do you need raid 4,
5, 6 anyway? In a first step, you should work with a standard
configuraiton, such as omap2plus_defconfig, or better, if you have it,
with a minimal configuration suited for your board, which will avoid:

> (...)
>   MODPOST 1950 modules

and build a kernel in just a few minutes.

-- 
                                                                Gilles.

Re: [Xenomai] Pre/Post patches

[Charles Steinkuehler]

On 4/21/2014 2:33 PM, Gilles Chanteperdrix wrote:
> On 04/21/2014 09:24 PM, Charles Steinkuehler wrote:
>>>
>> The next problem comes when building the kernel, I get an undefined
>> warning from the raid456.ko module (see below).  I'll keep plugging
>> away, but any advice would be appreciated.
> 
> probably something wrong with the configuration. Why do you need raid 4,
> 5, 6 anyway? 

I don't, of course, it's just what the config from RCN's build scripts
has setup for by default.

> In a first step, you should work with a standard
> configuraiton, such as omap2plus_defconfig, or better, if you have it,
> with a minimal configuration suited for your board, which will avoid:
> 
>> (...)
>>   MODPOST 1950 modules
> 
> and build a kernel in just a few minutes.

As mentioned, I'm still plugging along, and mostly learning as I go.

Thanks for the breadcrumbs!  :)

-- 
Charles Steinkuehler
charles@steinkuehler.net

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
URL: <http://www.xenomai.org/pipermail/xenomai/attachments/20140421/296c97ac/attachment.sig>

Re: [Xenomai] Pre/Post patches

[Gilles Chanteperdrix]

On 04/21/2014 09:39 PM, Charles Steinkuehler wrote:
> On 4/21/2014 2:33 PM, Gilles Chanteperdrix wrote:
>> On 04/21/2014 09:24 PM, Charles Steinkuehler wrote:
>>>> 
>>> The next problem comes when building the kernel, I get an
>>> undefined warning from the raid456.ko module (see below).  I'll
>>> keep plugging away, but any advice would be appreciated.
>> 
>> probably something wrong with the configuration. Why do you need
>> raid 4, 5, 6 anyway?
> 
> I don't, of course, it's just what the config from RCN's build
> scripts has setup for by default.
> 
>> In a first step, you should work with a standard configuraiton,
>> such as omap2plus_defconfig, or better, if you have it, with a
>> minimal configuration suited for your board, which will avoid:
>> 
>>> (...) MODPOST 1950 modules
>> 
>> and build a kernel in just a few minutes.
> 
> As mentioned, I'm still plugging along, and mostly learning as I
> go.
> 
> Thanks for the breadcrumbs!  :)

Our users have evolved. Somewhere along the road, Linux users stopped
being used to configuring their kernel, and we did not notice. I hope
to be able to propose something soon to help this situation.

Anyway, start with an unpatched kernel, tune the configuration until
you get a kernel which compiles and runs.

Then merge with the I-pipe tree. If the same configuration (minus the
usual forbidden options) does not compile, then there may be a problem
in the I-pipe tree.

Something to definitely disable for the time being is context tracking
(and so CONFIG_NOHZ_FULL).


-- 
                                                                Gilles.