[Xenomai] RTDM and PEAK PCIe

[Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

All,

I'm spec-ing out a new machine for realtime robot control over CANbus with
Xenomai. We currently use a machine with a PCI-based CANbus card from PEAK,
but we'd like to upgrade to the PCI Express version with our new machine.
Specifically the PCAN-PCI Express dual-channel card with galvanic isolation
(PCAN-003027) <http://www.peak-system.com/PCAN-PCI-Express.206.0.html>.

>From the looks of it, the current (and older) peak_pci driver in the
kernel<http://svn.berlios.de/svnroot/repos/socketcan/trunk/kernel/2.6/drivers/net/can/sja1000/peak_pci.c>
lists
the PCIe device, but the peak_pci driver in the Xenomai
sources<http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c>is
an older version that does not.

It looks like the PCI card and the PCIe card have the same device ID
(0x0001), but it's not clear if there are any other differences that would
keep it from working.

Does anyone know about the Xenomai support for this card?

thanks!
-j

Re: [Xenomai] RTDM and PEAK PCIe

[Wolfgang Grandegger]

On 10/01/2012 07:27 PM, Jonathan Bohren wrote:
> All,
> 
> I'm spec-ing out a new machine for realtime robot control over CANbus with
> Xenomai. We currently use a machine with a PCI-based CANbus card from PEAK,
> but we'd like to upgrade to the PCI Express version with our new machine.
> Specifically the PCAN-PCI Express dual-channel card with galvanic isolation
> (PCAN-003027) <http://www.peak-system.com/PCAN-PCI-Express.206.0.html>.
> 
>>From the looks of it, the current (and older) peak_pci driver in the
> kernel<http://svn.berlios.de/svnroot/repos/socketcan/trunk/kernel/2.6/drivers/net/can/sja1000/peak_pci.c>
> lists
> the PCIe device, but the peak_pci driver in the Xenomai
> sources<http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c>is
> an older version that does not.
> 
> It looks like the PCI card and the PCIe card have the same device ID
> (0x0001), but it's not clear if there are any other differences that would
> keep it from working.
> 
> Does anyone know about the Xenomai support for this card?

There is support for both of this card in mainline Linux-CAN:

http://lxr.linux.no/#linux+v3.6/drivers/net/can/sja1000/peak_pci.c#L64

There is no special treatment for the PCIe-version of that card and
therefore it should also work with the rtcan_peak_pci driver. But
support for 4 channels is still missing.

Wolfgang.

Re: [Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

On Mon, Oct 1, 2012 at 2:42 PM, Wolfgang Grandegger <wg@grandegger.com>wrote:

> On 10/01/2012 07:27 PM, Jonathan Bohren wrote:
> > All,
> >
> > I'm spec-ing out a new machine for realtime robot control over CANbus
> with
> > Xenomai. We currently use a machine with a PCI-based CANbus card from
> PEAK,
> > but we'd like to upgrade to the PCI Express version with our new machine.
> > Specifically the PCAN-PCI Express dual-channel card with galvanic
> isolation
> > (PCAN-003027) <http://www.peak-system.com/PCAN-PCI-Express.206.0.html>.
> >
> >>From the looks of it, the current (and older) peak_pci driver in the
> > kernel<
> http://svn.berlios.de/svnroot/repos/socketcan/trunk/kernel/2.6/drivers/net/can/sja1000/peak_pci.c
> >
> > lists
> > the PCIe device, but the peak_pci driver in the Xenomai
> > sources<
> http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> >is
> > an older version that does not.
> >
> > It looks like the PCI card and the PCIe card have the same device ID
> > (0x0001), but it's not clear if there are any other differences that
> would
> > keep it from working.
> >
> > Does anyone know about the Xenomai support for this card?
>
> There is support for both of this card in mainline Linux-CAN:
>
> http://lxr.linux.no/#linux+v3.6/drivers/net/can/sja1000/peak_pci.c#L64
>
> There is no special treatment for the PCIe-version of that card and
> therefore it should also work with the rtcan_peak_pci driver. But
> support for 4 channels is still missing.
>

Great, thanks for the quick response!

-j

Re: [Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

Wolfgang,

I got the PEAK PCAN-003027 card working yesterday, but it required
adding the device ID (0x0003 in addition to 0x0001) to the module
device table `peak_pci_tbl` here:
http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c;h=53c8fef71d6d9e71dce6001941cb5922f6c977d6;hb=HEAD#l79

I've committed my changes to a fork on github under the branch name
"feature_peak_pcie" here:
https://github.com/jbohren-forks/xenomai/compare/feature_peak_pcie

Is this the appropriate channel to make a pull request, or should I do
that somewhere else?

thanks again!
-j

On Mon, Oct 1, 2012 at 3:27 PM, Jonathan Bohren
<jonathan.bohren@gmail.com> wrote:
> On Mon, Oct 1, 2012 at 2:42 PM, Wolfgang Grandegger <wg@grandegger.com>
> wrote:
>>
>> On 10/01/2012 07:27 PM, Jonathan Bohren wrote:
>> > All,
>> >
>> > I'm spec-ing out a new machine for realtime robot control over CANbus
>> > with
>> > Xenomai. We currently use a machine with a PCI-based CANbus card from
>> > PEAK,
>> > but we'd like to upgrade to the PCI Express version with our new
>> > machine.
>> > Specifically the PCAN-PCI Express dual-channel card with galvanic
>> > isolation
>> > (PCAN-003027) <http://www.peak-system.com/PCAN-PCI-Express.206.0.html>.
>> >
>> >>From the looks of it, the current (and older) peak_pci driver in the
>> >
>> > kernel<http://svn.berlios.de/svnroot/repos/socketcan/trunk/kernel/2.6/drivers/net/can/sja1000/peak_pci.c>
>> > lists
>> > the PCIe device, but the peak_pci driver in the Xenomai
>> >
>> > sources<http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c>is
>> > an older version that does not.
>> >
>> > It looks like the PCI card and the PCIe card have the same device ID
>> > (0x0001), but it's not clear if there are any other differences that
>> > would
>> > keep it from working.
>> >
>> > Does anyone know about the Xenomai support for this card?
>>
>> There is support for both of this card in mainline Linux-CAN:
>>
>> http://lxr.linux.no/#linux+v3.6/drivers/net/can/sja1000/peak_pci.c#L64
>>
>> There is no special treatment for the PCIe-version of that card and
>> therefore it should also work with the rtcan_peak_pci driver. But
>> support for 4 channels is still missing.
>
>
> Great, thanks for the quick response!
>
> -j



-- 
Jonathan Bohren
PhD Student
Dynamical Systems and Control Laboratory
Laboratory for Computational Sensing and Robotics
The Johns Hopkins University

(707) 520-4736
jbo@jhu.edu

Re: [Xenomai] RTDM and PEAK PCIe

[Wolfgang Grandegger]

On 11/27/2012 03:33 PM, Jonathan Bohren wrote:
> Wolfgang,
> 
> I got the PEAK PCAN-003027 card working yesterday, but it required
> adding the device ID (0x0003 in addition to 0x0001) to the module
> device table `peak_pci_tbl` here:
> http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c;h=53c8fef71d6d9e71dce6001941cb5922f6c977d6;hb=HEAD#l79
> 
> I've committed my changes to a fork on github under the branch name
> "feature_peak_pcie" here:
> https://github.com/jbohren-forks/xenomai/compare/feature_peak_pcie
> 
> Is this the appropriate channel to make a pull request, or should I do
> that somewhere else?

Just post the patch here on this mailing list.

Wolfgang.

Re: [Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

On Tue, Nov 27, 2012 at 11:51 AM, Wolfgang Grandegger <wg@grandegger.com> wrote:
> On 11/27/2012 03:33 PM, Jonathan Bohren wrote:
>> Wolfgang,
>>
>> I got the PEAK PCAN-003027 card working yesterday, but it required
>> adding the device ID (0x0003 in addition to 0x0001) to the module
>> device table `peak_pci_tbl` here:
>> http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c;h=53c8fef71d6d9e71dce6001941cb5922f6c977d6;hb=HEAD#l79
>>
>> I've committed my changes to a fork on github under the branch name
>> "feature_peak_pcie" here:
>> https://github.com/jbohren-forks/xenomai/compare/feature_peak_pcie
>>
>> Is this the appropriate channel to make a pull request, or should I do
>> that somewhere else?
>
> Just post the patch here on this mailing list.
>
> Wolfgang.
>


Here you go:


>From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
From: Jonathan Bohren <jbo@jhu.edu>
Date: Tue, 27 Nov 2012 09:13:11 -0500
Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027

---
 ksrc/drivers/can/sja1000/rtcan_peak_pci.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
index 53c8fef..b5ad226 100644
--- a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
+++ b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
@@ -71,13 +71,15 @@ struct rtcan_peak_pci
 #define PITA_MISC        0x1C        // miscellanoes register

 #define PEAK_PCI_VENDOR_ID   0x001C  // the PCI device and vendor IDs
-#define PEAK_PCI_DEVICE_ID   0x0001
+#define PEAK_PCI_DEVICE_ID   0x0001  // Device ID for PCI and older PCIe cards
+#define PEAK_PCIE_DEVICE_ID  0x0003  // Device ID for newer PCIe
cards (IPEH-003027)

 #define PCI_CONFIG_PORT_SIZE 0x1000  // size of the config io-memory
 #define PCI_PORT_SIZE        0x0400  // size of a channel io-memory

 static struct pci_device_id peak_pci_tbl[] = {
  {PEAK_PCI_VENDOR_ID, PEAK_PCI_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
+ {PEAK_PCI_VENDOR_ID, PEAK_PCIE_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
  { }
 };
 MODULE_DEVICE_TABLE (pci, peak_pci_tbl);
--
1.7.9.5

Re: [Xenomai] RTDM and PEAK PCIe

[Wolfgang Grandegger]

On 11/27/2012 06:08 PM, Jonathan Bohren wrote:
> On Tue, Nov 27, 2012 at 11:51 AM, Wolfgang Grandegger <wg@grandegger.com> wrote:
>> On 11/27/2012 03:33 PM, Jonathan Bohren wrote:
>>> Wolfgang,
>>>
>>> I got the PEAK PCAN-003027 card working yesterday, but it required
>>> adding the device ID (0x0003 in addition to 0x0001) to the module
>>> device table `peak_pci_tbl` here:
>>> http://git.xenomai.org/?p=xenomai-2.6.git;a=blob;f=ksrc/drivers/can/sja1000/rtcan_peak_pci.c;h=53c8fef71d6d9e71dce6001941cb5922f6c977d6;hb=HEAD#l79
>>>
>>> I've committed my changes to a fork on github under the branch name
>>> "feature_peak_pcie" here:
>>> https://github.com/jbohren-forks/xenomai/compare/feature_peak_pcie
>>>
>>> Is this the appropriate channel to make a pull request, or should I do
>>> that somewhere else?
>>
>> Just post the patch here on this mailing list.
>>
>> Wolfgang.
>>
> 
> 
> Here you go:
> 
> 
>>From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
> From: Jonathan Bohren <jbo@jhu.edu>
> Date: Tue, 27 Nov 2012 09:13:11 -0500
> Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027

Thanks, just your Signed-off-by is missing.

Wolfgang.

Re: [Xenomai] RTDM and PEAK PCIe

[Wolfgang Grandegger]

On 12/03/2012 06:28 AM, Jonathan Bohren wrote:
> Anything else I need to do for the QA process?

Ah, yes, I just realized that the mail did not go to the list. I
re-added it. Please resend the patch. You can add my
"Acked-by: Wolfgang Grandegger <wg@grandegger.com>".

Wolfgang.

> On Wed, Nov 28, 2012 at 2:53 PM, Jonathan Bohren
> <jonathan.bohren@gmail.com>wrote:
> 
>> On Wed, Nov 28, 2012 at 2:10 PM, Wolfgang Grandegger <wg@grandegger.com>
>> wrote:
>>> Thanks, just your Signed-off-by is missing.
>>
>> Ah, here you go:
>>
>>
>> From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
>> From: Jonathan Bohren <jbo@jhu.edu>
>> Date: Tue, 27 Nov 2012 09:13:11 -0500
>> Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027
>> Signed-off-by: Jonathan Bohren <jbo@jhu.edu>
>> ---
>>  ksrc/drivers/can/sja1000/rtcan_peak_pci.c |    4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
>> b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
>> index 53c8fef..b5ad226 100644
>> --- a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
>> +++ b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
>> @@ -71,13 +71,15 @@ struct rtcan_peak_pci
>>  #define PITA_MISC        0x1C        // miscellanoes register
>>
>>  #define PEAK_PCI_VENDOR_ID   0x001C  // the PCI device and vendor IDs
>> -#define PEAK_PCI_DEVICE_ID   0x0001
>> +#define PEAK_PCI_DEVICE_ID   0x0001  // Device ID for PCI and older PCIe
>> cards
>> +#define PEAK_PCIE_DEVICE_ID  0x0003  // Device ID for newer PCIe
>> cards (IPEH-003027)
>>
>>  #define PCI_CONFIG_PORT_SIZE 0x1000  // size of the config io-memory
>>  #define PCI_PORT_SIZE        0x0400  // size of a channel io-memory
>>
>>  static struct pci_device_id peak_pci_tbl[] = {
>>   {PEAK_PCI_VENDOR_ID, PEAK_PCI_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
>> 0},
>> + {PEAK_PCI_VENDOR_ID, PEAK_PCIE_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
>> 0},
>>   { }
>>  };
>>  MODULE_DEVICE_TABLE (pci, peak_pci_tbl);
>> --
>> 1.7.9.5
>>
> 
> 
> 

Re: [Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

Ok, here's the PEAK CANBus PCIE patch with all the necessary metadata:
thanks!
-j

>From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
From: Jonathan Bohren <jbo@jhu.edu>
Date: Tue, 27 Nov 2012 09:13:11 -0500
Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027
Signed-off-by: Jonathan Bohren <jbo@jhu.edu>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
---
 #define PCI_CONFIG_PORT_SIZE 0x1000  // size of the config io-memory
 ksrc/drivers/can/sja1000/rtcan_peak_pci.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
index 53c8fef..b5ad226 100644
--- a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
+++ b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
@@ -71,13 +71,15 @@ struct rtcan_peak_pci
 #define PITA_MISC        0x1C        // miscellanoes register
 #define PEAK_PCI_VENDOR_ID   0x001C  // the PCI device and vendor IDs
-#define PEAK_PCI_DEVICE_ID   0x0001
+#define PEAK_PCI_DEVICE_ID   0x0001  // Device ID for PCI and older PCIe
cards
+#define PEAK_PCIE_DEVICE_ID  0x0003  // Device ID for newer PCIe
cards (IPEH-003027)
 #define PCI_PORT_SIZE        0x0400  // size of a channel io-memory

 static struct pci_device_id peak_pci_tbl[] = {
  {PEAK_PCI_VENDOR_ID, PEAK_PCI_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
+ {PEAK_PCI_VENDOR_ID, PEAK_PCIE_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
0},
  { }
 };
 MODULE_DEVICE_TABLE (pci, peak_pci_tbl);
--

>     1.7.9.5

Re: [Xenomai] RTDM and PEAK PCIe

[Wolfgang Grandegger]

On 12/03/2012 04:43 PM, Jonathan Bohren wrote:
> Ok, here's the PEAK CANBus PCIE patch with all the necessary metadata:
> thanks!
> -j
> 
>>From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
> From: Jonathan Bohren <jbo@jhu.edu>
> Date: Tue, 27 Nov 2012 09:13:11 -0500
> Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027

Please add an empty line between the subject and the commit message.

> Signed-off-by: Jonathan Bohren <jbo@jhu.edu>
> Acked-by: Wolfgang Grandegger <wg@grandegger.com>
> ---
>  #define PCI_CONFIG_PORT_SIZE 0x1000  // size of the config io-memory

Hm, where does that line come from. Maybe the result of the missing line
above.

>  ksrc/drivers/can/sja1000/rtcan_peak_pci.c |    4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> index 53c8fef..b5ad226 100644
> --- a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> +++ b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> @@ -71,13 +71,15 @@ struct rtcan_peak_pci
>  #define PITA_MISC        0x1C        // miscellanoes register
>  #define PEAK_PCI_VENDOR_ID   0x001C  // the PCI device and vendor IDs
> -#define PEAK_PCI_DEVICE_ID   0x0001
> +#define PEAK_PCI_DEVICE_ID   0x0001  // Device ID for PCI and older PCIe
> cards
> +#define PEAK_PCIE_DEVICE_ID  0x0003  // Device ID for newer PCIe
> cards (IPEH-003027)
>  #define PCI_PORT_SIZE        0x0400  // size of a channel io-memory
> 
>  static struct pci_device_id peak_pci_tbl[] = {
>   {PEAK_PCI_VENDOR_ID, PEAK_PCI_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
> + {PEAK_PCI_VENDOR_ID, PEAK_PCIE_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
> 0},

Unfortunately, you mail client has wrapped lines. This patch will not
apply. To avoid such problems I recommend to post the patch with  "git
send-email".

Thanks,

Wolfgang.

Re: [Xenomai] RTDM and PEAK PCIe

[Jonathan Bohren]

On Dec 4, 2012 8:22 AM, "Wolfgang Grandegger" <wg@grandegger.com> wrote:

> On 12/03/2012 04:43 PM, Jonathan Bohren wrote:
> > Ok, here's the PEAK CANBus PCIE patch with all the necessary metadata:
> > thanks!
> > -j
> >
> >>From d97ee63e108c3f6eff26368af0df3ba1d1050692 Mon Sep 17 00:00:00 2001
> > From: Jonathan Bohren <jbo@jhu.edu>
> > Date: Tue, 27 Nov 2012 09:13:11 -0500
> > Subject: [PATCH] Adding device ID for PEAK PCIe card IPEH-003027
>
> Please add an empty line between the subject and the commit message.
>
> > Signed-off-by: Jonathan Bohren <jbo@jhu.edu>
> > Acked-by: Wolfgang Grandegger <wg@grandegger.com>
> > ---
> >  #define PCI_CONFIG_PORT_SIZE 0x1000  // size of the config io-memory
>
> Hm, where does that line come from. Maybe the result of the missing line
> above.
>
> >  ksrc/drivers/can/sja1000/rtcan_peak_pci.c |    4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> > b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> > index 53c8fef..b5ad226 100644
> > --- a/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> > +++ b/ksrc/drivers/can/sja1000/rtcan_peak_pci.c
> > @@ -71,13 +71,15 @@ struct rtcan_peak_pci
> >  #define PITA_MISC        0x1C        // miscellanoes register
> >  #define PEAK_PCI_VENDOR_ID   0x001C  // the PCI device and vendor IDs
> > -#define PEAK_PCI_DEVICE_ID   0x0001
> > +#define PEAK_PCI_DEVICE_ID   0x0001  // Device ID for PCI and older PCIe
> > cards
> > +#define PEAK_PCIE_DEVICE_ID  0x0003  // Device ID for newer PCIe
> > cards (IPEH-003027)
> >  #define PCI_PORT_SIZE        0x0400  // size of a channel io-memory
> >
> >  static struct pci_device_id peak_pci_tbl[] = {
> >   {PEAK_PCI_VENDOR_ID, PEAK_PCI_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
> 0},
> > + {PEAK_PCI_VENDOR_ID, PEAK_PCIE_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
> > 0},
>
> Unfortunately, you mail client has wrapped lines. This patch will not
> apply. To avoid such problems I recommend to post the patch with  "git
> send-email".
>

Ah, I was worried about that.

How about this, I've put the patch up on gist here:
https://gist.github.com/4204869

thanks!
-j