[Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 513 bytes --]

Hello everyone,

I finally got my PCI CAN card today. It's an Advantech 1680 and provides, as
far as I'm informed, a memory mapped SJA1000 controller.
I´m trying to load the xeno_rtcan_mem module now, but I'm not sure what
parameters I should pass. More specific: What value(s) should I give to the
'mem' parameter? Is this an I/O address I found in /proc/pci under the 1680
device?

I might even be completely off-track. Please help me get back on.

Many thanks in advance!

Best regards,

Frits

[-- Attachment #2: Type: text/html, Size: 561 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 948 bytes --]

Frits de Klark wrote:
> Hello everyone,
> 
> I finally got my PCI CAN card today. It's an Advantech 1680 and
> provides, as
> far as I'm informed, a memory mapped SJA1000 controller.
> I´m trying to load the xeno_rtcan_mem module now, but I'm not sure what
> parameters I should pass. More specific: What value(s) should I give to the
> 'mem' parameter? Is this an I/O address I found in /proc/pci under the 1680
> device?

I'm not sure that this will work, but you can give it a try in any case.
Do you have a reference driver for that board? Or some manual describing
the mapping of the SJA1000 into the PCI address space?

AFAIK, Wolfgang is working on a generic SJA1000 PCI driver. So far we
only have a driver for the PEAK PCI adapter, but that one is going to be
generalised. So you could become a first-grade test candidate for
another piece of hardware supported by that driver. :)

Wolfgang, what the status?

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Wolfgang Grandegger]

Jan Kiszka wrote:
> Frits de Klark wrote:
>> Hello everyone,
>>
>> I finally got my PCI CAN card today. It's an Advantech 1680 and
>> provides, as
>> far as I'm informed, a memory mapped SJA1000 controller.
>> I´m trying to load the xeno_rtcan_mem module now, but I'm not sure what
>> parameters I should pass. More specific: What value(s) should I give to the
>> 'mem' parameter? Is this an I/O address I found in /proc/pci under the 1680
>> device?
> 
> I'm not sure that this will work, but you can give it a try in any case.
> Do you have a reference driver for that board? Or some manual describing
> the mapping of the SJA1000 into the PCI address space?

It was not working for the IXXAT PCI card, at least (but I have not yet 
understodd why). Check with "$ lspci -vv" how the device is mapped and 
what IRQ number is used. The right thing is to implement a PCI driver.

> AFAIK, Wolfgang is working on a generic SJA1000 PCI driver. So far we
> only have a driver for the PEAK PCI adapter, but that one is going to be
> generalised. So you could become a first-grade test candidate for
> another piece of hardware supported by that driver. :)
> 
> Wolfgang, what the status?

As already mention, I'm currently working on a driver for the IXXAT PCI 
card. As I see it, a generic driver is not feasible as the mapping, the 
recognition of the second channel and the IRQ handling is quite different.

Is there a Linux driver on the CRROM your got with your Advantech 1680 
PCI card? That would help a lot.

Wolfgang.

Wolfgang.

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 2881 bytes --]

Hi,

there is indeed a driver for the 2.4.x kernel series (including source), but
onfortunately also a BIG FAT copyright statement at the heading of the
sources. Something like:
#############################################################################

//
*****************************************************************************

//                  Copyright ( c ) 2003, Advantech Automation Corp.
//      THIS IS AN UNPUBLISHED WORK CONTAINING CONFIDENTIAL AND PROPRIETARY
//               INFORMATION WHICH IS THE PROPERTY OF ADVANTECH AUTOMATION
CORP.
//
//    ANY DISCLOSURE, USE, OR REPRODUCTION, WITHOUT WRITTEN AUTHORIZATION
FROM
//               ADVANTECH AUTOMATION CORP., IS STRICTLY PROHIBITED.
//
*****************************************************************************

//
#############################################################################

So, I'll have to inquire them first before I (and the company I work for)
take the risk of disclosure.
I'll get back on this as soon as I can. If I can be of any help in the
meantime, I'll be glad to hear from you.

Thank for your time and interest so far.

Frits

On 11/16/06, Wolfgang Grandegger <wg@domain.hid> wrote:
>
> Jan Kiszka wrote:
> > Frits de Klark wrote:
> >> Hello everyone,
> >>
> >> I finally got my PCI CAN card today. It's an Advantech 1680 and
> >> provides, as
> >> far as I'm informed, a memory mapped SJA1000 controller.
> >> I´m trying to load the xeno_rtcan_mem module now, but I'm not sure what
> >> parameters I should pass. More specific: What value(s) should I give to
> the
> >> 'mem' parameter? Is this an I/O address I found in /proc/pci under the
> 1680
> >> device?
> >
> > I'm not sure that this will work, but you can give it a try in any case.
> > Do you have a reference driver for that board? Or some manual describing
> > the mapping of the SJA1000 into the PCI address space?
>
> It was not working for the IXXAT PCI card, at least (but I have not yet
> understodd why). Check with "$ lspci -vv" how the device is mapped and
> what IRQ number is used. The right thing is to implement a PCI driver.
>
> > AFAIK, Wolfgang is working on a generic SJA1000 PCI driver. So far we
> > only have a driver for the PEAK PCI adapter, but that one is going to be
> > generalised. So you could become a first-grade test candidate for
> > another piece of hardware supported by that driver. :)
> >
> > Wolfgang, what the status?
>
> As already mention, I'm currently working on a driver for the IXXAT PCI
> card. As I see it, a generic driver is not feasible as the mapping, the
> recognition of the second channel and the IRQ handling is quite different.
>
> Is there a Linux driver on the CRROM your got with your Advantech 1680
> PCI card? That would help a lot.
>
> Wolfgang.
>
> Wolfgang.
>
>

[-- Attachment #2: Type: text/html, Size: 3618 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Wolfgang Grandegger]

Frits de Klark wrote:
> Hi,
> 
> there is indeed a driver for the 2.4.x kernel series (including source), 
> but onfortunately also a BIG FAT copyright statement at the heading of 
> the sources. Something like:
> ############################################################################# 
> 
> // 
> ***************************************************************************** 
> 
> //                  Copyright ( c ) 2003, Advantech Automation Corp.
> //      THIS IS AN UNPUBLISHED WORK CONTAINING CONFIDENTIAL AND PROPRIETARY
> //               INFORMATION WHICH IS THE PROPERTY OF ADVANTECH 
> AUTOMATION CORP.
> //
> //    ANY DISCLOSURE, USE, OR REPRODUCTION, WITHOUT WRITTEN 
> AUTHORIZATION FROM
> //               ADVANTECH AUTOMATION CORP., IS STRICTLY PROHIBITED.
> // 
> ***************************************************************************** 
> 
> // 
> #############################################################################
> 
> So, I'll have to inquire them first before I (and the company I work 
> for) take the risk of disclosure.
> I'll get back on this as soon as I can. If I can be of any help in the 
> meantime, I'll be glad to hear from you.

Well, it might be simpler to buy an already supported CAN PCI card.

> Thank for your time and interest so far.

Good luck.

Wolfgang.

> Frits
> 
> On 11/16/06, *Wolfgang Grandegger* <wg@domain.hid
> <mailto:wg@domain.hid>> wrote:
> 
>     Jan Kiszka wrote:
>      > Frits de Klark wrote:
>      >> Hello everyone,
>      >>
>      >> I finally got my PCI CAN card today. It's an Advantech 1680 and
>      >> provides, as
>      >> far as I'm informed, a memory mapped SJA1000 controller.
>      >> I´m trying to load the xeno_rtcan_mem module now, but I'm not
>     sure what
>      >> parameters I should pass. More specific: What value(s) should I
>     give to the
>      >> 'mem' parameter? Is this an I/O address I found in /proc/pci
>     under the 1680
>      >> device?
>      >
>      > I'm not sure that this will work, but you can give it a try in
>     any case.
>      > Do you have a reference driver for that board? Or some manual
>     describing
>      > the mapping of the SJA1000 into the PCI address space?
> 
>     It was not working for the IXXAT PCI card, at least (but I have not yet
>     understodd why). Check with "$ lspci -vv" how the device is mapped and
>     what IRQ number is used. The right thing is to implement a PCI driver.
> 
>      > AFAIK, Wolfgang is working on a generic SJA1000 PCI driver. So far we
>      > only have a driver for the PEAK PCI adapter, but that one is
>     going to be
>      > generalised. So you could become a first-grade test candidate for
>      > another piece of hardware supported by that driver. :)
>      >
>      > Wolfgang, what the status?
> 
>     As already mention, I'm currently working on a driver for the IXXAT PCI
>     card. As I see it, a generic driver is not feasible as the mapping, the
>     recognition of the second channel and the IRQ handling is quite
>     different.
> 
>     Is there a Linux driver on the CRROM your got with your Advantech 1680
>     PCI card? That would help a lot.
> 
>     Wolfgang.
> 
>     Wolfgang.
> 
> 

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 4045 bytes --]

Hi,

to get back on my earlier question: IF the memory mapping might work for me,
how should I call it?
The PCI card has two CAN ports. Here's (a part of) the output of lspci -vv:

Interrupt: pin A routed to IRQ 11
Region 0: Memory at e1003000 (32-bit, non prefetchable) [size=128]
Region 2: I/O ports at d800 [size=128]
Region 3: I/O ports at dc00 [size=128]
Region 4: I/O ports at e000 [size=4]

What should the insmod command to try if it might work?

Thanks again.

Frits

On 11/16/06, Wolfgang Grandegger <wg@domain.hid> wrote:
>
> Frits de Klark wrote:
> > Hi,
> >
> > there is indeed a driver for the 2.4.x kernel series (including source),
> > but onfortunately also a BIG FAT copyright statement at the heading of
> > the sources. Something like:
> >
> #############################################################################
> >
> > //
> >
> *****************************************************************************
> >
> > //                  Copyright ( c ) 2003, Advantech Automation Corp.
> > //      THIS IS AN UNPUBLISHED WORK CONTAINING CONFIDENTIAL AND
> PROPRIETARY
> > //               INFORMATION WHICH IS THE PROPERTY OF ADVANTECH
> > AUTOMATION CORP.
> > //
> > //    ANY DISCLOSURE, USE, OR REPRODUCTION, WITHOUT WRITTEN
> > AUTHORIZATION FROM
> > //               ADVANTECH AUTOMATION CORP., IS STRICTLY PROHIBITED.
> > //
> >
> *****************************************************************************
> >
> > //
> >
> #############################################################################
> >
> > So, I'll have to inquire them first before I (and the company I work
> > for) take the risk of disclosure.
> > I'll get back on this as soon as I can. If I can be of any help in the
> > meantime, I'll be glad to hear from you.
>
> Well, it might be simpler to buy an already supported CAN PCI card.
>
> > Thank for your time and interest so far.
>
> Good luck.
>
> Wolfgang.
>
> > Frits
> >
> > On 11/16/06, *Wolfgang Grandegger* <wg@domain.hid
> > <mailto:wg@domain.hid>> wrote:
> >
> >     Jan Kiszka wrote:
> >      > Frits de Klark wrote:
> >      >> Hello everyone,
> >      >>
> >      >> I finally got my PCI CAN card today. It's an Advantech 1680 and
> >      >> provides, as
> >      >> far as I'm informed, a memory mapped SJA1000 controller.
> >      >> I´m trying to load the xeno_rtcan_mem module now, but I'm not
> >     sure what
> >      >> parameters I should pass. More specific: What value(s) should I
> >     give to the
> >      >> 'mem' parameter? Is this an I/O address I found in /proc/pci
> >     under the 1680
> >      >> device?
> >      >
> >      > I'm not sure that this will work, but you can give it a try in
> >     any case.
> >      > Do you have a reference driver for that board? Or some manual
> >     describing
> >      > the mapping of the SJA1000 into the PCI address space?
> >
> >     It was not working for the IXXAT PCI card, at least (but I have not
> yet
> >     understodd why). Check with "$ lspci -vv" how the device is mapped
> and
> >     what IRQ number is used. The right thing is to implement a PCI
> driver.
> >
> >      > AFAIK, Wolfgang is working on a generic SJA1000 PCI driver. So
> far we
> >      > only have a driver for the PEAK PCI adapter, but that one is
> >     going to be
> >      > generalised. So you could become a first-grade test candidate for
> >      > another piece of hardware supported by that driver. :)
> >      >
> >      > Wolfgang, what the status?
> >
> >     As already mention, I'm currently working on a driver for the IXXAT
> PCI
> >     card. As I see it, a generic driver is not feasible as the mapping,
> the
> >     recognition of the second channel and the IRQ handling is quite
> >     different.
> >
> >     Is there a Linux driver on the CRROM your got with your Advantech
> 1680
> >     PCI card? That would help a lot.
> >
> >     Wolfgang.
> >
> >     Wolfgang.
> >
> >
>
>

[-- Attachment #2: Type: text/html, Size: 6030 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Sebastian Smolorz]

Frits de Klark wrote:
> Hi,
>
> to get back on my earlier question: IF the memory mapping might work for
> me, how should I call it?
> The PCI card has two CAN ports. Here's (a part of) the output of lspci -vv:
>
> Interrupt: pin A routed to IRQ 11
> Region 0: Memory at e1003000 (32-bit, non prefetchable) [size=128]
> Region 2: I/O ports at d800 [size=128]
> Region 3: I/O ports at dc00 [size=128]

Wild guess: Could it be that you can use io=0xd800,0xdc00 (Not memory mapped)?

--
Sebastian

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 872 bytes --]

Frits de Klark wrote:
> Hi,
> 
> to get back on my earlier question: IF the memory mapping might work for
> me,
> how should I call it?
> The PCI card has two CAN ports. Here's (a part of) the output of lspci -vv:
> 
> Interrupt: pin A routed to IRQ 11
> Region 0: Memory at e1003000 (32-bit, non prefetchable) [size=128]
> Region 2: I/O ports at d800 [size=128]
> Region 3: I/O ports at dc00 [size=128]
> Region 4: I/O ports at e000 [size=4]
> 
> What should the insmod command to try if it might work?

You could try handing region 2 or 3 to the rtcan_isa driver or region 0
to rtcan_mem. Maybe your adapter actually does 1:1 mapping of the
SJA1000 address space (it looks like). I just wonder what region 4 is
doing. Maybe it controls the content of region 0, which may either hold
the first or the second CAN controller. Just a theory...

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 930 bytes --]

So,

On 11/17/06, Jan Kiszka <jan.kiszka@domain.hid> wrote:
>
>
> You could try handing region 2 or 3 to the rtcan_isa driver or region 0
> to rtcan_mem. Maybe your adapter actually does 1:1 mapping of the
> SJA1000 address space (it looks like). I just wonder what region 4 is
> doing. Maybe it controls the content of region 0, which may either hold
> the first or the second CAN controller. Just a theory...
>
> Jan



I entered the following command:

modprobe  xeno_rtcan_isa io=55296,56320 irq=11

and the following output appeared:

RTCAN SJA1000 driver registered
rtcan: registered rtcan0
rtcan: registered rtcan1

and shortly thereafter:

Xenomai: xnintr_irq_handler: IRQ0 not handled. Disabling IRQ line.

The first part seems quite good to me. I don't know what to think of the
'disabling IRQ line' message.

I have to get hold of a CAN cable, so I can actually test if it works. I'll
get back on this one soon.

Frits

[-- Attachment #2: Type: text/html, Size: 1316 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 693 bytes --]

Frits de Klark wrote:
> So,
> 
> On 11/17/06, Jan Kiszka <jan.kiszka@domain.hid> wrote:
>>
>>
>> You could try handing region 2 or 3 to the rtcan_isa driver or region 0
>> to rtcan_mem. Maybe your adapter actually does 1:1 mapping of the
>> SJA1000 address space (it looks like). I just wonder what region 4 is
>> doing. Maybe it controls the content of region 0, which may either hold
>> the first or the second CAN controller. Just a theory...
>>
>> Jan
> 
> 
> 
> I entered the following command:
> 
> modprobe  xeno_rtcan_isa io=55296,56320 irq=11

Try "modprobe xeno_rtcan_isa io=55296,56320 irq=11,11" instead to assign
the irq of the second port as well.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 710 bytes --]

On 11/17/06, Jan Kiszka <jan.kiszka@domain.hid> wrote:
>
> > I entered the following command:
> >
> > modprobe  xeno_rtcan_isa io=55296,56320 irq=11
>
> Try "modprobe xeno_rtcan_isa io=55296,56320 irq=11,11" instead to assign
> the irq of the second port as well.
>
> Jan
>
>
This (two times irq 11) fails with the following output (only the last part
actually):

rtcan: registered rtcan0
ERROR! IRQ 11 busy or invalid (code=-16)!
ERROR -16 while trying to register SJA1000 device!
Unregistering SJA1000 device rtcan0
RTCAN: unregistered rtcan0
FATAL: Error inserting xeno_rtcan_isa (/lib/modules/../xeno_rtcan_isa.ko):
Device or resource busy

Then it just returns to the shell.

Any ideas about this?

Frits

[-- Attachment #2: Type: text/html, Size: 1090 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 829 bytes --]

Frits de Klark wrote:
> On 11/17/06, Jan Kiszka <jan.kiszka@domain.hid> wrote:
>>
>> > I entered the following command:
>> >
>> > modprobe  xeno_rtcan_isa io=55296,56320 irq=11
>>
>> Try "modprobe xeno_rtcan_isa io=55296,56320 irq=11,11" instead to assign
>> the irq of the second port as well.
>>
>> Jan
>>
>>
> This (two times irq 11) fails with the following output (only the last part
> actually):
> 
> rtcan: registered rtcan0
> ERROR! IRQ 11 busy or invalid (code=-16)!
> ERROR -16 while trying to register SJA1000 device!
> Unregistering SJA1000 device rtcan0
> RTCAN: unregistered rtcan0
> FATAL: Error inserting xeno_rtcan_isa (/lib/modules/../xeno_rtcan_isa.ko):
> Device or resource busy
> 
> Then it just returns to the shell.
> 
> Any ideas about this?

CONFIG_XENO_OPT_SHIRQ_*

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 448 bytes --]

On 11/17/06, Jan Kiszka <jan.kiszka@domain.hid> wrote:
>
>
> CONFIG_XENO_OPT_SHIRQ_*
>
> Jan



Ah yes. That works fine :) I´m sorry I didn't figure that out myself. I must
say that I'm REALLY glad that you guys are so responsive on this
mailinglist. Thanks a thousand times for the answers up till now!

I'll have myself a cable within an hour or so. I'll let you know how that
works out.

Best regards and thanks again!

Frits

[-- Attachment #2: Type: text/html, Size: 771 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Frits de Klark]

[-- Attachment #1: Type: text/plain, Size: 308 bytes --]

Hello,

Well, my first tests look very promising. Using your rtcansend and rtcanrecv
utilities and a cable between port 1 and port 2 on my card, the messages
arrive perfectly.

I'm off to do some more testing. If I run into any strange situations, I'll
let you know.

Thanks so far for all your help!

Frits

[-- Attachment #2: Type: text/html, Size: 345 bytes --]

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Wolfgang Grandegger]

Jan Kiszka wrote:
> Frits de Klark wrote:
>> Hi,
>>
>> to get back on my earlier question: IF the memory mapping might work for
>> me,
>> how should I call it?
>> The PCI card has two CAN ports. Here's (a part of) the output of lspci -vv:
>>
>> Interrupt: pin A routed to IRQ 11
>> Region 0: Memory at e1003000 (32-bit, non prefetchable) [size=128]
>> Region 2: I/O ports at d800 [size=128]
>> Region 3: I/O ports at dc00 [size=128]
>> Region 4: I/O ports at e000 [size=4]
>>
>> What should the insmod command to try if it might work?
> 
> You could try handing region 2 or 3 to the rtcan_isa driver or region 0
> to rtcan_mem. Maybe your adapter actually does 1:1 mapping of the
> SJA1000 address space (it looks like). I just wonder what region 4 is
> doing. Maybe it controls the content of region 0, which may either hold
> the first or the second CAN controller. Just a theory...

To check that, load the mem driver with:

# insmod xeno_rtcan_mem.o mem=0xe1003000 irq=11

maybe there are SJA1000 registers as well. But the extra registers might 
also be there to do a reset, or report some IRQ status.

Wolfgang.

Re: [Xenomai-help] xeno_rtcan_mem.ko, what are it's parameters?

[Wolfgang Grandegger]

Frits de Klark wrote:
> Hello,
> 
> Well, my first tests look very promising. Using your rtcansend and 
> rtcanrecv utilities and a cable between port 1 and port 2 on my card, 
> the messages arrive perfectly.
> 
> I'm off to do some more testing. If I run into any strange situations, 
> I'll let you know.
> 
> Thanks so far for all your help!

This already looks very good, indeed. The right thing now would be to 
integrate that into a PCI driver. Should not be that difficult.

Wolfgang.