[etherlab-users] Problem reading SII configuration from slave

[Graeme Foot]

Hi Jesper,

Thanks for the patch, it was very interesting.

Unfortunately our linux distro does not have the hotplug/udev framework
required for request_firmware to work.

So, what I have done is added the ability to "request_firmware_direct"
which will read firmware files direct from the kernel.  I've attached a
patch with my additions.  (Sorry, there's also a few formatting changes
in the patch to match the code to the etherlabs coding standards.)


Theres two defines:
- REQUEST_SII_DIRECT   to enable request_firmware_direct
                       (otherwise uses request_firmware)
- SII_PATH             the path to the firmware directory

Firmware files will also require read permissions for all users, eg:
# chmod a=r ec_[VENDOR_ID]_[PRODUCT_CODE].bin


Regards,
Graeme.


-----Original Message-----
From: etherlab-users-bounces at etherlab.org
[mailto:etherlab-users-bounces at etherlab.org] On Behalf Of Jesper Smith
Sent: Tuesday, 1 November 2011 10:54
To: Graeme Foot
Cc: etherlab-users at etherlab.org
Subject: Re: [etherlab-users] Problem reading SII configuration from
slave


The patch I have is for the latest stable-1.5 tag. To use the patch, put
the 
SII information (export using TwinCAT) in 

[FIRMWARE_LOCATION]/ethercat/ec_[VENDOR ID]_[PRODUCT CODE].bin

For my distro, OpenSuSE, the firmware location is 
/lib/firmware/[KERNEL_VERSION], but this can differ between distro's. 

Regards,

Jesper


Op dinsdag 1 november 2011 10:00:04 schreef Graeme Foot:
> Hi,
> 
> Yes, I'm interested in the patch.
> 
> Thanks,
> Graeme.
> 
> 
> -----Original Message-----
> From: etherlab-users-bounces at etherlab.org
> [mailto:etherlab-users-bounces at etherlab.org] On Behalf Of Jesper Smith
> Sent: Friday, 28 October 2011 04:02
> To: etherlab-users at etherlab.org
> Subject: Re: [etherlab-users] Problem reading SII configuration from
> slave
> 
> Hello,
> 
> Because the supplier of our slaves told us that their SII information
> will
> never fit in the slave (the SII binary is 3kb and the slave only has a
> 2kb
> memory location) and we're stuck with 14 of them I've cleaned up the
> hack to
> read the SII information from a binary file and made it work in
> combination
> with other slaves.  If someone is interested I can make a patch to the
> most
> recent master code.
> 
> I've adjusted the fsm_slave_scan state machine to works as follows
> 
> - Read vendor id + product code from slave
> - Use the kernels firmware_request to load
> ec_${vendor_id}_${product_code}.bin
> /lib/firmware/${uname -r}/ethercat (or whatever path your distribution
> uses)
> 
> If firmware is found,
>         use that to load the SII information
> else
>         read SII information from the slave
> 
> 
> I think this can be used as starting point for reading XML files. I'm
> thinking
> of the following solutions
> 
> - Make a separate user space program that compiles the XML files to
SII
> binaries and stores them in /lib/firmware
> - Store the XML file in /lib/firmware and parse it in the master
module
> - Create a separate communication between the kernel and user space
that
> 
> requests slave information from a daemon/udev.
> 
> Maybe someone on this list can voice their opinion what would be the
> neatest
> way to do it.
> 
> 
> Regards,
> 
> Jesper
-------------- next part --------------
A non-text attachment was scrubbed...
Name: etherlabmaster-1.5-2124-e_load_sii_from_file_changes.patch
Type: application/octet-stream
Size: 12979 bytes
Desc: etherlabmaster-1.5-2124-e_load_sii_from_file_changes.patch
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20111101/a82d8278/attachment-0003.obj>