RESEND: Generating interrupts from a USB device driver?

RESEND: Generating interrupts from a USB device driver?

[Daniel Santos]

Hello guys.  I didn't get a response the last time so hopefully with 
3.11 out I'll get one this time.

I need to be able to generate interrupts from a USB device driver while 
servicing the complete() function of an interrupt URB. While I realize 
that this may seem strange, the purpose is for a USB to SPI/GPIO bridge 
chip (the MCP2210). When something happens on the remote device where a 
chip is expecting it's interrupt out pin to trigger an interrupt on some 
local (to the board) microcontroller, the MCP2210 instead receives that 
signal and communicates it to the host the next time it's queried. This 
is the interrupt that I need to, in effect propagate locally. Since my 
spi_master and gpio_chip are all functioning now, this is the last step 
to get one of my spi protocol drivers working correctly.

I've been reading up on Documentation/IRQ-domain.txt, but IRQ 
controllers are all new to me and there's a lot for me to learn, so I 
just want to make sure I'm going down the correct path. Is the correct 
method to create an IRQ domain with irq_domain_add_linear() or some 
such? I can figure it out eventually, but if somebody can point me in 
the right direction, it can save me a WHOLE lot of time.

Thanks in advance
Daniel

Re: RESEND: Generating interrupts from a USB device driver?

[Greg KH]

On Mon, Sep 02, 2013 at 05:46:58PM -0500, Daniel Santos wrote:
> Hello guys.  I didn't get a response the last time so hopefully with 
> 3.11 out I'll get one this time.
> 
> I need to be able to generate interrupts from a USB device driver while 
> servicing the complete() function of an interrupt URB.

No you don't :)

> While I realize that this may seem strange, the purpose is for a USB
> to SPI/GPIO bridge chip (the MCP2210). When something happens on the
> remote device where a chip is expecting it's interrupt out pin to
> trigger an interrupt on some local (to the board) microcontroller, the
> MCP2210 instead receives that signal and communicates it to the host
> the next time it's queried. This is the interrupt that I need to, in
> effect propagate locally. Since my spi_master and gpio_chip are all
> functioning now, this is the last step to get one of my spi protocol
> drivers working correctly.

Just pass the data up the spi stack in your interrupt endpoint handler.
No need to try to create a "real" interrupt.  There are other USB SPI
drivers that should give you the idea of how to do it.

thanks,

greg k-h

Re: RESEND: Generating interrupts from a USB device driver?

[Daniel Santos]

On 09/02/2013 06:07 PM, Greg KH wrote:
> On Mon, Sep 02, 2013 at 05:46:58PM -0500, Daniel Santos wrote:
>> Hello guys.  I didn't get a response the last time so hopefully with
>> 3.11 out I'll get one this time.
>>
>> I need to be able to generate interrupts from a USB device driver while
>> servicing the complete() function of an interrupt URB.
> No you don't :)
>
>> While I realize that this may seem strange, the purpose is for a USB
>> to SPI/GPIO bridge chip (the MCP2210). When something happens on the
>> remote device where a chip is expecting it's interrupt out pin to
>> trigger an interrupt on some local (to the board) microcontroller, the
>> MCP2210 instead receives that signal and communicates it to the host
>> the next time it's queried. This is the interrupt that I need to, in
>> effect propagate locally. Since my spi_master and gpio_chip are all
>> functioning now, this is the last step to get one of my spi protocol
>> drivers working correctly.
> Just pass the data up the spi stack in your interrupt endpoint handler.
> No need to try to create a "real" interrupt.  There are other USB SPI
> drivers that should give you the idea of how to do it.
Thanks for your response! I haven't been able to find these drivers, can 
you please point me to one? I guess I don't know the "spi stack" well 
enough to know how to propagate that notification up to the driver for 
the spi device (let alone that it was called a "stack" :)

Thanks!!
Daniel
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: RESEND: Generating interrupts from a USB device driver?

[Greg KH]

On Mon, Sep 02, 2013 at 08:40:47PM -0500, Daniel Santos wrote:
> On 09/02/2013 06:07 PM, Greg KH wrote:
> >On Mon, Sep 02, 2013 at 05:46:58PM -0500, Daniel Santos wrote:
> >>Hello guys.  I didn't get a response the last time so hopefully with
> >>3.11 out I'll get one this time.
> >>
> >>I need to be able to generate interrupts from a USB device driver while
> >>servicing the complete() function of an interrupt URB.
> >No you don't :)
> >
> >>While I realize that this may seem strange, the purpose is for a USB
> >>to SPI/GPIO bridge chip (the MCP2210). When something happens on the
> >>remote device where a chip is expecting it's interrupt out pin to
> >>trigger an interrupt on some local (to the board) microcontroller, the
> >>MCP2210 instead receives that signal and communicates it to the host
> >>the next time it's queried. This is the interrupt that I need to, in
> >>effect propagate locally. Since my spi_master and gpio_chip are all
> >>functioning now, this is the last step to get one of my spi protocol
> >>drivers working correctly.
> >Just pass the data up the spi stack in your interrupt endpoint handler.
> >No need to try to create a "real" interrupt.  There are other USB SPI
> >drivers that should give you the idea of how to do it.
> Thanks for your response! I haven't been able to find these drivers,
> can you please point me to one? I guess I don't know the "spi stack"
> well enough to know how to propagate that notification up to the
> driver for the spi device (let alone that it was called a "stack" :)

Hm, I thought we used to have some of them, I guess people have been
saying they would write a driver for this type of hardware for a long
time now :(

Anyway, look at the spi core, I think you want to tie into the
spi_new_device() call in your usb driver, and start sending/receiving
data through the SPI interfaces the spi core provides.

The Linux SPI mailing list should be able to help you out a lot more
here than I can.

greg k-h

Re: RESEND: Generating interrupts from a USB device driver?

[Daniel Santos]

On 09/03/2013 12:59 PM, Greg KH wrote:
> Hm, I thought we used to have some of them, I guess people have been
> saying they would write a driver for this type of hardware for a long
> time now :(

Well, on the bright side, we have one now, albeit in a "this is my first 
driver that nobody has reviewed yet" state. :)  It works (even on RPi -- 
well, mostly).  As-is, it has a fair TODO & fix list, which I'm sure 
will grow rapidly with some review. I hope that as I learn the kernel 
better, I can generalize parts of this into some re-usable lib for other 
USB-to-SPI/I2C/GPIO drivers.

https://github.com/daniel-santos/mcp2210-linux/

> Anyway, look at the spi core, I think you want to tie into the
> spi_new_device() call in your usb driver, and start sending/receiving
> data through the SPI interfaces the spi core provides.

I'm actually using the alternative to that call, which is the 
spi_alloc_device() / spi_add_device() pair as I'm not using struct 
spi_board_info at all since it doesn't (currently) have even half of the 
fields I need for each device. However, struct spi_device does have the 
two void * fields controller_data and controller_state -- I suppose I 
can use those to pass a pointer to a add_notify() type of function.  I 
was just trying to seek a more generic, re-usable mechanism, but this 
should at least work for now.

> The Linux SPI mailing list should be able to help you out a lot more
> here than I can.
>
> greg k-h

Thanks again for your help!

Re: RESEND: Generating interrupts from a USB device driver? (USB to SPI/GPIO bridge)

[Daniel Santos]

On 09/02/2013 06:07 PM, Greg KH wrote:
> On Mon, Sep 02, 2013 at 05:46:58PM -0500, Daniel Santos wrote:
>> Hello guys.  I didn't get a response the last time so hopefully with
>> 3.11 out I'll get one this time.
>>
>> I need to be able to generate interrupts from a USB device driver while
>> servicing the complete() function of an interrupt URB.
> No you don't :)

Hmm, am I wrong that if I want my gpios to be poll()-able from userspace 
(via /sys/class/gpio/gpio<num>/value), that I must generate an IRQ and 
populate my .to_irq function of gpio_chip? Or is it cleaner to find the 
dirent and directly call sysfs_notify_dirent() to facilitate this 
behavior to userspace?

Of course, this will wont enable drivers that want an irq in the kernel, 
but aside from this ADNS-9800, I don't need that functionality.

Thanks,
Daniel