can not send interrupt with panda target

[Moirai7]

Hi there,

I tried to enable interrupt and inject interrupt, but the message sent to QMPProocol shows:

{"id": 3, "error": {"class": "CommandNotFound", "desc": "The command avatar-armv7m-enable-irq has not been found"}} '{"id": 5, "error": {"class": "CommandNotFound", "desc": "The command avatar-armv7m-inject-irq has not been found"}}

Another error message is:

avatar.targets.PandaTarget0.ARMV7MInterruptProtocol.ERROR | Unable to create rx_queue: No queue exists with the specified name /avatar_v7m_irq_rx_queue

It seems even though PandaTarget inherits QEMUTarget, it doesn't support interrupts yet.

[mariusmue]

Hi,

This is correct, IRQ forwarding, as described in pretender, is currently only supported on avatar2's QemuTarget - which uses a more recent version of QEMU and has a bit of differences in the QMP protocol. To enable IRQ injection via avatar2, I see two ways forward:

1) Backport IRQ injection from QEMU/Pretender:

In essence, to support IRQ injection via QMP, a couple of things are required:

• Implement Interrupt Logic in hw/avatar
• Expose these methods to QMP
• Adjust build files to include the interrupt implementation.

You can see according changes in this commit of the qemu target, which would needed to be backported: https://github.com/avatartwo/avatar-qemu/commit/2de8b91f36f71584c34df899e8c2cc57fbad806f

I don't have the time to do this backport right now, but if you want to give it a try, we can try to get it upstreamed to panda's dev branch once this is ready.

2) Using Pandas Python APIs/PyPanda: Using PANDA's recent extension to their python APIs, I was able to implement interrupt directly from Python. This would require to use the PyPanda target, which provides additional performance in any case. I can try to upstream these patches, or make it at least available in my PANDA fork. For this, it would be good to know what target you are interferring with - is it a Cortex-M class device, or something else? Interrupts implementation do change, unfortunately.

Let me know which of these two ways you prefer!

[Moirai7]

Hi Marius,

Thank you for the quick reply! We are working on Cortex-M class devices. It would be very helpful if the PyPanda target could be used to inject interrupt.

I tried to backport the injection implements of QEMU, but it takes time to locate all dependencies. I will try it again with the commit log you provide.

Thanks again! I really appreciate your help.

[mariusmue]

Hi! We finally released FirmWire, which has a modified Panda version with irq-injection over pypanda. You can find over here: https://github.com/FirmWire/panda

The important commit is this one: https://github.com/FirmWire/panda/commit/8eb9767dab63cce9f82946afe98c52076304ae0c

In essence, you need to add irq-logic to the configurable machine, and then expose the important functions in the header-file via comments which will be parsed by pypanda. That is, in the header file, you need to add those function between the following block:

// BEGIN_PYPANDA_NEEDS_THIS -- do not delete this comment bc pypanda
// api autogen needs it.  And don't put any compiler directives
// between this and END_PYPANDA_NEEDS_THIS except includes of other
// files in this directory that contain subsections like this one.

// from panda_api.c

[functions to expose]

// END_PYPANDA_NEEDS_THIS -- do not delete this comment!

Example for this can be found here: https://github.com/FirmWire/panda/blob/8eb9767dab63cce9f82946afe98c52076304ae0c/include/hw/avatar/configurable_machine.h