Skip to contentSkip to conversation

Help writing a custom HID driver

DISCUSSION Jose Diez 2 participants 7 messages Nov 6, 2014 – Nov 7, 2014 HID

lore
Share & tools
lore

Jose Diez

author
#0 lore raw

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

This is my code so far: http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while atomic" error.

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

#1 lore raw

Hi Jose,

Quoted 7 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 6 lines

This is my code so far: http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 3 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Cheers,
Benjamin

author
#2 lore raw

Thanks Benjamin. That fixed the issue. Have a great day.

Quoted 36 lines · Benjamin Tissoires

Hi Jose,

Quoted 7 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 6 lines

This is my code so far: http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 2 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Cheers,
Benjamin
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author
#3 lore raw

Okay, so I've solved the watchdog issue, and now I want the driver to be loaded automatically. I've copied the module to /lib/modules/<kernel>/ and run depmod -a, and I can see it in modules.alias, but it looks like hid_generic grabs it first upon boot - my module is loaded correctly, but it doesn't grab the HID device.

If I modprobe -r hid_generic and then load my module, everything works correctly.

Quoted 44 lines · Jose Diez

Thanks Benjamin. That fixed the issue. Have a great day.

Quoted 36 lines · Benjamin Tissoires

Hi Jose,

Quoted 7 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 6 lines

This is my code so far: http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 2 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Cheers,
Benjamin

Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#4 lore raw
Quoted 8 lines · Jose Diez

Okay, so I've solved the watchdog issue, and now I want the driver to be loaded automatically. I've copied the module to /lib/modules/<kernel>/ and run depmod -a, and I can see it in modules.alias, but it looks like hid_generic grabs it first upon boot - my module is loaded correctly, but it doesn't grab the HID device.

If I modprobe -r hid_generic and then load my module, everything works correctly.

Yep. You need to add your device VendorID/ProductID in hid_have_special_driver[] in hid-core.c.
This will tell hid-generic not to grab your device, and your custom module will take over.

Cheers,
Benjamin

Quoted 56 lines · Jose Diez
Quoted 52 lines

Thanks Benjamin. That fixed the issue. Have a great day.

Quoted 36 lines · Benjamin Tissoires

Hi Jose,

Quoted 9 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have
to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 7 lines

This is my code so far: http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while
atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 2 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Quoted 6 lines

Cheers,
Benjamin
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author
#5 lore raw

Ah, I was hoping there would be a way to avoid having to patch the kernel itself.

Thanks again,
Jose

Quoted 67 lines · Benjamin Tissoires
Quoted 8 lines · Jose Diez

Okay, so I've solved the watchdog issue, and now I want the driver to be loaded automatically. I've copied the module to /lib/modules/<kernel>/ and run depmod -a, and I can see it in modules.alias, but it looks like hid_generic grabs it first upon boot - my module is loaded correctly, but it doesn't grab the HID device.

If I modprobe -r hid_generic and then load my module, everything works correctly.

Yep. You need to add your device VendorID/ProductID in hid_have_special_driver[] in hid-core.c.
This will tell hid-generic not to grab your device, and your custom module will take over.

Cheers,
Benjamin

Quoted 42 lines · Jose Diez
Quoted 41 lines

Thanks Benjamin. That fixed the issue. Have a great day.

Quoted 38 lines · Benjamin Tissoires

Hi Jose,

Quoted 8 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have
to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 7 lines

This is my code so far:http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while
atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 2 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Cheers,
Benjamin
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html

Quoted 4 lines
Quoted 4 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html

author
#6 lore raw

Hi, it's me again.

So I'm trying to use the Linux input system to report key events from the device. This is my code so far: http://sprunge.us/KfhF

As you can see, I'm trying to send event 1 whenever the watchdog work is executed. I can see from dmesg that the task is actually being executed, but when I run evtest to check if I actually receive the messages, I see nothing.

Any help would be much appreciated.

Thanks,
Jose

Quoted 77 lines · Jose Diez

Ah, I was hoping there would be a way to avoid having to patch the kernel itself.

Thanks again,
Jose

Quoted 67 lines · Benjamin Tissoires
Quoted 8 lines · Jose Diez

Okay, so I've solved the watchdog issue, and now I want the driver to be loaded automatically. I've copied the module to /lib/modules/<kernel>/ and run depmod -a, and I can see it in modules.alias, but it looks like hid_generic grabs it first upon boot - my module is loaded correctly, but it doesn't grab the HID device.

If I modprobe -r hid_generic and then load my module, everything works correctly.

Yep. You need to add your device VendorID/ProductID in hid_have_special_driver[] in hid-core.c.
This will tell hid-generic not to grab your device, and your custom module will take over.

Cheers,
Benjamin

Quoted 46 lines · Jose Diez
Quoted 45 lines

Thanks Benjamin. That fixed the issue. Have a great day.

Quoted 38 lines · Benjamin Tissoires

Hi Jose,

Quoted 8 lines · Jose Diez

Hello linux-input,

I'm trying to write a custom HID driver. It works fine, and I can send reports just fine, but one of the requirements of this device is that I have
to reply to reports with code 62 with another report with code 62, which resets a watchdog in the device.

Quoted 7 lines

This is my code so far:http://codepad.org/m4QiWhDt

The problem is in line 40. It seems like I'm not allowed to call hid_hw_output_report from the raw_event callback handler. I've tried surrounding the call with spin_locks, but I still get the "scheduling while
atomic" error.

Yeah, when you are in the .event callback, you are basically called by an IRQ, so you can not schedule a potentially blocking operation.

Quoted 2 lines

I'm not sure how to approach this - can someone help? It would be much appreciated. Thanks.

I would use a worker to do what you are trying to do. You can have a look at the reset_worker we have in drivers/hid/hid-rmi.c. When the event is not one we expected, we schedule a worker thread which then sends an output report to the device. This way, the blocking operation is sent from a different thread than the IRQ one. It is kind of what you are willing to do.
There are many other examples of workers in the hid subtree, or you can refer to the doc to find out more.

Cheers,
Benjamin

Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html
Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html
Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message [email protected]
More majordomo info athttp://vger.kernel.org/majordomo-info.html
Signature · 3 lines
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html