Trouble with HOST HID examples

[robertofrechill]

Hi! I´m working with Bare metal USB Stack 4.1.1.

I´m debbugging this HOST HID examples in Codewarrior 10.3 and it doesn´t work! :smileysad:

At first appears the message: 

USB HID Mouse

Waiting for USB Mouse to be attached...

then, when i plug a mouse, the message: '' ISR is entered on vector number 0x3" and it repit forever while device is plugging.

Anybody could help me with this? thanks!

[robertofrechill]

Hi guys! I fixed the problem!! the solutions are comment out this lines of driver/khci_kinetis.c file:

49:  //#define _BDT_RESERVED_SECTION_  // TBD

.......

592://#if!(defined _BDT_RESERVED_SECTION_) /* << EST */

.......

594://#endif /* << EST */

Thanks all for your help

[robertofrechill]

sorry, i forget say that i´m working with TWR-K20D72M Tower kit complete.

[admin]

Hey,

For what it's worth... I don't think it's related to the hardware.  This hidmouse example works with USB stack 4.0.3, just not the latest OS-less 4.1.1 (I'm seeing the same Hard Fault on the TWR-K20D50M).  For what it's worth...according to the vector table (in vectors.c), that vector number corresponds to "ivINT_Hard_Fault". 

It may help to study the differences between the two stacks to see what key thing might have changed and caused this.  I'm interested to investigate this as well.

-Rich

[admin]

Hey,

I found where the issue was and was able to get this working.  Navigate to the driver/khci_kinetis.c file and comment out line 49:

49:  //#define _BDT_RESERVED_SECTION_  // TBD

This may not be the best solution, but it should get you up and going for your hid mouse example (this _BDT_RESERVED_SECTION_  was previously undefined in 4.0.3 and appeared to be the key difference for me).  It now works on my end on the TWR-K20D50M.  Also, since the modification is to the low-level driver code which is shared by the other USB classes on your machine, you might want to make a note of this in case this breaks other USB apps.

Let me know if you have any luck with this.

Rich

[BlackNight · Accepted solution]

That define is used to control if a global variable is reserved for the bdt USB buffer, or if the memory for the buffer shall by allocated dynamically (e.g. malloc()).

Are you using the Freescale (non-gcc) ARM compiler?

If that define is defined, then this code is active:

	#ifdef __CWCC__
	#pragma define_section usb_bdt ".usb_bdt" RW
	__declspec(usb_bdt) uint_8_ptr bdt;

It means that there needs to be a memory area allocated in the linker file for the USB bdt block, otherwise it will fail.

Obviously this is missing in the example :-(.

In my applications I don't want depend on malloc() for the buffer, so I use a static variable. But then it depends on the linker/compiler how to allocate this variable. The problem is that it needs to be aligned at a 512 byte boundary.

With ARM gcc this can be done in the source with

__attribute__((__aligned__(512))) uint8 bdt[512];

but not with the Freescale Kinetis compiler. To my knowledge you cannot align a variable in the source. This needs to be done in the linker file.

So I would say a better fix would be this:

/* Global variables */

#ifndef __CWCC__ /* Freescale (non-gcc) Kinetis compiler. If the macro below is enabled, the bdt buffer needs to be defined in the linker file */

  #define _BDT_RESERVED_SECTION_ /* if defined, we use a static buffer (instead of malloc()), on a 512 byte boundary */

#endif

I hope this helps to clarify the issue. I have passed this to the USB stack team.

[mauroansaloni]

Hi. Did you try with a different mouse?

just to see if the problem is general or maybe triggered by some devices