nuttx
nuttx copied to clipboard
apache
USB HID Keyboard causes crash
SAMA5D2 processor.
With the USB HID Keyboard driver enabled, a sequence of 2 or 3 attach-detach cycles causes a crash. Other USB host drivers (e.g. Mass Storage) behave OK.
This is the first stack dump crash I've encountered, and happy to investigate further with a bit of guidance, or maybe someone knows right away what the issue is? I have half a suspicion that it is the OHCI Monitor based on a message I saw once.
up_dump_register: R0: 6000005f R1: 00400416 R2: 00000000 R3: 20058c00
up_dump_register: R4: 00000000 R5: 00000000 R6: 00000008 R7: 00000000
up_dump_register: R8: 2005462c SB: 00000008 SL: ffffff80 FP: 00000000
up_dump_register: IP: 00000000 SP: 20064028 LR: 2001e5e0 PC: 2001e624
up_dump_register: CPSR: 200000df
show_tasks: PID PRI STACKBASE STACKSIZE COMMAND
show_tasks: ---- --- 0x20055000 1024 irq
dump_task: 0 0 0x2005ca40 1000 Idle Task
dump_task: 1 192 0x2005e2b8 1992 hpwork 0x200573dc
dump_task: 2 192 0x2005eec8 1992 hpwork 0x200573dc
dump_task: 3 100 0x2005fad8 968 lpwork 0x20057400
dump_task: 4 100 0x200602e8 968 lpwork 0x20057400
dump_task: 5 100 0x20060b08 968 lpwork 0x20057400
dump_task: 6 100 0x20061318 968 lpwork 0x20057400
dump_task: 8 110 0x20062d30 2000 FUSB302 Monitor
dump_task: 9 100 0x20063960 2000 OHCI Monitor
dump_task: 10 100 0x20064570 2000 EHCI Monitor
dump_task: 11 100 0x200722a8 4056 nsh
@TimJTi Did you test it on STM32F4Discovery board to confirm the issue is not in your arch driver?
@acassis - I only have sama5 boards[1] and not sure if my eval boards are still usable! My thinking was that since the USB memory stick works 100% reliably and I can plug/unplug many times without issue it was unlikely to be the SAMA5 usb code - but it is possible. I think one is ohci and the other ehci so i will investigate further. After all, I did find an issue with the SAMA5 device driver so further issues are I guess rather likely.
I did trace the connect, enumeration, and detach events and they superficially seemed to make sense. the keyboard actually works the first time, and it "feels" perhaps like a big chunk of memory is not being released.
What I've not got a good handle on is debugging full blown crashes as the screenful's of register dumps seem rather pointless!
[1] Edit: just remembered I have a samv71 eval board: I'll try that.
@TimJTi Check and see if #8879 fixes this problem.
I have a hardware problem on my development board it seems and I can't quickly test this...and I'm about to go on holiday for 2 weeks. I will check it out when I'm back - sorry not to be able to do this now.
@ldube With apologies, I haven't been able to verify it yet for various reasons. I can close if wanted but if I may end up reopening if it's still an issue.
What would your prefer - wait, pending me checking in the next month or so, or close it with the potential of it being re-opened?
What would your prefer - wait, pending me checking in the next month or so, or close it with the potential of it being re-opened?
Wait.