[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [bluetooth-dev] Alignement fix + kernel mode 2.4
Why should I care about kernel ioctl.h when I am building a module ?
I have gone through bluetooth.c and btd.c. It looks like the ttyBTC driver
(bluetooth.c) receives an ioctl that it doesn't recognize and forwards it to
the serial driver. But I don't see in btd.c where such an ioctl is sent.
Can you (maybe people from Axis) confirm that all ioctls sent by btd.c to the
ttyBTC device should be interpreted directly by the ttyBTC driver (and not
forwarded to the serial port) ?
Le Jeudi 01 Mars 2001 23:45, Gordon McNutt a écrit :
> david LIBAULT wrote:
> > Hi,
> > I did a cvs update on my modified version and tried to build a kernel
> > mode version of the stack. Compilation goes fine, with usual warnings
> > (not coming from my modifications... probably work in progress).
> > By the way, in the /drivers/char/bluetooth/Makefile, the ld should be
> > changed to $(LD), so it gives us a change to "cross-link", and the
> > -include directive shouldn't always point to /usr/include/... but again
> > to a variable that would point to the cross-compiler include files.
> > On the ARM, I can insmod the module, but when I start btd, the following
> > happens :
> > / # btd -m -u /dev/ttyS1 -s 57600
> > Bluetooth ControBT (ldisc) bt_tty_open
> > BT SYS: Setting BT driver to use serial tty
> > lBT (driver) bt_open on line 7
> > BT SYS: Registering tty on line 7
> > BT SYS: Now 1 open fd:s for ttyBTC
> > BT (driver) bt_ioctl: forwarding ioctl 0x540b to serial driver
> Just a guess at this point... did you compile against the right ioctl.h for
> the kernel you are running on the ARM?
> > Internal error: branch through zero: 0
> > CPU: 0
> > pc : [<00000004>] lr : [<400b1180>]
> > sp : bffffdd4 ip : c1635000 fp : c05c3f5c
> > r10: c178c480 r9 : c167c0e0 r8 : c178c480
> > r7 : c04b6000 r6 : 00000000 r5 : 00000002 r4 : 0000540b
> > r3 : 00000002 r2 : 0000540b r1 : c178c480 r0 : c1635000
> > Flags: NzCv IRQs on FIQs on Mode SVC_32 Segment user
> > Control: 017D Table: C05BC015 DAC: 00000015
> > Process btd (pid: 43, stackpage=c05c3000)
> > ...
> > and btd stops. Remark : it is impossible to rmmod the module as it says
> > it is busy. If I restart btd, then it says : BT SYS: Now 2 open fd:s for
> > ttyBTC...
> > Is this ioctl forwarded to the actual serial port (ie /dev/ttyS1) or one
> > of the ttyBTx ?
> Yes, but... since your program counter is so close to zero I wonder if, for
> some reason, the driver->ioctl is NULL? Without looking I'll bet
> bluetooth.c doesn't check this (because it shouldn't be NULL).
> > Should I have a look at the serial driver of my board ? what is
> > that ioctl supposed to do ?
> You might check that you're compiling against the correct kernel headers
> (most cross-compilers internally hardcode a path to a set of specific
> kernel headers, so if they don't match your kernel version it's possible
> that, perhaps, the tty_driver struct changed and the location of the ioctl
> field changed... this would explain the symptom, anyway...).
> If you're certain that you are you might add a debug statement to verify
> that driver->ioctl is not NULL before we call it in bluetooth.c.
> To unsubscribe from this list: send the line "unsubscribe bluetooth-dev" in
> the body of a message to email@example.com
To unsubscribe from this list: send the line "unsubscribe bluetooth-dev" in
the body of a message to firstname.lastname@example.org