[ECOS] BLX raises an exception in cyg_start

[ECOS] BLX raises an exception in cyg_start

[Davide Pippa]

Hi!

I'm still trying to get my first application run with eCos, on
TWR-K70F120M board.
I've got it compiling, and debugging thru openocd. I'm now using eCos
gnutools 4.6.3
for compilation, now my problem is that everything seems ok, but my
main() function
never gets called.
I've seen there is no reference to in in the whole disassembled image, I suppose
then the "hello world" tutorial in the docs is outdated?

I've tried changing main() into cyg_user_start().
Then I see a change, as in cyg_start() I see a cyg_user_start() call:

00001f64 <cyg_start>:
    1f64:	b508      	push	{r3, lr}
    1f66:	f000 f807 	bl	1f78 <cyg_prestart>
    1f6a:	f000 f807 	bl	1f7c <cyg_package_start>
    1f6e:	f7fe eab0 	blx	4d0 <cyg_user_start>
    1f72:	f002 f9b7 	bl	42e4 <_ZN13Cyg_Scheduler5startEv>
    1f76:	bf00      	nop
	
My problem now is that when I hit that "blx	4d0" line, instead of
correctly going into cyg_user_start at 0x4d0,
things go there:

0000043c <hal_default_exception_vsr>:
     43c:	f3ef 8009 	mrs	r0, PSP
     440:	f1a0 0130 	sub.w	r1, r0, #48	; 0x30
     444:	f381 8809 	msr	PSP, r1
     448:	f04f 0101 	mov.w	r1, #1
     44c:	f3ef 8205 	mrs	r2, IPSR
     450:	f3ef 8311 	mrs	r3, BASEPRI
     454:	e920 4ffe 	stmdb	r0!, {r1, r2, r3, r4, r5, r6, r7, r8, r9, sl, fp, lr}
     458:	4604      	mov	r4, r0
     45a:	f000 fc65 	bl	d28 <hal_deliver_exception>
     45e:	4620      	mov	r0, r4
     460:	e8b0 4ffe 	ldmia.w	r0!, {r1, r2, r3, r4, r5, r6, r7, r8, r9,
sl, fp, lr}
     464:	f380 8809 	msr	PSP, r0
     468:	f383 8811 	msr	BASEPRI, r3
     46c:	4770      	bx	lr

And i get, into openocd, a message that indicates that an exception has risen;
here is a debug session:

... reset halt thing ...

> bp 0x1f6e 4
 breakpoint set at 0x00001f6e
breakpoint set at 0x00001f6e
> resume
> target state: halted
target halted due to breakpoint, current mode: Thread
xPSR: 0x61000000 pc: 0x00001f6e psp: 0x2000ffe8
                                             target state: halted
target halted due to breakpoint, current mode: Thread
xPSR: 0x61000000 pc: 0x00001f6e psp: 0x2000ffe8
> rbp 0x1f6e
> step
 target state: halted
target halted due to single-step, current mode: Handler UsageFault
                                                                  xPSR: 0x610000
06 pc: 0x0000043c msp: 0x2000ffd0target state: halted

target halted due to single-step, current mode: Handler UsageFault
xPSR: 0x61000006 pc: 0x0000043c msp: 0x2000ffd0
>

I see "current mode" going to Handles UsageFault.

I looked into cortex-m architecture reference manual, and seen that "blx"
can raise such an exception. Anybody has any clue on what is happening?

Another thing I cannot understand is that: I see that line in disassembly:

    1f6e:	f7fe eab0 	blx	4d0 <cyg_user_start>

that seems to indicate a 32bit-encoded instruction, but on the Arm-M
Reference Manual
I can only find a 16-bit encoded "BLX" instruction, being referred to
a register, not
to an absolute / relative jump. Is that a mis-disassembly? Or am I
missing something?

I've tried compiling eCos in 2 configurations, the first using
defaults (twr-k70f120m with lwip stack),
the second changing compile flags to have -mcpu=cortex-m4 instead oc
-mcpu=cortex-m3;
anyway, the "BLX" problem happens...

Thanks for any help :)

Pyper

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: BLX raises an exception in cyg_start

[Davide Pippa]

Sorry I have found the problem!
I was compilinig my application (that single-line main) with wrong cpu
flags (default, then I suppose non-cortex-m and non-thumb),
then I got a (malformed?) blx; adding correct flags to my application
compilation got it right now, and I'm getting into cyg_user_start.

Davide

2013/2/16 Davide Pippa <thepyper@gmail.com>:
> Hi!
>
> I'm still trying to get my first application run with eCos, on
> TWR-K70F120M board.
> I've got it compiling, and debugging thru openocd. I'm now using eCos
> gnutools 4.6.3
> for compilation, now my problem is that everything seems ok, but my
> main() function
> never gets called.
> I've seen there is no reference to in in the whole disassembled image, I suppose
> then the "hello world" tutorial in the docs is outdated?
>
> I've tried changing main() into cyg_user_start().
> Then I see a change, as in cyg_start() I see a cyg_user_start() call:
>
> 00001f64 <cyg_start>:
>     1f64:       b508            push    {r3, lr}
>     1f66:       f000 f807       bl      1f78 <cyg_prestart>
>     1f6a:       f000 f807       bl      1f7c <cyg_package_start>
>     1f6e:       f7fe eab0       blx     4d0 <cyg_user_start>
>     1f72:       f002 f9b7       bl      42e4 <_ZN13Cyg_Scheduler5startEv>
>     1f76:       bf00            nop
>
> My problem now is that when I hit that "blx     4d0" line, instead of
> correctly going into cyg_user_start at 0x4d0,
> things go there:
>
> 0000043c <hal_default_exception_vsr>:
>      43c:       f3ef 8009       mrs     r0, PSP
>      440:       f1a0 0130       sub.w   r1, r0, #48     ; 0x30
>      444:       f381 8809       msr     PSP, r1
>      448:       f04f 0101       mov.w   r1, #1
>      44c:       f3ef 8205       mrs     r2, IPSR
>      450:       f3ef 8311       mrs     r3, BASEPRI
>      454:       e920 4ffe       stmdb   r0!, {r1, r2, r3, r4, r5, r6, r7, r8, r9, sl, fp, lr}
>      458:       4604            mov     r4, r0
>      45a:       f000 fc65       bl      d28 <hal_deliver_exception>
>      45e:       4620            mov     r0, r4
>      460:       e8b0 4ffe       ldmia.w r0!, {r1, r2, r3, r4, r5, r6, r7, r8, r9,
> sl, fp, lr}
>      464:       f380 8809       msr     PSP, r0
>      468:       f383 8811       msr     BASEPRI, r3
>      46c:       4770            bx      lr
>
> And i get, into openocd, a message that indicates that an exception has risen;
> here is a debug session:
>
> ... reset halt thing ...
>
>> bp 0x1f6e 4
>  breakpoint set at 0x00001f6e
> breakpoint set at 0x00001f6e
>> resume
>> target state: halted
> target halted due to breakpoint, current mode: Thread
> xPSR: 0x61000000 pc: 0x00001f6e psp: 0x2000ffe8
>                                              target state: halted
> target halted due to breakpoint, current mode: Thread
> xPSR: 0x61000000 pc: 0x00001f6e psp: 0x2000ffe8
>> rbp 0x1f6e
>> step
>  target state: halted
> target halted due to single-step, current mode: Handler UsageFault
>                                                                   xPSR: 0x610000
> 06 pc: 0x0000043c msp: 0x2000ffd0target state: halted
>
> target halted due to single-step, current mode: Handler UsageFault
> xPSR: 0x61000006 pc: 0x0000043c msp: 0x2000ffd0
>>
>
> I see "current mode" going to Handles UsageFault.
>
> I looked into cortex-m architecture reference manual, and seen that "blx"
> can raise such an exception. Anybody has any clue on what is happening?
>
> Another thing I cannot understand is that: I see that line in disassembly:
>
>     1f6e:       f7fe eab0       blx     4d0 <cyg_user_start>
>
> that seems to indicate a 32bit-encoded instruction, but on the Arm-M
> Reference Manual
> I can only find a 16-bit encoded "BLX" instruction, being referred to
> a register, not
> to an absolute / relative jump. Is that a mis-disassembly? Or am I
> missing something?
>
> I've tried compiling eCos in 2 configurations, the first using
> defaults (twr-k70f120m with lwip stack),
> the second changing compile flags to have -mcpu=cortex-m4 instead oc
> -mcpu=cortex-m3;
> anyway, the "BLX" problem happens...
>
> Thanks for any help :)
>
> Pyper

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss