pc686: --enable-rtems-debug not working

[Joel Sherrill]

On Tue, Apr 16, 2019 at 8:01 AM <Jan.Sommer at dlr.de> wrote:

> Hello,
>
> if I build the pc686 BSP with --enable-rtems-debug activated I can't run
> any of the testsuites examples, but get the following error:
>
> $ qemu-system-i386 -append --console=/dev/com1 -no-reboot -serial stdio
> -monitor none -nographic  -kernel
> ./build/i386-rtems5/c/pc686/testsuites/samples/ticker.exe
> i386: isr=0 irr=1
> assertion "bsp_interrupt_is_valid_vector(vector)" failed: file
> "../../../../../../../../refex-rtems/c/src/lib/libbsp/i386/pc386/../../../../../../bsps/i386/shared/irq/irq.c",
> line 275, function: bsp_interrupt_vector_enable
>
> *** FATAL ***
> fatal source: 7 (RTEMS_FATAL_SOURCE_ASSERT)
> fatal code: 1288832 (0x0013aa80)
> RTEMS version: 5.0.0.be50969881b97180bf4fc1e2975efd41169e08bb
> RTEMS tools: 7.4.0 20181206 (RTEMS 5, RSB
> 9a3e12e5820918057633798c3fe2a1f952fb4e56, Newlib 1d35a003f)
> executing thread is NULL
>
> It seems to stem from rtems/bsps/i386/pc386/clock/ckinit.c:165:
>
> > bsp_interrupt_vector_enable( BSP_PERIODIC_TIMER - BSP_IRQ_VECTOR_BASE );
>
> because "BSP_PERIODIC_TIMER - BSP_IRQ_VECTOR_BASE" equates to 0xFFFFFFE0,
> but the subsequent assert on bsp_interrupt_is_valid_vector() only allows
> vectors between 0 and 16.
> The BSP still works with the option disabled, but it makes debugging more
> difficult.
> Does someone has some guidance on this problem. I could remove the assert
> in bsp_interrupt_vector_enable, but I am a bit hesitant with simply
> removing things from core files.
>

I suspect based on the other uses and math that this check is valid but
that the calls to bsp_interrupt_vector_enable and
bsp_interrupt_vector_disable in pc386/clock/ckinit.c should not be
subtracting the BSP_IRQ_VECTOR_BASE. BSP_IRQ_VECTOR_BASE is set to
BSP_IRQ_VECTOR_BASE which is used in conjunction with IDT manipulation. The
vector passed into  bsp_interrupt_vector_*able directly corresponds to a
source on the two i2859s (0-15).

I would try deleting the subtraction in ckinit.c. The interrupt is likely
already on so as written it is probably a NOP.

The similar subtraction in btimer.c is probably right because the benchmark
timer is installing a raw timer and the math is subtracting from an IDT
index.

This is all a remnant from converting to the newer BSP interrupt Framework.
Just a nit.

Give that a try. It would be nice to see debug work on the pc386.

--joel


>
> Best regards,
>
>    Jan
>
>
> Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)
> German Aerospace Center
> Simulation and Software Technology | Software for Space Systems and
> Interactive Visualization | Lilienthalplatz 7 | 38108 Braunschweig | Germany
>
> Jan Sommer
> Telephone +49 531 295-2494 | Telefax 0531 295-2767 | jan.sommer at dlr.de
> DLR.de/SC
>
> _______________________________________________
> users mailing list
> users at rtems.org
> http://lists.rtems.org/mailman/listinfo/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20190418/f9713c3b/attachment.html>