Re: [ECOS] How DSR is executed in interrupt ?

["ariga masahiro" <ariga at link-lab dot co dot jp>]

Hi,

Thank you , Andrew,for your reply.

Andrew wrote--
> Are you sure the ISR is being called multiple times?
Definitely.Forever repeating.

> This masks the interrupt. i.e. it should disable the interrupt
> controller from causing further interrupts from this source. The
> interrupt is usually unmasked in the delivery function once the
> hardware has been reprogrammed to de-assert the interrupt. Since you
> say the DSR is not getting called, i don't see how the interrupt can
> get unmasked and so i don't see how the ISR can be called again.....
You are right I'm confirmed doing masking in ISR.

> This suggests that cyg_drv_interrupt_mask() is not working. It is not
> masking the interrupt. I would take a look at your HAL and check the
> interrupt handling code.
I studied cyg_drv_interrupt_mask() and learned it calls
packages\kernel\v2_0\src\intr\intr.cxx(614): 
Cyg_Interrupt::mask_interrupt(cyg_vector vector)
But it's too complicated.
Is there any way (like diag_printf) I could check entering this class 
function and working correctly ?
I couldn't break in this function in ICE.

If this function is working properly , it's outside of  ISR where unmasking 
is executed.
Is it possible  ISR returns in VSR and  somewhere did unmask and entering 
ISR repeatedly ?
Since DSR was never called,I think it's somewhere except DSR.

> As i've said before v2_0 is very old. It could be this has been fixed
> in newer code....
I am desperately trying to revise source using WinCVS,but WinCVS
can't operate properly.I still am trying.
Do I need to revise configtool also ?

Masahiro Ariga

-----------------------------------------------------------------------------------
> On Mon, Sep 10, 2007 at 05:21:21PM +0900, ariga masahiro wrote:
> > Hi,
> >
> > Please help me next question ?
> >
> > I am trying to run LAN91CXX drive on interrupt,
> > and it became to enter ISR routine when IRQ3 Line is asserted.
> >
> > But it looks like IRQ3 Line is ON-State forever and never become 
> > OFF-State.
> > ISR is repetedly entered but never enter into DSR routine.
> >
> > ISR/DSR are setted like this in smsc_lan91cxx_init(),
> >    // Initialize environment, setup interrupt handler
> >    cyg_drv_interrupt_create(cpd->interrupt,
> >                            8, // Priority - what goes here?
> >                             (cyg_addrword_t)sc, //  Data item passed to
> > interrupt handler
> >                             (cyg_ISR_t *)lan91cxx_isr,
> >                             (cyg_DSR_t *)eth_drv_dsr, // The logical 
> > driver
> > DSR
> >                             &lan91cxx_interrupt_handle,
> >                             &lan91cxx_interrupt);
> >    cyg_drv_interrupt_attach(lan91cxx_interrupt_handle);
> >
> > ISR is like below,
> > static int
> > lan91cxx_isr(cyg_vector_t vector, cyg_addrword_t data
> >             /* , HAL_SavedRegisters *regs */ )
> > {
> >    struct eth_drv_sc *sc = (struct eth_drv_sc *)data;
> >    struct lan91cxx_priv_data *cpd =
> >        (struct lan91cxx_priv_data *)sc->driver_private;
> >
> >    DEBUG_FUNCTION();
> >
> >    INCR_STAT( interrupts );
> >
> >    cyg_drv_interrupt_mask(cpd->interrupt);
> >    cyg_drv_interrupt_acknowledge(cpd->interrupt);
> >    return (CYG_ISR_HANDLED|CYG_ISR_CALL_DSR);  // Run the DSR
> > }
>
> Are you sure the ISR is being called multiple times?
>
> This masks the interrupt. i.e. it should disable the interrupt
> controller from causing further interrupts from this source. The
> interrupt is usually unmasked in the delivery function once the
> hardware has been reprogrammed to de-assert the interrupt. Since you
> say the DSR is not getting called, i don't see how the interrupt can
> get unmasked and so i don't see how the ISR can be called again.....
>
> This suggests that cyg_drv_interrupt_mask() is not working. It is not
> masking the interrupt. I would take a look at your HAL and check the
> interrupt handling code.
>
> > I think /packages/io/eth/v2_0/src/net/eth_drv.c file's
>
> As i've said before v2_0 is very old. It could be this has been fixed
> in newer code....
>
>   Andrew


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