Re: [microblaze-uclinux] Heavy packet load and the sigaction..

[Prasad <ndprasad@xxxxxxxxx>]

Yes, this is a regular pattern. As soon as i kill the test code and
re-run it, it comes ok for few secs under heavy load and the first
occurence of calling signal handler in the IRQ context, i start to
observe the signal handler running with the interrupts disabled.

(PS: I used setitimer to start a periodic timer. I tried with both
sigaction/signal to register for  function to capture the SIGALRM.
Both behaves identically)

- prasad


On 11/16/06, John Williams <jwilliams@xxxxxxxxxxxxxx> wrote:
> Hi Prasad,
>
> Prasad wrote:
>
> > ==================================================
> >  /* Signal handler args: */
> >    regs->gpr[GPR_ARG0] = signal; /* Arg 0: signum */
> >    regs->gpr[GPR_ARG1] = &frame->sc; /* arg 1: sigcontext */
> >
> >   __asm__ __volatile__("mfs %0, rmsr; nop;" : "=r" (flags));
> >
> >    printk ("inside do_signal: are 0x%08x\n", flags);
> >    regs->pc = ((unsigned long)ka->sa.sa_handler);
> > ===================================================
> >
> > and the output i got is this when i send  a packet burst..
> > =======================================================
> > inside setup_frame: are 0x000000aa
> > flags are 0x000000a2
> > inside setup_frame: are 0x000000aa
> > flags are 0x000000a2
> > inside setup_frame: are 0x000000aa
> > flags are 0x000000a2
> > inside setup_frame: are 0x000000aa
>
> 0xaa shows BIP and EI are set - setup_frame is called during a syscall (brki)
> context.
>
> > flags are 0x000000a2
> > inside setup_frame: are 0x000000a2
>
> 0xa2 means BIP is not set - we are in IRQ context.
>
> > flags are 0x000000a2
> > inside setup_frame: are 0x000000aa
> > flags are 0x000000a0
> > inside setup_frame: are 0x000000aa
> > flags are 0x000000a0
>
> This is interesting, it's like the first time the sighandler runs after
> setup_frame occurs in IRQ context, it's fine.  Then the next time after
> setup_frame runs in syscall context, the handler runs with interrupts disabled.
>
> Is this a regular pattern?
>
> > I am not sure whether this points to something significant to debug.
> > If so, please let me know what else i can debug.
>
> A useful thing you can do is try to identify any patterns in sequences like
> above.  Unfortunately I won't have any time to look at this until next week at
> the earliest, so the clearer picture you can give me on what's going on, the
> better it will be when I do get to it.
>
> Cheers,
>
> John
>
> ___________________________
> microblaze-uclinux mailing list
> microblaze-uclinux@xxxxxxxxxxxxxx
> Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
> Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/
___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/